literature review & thesis aims

Effect of Perivascular Cells on Retinal Endothelial Cell Permeability

1

CHAPTER 1

LITERATURE REVIEW &

THESIS AIMS


2

1.1 RETINAL ANATOMY AND PHYSIOLOGY

1.1.1 Background

The retina consists of two primary layers: an inner neurosensory layer and an outer

layer comprised of epithelial cells known as retinal pigmented epithelium (RPE). The

neurosensory retina is further divided into 10 distinct layers, including (from the

vitreous) the nerve fibre layer (NFL), the ganglion cell layer (GCL), inner plexiform

layer (IPL), inner nuclear layer (INL), outer plexiform layer (OPL), outer nuclear

layer (ONL) and the photoreceptor layer (PRL) (Figure 1.1A, B).

The retinal layer adjacent to the RPE is composed exclusively of photoreceptors

that convert light photons into neural signals. These signals are further processed by

neighbouring layers of neurons and ultimately transmitted along ganglion cell axons

which converge in the optic nerve, for eventual integration by the brain (reviewed

Forrester et al, 1996). The three principle neuronal cells that relay impulses generated

by light are photoreceptors, bipolar cells and ganglion cells. Activity of these cells is

modulated by other cell types including horizontal cells, amacrine cells and possibly by

non-neuronal elements such as macroglia and microglial cells (Forrester et al, 1996).

1.1.2 The Neural Retina: Vertical and Horizontal Processing

a) Photoreceptor cells

Rod and cone photoreceptor cells are situated at the outer aspect of the retina (Figure

1.1B). Rods are responsible for sensing motion, contrast and brightness, while cones

are necessary for spatial resolution, colour vision and resolution of fine detail

(reviewed Kolb et al, 2001). Photoreceptor outer segments contain stacks of

membranous disks loaded with photopigments and proteins required to transduce

light into neural signals (reviewed Blanks, 2001). Photoreceptor inner segments

contain a nucleus, mitochondria and organelles needed for biosynthesis of disks and

other molecules. As new disks are formed at the base of the outer segment, the oldest

disks at the tip of the outer segment are shed into the subretinal space between the

neural retina and the RPE (Young, 1976). Cell bodies of rods and cones are

connected to specialised synaptic terminals known as spherules and pedicles,

respectively. Photoreceptor terminals synapse with bipolar and horizontal cells. The


3

abundance of rods and cones varies within different retinal regions, with rods

dominanting at the retina periphery and cones dominating within the macula.

b) Bipolar cells

Bipolar cells are primarily responsible for transmitting signals from photoreceptors

to ganglion cells (Kolb et al, 2001). Multiple bipolar cell dendrites reach out to

photoreceptors, while only a single axon synapses with ganglion and amacrine cells

(reviewed Wassle and Boycott, 1991). Bipolar cell bodies lie within the INL and are

oriented parallel to photoreceptor cells (Figure 1.1B). In the foveal region of the

central retina, the ratio of cones to bipolar cells to ganglion cells can be as high as

1:1:1, while in the peripheral retina one bipolar cell can receive stimuli from up to

50-100 rods.

c) Retinal ganglion cells

Retinal ganglion cell (RGC) bodies lie on the vitreal aspect of the IPL (Figure 1.1B).

Ganglion cell dendrites receive impulses from bipolar cells and amacrine cells

(Blanks, 2001). Six types of RGC have been described in human and primate retinas

including midget, parasol, shrub, small diffuse, garland and giant (Polyak, 1941).

While there may be up to seven layers of ganglion cell bodies in the central retina or

fovea, within the peripheral retina there may be as few as one cell layer (Curcio and

Allen, 1990; Gao and Hollyfield, 1992; Wassle et al, 1989; Wassle et al, 1990).

Ganglion cell axons form the NFL; these axons form bundles that are separated and

ensheathed by glial cells. The bundles leave the eye to form the optic nerve.

d) Horizontal cells and amacrine cells

Horizontal cells and amacrine cells mediate lateral interactions between adjacent

groups of photoreceptors, bipolar and RGC (Figure 1.1B), enabling adjacent regions

within the retina to compare the intensity of light arising from contiguous regions of

the visual field (Blanks, 2001). In the primate retina, there are two morphologically

distinct types of horizontal cells (reviewed Gallego, 1986; also Boycott et al, 1986).

The type I horizontal cell contacts only rods, while type II horizontal cells contact

only cones (Kolb et al, 1980). Stratified subpopulations of amacrine cells have


4

different connections within the retinal circuits, playing different roles in creating

and shaping the retina’s final physiologic output (Strettoi and Masland, 1996).

1.1.3 Retinal Neurogenesis

In species with highly developed visual functions such as primates, the foveal region

in the mature retina is arranged for high resolution colour vision, and is characterised

by modification of retinal lamination, specialisation of photoreceptors and variation

of the retinal vascular pattern (reviewed Provis et al, 1998). In retinal cross sections,

the foveal region is identified by a depression, in which both the GCL and INLs are

absent (Stone et al, 1973; Bunt and Armitage, 1977; Stone, 1983; Leventhal et al,

1988; Leventhal et al, 1993) (Figure 1.1C). The area is also devoid of retinal blood

vessels that might cause optical interference (Weale, 1966; Rohen and Castenholz,

1967; Wolin and Massopust, 1967). In humans, the macula region (which includes

the fovea and perifoveal regions) is subject to degenerative diseases such as age-

related macular degeneration (AMD) which affects the RPE, and diabetic retinopathy

which affects the inner retinal vasculature (reviewed Arden et al, 2005).

Retinal neurogenesis occurs in a centro-peripheral sequence around the

incipient fovea. Foveal cones are the first retinal cells generated at 12 weeks

gestation (WG) (Diaz-Araya and Provis, 1992), reaching their mature morphology

and density about 3-4 years post natal (Yuodelis and Hendrickson, 1986). Initially,

the area that will become the fovea centralis is domed in appearance (around 17-20

WG) due to dense cellular stacking of the developing neurons (Mann, 1964). In this

domed configuration, RGC reside at a maximal distance from the nearest oxygen

source, the choriocapillaris (Provis et al, 1998). The overall retinal thickness is

therefore likely to be limited by metabolic factors, and formation of the foveal

depression (in this formerly domed region) may be precipitated by a combination of

ischemia within the inner retina, underlied by centrifugal migration towards the

developing (inner retinal) vasculature and cell death (Provis et al, 1998) (For more

information about development of the inner retinal vasculature, see Chapter 1.1.5

Section a).


5

a) VEGF is a critical survival factor in the retina

Vascular endothelial growth factor (VEGF) is a 45 kDa dimeric heparin-binding

glycoprotein that is a potent endothelial cell mitogen and chemoattractant

(Gospodarowicz et al, 1989; Connolly, 1991; Ferrara et al, 1991). VEGF-A is the

most abundant type within the retina (Ahmed et al, 2000). VEGF-A has several

isoforms VEGF121, VEGF189, VEGF206 and VEGF165, which differ primarily in

the ability to bind heparin (Ahmed et al, 2000). The longer isoforms are matrix-

bound while the shorter forms are freely diffusible. More recent studies of the VEGF

family and VEGF receptors (VEGFR) have shown that there are 6 members: VEGF-

A, -B, -C, -D, -E, and placental growth factor (PlGF) all of which (except –E) are

expressed in the retina (reviewed Witmer et al., 2003). The various subtypes of

VEGF have different affinities for the receptors: VEGFR1 (flt), VEGFR2

(Flk/KDR), VEGFR3 (flt4) and VEGF co-receptors - Neutropilin 1 and 2 (NP-1, NP-

2) - which have also been identified in the retina (Witmer et al., 2003).

Vascular endothelial growth factor is a mitogen for endothelial cells (Ferrara

et al, 1991) and plays an important role in retinal vascular development as the major

hypoxia-related angiogenic factor expressed by retinal macroglia (Stone et al, 1995;

Sandercoe et al, 2003). Receptors for VEGF have also been detected in the murine

neural retina during an avascular phase of development (Robinson et al, 2001) and

lately, VEGF is recognised to be an important trophic factor with neuroprotective

potential (reviewed Storkebaum et al, 2004; also Jin et al, 2000; Yasuhara et al,

2005; Gora-Kupilas and Josko, 2005). In response to ischemic stress, retinal

photoreceptors may upregulate VEGF expression to enhance retinal cell survival

(Arden et al, 2005). In the brain, VEGF upregulation by neuronal cells preceded

angiogenesis and glial cell proliferation (reviewed Sun and Guo, 2005) which are

common sequelae of ischemic injury.


6

1.1.4 Retinal Glia

There are two types of glia in the human retina – microglia and macroglia (astrocytes

and Müller cells). Müller cells are the principal glial cell of the retina, astrocytes are

functional components of the NFL, and microglia are immune system-derived

elements that react to injury or infection in the CNS (Kolb et al, 2001).

a) Microglia

Retinal microglia occur in three forms: parenchymal microglia associated with

neurons, paravascular macrophages associated with blood vessels, and perivascular

macrophages found within the perivascular space (Penfold et al, 1991; Provis et al,

1995). Originating from the bone marrow, microglia are comprised of both

dendritiform cells (tissue microglia) and macrophages expressing leucocyte common

antigen (CD45) (Penfold et al, 1991; Penfold et al, 1993; Provis et al, 1995). During

the course of normal development, microglia clear DNA from retinal cells dying

during development (Egensperger et al, 1996) and may become activated by ex vivo

culturing (Becher and Antel, 1996) and a variety of infectious or inflammatory

neurological processes (reviewed Ling et al, 2001), including neurodegenerative

disorders in the mature retina (Zeng et al, 2000).

b) Astrocytes

Astrocytes originate from stem cells near and within the optic disc (Watanabe and

Raff, 1988) but also proliferate within the fetal retina itself (Sandercoe et al, 1999).

In the mature retina, astrocytes are largely restricted to the NFL (Ogden, 1978).

Astrocytes also have a close anatomical relationship with neuronal components in the

retina, particularly ganglion cell axons (Hollander et al, 1991). In the developing rat

retina, platelet-derived growth factor (PDGF) is expressed by RGC (PDGF-AA)

(Mudhar et al, 1993) and retinal astrocytes express the corresponding PDGF receptor

(PDGFR); this growth factor is likely to be responsible for stimulation of astrocyte

proliferation or migration into the neural retina (Mudhar et al, 1993). During

development of the retinal vasculature, astrocytes preceding the developing vessels

release VEGF in response to hypoxia that is thought to be caused by maturing

neuronal cells (Provis et al, 1997) (For further details regarding astrocyte


7

involvement in vascular development, see Chapter 1.1.5 Section a). Astrocytes and

Müller cells may function interchangeably in many respects in the mature retina

(Hollander et al, 1991) and both appear to play a role in formation of the glia

limitans around blood vessels at the inner retinal surface (Dreher et al, 1988). The

close affiliation of astrocytes with retinal vessels is highlighted by the observation

that they only occur in vascularised retinas including those of rodents, cat, primate

and humans (Schnitzer, 1987; Stone and Dreher, 1987; Schnitzer, 1988).

c) Müller cells

Müller cells provide the scaffolding around which the retinal architecture is built.

Stretching radially across the retina to form both the outer and inner limiting

membranes, Müller cells may be the most important retinal cell by supporting

neuronal functions, and influencing blood flow and permeability of retinal capillary

endothelial cells (Kolb et al, 2001). Müller cells originate from the same progenitor

stem cell that gives rise to retinal neurons (reviewed Fischer and Reh, 2003) (For

more information about the retinal progenitor cell marker nestin, see Chapter 1.1.4

Section ci). Müller cells play a role in the metabolism of glucose to lactose which is

used by photoreceptors as an energy source; in addition, they maintain retinal

homeostasis by scavenging neurotransmitter endproducts such as glutamate in the

extracellular space (reviewed Puro, 1995). Many growth factors, cytokines and

neurotransmitters are produced by Müller cells (Puro, 1995) and at the inner retina

Müller cells clear extracellular fluid from tissue spaces by means of ion channels and

aquaporins (reviewed Bringmann et al, 2004) (Refer to Chapter 1.1.4 Section cv).

i) Immunohistochemical markers of Müller cells

Cellular retinaldehyde binding protein (CRALBP) is a water soluble vitamin A-

binding protein with specific immunogenicity for retinal Müller cells and RPE

(Crabb et al, 1988). Vitamin A is an important ingredient of the (dark-adaptation)

visual cycle and undergoes repeated shuffling between RPE and photoreceptors as

bleaching of rhodopsin occurs (Bridges, 1976). Transportation of vitamin A to target

cells occurs by means of serum retinol binding protein which acts to solubilise and

enhance the vitamin’s stability to prevent its non-specific absorption (Kanai et al,


8

1968). The problem of vitamin A toxicity and limited solubility appears to be solved

by containment of the vitamin between three compartments surrounding

photoreceptor cells: RPE cells, Müller cells and the interphotoreceptor matrix (IPM)

(Bunt-Milam and Saari, 1983). Within these compartments, specific carrier proteins

may act as transporters or in metabolism of the vitamin A derivatives, or retinoids.

CRALBP forms complexes with endogenous retinoid within RPE cells and Müller

cells, while interstitial retinol-binding protein (IRBP) complexes with extracellular

retinoid within the IPM (Bunt-Milam and Saari, 1983).

As Müller cells dedifferentiate under prolonged culture conditions, CRALBP

undergoes rapid downregulation (Hauck et al, 2003) while other markers (including

GFAP - see below) are simultaneously upregulated. CRALBP is expressed within

oligodendrocytes in optic nerves and brain (Saari et al, 1997), suggesting additional

functions for CRALBP in extraretinal tissues.

The presence of both glutamine synthetase (GS) and carbonic anhydrase (CA)

in Müller cells indicates that retinal glia combine functional roles that in the brain are

undertaken by astrocytes and oligodendrocytes (Linser and Moscona, 1979; 1981). In

adult human retinas, GS immunoreactivity occurs throughout the Müller cell

cytoplasm, with intense labelling of Müller cell endfeet at the inner limiting

membrane and radial processes in the outer retina (Linser and Moscona, 1979). After

8 days in culture, GS expression was diminished (Hauck et al, 2003) because protein

expression and activity require direct neuronal-glial cell contacts (reviewed Garcia

and Vecino, 2003). Carbonic anhydrase is detectable in virtually all cells in the

undifferentiated retina. As cell specialisation progresses, the level of CA declines

rapidly in the emerging neurons and increases in Müller cells (Linser and Moscona,

1981). In the adult retina, CA activity is confined to Müller cells, the tips of the rod

outer segments and in RPE (Terashima et al, 1996). Inhibitors of CA have been used

to dilate blood vessels in animal models of ischemic retinal disease, possibly by an

indirect effect on Müller cells (Reber et al, 2003; Pedersen et al, 2005).

Both Müller cells and astrocytes contain Type III intermediate filaments, a

heterogenous group of proteins that includes glial fibrillary acidic protein (GFAP),

vimentin and nestin – see following sections. In the healthy retina, GFAP is localised

to Müller cell end-foot domains, while in astrocytes, GFAP is evenly distributed


9

throughout the cell cytoplasm (reviewed Lewis and Fisher, 2003). Müller cells

dramatically upregulate GFAP expression in response to injury (Lewis and Fischer,

2003). Upregulation of GFAP may be associated with down-regulation of

physiological markers of Müller cells such as CRALBP, CA and GS (reviewed

Guidry, 2005; also Hauck et al, 2003). Activation markers such as GFAP in Müller

cells may be a useful indicator about disease progression. After laser

photocoagulation, GFAP is upregulated in Müller cells distant from the site of injury

(Humphrey et al, 1997; Humphrey et al, 1993) indicating that Müller cell activation

may be driven by cell-mediated signals. In the CNS, upregulation of GFAP in

astrocytes is a well-recognised response to injury (reviewed Pekny and Pekna, 2004).

Mice deficient in GFAP (GFAP -/-) develop normally but exhibit polymerisation

defects in response to stress (reviewed Pekny, 2001). In addition there may be

consequences for astrocyte-neuron interactions, blood-brain barrier integrity and glial

scar formation in the central nervous system (CNS) of knockout mice (Pekny, 2001).

Vimentin is a specific marker for CNS glia (reviewed Reichenbach and

Robinson, 2005). Retinal Müller cells are frequently labelled with vimentin which

localises to cellular intermediate filaments (Famiglietti et al, 2003; Sakamoto et al,

1998), as discussed in the GFAP section above. Both vimentin and GFAP proteins

are increasingly expressed with age in the human retina (Madigan et al, 1994; Wu et

al, 2003). In human foetal retina, CRALBP-positive Müller cells also express nestin

(an intermediate filament protein expressed by neural progenitor cells) and Ki67,

suggesting that Müller cells are end-stage progenitor cells (Walcott and Provis,

2003).

Although not a characteristic marker of retinal Müller cells, smooth muscle

actin (SMA) protein is recognised as another stress-related marker upregulated in

long-term cultures as Müller cells transform into a myofibroblastic cell type (Guidry,

1996; Hauck et al, 2003). Upregulation of SMA expression is thought to be

necessary for generation of tractional forces to drive extracellular matrix (ECM)

contractions and to form the scar tissue associated with fibrocontractive retinal

disorders (Guidry, 2005).


10

ii) Role of Müller cells in neuroprotection

Müller cells and retinal neuronal cells are mutually interdependent; Müller cell

processes make physical contact with neurons which in turn, determine the formation

and characteristics of those processes (reviewed Newman and Reichenbach, 1996;

also Reichenbach and Robinson, 2005). Chao et al, (1997) observed that Müller cells

are greatly elongated to span the entire width of the retina in thick retinas of cone-

dominated species. Presumably this places a heavy metabolic burden on Müller cells

which may stimulate vascularisation by production of angiogenic factors (Chao et al,

1997).

The Müller cell p75 neurotrophin receptor is a pro-apoptotic receptor that

may enhance the Müller cell response to signals from distressed cells (Garcia and

Vecino, 2003). The receptor is located within Müller cell processes, forming 3

horizontal layers of immunoreactivity in the IPL and 1 layer within the OPL in the

rat retina. The receptor is also located on radial Müller cell processes stretching from

the inner to the outer retinal regions. The choice between survival or death signals

mediated by the p75 receptor varies depending on the cellular microenvironment

(Hammes et al, 1995) (For more information about the p75 receptor, see Chapter

1.2.7 Section b).

Neurotrophic factors from Müller cells play a significant role in cell survival

and regulation in the retina. Growth and survival characteristics of RGC are

enhanced when cells are co-cultured directly onto confluent Müller cells, or in

Müller cell conditioned medium (Garcia et al, 2002), and endothelial cell barrier

integrity is enhanced by Müller cell expression of glial cell line-derived neurotrophic

factor (GDNF) and neurturin (NTN) (Igarashi et al, 2000).

Müller cells participate in neuronal regeneration after injury by re-entering

the cell cycle, dedifferentiating, and acquisition of a progenitor phenotype in order to

give rise to different neuronal cell types (Yurco and Cameron, 2005; Fischer and

Reh, 2000).

iii) Glutamate recycling in Müller cells – GLAST and glutamine synthetase

One of the major functions of Müller cells is to prevent damage to the neural retina

by recycling of glutamate (Garcia and Vecino, 2003) which is a neurotoxic by-


11

product of neuronal metabolism. In humans, GLAST (L-glutamate-L-aspartate

transporter) immunoreactivity is localised in the cytoplasm and radial processes of

Müller cells in all retinal layers (Barnett and Pow, 2000). Uptake of glutamate by the

Müller cell GLAST is regulated by neuronal-glial cell contacts (Choi and Chiu,

1997). Glutamate is converted into the non-neuroactive compound glutamine by the

Müller cell specific enzyme, glutamine synthetase (For more details about GS, see

Chapter 1.1.4 Section ci). Changes in GLAST expression have been reported in some

ocular diseases. For example, decreased activity of both GLAST and GS has been

described in the diabetic retina (Li and Puro, 2002; Lieth et al, 2000) and after an

acute ischemic episode in rat retinas (Barnett et al, 2001). Glutamate toxicity in

Müller cells is prevented by downregulation of NMDA (N-methyl-D-aspartate)

glutamate receptors and upregulation of neurotrophin expression, including brain-

derived neurotrophic factor (BDNF) (Taylor et al, 2003).

iv) Müller cells and glutathione

Due to their ubiquitous presence in the retina, Müller cells are an important source of

glutathione for vulnerable cell species such as photoreceptors and RGC (Schutte and

Werner, 1998). Glutathione is a tripeptide of glutamine, cysteine and glycine with

antioxidant properties. In the retina, glutathione is located in horizontal cells, Müller

cells (Pow and Crook, 1995) and RPE (Huster et al, 1998). Physiological

concentrations of glutathione protect Müller cells from oxidative injury, and under

ischemic conditions in rat retina the transfer of glutathione from Müller cells to

neurons has been reported (Schutte and Werner, 1998). When glutamate uptake is

impaired, glutamate is preferentially delivered to the glutamate-glutamine pathway at

the expense of glutathione, resulting in increased production of reactive oxygen

species (ROS) (Huster et al, 2000).

v) K+ channels and aquaporins in Müller cells

Neuronal function is dependent upon the removal of extracellular potassium (K+)

ions by Müller cells (Newman and Reichenbach, 1996). Inwardly rectifying K+ ion

channels move K+ into Müller cells in retinal layers where K+ levels are increased

during illumination (Kofuji et al, 2000). Transretinal water fluxes are osmotically


12

coupled to osmolyte movement therefore, where K+ goes, water follows through

water selective channels (the aquaporins) in the plasma membrane of Müller cells

(Bringmann et al, 2004). K+ ions (and water) are released from Müller cells into

other cellular sites including blood vessels, vitreous and the subretinal space

(Newman and Reichenbach, 1996). Pathological changes in extracellular K+ leads to

disturbance of neuronal and glial cell membrane potentials (Francke et al, 1997). The

membrane characteristics of Müller cells from diseased retinas may adversely affect

retinal K+ homeostasis and other transport systems of Müller cells.

The aquaporins (AQP) are a family of homologous water-transporting

proteins that are expressed in many tissues (reviewed Verkman, 2002). Aquaporins

enhance water permeability of cellular membranes and mediate the bidirectional

movement of water in response to osmotic gradients or differences in hydrostatic

pressure. Aquaporins have not been identified in brain endothelium, although AQP-4

is strongly expressed in astrocytic foot processes that comprise the blood-brain

barrier (BBB) in close contact with endothelial cells (Nielsen et al, 1997). AQP-4 is

thought to be involved in cerebral oedema via interactions between astroglia, neurons

and endothelium (Verkman, 2002). Within the eye, there are at least 4 aquaporins

(AQP-1, AQP3, AQP4 and AQP5) (Patil et al, 1997). Two aquaporins (AQP-1 and

AQP4) have been found in RPE and glial cells in the retina (Frigeri et al, 1995; Patil

et al, 1997; Nagelhus et al, 1998) (For more detail about pathomechanisms involving

K+ ion channels and aquaporins in macular oedema, see Chapter 1.2.1).

vi) Müller cells and the blood-retinal barrier

Retinal vascular endothelium that comprises the inner blood-retinal barrier (BRB)

forms a barrier that prevents diffusion of ions and small molecules out of blood

vessels. Glial cells induce the formation of tight junctions in retinal endothelial cells

(reviewed Risau, 1991; also Janzer and Raff, 1987; Laterra et al, 1990; Tout et al,

1993) initially by association with astrocytes in the NFL, and later with Müller cells

in the deeper retinal layers (reviewed Newman, 2001). Müller cells may act as a

conduit for metabolic exchanges between the vasculature and neurons in much the

same way that has been suggested for brain astrocytes (Garcia and Vecino, 2003)

(For additional information about the BRB, see Chapter 1.3.2).


13

vii) Immunoregulatory and phagocytic roles of Müller cells

The immunoregulatory role of Müller cells in vivo is not clear. Phagocytosis by

Müller cells of apoptotic neuronal cells during development (Egensperger et al,

1996) and debris liberated into the subretinal space after experimental transplantation

of RPE (Crafoord et al, 2000) may be important secondary functions for tissue

repair. In explant cultures, Müller cells were observed ingesting dying photoreceptor

cells (Burke and Foster, 1984).

viii) Reactive gliosis in Müller cells

Müller cells respond to environmental insults in a graded or incremental fashion

(Guidry, 2005). Long before retinal features become clinically obvious, Müller cells

react to the diabetic milieu by rearrangement of internal and external cellular

structures (including hypertrophy, swelling of cellular processes and cell

proliferation) initiating a series of gliotic changes (Guidry et al, 2003). A serious

consequence of Müller cell activation due to injury or stress is irreversible scar

formation in the retina (Guidry et al, 2004). Upregulation of cytokine production and

neurotrophic factor expression by gliotic Müller cells may adversely affect the

function and survival of both neuronal and vascular elements of the retina (Lieth et

al., 2000; also Mizutani et al, 1998) (For more information about Müller cell effects

in early diabetes, see Chapter 1.2.7).

Structural changes in Müller cells may be precipitated by lesions in neuronal

tissues (Reichenbach and Robinson, 2005). Müller cell morphology alters

dramatically after destruction of RPE cells and photoreceptors by laser

photocoagulation or as a result of retinal detachment. Intravenous injections of

sodium iodate in rabbit eyes caused subretinal scar formation when Müller cells

migrated into the areas formerly occupied by RPE cells, photoreceptors and the

interphotoreceptor matrix (Korte et al, 1992) (For more information about retinal

scar formation after laser, refer to Chapter 5).


14

1.1.5 Retinal Vasculature

a) Vascular endothelium

The signal for initiation of new blood vessel growth in the developing human retina

is thought to be mediated by maturation of photoreceptors and neurons (van Driel et

al, 1990; Provis et al, 1985). Astrocytes migrate into the retina through the optic

nerve head and guide the growth of new blood vessels along cellular processes which

are laid out along the inner surface of the retina (Ling and Stone, 1988). The earliest

vessels form in the vicinity of the optic disc around 14-15 WG (Provis et al, 1997;

Sandercoe et al, 1999; Hughes et al, 2000) and spread peripherally to reach the

retinal margin at around 34 WG (Gariano et al, 1994). During the earliest stages of

vasculogenesis, the density of the developing retinal vasculature may be dependent

on the degree of differentiation and the metabolic activity of local neural elements

(Provis et al, 1998). Ashton (1966, 1970) emphasised the importance of oxygen

tension in regulating spread and modelling of the retinal vasculature. Glial cells in

the hypoxic environment stimulate new vessel growth by release of the angiogenic

growth factor, VEGF (Provis et al, 1997).

The human retina has a dual blood supply; the inner two thirds being

nourished by branches from the central retinal blood vessels and the outer third being

nourished by the choroidal circulation. The primary capillary beds of the retina are

located in the nerve fibre-ganglion cell layers (inner plexus) and within the INL

(outer plexus). Human retinal capillaries pass only as far as the sclerad margin of the

INL, the outer retina being normally avascular. Neural tissue in the retina is

extremely metabolically active, with the highest oxygen consumption of any human

organ (reviewed Yu and Cringle, 2001). The dominant oxygen-consuming layers in

the adult rat retina are photoreceptor inner segments (Linsenmeier, 1986) the OPL

and the inner portion of the IPL (Yu and Cringle, 2001). Maintenance of an adequate

oxygen supply to the retina is critical for function.

The mature vasculature consists of endothelial cells surrounded by a

perivascular space comprised of basement membrane, pericytes and perivascular

microglia (reviewed Provis, 2001). Separating the perivascular space from retinal

tissue is the glia limitans formed by astrocytic processes, Müller cell end-feet, and

paravascular microglia (Dreher et al, 1988; Provis et al, 1995). Amacrine-like cells


15

containing dopamine, substance P (SP) or nitric oxide synthase (NOS) contribute to

the glia limitans and may play a role in blood vessel autoregulation (Provis, 2001).

Autoregulation of retinal blood flow is a unique feature of the retina by which the

rate of blood flow is held constant in the face of variable arterial perfusion pressure

(Ffytche et al, 1974). The intrinsic factor that affects blood flow is the tissue oxygen

levels (Deutsch et al, 1982; Frayser and Hickham, 1964).

Retinal capillaries are characterised by a complete circumferentially oriented

endothelial cell with a lumen diameter between 3.5-6m. Endothelial cells in retinal

capillaries are non-fenestrated (reviewed Tornquist el al, 1990) however, the high

number of endocytotic vesicles suggests they are more permeable than brain

capillaries. The inner BRB exists at the level of the vascular endothelium (For further

information about endothelial cell tight junctions and the BRB, see Chapter 1.3

Section a, and 1.3.2).

Retinal endothelial cells express a variety of surface antigens and react

uniquely to various stimuli (Pohl and Kaas, 1994). The phenotypic expression of an

endothelial cell can be altered by the local ECM, soluble growth factors, and

heterotypic and homotypic cellular interactions through intercellular junctions

(reviewed Geiger and Ayalon, 1992; also Hynes, 1992).

b) Pericytes

Within the retina there is a significantly greater coverage of endothelium by

pericytes compared with brain, and basement membrane thickness between

pericytes and endothelial cells in the retina is thinner, presumably allowing for more

intimate cellular interactions (Frank et al, 1990). In human retinal capillaries,

pericytes occur at a ratio of 1:1 with endothelial cells.

Platelet-derived growth factor (PDGF-B) derived from endothelial cells

plays a critical role in pericyte recruitment to newly formed vascular beds during

development (reviewed Betsholtz, 1995). PDGF-B/PDGF receptor-deficient mice

lack microvascular pericytes in brain vessels, forming capillary microaneurysms

during late gestation (Lindahl et al, 1997) (For more information about PDGF-

deficient mice, see Chapter 1.3.3 Section bi).


16

Pericytes appear to have important contractile functions that may be

controlled by endothelial cell signalling in vivo. Endothelial cell-pericyte ‘peg and

socket’ type junctions and cellular gap junctions enable direct communication

between cells (Forbes et al, 1977). Sakagami et al, (2001) reported that PDGF

activated different ion channels in retinal pericytes according to the metabolic

conditions, having an (alternatively) vasodilator or vasoconstrictor effect. Inhibition

of contractile function in cultured pericytes under high glucose conditions has been

reported (Gillies and Su, 1993). Capillary dilation and loss of endothelial cell

autoregulatory functions are considered to be a consequence of pericyte dropout in

the clinical setting of diabetic retinopathy. The inability of retinal pericytes to

reproliferate as quickly as brain pericytes in vitro might explain the selective loss of

pericytes in diabetic retinopathy (Wong et al, 1992).


17

1.1.6 Retinal Pigmented Epithelium

The RPE is the outermost layer of the retina and has important functions relating to

nutrition and support of the outer retina (reviewed Thumann and Hinton, 2001).

Retinal pigmented epithelium is bounded by Bruch’s membrane basally and the

interphotoreceptor matrix (IPM) apically, forming a selectively permeable barrier

(known as the outer BRB) between the choroidal plexus and the neurosensory retina

(Thumann and Hinton, 2001). Mature RPE cells are polygonal in shape with tight

junctions, apical microvilli and basal membrane infoldings. Melanin granules within

RPE absorb stray light and give the cell layer its pigmented appearance. The highest

concentration of pigment occurs in the peripheral areas and the fovea (Schmidt and

Peisch, 1986).

The proximity of RPE to photoreceptors in the outer layer of the retina

reflects the dependence of photoreceptors on the RPE. Retinal pigmented epithelium

apical membranes surround portions of the outer segments of rod and cone

photoreceptor cells, and shed disks of photoreceptor outer segments are taken into

the RPE cell cytoplasm and degraded by phagolysosomes (Marshall, 1987).

While there is a richly vascular choriocapillaris at the basal side of RPE, the

photoreceptor layer at the apical side of RPE remains avascular. Furthermore,

choriocapillaris vessels facing the RPE are fenestrated (Figure 1.2), suggesting that

an important trophic relationship exists between RPE and choriocapillaris. VEGF is

preferentially secreted to the basal (choroidal) aspect of RPE, and is believed to play

a dual role in survival and maintenance of the fenestrated choriocapillaris

(Blaauwgeers et al, 1999).

Other important functions of RPE include recycling of rhodopsin (see

Chapter 1.1.4 Section ci) and antioxidant activity. Excess fluid is absorbed by RPE

from the sub-retinal space preventing retinal detachment due to extracellular fluid

accumulation in retinal tissues (Marmor et al, 1980; Miller et al, 1982). Retinal

pigmented epithelium plays a role in visual processing by providing storage and

transport functions for metabolites and vitamins (Thumann and Hinton, 2001). In the

normal eye, RPE has a low regenerative capacity; cell loss is accommodated by

hyperplasia of the adjacent cells. CRALBP and cytokeratins are specific markers of

RPE (Crabb et al, 1988; McKechnie et al, 1988).


18

1.1.7 Choroid and Choriocapillaris

The choroid is a loose, thin, vascularised, connective tissue layer situated between

the sclera and the neural retina. The principal function of the choroid is to nourish the

outer layers of the retina as well as to act as a conduit for vessels travelling to other

parts of the eye (reviewed Guyer et al, 1989). Absorption of light by choroidal

melanocytes aids vision by preventing unwanted light from reflecting back

throughout the retina. The choriocapillaris is the capillary layer of the choroid.

Capillaries have a large diameter lumen and thin walls, allowing 2-3 blood cells to

pass through at a time (Guyer et al, 2001). Choriocapillaris endothelium is polarised

with a thin, fenestrated inner portion facing the RPE (Figure 1.2) and a thick, non-

fenestrated outer portion facing the deeper layers of the choroid (Mancini et al,

1986). While the retinal circulation is characterised by a low flow rate (25mm/s) and

high oxygen exchange, choroidal circulation is characterised by a high flow rate

(150mm/s) and low oxygen exchange (Alm and Bill, 1973). The choroid has an

extensive nervous system incorporating both parasympathetic and sympathetic

nerves (Wolter, 1960). The sympathetic nervous system autoregulates blood flow in

the choroid (Alm, 1977).


19

1.2 MACULAR OEDEMA

Macular oedema is one of the earliest features of a continuum of cellular responses to

a wide variety of ocular conditions - such as uveitis, retinitis pigmentosa, various

vascular disorders, epiretinal membranes and the vitreoretinal traction syndrome - as

well as being a common postoperative complication of ocular surgery (reviewed

Dick II et al, 2001). The causes are varied and the mechanism(s) by which macular

oedema occurs are not clearly understood (reviewed Wilkinson-Berka et al, 2001;

also Eagle, 1984). Early reports on macular oedema were primarily concerned with

the aetiology of diabetic retinopathy (reviewed Wolfensberger, 1999). Ballantyne

and Loewenstein (1943) were the first to describe the capillary wall alterations, as

well as the presence of deep waxy exudates in the OPL, that are characteristic of the

condition.

Macular oedema that is associated with diabetes mellitus is one of the first

indicators of retinal microvascular dysfunction (also known as background, or non-

proliferative diabetic retinopathy - NPDR) and may be followed by capillary

occlusion and neovascularisation in the chronic disease phase (or proliferative

diabetic retinopathy - PDR) (reviewed Gardner et al, 2002). Recognition of oedema

may be difficult since clinical leakage does not always correspond with tissue

swelling or functional visual loss (reviewed Marmor, 1999). Extravascular

erythrocytes may be detected ophthalmoscopically (Figure 1.3) and lipids deposited

adjacent to the areas of oedema aggregate as “hard exudates” within the outer

plexiform layer (Figure 1.4), however these features are associated with advanced

disease.

