INTRODUCTION

Extracellular matrix (ECM) plays important roles in develop-ment not only by providing structural support to cells but alsoby influencing such cellular processes as cell proliferation, celldifferentiation, cell adhesion and cell migration (see review byHay, 1991). In order to study the influences of ECM on cellmigration under in vivo conditions, we have utilized the freshwater invertebrate,

Hydra vulgaris as a model system foranalysis. Hydra was chosen because of its simple bodystructure and its high capacity for regeneration. Structurally,hydra is composed of a gastric tube with a head at one poleand a foot process at the other pole. Its entire body wall isformed by an epithelial bilayer with an intervening ECMtermed the mesoglea (Campbell and Bode, 1983). Hydramesoglea is known to contain the major ECM components (i.e.fibronectin, laminin, type IV collagen and heparan sulfate pro-teoglycan) found in vertebrate and more complex invertebratespecies (Sarras et al., 1991a). Recent functional studies havedemonstrated that these ECM components play an importantrole in hydra head regeneration (Sarras et al., 1991b) and inhydra cell aggregate development (Sarras et al., 1993).

The regional cell differentiation pattern of hydra isdependent on migration of interstitial cells (I-cells). Under in

vitro conditions, Day and Lenhoff (1981) have demonstratedthat hydra cells attach to and spread on isolated mesoglea.More recently, Agosti and Stidwill (1991) have demonstratedthat hydra nematocytes attach to and migrate on substratacoated with isolated mesoglea or coated with purified ECMcomponents such as type IV collagen and laminin. Hydranematocytes have also been shown to bind to fibronectin in aRGD-dependent manner (Ziegler and Stidwill, 1992). Previousin vivo studies by Campbell and Marcum (1980) also indicatedthat nematocytes migrate between ectodermal epitheliomuscu-lar cells via cell-cell contact guidance mechanisms. While allprevious in vivo studies have indicated that cell migration inhydra depends on cell-cell interactions (Campbell andMarcum, 1980) and chemotaxic gradients (Teragawa andBode, 1991), we propose that cell-ECM interactions are alsocritical to this process.

In order to determine if hydra cell-ECM interactions dooccur in situ, we have developed an in vivo bioassay thatcombines a number of previously published procedures. Oneprocedure involves the use of hydra grafting techniques thatallow quantification of the migration of I-cells from a donorhydra (basal half of graft) to a host hydra (apical half of graft)(Teragawa and Bode, 1990, 1991). This technique was thencombined with two other procedures. In one case, hydra grafts

425Development 120, 425-432 (1994)Printed in Great Britain © The Company of Biologists Limited 1994

Interstitial cell (I-cell) migration in hydra is essential forestablishment of the regional cell differentiation pattern inthe organism. All previous in vivo studies have indicatedthat cell migration in hydra is a result of cell-cell interac-tions and chemotaxic gradients. Recently, in vitro celladhesion studies indicated that isolated nematocytes couldbind to substrata coated with isolated hydra mesoglea,fibronectin and type IV collagen. Under these conditions,nematocytes could be observed to migrate on some of theseextracellular matrix components. By modifying previouslydescribed hydra grafting techniques, two procedures weredeveloped to test specifically the role of extracellular matrixcomponents during in vivo I-cell migration in hydra. In oneapproach, the extracellular matrix structure of the apicalhalf of the hydra graft was perturbed using

β-aminopropi-onitrile and β-xyloside. In the second approach, grafts were

treated with fibronectin, RGDS synthetic peptide andantibody to fibronectin after grafting was performed. Inboth cases, I-cell migration from the basal half to the apicalhalf of the grafts was quantitatively analyzed. Statisticalanalysis indicated that β-aminopropionitrile, fibronectin,RGDS synthetic peptide and antibody to fibronectin allwere inhibitory to I-cell migration as compared to theirrespective controls.

β-xyloside treatment had no effect on interstitial cellmigration. These results indicate the potential importanceof cell-extracellular matrix interactions during in vivo I-cell migration in hydra.