Objective clinical measurements of retinal thickness can now be made with

optical coherence tomography (OCT) (reviewed Frank, 2004) or retinal thickness

analysis (Dick II et al, 2001). A recent study reporting an early morning reduction in

visual acuity was related to overnight retinal thickening in 12 patients with diabetic

macular oedema (Larsen et al, 2005). The observed evening-to-morning variation in

retinal thickness (within retinas of diabetic patients) was thought to be due to a

deficiency in the removal of extracellular fluid in the event of nocturnal variations in

arterial blood pressure (Larsen et al, 2005).


20

However, while oedema may cause cellular damage, tissue swelling does not

automatically translate into neuronal dysfunction. The possibility that visual acuity

changes may be secondary to other factors including ischemia (which is an

underlying cause of oedema) should also be considered (Marmor, 1999).

1.2.1 Retinal Ischemia

As previously mentioned in Chapter 1.1.5 Section a, blood supply of the retinal

vasculature is autoregulated. Autoregulation occurs by means of interactions between

endothelium and perivascular elements for the purpose of adjusting capillary

perfusion to meet local metabolic demand (reviewed Harris et al, 2001). Inner retinal

oxygen levels are maintained within narrow confines irrespective of increased

oxygen availability, reflecting an ability to adapt to subtle variations in oxygen

supply and demand (Yu and Cringle, 2001). High oxygen demands together with the

relatively sparse nature of the inner retinal blood supply are thought to contribute to

the particular vulnerability of the retina to metabolic disturbances affecting the

vasculature (Yu and Cringle, 2001). Retinal ischemia is a condition in which the

blood supply does not meet the metabolic demands of the retina. Hypoxic/ischaemic

(hypoxia plus the absence of glucose) conditions may be crucial factors in the

pathogenesis of cystoid macular oedema (CMO) (reviewed Aiello, 2002; also Tso,

1982) resulting in neuronal death and gliotic changes in Müller cells (Bringmann et

al, 2004).

1.2.2 Intracellular or Extracellular Oedema?

Neuronal cells in the retina require the same protection from excess extracellular

fluid as brain tissue, therefore interstitial spaces within the retina are relatively

dehydrated (Marmor, 1999). In the normal eye, there are passive and active forces at

work to move water across the retina and out of the subretinal space (Negi and

Marmor, 1986). Macular oedema is characterised by expansion of the extracellular

and/or intracellular spaces in the macular area of the retina (Marmor, 1999).

Extracellular swelling can occur anywhere between the internal and external

limiting membranes, both of which prevent the clearance of proteins and other

osmotically-active molecules that bind water to cause oedema. By comparison, when


21

fluid accumulation is localised to cystic spaces of the OPL and INL of the parafoveal

retina, the oedema is believed to be intracellular, and is known as CMO (Figures 1.5,

1.6, 1.8).

Ischemia has been proposed to be the trigger of many retinal conditions,

including diabetic retinopathy (Arden et al, 2005). Using a well-established animal

model of ischemia, the controversy about whether macular oedema is localised

intracellularly or extracellularly has been clarified (Pannicke et al, 2004). In these

studies, transient elevation of intraocular pressure precipitates reproducible cellular

responses in the vascularised rat retina. Müller cell expression of bidirectional

potassium (K+ ) (KIR 4.1) channels were significantly downregulated after transient

ischemia, while the expression of AQP-4 water channels remain unaltered. As a

consequence, Müller cells accumulate K+ ions intracellularly - providing an osmotic

gradient that drives water into Müller cells through AQP-4 channels (Figure 1.7).

Disturbances in K+ efflux out of Müller cells and into blood vessels and the vitreous

provides a plausible explanation for the development of oedema in human patients

suffering from retinal hypoxia (reviewed Bringmann et al., 2004). Deletion of the

AQP-4 gene in mice was neuroprotective under the same experimental conditions of

transient ischemia (Da and Verkman, 2004), suggesting that reduced Müller cell

swelling associated with AQP-4 deficiency is likely to be responsible for retinal

protection in cytotoxic oedema.

Swelling of Müller cells (intracellular oedema) in the macula has been

suggested to precede extracellular oedema, which is usually associated with BRB

breakdown (reviewed Yanoff et al, 1984). A better way to discriminate between the

types of macular oedema may be to refer to primary transcellular fluid exudation via

disturbed K+ and/or water channel expression and without defects in the BRB

(intracellular oedema), and primary paracellular fluid exudation via defects in tight

junctions of the BRB (extracellular oedema) (Bringmann et al, 2004).

1.2.3 Diabetes and Diabetic Retinopathy

The life expectancy of adults with diabetes is decreased on average by 5-10 years

(reviewed Donnelly et al, 2000). Type 2 diabetes (characterised by insulin resistance

and/or abnormal insulin secretion) accounts for over 90% of cases globally (reviewed


22

Zimmet et al, 2001) and is most pronounced in native populations such as Australian

Aborigines who have abandoned a traditional hunter-gatherer lifestyle (O'Dea,

1991). Type 2 diabetes is predominately associated with a sedentary lifestyle and

obesity (Zimmet et al, 2001). Type 1 diabetes which often presents precipitately with

ketoacidosis, generally occurs in younger people and is due to loss of insulin

production by pancreatic islet cells (reviewed Kim, 2004).

A diagnosis of diabetes increases the risk of developing both macrovascular

and microvascular complications by 25 times (Aiello, 2002). Diabetic retinopathy is

the most common cause of blindness in working-aged people, arising from a

combination of microvascular leakage and occlusion (Donnelly et al, 2000). Within

Australia, the prevalence of retinopathy in the diabetic population was shown to

increase in accordance with the duration of diabetes: 0-4 year duration, 9.2%; 5-9

years, 23.1%; 10-19 years, 33.3%; ≥20 years, 57.1% (Australian Diabetes, Obesity

and Lifestyle Study) (Tapp et al, 2003). Although there is a lower overall incidence

of diabetic retinopathy within the diabetic Aboriginal population (compared to the

general Australian diabetic population), this is probably due to the limited lifespans

of indigenous people which does not allow them to survive long enough to develop

complications. Indigenous Australians are reported to have the highest incidence of

CSMO (clinically significant macular oedema) ever described (Katherine Region

Diabetic Retinopathy Study) (Jaross et al, 2005).

Cost/burden-of-illness studies have estimated that primary prevention of type

2 diabetes (particularly in high-risk groups) appears to be the most cost-effective

approach to dealing with the growing epidemic due to the debilitating nature of

diabetes-related complications (Raikou and McGuire, 2003). The gradual loss of

neurons in diabetic retinopathy (see Chapter 1.2.4) suggests that progress of the

disease is ultimately irreversible, since these cells cannot be replaced (Barber,

2003a). Due to the progressive nature of the disease, intensive treatment is favoured

over more conservative strategies (Meyer-Schwickerath and Fried, 1981) (For more

information about the treatments used in diabetic retinopathy, see Chapter 1.2.5).


23

1.2.4 Effect of Diabetes at a Cellular Level

Endothelial cells are the main regulators of homeostasis in the vasculature and are

extremely sensitive to changes in blood composition and flow (reviewed Michiels,

2003). Being so intimately associated with blood, vascular endothelium is the first

point of contact with the metabolic byproducts of hyperglycemia in diabetes mellitus.

The earliest detectable evidence of vascular dysfunction induced by diabetes is

altered blood flow in the retina, kidney and peripheral nerves (reviewed Ido et al,

1997). Retinal blood flow abnormalities in diabetic animals and patients precedes

clinical diabetic retinopathy (Bursell et al, 1992; Higashi et al, 1998; Bursell et al,

1996; Patel et al, 1992) and may set in train cellular consequences that progress even

after glycaemic control has been reinstated. For example, changes in basement

membrane thickness in diabetes often follows from blood flow abnormalities

(McMillan, 1997) and this must clearly interfere with cell-cell signalling and

metabolic exchanges generally, and specifically with respect to nutrition of the inner

retina (reviewed Ashton, 1974).

Reduction in endothelial cell-pericyte coupling was demonstrated in an

experimental model of diabetes within six days after the onset of hyperglycemia in

streptozotocin (STZ)-induced diabetic rats (Oku et al, 2001). Dysfunctional

communication between pericytes and endothelial cells may metabolically isolate

pericytes and contribute to their demise early in the course of diabetic retinopathy

(Oku et al, 2001). Loss of an inhibitory influence on vascular endothelium in the

form of pericyte dropout, together with an increase of local stimulators as a result of

hypoxic damage to neural tissue, may facilitate conditions that lead to

vasoproliferation (reviewed Dodge and D’Amore, 1992) (For more information

about the effect of diabetes on retinal pericytes, refer to Chapter 1.3.3 Section bi).

Alteration of the rate of retinal blood flow (Bursell et al, 1992) and

breakdown of (primarily) the inner BRB (Do carmo et al, 1998) occurs by the

beginning of the second week in the same diabetic rat model. Autoregulatory

mechanisms become increasingly deranged with disease progression (reviewed

Kohner et al, 1995). Studies involving diabetic humans and dogs have shown that

metabolic changes that occur during the initial period of chronic hyperglycemia, set


24

the stage for subsequent anatomic abnormalities, even though these lesions may not

be detectable clinically for several months or years (Frank, 2004).

Numerous biochemical factors are thought to play a role in the development

of diabetic retinopathy. The renin-angiotensin system (RAS) modulates blood

volume via specific receptors on endothelial cells that effect changes in

vasoconstriction or vasodilation (reviewed Fletcher et al, 2005). The human retina

contains RAS components within neurons, glia and retinal blood vessels, and the

RAS has been found to be dysregulated in diabetic retinopathy (Danser et al, 1989).

Understanding how early vascular changes alter glial and neuronal functions -

which in turn may further exacerbate vascular pathology - could assist in clarification

of the primary events in diabetic retinopathology. Additional biochemical pathways

likely to be of importance in the progression and development of diabetic retinopathy

are included in Table 1.1; although no single mechanism has yet led to an effective

therapy (Frank, 2004).

1.2.5 Prophylaxis and Retinal Laser Treatment

Photocoagulation, corticosteroids, and non-steroidal anti-inflammatory drugs have

been used in the treatment of macular oedema (reviewed Jonas, 2005; also Marmor,

1999). Laser treatment has proven effective in slowing the progression of visual loss

from diabetic macular oedema (Early Treatment Diabetic Retinopathy Study Report

number 9) although the mechanisms behind the resolution of oedema are unknown

(ETDRS, 1991). Recently it was suggested that laser therapy may work by

destroying photoreceptors (especially rods which place a high demand on the

available oxygen supply in the dark adapted retina) and increasing retinal pO2 (Arden

et al, 2005). Visual acuity is generally not improved by laser treatment and vision

may be damaged when laser is applied too often, since scars from the laser burns

tend to enlarge over many years (For more detail about retinal laser, refer to Chapter

5).

Of more interest to provide an insight into the mechanisms of macular

oedema are studies using pharmacological compounds. The rationale behind recent

approaches is to target the early features of diabetic retinopathy before the cellular

changes become irreversible. Macular oedema appears to result from an underlying


25

defect in retinal Müller cells that maintain the BRB, comprising both an ischemic

(Pannicke et al, 2004) and an inflammatory component (reviewed Adamis, 2002).

Triamcinolone acetate (TA) has been found to inhibit intracellular Müller cell

swelling in a rat model of concurrent ischemia and inflammation, thereby resolving

oedema by restoration of Müller cell homeostasis (Bringmann et al, 2005). In

another study using a rabbit model of transient ischemia, TA had no effect on Müller

cell gliosis, but decreased the number of microglial and immune cells (Uckermann et

al, 2005). TA has also been reported to reduce the expression of adhesion molecules

and permeability of human choroidal endothelial cells treated with inflammatory

agents in vitro (Penfold et al, 2002). In a human trial investigating the use of

intravitreal triamcinolone (IVTA) for persistent macular oedema after laser

treatment, visual acuity was improved and retinal thickness was decreased after 3

months, albeit with the development of some adverse effects reported in a 2 year

follow up study (Sutter et al, 2004; Gillies, 2005 in press).

The use of IVTA raises the possibility that retinal diseases could be treated

locally however, more long-term studies are necessary to monitor adverse outcomes

(Jonas et al, 2005). As more becomes known about early cellular changes in diabetic

retinopathy, treatment modalities may consider different aspects of the disease

pathology such as interference of the biological systems that cause BRB breakdown,

blockage of gliotic changes, or enhancement of absorptive forces (Marmor, 1999).

1.2.6 Non-Vascular Cells in the Pathogenesis of Diabetic Retinopathy

Until recently, diabetic retinopathy was thought of in terms of the vascular features,

which become clinically obvious after 10-15 years of diabetes in humans (Lieth et al,

1998). As new technologies become available in the diagnostic and research fields, a

better understanding of the aetiology of the disease has emerged. In addition to the

microvascular component, diabetic retinopathy may also comprise both

neurodegenerative (Abu-El-Asrar et al, 2004; Lieth et al, 1998; Barber et al, 1998;

Hammes et al, 1995) and inflammatory features (Mamputu et al, 2004; Adamis,

2002; Joussen et al, 2002).

It is increasingly recognised that ongoing early changes in neuronal cells and

Müller cells may initiate pathological changes prior to the development of clinically


26

detectable microvascular damage (reviewed Lieth et al, 2000; also Barber et al,

1998). Once vascular changes become clinically evident, the disease processes may

have gone beyond containment by the currently available therapies. The aim of the

studies described in this thesis was to investigate how perivascular cells may

contribute to early changes in BRB integrity.

1.2.7 Evidence for Early Retinal Müller Cell Pathology and the Relationship to

Neuronal and Vascular Changes

Müller cells have a close relationship with both neuronal and vascular elements of

the retina and both these cell populations are likely to be adversely affected by

pathological changes in Müller cells. In the early stages of diabetic retinopathy,

metabolic and cellular changes ongoing in the retina may precede the onset of

clinically detectable retinal vascular lesions. Here, the case will be made that tissue

ischemia and the effect of ischemia on Müller cells is the insult that underlies the

initiation and development of macular oedema.

In the adult macula there is a critical balance between the vasculature and

high metabolic demand such that even minor perturbations of circulation, as may

occur in vascular disease, could lead to metabolic stress in foveal neurons and/or

glial cells (reviewed Penfold et al, 2001). Elevated blood glucose causes increased

retinal tissue glucose that may result in a glucose-induced redox imbalance

(hyperglycaemic pseudohypoxia) which may contribute to the ischemia that precedes

the development of diabetic retinopathy (reviewed Williamson et al, 1993).

Hyperglycaemic pseudohypoxia develops in poorly controlled diabetes due to the

increased cytosolic ratio of free NADH/NAD+ caused by hyperglycemia, and

referred to as pseudohypoxia because the partial pressure of oxygen in tissue is

normal. Observers have suggested that the NPDR environment is not hypoxic

because retinal capillaries are not yet occluded (Hammes et al, 1998) or because

VEGF expression is not substantially increased (Gerhardinger et al, 1998). However,

an important feature of the glucose-induced redox imbalance is that relatively mild

hypoxic or ischemic episodes that are insufficient to cause dysfunction in nondiabetic

subjects - when superimposed on pre-existing pseudohypoxia induced by

hyperglycemia - results in a higher cytosolic NADH/NAD+ that causes tissue


27

dysfunction and injury in diabetic subjects (Williamson et al, 1993). Williamson’s

hypothesis provides an explanation for the increased susceptibility of diabetic

subjects to hypoxic and ischemic injury.

At a cellular level, the supportive functions of Müller cells may be

compromised by the diabetic/ischemic environment. The effect of dysfunctional

Müller cells on neuronal cells and the vasculature may be simultaneous or

incremental in NPDR - perhaps beginning with neuronal cell apoptosis - followed by

activation of endothelial cells that form the inner BRB. An insight into progression

of the disease is illustrated by the full spectrum of Müller cell responses that are

likely to occur during the course of diabetic retinopathy - from initial upregulation of

intracellular stress-related proteins GFAP and vimentin - to the end-stage gliotic

changes that are seen in Müller cell fibrocontractive elements in PDR (Barber et al,

2000; Hammes et al, 1995; Guidry, 2005). These changes are likely to be mediated

by upregulated growth factors in retinal tissue (reviewed Chiarelli et al, 2000). An

informative picture about the early diabetes-induced changes that occur in the retina

is emerging from in vivo studies with rodents, and post mortem features in human

eyes from patients with NPDR.

a) Early Müller cell and neuronal changes in NPDR

The electroretinogram (ERG) output - measured as a massed retinal potential - can be

broken down into several components, each reflecting underlying neuronal and/or

glial origins (Granit, 1933) and has been used to study functional losses in animal

models of diabetes (Fletcher et al, 2005). In streptozotocin (STZ)-induced diabetic

rats, the most common finding is an inner retinal dysfunction with decreased

amplitude and delayed timing of the oscillatory potential (Bui et al, 2003; Aizu et al,

2002; Sakai et al, 1995). More recently, a hypothetical association has been

suggested between early degenerative processes in the neural retina, and a defect in

Müller cells. Electrophysiological and psychometric vision testing in humans soon

after the onset of diabetes provides further support for this idea. Loss in sensitivity of

phototransduction for both rods and cones has been reported in human diabetic

patients (Holopigian et al , 1997). Colour vision defects are significantly higher in the


28

eyes from diabetic patients with minimal retinopathy, compared to control eyes from

non-diabetic patients (Roy et al, 1986).

Animal studies have also shown specific metabolic defects in retinal cells.

Leith et al, (1998) demonstrated that glial reactivity is an early event that may

precede vascular abnormalities; GFAP levels in Müller cells and astrocytes were

upregulated by 5-fold after 3 months in STZ-induced diabetic rats. In addition,

glutamate levels were 1.6-fold higher in diabetic rat retinas, suggesting that

metabolism of retinal glutamate by Müller cells may be compromised in the diabetic

environment (Lieth et al, 1998). In a follow up study, Leith et al, (2000) showed that

GS activity was decreased after only 2 months of diabetes. Using the same rat model,

another report suggested that there was enhanced uptake of glutamate after 3 months

of diabetes (Ward et al, 2005). Ward et al, (2005) argued that increased uptake of

glutamate would reduce the likelihood of neuronal toxicity due to extracellular

glutamate, although it is possible that some degree of neuronal dysfunction might

arise because of abnormal glutamate transport.

Müller cells exposed to continuously raised glutamate levels, either due to

prolonged release and/or diminished uptake, become activated and proliferate in vitro

(Uchihori and Puro, 1993), suggesting that glutamate may have a mitogenic effect on

Müller cells under pathophysiological conditions in vivo.

b) Müller cell involvement in neuronal apoptosis in NPDR

Apoptosis of certain vulnerable cell populations in NPDR has gained increasing

popularity as an explanation for losses in vision sensitivity. Apoptosis is the result of

a genetically encoded intrinsic cell suicide program also known as programmed cell

death (PCD) (reviewed Kerr, 2002; reviewed Majno and Joris, 1995). In histological

sections, apoptosis is recognised by pyknotic nuclei, cytoplasmic condensation and

DNA fragmentation (Gavrieli et al, 1992; Iseki, 1986; Wijsman et al, 1993).

Apoptotic markers - including in situ DNA terminal dUTP nick end labelling

(TUNEL) - together with retinal thickness measurements provide strong evidence

that diabetic retinopathy has a neurodegenerative component (Barber et al, 1998;

Rosenbaum et al, 1998).


29

A 10-fold increase in apoptosis in neural retinal cells together with decreased

retinal thickness is seen after only 1 month of STZ-induced diabetes in rats (Barber

et al, 1998). The levels of apoptotic cell death (predominately in the RGC layer)

correspond to a disease duration of 6 years of diabetes in human post mortem retinas

(Barber et al, 1998). Recently, Abu El-Asrar et al, (2004) investigated the possible

mediators of apoptotic cell death in post mortem retinas from 5 patients with NPDR.

The apoptosis-promoting factors, caspase 3, Fas and Bax were upregulated by RGC

in diabetic retinas, and GFAP-reactive Müller cells overexpressed the anti-apoptotic

markers ERK 1 / 2 and Bcl-2, in line with their neuroprotective functions (Abu-El-

Asrar et al, 2004). It is anticipated that activated Müller cells may have a capacity for

both supportive and destructive roles in the apoptogenic diabetic retinal environment.

Fas ligand (FasL) localised to Müller glial cells may be involved in the induction of

apoptotic cell death in RGC (Abu-El-Asrar et al, 2004).

An important link between the neural and vascular retina was made when

neuroretinal apoptosis and retinal capillary damage were prevented in diabetic rats,

by treatment with nerve growth factor (NGF). Hammes et al, (1995) predicted that

the characteristic vascular lesions in diabetic retinopathy result from a loss of trophic

support by apoptotic cells in the neural retina. Apoptosis was associated with

upregulated expression of p75NGFR in RGC and Müller cells. Apoptotic RGC and

Müller cells, observed in the INL of early (15 week) STZ-induced diabetic rats,

corresponded to areas of upregulated GFAP and vimentin immunoreactivity

(Hammes et al, 1995). At this time, there were also significant increases in pericyte

loss and capillary occlusion in diabetic rat retinas.

It had not previously been appreciated that injury to the neuroretina could

influence retinal vessels, and it was unclear whether diabetes-induced degeneration

of RGC and Müller cells occurred in parallel with - but independent of - diabetic

vascular changes, or whether cell degeneration caused, or resulted from the vascular

changes. Barber et al., (1998) also demonstrated a reversal of neurodegenerative

changes in rat retinas by delivery of insulin in the first month of diabetes induction,

although there was no mention about resolution of vascular lesions in this study.

Here, it is proposed that insulin may act by reversing some aspect of the physiology


30

of diabetes, or as a possible trophic factor affecting cell survival in the CNS (Barber

et al, 1998).

As discussed earlier in Chapter 1.2.2, a recent approach to studying the

progression of cellular pathology after ischemic injury is by induction of transient

ischemia in the high intraocular pressure rat model (Pannicke et al, 2004). The rat is

preferable to other species which lack an inner retinal blood supply and associated

astrocytes (Uckermann et al, 2005). The ischemic injury in the rat model is

characterised by activation of retinal glial cells and neuronal cell degeneration,

primarily in the GCL and INL (Rosenbaum et al, 1998; Shibuki et al, 1998). Injuries

precipitated by ischemia persist even after removal of ischemic conditions.

Rosenbaum et al, (1998) reported a significant thinning in the inner retinal layers at 7

days after 60 minutes of ischemia in the rat eye. The number of TUNEL-positive

cells peaks at 24-48 hours and persists for 7 days (Rosenbaum et al, 1998),

supporting the hypothesis that delayed apoptotic cell death occurs, possibly mediated

by activated Müller cells (Abu-El-Asrar et al, 2004).

As previously mentioned, RGC appear to be especially vulnerable to

apoptotic cell death in NPDR (Barber et al, 1998; Abu-El-Asrar et al, 2004; Hammes

et al, 1995; Rosenbaum et al, 1998; Shibuki et al, 1998). A variety of pathogenic

mechanisms may predispose the inner retina to degeneration. Neufeld et al., (2002)

reported that blood-derived leucocytes enter retinal tissue shortly after transient

ischemia, surrounding neurons in the GCL and releasing cytotoxic free radicals.

Upregulated cytokines in NPDR that may mediate leucocyte entry into retinal tissue

are discussed further in Chapter 1.2.7 Section d, and for additional information about

the effects of leucocyte activity on the BRB, see Chapter 1.3.3 Section biv.

Activated microglial cells may do further damage to inner retinal neurons by

the release of nitric oxide (NO) (Neufeld, 1999). Neuronal pathology was associated

with activation of both Müller cells and microglia in early (1-4 months) STZ-induced

diabetic rats (Zeng et al, 2000). Neurons in the GCL and the inner part of the INL

were moderately reduced after 1 month of diabetes in rat retinas, corresponding with

upregulation of GFAP in Müller cells. After 4 months of diabetes, neurons were

significantly reduced; at this time, microglia had relocated throughout all of the


31

retinal layers and appeared to be involved in clearing of degenerated neuronal

elements.

c) Müller cell involvement in BRB breakdown in NPDR

The consequences of increasingly reactive Müller cells on neuronal cell populations

in the retina in NPDR have been alluded to above. At the pre-clinical stage, Müller

cells may play conflicting roles in retinal homeostasis, having both protective and

cytotoxic effects on retinal neurons. An important feature of NPDR that may be

related to Müller cell dysfunction and altered neuronal interactions, is inner BRB

breakdown.

The association between activated Müller cells and BRB breakdown in

NPDR corresponds with the distribution of GFAP and ZO-1 (tight junction) protein

(Barber et al , 2000). After 2 months of diabetes in STZ-induced diabetic rats, GFAP-

positive Müller cells were evident across the retina, and by 4 months, GFAP

upregulation was even more pronounced (Barber et al, 2000). Occludin

immunoreactivity was reduced in the outer plexiform capillary bed at 4 months, and

in large arterioles, occludin had redistributed from endothelial cell borders to the

cytoplasm (Barber et al, 2000) (For more detail about the role of occludin in BRB

breakdown, see Chapter 1.3 Section b).

d) Müller cell and neuronal cell expression of VEGF in NPDR

VEGF protein occurs in nonvascular cells within the eyes of diabetic patients without

retinopathy (Famiglietti et al, 2003) supporting the hypothesis that diabetic

retinopathy begins as a disease of retinal neurons and Müller cells, only later

involving the retinal vasculature (Frank, 2004). Because VEGF has been shown to

play a key role in retinal neovascularisation in PDR (Aiello et al, 1994a; Pierce et al,

1995), it was assumed that increasing levels of VEGF would be demonstrable in all

stages of diabetic retinopathy to different degrees, and may play a role in breakdown

of the BRB in NPDR (Amin et al, 1997; Gerhardinger et al, 1998). Interestingly,

although VEGF has been shown to be upregulated in NPDR - and in a variety of

other non-ischemic retinal disorders - neovascularisation often does not occur. This

may be due to the presence of inhibitors of neovascularisation, to a deficiency of


32

specific receptors, or the requirement for an appropriate metabolic environment

(Eichler et al, 2001; Brooks et al, 1998). Upregulated VEGF during NPDR may be a

non-specific response to the general metabolic changes occurring in the diabetic

environment (Hammes et al, 1998). More likely, VEGF upregulation during the early

changes in diabetic retinopathy is relative to its importance as a retinal survival factor

(Arden et al, 2005; Jin et al, 2000; Gora-Kupilas and Josko, 2005; Alon et al, 1995).

Eichler et al. (2001) considered what might be going on in NPDR by using

cultured Müller cells and retinal explants to show (as anticipated) that VEGF is

significantly upregulated under hypoxic conditions compared to normoxia, and the

major source of VEGF was Müller cells (Eichler et al, 2001). Medium conditioned

by hypoxic Müller cells had no more effect on retinal endothelial cell proliferation,

compared to medium from normoxic Müller cells; in some cases inhibition of

endothelial cell proliferation occurred (Eichler et al, 2001). Overall, anti-proliferative

factors appear to be more effective in preventing endothelial cell proliferation, even

when the balance of anti-proliferative factors and mitogenic factors is biased towards

proliferation (Eichler et al, 2001). In NPDR, although more VEGF is expressed, anti-

proliferative factors prevent the growth of new blood vessels. Neovascularisation

may occur when the relative amount of mitogenic factors and/or their (relative)

signalling efficacy is dramatically increased (Eichler et al, 2001).

Another study reported that basal levels of VEGF are always present in the

retina in healthy eyes in the absence of neovascularisation, supporting the

proposition that naturally-occurring inhibitors of angiogenesis must exist (Eichler et

al, 2004). The ratio between the levels of angiogenic stimulators (VEGF) and

inhibitors (TGF-β2, PEDF, TSP-1) was assessed in vitro using an immortalised

Müller cell line (MIO-M1). VEGF production increased by 6.7-fold in MIO-M1 cells

and in guinea pig Müller cells by 25-fold under hypoxic conditions (Eichler et al,

2004). When the ratios of VEGF/TGF- β2, VEGF/PEDF and VEGF/TSP-1 were

estimated under normoxic and hypoxic conditions, it was demonstrated that although

the release of angiogenic-relevant factors are regulated by hypoxia so that the net

effect should be permissive for angiogenesis, these factors inhibited retinal

endothelial cell proliferation (Eichler et al, 2004). These findings suggest that Müller


33

cells control endothelial cell activation and neovascularisation by the release of anti-

angiogenic cytokines (Eichler et al, 2004).

There is increasing evidence of Müller cell dysfunction in NPDR. It may be

the case that cellular products including (VEGF) are significantly upregulated in the

diabetic environment to an extent that irreversible functional changes occur.

Alternatively, cell-cell signalling may become dysfunctional in the diabetic

environment so that homeostatic or inhibitory mechanisms are overcome. Müller

cells have been provided with the machinery to support and maintain most of the

retinal cell populations by their ubiquitous presence in all retinal layers.

Given the above observations, when Müller cell derived anti-proliferative

factors are overwhelmed in PDR, VEGF (expressed by Müller cells) may play an

integral role in the initiation and progression of retinal neovascularisation, possibly

in synergy with other growth factors (Ruberte et al, 2004; Castellon et al, 2002).

Hypertrophy and proliferation of retinal Müller cells is a well-recognised late change

in PDR contributing to epiretinal membrane formation (Ohira and de Juan, 1990).


34

1.3 BLOOD-RETINAL BARRIER

a) Anatomy

The BRB exists at two principal sites: an outer barrier consisting of RPE cells and an

inner barrier comprised of retinal vascular endothelial cells, and is dependent upon

integrity of the RPE, the retinal vasculature and a glia limitans that restricts contact

between blood vessels and the neural retina (reviewed Penfold et al, 2005). The glia

limitans is comprised of contributions from at least five cell types: astrocytes,

microglia, the terminals of NOS and SP–immunoreactive amacrine cells in the inner

retina, and Müller cells in the deeper retina (Penfold et al, 2005).

Increases in BRB permeability may involve an increase in the movement of

solutes between cells (paracellular flux) or through cells (trancellular flux).

Paracellular flux is dependent upon intercellular structures, whereas transcellular flux

is a membrane-associated, receptor-mediated process. Intercellular structures that

mediate paracellular flux consist of a protein complex that is comprised of tight

junctions (known as zonula occludens), adherens junctions (Figure 1.9), desmosomes

and gap junctions (Miyoshi and Takai, 2005). The integrity of the cellular barrier

appears to be dependent upon regulation of these junctional complex-associated

proteins. At least 8 structural proteins are involved: zonula occludens-1, -2 and -3

(ZO-1, ZO2, ZO3), occludin (65kDa), cingulin, 7H6, symplekin and claudin (in tight

junctions) (Citi and Cordenonsi, 1998; Denker and Nigam, 1998; Mitic and

Anderson, 1998; Stevenson and Keon, 1998) and the catenins: p120 (Figure 1.9),

plakoglobin and b-catenin (in adherens junctions) (Miller and Moon, 1996; Peifer,

1995; Hulsken et al, 1994). As well, Russ et al, (1998) have identified cadherin-5

(also known as VE-cadherin) and beta-catenin in cultured human retinal endothelial

cells. The major integral proteins in tight junctions and adherens junctions are

claudins, and cadherins respectively.

A number of additional ‘linker’ proteins have been identified that are

associated with the cytoplasmic domains of these integral proteins and the

cytoskeleton. Unlike epithelial cells, endothelial cells do not possess classic

desmosomes at interendothelial junctions (Valiron et al, 1996). A simplified

desmosomal structure (also known as the complexus adherens) (Figure 1.9) is

comprised of desmoplakin which co-localises with VE-cadherin, plakoglobin and


35

vimentin in endothelial cells (Valiron et al, 1996). Platelet/endothelial cell adhesion

molecule (PECAM) (Figure 1.9) is also found concentrated at sites of endothelial

cell-cell contacts, but is not directly associated with adherens or tight junctions.

PECAM appears to participate in the signalling cascades of different growth factors

via Src homology 2 containing PTP-2 (SHP-2) (reviewed Dejana, 1996).

An increasing number of in vivo and in vitro examples of endothelial cell

heterogeneity within brain and retinal microvasculature suggest that only specific

branches of the microvascular tree harbour the BBB/BRB (Song and Pachter, 2003;

Barber et al, 2000; Rajah and Grammas, 2002). The possibility that the variability in

electrical resistances (TEER) generated between different retinal endothelial cell

isolations may be a consequence of the inclusion of heterogenous cell lines from

different branches of the vascular tree, is discussed in Chapter 3.

b) Functions of tight junctions

i) Paracellular permeability

Tight junctions are traditionally recognised as the moderators of paracellular

permeability (see Chapter 1.3 Section a) and cellular polarity within epithelium and

endothelium (reviewed Anderson and van Itallie, 1995). Endothelial junctions are

generally less restrictive than epithelial junctional complexes and behave more like

diffusive pathways (Figure 1.10) (reviewed Schneeberger and Lynch, 1992).

Leucocyte extravasation most probably occurs via the paracellular pathway, through

cellular tight junctions (reviewed Lum and Malik, 1996) (see Chapter 1.3.3 Section

biv).

In terms of functional effects, ZO-1 (225 kDa) is the best studied protein of

tight junctions. Tyrosine phosphorylation of ZO-1 is associated with increased

paracellular permeability in both epithelial and endothelial cells (Staddon et al,

1995). While there is some evidence to suggest that occludin may directly contribute

to paracellular permeability (Barber and Antonetti, 2003; McCarthy et al, 1996),

when occludin is transfected into an endothelial cell line, no alteration of paracellular

barrier function is seen, suggesting that the purpose of occludin is primarily to

anchor tight junctions rather than direct control of barrier function (Kuwabara et al,


36

2001). More recently claudins are also reported to be integral proteins of the

interendothelial cell tight junction (Tsukita and Furuse, 2000).

ii) Transcellular permeability

Endothelial vesicles may be important in the transport of albumin through the

endothelial cell (Siflinger-Birnboim et al, 1991). Albumin may be absorbed onto the

luminal vesicular surface (into caveolae) then pinched off and transported

intracellularly to the abluminal surface. Here, they again fuse with the cellular

membrane and extrude the albumin molecule (Ghitescu et al, 1986). Alternatively,

albumin may bind to specific albumin receptors. An albumin receptor glycoprotein

of 60 kDa (gp60) has been identified on the plasma membrane of endothelial cells

(Schnitzer et al, 1988).

1.3.1 Outer Blood-Retinal Barrier

The RPE forms the outer BRB lying between the retina and the choriocapillaris. The

barrier provided by the RPE enables selective transport of nutrients from the

choroidal circulation to the outer retina, and vice-versa. The barrier characteristics of

mature RPE appear to be innate and not induced by the presence of other cell types

(Steuer et al, 2005). However, during development, RPE cells may require continual

exposure to retinal-derived factors for expression of phenotypic features including

barrier characteristics (Ban et al, 2000).

Epithelial tight junctions regulate the paracellular flux of ions which

generates a transcellular electrical resistance (TER), generally measured in Ω (ohms)

per cm2 (Rajasekaran et al, 2003; Giebel et al, 2005; Stanzel et al, 2005; Abe et al,

2003; Zech et al, 1998) (Table 1.2). The TER measures one aspect of junctional

permeability. The inverse of the TER is proportional to the conductance of ions

down a voltage gradient (see Chapter 3 for further details). By contrast, tracer studies

measure passive diffusion. Studies have shown that different mechanisms regulate

TER and permeation by passive diffusion (Balda et al, 1996) (For a description about

the use of tracers in permeability assays, see Chapters 4 and 5).


37

Table 1.2 Examples of baseline TERs in RPE cells from different sources.

Study TERΩ(ohms).cm2

____________________________________________________________

Human RPE cells (Rajasekaran et al, 2003) 525

Human RPE cell line, ARPE-19* (Giebel et al, 2005) 181

Rabbit RPE cells (Stanzel et al, 2005) 118

ARPE-19 (Abe et al, 2003) 80

Rat RPE cells (Zech et al, 1998) 67

_____________________________________________________________

*ARPE-19 is a spontaneously immortalised RPE cell line from a human donor

1.3.2 Inner Blood-Retinal Barrier

The inner BRB is formed by endothelial cells rather than by epithelial cells (outer

BRB). Endothelial cells have less rigidly organised junctions than epithelial cells

(reviewed Dejana, 2004). Endothelial cell tight junctions physically occlude aqueous

channels between cells (reviewed Ge et al., 2005). As with the outer BRB, the barrier

function of endothelial cells of the inner BRB can be measured by assaying the

transendothelial electrical resistance (TEER) (Gillies et al, 1995; Gillies et al, 1997;

Gillies and Su, 1995; Giebel et al, 2005; Feng et al, 1999; Gu et al, 2003). Within the

retina, endothelial cell barrier characteristics are thought to be induced by factors

derived from perivascular cells, especially Müller cells and astrocytes (Tout et al,

1993; Janzer and Raff, 1987) (refer to Chapter 1.1.5 Section a).


38

Table 1.3 Examples of baseline TEERs in bovine REC.

Study TEER Ω(ohms).cm2

____________________________________________________________

Retinal endothelial cells, REC (Gillies et al, 1995) 186

REC (Gillies et al, 1997) 168

REC (Gillies and Su, 1995) 137

Retinal microvascular endothelial cells, RMVEC 76

(Giebel et al, 2005)

REC (Feng et al, 1999) 50

BRE cell line* (Gu et al, 2003) 23

_____________________________________________________________

*hTERT-BREC is a conditionally immortalised endothelial cell line from bovine retina

1.3.3 BRB BREAKDOWN

It is increasingly recognised that endothelium actively participates in both

physiological and pathophysiological processes (Michiels, 2003). Cell-cell adhesive

junctions are important not only as the sites of attachment between endothelial cells

and for the control of vascular permeability, but also for intercellular signalling in

order to regulate tissue homeostasis (reviewed Lampugnani and Dejana, 1997).

Transport systems within endothelial cells such as vesiculo-vacuolar organelles are

precisely regulated to maintain barrier integrity and to protect blood vessels from

inappropriate increases in permeability, inflammation or thrombotic reactions

(Dejana, 2004).

The functional state of interendothelial junctions can be changed by the state

of growth and activation of endothelial cells, for example inflammatory cytokines or

growth factors may change the expression or phosphorylation state of the junctional

proteins (Dejana, 1996). Reassembly of tight junctions after leucocyte diapedesis

occurs rapidly in a mouse model of autoimmune uveoretinitis (Xu et al, 2005).


39

The endothelial basement membrane may also play a significant role in

endothelial cell barrier integrity (Lum and Malik, 1996). The basement membrane is

a heterogeneous and complex mixture of high molecular weight molecules secreted

by both glial cells and endothelial cells (Prat et al, 2001). The major molecular

components of basement membrane are laminin, collagen type IV, proteoglycans and

fibronectin. Basement membrane constituents can be degraded by proteinases

including matrix-metalloproteinases (MMPs) (Yong et al, 1998).

a) Indicators of BRB breakdown: GLUT-1 and PAL-E

GLUT1 is a major membrane glucose transporter protein (55 kDa) of brain and

retinal endothelial cells that is expressed in functional blood-tissue barriers (Gerhart

et al, 1989; Pardridge et al, 1990). GLUT1 may be a useful in vivo marker of BRB

breakdown during different stages of diabetic retinopathy. GLUT1 labelling occurs

in a large number of ocular cells besides capillaries and RPE in non-diabetic control

eyes (Kumagai et al, 1994). GLUT1 immunoreactivity was similar in diabetic and

non-diabetic eyes, except that the neovascular endothelium of proliferative

retinopathy (PDR) did not label with GLUT1 (Kumagai et al, 1994). Furthermore,

GLUT1 expression was focally upregulated on microvessels in retinas from patients

with long-standing diabetes with no (or minimal) diabetic retinopathy (NPDR)

(Kumagai et al, 1996). These studies show that GLUT1 levels may be indicative of

BRB changes that are ongoing throughout the course of diabetic retinopathy. PDR

appears to be associated with a loss of GLUT1 reactivity and loss of barrier integrity

(Kumagai et al, 1994; Kumagai et al, 1996). NPDR by comparison is characterised

by focal upregulation of GLUT1, that may amplify the toxic effect of hyperglycemia

on retinal vasculature and profoundly effect glucose availability to the biochemical

pathways leading to the development of diabetic retinopathy (Kumagai et al, 1994;

Kumagai et al, 1996).

Pathologische Anatomie Leiden-Endothelium (PAL-E) is an endothelial

specific antigen associated with endothelial cell caveolae (Schlingemann et al, 1999).

PAL-E is absent in microvessels with an intact BRB and becomes upregulated in

retinal vessels of patients with diabetic retinopathy, correlating with microvascular

leakage of plasma proteins (Schlingemann et al, 1999). In this study using post


40

mortem human eyes, Schlingemann et al, (1999) verified that the microvascular

leakage which occurs in diabetic retinopathy involves actively dysfunctional

endothelium rather than passive endothelial cell damage.

b) BRB pathology in NPDR

i) Cell-cell interactions and the BRB

Pericytes appear to be selectively located where they offer the most good in

terms of maintaining vessel integrity. At the capillary level, pericytes are located

around endothelial cells to complement vascular function and may play an

important role in modulation of inflammatory events, such as leakage of plasma

proteins (reviewed Sims, 2000). Degeneration of blood vessels associated with

pericyte dropout causes ischemia leading to neovascularisation in diabetic

retinopathy. The mechanism of blood vessel degeneration in diabetic retinopathy is

not well understood.

Interactions between endothelial cells and pericytes appear to be vital for the

health of retinal capillaries (refer to Chapter 1.1.5 Section b). In this regard, the

trophic growth factors PDGF and VEGF play a critical survival-promoting role in

pericytes and endothelial cells, respectively. Hammes et al., (2002) demonstrated

negligible pericyte loss in the heterozygous diabetic PDGF-B+/- mouse is

characterised by focal endothelial degeneration, compared to the homozygous

PDGF-B-deficient mouse (PDGF-B-/-) which is characterised by major pericyte loss

with capillary dilatation and bleeding (Lindahl et al, 1997). Mice lacking PDGF-B

in endothelial cells show compromised pericyte recruitment and vessel loss is

similar to that of NPDR (Enge et al, 2002).

Furthermore, VEGFR activation in endothelial cells appears to be critical to

retinal vessel survival (Shih et al, 2003a; Shih et al, 2003b). In vitro studies show

that pericyte and endothelial cell co-cultures produce activated TGF(Antonelli-

Orlidge et al, 1989; Sato et al, 1990). Shih et al, (2003a; 2003b) observed that

vessels were protected from oxygen-induced degeneration by TGF-1 induction of

VEGFR-1 in endothelial cells. Patients with PDR have decreased vitreal TGF-1

compared with patients with NPDR, and this is associated with increased vessel loss


41

in PDR, corresponding with a role for TGF-1 as a blood vessel survival factor

(Spranger et al, 1999).

TGF-1 also has a role in the suppression of inflammation. Early diabetic

retinal vascular loss (associated with pericyte dropout) is followed by inflammatory

cell infiltration (Shull et al, 1992). Surviving TGF-1-/- mouse pups, soon die from

inflammatory infiltrates (Shull et al, 1992). Similar processes may occur in NPDR

(For additional information regarding the effects of leucocytes on the BRB, see

Chapter 1.3.3 Section biv).

Impaired vascular-glial cell interactions occur in the development of diabetic

retinopathy (Barber et al, 2000). Janzer and Raff (1987) demonstrated that glial cells

induce BBB properties in non-neural endothelial cells, and retinal Müller cells

contribute to the formation of a tight BRB in an analogous way to astrocytes at the

BBB (Tout et al, 1993) (refer to Chapter 1.1.4 Section bvi). Glial cells are resistant

to anoxia and maintain their energy reservoirs by anaerobic glycolysis (reviewed

Juurlink, 1997). In chronic diabetes, Müller cells may be unable to fulfil the normal

supportive functions required for neurons and the vasculature (see Chapter 4).

Depletion of glycogen reserves may contribute to the early derangement of Müller

cell function, leading to a down-regulation of energy-dependent mechanisms

(including glutathione production) required to regulate the extracellular environment

of the retina (Juurlink, 1997) (see also Chapter 1.1.4 Section civ).

ii) Growth factors, cytokines and the BRB

Growth factors mediate physiological effects in virtually every organ and tissue. The

metabolic milieu of the diabetic patient may trigger activation or repression of genes

that could lead to imbalances in growth factor expression, causing derangements of

cellular metabolism and proliferation in genetically susceptible individuals (Chiarelli

et al, 2000).

Local factors in the retina may play an important role in the progression of

diabetic retinopathy. VEGF upregulation and overexpression initiates a sequence of

events that is characterised by increased vascular permeability (NPDR) leading to

new vessel formation, or angiogensis (PDR) (reviewed Dvorak et al, 1995). Potential

triggers of the VEGF-induced cascade include local hypoxia (Shweiki et al, 1992;


42

Forsythe et al, 1996; Shima et al, 1996; Semenza, 1998) and/or other target cell

specific (angiogenic) growth factors and cytokines that (in themselves) are unable to

elicit a potent hyperpermeability effect (Dvorak et al, 1995). The increase in

permeability and angiogenesis that occurs in diabetic retinopathy may result from an

increase in permeabilising factors and/or a decrease in inhibitory factors (Eichler et

al, 2001; Eichler et al, 2004). Vasoactive factors such as VEGF may act directly on

endothelial cell tight junctions to decrease their protein content or to increase their

phosphorylation.

Qaum et al, (2001) considered the role of VEGF and the precise vessels that

might be involved in BRB breakdown in NPDR. Specific vessel phenotypes

responsible for diabetic BRB breakdown were identified by labelling retinas with

fluorescent microspheres that became trapped in ECM of the hyperpermeable

vessels. High affinity soluble VEGFR administered to STZ-induced diabetic rats

inhibited VEGF bioactivity. BRB breakdown temporarily coincided with increased

retinal VEGF levels in rats after 1 week of diabetes. Soluble VEGFR restored

diabetic BRB breakdown to non-diabetic levels. Localisation studies identified

retinal capillaries and venules of the superficial inner retinal vasculature as the

primary sites of early BRB breakdown.

Insulin-like growth factor (IGF) has been shown to stimulate Müller cell

tractional force generation, thereby exaggerating the activation signals that

precipitate stress-induced responses in Müller cells in NPDR (Guidry et al, 2003).

The biological activity of IGF in normal vitreous is low or undetectable, and yet

these low levels are well above the threshold of Müller cell sensitivity (Guidry,

1997). Under normal conditions, IGF is controlled or attenuated (Hardwick et al,

1997; Guidry et al, 2004). In the diabetic retina IGF is significantly upregulated,

possibly related to plasma leakage as a consequence of BRB breakdown.

Recently, there is growing interest into understanding the mechanisms of

natural controlling factors and disease-related changes at the cellular level that may

contribute to loss of growth factor control (Meyer-Schwickerath et al, 1993; Burgos

et al, 2000). To this end, King and Guidry (2004) investigated IGF binding protein

(IGFBP) production by Müller cells. At least six IGFBPs are known with the

capacity to inhibit and/or potentiate growth factor activities (reviewed Hwa et al,


43

1999 and Arnold et al, 1993). Accumulation of different levels of growth factors in

the vitreous has an effect on Müller cell responses (Guidry et al , 2004) therefore it

was necessary to look at local IGFBP production and to see whether IGFBP

production is changed as Müller cell physiology alters between the normal,

proliferative and myofibroblastic states (King and Guidry, 2004). Müller cells

produced 5-6 IGFBPs and there was a significant increase in IGFBP production in

myofibroblastic Müller cells (King and Guidry, 2004). This was interpreted as a

response by Müller cells to attenuate the actions of increasing levels of free growth

factor in the environment (King and Guidry, 2004).

Transgenic mice expressing IGF-1 in the retina mimic most features of

human diabetic eye disease suggesting that IGF-1 has an important role in the

development of ocular complications (Ruberte et al, 2004). NPDR features were

observed in 2 month old mice including loss of pericytes, acellular capillaries,

thickened BM (Ruberte et al, 2004). Transgenic mice aged 6 months and older

showed altered retinal neovascularisation and most features found in human diabetic

retinopathy. In addition, retinas from transgenic mice overexpressed GFAP at 3 and

15 months suggesting that upregulation of IGF-1 may be associated with Müller cell

activation (Ruberte et al, 2004). Furthermore, hypoxia-stimulated retinal

neovascularisation was inhibited in mice that expressed a growth hormone antagonist

leading to low circulating IGF-1. Vascular alterations were associated with increased

VEGF in the early and late stage of diabetic retinopathy in transgenic mice (Ruberte

et al, 2004). It is highly likely that IGF-1 and VEGF act co-operatively to precipitate

neovascularisation in the advanced stage of diabetic retinopathy.

Many other growth factors with angiogenic potential have been identified as

playing a role in PDR, including acidic and basic fibroblast growth factors (aFGF

and bFGF) (Sivalingam et al, 1990; Fredj-Reygrobellet et al, 1991; Morishita et al,

1997), PDGF, tumor necrosis factor alpha (TNFa) (Armstrong et al, 1998) and

hepatocyte growth factor (HGF) (Morishita et al, 1997; Katsura et al, 1998;

Nishimura et al, 1998).

Increased growth factor levels in the vitreous in late diabetic retinopathy may

be the result of excess growth factor secretion by cells in the ischemic retina, a


44

contribution from systemic sources, or due to decreased degradation of secreted

growth factors (Chiarelli et al, 2000).

Recently, Castellon et al, (2002) explored the individual and collective

actions of IGF-1, VEGF PDGF-BB, FGF-2 and PlGF on cell migration and

survival/proliferation using primary cultures of bovine retinal endothelial cells.

Growth factors enhanced the angiogenic characteristics of cultured cells, specifically

with respect to tube formation, proliferation, secondary sprouting and migration

(Castellon et al, 2002). While some growth factors exerted minimal or no action

individually, the effects could be greatly augmented in combinations with other

factors, suggesting that diabetic retinal neovascularisation may result from the

additive or synergistic action of several growth factors (Castellon et al, 2002).

iii) Decreased adhesion molecule expression and the BRB

Expression of the tight junction proteins occludin and zonula occludens-1 (ZO-1) and

the adherens junction protein cadherin-5 (VE-cadherin) were investigated in retinal

vessels from a 73 year old diabetic patient with progressive NPDR in one eye and a

72 year old normal control patient (Davidson et al, 2000). Positive labelling of retinal

vessels for occludin and ZO-1 occurred in both the normal and the diabetic eye.

Large retinal vessels and capillaries in the normal and diabetic eye exhibited similar

tight junction protein labelling patterns, however cadherin-5 labelling was

undetectable in most of the retinal capillaries of the diabetic eye (Davidson et al,

2000). In large vessels, reduced cadherin-5 labelling occurred in the diabetic eye,

compared with the non-diabetic eye. This pattern was observed in all regions of the

diabetic retina. Decreased expression of cadherin-5 in the diabetic eye is consistent

with the hypothesis that junctional proteins are altered in diabetic retinopathy

(Davidson et al, 2000). Junctional proteins directly involved in the BRB might be

useful indicators of disease progression.

iv) Leucocyte activity and the effect on the BRB

Properties of endothelial cells that contribute to the BRB can also be regulated by

signals derived from the immune system and from Müller cells, astrocytes and


45

microglia that contribute to barrier integrity (Prat et al, 2001). Under basal conditions

the input from glial cells favours restriction of entry to immune cells.

Müller cells that become activated in response to altered neuronal signals in

the diabetic environment may produce an array of cytokines that increases

endothelial cell barrier permeability and promote leucocyte recruitment (Prat et al,

2001; Mamputu et al, 2004). In turn, activated endothelial cells express adhesion

molecules that lead to leucocyte-endothelial cell interactions (Sugama et al, 1992).

Leucocytes interact with endothelium in a sequential process involving primary

contact (mediated by P-selectin molecules) and leading to firm adhesion with

intercellular adhesion molecule-1 (ICAM-1, CD54) on endothelial cells (Musashi et

al, 2005) and the leucocyte receptor, CD18. In this adhesion cascade, leucocytes

become activated, injure tissue, and may be responsible for increased microvascular

permeability under inflammatory conditions (Matsuo et al, 1995).

Adhesion of leucocytes to endothelial cells enhances the ability of leucocytes

to generate reactive oxygen species (ROS) (reviewed Miyamoto and Ogura, 1999)

and has been shown to induce disorganisation of adherens junctions (Del Maschio et

al, 1996) and tight junctions (Bolton et al, 1998; Xu et al, 2005) between endothelial

cells, thereby increasing vascular permeability. Others have shown that leucocytes

induce endothelial cell death (Joussen et al, 2001). BRB breakdown is associated

with increased expression of ICAM-1 and CD18 (Miyamoto and Ogura, 1999;

Barouch et al, 2000). Inhibition of ICAM-1 and CD18 inhibits leucocyte adhesion

and endothelial cell death in diabetic rats (Joussen et al, 2001). It remains to be

established whether leucocyte adhesion is causal or secondary to BRB breakdown.

Retinal leucostasis is an important event in the development of vascular

leakage and capillary non-perfusion in NPDR (Miyamoto et al, 1999). Within 3 days

of diabetes induction in a STZ-diabetic rat model, retinal leucostasis had increased

by 1.9-fold (Miyamoto et al, 1999) with a 3.2-fold increase after 1 week, and this

level remained unchanged for 3 weeks. BRB breakdown had increased by 2.9-fold

after 1 week and by 10.7-fold after 4 weeks of diabetes. Leucostasis and vascular

leakage were prevented by pre-treatment with an anti-ICAM mAb (Miyamoto et al,

1999).


46

In a mouse model of autoimmune uveoretinitis, leucocyte recruitment to the

retina occurred through venule endothelium (Xu et al, 2005). In these locations,

endothelium-astrocyte contacts were lost, corresponding to decreased tight junction

protein expression. Redistribution of astrocytes and disensheathment of retinal

venules affected tight junction protein integrity in retinal venules however, the cause

of astrocyte redistribution is unknown (Xu et al, 2005).

Mamputu and Renier (2004) hypothesised that VEGF is a key regulator of

leucostasis in the diabetic retina. VEGF exerted a stimulatory effect on human

monocyte adhesion to cultured bovine retinal endothelial cells mediated by

upregulation of ICAM-1 expression; the effect was entirely abrogated by

immunoneutralisation of ICAM-1 (Mamputu et al, 2004). This study demonstrated

that VEGF and ICAM-1 pathways are mechanistically linked in NPDR.

The consequences of increased leucocyte adhesion on BRB breakdown are

well established. Recently, the chronic effects of leucocyte adhesion in diabetic

retinopathy have been studied in mice gene-deficient for ICAM-1 and CD18

(Joussen et al, 2004). Long-term wild type diabetic animals in this study manifested

BRB breakdown, and this was almost totally suppressed in ICAM -/- and CD18 -/-

animals (Joussen et al, 2004). Pericyte loss was also prevented in these animals. The

reduction in acellular capillaries observed in diabetic ICAM -/- and CD18 -/- mice was

directly associated with the protective effect of inflammation suppression on cell

death in the vasculature (Joussen et al, 2004).

Further support for the hypothesis of leucocyte-mediated destruction of the

BRB was shown by Sander et al, (2001) who demonstrated that the major change in

transport in the advanced stage of diabetic retinopathy is due to increased passive

leak through a damaged barrier. In this study, fluorescein was used as a marker of

barrier breakdown in patients with clinically significant macular oedema (Sander et

al, 2001). The active transport mechanism at the outer BRB (mediated by RPE)

remained intact and was probably responsible for the increased absorptive activity

that was seen in response to disruption of the inner BRB (Sander et al, 2001).


47

v) Matrix metalloproteinases and the BRB

Matrix metalloproteinases (MMPs) are a large family of zinc-dependent membrane-

degrading endopeptidases involved in a diversity of homeostatic and pathological

processes (reviewed Cunningham et al, 2005). MMPs participate in the degradation

of ECM components, cell adhesion, proteolytic release of ECM-sequestered

molecules and shedding of cell surface proteins that translate signals from the

extracellular environment (Cunningham et al, 2005). MMPs are secreted as

proenzymes that are activated by other proteases such as the serine proteinase,

urokinase plasminogen activator (uPA). Urokinase binds to the urokinase

plasminogen activator receptor (uPAR) on the cell surface and converts plasminogen

to another proteinase, plasmin, which degrades matrix components and may be

involved in the activation of latent MMPs (Giebel et al, 2005).

MMP substrates include all ECM components such as collagens,

proteoglycans, fibronectin, laminin and gelatin. The gelatinases MMP-2 (72 kDa),

MMP-9 (92 kDa) and uPA have been shown to be upregulated in epiretinal

neovascular membranes of patients with PDR (Das et al, 1999). Behzadian et al,

(2003) showed that uPA is involved in regulation of the paracellular pathway in

vitro; although this study could not define whether elevated levels of uPA had a

direct effect on permeability, or whether permeability increases in VEGF-treated

endothelial cells were mediated by MMPs.

Upregulation of MMP-2, MMP-9 and MMP-14 is associated with increased

BRB permeability after 12 weeks in retinas of STZ-diabetic rats (Giebel et al, 2005).

Glucose levels increased MMP-9 production in both ARPE-19 (an immortalised

human RPE cell line) and in bovine retinal microvessel endothelial cells (Giebel et

al, 2005). Furthermore, MMP-2 and MMP-9 increased paracellular permeability in

both cell types. MMP degradation of the tight junction protein occludin may have

been responsible for the increase in cell permeability in this study (Giebel et al,

2005).

Using an in vitro model of the BRB, Behzadian et al, (2001) showed that

endothelial cell production of MMP-9 was regulated by Müller cells via TGF-

expression, and thereby demonstrated a direct relationship between TGF-induced

MMP-9 activity, and increased endothelial cell permeability.


48

MMPs are inhibited by the tissue inhibitors of metalloproteinases (TIMPs)

which regulate cell behaviours such as proliferation, apoptosis and angiogenic

processes by mechanisms independent of MMP inhibition (reviewed Baker et al.,

2002; Mannello and Gazanelli, 2001; also Qi et al, 2003). Recent work in the area of

cerebral edema suggests that MMPs and TIMPs play an important role in the

regulation of neuronal cell death and apoptosis through MMP regulation of

excitotoxicity (Jourquin et al, 2003), death receptor activation (Wetzel et al, 2003)

and neurotrophic factor bioavailability (Lee et al, 2001). These actions may have

consequences in brain injury and repair mechanisms (Cunningham et al, 2005) that

could be equally important in the pathology associated with diabetic macular

oedema.

vi) Reactive oxygen species and the BRB

Production of ROS is likely to be elevated in the retina in diabetic retinopathy

(Frank, 2004). Certain regions in the retina are enriched in substrates for lipid

peroxidation that may create an environment susceptible to oxidative damage

(reviewed van Reyk et al, 2003). Imbalances in lipid metabolism, increased O2.

(super oxide anion) formation and possibly NO (nitric oxide) production may play a

role in mediating early vascular and neural dysfunction linked to hyperglycaemic

pseudohypoxia (Williamson et al, 1993). NO-quenching due to generation of excess

peroxynitrite would have consequences for the maintenance of endothelial

homeostasis (reviewed Santilli et al, 2004). Oxidant production by leucocytes that

migrate into retinal tissue after BRB breakdown may precipitate further damage to

integrity of the vascular barrier. Significant cytotoxicity was demonstrated in

cultured pericytes exposed to glycated albumin, while certain antioxidant treatments

had a beneficial effect on cell viability after exposure to ROS (Kim, 2004).

NO is a free radical gas produced by inducible nitric oxide synthasae (iNOS).

NO can be beneficial as a vasodilator but may be neurotoxic in excessive

concentrations (reviewed Nathan, 1992, 1997). iNOS is expressed in response to

stimulation by cytokines and endotoxins and is capable of producing large,

continuous amounts of NO which exerts cytotoxic and cytostatic effects largely

associated with inflammation and injury (Nathan, 1992; Nathan, 1997).


49

NO may be an important mediator of neovascularisation and permeability in

endothelial cells in diabetic retinopathy. iNOS expression corresponded to the

location of GFAP- and vimentin-positive Müller cells in retinas from diabetic

patients with NPDR or with no retinopathy (Abu El-Asrar et al, 2001).

Another report suggested that early BRB breakdown is associated with

increases in the expression of constitutive NOS, rather than iNOS (El-Remessy et al,

2003). Constitutive NOS generates low levels of NO for short periods and is thought

to be important in signal transduction mechanisms, including regulation of blood

vascular tone and blood pressure.

El-Remessy et al, (2003) hypothesised that the effects of oxidative stress in

NPDR may be mediated by VEGF-induced increases in uPAR expression. VEGF

mRNA expression was increased, consistent with previous findings that ischemia,

hyperglycemia and oxidative stress induce growth factor expression; and VEGF-

induced permeability increases were mediated by activation of urokinase and

expression of the uPAR (El-Remessy et al, 2003). uPA expression was inhibited by

L-NAME (a selective constituitive NOS inhibitor) or uric acid (selectively inhibits

tyrosine nitration) supporting the role of ROS-mediated VEGF expression in the

diabetic retina (El-Remessy et al, 2003).


50

1.4 SUMMARY OF THE LITERATURE & THESIS AIMS

The injury caused by elevated blood sugar in diabetes appears to precipitate cellular

responses that go beyond the original insult. Even after strict glycemic control has

been achieved, diabetic retinopathy often continues to progress, indicating that

cellular damage may in some respects lead to irreversible changes.

The literature stemming from recent studies focuses on early disease

processes in an effort to mitigate changes before they progress to a stage beyond

which they cannot be reversed. This review highlights a key role for retinal glia and

other supporting cells in the development of diabetic retinopathy, both in terms of

neuropathic changes and vascular dysfunction.

Heterogeneous cell populations in the retina may explain the geographic

distribution of microvascular lesions. Proteomic profiling and molecular techniques

may explain the extent of cellular diversity within individual populations in the retina

and elsewhere. Results gleaned from primary endothelial cell cultures may be

difficult to interpret where the population is comprised of a heterogenous mix of cells

from different branches of the vascular tree. There is an increasing recognition that

cells do not respond to microenvironmental changes in isolation. Investigators are

trying to address this problem by developing in vitro co-culture models to replicate

the in vivo situation more closely.

1.4.1 Project Aims

The studies in this thesis aim to:

1. Use established methods for the isolation and characterisation of bovine

retinal endothelial cells (BREC) and perivascular cells. (Chapter 2).

2. Compare the functional and morphological characteristics of a co-culture

model using BREC and Müller cells (Chapter 3).

3. Investigate the effect of retinal Müller cells on endothelial cell permeability

under normoxic and hypoxic conditions (Chapter 4).

4. Investigate the effect of co-culture of Müller cells and RPE - proposed as an

in vitro model of retinal laser therapy - on reduction of endothelial cell

permeability (Chapter 5).


51

CHAPTER 2

ISOLATION

AND CHARACTERISATION

OF BOVINE RETINAL CELLS


52

2.1 INTRODUCTION

Diabetic animal models including rats (Strother et al, 2001; Glover et al, 2000; Kern

and Engerman, 1995), dogs (Ammar et al, 2000; Engerman and Kern, 1993) and

primates (Birrell et al, 2002; Comuzzie et al, 2003) have significantly advanced our

understanding of both the metabolic disorder and the complications associated with

diabetes. One major disadvantage with animal models is the finding that

characteristic diabetic lesions seen in humans are often not easily reproducible in

animals. An extended duration of disease is necessary before these models develop

clinically visible lesions that are comparable to the human condition (reviewed

Engerman and Kern, 1995). In addition, species-specific anatomical differences

could influence progression of the disease in a way that does not correspond to the

human condition. Genetic manipulation of the metabolic pathways within knockout

animals may offer the best promise of understanding diabetic processes (Engerman

and Kern, 1995).

The endothelium plays a key role in dictating the variable patterns of

systemic disease (reviewed Aird, 2003). Each vascular bed responds to systemic

changes in a unique way, on both an intrinsic level (that is pre-determined at the

genetic level) and on an extrinsic level (that is governed by the local

microenvironment). It is well known that diabetes-induced vascular lesions are

different in the brain and the retina, even though these tissues are embryologically

similar (Kern and Engerman, 1996). One plausible explanation for this diversity in

diabetic lesions may be the unique location of retinal microvessels which brings

them into close association with several cell layers and types with very different

functions and characteristics (reviewed Ruggiero et al, 1997). Interaction of retinal

microvascular cells with perivascular cells may occur through direct anatomical

contacts or indirect humoral contacts. Evidence for endothelial cell heterogeneity

within different tissues (Rajah and Grammas, 2002; Thieme et al, 1995) and even

between neighbouring endothelial cells of a single vessel (Barber et al, 2000; Michel

and Curry, 1999) has also been reported.

The concept of diversity of the microvasculature within the same tissue

obviously has important consequences for in vitro efforts to model the characteristic

phenotypes of endothelium from the blood-brain and blood-retinal barriers (reviewed


53

Ge et al, 2005). However, a great deal of useful information has already been

gleaned from in vitro investigations including mechanisms of retinal capillary

survival and proliferation (Yafai et al, 2004; Murata et al, 1994; Orlidge and

D'Amore, 1987), capillary permeability characteristics (Behzadian et al, 2001; Dente

et al, 2001; Wolburg et al, 1994) and other endothelial cell processes which have

been dissected using inhibition of specific signal transduction pathways (Raub,

1996). Until more is known about differences within endothelial cell populations

through microarray and proteomic analyses, it may be of limited value to extrapolate

findings from large vessels to microvessels, or to vessels from other tissues

(reviewed Gerritsen, 1987).

In vitro models of the retina provide a useful and relatively inexpensive first

line of investigation for studies that can subsequently be confirmed in situ. Bovine

retinas have been used for these studies because of the easy access to material and

the high yield of cells obtainable. Diabetes-related symptoms may be readily induced

in cows, as already shown in recently developed diabetic sheep models (Ramanathan

et al, 2004; Ramanathan et al, 2002) although cells from non-diabetic animals grown

in high glucose may offer valid insights into the cellular pathogenesis of diabetic

retinopathy.

Prior to developing the co-culture models used in this thesis (Chapters 3-5), it

was crucial to ensure the purity and define the characteristic features of the

endothelial cells and perivascular cells from bovine neural retina.

As such, the present studies aimed to:

1. Refine methods for isolating bovine retinal endothelial cells and perivascular cells.

2. Define the phenotypic characteristics of these cells, including morphology and

expression of specific protein markers, using immunocytochemistry.


54

2.2 MATERIALS & METHODS

2.2.1 Bovine Retinal Cell Isolation

a) Bovine retinal endothelial (BRE) cells

BRE cells were isolated using methods developed in our laboratory (Su and Gillies,

1992) (Gillies et al, 1995) with minor modifications to culture medium composition.

Bovine eyes were obtained from an abattoir and dissected within 10 h of death. The

eyeballs were initially tidied of extraneous fat and muscle and washed briefly in

iodine. A coronal section of the globe was made posterior to the limbus so that

anterior segment (including the lens and vitreous) could be removed in its entirety.

Retinas were removed from the eye cups by gently teasing the neurosensory retina

away from the pigmented layer, starting from the periphery and folding the edges

towards the centre. The neurosensory retina was removed as one piece by pinching

with forceps at the optic nerve. Six retinas were rinsed in a total of 3X cold Iscoves

modification of Dulbecco’s modified Eagles medium (IMDM) (ThermoElectron,

Melbourne, VIC) washes supplemented with 100 IU/ml penicillin and 100 g/ml

streptomycin (ThermoElectron) for 1 h in order to remove adherent RPE cells.

Retinas were removed from the IMDM wash, cut into small pieces and placed in an

enzyme digestion mixture (see Appendix I) containing 500 g/ml collagenase type I

(Roche Diagnostics Australia P/L, Castle Hill, NSW), 200 g/ml pronase (Roche

Diagnostics Australia P/L) and 200 g/ml DNase (Roche Diagnostics Australia

P/L). Retinal fragments were briefly vortexed and incubated for 28.5 mins in a

shaking water bath at 37oC, until the suspension was uniform. Microvessels were

trapped on a 53 m nylon mesh and washed off into cold IMDM. The suspension

was centrifuged at 400 g for 7 mins and the pellet was resuspended into 10 ml of

EC:C6 culture medium (Appendix II) containing 20l/ml bovine retinal extract

(BRE) which is a growth supplement for endothelial cells. The suspension was

placed into two 60 mm tissue culture dishes (Costar Inc, Acton, MA, USA) coated

with 100 g fibronectin (ThermoElectron) and 50 g collagen Type IV (BD

Australia P/L, North Ryde, NSW) and incubated overnight at 37oC. On Day 1 after

the isolation, the unattached material was gently washed away with 2 changes of

warm IMDM. EC:C6 medium was replaced on endothelial cells which were then

left until Day 4 when the culture was between 60-80% confluent. Low pericyte


55

contamination was achieved by careful passaging, using low dose TE (0.05%

trypsin and 0.02% EDTA) for 1 ½ mins at 37oC, effectively skimming endothelial

cells off the dish surface. Suspended BRE cells were moved into two coated 25 cm2

flasks (Nunc A/S, Roskilde, Denmark) (P1). An inverted microscope (Zeiss Telaval

31, Carl Zeiss, North Ryde, NSW) with a phase contrast filter was used to observe

the cells. Second passage endothelial cells were used for all experiments.

b) Bovine retinal pericytes

Bovine pericytes were isolated using the method of Gillies and Su (1993). This

method varies in only minor details from that described for the isolation of BRE

cells (refer to Chapter 2.2.1 Section a) taking into account that retinal capillaries are

closely invested with pericytes. For pericyte isolation retinal fragments were

digested in the enzyme mixture for a shorter time, 20 mins at 37oC. The culture

medium required was not as complex for pericytes: cells were maintained in

Dulbecco’s modified Eagles medium (DMEM) (ThermoElectron, Melbourne, VIC)

containing 20% heat-inactivated fetal bovine serum (FBS) and supplemented with

2mM glutamine, 100 IU/ml penicillin and 100 g/ml streptomycin. Retinal

pericytes were passaged by incubating cells at 37oC with 0.25% trypsin

ethylenediamine tetra-acetic acid (TE) (ThermoElectron) for 3-5 mins.

c) Bovine retinal Müller cells

Müller cells were isolated using a method previously described for pericytes (Gillies

and Su, 1993) with modifications. For each isolation, two eye globes were stored

overnight at room temperature in DMEM with 2mM glutamine, 100 IU/ml

penicillin and 100 g/ml streptomycin. The following day, the neural retina was

isolated and washed in IMDM. Small pieces were cut from the retina and digested

in an enzyme cocktail consisting of 500 g/ml collagenase Type I, 200 g/ml

pronase and 200 g/ml DNase at 37oC for 20 mins. Digested fragments were

filtered through a 53 m nylon mesh sieve. The filtrate was discarded and the mesh

rinsed in IMDM to resuspend the tissue which was then centrifuged for 5 mins at

400 g. The supernatant was discarded and the pellet resuspended in DMEM with

20% FBS, 2mM glutamine, 100 IU/ml penicillin and 100 g/ml streptomycin.


56

Tissue fragments were seeded into a 25 cm2 flask (Nunc A/S, Roskilde, Denmark)

and incubated in a humidified atmosphere with 5% CO2 at 37oC for 7 days.

Thereafter, medium was replaced and changed every 2-3 days until the cells became

confluent. Müller cells were dissociated from the flask surface by incubation at

37oC with 0.25% TE for 3-5 mins.

d) Bovine retinal pigmented epithelium (BRPE)

BRPE were isolated after removal of the neurosensory retina from the posterior eye

cup, and after washing the pigmented layer to remove contaminating retinal cells

that may be adherent to the pigmented surface. Two eyecups were filled with 0.25%

TE and incubated at 37oC for 15 mins. The enzyme mixture was gently rinsed

against the pigmented surface and then discarded. Eyecups were filled with fresh TE

and incubated for a further 30-60 mins. The pigmented surface was again rinsed and

the product of the second digestion was removed and transferred into DMEM with

20% FBS, 2 mM glutamine, 100 IU/ml penicillin and 100 g/ml streptomycin. Cells

were centrifuged at 163 g for 6 mins, resuspended in fresh medium and seeded into

a 25 cm2 flask. Cells were incubated at 37oC for 1-2 weeks.

e) Mixed BRPE and Müller cells

A mixed cell population of activated RPE and migrating Müller cells was isolated

from 24-48 h post-mortem bovine eyes using a modification of Edwards’ method

(1982) and based upon previous observations on the characteristics of post-mortem

Müller cells (Winkler et al, 2002; Smith, 2001; Roque et al, 1992; Hicks and

Courtois, 1990; Burke and Foster, 1984). These studies observed that Müller cells

migrated after death.

A coronal section of the globe was made and the cornea, lens and vitreous

tissues were removed. The retina was carefully dislodged, and the remaining traces

of retinal tissue were removed from the optic nerve with a sterile scalpel blade. Two

eyecups were filled with 0.25% TE and incubated at 37oC for 15 mins. The enzyme

mixture was pipetted to gently loosen adherent cells, and the eyecups containing TE

were incubated for a further 15 mins at 37oC. Dissociated cells were removed and

transferred into DMEM with 20% FBS. Cells were centrifuged at 163 g for 6 mins.


57

The cells were resuspended in fresh medium [DMEM with 20% FBS, 2 mM

glutamine, 100 IU/ml penicillin and 100 g/ml streptomycin] and seeded into a 25

cm2 flask. Cells were incubated at 37oC for 1-2 weeks, as above.


58

2.2.2 Retinal Cell Characterisation

a) FACS analysis: BRE cells

Identification and purity of early BRE cell cultures was carried out by flow

cytometry (FCM) (FACS Calibur, Becton Dickinson, North Ryde, NSW, Australia)

using a panel of antibodies predominantly raised against human proteins (Table

2.1a). Bound antibody was detected with an FITC-conjugated secondary antibody

(Table 2.1b) as per standard methods (Su et al, 2003). Fluorescence between 515

and 545 nm was measured using an argon laser at 488 nm for excitation of FITC.

Forward and side scatter measurements were within the same range for all

populations and 104 events were collected for each sample. Results were presented

as histograms with the number of events versus log10 fluorescence intensity.

b) Indirect immunoperoxidase: BRE cells, pericytes, BRPE and Müller cells

For all experiments, a negative control omitting the primary antibody was included.

All cells (except endothelial cells) were grown on uncoated Thermanox coverslips;

for endothelial cells coverslips were pre-coated with 1.5 g fibronectin, 1 g

collagen IV and 1 g laminin (BD Australia P/L) at room temperature for 2 h, then

briefly rinsed with phosphate buffered saline (PBS) (0.1M, pH7.2). Cells at a

density of 1 X 105 were seeded onto coverslips in 24 well plates and grown in the

appropriate medium. Sub-confluent or confluent cells were washed twice with PBS,

fixed with acetone or methanol at -20oC for 5 mins and rinsed with PBS.

For von Willebrand’s Factor (vWF) immunohistochemistry staining, cells

were incubated with 10% donkey serum/PBS to reduce non-specific binding, then

incubated with rabbit anti-human vWF (Table 2.1a) overnight at 4oC. Cells were

washed 3 times with PBS and incubated with biotinylated donkey anti-rabbit

secondary antibody (Table 2.1b) at 4oC for 1 h. After washing with PBS,

ExtraAvidin peroxidase (Table 2.1b) was added for 45 mins. Bound antibody was

detected with DAB chromagen (Dako Cytomation, Australia). Coverslips were

dehydrated through a series of alcohols and xylene and mounted on slides with

DePeX (BDH AnalaR, Australia). Bovine aortic endothelial cells were used as a

positive control for vWF labelling.


59

For Smooth Muscle Actin (SMA) staining, cells were incubated in 10%

sheep serum, then in mouse anti-human SMA (Table 2.1a) followed by

biotinylated sheep anti-mouse secondary antibody (Table 2.1b). After incubation in

ExtraAvidin peroxidase, immunoreactivity was detected with Vector Red

chromagen (Vector NovaRED, Vector Laboratories, Burlingame, CA, USA).

Coverslips were then dehydrated and mounted, as above.

For cellular retinaldehyde binding protein (CRALBP) staining, cells were

initially incubated in either 10% donkey or sheep serum in PBS, followed by a

polyclonal CRALBP raised in rabbit (Table 2.1a) or a monoclonal CRALBP raised

in mouse (Table 2.1a), respectively. Cells were washed and incubated with a

biotinylated secondary antibody and ExtraAvidin peroxidase as above. Bound

antibody was detected with Vector Red chromagen. A spontaneously immortalised

human cell line MIO-M1 (P76) (Limb et al, 2002) was used as a positive control for

CRALBP immunolabelling of bovine Müller cells.

c) Indirect immunofluorescence: BRE cells, BRPE and Müller cells

Cells were seeded at a density of 1 X 10 4 onto glass coverslips and grown until sub-

confluent, rinsed with PBS and fixed in 2% paraformaldehyde in 0.1M PBS (pH

7.4) at 4oC for 1 h. Cells were incubated in 10% sheep or donkey serum in PBS,

followed by overnight incubation at 4oC with the primary antibody. BRPE cells

were immunolabelled with mouse anti-human cytokeratin (Table 2.1a) using human

RPE cells as a positive control.

BRE cells and mixed BRPE and Müller cells were immunolabelled with

rabbit anti-human ZO-1; Müller cell cultures were labelled with mouse anti-swine

vimentin (Table 2.1a). Cells were washed and incubated with an appropriate

biotinylated secondary antibody. After rinsing in PBS, bound antibody was detected

with a streptavidin-Cy-3 conjugate (Table 2.1b). Coverslips were mounted on slides

in glycerol and examined with a Leitz Diaplan fluorescence microscope (Leitz

Messtechnik GmbH, Wetzlar, Germany). Images were captured with Leica DC

Viewer Computer Software (Version 3) (Leica Microsystems Ltd., Heerbrugg,

Switzerland).


60

2.3 RESULTS

2.3.1 Retinal Cell Isolation

a) BRE cells

By Day 1 after the isolation there were as many as 10 individual colonies of BRE

cells to each grid square in the tissue culture dish (Figure 2.1A). These clones

multiplied quickly so that after 1 week, cells formed confluent monolayers (Figure

2.1B,C). Occasional pericyte contamination of endothelial cell isolations occurred in

primary culture (Figure 2.1D). Pericyte contamination of subsequent passages could

be minimised by selective trypsinisation as cells were passaged to P1.

b) Pericytes

Retinal pericytes were large, flat cells with cytoplasmic filaments and irregular

processes (Figure 2.1E,F). Pericytes grew slower than BRE cells (Table 2.2),

eventually forming individual, multilayered nodules (Figure 2.1G) after eight weeks

in culture. Occasionally RPE cells contaminated pericyte cultures and could be

mistaken for pericytes when RPE cells had degranulated in culture (Figure 2.1H).

c) Müller cells

Once attached to the tissue culture dish surface in the primary culture, bovine retinal

Müller cells grew rapidly becoming confluent 1-2 weeks after isolation (Figure

2.2A,B). When Müller cells were cultured in serum-free medium [DMEM with

2mM glutamine, 100 IU/ml penicillin and 100 g/ml streptomycin] they exhibited

long, delicate radial fibre structures and distinctive varicosities around the cell body

(Figure 2.2C), however in serum-containing medium (which was required to expand

the cell population) cells became flat and hypertrophied (Figure 2.2D).

d) BRPE

Primary RPE cells often failed to attach to the isolation dish surface (Figure 2.3A,B)

and remained suspended in the medium for up to 1 week after the isolation. These

cells could generally be ‘induced’ to attach by centrifugation and replating in a new

tissue culture dish, thereby increasing the overall yield. Attached cells grew to

confluence in about 2 weeks (Figure 2.3C). Subsequent passaging of RPE cells was


61

characterised by gradual loss of pigmentation and increasingly disordered growth

(Figure 2.3D). Occasionally Müller cells contaminated RPE cultures (Figure 2.3E).

When this occurred, Müller cells readily overgrew the slower growing RPE cultures.

Only cultures that were >95% pure were used in experiments.

2.3.2 Retinal Cell Characterisation

A summary of antibody affinities in cultured bovine retinal cells is provided in

Table 2.1a. The purity of the endothelial cell isolation was confirmed by FACS

analysis, displaying positive immunolabelling for vWF (Factor VIII) compared to

the rabbit IgG control (Figure 2.4). Factor VIII antibody raised in mouse was

negative by comparison, as was CD31 and CD34.

Characterisation of retinal cells showed positive immunolabelling of BRE

cells with polyclonal anti-vWF (Figure 2.5C), pericytes with monoclonal anti-

SMA (Figure 2.5F), and Müller cells and RPE with polyclonal anti-CRALBP

(Figure 2.5I, L respectively).

Bovine RPE cells did not immunolabel with monoclonal anti-human

cytokeratin (not shown), however polyclonal anti-human ZO-1 staining was a

consistent immunomarker for both BRE cells (see Chapter 3, Figure 3.1B) and RPE

cells (Figure 2.6A). ZO-1 immunoreactivity was localised to RPE cell borders and

at the ends of Müller cell processes in mixed RPE and Müller cell cultures (Figure

2.6C,D). Müller cells reliably labelled with a monoclonal anti-swine vimentin

antibody (see Chapter 3, Figure 3.1C).


62

Table 2.2 Summary of retinal cell growth characteristics

Cell type Relative cell Potential contaminating cell type

growth rate

____________________________________________________________

BRE cells fast pericyte, fibroblast, Müller cell, RPE

Pericyte slow BRE cells, RPE

Müller cell fast RPE

RPE slow Müller cell

_____________________________________________________________

Key: BRE cells, bovine retinal endothelial cells; BRPE, bovine retinal pigmented epithelium


63

2.4 DISCUSSION

BRE Cell Isolation and Characterisation

The present studies used a combined digestion, filtration and extracellular matrix

(ECM) coating approach to select bovine endothelial cells from whole retina.

Alternative approaches have also been used to isolate BRE cells including CD31-

coated Dynabeads (Dynal Biotech, NY, USA). Dynabeads have been routinely used

in our laboratory for human retinal endothelial cell isolations (L. Wen, J. Wyndham,

personal communication). However, the lack of success with monoclonal CD31

antibody coated Dynabeads for isolation of BRE cells is consistent with the absence

of immunolabelling of bovine cells (using the same antibody) in the FACS analysis

(see below). A recent report detailed the use of monoclonal anti-rat CD31-coated

Dynabeads for the isolation of rat retinal vascular endothelial cells (Tomi and

Hosoya, 2004). An endothelial cell-specific ligand such as Ulex europaeus I (UEAI)

lectin (Sigma-Aldrich P/L, Sydney, NSW) has also been used to coat Dynabeads for

the isolation of bovine mesenteric lymphatic endothelium (Jones and Yong, 1987)

and for isolation of human choroidal endothelial cells (Penfold et al, 2002). UEAI

lectin could be used for isolation of endothelial cells from bovine retinas in future

studies.

The critical features of the endothelial cell isolation in this study were the

wash steps (to remove contaminating RPE cells), the fine mincing of retinal tissue

into small (approx. 0.3 cm2) pieces, and thorough enzyme digestion, that ultimately

produced only capillary fragments in the culture dish. The importance of keeping

eyes and retinal tissue at 4oC from the point of death and for the duration of the

isolation cannot be overstated, since cells of the retinal vasculature have only a

limited lifespan once post mortem changes begin to occur.

Conditioned medium from C6 (rat glioma) astrocytes was added to the

endothelial cell specific media (EC:C6 medium) in an effort to promote cell growth

and viability, and to improve BRB characteristics (Arthur et al, 1987). Although

there are conflicting reports in the literature with regard to the effect of conditioned

medium from co-cultured C6 cells on endothelial cell permeability (Fischer et al,

2000; Abbruscato and Davis, 1999) no adverse effects were seen in endothelial cell

cultures using medium with C6 supplement. Other groups have maintained primary


64

cultures of bovine retinal endothelial cells in a less complex medium (Behzadian et

al, 2001). Various combinations of medium using different concentrations of horse

or bovine serum in DMEM were investigated - during isolation and characterisation

studies described in this chapter - without success, due to the rapid overgrowth of

contaminating cells.

The EC:C6 culture medium was designed to optimise endothelial cell growth

and retard contaminating cell types. The EC:C6 media component bovine retinal

extract (BRE) was used as a substitute for commercially derived endothelial cell

growth factor supplement, and most likely contained many critical factors expressed

by perivascular cells. Heparin and human platelet poor serum, selected for

endothelial cells by inhibiting pericyte activation and growth respectively (Clowes

and Karnowsky, 1977; Buzney et al, 1983). Contaminating pericytes can be further

reduced by selective trypsinisation using low dose TE for a prescribed time (1 ½

mins at 37oC) which releases endothelial cells during passaging, leaving pericytes

attached to the tissue culture dish surface. Previously, a method for mechanical

weeding of contaminating cells with a diathermy needle has been used (unpublished

observation) and there are many reports of removal of non-endothelial cells by

sterilised probes (Yan et al, 1996; Hull et al, 1996). However, manual weeding

appeared to encourage more aggressive growth by contaminating cells, and this

technique was not pursued.

FACS analysis using a variety of antibodies raised against human proteins

showed a largely negative reaction in bovine tissue for endothelial cell-specific

proteins such as CD31 and CD34 (Figure 2.4). Polyclonal rabbit anti-human vWF

displayed positive immunolabelling for vWF compared to the IgG control (Figure

2.4). This antibody displayed positive immunoreactivity for large and small vessel

bovine endothelial cells (Figure 2.5A and C, respectively). Polyclonal vWF was

subsequently used for all retinal cell characterisation studies.

vWF labelling is a universally accepted as an unequivocal marker of

endothelial cells (reviewed Nachman and Jaffe, 2004). Uptake of acetylated LDL is

also commonly used (Kondo et al, 2003; Behzadian et al, 2001; Dente et al, 2001;

Orlidge and D'Amore, 1987). The absence of cytokeratin, smooth muscle actin and

GFAP expression are also useful indicators for the presence of endothelial cells.


65

Pericytes

Pericytes are closely associated with endothelial cells, occurring in a ratio thought to

be as high as 1:1 within retinal microvessels (reviewed Chakravarthy and Gardiner,

1999). Although the possibility of endothelial cell contamination of pericyte

cultures was high, under the conditions that are optimal for pericyte growth and

survival, endothelial cells generally did not proliferate for a number of reasons.

Directly contacting pericytes have an inhibitory effect on endothelial cell growth in

vitro (Orlidge and D'Amore, 1987), endothelial cells do not grow optimally on

uncoated plastic surfaces (D'Amore, 1990) and the simplified pericyte medium

(without BRE) does not select for endothelial cells. The finding that post-confluent

pericyte cultures formed isolated, multilayered nodules has been previously

observed for bovine (Wong et al, 1992; D'Amore, 1990) and rat retinal pericytes

(Kondo et al, 2003). Although isolation conditions were optimised for pericytes,

pericyte growth was slow in comparison to other retinal cells (Table 2.2). Wong et

al., (1992) has observed that brain pericytes grow up to 10 times faster than retinal

pericytes and has suggested that the relative inability of retinal pericytes to re-

proliferate may explain the selective pericyte degeneration in the retinal circulation

associated with diabetic retinopathy.

SMA immunoreactivity was localised to cytoplasmic filaments in the

isolated retinal pericytes.SMA is the most commonly used in vitro marker of

pericytes and smooth muscle cells (SMC) (Kondo et al, 2003; Dente et al, 2001;

Katsura et al, 2000; Murata et al, 1994; Saunders and D'Amore, 1992). It is widely

believed that SMA-expressing pericytes function as important regulators of blood

flow in the vasculature (reviewed Hirschi and D’Amore, 1996). However, retinal

pericytes of BRB capillaries may not express SMA in vivo (Nehls and

Drenckhahn, 1991) suggesting that some pericytes have different functions

(reviewed Pardridge, 1999). It is unclear whether the expression status of SMA

changes under pathophysiologic conditions. Future studies investigating different

pericyte functions should incorporate a wider panel of pericyte-specific markers

including platelet derived growth factor receptor (PDGFR) (Fruttiger, 2002) and

chondroitin sulphate proteoglycan NG2 (Ozerdem et al, 2002).


66

Müller cells

The explant method of Burke and Foster (1984) provided useful Müller cell cultures

although the procedure was time-consuming, requiring a 14 day initial incubation

before cells could be plated out for attachment. To achieve a faster outcome, the

method for isolation of BRE cells described by Gillies et al, (1995) was modified

and used together with the recommendation of Hicks and Courtois (1990) to soak

eyeballs overnight in DMEM at room temperature prior to isolating retinal tissue.

This step relies upon the ability of Müller cells to survive and to become activated

in post mortem retinal tissue (Winkler et al, 2002; Smith, 2001; Roque et al, 1992;

Hicks and Courtois, 1990; Burke and Foster, 1984), where other more delicate cell

types have only a limited capacity for survival under anaerobic conditions. Once

established in primary culture, Müller cells were fast growing and aggressively

overgrew any cultures in which they were the contaminating cell type. As expected,

Müller cell growth was usually less robust the more times that cells were passaged.

Many other protocols for Müller cell isolation describe a mechanical

trituration technique that dissociates the retinal cell layers in a non-specific manner

(Jingjing et al, 1999; Arroyo et al, 1997). These techniques were attempted many

times during the present studies without success, perhaps related to the mechanical

disruption of cells, where the long, delicate radial processes of Müller cells may

become damaged if the isolation technique is too robust.

In the present studies, bovine Müller cell isolates were characterised using a

panel of antibodies including anti-CRALBP, anti-vimentin, anti-GFAP and anti-

NCAM (Table 2.1a). CRALBP is involved in the (dark adaptation) visual cycle as a

transporter protein of vitamin A derivatives (Bunt-Milam and Saari, 1983) and is

accepted as a specific marker for retinal Müller cells and RPE (refer to Chapter 1,

literature review). Müller cells are highly polarised in vivo with specific intracellular

protein localisations that may be disrupted by culturing. Several groups have

reported variable success with CRALBP labelling in cultured cells (Limb et al,

2002; Guidry et al, 2003). Guidry (1996) observed that CRALBP labelling was lost

in primary Müller cell cultures after 2 weeks, most likely due to lack of essential

contacts within the tissue environment. In a report describing experimental retinal

detachment in cat eyes (Fisher et al, 2001), CRALBP labelling of Müller cells


67

rapidly decreased within 3 days. Downregulation of CRALBP was associated with

gliotic changes in Müller cells. The effect was reversed by enriching atmospheric

oxygen (Fisher et al, 2001).

In this study, CRALBP immunoreactivity was localised within the

cytoplasm and along Müller cell processes. Cultured Müller cells without extensive

cellular contacts required higher concentrations of CRALBP antibody (1:100

dilution) as described, together with a 3-step immunohistochemical protocol that

significantly amplified the weaker expression signal. Where the reactivity to anti-

GFAP was at times variable in cultured Müller cells (refer to Chapter 1, literature

review), expression of vimentin protein was usually robust (Figure 3.1C). Vimentin

is expressed in a high proportion of brain cells during development (Elmquist et al,

1994), and throughout adulthood vimentin continues to be expressed in astrocytes

and Schwann cells (Sancho-Tello et al, 1995; Pulido-Caballero et al, 1994; Kameda,

1996). Vimentin is also a marker of CNS ependymoglia, a group of radial glia that

is comprised of tanycytes in the brain and spinal cord and Müller cells in the retina

(Reichenbach and Robinson, 2005). Within adult human retinal Müller cells,

vimentin immunoreactivity is localised to cellular (Type III) intermediate filaments

(Famiglietti et al, 2003). In normal retinas, vimentin expression in Müller cells is

more intense than GFAP, and expression of both GFAP and vimentin often

increases as a response to retinal stress (Lewis and Fischer, 2003).

Some studies have reported -SMA protein expression in late-passage (P5

and later) Müller cells (Guidry, 1996) and in MIO-M1 cells (Limb et al, 2002). This

has generally been thought to be indicative of gliotic changes that occur when

Müller cells are removed from the retinal microenvironment or as a result of chronic

stress in vivo (Guidry, 2005). -SMA expression was not apparent in the early

passage Müller cell cultures in the present studies.

Together with the antibodies described above, Müller cell identification was

made from the distinctive cell morphology and rapid proliferation of cultured cells

in the early passages.


68

BRPE

The method for isolation of RPE cells was adapted from Edwards (1982), who

recommended incubation of trypsin-EDTA in eyecups for up to 2½ hours. In the

present study, 1 hour incubation was found to be effective for isolating RPE cells

from bovine eyes. For the purposes of the experiments described in Chapter 5, long

term cultures of RPE were grown. Observations of RPE characteristics in long term

cultures agree with those reported by Glaser and colleagues (1987), who found that

superconfluent RPE cells did not maintain the ordered cell-cell relationships that are

seen in vivo. At the molecular level, Kaida et al, (2000) reported that long term RPE

cultures may be closer to the in vivo situation because cellular junctions were more

mature compared to RPE cells from early confluent cultures.

Early passage RPE cells could be readily identified by the intense

pigmentation and characteristic cobblestone morphology (Figure 2.3A-C). CRALBP

is considered to be a specific marker of RPE cells and retinal Müller cells (refer to

Chapter 1, literature review) and labelling was observed in both intranuclear and

perinuclear locations. Type I and Type II intermediate filaments, also known as

cytokeratins have been frequently used to characterise RPE, both in vivo and in vitro

(McKechnie et al, 1988) however, the monoclonal antibody used here was raised

against human cytokeratin and did not label bovine RPE cells. ZO-1 by comparison

(also raised against human protein) was a polyclonal antibody, raised in rabbit and

cell membrane ZO-1 immunoreactivity was found between bovine RPE cells,

consistent with tight junctions.

Mixed RPE and Müller cell characterisation with ZO-1

When neuronal cells are destroyed as a result of laser photocoagulation in the

treatment of macular oedema (refer to Chapter 5) Müller cells and RPE cells

migrate into laser-affected areas and form a ‘glial scar’ apparently in an effort to

repair damage to the outer BRB (Roider et al, 1992). Although tight junctions have

not been unequivocally demonstrated between Müller cells in situ, this lack of

expression may be due to inhibitory influences from surrounding retinal neurons.

Alternatively, the presence of abundant neuronal cells could physically make glia-

glial contacts difficult (Wolburg et al, 1990). Others have reported formation of glia


69

limitans –like specialisations in Müller cell plasma membranes after induction of

subretinal gliosis in rabbit eyes (Korte et al, 1992). In the present study, punctate

ZO-1 immunolabelling was seen at the ends of Müller cell processes, and

circumferentially in a continuous band around RPE cells in mixed RPE and Müller

cell cultures. Further studies of the formation of specialised junctions by Müller

cells are warranted. The differential expression of stress-related proteins vimentin

and GFAP in Müller cells may provide a better insight into the ‘laser scar’

formation after photocoagulation therapy (Lewis and Fischer, 2003).

These studies established methods for the isolation and phenotypic

characterisation of primary bovine retinal cell populations. The capacity to culture

and identify BRE cells, pericytes, RPE and Müller cell populations provides the

basis for the experiments described in Chapters 3, 4 and 5. Co-culturing methods

were used in these chapters to define the in vitro effects of cellular interactions.


70

CHAPTER 3

TRANSENDOTHELIAL

ELECTRICAL RESISTANCE OF

BOVINE RETINAL

ENDOTHELIAL CELLS

IS INFLUENCED BY CELL

GROWTH PATTERNS: AN

ULTRASTRUCTURAL STUDY

This work appeared in the publication:

Tretiach M, van Driel D, Gillies MC. Transendothelial electrical resistance of bovine

retinal capillary endothelial cells is influenced by cell growth patterns: an

ultrastructural study. Clin Exp Ophthalmol 2003; 31: 348-353


71

3.1 INTRODUCTION

In vitro systems of varying degrees of complexity have been established to model

the vasculature and to study interactions between endothelial cells and smooth

muscle cells (Weber et al, 1986; van Buul-Wortelboer et al, 1986), endothelial cells

and pericytes (King et al, 1987), endothelial cells and microglial cells (Diaz et al,

1998; Stanness et al, 1999), and endothelial cells with neuronal cells (Cestelli et al,

2001). The variability associated with primary cultures has led to the increased use

of immortalised cell lines (Diaz et al, 1998; Penfold et al, 2002; Albelda et al, 1988)

and microvascular endothelial cells are now available both in immortalised (Roux et

al, 1994) and conditionally immortalised forms (Hosoya et al , 2001). Medium

conditioned by non-endothelial cells has been included as an additive to the standard

medium to stimulate expression of endothelial cell specific characteristics that may

be lost in vitro (Penfold et al, 2002; Wong et al, 1987; Gardner et al, 1997;

Behzadian et al, 1998). Co-cultures, where endothelial cells are grown in close

proximity to cells of another lineage have also been used (Diaz et al, 1998; Hayashi

et al, 1997).

Glial cells have been found to affect endothelial cell permeability within the

blood-brain barrier (Janzer and Raff, 1987; Igarashi et al, 1999) and Müller (glial)

cells of the retina are known to be involved in the BRB, mediating both barrier-

enhancing (Tout et al, 1993) and barrier-impairing (Behzadian et al, 2001)

properties in endothelial cells within the (outer layer of the) inner BRB. In the

present study, Müller cells were co-cultured with endothelial cells in an effort to

improve the barrier resistance of our model. The ultrastructural morphology of an in

vitro retinal endothelial cell model was compared with electrical resistance

measurements, which reflect the permeability of a barrier to electrolytes (Gillies and

Su, 1995). Morphological differences between preparations exhibiting high and low

TEER were investigated. In particular, interactions between Müller cells and

endothelial cells in co-culture (and whether these cells made direct contact through

the filter pores) were examined.


72

3.2 MATERIALS AND METHODS

3.2.1 Cell isolation and culture

BRE cells were isolated using the method of Gillies et al, (1995) (For a complete

description of endothelial cell isolation, see Chapter 2.2.1 Section a). Bovine Müller

cells were isolated using a modified method for pericyte isolation (Gillies and Su,

1993) (see Chapter 2.2.1 Section c).

Müller cell cultures were immunostained with antibodies to vimentin (V9

clone) and NCAM (Table 2.1a) to confirm purity as described in Chapter 2.2.2

Section c.

3.2.2 Electrical resistance studies

TEER measurements were used to assess the paracellular permeability of endothelial

cells. Two-chamber 0.33cm2 polycarbonate Transwell filters (Corning Costar Corp.,

Cambridge, MA) were used. The inner (luminal) chamber of the filter was pre-coated

with ECM components for BRE cell cultures including 0.1% gelatin, followed by

100 g fibronectin, 50 g collagen IV and 50 g laminin (see Appendix 1). The

outer (abluminal) chamber was left uncoated for Müller cells that readily adhere to

plastic. The co-culture group consisted of BRE cells and Müller cells seeded on the

luminal and abluminal surfaces of the filter, respectively.

Three control groups were assessed: Two ‘no cell’ wells (for calculation of the

resistance generated by ECM alone, as well as BRE cell and Müller cell

monocultures (see preface page xv). There were between 4 to 6 filters in each group.

Cells were cultured on both 3.0m and 0.4 m pore size filters.

Early passage Müller cells were conditioned in EC:C6 medium for 24 h prior to

seeding on filters. Between 2,500-5,000 Müller cells per well were seeded to the

abluminal filter surface and incubated for two hours at 37oC in humidified

conditions. The plastic inserts were turned over and BRE cells (passage 2) were

seeded onto the luminal filter surface (35,000 cells/well) on Day 0 (co-cultured cells,

see preface page xv). Media was changed on Day 1, and then every second day.

Electrical resistance measurements were made on Day 3, 5, 7 and 9 with a Millipore

ERS resistance meter (Millipore Australia, North Ryde, NSW). Resistances were

calculated as the average resistance of the different groups (raw reading), minus the


73

average reading from the ‘no cell’ wells, multiplied by the area of the Transwell filter

(0.33cm2).

3.2.3 Immunocytochemistry

At the completion of TEER experiments, cells attached to polycarbonate filters were

initially rinsed with PBS (pH 7.4), fixed in 2% paraformaldehyde at 4oC for 1 h and

washed again in PBS. Cells were blocked and stained (as described in Chapter 2.2.2

Section c) with a polyclonal antibody to the zonula occludens (ZO-1) protein (Table

2.1a). Bound antibody was detected with a streptavidin-Cy-3 conjugate (Table 2.1b).

Preparations were cut from the plastic wells, mounted on slides in glycerol and

examined with an argon krypton confocal microscope (Leica, Solms, Germany) with

a filter that allowed 568 nm wavelengths to stimulate the fluorochrome label.

3.2.4 Transmission electron microscopy

Filters with monocultures and co-cultured cells were selected for electron

microscopy to compare filter pore size and cell growth characteristics. From a total

of six preparations, the preparation with the highest and lowest electrical resistance

in each group was fixed and processed for electron microscopy after 8-9 days in

culture. Transwell inserts were washed in PBS (pH 7.4) fixed with 2.5%

glutaraldehyde in 0.1M sodium cacodylate and 22mM betaine (pH7.4) and

embedded in epon-araldite by standard methods. Thick sections (m) were cut and

stained with toluidine blue (1% solution) for light microscopy. Between 4-6 ultrathin

sections were cut from the widest face of each block, collected on grids and stained

with uranyl acetate and Reynold’s lead citrate and examined in a 7100FA electron

microscope (Hitachi Koki, Tokyo, Japan).


74

3.3 RESULTS

Cell isolation and ultrastructure

Endothelial cell purity was improved by selective trypsinisation (Su and Gillies,

1992) (Figure 3.1A). Endothelial cells formed thin layers with close interdigitations

on the Transwell filter as seen with immunocytochemistry for ZO-1 (Figure 3.1B).

Ultrastructurally, endothelial cells exhibited numerous junctional structures

including adherens junctions and tight junctions at zones of cell-cell contact (Figure

3.2). Cell polarity was verified by observation of junctional structures at the luminal

cellular surfaces and basally positioned nuclei in endothelial cells (not shown).

Following the method of Roque et al, (1992) in which globes are incubated

overnight, Müller cells were found to be more than 95% pure by light microscopy

and field counts after staining to the vimentin antibody (Figure 3.1C). In primary

cultures grown with low serum content (below 2%) medium, Müller cells exhibited

a characteristic morphology (Figure 3.1D). The NCAM labelling detected only

minimal contamination by neuronal cells (not shown). Based upon previous

observations that astrocytes appear to be adversely affected after the overnight

soaking step, GFAP staining was not carried out. Ultrastructurally, Müller cells

were readily distinguished from endothelial cells because they formed loose

aggregations or were arranged in irregular multilayers; they displayed no tight

junctions and were not closely associated with neighbouring cells.

Cell growth on Transwell filters

Endothelial cell and Müller cell monocultures grew on only one side of 0.4 m filters

but on both sides of 3.0 m filters. Co-cultured cells grew on both sides of 0.4m

and 3.0 m filters, as expected. Müller cells seeded on the abluminal surface of 3.0

m filters translocated to the luminal surface and became established on both sides

of the filter. Cell processes could be seen passing through the pores (Figure 3.2A). In

co-cultures on 3.0 m filters, endothelial cells formed a superficial layer around the

multilayered Müller cells on both the luminal and abluminal surfaces (Figure 3.2B).

However, endothelial cells and Müller cells seeded on opposite sides of 0.4 m

filters remained separated.


75

Correlation of cell morphology and TEER

Filters with high TEER had 1-3 layers of endothelial cells on the luminal surface

(Figure 3.2C, D). Filters with low TEER exhibited irregular endothelial cell growth

with incomplete coverage of the filters and/or multilayering (not shown). The TEER

increased from Day 3 to 7 in the high series of resistances (Table 3.1) and by Day 7,

endothelial cells co-cultured with Müller cells on 0.4 m filters displayed a marked

(but not significant) increase in mean +/- SD electrical resistance above endothelial

cell monocultures (57.06 +/- 15.38 ohms.cm2 vs. 34.13 +/- 9.87 ohms.cm2,

respectively). In most cases, TEER had begun to fall by Day 9 indicating loss of a

functional barrier. After 9 days in culture, cell viability appeared diminished with

ultrastructural evidence of apoptotic and necrotic cell death and abundant

microfilament expression (not shown).


76

3.4 DISCUSSION

The present investigation examined the variability of resistances in two permeability

studies by selecting individual filters that generated both a high and a low TEER.

The main finding of the study is that barrier formation in endothelial cells, as

assessed by the TEER assay, may be positively affected by co-culture with Müller

cells.

Other investigators of endothelial cell permeability performed ultrastructural

studies only to verify the similarity of their model to the in vivo situation (Dehouck

et al, 1990; Furie et al, 1984). In this report, an attempt was made to expand these

findings in order to correlate a functional measurement with cellular ultrastructure.

Although the number of filters examined by electron microscopy in this study was

limited, the findings – that TEER appears to be adversely affected by unusual cell

growth characteristics, incomplete coverage of the filter surface, filter pore size

and/or age of the culture – were expected.

Measurement of electrical resistance was originally used to assess epithelial

barrier function or paracellular permeability (Cereijido et al, 1978). Since then, the

acronym TER has been used to describe a number of different concepts:

transepithelial (Rizzolo and Li, 1993; Lo et al, 1999), transmonolayer (Albelda et al,

1988; Milton and Knutson, 1990), transendothelial (Dye et al, 2001), and even

(although erroneously) transcellular (Tilling et al, 1998; Wang et al, 2001) electrical

resistance. Recently, a more appropriate term for endothelial cell barrier function –

transendothelial electrical resistance, or TEER – has been proposed (Tan et al, 2001;

Iwasaki et al, 1999; Cucullo et al, 2002).

It is generally assumed that cells in barrier assays (epithelium or

endothelium) form monolayers in vitro (van Buul-Wortelboer et al, 1986; King et

al, 1987; Albelda et al, 1988; Haudenschild, 1984; Schirmacher et al, 2000) as they

characteristically do in vivo. However in the present study, endothelial cells were

frequently found growing in thin multilayers, sometimes up to six cells thick. When

preparations from the high and low series of experiments were compared

ultrastructurally, it was observed those that achieved a high electrical resistance

grew confluent endothelial cells, while low electrical resistance preparations

displayed irregular cell growth with incomplete coverage of the filter surface.


77

Monocultures and co-cultured cells were seeded on different pore size filters

to determine whether cellular processes grew into pore spaces, made contact with

cells on the opposite side of the filter and to assess whether these contacts

influenced TEER. There was a marked (but not significant) increase in TEER in co-

cultured endothelial cells and Müller cells compared with endothelial cell

monocultures on 0.4m filters in the high series, although there was no evidence of

direct contact between cells on either side of the filter. This suggests that Müller

cells in close proximity to endothelial cells have a positive effect on barrier

function, beyond the additive effect of endothelial cells and Müller cells alone,

however these results are only preliminary and require further investigation.

There was evidence of cellular migration through the large pore size filters

of both endothelial cells and Müller cells. Others have observed cellular lamellipoda

of smooth muscle cells within 5.0 m pore spaces (Weber et al, 1986). Albelda et

al. have observed bovine aortic endothelial cells migrating into 3.0 m pore spaces

(Albelda et al, 1988), however Hayashi et al did not, and instead noted that rat brain

astrocytes made direct contact with human umbilical vein endothelial cells through

the large pore size filters (Hayashi et al, 1997). Translocating endothelial cells in

this study had the undesired effect of forming a double layer of cells on the luminal

and abluminal filter surfaces. Given these observations, only 0.4 m pore filters are

now used in the permeability experiments.

Using the same methods, the high resistances of Gillies et al, (1995) could

not be replicated. Conditioned medium from (rat glioma) astrocytes has been added

to the endothelial cell-specific medium in an attempt to improve the barrier

resistance; however, the baseline resistance of endothelial cell barriers usually

remained between 20 and 30 ohms.cm2 (unpublished data). It is difficult to account

for the differences in the overall resistances between the different studies. Some

unidentified change in the medium, basement membrane components, or filters is

likely. Recently, it was suggested that endothelial cells with ‘barrier characteristics’

may comprise only a limited number of vessels within retinal tissues (Ge et al,

2005).

The filters in the present study were fixed for electron microscopy after cells

had been in culture for between 7 and 9 days, in keeping with a report that the limit


78

of endothelial cells on gelatin-coated plastic is between 7 and 10 days (Furie et al,

1984). There was evidence of diminished viability (apoptotic and necrotic cell

death) in some cultures, and high metabolic activity (with abundant rough

endoplasmic reticulum and mitochondria) in others. Other researchers have

observed co-cultures of endothelial cells and smooth muscle cells that remained

stable for more than 10 days and thereafter broke down related to collagenolytic

activity of smooth muscle cells (van Buul-Wortelboer et al, 1986). It has been

suggested that ‘culture activated’ cells may secrete factors that adversely effect

barrier function (King et al, 1987; Dodge and D'Amore, 1992) and these

observations should be taken into consideration when analysing the results of a co-

culture assay using freshly isolated cells.

In this study a functional measurement (TEER) was compared with cellular

ultrastructure to investigate the variability of electrical resistance measurements

within a series of experiments. These results indicate that it may be misleading to

use the term ‘transmonolayer electrical resistance’, originally intended to describe

an epithelial cell monolayer, for a barrier assay involving endothelial cells. A

preferable term is ‘transendothelial electrical resistance’, consistent with the barrier

characteristics of endothelial cells being different from epithelia. This terminology

has been increasingly used in endothelial barrier studies in recent years (Hayashi et

al, 2004; Tan et al, 2001). There are several limitations to using endothelial cells in

vitro including the (previously mentioned) observation that endothelial cells may

not always grow as a monolayer in culture. As such, the usefulness of this in vitro

system is best restricted to endothelial cell cultures that achieve a high TEER

(Tretiach and Gillies, 2001). This study has shown that a TEER of at least 20

ohms.cm2 by Day 5 reflects acceptable barrier formation in endothelial cell

monocultures. Finally, co-culturing endothelial cells with Müller cells has the

potential to improve the overall barrier resistance, as well as to provide useful

information about cell interactions in vitro and possibly in vivo (See Chapters 4 and

5).


79

CHAPTER 4

EFFECT OF MÜLLER CELL

CO-CULTURE ON IN VITRO

PERMEABILITY OF

BOVINE RETINAL VASCULAR

ENDOTHELIUM IN NORMOXIC

AND HYPOXIC CONDITIONS


Tretiach M, Madigan MC, Wen L, Gillies MC. Effect of Müller cell co-culture on in

vitro permeability of bovine retinal vascular endothelium in normoxic and hypoxic

conditions Neuroscience Letters 2005; 378: 160-165


80

4.1 INTRODUCTION

The BRB exists at the level of the retinal capillary endothelium [‘inner’ BRB] and

the retinal pigmented epithelium (RPE) [‘outer’ BRB]. Breakdown of the BRB

appears to be a primary event in the pathogenesis of diabetic retinopathy in humans

(Cunha-Vaz et al , 1975), and in rats within 1-2 weeks of the onset of streptozotocin-

induced diabetes (Xu et al, 2004; Qaum et al, 2001; Do carmo et al, 1998). Macular

edema secondary to breakdown of the inner BRB is the most common cause of

vision impairment in diabetic retinopathy (Kent et al, 2000). Although increased

retinal vasculature leak can be detected very early in diabetic retinopathy, clinically

apparent macular edema presumably does not develop until local compensatory

mechanisms are overcome.

Hyperpermeability of the retinal vasculature is thought to be caused by local

hypoxia. Endothelial cells by their nature can tolerate large variations in oxygen

tension however signals from endothelial cells can influence the microenvironment,

with potential for an overcompensatory response induced in surrounding tissues

(Faller, 1999). In the retina for example, metabolic changes in the neural

environment are initially detected by macroglia (astrocytes and Müller cells) and

microglial cells, which have a wide array of responses to maintain homeostasis for

neuronal and vascular elements (Rungger-Brandle et al, 2000).

A number of early changes have been described in perivascular elements

associated with macula edema secondary to diabetic retinopathy. These include

increased levels of VEGF protein localised to glial and some vascular elements

(Amin et al, 1997) and upregulation of GFAP in Müller cells in humans with non-

proliferative diabetes (Mizutani et al, 1998). Impairment of the Müller cell

glutamate transporter system in diabetic rat retina has also been reported (Puro,

2002). Müller cells can enhance the barrier properties of retinal blood vessels by

production of factors that contribute to tight junction formation (Igarashi et al, 2000;

Tout et al, 1993). The object of the present study was to establish an in vitro

permeability model of the trophic effect of Müller cells on retinal vascular

endothelial cell barriers, and to subsequently test the hypothesis that hypoxia

abrogates this trophic effect.


81


4.2.1 Cell isolation and primary culture

Retinal Müller cells were isolated from post-mortem bovine eyes as previously

described (refer to Chapter 2.2.1 Section c). Cells were maintained in DMEM

containing 20% heat-inactivated foetal bovine serum supplemented with 2 mmol/L

glutamine, 100 IU/ml penicillin and 100 g/ml streptomycin. Only Müller cells

from an early passage (P1-2) were used for the permeability studies. BRE cells were

isolated as described previously (refer to Chapter 2.2.1 Section a). Primary

endothelial cell colonies grew to sub-confluency after 4 days in culture when they

were moved into T25cm2 flasks and expanded by passaging (P1) prior to use in the

permeability experiment.

4.2.2 Cell characterisation

Purity and identity of early passage BRE cells was determined by fluorescent

activated cell sorter (FACS) analysis (Chapter 2.2.2 Section a).

Endothelial cells on Transwell filters were (post-experimentally) labelled with anti-

ZO-1 as described in Chapter 2.2.2 Section c. Müller cell characterisation is

described in Chapter 2.2.2 Section b.

4.2.3 Co-culture groups

To determine whether interactions occur between endothelial cells and Müller cells,

two co-culture groups and one endothelial cell control group (monocultured cells)

were compared (see preface pages xv-xvi for definition of monocultured and co-

cultured cells). In both of the co-culture groups, Müller cells were situated on the

abluminal aspect of the endothelial cell barrier as might occur in vivo. Any

substance that was generated by Müller cells in the hypoxic environment would

presumably have an effect on endothelial cell barrier function. Groups are described

in Table 4.1.


82

Table 4.1 Co-cultured cells and control groups used in the barrier assay

Group Description Number of wells

number

____________________________________________________________

1 BRE cells and Müller cells on Transwell filter (MCC) 12

2 BRE cells in Müller cell conditioned medium (MCM) 12

3** BRE cells 12

4** Müller cells alone 8

5** Coated filters without cells 4

_____________________________________________________________

Key: ** Control groups; BRE, bovine retinal endothelial cells

1. Control BRE cells

Only endothelial cells from the second passage were used for the permeability

studies. Cells were grown in 24-well (two-chamber, 0.33 cm2) Transwell plates with

0.4 m pore size polycarbonate filters coated with extracellular matrix materials as

previously described (Chapter 3, Section 3.2.2). Cells were seeded to the upper filter

surface of the inner chamber (35,000 cells/well) on Day 0. Medium was replaced in

the luminal (150 l) and abluminal chambers (700 l) on Day 1 after seeding, and

then every second day following.

2. Müller cells co-cultured on the underside of the Transwell filter (MCC group)

Müller cells were seeded to the underside (abluminal surface) of the Transwell filter

insert and allowed to attach for 2 h at 37oC under humidified conditions before the

well was turned right-side up (Chapter 3, Section 3.2.2). Cells became stabilised on

the filter overnight prior to adjusting cells in EC:C6 medium for 24 h. Endothelial

cells were seeded to the pre-coated upper filter surface of the inner well, on Day 0.


83

In this instance, Müller cells were separated from the endothelial cell barrier by a

filter with a membrane thickness of 10m. Preliminary studies were carried out to

determine the appropriate Müller cell numbers cultured on the underside of the

filter, and transmission electron microscopy was used to verify cell growth

characteristics on the Transwell filter surface (Chapter 3).

3. Müller cells providing continually conditioned medium (MCM group)

Müller cells were seeded to the base of the outer chamber of Transwell plates to

study the effects of Müller cell conditioned medium on BRE cell permeability.

Müller cells were pre-conditioned in EC:C6 medium for 24 h prior to co-culture

with endothelial cells. Endothelial cells were seeded to the filter of the inner well as

above, on Day 0.

4.2.4 Permeability studies in normoxic and hypoxic conditions

On Day 5 of culture (by which time control endothelial cell cultures had achieved

stable transendothelial electrical resistance) medium was changed to serum-

lightened (5% HPPS) medium to minimise the effect of extraneous growth factors.

The 12 wells in each group were divided into normoxic or hypoxic treatments (6

wells/group). A mixture of radiolabelled tracers [methoxy-3H] inulin (NEN Life

Science Products Inc, Boston, MA, USA) and [methyl-14C] methylated (bovine

serum) albumin (NEN Life Science Products Inc) was added to medium in the upper

Transwell chamber. The concentration of tracer was predetermined to provide a

sufficient number of counts (5000-50,000 dpm) in the final volume added to the

luminal chamber (total count). Hypoxic conditions were provided by placing

cultures in a modular incubator chamber (Billups-rothenberg Inc., Del Mar, CA)

flushed with 1% oxygen and 5% carbon dioxide in a nitrogen gas mixture (BOC

Gases, Wetherill Park, Aust.) for 5-10 mins. The chamber was sealed and placed

into a humidified 37oC incubator. At 1, 4, 12 and 24 h medium was removed from

the lower chambers and transferred into a plastic tube. 2ml of UG – Ultima Gold

scintillant (Packard Instrument Co., Meriden CT. USA) was added to each tube to

disperse the tracer within the sample. An equal volume of serum-lightened medium

was replaced in the lower wells. Plates were returned to hypoxic conditions after


84

sampling at each timepoint. There were no adverse effects on endothelial cell

morphology for up to 24 h hypoxia and medium pH was unchanged for the duration

of the permeability experiment. After 24 h hypoxia, cultures were replaced in

normoxic conditions. A Tricarb 2100TR liquid scintillationcounter (Packard

Instrument Co., Meriden, CT, USA) using the full spectrum DPM technique

(Operation Manual, Packard Instrument Co.) was used to calculate radioactivity of

the individual radionuclides. A ratio of the radioactivity from each of the lower

wells to the total count was determined for each timepoint and expressed as the

percentage equilibration.

Endothelial cell confluence and viability was verified by TEER

measurements using a Millipore ERS resistance meter as described in Chapter 3.

Electrical resistance measurements were made on Day 3, 5, 6 (24 h), 7 (48 h), and at

1, 4, 12 and 24 h during hypoxic conditions. Resistances were calculated as the

average (mean) resistance of the different groups (raw reading), minus the average

reading from the ‘no cell’ wells, multiplied by the area of the Transwell filter

(0.33cm2). Results were expressed in ohms.cm2.

4.2.5 Confirmation of hypoxic conditions

To confirm the induction of hypoxia using the system described above, lactate

release from primary bovine Müller cells into the medium was measured (Brooks et

al, 1998). Confluent Müller cells in 6-well plates were exposed to normoxic or

hypoxic conditions (as above) in serum-free medium. After 24 h conditioned

medium was collected and lactate concentration was measured with a Cobas Fara

centrifugal batch analyser (Roche Diagnostics Australia P/L, Castle Hill, NSW,

Australia).

4.2.6 Statistics

Results were calculated as 1) mean standard error (SE) electrical resistance

(ohms.cm2) for TEER, and 2) equilibration of tracers, expressed as mean (SE)

percent (%) of total potential equilibration (flux) for the permeability studies. An

unpaired Student’s t-test or repeated measures ANOVA (RANOVA) followed by


85

linear contrast was used to analyse the data. The level of significance was set at p =

0.05.


86

4.3 RESULTS

Cell isolation and characterisation

As observed previously, primary bovine retinal endothelial cells could be seen

growing out from capillary fragments adhered to the tissue culture dishes after one

day in culture (Chapter 2).

Endothelial cells were determined to be >95% pure by FACS analysis displaying

positive immunolabelling for von Willebrand’s Factor VIII compared to the IgG

control (Figure 4.1).

Confirmation of hypoxic conditions

Lactate was increased by 1.6-fold in medium from primary bovine Müller cells

grown under hypoxic conditions after 24 h: 9.8 +/- 0.1 mmol/L (hypoxic conditions)

v. 6.1 +/- 0.3 (normoxic conditions), (Figure 4.2).

Cell confluence and health under normoxic conditions

Inulin flux was restricted by the endothelial cell barrier compared to filters without

cells (Figure 4.3).

The integrity of the endothelial cell barrier and the effect of co-culturing

endothelial cells with Müller cells was followed by serial measurement of TEER.

Between days 3 and 7, TEER of endothelial cells in the MCC group was increased

above that of the control endothelial cell barrier, shown by the significant

interaction (between time and group) term in the RANOVA: 41.0 ohms.cm2 +/- 3.2

MCC group v. 24.4 +/- 3.2 control endothelial cells (P < 0.05; Figure 4.4A). This

trend of increased TEER in the MCC group under normoxic conditions was

associated with a corresponding reduction of inulin flux: 13.3 +/- 2.0% MCC group

v. 18.1 +/- 2.0 control endothelial cells, (RANOVA, P < 0.05; Figure 4.4B).

Müller cells in the MCM group did not induce a corresponding effect on the

endothelial cell barrier: 19.9 ohms.cm2 +/- 2.8 MCM group v. 24.4 +/- 3.2 control

endothelial cells (TEER); 23.9 +/- 1.7% MCM group v. 18.1 +/- 2.0 control

endothelial cells (inulin flux).


87

Permeability of macromolecules under hypoxic conditions

Endothelial cells cultured alone were unaffected by hypoxic conditions with no

significant changes in inulin flux (Figure 4.3) or TEER (not shown) compared with

cells in normoxia at all timepoints up to 24 h.

There was reduced inulin leakage in the MCC group (compared with

endothelial cells alone) up to 12 h of hypoxia: 2.7 +/- 0.5% MCC group v. 4.3 +/-

0.5 control endothelial cells (RANOVA).

Hypoxia-induced permeability changes were first seen in the MCC group at

the 12 h timepoint when there was increased equilibration of both inulin (Figure

4.5A) and albumin (Figure 4.5B). Significant interaction was detected for the two-

way (2x3) ANOVA (P < 0.05). Hence, the Student’s t-test (with a Bonferroni

adjustment) was used to compare the inulin leak within each group under normoxia

and hypoxia. Inulin equilibration in the MCC group at 12 h was 34.7 +/- 5.0%

(hypoxic conditions) v. 15.8 +/- 2.9 (normoxic conditions), (P < 0.05; Figure 4.5A).

Similarly, albumin equilibration in the MCC group at 12 h was 34.5 +/- 4.3%

(hypoxic conditions) v. 9.2 +/- 0.7 (normoxic conditions), (P < 0.05; Figure 4.5B).

By comparison, permeability of cells in the MCM group was similar to the

control endothelial cells up to the 12 h timepoint; however at this time, inulin

equilibration was increased under both normoxic and hypoxic conditions (Figure

4.5A). Although albumin equilibration in the MCM group was increased, the

difference was not statistically significant (Figure 4.5B).


88

4.4 DISCUSSION

An in vitro model was used to confirm the induction of barrier-enhancing properties

in endothelial cells by Müller cells alluded to in a previous in vivo study (Tout et al,

1993). Endothelial cells grown in close association with Müller cells (MCC group)

were compared with those grown separately (MCM group) in order to assess

whether Müller cell-induced effects on endothelial cells required intimate cell

contact or a diffusible factor(s). Earlier morphological studies found no evidence of

direct contact when cells were grown on either side of the Transwell filter (Chapter

3) and this model may therefore replicate the in vivo situation where Müller cells are

a constituent of the glia limitans and do not physically contact endothelial cells. It

was observed elsewhere, that continually conditioned medium from long-term

Müller cell cultures grown in the lower chamber did not increase the TEER of

retinal endothelial cells using same the experimental system (Chapter 5). The

present study extends these previous findings to show that an intimate association of

endothelial cells and Müller cells is necessary to induce the barrier-protective effect

under normoxic conditions and barrier-breakdown under hypoxic conditions. The

findings are consistent with the existence of a reciprocal relationship or two-way

communication between retinal endothelial cells and Müller cells.

Under normoxic conditions, Müller cells closely associated with endothelial

cells (MCC group) improved the endothelial cell barrier function (measured by

TEER and inulin flux) presumably by diffusible factor(s) that regulate tight

junctions involved in paracellular permeability. This barrier-enhancing factor may

be concentration-dependent or short-lived, since Müller cells in the MCM group

(separated from endothelial cells) did not enhance endothelial cell barrier integrity.

Under hypoxic conditions it was found that Müller cells still conferred a ‘protective

effect’ on the endothelial cell barrier for a short period (greater than 4 hours, but less

than 12 hours). Thereafter, Müller cells (MCC group) induced increased leakage of

inulin and albumin (reflecting increases in both paracellular and transcellular

permeability) in retinal vascular endothelial cells, while the permeability of

endothelial cells cultured alone was unaffected by hypoxia. These observations

suggest that Müller cells produce factor(s) that can mediate pro- and anti-barrier

effects on retinal vascular endothelial cells; alternatively under hypoxia, Müller cell


89

production of barrier-enhancing factor(s) may be inhibited. The identity of these

factors remains to be established. This dual effect of the same cells under different

environmental conditions underscores the critical role of Müller cells in regulating

the inner blood-retinal barrier.

Growth factors TGF-and VEGF are increasingly recognised for their

apparently antagonistic effects in both angiogenesis and hyperpermeability of

hypoxic tissues (Eichler et al, 2001; Behzadian et al, 1998; Hata et al, 1995).

However, the use of neutralising antibodies in studies where TGF-and VEGF are

believed to be mediating a permeability effect (Behzadian et al, 2001; Hartnett et al ,

2003, respectively) have only partly abrogated the effect, suggesting that additional

elements and/or co-factors may be involved. The kinetics of this study correspond

with those of others who found exogenously added VEGF-mediated permeability

increases occur in both transcellular (Feng et al, 1999) and paracellular (Behzadian

MA. et al, 2003) pathways of bovine retinal endothelial cells. More research is

required to investigate the mechanisms by which retinal Müller cells can modulate

endothelial cell barrier integrity, including growth factors and signalling pathways.

The present study shows that the BRB can be closely regulated by Müller

cells and the capacity of Müller cells to maintain the integrity of the BRB is

diminished under hypoxic conditions. These results highlight an important role for

Müller cells in the pathogenesis of macular edema which is a leading cause of

blindness.


90

CHAPTER 5

CONDITIONED MEDIUM FROM

MIXED RETINAL PIGMENTED

EPITHELIUM AND MULLER

CELL CULTURES REDUCES

IN VITRO PERMEABILITY

OF RETINAL VASCULAR

ENDOTHELIAL CELLLS


Tretiach M, Madigan MC, Gillies MC. Conditioned medium from mixed retinal

pigmented epithelium and Müller cells reduces in vitro permeability of retinal

vascular endothelial cells Br J Ophthalmol 2004; 88: 957-961


91

5.1 INTRODUCTION

The BRB exists at the level of the retinal capillary endothelium [‘inner’ BRB] and

the retinal pigmented epithelium (RPE) [‘outer’ BRB]. Macular edema secondary to

breakdown of the inner BRB is the most common cause of loss of vision in diabetic

retinopathy (Ferris et al, 1984). Retinal laser therapy reduces the risk of blindness in

eyes with diabetic macular edema (Aiello, 2003). However, laser photocoagulation

which is generally administered late in the course of the disease when vision loss is

imminent, may not always work, and is inherently destructive. Understanding how

retinal laser treatment affects a leaking BRB is important for developing better

treatments of macular edema.

Changes in retinal morphology after laser have been well described in rats

(Pollack and Korte, 1997), rabbits (Roider et al, 1992), monkeys (Wallow, 1984),

and humans (Marshall et al, 1984; Marshall, 1981). Although some laser energy

may directly affect the retinal vessels, it is generally accepted that the major site of

absorption is the RPE and choroid (Marshall et al, 1984). The laser-affected areas of

the photoreceptor outer segments and RPE exhibit signs of necrosis including cell

disruption, vacuolization, and condensation of cytoplasmic proteins within a few

hours after treatment (Roider et al, 1992) to an extent that is commensurate with the

intensity of the burn (Wallow, 1984). Within days, RPE cells migrate across

Bruch’s membrane to fill the lesion with subsequent scar formation (Del Priore et

al, 1989; Smiddy et al, 1986). Müller cells and astrocytes replace the damaged outer

nuclear layer of the retina, interdigitating with the migrated RPE cells (Marshall,

1981; Johnson et al, 1977). Müller cells undergo widespread and long-lasting

changes after photocoagulation including increased expression of GFAP associated

with hypertrophy, migration and scar tissue formation (Humphrey et al, 1997).

Laser photocoagulation may stimulate cells to produce soluble factor(s) that

can restore a leaky BRB. In this study, supernatants from RPE, Müller cells,

pericytes, and control ECV304 cells were examined for their ability to reduce the in

vitro permeability of a BRE cell barrier.


92


5.2.1 Cell isolation and culture

A mixed cell population of post-mortem activated RPE and migrating Müller cells

was isolated from 24-48 h post-mortem bovine eyes using a modification of

Edwards’ method (Edwards, 1982) (Chapter 2 Section e). At the first or second

passage, cells were seeded (15,000 cells per well) into the lower chamber of 24 well

Costar Transwell plates (Corning Inc, Acton, MA, USA) with DMEM containing

10% FBS. Medium was replaced 24 h after seeding, and then twice weekly for 3

months as described by Kaida et al, (2000) for long term RPE cultures. Other cell

types were cultured in 24 well plates as described above. Pericytes and Müller cells

were isolated from bovine retinas as described in Chapter 2 Section b, c

respectively, and BRE cells were isolated using an enzyme digestion technique

followed by filtration to collect the microvessel fragments. (Chapter 2 Section a) A

human bladder carcinoma-derived epithelial cell line (ECV304, European

Collection of Cell Cultures, Salisbury, UK) was included as an epithelial cell control

(Brown et al, 2000; Penfold et al, 2000). Cultures were photographed with Kodak

Ektachrome T160 (Kodak, Rochester, NY, USA) film using a Zeiss Telaval 31

inverted microscope (Carl Zeiss, North Ryde, NSW, Australia).

5.2.2 Immunohistochemistry

Cells from primary cultures were routinely immunostained with a panel of

antibodies (Table 2.1a). Antibodies, except anti-CRALBP, were visualised using

either Alexa 488- or 568-conjugated secondary antibodies. The method for

localisation of CRALBP is described in Chapter 2.2.2 Section b. Images were

captured with Leica DC Viewer Computer Software (Version 3) (Leica

Microsystems Ltd., Heerbrugg, Switzerland) using a Leitz Diaplan light microscope

(Leitz Messtechnik GmbH, Wetzlar, Germany).

5.2.3 Conditioned medium and argon laser studies

Second passage BRE cells (5,000 cells per well) were seeded onto coated 0.4 m

pore size polycarbonate filter (0.33 cm2) inserts of two-chamber Transwell plates.

Inserts were placed into wells containing the long term cultured cells (described


93

above) to study the effects of conditioned medium on BRE cell permeability. The

conditioned medium and control cell groups (7-8 wells per group) are described in

Table 5.1. (Group number 6 was a control group containing BRE cells on the insert

filter, and no cells in the bottom chamber.) Long term cultured cells were adapted in

endothelial cell medium for 24 h before co-culturing with BRE cells. Seven days

later, cells in the lower chamber were lasered as follows. Filter inserts were

removed from the 24 well Transwell plate and returned to the incubator in

humidified dishes. Medium was decanted from cells in the lower wells and the

plates were held sideways in situ on the chin rest of a Coherant argon blue-green

laser. Four wells per group received 70 shots per well with 200 m spot size, pulse

duration 0.1 s, 150 mW. The lowest dose that caused a visible reaction in the RPE

monolayer on phase contrast microscopy had been established previously in dose-

response studies (not shown). After laser treatment, fresh medium was added to

cells in the lower chamber. The Transwell filter inserts containing BRE cells were

replaced into medium that was now conditioned by the ‘lasered’ and ‘unlasered’

cells.

Table 5.1 Conditioned medium and control cell groups used in the barrier assay

Group number Description Number of wells

____________________________________________________________

1 Mixed RPE & Müller cells 8

2 Mixed pericyte & Müller cells 8

3 RPE cells alone 8

4 Müller cells alone 8

5 ECV 304 cells 8

6 ** BRE cells 7

7 ** Coated filters only 2

_____________________________________________________________

Key: ** Control groups; BRE, bovine retinal endothelial cells; RPE, retinal pigmented

epithelium


94

5.2.4 Permeability studies

TEER was measured using a Millipore ERS resistance meter (Millipore, NSW,

Australia) as described in Chapter 3.2.2. TEER was recorded from Day 3 after the

filter inserts containing BRE cells had been added to the cell groups providing

conditioned medium, and then every second day until Day 7, when these groups

were lasered. TEER was measured every 12 hours thereafter. It was prospectively

determined that comparisons between groups should be carried out when control

BRE cells reached peak resistance. Presumably this situation best reflects the in vivo

BRB. The mean resistance of the different groups was calculated by subtracting the

average reading from the ‘no cell’ wells (Group 7) multiplied by the area of the

Transwell filter (0.33 cm2). Results were expressed in ohms.cm2. The experiment

was repeated 3 times. Permeability of radiolabelled macromolecular tracers across

the EC barrier was determined as follows. Within one hour of laser treatment, a

mixture of radiolabelled tracers [methoxy-3H]-inulin and [methyl-14C] methylated

(bovine serum) albumin (see Chapter 4) was added to medium in the upper

Transwell chamber. The concentration of tracer was predetermined to provide a

sufficient number of counts (5000-50 000 dpm) in the final volume added to the

luminal chamber (total count). Medium was removed from the lower chamber at 24,

36 and 48 h. Radioactivity was measured as described in Chapter 4 with a liquid

scintillationcounter.

5.2.5 Statistics

As it was anticipated that the in vitro endothelial cell barrier-enhancing effect

occurred after a delayed response to photocoagulation, the experimental groups

were compared at the later timepoints: between 24-48 h after lasering. Results were

calculated as follows: (1) mean (standard deviation) electrical resistance (ohms.cm2)

for the TEER, and (2) equilibration of tracers was expressed as mean (SD) percent

(%) for the permeability studies. Repeated measures ANOVA followed by linear

contrast was used to analyse the data. Time was treated as the within-subject factor

and group as the between-subject factor. The level of significance was set at p =

0.05.


95

5.3 RESULTS

Cell morphology and immunohistochemistry

All primary cells grew to confluence within one week after seeding into culture

flasks. Cells cultured from the first digest of the eyecups contained a mixed

population of two distinct cell types (Figure 5.1A). RPE and Müller cells were

identified in these cultures, where RPE displayed intense pigmentation and

characteristic cobblestone morphology and Müller cells exhibited long, delicate

radial fibre structures and distinctive varicosities around the cell body. Cells

cultured from the second digest consisted of >90% RPE (Figure 5.1B) as

determined with specific antibody labelling (Figure 5.1C). Müller cells isolated

from retinal tissue were identified by anti-CRALBP (Figure 5.1D) and anti-vimentin

immunolabelling (not shown). Cultured pericytes labelled for -SMA (not shown)

showed distinct actin filaments within large, amorphous cells. Contamination of

primary cultures by neuronal cells and/or astrocytes was estimated to be <1% based

on immunolabelling for NCAM and GFAP (not shown).

Long term cultures

With phase microscopy, long term mixed RPE and Müller cell cultures (P1-2)

grown in 24 well plates appeared as flat, uniform sheets with foci of darkly

pigmented areas (Figure 5.1E). RPE cells alone appeared to grow irregularly with

areas of multilayering (Figure 5.1F). Mixed pericytes and Müller cells grew without

extensive contact in the long term cultures albeit with good coverage of the dish

surface (Figure 5.1G). Many pericytes became detached from the lower well surface

by completion of the experiment, when the upper Transwells were removed. The

predominant population remaining on the lower well surface were Müller cells.

ECV304 cells remained as a stable monolayer with characteristic cobblestone

morphology (Figure 5.1H) for up to 4 months in culture. Lasering of the long term

cultured cells did not appear to change cell morphology dramatically.

Permeability results

TEER of control BRE cells reached 9.0 ohms.cm2 on Day 5 and remained fairly

constant thereafter, peaking at 11.7 ohms.cm2 on Day 8 (thick line in Figure 5.2A).


96

By Day 9, TEER of BRE cells grown in medium conditioned by mixed RPE and

Müller cell cultures (Figure 5.2A) was four-fold that of the control BRE cells and

the other conditioned medium groups (Figure 5.2B). Conditioned medium from

mixed RPE and Müller cell cultures that were laser-treated did not significantly

affect TEER of the BRE cell barrier (not shown).

As there was no difference in results between the lasered and unlasered wells

in each group, the outputs from both groups were combined to improve the

statistical power. Groups in which there was an obvious effect on BRE cell

permeability were analysed using repeated measures ANOVA as described above.

The difference between BRE cells exposed to supernatants from the mixed RPE and

Müller cell group (43.2% +/- 6.5 equilibration) and control BRE cells (59.8% +/-

7.0 equilibration) was significant for inulin (p < 0.05) (Figure 5.2C) and albumin

leakage (15.1% +/- 3.8 v. 31.1% +/- 6.7, p < 0.05) (Figure 5.2D). Conditioned

medium from other groups had no discernible effect on the overlying BRE cells.


97

5.4 DISCUSSION

In this study, the ability of supernatants from a variety of lasered cells to reduce the

permeability of a BRE cell layer using a two-chamber in vitro assay was examined.

Although laser treatment of cells in the lower chamber did not have any effect on

BRE cell permeability, it was observed that conditioned medium from mixed RPE

and Müller cells significantly reduced the cell barrier permeability. This observation

is consistent with the hypothesis that barrier enhancing factor(s) are released from

cells (such as RPE and Müller cells) that occur within the laser scar, which do not

normally interact in vivo.

Various theories about the mechanism(s) of retinal laser therapy have been

proposed (L'Esperance, 1968; Stefansson, 2001). The therapeutic benefit of retinal

laser appears to be an indirect effect related to a secondary tissue response, rather

than to the immediate burn (Marshall et al, 1984). Incompetent retinal vessels regain

patency when the outer BRB is repaired after laser therapy (Roider et al, 1992).

These observations suggest that the laser scar that forms after photocoagulation may

be a source of factors that restore leaking retinal vessels.

RPE and Müller cells react to tissue destruction by assembling at the site in

the immediate and early phases, suggesting that they play an important role in repair

of the outer BRB (Hara et al, 2000). Although the response of RPE cells to injury

appears to be rapid, that of the Müller cells is temporarily delayed.

Patterns of growth factor expression following photocoagulation in normal

pig retinas have been studied to understand how RPE and Müller cells might play a

contributory role in retinal wound healing (Xiao et al, 1999). RPE may orchestrate

the initial response(s) via TGF-which is a chemoattractant for inflammatory cells

and promotes matrix deposition - as well as PDGF, EGF, TGF-and FGF - to

promote the proliferation of RPE and other cells. Thereafter the reparative effects

appear to be mediated by a combination of autocrine and paracrine signals from the

major cell types - including RPE and Müller cells (Xiao et al, 1999).

ELISA studies show that immortalised RPE and Müller cell lines (ARPE and

MIO-M1, respectively) secret large quantities of VEGF protein when cultured

alone, compared with co-cultures of ARPE and MIO-M1 cells that had lower

expression of VEGF. The improved barrier effect may be partly explained by


98

reduction in the production of VEGF protein (personal communication, Li Wen –

unpublished data).

Long term cell cultures were used to control for artefacts that that might be

mistaken for a ‘laser-induced’ response, as it was found that short term cultured

cells appear to have an activated phenotype (unpublished). Other in vitro studies

have used short term RPE cultures to investigate the effect of conditioned medium

on endothelial cell proliferation (Glaser et al, 1987; Yoshimura et al, 1995). The

present study is the first to investigate the effect of diffusible substances from RPE

cells on BRE cell permeability. The short- versus long term (present study) culture

conditions and variations in cell confluency may elicit different factors from RPE.

Endothelial cell responses to RPE grown under different culture conditions are

likely to be variable.

Inulin is a low molecular weight molecule (MW 5,000-5,500). Its movement

across BRE cell layer reflects TEER and is an accepted measure of paracellular

permeability (Milton and Knutson, 1990). Here, the larger albumin molecule (MW

69,000) was more successfully retarded by the BRE cell barrier (see Figures 5.2C

and D). Previously, it was observed that in vitro BRE cells may grow unreliably;

forming multilayers that do not always achieve complete barrier formation (Chapter

3). Nevertheless, the significant degree to which conditioned medium from the

mixed RPE and Müller cells contributed to decreased permeability of the

abovementioned macromolecules provides further evidence of the in vitro plasticity

of the BRE cell layer.

It is well established that the scar formed after laser therapy is comprised

predominately of RPE and Müller cells. In this study it was found that only the

supernatants from mixed RPE and Müller cells significantly decreased leakage for

three measures of BRE cell layer permeability. These results support the suggestion

that a secretory product(s) from the laser induced scar may contribute to tightening

leaky retinal blood vessels. Further work is necessary to identify factor(s) that may

contribute to the barrier-tightening effect.


99

CHAPTER 6

CONCLUSIONS


100

6.1 OVERVIEW

A review of the literature brings together some recent ideas about possible

pathogenic mechanisms that occur in early diabetic retinopathy (Chapter 1). In

particular, the evidence suggests that early degenerative features in the diabetic

retina may result from loss of Müller cell trophic support of neuronal and vascular

elements. At the inner blood-retinal barrier, this may precipitate leucocyte

infiltration into neural tissue and focal breakdown of barrier integrity. Moreover, in

the chronic condition and long before vascular changes become clinically evident,

Müller cell responses may set in motion a series of cellular reactions that continue

even after the original insult is removed. This thesis aimed to investigate the effect

of perivascular cells on retinal endothelial cell permeability so as to better

understand the processes involved in blood-retinal barrier leakage, a feature of early

diabetic retinopathy.

In Chapter 2, the methods for isolation and characterisation of bovine retinal

cells are described. These studies form the basis for the in vitro models of the blood-

retinal barrier used in Chapters 3, 4 and 5. Table 2.2 summarises findings on cell

growth rates and potential contaminating cells for the isolation of specific cell types.

Characterisation of retinal cells should be made with regard to careful selection of

specific antibodies as well as choosing appropriate positive and negative control

cells. any antibodies raised against human proteins are unsuitable for bovine

tissue, however the following proved useful: monoclonal anti-humanSMA

(pericytes), monoclonal anti-human NCAM (neurons), monoclonal anti-swine

vimentin (macroglia) and polyclonal anti-human vWF (endothelial cells). Although

antibodies were raised against proteins from species other than cow, protein

homology was apparently sufficient to be cross-reactive in bovine tissues. A more

detailed investigation of protein sequences may confirm the degree of homology

between bovine and other species.

The ultrastructural features of an in vitro retinal endothelial cell model of

blood-retinal barrier permeability are described in Chapter 3. These morphological

features were compared with transendothelial electrical resistance. The electrical

resistance of confluent endothelial cells grown on small and large pore size

polycarbonate Transwell filters was measured and compared with co-cultures of


101

endothelial cells and Müller cells. Electrical resistance measurements were variable,

with many preparations not achieving a functional barrier. The ultrastructural

features associated with barrier function in vitro were studied by comparing cultures

that exhibited a ‘tight’ or ‘leaky’ barrier when measured immediately prior to

processing for electron microscopy. ‘Leaky’ preparations with low transendothelial

electrical resistance were associated with irregular cell growth when studied

morphologically. It was concluded that parallel light and electron microscopic

studies are important for validation of in vitro models of vascular endothelial

permeability.

An in vitro model of the blood-retinal barrier was used to investigate Müller

cell effects on retinal vascular endothelial cell permeability under normoxic (20%

oxygen) and hypoxic (1% oxygen) conditions in Chapter 4. Second passage bovine

retinal endothelial cells were co-cultured with retinal Müller cells on opposite sides

of a 0.4 m pore size polycarbonate Transwell filter or in medium that was

continually conditioned by Müller cells. Permeability changes were observed for up

to 24 hours of hypoxia by measurement of [3H]-inulin and [14C]-albumin flux across

the endothelial cell barrier. Endothelial cell barrier function was enhanced by co-

culturing with Müller cells under normoxic conditions. Under hypoxic conditions

however, the barrier was significantly impaired after 12 hours of co-culture with

Müller cells. This study showed that the blood-retinal barrier can be closely

regulated by Müller cells, and that the capacity of Müller cells to maintain the

integrity of the barrier is diminished under hypoxic conditions. Müller cells clearly

have an important role in the pathogenesis of macular oedema which is a leading

cause of blindness.

Laser therapy is commonly used in the treatment of macular oedema and in

Chapter 5 the in vitro effect of laser photocoagulation on blood-retinal barrier

permeability was investigated. Retinal endothelial cells were exposed to

supernatants from long term co-cultures of Müller cells and RPE that were argon

laser treated. Endothelial cell permeability was analysed by measurement of

transendothelial electrical resistance and equilibration of [3H]-inulin and [14C]-

albumin across the cell barrier. Laser photocoagulation of retinal cells (including

Müller cells, RPE and pericytes) and control ECV304 cells (an epithelial cell line


102

derived from a human bladder carcinoma) in the lower chamber did not appreciably

improve permeability of endothelial cell monocultures compared with that of

unlasered cells. However, medium that was conditioned by mixed RPE and Müller

cells significantly reduced both inulin and albumin permeability of the endothelial

cell barrier. A 4-fold increase in transendothelial electrical resistance was also seen.

These results are consistent with the hypothesis that the interaction of Müller cells

with RPE that occurs subsequent to laser treatment results in secretion of soluble

factor(s) that reduce the permeability of retinal vascular endothelium. Future studies

to identify these factor(s) may have implications for the clinical treatment of

macular oedema secondary to diabetic retinopathy and other diseases.

6.2 FUTURE DIRECTIONS

Irrespective of the limitations of in vitro models, these studies show that useful

information can be extracted provided the experimental conditions are established

and carefully controlled beforehand. In the case of the permeability assay for

example, Chapter 3 discusses conditions under which the assay will provide the

most useful results: that is when high TEERs are achieved, indicating good barrier

integrity. Future developments to refine in vitro modelling of the blood-retinal

barrier may involve studies of the heterogeneity of retinal endothelial cells. These

studies may determine which arms of the microvascular tree comprise cells that

exhibit blood-retinal barrier characteristics. One approach would be to use

differential filtration of microvessels in the isolation step, followed by comparison

of the barrier characteristics generated by each clone; to more sophisticated analyses

of retinal endothelial cell populations by microarray and proteomic techniques.

Further analysis of endothelial cell markers specific for blood-retinal barrier

characteristics may also be a useful line of investigation. Until more is known about

endothelial cell heterogeneity, extrapolation of results using cells derived from other

tissues (ie. umbilical vein endothelium) may be misleading. Studies on the blood-

retinal barrier are likely to be most informative when only retina-derived endothelial

cells are used.

Investigation of cell-derived factors that influence retinal endothelial cell

barrier integrity may have important applications in diabetic macular oedema and


103

other systemic conditions in which microvessel leakage is a factor. If the barrier-

enhancing factor is stable in conditioned medium - such as that from mixed cultures

of RPE and Müller cells that influenced barrier integrity over some distance

(Chapter 5) biochemical fractionation methods may be appropriate. Barrier-

enhancing factors that act over short distances with closely apposed cells (Chapter

4) act rapidly and are quickly degraded, making analysis difficult. In this situation, it

might be more productive to study gene expression profiles of the cells using

microarray technology.

In this thesis, the blood-retinal barrier was found to be closely regulated by

Müller cells; the capacity of Müller cells to maintain the integrity of the barrier was

also diminished under hypoxic conditions. Müller cells clearly have an important

role in the pathogenesis of macular oedema. Future studies to investigate the

expression of aquaporins and water movement in retinal tissues at the inner and

outer blood-retinal barriers will be of critical importance to understanding

mechanisms of oedema formation.


104

APPENDICES


105

Appendix I

Buffers and enzyme cocktail for BRE cell isolation

Enzyme diluent buffer

Enzyme diluent was made with 50 mM Tris and 5 mM calcium chloride in a total of500 ml double distilled water. pH was adjusted to 7.5 with 1 M HCl.

Collagenase Type 1A (500g/ml) catalogue # C103586 (Roche) # C9891 (Sigma)

A stock solution of 50 mg/ml was made by adding 10 ml of enzyme buffer to 500 mgof Collagenase A enzyme in lyophilised powder form. Stock solution needs to befurther diluted 1 in 100 to give 500g/ml in the final concentration. In a final volumeof 20 ml therefore, a 200 l aliquot of stock solution is required.

Pronase (200g/ml) catalogue # 165921(Roche)

A stock solution of 100 mg/ml was made by adding 1ml of enzyme buffer to 100 mgof pronase enzyme in lyophilised powder form. Stock solution needs to be furtherdiluted 1 in 500 to give 200 g/ml in the final concentration. In a final volume of 20ml therefore, a 40 l aliquot of stock is required.

DNase I (200g/ml) catalogue # C128493 (Roche)

A stock solution of 100 mg/ml was made by adding 1 ml of enzyme buffer to 100 mgof DNase I enzyme in lyophilised powder form. Stock solution needs to be furtherdiluted 1 in 500 to give 200 g/ml in the final concentration. In a final volume of 20ml therefore, a 40 l aliquot of stock is required.

Coating reagents for Transwell filters

100 g fibronectin (ThermoElectron, Melbourne, VIC) catalogue # 21-177-0001V50 g collagen Type IV (BD Australia P/L, North Ryde, NSW) catalogue # 354233350 g laminin (BD Australia P/L) catalogue # CR3542320.1% gelatin Type B (Sigma-Adrich P/L, Sydney, NSW) catalogue # G6269


106

Appendix II

Consumables

Tissue culture medium

Dulbecco’s modified Eagles medium (DMEM) (ThermoElectron, Melbourne, VIC)2 mmol/L glutamine (ThermoElectron)90 g/ml heparin (ThermoElectron)0.2 g/ml insulin (Sigma-Aldrich P/L, Sydney, NSW)2.5 g/ml transferrin (Sigma-Aldrich P/L)5 g/ml ascorbic acid (Sigma-Aldrich P/L)100 IU/ml penicillin/100 g/ml streptomycin (ThermoElectron)

Formulation for tissue culture medium

Cell culture medium for bovine retinal endothelial cells (EC:C6 medium)

DMEM supplemented with 15% pooled HPPS (see below), 20 l/ml bovine retinalextract, 2 mM glutamine, 90 g/ml heparin, 0.2 g/ml insulin, 2.5 g/ml transferrin,5 g/ml ascorbic acid, 100 IU/ml penicillin and 100 g/ml streptomycin.

The above reagents were made up to a final volume of 50 ml. The mixture was thencombined on a 1:1 basis with conditioned medium from rat glioma (C6) cells (seebelow), (American Type Culture Collection, ATCC #CCL-107, Rockville, MD,USA).

Human platelet poor serum (HPPS) for EC:C6 medium

Blood samples from healthy human donors were collected into 8 ml vacuette (serum)tubes (Greiner Labortechnik, Germany), incubated at 37oC for 1 h and centrifuged at2905 g for 10 mins. Serum was poured into 50 ml Falcon tubes, stored overnight at4oC to precipitate platelets and centrifuged again at 2905 g for 10 mins. Platelet poorserum was filter sterilised and stored at -20oC.

Bovine retinal extract (BRE) for EC:C6 medium

Bovine retinas were rinsed in buffered salt solution (BSS) (1 ml per retina) for 3 h atroom temperature. The suspension was centrifuged at 850 g for 5 mins. Supernatantwas filter sterilised and stored at -20oC.

BSS was made by adding 40 mg glucose in 200 ml of PBS (Ca2+ and Mg2+ free) toprovide a final concentration of 0.2 g/L. (pH was adjusted to 7.2)

C6 conditioned medium for EC:C6 mediumRat glioma (C6) cells were grown in 75 cm2 flasks with Hams F12 with 15% horseserum and 2.5% fetal bovine serum, 2 mM glutamine and penicillin/streptomycin.


107

REFERENCES


108

Abbruscato TJ, Davis TP (1999) Combination of hypoxia/aglycemia compromises invitro blood-brain barrier integrity. J Pharmacol Exp Ther, 289, 668-75.

Abe T, Sugano E, Saigo Y, Tamai M (2003) Interleukin-1beta and barrier function ofretinal pigment epithelial cells (ARPE-19): aberrant expression of junctionalcomplex molecules. Invest Ophthalmol Vis Sci, 44, 4097-104.

Abu El-Asrar AM, Desmet S, Meersschaert A, Dralands L, Missotten L, Geboes K(2001) Expression of the inducible isoform of nitric oxide synthase in the retinas ofhuman subjects with diabetes mellitus. Am J Ophthalmol, 132, 551-6.

Abu-El-Asrar AM, Dralands L, Missotten L, Al-Jadaan IA, Geboes K (2004)Expression of apoptosis markers in the retinas of human subjects with diabetes.Invest Ophthalmol Vis Sci, 45, 2760-6.

Adamis AP (2002) Is diabetic retinopathy an inflammatory disease? Br JOphthalmol, 86, 363-5.

Ahmed A, Dunk C, Ahmad S , Khaliq A (2000) Regulation of placental vascularendothelial growth factor (VEGF) and placenta growth factor (PIGF) and soluble Flt-1 by oxygen--a review. Placenta, 21 Suppl A, S16-24.

Aiello LM (2003) Perspectives on diabetic retinopathy. Am J Ophthalmol, 136, 122-35.

Aiello LP (2002) The potential role of PKC beta in diabetic retinopathy and macularedema. Surv Ophthalmol, 47 Suppl 2, S263-9.

Aiello LP, Avery RL, Arrigg PG, Keyt BA, Jampel HD, Shah ST, Pasquale LR,Thieme H, Iwamoto MA, Park JE , et al. (1994a) Vascular endothelial growth factorin ocular fluid of patients with diabetic retinopathy and other retinal disorders. NEngl J Med, 331, 1480-7.

Aiello LP, Robinson GS, Lin YW, Nishio Y, King GL (1994b) Identification ofmultiple genes in bovine retinal pericytes altered by exposure to elevated levels ofglucose by using mRNA differential display. Proc Natl Acad Sci USA, 91, 6231-5.

Aird WC (2003) Endothelial cell heterogeneity. Crit Care Med, 31 Suppl S221-30.

Aizu Y, Oyanagi K, Hu J , Nakagawa H (2002) Degeneration of retinal neuronalprocesses and pigment epithelium in the early stage of the streptozotocin-diabeticrats. Neuropathology, 22, 161-70.

Albelda SM, Sampson PM, Haselton FR, McNiff JM, Mueller SN, Williams SK,Fishman AP, Levine EM (1988) Permeability characteristics of cultured endothelialcell monolayers. J Appl Physiol, 64, 308-22.


109

Alm A (1977) The effect of sympathetic stimulation on blood flow through the uvea,retina and optic nerve in monkeys (Macacca irus). Exp Eye Res, 25, 19-24.

Alm A, Bill A (1973) Ocular and optic nerve blood flow at normal and increasedintraocular pressures in monkeys (Macaca irus): a study with radioactively labelledmicrospheres including flow determinations in brain and some other tissues. Exp EyeRes, 15, 15-29.

Alon T, Hemo I, Itin A, Pe'er J, Stone J, Keshet E (1995) Vascular endothelialgrowth factor acts as a survival factor for newly formed retinal vessels and hasimplications for retinopathy of prematurity. Nat Med, 1, 1024-8.

Amin RH, Frank RN, Kennedy A, Eliott D, Puklin JE , Abrams GW (1997) Vascularendothelial growth factor is present in glial cells of the retina and optic nerve ofhuman subjects with nonproliferative diabetic retinopathy. Invest Ophthalmol VisSci, 38, 36-47.

Ammar RF, Jr., Gutterman DD, Brooks LA, Dellsperger KC (2000) Free radicalsmediate endothelial dysfunction of coronary arterioles in diabetes. Cardiovasc Res,47, 595-601.

Anderson JM, van Itallie CM (1995) Tight junctions and the molecular basis forregulation of paracellular permeability. Am J Physiol, 269 (4 Pt 1), G467-75.

Antonelli-Orlidge A, Saunders KB, Smith SR, D'Amore PA (1989) An activatedform of transforming growth factor beta is produced by cocultures of endothelialcells and pericytes. Proc Natl Acad Sci USA, 86, 4544-8.

Arden GB, Sidman RL, Arap W, Schlingemann RO (2005) Spare the rod and spoilthe eye. Br J Ophthalmol, 89, 764-9.

Armstrong D, Augustin AJ, Spengler R, Al-Jada A, Nickola T, Grus F , Koch F(1998) Detection of vascular endothelial growth factor and tumor necrosis factoralpha in epiretinal membranes of proliferative diabetic retinopathy, proliferativevitreoretinopathy and macular pucker. Ophthalmologica, 212, 410-4.

Arnold DR, Moshayedi P, Schoen TJ, Jones BE, Chader GJ , Waldbillig RJ (1993)Distribution of IGF-I and -II, IGF binding proteins (IGFBPs) and IGFBP mRNA inocular fluids and tissues: potential sites of synthesis of IGFBPs in aqueous andvitreous. Exp Eye Res, 56, 555-65.

Arroyo JG, Ghazvini S, Char DH (1997) An immunocytochemical study of isolatedhuman retinal Muller cells in culture. Graefes Arch Clin Exp Ophthalmol, 235, 411-4.

Arthur FE, Shivers RR, Bowman PD (1987) Astrocyte-mediated induction of tightjunctions in brain capillary endothelium: an efficient in vitro model. Brain Res, 433,155-9.


110

Baker AH, Edwards DR, Murphy G (2002) Metalloproteinase inhibitors: biologicalactions and therapeutic opportunities. J Cell Sci, 115 (Pt19), 3719-27.

Balda MS, Whitney JA, Flores C, Gonzalez S, Cereijido M , Matter K (1996)Functional dissociation of paracellular permeability and transepithelial electricalresistance and disruption of the apical-basolateral intramembrane diffusion barrier byexpression of a mutant tight junction membrane protein. J Cell Biol, 134, 1031-49.

Ballantyne AJ, Loewenstein A (1943) Exudates in diabetic retinopathy. TransOphthalmol Soc UK, 63, 95.

Ban Y, Wilt SD, Rizzolo LJ (2000) Two secreted retinal factors regulate differentstages of development of the outer blood-retinal barrier. Brain Res Dev Brain Res,119, 259-67.

Barber AJ (2003) A new view of diabetic retinopathy: a neurodegenerative disease ofthe eye. Prog Neuropsychopharmacol Biol Psychiatry, 27, 283-90.

Barber AJ, Antonetti DA (2003) Mapping the blood vessels with paracellularpermeability in the retinas of diabetic rats. Invest Ophthalmol Vis Sci, 44, 5410-6.

Barber AJ, Antonetti DA, Gardner TW (2000) Altered expression of retinal occludinand glial fibrillary acidic protein in experimental diabetes. The Penn State RetinaResearch Group. Invest Ophthalmol Vis Sci, 41, 3561-8.

Barber AJ, Lieth E, Khin SA, Antonetti DA, Buchanan AG, Gardner TW (1998)Neural apoptosis in the retina during experimental and human diabetes. Early onsetand effect of insulin. J Clin Invest, 102, 783-91.

Barnett NL, Pow DV (2000) Antisense knockdown of GLAST, a glial glutamatetransporter, compromises retinal function. Invest Ophthalmol Vis Sci, 41, 585-91.

Barnett NL, Pow DV, Bull ND (2001) Differential perturbation of neuronal and glialglutamate transport systems in retinal ischaemia. Neurochem Int, 39, 291-9.

Barouch FC, Miyamoto K, Allport JR, Fujita K, Bursell SE, Aiello LP, LuscinskasFW, Adamis AP (2000) Integrin-mediated neutrophil adhesion and retinalleukostasis in diabetes. Invest Ophthalmol Vis Sci, 41, 1153-8.

Becher B, Antel JP (1996) Comparison of phenotypic and functional properties ofimmediately ex vivo and cultured human adult microglia. Glia, 18, 1-10.

Behzadian MA, Wang XL, Shabrawey M, Caldwell RB (1998) Effects of hypoxia onglial cell expression of angiogenesis-regulating factors VEGF and TGF-beta. Glia,24, 216-25.


111

Behzadian MA, Wang XL, Windsor LJ, Ghaly N, Caldwell RB (2001) TGF-betaincreases retinal endothelial cell permeability by increasing MMP-9: possible role ofglial cells in endothelial barrier function. Invest Ophthalmol Vis Sci, 42, 853-9.

Behzadian MA, Windsor LJ, Ghaly N, Liou G, Tsai NT, Caldwell RB (2003) VEGF-induced paracellular permeability in cultured endothelial cells involves urokinase andits receptor. FASEB J, 17, 752-754.

Betsholtz C (1995) Role of platelet-derived growth factors in mouse development.Int J Dev Biol, 39, 817-25.

Birrell AM, Heffernan SJ, Kirwan P, McLennan S, Gillin AG, Yue DK (2002) Theeffects of aminoguanidine on renal changes in a baboon model of Type 1 diabetes. JDiabetes Complications, 16, 301-9.

Blaauwgeers HG, Holtkamp GM, Rutten H, Witmer AN, Koolwijk P, Partanen TA,Alitalo K, Kroon ME, Kijlstra A, van Hinsbergh VW , Schlingemann RO (1999)Polarized vascular endothelial growth factor secretion by human retinal pigmentepithelium and localization of vascular endothelial growth factor receptors on theinner choriocapillaris. Evidence for a trophic paracrine relation. Am J Pathol, 155,421-8.

Blanks J (2001) Morphology and topography of the retina (Chapter 3). In Retina (3rdedition), vol. 1 (ed. Ryan S), pp. 32-53. St Louis: Mosby.

Bolton SJ, Anthony DC, Perry VH (1998) Loss of the tight junction proteinsoccludin and zonula occludens-1 from cerebral vascular endothelium duringneutrophil-induced blood-brain barrier breakdown in vivo. Neuroscience, 86, 1245-57.

Boycott BB, Hopkins JM, Sperling HG (1986) Cone connections of the horizontalcells of the rhesus monkey's retina. Proc R Soc Lond [Biol], 229, 345-379.

Bridges CD (1976) Vitamin A and the role of the pigment epithelium duringbleaching and regeneration of rhodopsin in the frog eye. Exp Eye Res, 22, 435-55.

Bringmann A, Reichenbach A, Wiedemann P (2004) Pathomechanisms of cystoidmacular edema. Ophthalmic Res, 36, 241-9.

Bringmann A, Uckermann O, Pannicke T, Iandiev I, Wolf A, Kutzera F,Reichenbach A, Wolf S , Weidermann P (2005) Triamcinolone acetate inhibitshypotonic glial cell swelling in the rat retina. In ARVO, vol. 46. Florida.

Brooks SE, Gu X, Kaufmann PM, Marcus DM, Caldwell RB (1998) Modulation ofVEGF production by pH and glucose in retinal Muller cells. Curr Eye Res, 17, 875-82.


112

Brown J, Reading SJ, Jones S, Fitchett CJ, Howl J, Martin A, Longland CL,Michelangeli F, Dubrova YE, Brown CA (2000) Critical evaluation of ECV304 as ahuman endothelial cell model defined by genetic analysis and functional responses: acomparison with the human bladder cancer derived epithelial cell line T24/83. LabInvest, 80.

Brownlee M, Cerami A (1981) The biochemistry of the complications of diabetesmellitus. Annu Rev Biochem, 50, 385-432.

Bui BV, Armitage JA, Tolcos M, Cooper ME, Vingrys AJ (2003) ACE inhibitionsalvages the visual loss caused by diabetes. Diabetologia, 46, 401-8.

Bunt AH, Minckler DS (1977) Foveal sparing. New anatomical evidence for bilateralrepresentation of the central retina. Arch Ophthalmol, 95, 1445-7.

Bunt-Milam AH, Saari JC (1983) Immunocytochemical localization of two retinoid-binding proteins in vertebrate retina. J Cell Biol, 97, 703-12.

Burgos R, Mateo C, Canton A, Hernandez C, Mesa J , Simo R (2000) Vitreous levelsof IGF-I, IGF binding protein 1, and IGF binding protein 3 in proliferative diabeticretinopathy: a case-control study. Diabetes Care, 23, 80-3.

Burke JM, Foster SJ (1984) Culture of adult rabbit retinal glial cells: methods andcellular origin of explant outgrowth. Curr Eye Res, 3, 1169-78.

Bursell SE, Clermont AC, Kinsley BT, Simonson DC, Aiello LM , Wolpert HA(1996) Retinal blood flow changes in patients with insulin-dependent diabetesmellitus and no diabetic retinopathy. Invest Ophthalmol Vis Sci, 37, 886-97.

Bursell SE, Clermont AC, Shiba T , King GL (1992) Evaluating retinal circulationusing video fluorescein angiography in control and diabetic rats. Curr Eye Res, 11,287-95.

Buzney SM, Massicotte SJ, Hetu N , Zetter BR (1983) Retinal vascular endothelialcells and pericytes. Differential growth characteristics in vitro. Invest Ophthalmol VisSci, 24, 470-80.

Castellon R, Hamdi HK, Sacerio I, Aoki AM, Kenney MC, Ljubimov AV (2002)Effects of angiogenic growth factor combinations on retinal endothelial cells. ExpEye Res, 74, 523-35.

Cereijido M, Robbins ES, Dolan WJ, Rotunno CA , Sabatini DD (1978) Polarizedmonolayers formed by epithelial cells on a permeable and translucent support. J CellBiol, 77, 853-80.

Cestelli A, Catania C, D'Agostino S, Di Liegro I, Licata L, Schiera G, Pitarresi GL,Savettieri G, De Caro V, Giandalia G, Giannola LI (2001) Functional feature of a


113

novel model of blood brain barrier: studies on permeation of test compounds. JControl Release, 76, 139-47.

Chakravarthy U, Gardiner TA (1999) Endothelium-derived agents in pericytefunction/dysfunction. Prog Retin Eye Res, 18, 511-27.

Chao TI, Grosche J, Friedrich KJ, Biedermann B, Francke M, Pannicke T, ReicheltW, Wulst M, Muhle C, Pritz-Hohmeier S, Kuhrt H, Faude F, Drommer W, KasperM, Buse E , Reichenbach A (1997) Comparative studies on mammalian Muller(retinal glial) cells. J Neurocytol, 26, 439-54.

Chiarelli F, Santilli F, Mohn A (2000) Role of growth factors in the development ofdiabetic complications. Horm Res, 53, 53-67.

Choi I, Chiu SY (1997) Expression of high-affinity neuronal and glial glutamatetransporters in the rat optic nerve. Glia, 20, 184-92.

Citi S, Cordenonsi M (1998) Tight junction proteins. Biochim Biophys Acta, 1448, 1-11.

Clowes AW, Karnowsky MJ (1977) Suppression by heparin of smooth muscle cellproliferation in injured arteries. Nature, 265, 625-6.

Comuzzie AG, Cole SA, Martin L, Carey KD, Mahaney MC, Blangero J ,VandeBerg JL (2003) The baboon as a nonhuman primate model for the study of thegenetics of obesity. Obes Res, 11, 75-80.

Connolly DT (1991) Vascular permeability factor: a unique regulator of blood vesselfunction. J Cell Biochem, 47, 219-23.

Crabb JW, Goldflam S, Harris SE, Saari JC (1988) Cloning of the cDNAs encodingthe cellular retinaldehyde-binding protein from bovine and human retina andcomparison of the protein structures. J Biol Chem, 263, 18688-92.

Crafoord S, Dafgard Kopp E, Seregard S, Algvere PV (2000) Cellular migration intoneural retina following implantation of melanin granules in the subretinal space.Graefes Arch Clin Exp Ophthalmol, 238, 682-9.

Cucullo L, McAllister MS, Kight K, Krizanac-Bengez L, Marroni M, Mayberg MR,Stanness KA, Janigro D (2002) A new dynamic in vitro model for themultidimensional study of astrocyte-endothelial cell interactions at the blood-brainbarrier. Brain Res, 951, 243-54.

Cunha-Vaz J, Faria de Abreu JR, Campos AJ (1975) Early breakdown of the blood-retinal barrier in diabetes. Br J Ophthalmol, 59, 649-56.

Cunningham LA, Wetzel M, Rosenberg GA (2005) Multiple roles for MMPs andTIMPs in cerebral ischemia. Glia, 50, 329-39.


114

Curcio CA, Allen KA (1990) Topography of ganglion cells in human retina. J CompNeurol, 300, 5-25.

Da T, Verkman AS (2004) Aquaporin-4 gene disruption in mice protects againstimpaired retinal function and cell death after ischemia. Invest Ophthalmol Vis Sci,45, 4477-83.

D'Amore PA (1990) Culture and study of pericytes. In Cell Culture Techniques inHeart and Vessel Research (ed. Piper H), pp. 300-314. Berlin: Springer-Verlag.

Danser AH, Derkx FH, Admiraal PJ, Deinum J, de Jong PT, Schalekamp MA (1994)Angiotensin levels in the eye. Invest Ophthalmol Vis Sci, 35, 1008-18.

Danser AH, van den Dorpel MA, Deinum J, Derkx FH, Franken AA, Peperkamp E,de Jong PT, Schalekamp MA (1989) Renin, prorenin, and immunoreactive renin invitreous fluid from eyes with and without diabetic retinopathy. J Clin EndocrinolMetab, 68, 160-7.

Das A, McGuire PG, Eriqat C, Ober RR, DeJuan E, Jr., Williams GA, McLamore A,Biswas J , Johnson DW (1999) Human diabetic neovascular membranes contain highlevels of urokinase and metalloproteinase enzymes. Invest Ophthalmol Vis Sci, 40,809-13.

Davidson MK, Russ PK, Glick GG, Hoffman LH, Chang MS, Haselton FR (2000)Reduced expression of the adherens junction protein cadherin-5 in a diabetic retina.Am J Ophthalmol, 129, 267-9.

Dehouck MP, Meresse S, Delorme P, Fruchart JC, Cecchelli R (1990) An easier,reproducible, and mass-production method to study the blood-brain barrier in vitro. JNeurochem, 54, 1798-801.

Dejana E (1996) Endothelial adherens junctions: implications in the control ofvascular permeability and angiogenesis. J Clin Invest, 98, 1949-53.

Dejana E (2004) Endothelial cell-cell junctions: happy together. Nat Rev Mol CellBiol, 5, 261-70.

Del Maschio A, Zanetti A, Corada M, Rival Y, Ruco L, Lampugnani MG, Dejana E(1996) Polymorphonuclear leukocyte adhesion triggers the disorganization ofendothelial cell-to-cell adherens junctions. J Cell Biol, 135, 497-510.

Del Priore LV, Glaser BM, Quigley HA, Green WR (1989) Response of pig retinalpigment epithelium to laser photocoagulation in organ culture. Arch Ophthalmol,107, 119-22.

Denker BM, Nigam SK (1998) Molecular structure and assembly of the tightjunction. Am J Physiol, 274, F1-9.


115

Dente CJ, Steffes CP, Speyer C, Tyburski JG (2001) Pericytes augment the capillarybarrier in in vitro cocultures. J Surg Res, 97, 85-91.

Deutsch TA, Read JS, Ernest JT (1982) Effects of oxygen and carbon dioxide on theretinal circulation in man. In ARVO, vol. 22, pp. S195. Florida.

Diaz CM, Penfold PL, Provis JM (1998) Modulation of the resistance of a humanendothelial cell line by human retinal glia. Aust N Z J Ophthalmol, 26 Suppl 1, S62-4.

Diaz-Araya C, Provis JM (1992) Evidence of photoreceptor migration during earlyfoveal development: a quantitative analysis of human fetal retinae. Vis Neurosci, 8,505-14.

Dick II J, Jampol L, Haller J (2001) Macular edema (Chapter 57). In Retina (3rdedition), vol. 2 (ed. Ryan S), pp. 967-981. St Louis: Mosby.

Do carmo A, Ramos P, Reis A, Proenca R, Cunha-vaz JG (1998) Breakdown of theinner and outer blood retinal barrier in streptozotocin-induced diabetes. Exp Eye Res,67, 569-75.

Dodge AB, D'Amore PA (1992) Cell-cell interactions in diabetic angiopathy.Diabetes Care, 15, 1168-80.

Donnelly R, Emslie-Smith AM, Gardner ID, Morris AD (2000) ABC of arterial andvenous disease: vascular complications of diabetes. BMJ, 320, 1062-6.

Dreher Z, Wegner M, Stone J (1988) Muller cell endfeet at the inner surface of theretina: light microscopy. Vis Neurosci, 1, 169-80.

Dvorak HF, Brown LF, Detmar M, Dvorak AM (1995) Vascular permeabilityfactor/vascular endothelial growth factor, microvascular hyperpermeability, andangiogenesis. Am J Pathol, 146, 1029-39.

Dye JF, Leach L, Clark P, Firth JA (2001) Cyclic AMP and acidic fibroblast growthfactor have opposing effects on tight and adherens junctions in microvascularendothelial cells in vitro. Microvasc Res, 62, 94-113.

Eagle RC Jr (1984) Mechanisms of maculopathy. Ophthalmol, 91, 613-25.

Edwards RB (1982) Culture of mammalian retinal pigment epithelium and neuralretina. Methods Enzymol, 81, 39-43.

Egensperger R, Maslim J, Bisti S, Hollander H, Stone J (1996) Fate of DNA fromretinal cells dying during development: uptake by microglia and macroglia (Mullercells). Brain Res Dev Brain Res, 97, 1-8.


116

Eichler W, Yafai Y, Kuhrt H, Grater R, Hoffmann S, Wiedemann P, Reichenbach A(2001) Hypoxia: modulation of endothelial cell proliferation by soluble factorsreleased by retinal cells. Neuroreport, 12, 4103-8.

Eichler W, Yafai Y, Wiedemann P, Reichenbach A (2004) Angiogenesis-relatedfactors derived from retinal glial (Muller) cells in hypoxia. Neuroreport, 15, 1633-7.

Ellis EA, Guberski DL, Somogyi-Mann M, Grant MB (2000) Increased H2O2,vascular endothelial growth factor and receptors in the retina of the BBZ/Wordiabetic rat. Free Radic Biol Med, 28, 91-101.

Elmquist JK, Swanson JJ, Sakaguchi DS, Ross LR, Jacobson CD (1994)Developmental distribution of GFAP and vimentin in the Brazilian opossum brain. JComp Neurol, 344, 283-96.

El-Remessy AB, Behzadian MA, Abou-Mohamed G, Franklin T, Caldwell RW ,Caldwell RB (2003) Experimental diabetes causes breakdown of the blood-retinabarrier by a mechanism involving tyrosine nitration and increases in expression ofvascular endothelial growth factor and urokinase plasminogen activator receptor. AmJ Pathol, 162, 1995-2004.

Enge M, Bjarnegard M, Gerhardt H, Gustafsson E, Kalen M, Asker N, Hammes HP,Shani M, Fassler R, Betsholtz C (2002) Endothelium-specific platelet-derived growthfactor-B ablation mimics diabetic retinopathy. EMBO J, 21, 4307-16.

Engerman RL, Kern TS (1993) Aldose reductase inhibition fails to preventretinopathy in diabetic and galactosemic dogs. Diabetes, 42, 820-5.

Engerman RL, Kern TS (1995) Retinopathy in animal models of diabetes. DiabetesMetab Rev, 11, 109-20.

ETDRS (1991) Early photocoagulation for diabetic retinopathy. ETDRS reportnumber 9. Early Treatment Diabetic Retinopathy Study Research Group.Ophthalmology, 98, 766-85.

Faller DV (1999) Endothelial cell responses to hypoxic stress. Clin Exp PharmacolPhysiol, 26, 74-84.

Famiglietti EV, Stopa EG, McGookin ED, Song P, LeBlanc V, Streeten BW (2003)Immunocytochemical localization of vascular endothelial growth factor in neuronsand glial cells of human retina. Brain Res, 969, 195-204.

Feng Y, Venema VJ, Venema RC, Tsai N, Behzadian MA, Caldwell RB (1999)VEGF-induced permeability increase is mediated by caveolae. Invest Ophthalmol VisSci, 40, 157-67.

Ferrara N, Houck KA, Jakeman LB, Winer J, Leung DW (1991) The vascularendothelial growth factor family of polypeptides. J Cell Biochem, 47, 211-8.


117

Ferrari-Dileo G, Davis EB, Anderson DR (1987) Angiotensin binding sites in bovineand human retinal blood vessels. Invest Ophthalmol Vis Sci, 28, 1747-51.

Ferris FL, 3rd, Fine SL, Hyman L (1984) Age-related macular degeneration andblindness due to neovascular maculopathy. Arch Ophthalmol, 102, 1640-2.

Ffytche TJ, Bulpitt CJ, Kohner EM, Archer D, Dollery CT (1974) Effect of changesin intraocular pressure on the retinal microcirculation. Br J Ophthalmol, 58, 514-22.

Fischer AJ, Reh TA (2003) Potential of Müller glia to become neurogenic retinalprogenitor cells. Glia, 43, 70-6.

Fischer AJ, Reh TA (2000) Identification of a proliferating marginal zone of retinalprogenitors in postnatal chickens. Dev Biol, 220, 197-210.

Fischer S, Wobben M, Kleinstuck J, Renz D, Schaper W (2000) Effect of astroglialcells on hypoxia-induced permeability in PBMEC cells. Am J Physiol Cell Physiol,279, C935-44.

Fisher SK, Stone J, Rex TS, Linberg KA, Lewis GP (2001) Experimental retinaldetachment: a paradigm for understanding the effects of induced photoreceptordegeneration. Prog Brain Res, 131, 679-98.

Fletcher EL, Phipps JA, Wilkinson-Berka JL (2005) Dysfunction of retinal neuronsand glia during diabetes. Clin Exp Optom, 88, 132-45.

Forbes MS, Rennels ML, Nelson E (1977) Ultrastructure of pericytes in mouse heart.Am J Anat, 149, 47-70.

Forrester J, Dick A, McMenamin P, Lee W (1996) Anatomy of the eye (Chapter 1).In The Eye: Basic Sciences in Practice, pp. 1-87. Edinburgh: WB Saunders.

Forsythe JA, Jiang BH, Iyer NV, Agani F, Leung SW, Koos RD, Semenza GL(1996) Activation of vascular endothelial growth factor gene transcription byhypoxia-inducible factor 1. Mol Cell Biol, 16, 4604-13.

Francke M, Pannicke T, Biedermann B, Faude F, Wiedemann P, Reichenbach A,Reichelt W (1997) Loss of inwardly rectifying potassium currents by human retinalglial cells in diseases of the eye. Glia, 20, 210-8.

Frank RN (2004) Diabetic retinopathy. N Engl J Med, 350, 48-58.

Frank RN, Turczyn TJ, Das A (1990) Pericyte coverage of retinal and cerebralcapillaries. Invest Ophthalmol Vis Sci, 31, 999-1007.

Frayser R, Hickham JB (1964) Retinal vascular response to breathing increasedcarbon dioxide and oxygen concentrations. Invest Ophthalmol, 32, 427-431.


118

Fredj-Reygrobellet D, Baudouin C, Negre F, Caruelle JP, Gastaud P, Lapalus P(1991) Acidic FGF and other growth factors in preretinal membranes from patientswith diabetic retinopathy and proliferative vitreoretinopathy. Ophthalmic Res, 23,154-61.

Frigeri A, Gropper MA, Turck CW, Verkman AS (1995) Immunolocalization of themercurial-insensitive water channel and glycerol intrinsic protein in epithelial cellplasma membranes. Proc Natl Acad Sci U S A, 92, 4328-31.

Fruttiger M (2002) Development of the mouse retinal vasculature: angiogenesisversus vasculogenesis. Invest Ophthalmol Vis Sci, 43, 522-7.

Furie MB, Cramer EB, Naprstek BL, Silverstein SC (1984) Cultured endothelial cellmonolayers that restrict the transendothelial passage of macromolecules andelectrical current. J Cell Biol, 98, 1033-41.

Gallego A (1986) Comparative studies of horizontal cells and a note on microglialcells. In Progress in Retinal Research, vol 5 (ed. Osborne N and Chader G) Oxford:Pergamon Press.

Gao H, Hollyfield JG (1992) Aging of the human retina. Differential loss of neuronsand retinal pigment epithelial cells. Invest Ophthalmol Vis Sci, 33, 1-17.

Garcia M, Forster V, Hicks D, Vecino E (2002) Effects of muller glia on cellsurvival and neuritogenesis in adult porcine retina in vitro. Invest Ophthalmol VisSci, 43, 3735-43.

Garcia M, Vecino E (2003) Role of Muller glia in neuroprotection and regenerationin the retina. Histol Histopathol, 18, 1205-18.

Gardner TW, Antonetti DA, Barber AJ, LaNoue KF, Levison SW (2002) Diabeticretinopathy: more than meets the eye. Surv Ophthalmol, 47 Suppl 2, S253-62.

Gardner TW, Lieth E, Khin SA, Barber AJ, Bonsall DJ, Lesher T, Rice K, BrennanWA, Jr. (1997) Astrocytes increase barrier properties and ZO-1 expression in retinalvascular endothelial cells. Invest Ophthalmol Vis Sci, 38, 2423-7.

Gariano RF, Iruela-Arispe ML, Hendrickson AE (1994) Vascular development inprimate retina: comparison of laminar plexus formation in monkey and human.Invest Ophthalmol Vis Sci, 35, 3442-55.

Gavrieli Y, Sherman Y, Ben-Sasson SA (1992) Identification of programmed celldeath in situ via specific labeling of nuclear DNA fragmentation. J Cell Biol, 119,493-501.

Ge S, Song L, Pachter JS (2005) Where is the blood-brain barrier ... really? JNeurosci Res, 79, 421-7.


119

Geiger B, Ayalon O (1992) Cadherins. Annu Rev Cell Biol, 8, 307-32.

Gerhardinger C, Brown LF, Roy S, Mizutani M, Zucker CL, Lorenzi M (1998)Expression of vascular endothelial growth factor in the human retina and innonproliferative diabetic retinopathy. Am J Pathol, 152, 1453-62.

Gerhart DZ, LeVasseur RJ, Broderius MA, Drewes LR (1989) Glucose transporterlocalization in brain using light and electron immunocytochemistry. J Neurosci Res,22, 464-72.

Gerritsen ME (1987) Functional heterogeneity of vascular endothelial cells. BiochemPharmacol, 36, 2701-11.

Ghitescu L, Fixman A, Simionescu M, Simionescu N (1986) Specific binding sitesfor albumin restricted to plasmalemmal vesicles of continuous capillary endothelium:receptor-mediated transcytosis. J Cell Biol, 102, 1304-11.

Giebel SJ, Menicucci G, McGuire PG, Das A (2005) Matrix metalloproteinases inearly diabetic retinopathy and their role in alteration of the blood-retinal barrier. LabInvest, 85, 597-607.

Gillies MC, Su T (1993) High glucose inhibits retinal capillary pericyte contractilityin vitro. Invest Ophthalmol Vis Sci, 34, 3396-401.

Gillies MC, Su T (1995) Interferon-alpha 2b enhances barrier function of bovineretinal microvascular endothelium in vitro. Microvasc Res, 49, 277-88.

Gillies MC, Su T, Naidoo D (1995) Electrical resistance and macromolecularpermeability of retinal capillary endothelial cells in vitro. Curr Eye Res, 14, 435-42.

Gillies MC, Su T, Stayt J, Simpson JM, Naidoo D, Salonikas C (1997) Effect of highglucose on permeability of retinal capillary endothelium in vitro. Invest OphthalmolVis Sci, 38, 635-42.

Glaser BM, Campochiaro PA, Davis JL, Jr., Jerdan JA (1987) Retinal pigmentepithelial cells release inhibitors of neovascularization. Ophthalmology, 94, 780-4.

Glover JP, Jacot JL, Basso MD, Hohman TC , Robison WG, Jr. (2000) Retinalcapillary dilation: early diabetic-like retinopathy in the galactose-fed rat model. JOcul Pharmacol Ther, 16, 167-72.

Gora-Kupilas K, Josko J (2005) The neuroprotective function of vascular endothelialgrowth factor (VEGF). Folia Neuropathol, 43, 31-9.

Gospodarowicz D, Abraham JA, Schilling J (1989) Isolation and characterization ofa vascular endothelial cell mitogen produced by pituitary-derived folliculo stellatecells. Proc Natl Acad Sci U S A, 86, 7311-5.


120

Granit R (1933) Components of the retinal action potential in mammals and theirrelations to the discharge in the optic nerve. J Physiol (Lond), 77, 207-238.

Gu X, Zhang J, Brann DW, Yu FS (2003) Brain and retinal vascular endothelial cellswith extended life span established by ectopic expression of telomerase. InvestOphthalmol Vis Sci, 44, 3219-25.

Guidry C (1996) Isolation and characterization of porcine Muller cells.Myofibroblastic dedifferentiation in culture. Invest Ophthalmol Vis Sci, 37, 740-52.

Guidry C (1997) Tractional force generation by porcine Muller cells. Developmentand differential stimulation by growth factors. Invest Ophthalmol Vis Sci, 38, 456-68.

Guidry C (2005) The role of Muller cells in fibrocontractive retinal disorders. ProgRetin Eye Res, 24, 75-86.

Guidry C, Bradley KM, King JL (2003) Tractional force generation by human mullercells: growth factor responsiveness and integrin receptor involvement. InvestOphthalmol Vis Sci, 44, 1355-63.

Guidry C, Feist R, Morris R, Hardwick CW (2004) Changes in IGF activities inhuman diabetic vitreous. Diabetes, 53, 2428-35.

Guyer DR, Schachat AP, Green WR (2001) The choroid: structural considerations(Chapter 2) In Retina (3rd edition), vol. 1 (ed. Ryan SJ), pp. 21-31. St Louis: Mosby.

Hammes HP, Lin J, Renner O, Shani M, Lundqvist A, Betsholtz C, Brownlee M,Deutsch U (2002) Pericytes and the pathogenesis of diabetic retinopathy. Diabetes,51, 3107-12.

Hammes HP, Federoff HJ, Brownlee M (1995) Nerve growth factor prevents bothneuroretinal programmed cell death and capillary pathology in experimentaldiabetes. Mol Med, 1, 527-34.

Hammes HP, Lin J, Bretzel RG, Brownlee M, Breier G (1998) Upregulation of thevascular endothelial growth factor/vascular endothelial growth factor receptor systemin experimental background diabetic retinopathy of the rat. Diabetes, 47, 401-6.

Hara S, Sakuraba T, Nakazawa M (2000) Morphological changes of retinal pigmentepithelial and glial cells at the site of experimental retinal holes. Graefes Arch ClinExp Ophthalmol, 238, 690-5.

Hardwick C, Feist R, Morris R, White M, Witherspoon D, Angus R, Guidry C(1997) Tractional force generation by porcine Muller cells: stimulation by growthfactors in human vitreous. Invest Ophthalmol Vis Sci, 38, 2053-63.


121

Harris A, Bingaman DP, Ciulla TA, Martin BJ (2001) Retinal and choroidal bloodflow in health and disease (Chapter 5) In Retina (3rd edition) vol 2 (ed. Ryan SJ)pp.68-88 St Louis: Mosby.

Hartnett ME, Lappas A, Darland D, McColm JR, Lovejoy S, D'Amore PA (2003)Retinal pigment epithelium and endothelial cell interaction causes retinal pigmentepithelial barrier dysfunction via a soluble VEGF-dependent mechanism. Exp EyeRes, 77, 593-9.

Hata Y, Nakagawa K, Ishibashi T, Inomata H, Ueno H, Sueishi K (1995) Hypoxia-induced expression of vascular endothelial growth factor by retinal glial cellspromotes in vitro angiogenesis. Virchows Arch, 426, 479-86.

Hauck SM, Suppmann S, Ueffing M (2003) Proteomic profiling of primary retinalMuller glia cells reveals a shift in expression patterns upon adaptation to in vitroconditions. Glia, 44, 251-63.

Haudenschild CC (1984) Morphology of vascular endothelial cells in culture. InBiology of endothelial cells (ed. Jaffe EA), pp. 129-40. (Netherlands): Martinus-Nijhoff.

Hayashi Y, Nomura M, Yamagishi S, Harada S, Yamashita J, Yamamoto H (1997)Induction of various blood-brain barrier properties in non-neural endothelial cells byclose apposition to co-cultured astrocytes. Glia, 19, 13-26.

Hicks D, Courtois Y (1990) The growth and behaviour of rat retinal Muller cells invitro. 1. An improved method for isolation and culture. Exp Eye Res, 51, 119-29.

Higashi S, Clermont AC, Dhir V, Bursell SE (1998) Reversibility of retinal flowabnormalities is disease-duration dependent in diabetic rats. Diabetes, 47, 653-9.

Hirschi KK, D'Amore PA (1996) Pericytes in the microvasculature. Cardiovasc Res,32, 687-98.

Hollander H, Makarov F, Dreher Z, van Driel D, Chan-Ling TL, Stone J (1991)Structure of the macroglia of the retina: sharing and division of labour betweenastrocytes and Muller cells. J Comp Neurol, 313, 587-603.

Holopigian K, Greenstein VC, Seiple W, Hood DC, Carr RE (1997) Evidence forphotoreceptor changes in patients with diabetic retinopathy. Invest Ophthalmol VisSci, 38, 2355-65.

Hosoya K, Kondo T, Tomi M, Takanaga H, Ohtsuki S, Terasaki T (2001) MCT1-mediated transport of L-lactic acid at the inner blood-retinal barrier: a possible routefor delivery of monocarboxylic acid drugs to the retina. Pharm Res, 18, 1669-76.

Hughes S, Yang H, Chan-Ling T (2000) Vascularization of the human fetal retina:roles of vasculogenesis and angiogenesis. Invest Ophthalmol Vis Sci, 41, 1217-28.


122

Hull MA, Hewett PW, Brough JL, Hawkey CJ (1996) Isolation and culture of humangastric endothelial cells. Gastroenterology, 111, 1230-40.

Hulsken J, Behrens J, Birchmeier W (1994) Tumor-suppressor gene products in cellcontacts: the cadherin-APC-armadillo connection. Curr Opin Cell Biol, 6, 711-6.

Humphrey MF, Chu Y, Mann K, Rakoczy P (1997) Retinal GFAP and bFGFexpression after multiple argon laser photocoagulation injuries assessed by bothimmunoreactivity and mRNA levels. Exp Eye Res, 64, 361-9.

Humphrey MF, Constable IJ, Chu Y, Wiffen S (1993) A quantitative study of thelateral spread of Muller cell responses to retinal lesions in the rabbit. J Comp Neurol,334, 545-58.

Huster D, Hjelle OP, Haug FM, Nagelhus EA, Reichelt W, Ottersen OP (1998)Subcellular compartmentation of glutathione and glutathione precursors. A highresolution immunogold analysis of the outer retina of guinea pig. Anat Embryol(Berl), 198, 277-87.

Huster D, Reichenbach A, Reichelt W (2000) The glutathione content of retinalMuller (glial) cells: effect of pathological conditions. Neurochem Int, 36, 461-9.

Hwa V, Oh Y, Rosenfeld RG (1999) The insulin-like growth factor-binding protein(IGFBP) superfamily. Endocr Rev, 20, 761-87.

Hynes RO (1992) Integrins: versatility, modulation, and signaling in cell adhesion.Cell, 69, 11-25.

Ido Y, Kilo C, Williamson JR (1997) Cytosolic NADH/NAD+, free radicals, andvascular dysfunction in early diabetes mellitus. Diabetologica, 40, S115-7.

Igarashi Y, Chiba H, Utsumi H, Miyajima H, Ishizaki T, Gotoh T, Kuwahara K,Tobioka H, Satoh M, Mori M, Sawada N (2000) Expression of receptors for glial cellline-derived neurotrophic factor (GDNF) and neurturin in the inner blood-retinalbarrier of rats. Cell Struct Funct, 25, 237-41.

Igarashi Y, Utsumi H, Chiba H, Yamada-Sasamori Y, Tobioka H, Kamimura Y,Furuuchi K, Kokai Y, Nakagawa T, Mori M , Sawada N (1999) Glial cell line-derived neurotrophic factor induces barrier function of endothelial cells forming theblood-brain barrier. Biochem Biophys Res Commun, 261, 108-12.

Iseki S (1986) DNA strand breaks in rat tissues as detected by in situ nick translation.Exp Cell Res, 167, 311-26.

Iwasaki T, Kanda T, Mizusawa H (1999) Effects of pericytes and various cytokineson integrity of endothelial monolayer originated from blood-nerve barrier: an in vitrostudy. J Med Dent Sci, 46, 31-40.


123

Janzer RC, Raff MC (1987) Astrocytes induce blood-brain barrier properties inendothelial cells. Nature, 325, 253-7.

Jaross N, Ryan P, Newland H (2005) Incidence and progression of diabeticretinopathy in an Aboriginal Australian population: results from the KatherineRegion Diabetic Retinopathy Study (KRDRS). Report no. 2. Clin ExperimentOphthalmol, 33, 26-33.

Jin KL, Mao XO, Greenberg DA (2000) Vascular endothelial growth factor: directneuroprotective effect in in vitro ischemia. Proc Natl Acad Sci U S A, 97, 10242-7.

Jingjing L, Xue Y, Agarwal N, Roque RS (1999) Human Muller cells expressVEGF183, a novel spliced variant of vascular endothelial growth factor. InvestOphthalmol Vis Sci, 40, 752-9.

Johnson NF, McNaught EI, Foulds WS (1977) Effect of photocoagulation on thebarrier function of the pigment epithelium. II. A study by electron microscopy. TransOphthalmol Soc U K, 97, 640-51.

Jonas JB (2005) Intravitreal triamcinolone acetonide for treatment of intraocularoedematous and neovascular diseases. Acta Ophthalmol Scand, 83, 645-63.

Jonas JB, Kreissig I, Degenring R (2005) Intravitreal triamcinolone acetonide fortreatment of intraocular proliferative, exudative, and neovascular diseases. ProgRetin Eye Res, 24, 587-611.

Jones BE, Yong LC (1987) Culture and characterization of bovine mesentericlymphatic endothelium. In Vitro Cell Dev Biol, 23, 698-706.

Jourquin J, Tremblay E, Decanis N, Charton G, Hanessian S, Chollet AM, LeDiguardher T, Khrestchatisky M , Rivera S (2003) Neuronal activity-dependentincrease of net matrix metalloproteinase activity is associated with MMP-9neurotoxicity after kainate. Eur J Neurosci, 18, 1507-17.

Joussen AM, Murata T, Tsujikawa A, Kirchhof B, Bursell SE, Adamis AP (2001)Leukocyte-mediated endothelial cell injury and death in the diabetic retina. Am JPathol, 158, 147-52.

Joussen AM, Poulaki V, Le ML, Koizumi K, Esser C, Janicki H, Schraermeyer U,Kociok N, Fauser S, Kirchhof B, Kern TS , Adamis AP (2004) A central role forinflammation in the pathogenesis of diabetic retinopathy. FASEB J, 18, 1450-2.

Joussen AM, Poulaki V, Mitsiades N, Kirchhof B, Koizumi K, Dohmen S, AdamisAP (2002) Nonsteroidal anti-inflammatory drugs prevent early diabetic retinopathyvia TNF-alpha suppression. FASEB J, 16, 438-40.


124

Juurlink BH (1997) Response of glial cells to ischemia: roles of reactive oxygenspecies and glutathione. Neurosci Biobehav Rev, 21, 151-66.

Kaida M, Cao F, Skumatz CM, Irving PE, Burke JM (2000) Time at confluence forhuman RPE cells: effects on the adherens junction and in vitro wound closure. InvestOphthalmol Vis Sci, 41, 3215-24.

Kameda Y (1996) Immunoelectron microscopic localization of vimentin insustentacular cells of the carotid body and the adrenal medulla of guinea pigs. JHistochem Cytochem, 44, 1439-49.

Kanai M, Raz A, Goodman DS (1968) Retinol-binding protein: the transport proteinfor vitamin A in human plasma. J Clin Invest, 47, 2025-44.

Katsura MK, Mishima HK, Minamoto A, Ishibashi F , Yamashita H (2000) Growthregulation of bovine retinal pericytes by transforming growth factor-beta2 andplasmin. Curr Eye Res, 20, 166-72.

Katsura Y, Okano T, Noritake M, Kosano H, Nishigori H, Kado S , Matsuoka T(1998) Hepatocyte growth factor in vitreous fluid of patients with proliferativediabetic retinopathy and other retinal disorders. Diabetes Care, 21, 1759-63.

Kent D, Vinores SA, Campochiaro PA (2000) Macular oedema: the role of solublemediators. Br J Ophthalmol, 84, 542-5.

Kern TS, Engerman RL (1995) Galactose-induced retinal microangiopathy in rats.Invest Ophthalmol Vis Sci, 36, 490-6.

Kern TS, Engerman RL (1996) Capillary lesions develop in retina rather thancerebral cortex in diabetes and experimental galactosemia. Arch Ophthalmol, 114,306-10.

Kerr JF (2002) History of the events leading to the formation of the apoptosisconcept. Toxicology, 181-182, 471-4.

Kim J (2004) Pericytes and the prevention of diabetic retinopathy. Diabetes Res ClinPract, 66 Suppl 1, S49-51.

King GL, Berman AB, Bonner-Weir S, Carson MP (1987) Regulation of vascularpermeability in cell culture. Diabetes, 36, 1460-7.

King JL, Guidry C (2004) Muller cell production of insulin-like growth factor-binding proteins in vitro: modulation with phenotype and growth factor stimulation.Invest Ophthalmol Vis Sci, 45, 4535-42.

Kofuji P, Ceelen P, Zahs KR, Surbeck LW, Lester HA , Newman EA (2000) Geneticinactivation of an inwardly rectifying potassium channel (Kir4.1 subunit) in mice:phenotypic impact in retina. J Neurosci, 20, 5733-40.


125

Kolb H, Mariani A, Gallego A (1980) A second type of horizontal cell in the monkeyretina. J Comp Neurol, 189, 31-44.

Kolb H, Nelson R, Ahnelt P, Cuenca N (2001) Cellular organization of the vertebrateretina. Prog Brain Res, 131, 3-26.

Kondo T, Hosoya K, Hori S, Tomi M, Ohtsuki S, Takanaga H, Nakashima E, IizasaH, Asashima T, Ueda M, Obinata M , Terasaki T (2003) Establishment ofconditionally immortalized rat retinal pericyte cell lines (TR-rPCT) and theirapplication in a co-culture system using retinal capillary endothelial cell line (TR-iBRB2). Cell Struct Funct, 28, 145-53.

Korte GE, Hageman GS, Pratt DV, Glusman S, Marko M, Ophir A (1992) Changesin Muller cell plasma membrane specializations during subretinal scar formation inthe rabbit. Exp Eye Res, 55, 155-62.

Kumagai AK, Glasgow BJ, Pardridge WM (1994) GLUT1 glucose transporterexpression in the diabetic and nondiabetic human eye. Invest Ophthalmol Vis Sci, 35,2887-94.

Kumagai AK, Vinores SA, Pardridge WM (1996) Pathological upregulation of innerblood-retinal barrier Glut1 glucose transporter expression in diabetes mellitus. BrainRes, 706, 313-7.

Kuwabara H, Kokai Y, Kojima T, Takakuwa R, Mori M, Sawada N (2001) Occludinregulates actin cytoskeleton in endothelial cells. Cell Struct Funct, 26, 109-16.

Lampugnani MG, Dejana E (1997) Interendothelial junctions: structure, signallingand functional roles. Curr Opin Cell Biol, 9, 674-82.

Larsen M, Wang M, Sander B (2005) Overnight thickness variation in diabeticmacular edema. Invest Ophthalmol Vis Sci, 46, 2313-6.

Laterra J, Guerin C, Goldstein GW (1990) Astrocytes induce neural microvascularendothelial cells to form capillary-like structures in vitro. J Cell Physiol, 144, 204-15.

Lee R, Kermani P, Teng KK, Hempstead BL (2001) Regulation of cell survival bysecreted proneurotrophins. Science, 294, 1945-8.

Lee TS, Saltsman KA, Ohashi H, King GL (1989) Activation of protein kinase C byelevation of glucose concentration: proposal for a mechanism in the development ofdiabetic vascular complications. Proc Natl Acad Sci USA, 86, 5141-5.

L'Esperance FA, Jr. (1968) An opthalmic argon laser photocoagulation system:design, construction, and laboratory investigations. Trans Am Ophthalmol Soc, 66,827-904.


126

Leventhal AG, Schall JD, Ault SJ, Provis JM, Vitek DJ (1988) Class-specific celldeath shapes the distribution and pattern of central projection of cat retinal ganglioncells. J Neurosci, 8, 2011-27.

Leventhal AG, Thompson KG, Liu D (1993) Retinal ganglion cells within thefoveola of New World (Saimiri sciureus) and Old World (Macaca fascicularis)monkeys. J Comp Neurol, 338, 242-54.

Lewis GP, Fisher SK (2003) Up-regulation of glial fibrillary acidic protein inresponse to retinal injury: its potential role in glial remodeling and a comparison tovimentin expression. Int Rev Cytol, 230, 263-90.

Li Q, Puro DG (2002) Diabetes-induced dysfunction of the glutamate transporter inretinal Muller cells. Invest Ophthalmol Vis Sci, 43, 3109-16.

Lieth E, Barber AJ, Xu B, Dice C, Ratz MJ, Tanase D, Strother JM (1998) Glialreactivity and impaired glutamate metabolism in short-term experimental diabeticretinopathy. Penn State Retina Research Group. Diabetes, 47, 815-20.

Lieth E, Gardner TW, Barber AJ, Antonetti DA (2000) Retinal neurodegeneration:early pathology in diabetes. Clin Experiment Ophthalmol, 28, 3-8.

Limb GA, Salt TE, Munro PM, Moss SE, Khaw PT (2002) In vitro characterizationof a spontaneously immortalized human Muller cell line (MIO-M1). InvestOphthalmol Vis Sci, 43, 864-9.

Lindahl P, Johansson BR, Leveen P, Betsholtz C (1997) Pericyte loss andmicroaneurysm formation in PDGF-B-deficient mice. Science, 277, 242-5.

Ling EA, Ng YK, Wu CH, Kaur C (2001) Microglia: its development and role as aneuropathology sensor. Prog Brain Res, 132, 61-79.

Ling TL, Stone J (1988) The development of astrocytes in the cat retina: evidence ofmigration from the optic nerve. Brain Res Dev Brain Res, 44, 73-85.

Linsenmeier RA (1986) Effects of light and darkness on oxygen distribution andconsumption in the cat retina. J Gen Physiol, 88, 521-42.

Linser P, Moscona AA (1979) Induction of glutamine synthetase in embryonicneural retina: localization in Muller fibers and dependence on cell interactions. ProcNatl Acad Sci U S A, 76, 6476-80.

Linser P, Moscona AA (1981) Carbonic anhydrase C in the neural retina: transitionfrom generalized to glia-specific cell localization during embryonic development.Proc Natl Acad Sci U S A, 78, 7190-4.


127

Lo CM, Keese CR, Giaever I (1999) Cell-substrate contact: another factor mayinfluence transepithelial electrical resistance of cell layers cultured on permeablefilters. Exp Cell Res, 250, 576-80.

Lum H, Malik AB (1996) Mechanisms of increased endothelial permeability. Can JPhysiol Pharmacol, 74, 787-800.

Madigan MC, Penfold PL, Provis JM, Balind TK, Billson FA (1994) Intermediatefilament expression in human retinal macroglia. Histopathologic changes associatedwith age-related macular degeneration. Retina, 14, 65-74.

Majno G, Joris I (1995) Apoptosis, oncosis, and necrosis. An overview of cell death.Am J Pathol, 146, 3-15.

Mamputu JC, Renier G (2004) Advanced glycation end-products increase monocyteadhesion to retinal endothelial cells through vascular endothelial growth factor-induced ICAM-1 expression: inhibitory effect of antioxidants. J Leukoc Biol, 75,1062-9.

Mancini MA, Frank RN, Keirn RJ, Kennedy A, Khoury JK (1986) Does the retinalpigment epithelium polarize the choriocapillaris? Invest Ophthalmol Vis Sci, 27, 336-45.

Mann I (1964) Geographic Ophthalmology. A Review of the Possibilities. ArchOphthalmol, 72, 632-6.

Marmor MF (1999) Mechanisms of fluid accumulation in retinal edema. DocOphthalmol, 97, 239-49.

Marmor MF, Abdul-Rahim AS, Cohen DS (1980) The effect of metabolic inhibitorson retinal adhesion and subretinal fluid resorption. Invest Ophthalmol Vis Sci, 19,893-903.

Marshall J (1981) Interactions between sensory cells, glial cells and the retinalpigment epithelium and their response to photocoagulation. Dev Ophthalmol, 2, 308-17.

Marshall J (1987) The ageing retina: physiology or pathology. Eye, 1 (Pt 2), 282-95.

Marshall J, Clover G, Rothery S (1984) 1.3 Some new findings on retinal irradiationby krypton and argon lasers. In Documenta Ophthalmologica Proceedings Series 36,vol. 36 (eds Birngruber R, Gabel V-P), pp. 21-37. The Hague: Dr W. JunkPublishers.

Matsuo Y, Kihara T, Ikeda M, Ninomiya M, Onodera H, Kogure K (1995) Role ofneutrophils in radical production during ischemia and reperfusion of the rat brain:effect of neutrophil depletion on extracellular ascorbyl radical formation. J CerebBlood Flow Metab, 15, 941-7.


128

McCarthy KM, Skare IB, Stankewich MC, Furuse M, Tsukita S, Rogers RA, LynchRD, Schneeberger EE (1996) Occludin is a functional component of the tightjunction. J Cell Sci, 109 (Pt 9), 2287-98.

McKechnie NM, Boulton M, Robey HL, Savage FJ, Grierson I (1988) Thecytoskeletal elements of human retinal pigment epithelium: in vitro and in vivo. JCell Sci, 91 (Pt 2), 303-12.

McMillan DE (1997) Development of vascular complications in diabetes. Vasc Med,2, 132-42.

Meyer-Schwickerath G, Fried M (1981) Treatment of diabetic retinopathy withphotocoagulation. How many coagulations have to be performed in the individualcase? Dev Ophthalmol, 2, 265-73.

Meyer-Schwickerath R, Pfeiffer A, Blum WF, Freyberger H, Klein M, Losche C,Rollmann R , Schatz H (1993) Vitreous levels of the insulin-like growth factors I andII, and the insulin-like growth factor binding proteins 2 and 3, increase inneovascular eye disease. Studies in nondiabetic and diabetic subjects. J Clin Invest,92, 2620-5.

Michel CC, Curry FE (1999) Microvascular permeability. Physiol Rev, 79, 703-61.

Michiels C (2003) Endothelial cell functions. J Cell Physiol, 196, 430-43.

Miller JR, Moon RT (1996) Signal transduction through beta-catenin andspecification of cell fate during embryogenesis. Genes Dev, 10, 2527-39.

Miller SS, Hughes BA, Machen TE (1982) Fluid transport across retinal pigmentepithelium is inhibited by cyclic AMP. Proc Natl Acad Sci U S A, 79, 2111-5.

Milton SG, Knutson VP (1990) Comparison of the function of the tight junctions ofendothelial cells and epithelial cells in regulating the movement of electrolytes andmacromolecules across the cell monolayer. J Cell Physiol, 144, 498-504.

Mitic LL, Anderson JM (1998) Molecular architecture of tight junctions. Annu RevPhysiol, 60, 121-42.

Miyamoto K, Khosrof S, Bursell SE, Rohan R, Murata T, Clermont AC, Aiello LP,Ogura Y , Adamis AP (1999) Prevention of leukostasis and vascular leakage instreptozotocin-induced diabetic retinopathy via intercellular adhesion molecule-1inhibition. Proc Natl Acad Sci U S A, 96, 10836-41.

Miyamoto K, Ogura Y (1999) Pathogenetic potential of leukocytes in diabeticretinopathy. Semin Ophthalmol, 14, 233-9.


129

Miyoshi J, Takai Y (2005) Molecular perspective on tight-junction assembly andepithelial polarity. Adv Drug Deliv Rev, 57, 815-55.

Mizutani M, Gerhardinger C, Lorenzi M (1998) Muller cell changes in humandiabetic retinopathy. Diabetes, 47, 445-9.

Morishita R, Nakamura S, Nakamura Y, Aoki M, Moriguchi A, Kida I, Yo Y,Matsumoto K, Nakamura T, Higaki J, Ogihara T (1997) Potential role of anendothelium-specific growth factor, hepatocyte growth factor, on endothelial damagein diabetes. Diabetes, 46, 138-42.

Mudhar HS, Pollock RA, Wang C, Stiles CD, Richardson WD (1993) PDGF and itsreceptors in the developing rodent retina and optic nerve. Development, 118, 539-52.

Murata T, Ishibashi T, Inomata H, Sueishi K (1994) Media conditioned by cocultureof pericytes and endothelial cells under a hypoxic state stimulate in vitroangiogenesis. Ophthalmic Res, 26, 23-31.

Musashi K, Kiryu J, Miyamoto K, Miyahara S, Katsuta H, Tamura H, Hirose F,Yoshimura N (2005) Thrombin inhibitor reduces leukocyte-endothelial cellinteractions and vascular leakage after scatter laser photocoagulation. InvestOphthalmol Vis Sci, 46, 2561-6.

Nagelhus EA, Veruki ML, Torp R, Haug FM, Laake JH, Nielsen S, Agre P, OttersenOP (1998) Aquaporin-4 water channel protein in the rat retina and optic nerve:polarized expression in Muller cells and fibrous astrocytes. J Neurosci, 18, 2506-19.

Nathan C (1992) Nitric oxide as a secretory product of mammalian cells. FASEB J,6, 3051-64.

Nathan C (1997) Inducible nitric oxide synthase: what difference does it make? JClin Invest, 100, 2417-23.

Negi A, Marmor MF (1986) Quantitative estimation of metabolic transport ofsubretinal fluid. Invest Ophthalmol Vis Sci, 27, 1564-8.

Nehls V, Drenckhahn D (1991) Heterogeneity of microvascular pericytes for smoothmuscle type alpha-actin. J Cell Biol, 113, 147-54.

Neufeld AH, Kawai S, Das S, Vora S, Gachie E, Connor JR, Manning PT (2002)Loss of retinal ganglion cells following retinal ischemia: the role of inducible nitricoxide synthase. Exp Eye Res, 75, 521-8.

Neufeld AH (1999) Nitric oxide: a potential mediator of retinal ganglion cell damagein glaucoma. Surv Ophthalmol, 43 Suppl 1, S129-35.

Newman E, Reichenbach A (1996) The Muller cell: a functional element of theretina. Trends Neurosci, 19, 307-12.


130

Nielsen S, Nagelhus EA, Amiry-Moghaddam M, Bourque C, Agre P, Ottersen OP(1997) Specialized membrane domains for water transport in glial cells: high-resolution immunogold cytochemistry of aquaporin-4 in rat brain. J Neurosci, 17,171-80.

Nishimura M, Nakano K, Ushiyama M, Nanbu A, Ohtsuka K, Takahashi H,Yoshimura M (1998) Increased serum concentrations of human hepatocyte growthfactor in proliferative diabetic retinopathy. J Clin Endocrinol Metab, 83, 195-8.

O'Dea K (1991) Westernisation, insulin resistance and diabetes in Australianaborigines. Med J Aust, 155, 258-64.

Ogden TE (1978) Nerve fiber layer astrocytes of the primate retina: morphology,distribution, and density. Invest Ophthalmol Vis Sci, 17, 499-510.

Ohira A, de Juan E, Jr. (1990) Characterization of glial involvement in proliferativediabetic retinopathy. Ophthalmologica, 201, 187-95.

Oku H, Kodama T, Sakagami K, Puro DG (2001) Diabetes-induced disruption of gapjunction pathways within the retinal microvasculature. Invest Ophthalmol Vis Sci, 42,1915-20.

Orlidge A, D'Amore PA (1987) Inhibition of capillary endothelial cell growth bypericytes and smooth muscle cells. J Cell Biol, 105, 1455-62.

Ozerdem U, Monosov E, Stallcup WB (2002) NG2 proteoglycan expression bypericytes in pathological microvasculature. Microvasc Res, 63, 129-34.

Pannicke T, Iandiev I, Uckermann O, Biedermann B, Kutzera F, Wiedemann P,Wolburg H, Reichenbach A, Bringmann A (2004) A potassium channel-linkedmechanism of glial cell swelling in the postischemic retina. Mol Cell Neurosci, 26,493-502.

Pardridge WM (1999) Blood-brain barrier biology and methology. J Neurovirol, 5,556-69.

Pardridge WM, Boado RJ, Farrell CR (1990) Brain-type glucose transporter (GLUT-1) is selectively localized to the blood-brain barrier. Studies with quantitativewestern blotting and in situ hybridization. J Biol Chem, 265, 18035-40.

Patel V, Rassam S, Newsom R, Wiek J, Kohner E (1992) Retinal blood flow indiabetic retinopathy. BMJ, 305, 678-83.

Patil RV, Saito I, Yang X, Wax MB (1997) Expression of aquaporins in the ratocular tissue. Exp Eye Res, 64, 203-9.


131

Pedersen DB, Jensen PK, la Cour M, Kiilgaard JF, Eysteinsson T, Bang K, WeinckeAK, Stefansson E (2005) Carbonic anhydrase inhibition increases retinal oxygentension and dilates retinal vessels. Graefes Arch Clin Exp Ophthalmol, 243, 163-168.

Peifer M (1995) Cell adhesion and signal transduction: the Armadillo connection.Trends Cell Biol, 5, 224-9.

Pekny M (2001) Astrocytic intermediate filaments: lessons from GFAP and vimentinknock-out mice. Prog Brain Res, 132, 23-30.

Pekny M, Pekna M (2004) Astrocytic intermediate filaments in CNS pathologies andregeneration. J Pathol, 204, 428-37.

Penfold PL, Madigan MC, Gillies MC, Provis JM (2001) Immunological andaetiological aspects of macular degeneration. Prog Retin Eye Res, 20, 385-414.

Penfold PL, Madigan MC, Provis JM (1991) Antibodies to human leucocyte antigensindicate subpopulations of microglia in human retina. Vis Neurosci, 7, 383-8.

Penfold PL, Provis JM, Liew SC (1993) Human retinal microglia express phenotypiccharacteristics in common with dendritic antigen-presenting cells. J Neuroimmunol,45, 183-91.

Penfold PL, Wen L, Madigan MC, Gillies MC, King NJ, Provis JM (2000)Triamcinolone acetonide modulates permeability and intercellular adhesionmolecule-1 (ICAM-1) expression of the ECV304 cell line: implications for maculardegeneration. Clin Exp Immunol, 121, 458-65.

Penfold PL, Wen L, Madigan MC, King NJ, Provis JM (2002) Modulation ofpermeability and adhesion molecule expression by human choroidal endothelialcells. Invest Ophthalmol Vis Sci, 43, 3125-30.

Penfold PL, Wong J, van Driel D, Provis JM, Madigan MC (2005) Chapter 2Immunology and Age related Macular Degeneration. In Macular Degeneration (edsPenfold PL, Provis JM), pp. 25-44. Berlin: Springer Press.

Pierce EA, Avery RL, Foley ED, Aiello LP, Smith LE (1995) Vascular endothelialgrowth factor/vascular permeability factor expression in a mouse model of retinalneovascularization. Proc Natl Acad Sci U S A, 92, 905-9.

Pohl U, Kaas J (1994) Interactions of hormones with the vascular endothelium.Effects on the control of vascular tone. Arzneimittelforschung, 44, 459-61.

Pollack A, Korte GE (1997) Repair of retinal pigment epithelium andchoriocapillaries after laser photocoagulation: correlations between scanningelectron, transmission electron and light microscopy. Ophthalmic Res, 29, 393-404.

Polyak S (1941) The Retina. Chicago: University of Chicago Press.


132

Pow DV, Crook DK (1995) Immunocytochemical evidence for the presence of highlevels of reduced glutathione in radial glial cells and horizontal cells in the rabbitretina. Neurosci Lett, 193, 25-8.

Prat A, Biernacki K, Wosik K, Antel JP (2001) Glial cell influence on the humanblood-brain barrier. Glia, 36, 145-55.

Pritchard TC, Alloway KD (1999) Visual system (Chapter 10) In: MedicalNeuroscience, pp.283- Madison:Fence Creek Publishing

Provis JM (2001) Development of the primate retinal vasculature. Prog Retin EyeRes, 20, 799-821.

Provis JM, Diaz CM, Dreher B (1998) Ontogeny of the primate fovea: a central issuein retinal development. Prog Neurobiol, 54, 549-80.

Provis JM, Leech J, Diaz CM, Penfold PL, Stone J, Keshet E (1997) Development ofthe human retinal vasculature: cellular relations and VEGF expression. Exp Eye Res,65, 555-68.

Provis JM, Penfold PL, Edwards AJ, van Driel D (1995) Human retinal microglia:expression of immune markers and relationship to the glia limitans. Glia, 14, 243-56.

Provis JM, van Driel D, Billson FA, Russell P (1985) Development of the humanretina: patterns of cell distribution and redistribution in the ganglion cell layer. JComp Neurol, 233, 429-51.

Pulido-Caballero J, Jimenez-Sampedro F, Echevarria-Aza D, Martinez-Millan L(1994) Postnatal development of vimentin-positive cells in the rabbit superiorcolliculus. J Comp Neurol, 343, 102-12.

Puro DG (2002) Diabetes-induced dysfunction of retinal Muller cells. Trans AmOphthalmol Soc, 100, 339-52.

Puro DG (1995) Growth factors and Müller cells. Prog Retin Eye Res, 15, 89-101.

Qaum T, Xu Q, Joussen AM, Clemens MW, Qin W, Miyamoto K, Hassessian H,Wiegand SJ, Rudge J, Yancopoulos GD, Adamis AP (2001) VEGF-initiated blood-retinal barrier breakdown in early diabetes. Invest Ophthalmol Vis Sci, 42, 2408-13.

Qi JH, Ebrahem Q, Moore N, Murphy G, Claesson-Welsh L, Bond M, Baker A,Anand-Apte B (2003) A novel function for tissue inhibitor of metalloproteinases-3(TIMP3): inhibition of angiogenesis by blockage of VEGF binding to VEGFreceptor-2. Nat Med, 9, 407-15.

Raikou M, McGuire A (2003) The economics of screening and treatment in type 2diabetes mellitus. Pharmacoeconomics, 21, 543-64.


133

Rajah TT, Grammas P (2002) VEGF and VEGF receptor levels in retinal and brain-derived endothelial cells. Biochem Biophys Res Commun, 293, 710-3.

Rajasekaran SA, Hu J, Gopal J, Gallemore R, Ryazantsev S, Bok D, Rajasekaran AK(2003) Na,K-ATPase inhibition alters tight junction structure and permeability inhuman retinal pigment epithelial cells. Am J Physiol Cell Physiol, 284, C1497-507.

Ramanathan T, Morita S, Huang Y, Shirota K, Nishimura T, Zheng X , Hunyor SN(2004) Glucose-insulin-potassium solution improves left ventricular energetics inchronic ovine diabetes. Ann Thorac Surg, 77, 1408-14.

Ramanathan T, Shirota K, Morita S, Nishimura T, Huang Y, Zheng X, Hunyor S(2002) Left ventricular oxygen utilization efficiency is impaired in chronicstreptozotocin-diabetic sheep. Cardiovasc Res, 55, 749-56.

Raub TJ (1996) Signal transduction and glial cell modulation of cultured brainmicrovessel endothelial cell tight junctions. Am J Physiol, 271, C495-503.

Reber F, Gersch U, Funk RW (2003) Blockers of carbonic anhydrase can causeincrease of retinal capillary diameter, decrease of extracellular and increase ofintracellular pH in rat retinal organ culture. Graefes Arch Clin Exp Ophthalmol, 241,140-8.

Reichenbach A, Robinson SR (2005) Ependymoglia and ependymoglia-like cells. InNeuroglia (eds Kettenmann H, Ranson B), pp. 58-84: Oxford.

Risau W (1991) Induction of blood-brain barrier endothelial cell differentiation. AnnNY Acad Sci, 633, 405-19.

Rizzolo LJ, Li ZQ (1993) Diffusible, retinal factors stimulate the barrier properties ofjunctional complexes in the retinal pigment epithelium. J Cell Sci, 106 (Pt 3), 859-67.

Robinson GS, Ju M, Shih SC, Xu X, McMahon G, Caldwell RB, Smith LE (2001)Nonvascular role for VEGF: VEGFR-1, 2 activity is critical for neural retinaldevelopment. Faseb J, 15, 1215-7.

Rohen JW, Castenholz A (1967) [On the centralization of the retina in primates].Folia Primatol (Basel), 5, 92-147.

Roider J, Michaud NA, Flotte TJ, Birngruber R (1992) Response of the retinalpigment epithelium to selective photocoagulation. Arch Ophthalmol, 110, 1786-92.

Roque RS, Caldwell RB, Behzadian MA (1992) Cultured Muller cells have highlevels of epidermal growth factor receptors. Invest Ophthalmol Vis Sci, 33, 2587-95.


134

Rosenbaum DM, Rosenbaum PS, Gupta H, Singh M, Aggarwal A, Hall DH, Roth S ,Kessler JA (1998) The role of the p53 protein in the selective vulnerability of theinner retina to transient ischemia. Invest Ophthalmol Vis Sci, 39, 2132-9.

Roux F, Durieu-Trautmann O, Chaverot N, Claire M, Mailly P, Bourre JM,Strosberg AD , Couraud PO (1994) Regulation of gamma-glutamyl transpeptidaseand alkaline phosphatase activities in immortalized rat brain microvessel endothelialcells. J Cell Physiol, 159, 101-13.

Roy MS, Gunkel RD, Podgor MJ (1986) Color vision defects in early diabeticretinopathy. Arch Ophthalmol, 104, 225-8.

Ruberte J, Ayuso E, Navarro M, Carretero A, Nacher V, Haurigot V, George M,Llombart C, Casellas A, Costa C, Bosch A , Bosch F (2004) Increased ocular levelsof IGF-1 in transgenic mice lead to diabetes-like eye disease. J Clin Invest, 113,1149-57.

Ruggeiro D, Lecomte M, Michoud E, Lagarde M, Wiernsperger N (1997)Involvement of cell-cell interactions in the pathogenesis of diabetic retinopathy.Diabetes Metabol, 23, 30-42.

Rungger-Brandle E, Dosso AA, Leuenberger PM (2000) Glial reactivity, an earlyfeature of diabetic retinopathy. Invest Ophthalmol Vis Sci, 41, 1971-80.

Russ PK, Davidson MK, Hoffman LH, Haselton FR (1998) Partial characterisationof the human retinal endothelial cell tight and adherens junction complexes. InvestOphthalmol Vis Sci, 39, 2479-85.

Saari JC, Huang J, Possin DE, Fariss RN, Leonard J, Garwin GG, Crabb JW, MilamAH (1997) Cellular retinaldehyde-binding protein is expressed by oligodendrocytesin optic nerve and brain. Glia, 21, 259-68.

Sakai H, Tani Y, Shirasawa E, Shirao Y , Kawasaki K (1995) Development ofelectroretinographic alterations in streptozotocin-induced diabetes in rats.Ophthalmic Res, 27, 57-63.

Sakagami K, Kodama T, Puro DG (2001) PDGF-induced coupling of function withmetabolism in microvascular pericytes of the retina. Invest Ophthalmol Vis Sci, 42,1939-44.

Sakamoto T, Ueno H, Goto Y, Oshima Y, Ishibashi T, Inomata H (1998) Avitrectomy improves the transfection efficiency of adenoviral vector-mediated genetransfer to Muller cells. Gene Ther, 5, 1088-97.

Sancho-Tello M, Valles S, Montoliu C, Renau-Piqueras J, Guerri C (1995)Developmental pattern of GFAP and vimentin gene expression in rat brain and inradial glial cultures. Glia, 15, 157-66.


135

Sander B, Larsen M, Moldow B, Lund-Andersen H (2001) Diabetic macular edema:passive and active transport of fluorescein through the blood-retina barrier. InvestOphthalmol Vis Sci, 42, 433-8.

Sandercoe TM, Geller SF, Hendrickson AE, Stone J, Provis JM (2003) VEGFexpression by ganglion cells in central retina before formation of the fovealdepression in monkey retina: evidence of developmental hypoxia. J Comp Neurol,462, 42-54.

Sandercoe TM, Madigan MC, Billson FA, Penfold PL, Provis JM (1999) Astrocyteproliferation during development of the human retinal vasculature. Exp Eye Res, 69,511-23.

Santilli F, Cipollone F, Mezzetti A, Chiarelli F (2004) The role of nitric oxide in thedevelopment of diabetic angiopathy. Hormone Metabol Res, 36, 319-35.

Sato Y, Tsuboi R, Lyons R, Moses H, Rifkin DB (1990) Characterization of theactivation of latent TGF-beta by co-cultures of endothelial cells and pericytes orsmooth muscle cells: a self-regulating system. J Cell Biol, 111, 757-63.

Saunders KB, D'Amore PA (1992) An in vitro model for cell-cell interactions. InVitro Cell Dev Biol, 28A, 521-8.

Schirmacher A, Winters S, Fischer S, Goeke J, Galla HJ, Kullnick U, RingelsteinEB, Stogbauer F (2000) Electromagnetic fields (1.8 GHz) increase the permeabilityto sucrose of the blood-brain barrier in vitro. Bioelectromagnetics, 21, 338-45.

Schlingemann RO, Hofman P, Vrensen GF, Blaauwgeers HG (1999) Increasedexpression of endothelial antigen PAL-E in human diabetic retinopathy correlateswith microvascular leakage. Diabetologia, 42, 596-602.

Schmidt E, Peisch RD (1986) Melanin concentration in normal human retinalpigment epithelium: regional variation and age-related reduction. In: ARVO, 37, pp.S1785. Florida.

Schneeburger EE, Lynch RD (1992) Structure, function, and regulation of cellulartight junctions. Am J Physiol, 262 (6 Pt 1), L647-61.

Schnitzer J (1987) Retinal astrocytes: their restriction to vascularized parts of themammalian retina. Neurosci Lett, 78, 29-34.

Schnitzer J (1988) Astrocytes in the guinea pig, horse, and monkey retina: theiroccurrence coincides with the presence of blood vessels. Glia, 1, 74-89.

Schnitzer JE, Carley WW, Palade GE (1988) Albumin interacts specifically with a60-kDa microvascular endothelial glycoprotein. Proc Natl Acad Sci U S A, 85, 6773-7.


136

Schutte M, Werner P (1998) Redistribution of glutathione in the ischemic rat retina.Neurosci Lett, 246, 53-6.

Semenza GL (1998) Hypoxia-inducible factor 1 and the molecular physiology ofoxygen homeostasis. J Lab Clin Med, 131, 207-14.

Shibuki H, Katai N, Kuroiwa S, Kurokawa T, Yodoi J, Yoshimura N (1998)Protective effect of adult T-cell leukemia-derived factor on retinal ischemia-reperfusion injury in the rat. Invest Ophthalmol Vis Sci, 39, 1470-7.

Shih SC, Ju M, Liu N, Mo JR, Ney JJ, Smith LE (2003a) Transforming growth factorbeta1 induction of vascular endothelial growth factor receptor 1: mechanism ofpericyte-induced vascular survival in vivo. Proc Natl Acad Sci U S A, 100, 15859-64.

Shih SC, Ju M, Liu N, Smith LE (2003b) Selective stimulation of VEGFR-1 preventsoxygen-induced retinal vascular degeneration in retinopathy of prematurity. J ClinInvest, 112, 50-7.

Shima DT, Gougos A, Miller JW, Tolentino M, Robinson G, Adamis AP, D'AmorePA (1996) Cloning and mRNA expression of vascular endothelial growth factor inischemic retinas of Macaca fascicularis. Invest Ophthalmol Vis Sci, 37, 1334-40.

Shull MM, Ormsby I, Kier AB, Pawlowski S, Diebold RJ, Yin M, Allen R, SidmanC, Proetzel G, Calvin D, et al. (1992) Targeted disruption of the mouse transforminggrowth factor-beta 1 gene results in multifocal inflammatory disease. Nature, 359,693-9.

Shweiki D, Itin A, Soffer D, Keshet E (1992) Vascular endothelial growth factorinduced by hypoxia may mediate hypoxia-initiated angiogenesis. Nature, 359, 843-5.

Siflinger-Birnboim A, Schnitzer J, Lum H, Blumenstock FA, Shen CP, Del VecchioPJ, Malik AB (1991) Lectin binding to gp60 decreases specific albumin binding andtransport in pulmonary artery endothelial monolayers. J Cell Physiol, 149, 575-84.

Sims DE (2000) Diversity within pericytes. Clin Exp Pharmacol Physiol, 27, 842-6.

Sivalingam A, Kenney J, Brown GC, Benson WE , Donoso L (1990) Basic fibroblastgrowth factor levels in the vitreous of patients with proliferative diabetic retinopathy.Arch Ophthalmol, 108, 869-72.

Smiddy WE, Fine SL, Quigley HA, Dunkelberger G, Hohman RM , Addicks EM(1986) Cell proliferation after laser photocoagulation in primate retina. Anautoradiographic study. Arch Ophthalmol, 104, 1065-9.

Smith ME (2001) Phagocytic properties of microglia in vitro: implications for a rolein multiple sclerosis and EAE. Microsc Res Tech, 54, 81-94.


137

Smith LEH, Shen W, Perruzzi C, Soker S, Kinose F, Xu X, Robinson G, Driver S,Bischoff J, Zhang B, Schaeffer JM, Senger DR (1999) Regulation of vascularendothelial growth factor-dependent retinal neovascularisation by insulin-like growthfactor-1 receptor. Nature Med, 5, 1390-5.

Song L, Pachter JS (2003) Culture of murine brain microvascular endothelial cellsthat maintain expression and cytoskeletal association of tight junction-associatedproteins. In Vitro Cell Dev Biol Anim, 39, 313-20.

Spranger J, Meyer-Schwickerath R, Klein M, Schatz H, Pfeiffer A (1999) Deficientactivation and different expression of transforming growth factor-beta isoforms inactive proliferative diabetic retinopathy and neovascular eye disease. Exp ClinEndocrinol Diabetes, 107, 21-8.

Staddon JM, Herrenknecht K, Smales C, Rubin LL (1995) Evidence that tyrosinephosphorylation may increase tight junction permeability. J Cell Sci, 108 ( Pt 2),609-19.

Stanness KA, Neumaier JF, Sexton TJ, Grant GA, Emmi A, Maris DO, Janigro D(1999) A new model of the blood--brain barrier: co-culture of neuronal, endothelialand glial cells under dynamic conditions. Neuroreport, 10, 3725-31.

Stanzel BV, Espana EM, Grueterich M, Kawakita T, Parel JM, Tseng SC, Binder S(2005) Amniotic membrane maintains the phenotype of rabbit retinal pigmentepithelial cells in culture. Exp Eye Res, 80, 103-12.

Stefansson E (2001) The therapeutic effects of retinal laser treatment and vitrectomy.A theory based on oxygen and vascular physiology. Acta Ophthalmol Scand, 79,435-40.

Steuer H, Jaworski A, Elger B, Kaussmann M, Keldenich J, Schneider H, Stoll D ,Schlosshauer B (2005) Functional characterization and comparison of the outerblood-retina barrier and the blood-brain barrier. Invest Ophthalmol Vis Sci, 46, 1047-53.

Stevenson BR, Keon BH (1998) The tight junction: morphology to molecules. AnnuRev Cell Dev Biol, 14, 89-109.

Stone J (1983) Topographical organisation of the retina in a monotreme: Australianspiny anteater Tachyglossus aculeatus. Brain Behav Evol, 22, 175-84.

Stone J, Dreher Z (1987) Relationship between astrocytes, ganglion cells andvasculature of the retina. J Comp Neurol, 255, 35-49.

Stone J, Itin A, Alon T, Pe'er J, Gnessin H, Chan-Ling T, Keshet E (1995)Development of retinal vasculature is mediated by hypoxia-induced vascularendothelial growth factor (VEGF) expression by neuroglia. J Neurosci, 15, 4738-47.


138

Stone J, Leicester J, Sherman SM (1973) The naso-temporal division of the monkey'sretina. J Comp Neurol, 150, 333-48.

Storkebaum E, Lambrechts D, Carmeliet P (2004) VEGF: once regarded as a specificangiogenic factor, now implicated in neuroprotection. Bioessays, 26, 943-54.

Strettoi E, Masland RH (1996) The number of unidentified amacrine cells in themammalian retina. Proc Natl Acad Sci U S A, 93, 14906-11.

Strother RM, Thomas TG, Otsyula M, Sanders RA, Watkins JB, 3rd (2001)Characterization of oxidative stress in various tissues of diabetic and galactose-fedrats. Int J Exp Diabetes Res, 2, 211-6.

Su T, Gillies MC (1992) A simple method for the in vitro culture of human retinalcapillary endothelial cells. Invest Ophthalmol Vis Sci, 33, 2809-13.

Su X, Sorenson CM, Sheibani N (2003) Isolation and characterization of murineretinal endothelial cells. Mol Vis, 9, 171-8.

Sugama Y, Tiruppathi C, Offakidevi K, Andersen TT, Fenton JW, 2nd , Malik AB(1992) Thrombin-induced expression of endothelial P-selectin and intercellularadhesion molecule-1: a mechanism for stabilizing neutrophil adhesion. J Cell Biol,119, 935-44.

Sun FY, Guo X (2005) Molecular and cellular mechanisms of neuroprotection byvascular endothelial growth factor. J Neurosci Res, 79, 180-4.

Sutter FK, Simpson JM, Gillies MC (2004) Intravitreal triamcinolone for diabeticmacular edema that persists after laser treatment: three-month efficacy and safetyresults of a prospective, randomized, double-masked, placebo-controlled clinicaltrial. Ophthalmology, 111, 2044-9.

Tan KH, Dobbie MS, Felix RA, Barrand MA, Hurst RD (2001) A comparison of theinduction of immortalized endothelial cell impermeability by astrocytes.Neuroreport, 12, 1329-34.

Tapp RJ, Shaw JE, Harper CA, de Courten MP, Balkau B, McCarty DJ, Taylor HR,Welborn TA , Zimmet PZ (2003) The prevalence of and factors associated withdiabetic retinopathy in the Australian population. Diabetes Care, 26, 1731-7.

Taylor S, Srinivasan B, Wordinger RJ, Roque RS (2003) Glutamate stimulatesneurotrophin expression in cultured Muller cells. Brain Res Mol Brain Res, 111, 189-97.

Terashima H, Suzuki K, Kato K, Sugai N (1996) Membrane-bound carbonicanhydrase activity in the rat corneal endothelium and retina. Jpn J Ophthalmol, 40,142-53.


139

Thieme H, Aiello LP, Takagi H, Ferrara N, King GL (1995) Comparative analysis ofvascular endothelial growth factor receptors on retinal and aortic vascular endothelialcells. Diabetes, 44, 98-103.

Thumann G, Hinton DR (2001) Cell biology of the retinal pigment epithelium(Chapter7). In Retina (3rd edition), vol. 1 (ed. Ryan SJ), pp. 104-121. St Louis:Mosby.

Tilling T, Korte D, Hoheisel D, Galla HJ (1998) Basement membrane proteinsinfluence brain capillary endothelial barrier function in vitro. J Neurochem, 71,1151-7.

Tomi M, Hosoya K (2004) Application of magnetically isolated rat retinal vascularendothelial cells for the determination of transporter gene expression levels at theinner blood-retinal barrier. J Neurochem, 91, 1244-8.

Tornquist P, Alm A, Bill A (1990) Permeability of ocular vessels and transportacross the blood-retinal-barrier. Eye, 4 (Pt2), 303-9.

Tout S, Chan-Ling T, Hollander H, Stone J (1993) The role of Muller cells in theformation of the blood-retinal barrier. Neuroscience, 55, 291-301.

Tretiach M, Gillies MC (2001) Effect of laser treatment of RPE and retinal glial cellsin vitro on the permeability of retinal vascular endothelial cell monolayers. In:ARVO, 42, pp. S207. Florida.

Tso MO (1982) Pathology of cystoid macular edema. Ophthalmology, 89, 902-15.

Tsukita S, Furuse M (2000) The structure and function of claudins, cell adhesionmolecules at tight junctions. Ann N Y Acad Sci, 915, 129-35.

Uchihori Y, Puro DG (1993) Glutamate as a neuron-to-glial signal for mitogenesis:role of glial N-methyl-D-aspartate receptors. Brain Res, 613, 212-20.

Uckermann O, Uhlmann S, Pannicke T, Francke M, Gamsalijew R, Makarov F,Ulbricht E, Wiedemann P, Reichenbach A, Osborne NN , Bringmann A (2005)Ischemia-reperfusion causes exudative detachment of the rabbit retina. InvestOphthalmol Vis Sci, 46, 2592-600.

Valiron O, Chevrier V, Usson Y, Breviario F, Job D , Dejana E (1996) Desmoplakinexpression and organization at human umbilical vein endothelial cell-to-celljunctions. J Cell Sci, 109 (Pt 8), 2141-9.

van Buul-Wortelboer MF, Brinkman HJ, Dingemans KP, de Groot PG, van AkenWG, van Mourik JA (1986) Reconstitution of the vascular wall in vitro. A novelmodel to study interactions between endothelial and smooth muscle cells. Exp CellRes, 162, 151-8.


140

van Driel D, Provis JM, Billson FA (1990) Early differentiation of ganglion,amacrine, bipolar, and Muller cells in the developing fovea of human retina. J CompNeurol, 291, 203-19.

van Reyk DM, Gillies MC, Davies MJ (2003) The retina: oxidative stress anddiabetes. Redox Report, 8, 187-92.

Verkman AS (2002) Aquaporin water channels and endothelial cell function. J Anat,200, 617-27.

Walcott JC, Provis JM (2003) Muller cells express the neuronal progenitor cellmarker nestin in both differentiated and undifferentiated human foetal retina. ClinExperiment Ophthalmol, 31, 246-9.

Wallow IH (1984) Repair of the pigment epithelial barrier followingphotocoagulation. Arch Ophthalmol, 102, 126-35.

Wang P, Verin AD, Birukova A, Gilbert-McClain LI, Jacobs K, Garcia JG (2001)Mechanisms of sodium fluoride-induced endothelial cell barrier dysfunction: role ofMLC phosphorylation. Am J Physiol Lung Cell Mol Physiol, 281, L1472-83.

Ward MM, Jobling AI, Kalloniatis M, Fletcher EL (2005) Glutamate uptake inretinal glial cells during diabetes. Diabetologia, 48, 351-60.

Wassle H, Grunert U, Rohrenbeck J, Boycott BB (1989) Cortical magnificationfactor and the ganglion cell density of the primate retina. Nature, 341, 643-6.

Wassle H, Grunert U, Rohrenbeck J, Boycott BB (1990) Retinal ganglion celldensity and cortical magnification factor in the primate. Vision Res, 30, 1897-911.

Wassle H, Boycott BB (1991) Functional architecture of the mammalian retina.Physiol Rev, 71, 447-80.

Watanabe T, Raff MC (1988) Retinal astrocytes are immigrants from the optic nerve.Nature, 332, 834-7.

Weale RA (1966) Why does the human retina possess a fovea? Nature, 212, 255-6.

Weber E, Hammerle H, Vatti R, Berti G, Betz E (1986) Co-cultivation of endothelialand smooth muscle cells on opposite sides of a porous membrane. Appl Pathol, 4,246-52.

Wetzel M, Rosenberg GA, Cunningham LA (2003) Tissue inhibitor ofmetalloproteinases-3 and matrix metalloproteinase-3 regulate neuronal sensitivity todoxorubicin-induced apoptosis. Eur J Neurosci, 18, 1050-60.


141

Wijsman JH, Jonker RR, Keijzer R, van de Velde CJ, Cornelisse CJ, vanDierendonck JH (1993) A new method to detect apoptosis in paraffin sections: in situend-labeling of fragmented DNA. J Histochem Cytochem, 41, 7-12.

Wilkinson-Berka JL, Kelly DJ, Gilbert RE (2001) The interaction between the renin-angiotensin system and vascular endothelial growth factor in the pathogenesis ofretinal neovascularisation in diabetes. J Vasc Res, 38, 527-35.

Williamson JR, Chang K, Frangos M, Hasan KS, Ido Y, Kawamura T, NyengaardJR, van den Enden M, Kilo C, Tilton RG (1993) Hyperglycemic pseudohypoxia anddiabetic complications. Diabetes, 42, 801-13.

Winkler J, Hagelstein S, Rohde M, Laqua H (2002) Cellular and cytoskeletaldynamics within organ cultures of porcine neuroretina. Exp Eye Res, 74, 777-88.

Witmer AN, Vrensen GFJM, van Noorden CJF, Schlingemann RO (2003) Vascularendothelial growth factors and angiogenesis in eye disease. Prog Retin Eye Res, 22,1-29.

Wolburg H, Neuhaus J, Kniesel U, Krauss B, Schmid EM, Ocalan M, Farrell C,Risau W (1994) Modulation of tight junction structure in blood-brain barrierendothelial cells. Effects of tissue culture, second messengers and coculturedastrocytes. J Cell Sci, 107 (Pt 5), 1347-57.

Wolburg H, Reichelt W, Stolzenburg JU, Richter W, Reichenbach A (1990) Rabbitretinal Muller cells in cell culture show gap and tight junctions which they do notexpress in situ. Neurosci Lett, 111, 58-63.

Wolfensberger TJ (1999) The historical discovery of macular edema. DocOphthalmol, 97, 207-16.

Wolin LR, Massopust LC, Jr. (1967) Characteristics of the ocular fundus in primates.J Anat, 101, 693-9.

Wolter JR (1960) Nerves of the normal human choroid. Arch Ophthalmol, 64, 120-4.

Wong HC, Boulton M, Marshall J, Clark P (1987) Growth of retinal capillaryendothelia using pericyte conditioned medium. Invest Ophthalmol Vis Sci, 28, 1767-75.

Wong HC, Elts SM, Phillips JW, Williams KA (1992) Differential growth of brainand retinal bovine pericytes. Diabetologia, 35, 818-27.

Wu KH, Madigan MC, Billson FA, Penfold PL (2003) Differential expression ofGFAP in early v late AMD: a quantitative analysis. Br J Ophthalmol, 87, 1159-66.


142

Xiao M, McLeod D, Cranley J, Williams G, Boulton M (1999) Growth factorstaining patterns in the pig retina following retinal laser photocoagulation. Br JOphthalmol, 83, 728-36.

Xu H, Dawson R, Crane IJ, Liversidge J (2005) Leukocyte diapedesis in vivoinduces transient loss of tight junction protein at the blood-retina barrier. InvestOphthalmol Vis Sci, 46, 2487-94.

Xu X, Zhu Q, Xia X, Zhang S, Gu Q, Luo D (2004) Blood-retinal barrier breakdowninduced by activation of protein kinase C via vascular endothelial growth factor instreptozotocin-induced diabetic rats. Curr Eye Res, 28, 251-6.

Yafai Y, Iandiev I, Wiedemann P, Reichenbach A, Eichler W (2004) Retinalendothelial angiogenic activity: effects of hypoxia and glial (Muller) cells.Microcirculation, 11, 577-86.

Yan Q, Vernon RB, Hendrickson AE, Sage EH (1996) Primary culture andcharacterization of microvascular endothelial cells from Macaca monkey retina.Invest Ophthalmol Vis Sci, 37, 2185-94.

Yanoff M, Fine BS, Brucker AJ, Eagle RC Jr (1984) Pathology of human cystoidmacular edema. Surv Ophthalmol, 28 Suppl, 505-11.

Yasuhara T, Shingo T, Muraoka K, Kameda M, Agari T, Wen Ji Y, Hayase H,Hamada H, Borlongan CV, Date I (2005) Neurorescue effects of VEGF on a ratmodel of Parkinson's disease. Brain Res, 1053, 10-8.

Yong VW, Krekoski CA, Forsyth PA, Bell R, Edwards DR (1998) Matrixmetalloproteinases and diseases of the CNS. Trends Neurosci, 21, 75-80.

Yoshimura N, Matsumoto M, Shimizu H, Mandai M, Hata Y, Ishibashi T (1995)Photocoagulated human retinal pigment epithelial cells produce an inhibitor ofvascular endothelial cell proliferation. Invest Ophthalmol Vis Sci, 36, 1686-91.

Young RW (1976) Visual cells and the concept of renewal. Invest Ophthalmol VisSci, 15, 700-25.

Yu DY, Cringle SJ (2001) Oxygen distribution and consumption within the retina invascularised and avascular retinas and in animal models of retinal disease. ProgRetin Eye Res, 20, 175-208.

Yuodelis C, Hendrickson A (1986) A qualitative and quantitative analysis of thehuman fovea during development. Vision Res, 26, 847-55.

Yurco P, Cameron DA (2005) Responses of Muller glia to retinal injury in adultzebrafish. Vision Res, 45, 991-1002.


143

Zech JC, Pouvreau I, Cotinet A, Goureau O, Le Varlet B, de Kozak Y (1998) Effectof cytokines and nitric oxide on tight junctions in cultured rat retinal pigmentepithelium. Invest Ophthalmol Vis Sci, 39, 1600-8.

Zeng XX, Ng YK, Ling EA (2000) Neuronal and microglial response in the retina ofstreptozotocin-induced diabetic rats. Vis Neurosci, 17, 463-71.

Zhang SX, Ma JX, Sima J, Chen Y, Hu MS, Ottlecz A, Lambrou GN (2005) Geneticdifference in susceptibility to the blood-retina barrier breakdown in diabetes andoxygen-induced retinopathy. Am J Pathol, 166, 313-21.

Zimmet P, Alberti KG, Shaw J (2001) Global and societal implications of thediabetes epidemic. Nature, 414, 782-7.

literature review & thesis aims

Documents