Key words: Hydra, mesoglea, extracellular matrix, fibronectin,collagen, proteoglycans, cell migration

SUMMARY

Cell-extracellular matrix interactions under in vivo conditions during

interstitial cell migration in

Hydra vulgaris

Xiaoming Zhang and Michael P. Sarras Jr*

Department of Anatomy and Cell Biology, University of Kansas Medical Center, 3901 Rainbow Blvd, Kansas City, Kansas 66160-7400, USA

*Author for correspondence

426

were made using animals in which the structural integrity ofthe mesoglea was perturbed using drugs that interfere with thesynthesis and processing of matrix components. In the secondcase, procedures were employed that allow one to introducemacromolecules between the epithelium and mesogleautilizing a dimethylsulfoxide (DMSO)-loading procedure(Fraser et al., 1987). Passage of macromolecules betweenepithelial cells in hydra is normally prevented by septatejunctions (Wood and Kuda, 1980), but low levels of DMSOhave been shown to temporarily open these junctions (HansBode, personal communication). We utilized this DMSOloading procedure to introduce macromolecules into hydragrafts that could potentially interfere with normal cell-ECMinteractions (e.g. ECM components, synthetic peptide, or anti-bodies to ECM components). Using these approaches, thecurrent study therefore focused on the effect of alterations inECM structure and the role of fibronectin on in vivo I-cellmigration in hydra.

MATERIALS AND METHODS

Culture of animalsHydra vulgaris (previously named Hydra attenuata) were used in allexperiments. Animals were cultured in hydra medium (HM) as pre-viously described by Sarras et al. (1991a).

Depletion of I-cells in hydra using hydroxyurea and use ofdrugs to alter mesoglea structureTo deplete hydra of I-cells, similar size hydra polyps (about 30 pergroup) were incubated in HM containing 0.01 M hydroxyurea (HU;Sigma Chemical Company, St Louis, MO) for 5 days with repeatedsolution changes. Animals were then transferred into fresh HM(without HU) for 2 days to allow recovery of the polyps from HUtreatment prior to use. Under these conditions, the I-cell population isreduced by approximately 99% as monitored by macerate analysis andas reported by Bode et al. (1976) and Teragawa and Bode (1990). I-cell-depleted hydra were used as host grafts (apical half of graft) inthe cell migration assay.

To disrupt mesoglea structure, two protocols were followed. In thefirst, polyps were treated with either 0.05 mM β-aminopropionitrile(β-APN) (Sigma) or 0.01 mM p-nitrophenyl β-D-xylopyranoside (β-xyloside; Sigma) for 15 days. 0.01 M HU was added to the abovesolutions at day 11. A 2-day recovery period in HM was employedafter 15 days of drug and HU treatment before grafting wasperformed. In the second protocol, although the length of treatmenttime was the same as previously described, the sequence of drugtreatment was reversed (i.e. polyps were first treated with HU and thenwith β-APN or β-xyloside) before grafting was performed. Previousbiochemical and autoradiography studies have shown that mesogleastructure is altered under these conditions of β-APN and β-xylosidetreatment (Sarras et al., 1991b). Two different sequences of HU andβ-APN or β-xyloside treatment were followed because of the concernthat initial treatment with the mesoglea perturbing drugs would com-pletely inhibit cell migration, thereby preventing the depletion of I-cells during HU treatment. As will be discussed in the results section,this was not the case.

Labeling of I-cells with 5-bromo-2′-deoxyuridineThe labeling procedure was carried out according to Teragawa andBode (1990, 1991). Briefly, animals were injected through thehypostome into the gastric cavity with an aqueous solution contain-ing 1.0 mM 5-bromo-2′-deoxyuridine (BrdU; Sigma) and 1-2% Indiaink (Pelikan C11/1431a, Bio/medical Specialties). As an analog of

thymidine, BrdU is incorporated into DNA during S-phase. BrdU-labeled cells can be distinguished from non-BrdU cells with immuno-histochemical staining using antibody against BrdU. In hydra, bothectodermal and endodermal cells can be labeled after 12 hoursfollowing BrdU injection (Teragawa and Bode, 1990). India ink canbe phagocytosed by endodermal epithelial cells and, thus, was usedin our grafting experiments to distinguish the graft junction linebetween BrdU-labeled tissues and those non-BrdU labeled. In orderto reduce experimental variations, efforts were taken to select similarsize polyps without buds for each experiment. Buds were removedbefore grafting from any animal that entered the budding process afterbeing selected and during pregrafting treatment. Considering the S-phase in hydra interstitial cells is 12 hours (Campbell and David,1974), BrdU injections were carried out 12 hours before grafting andrepeated 1 hour before grafting began. These BrdU-injected animalswere used as I-cell donors (basal half of graft) in hydra grafts.

Grafting proceduresThe grafting procedure is illustrated in Fig. 1. For these experiments,the division between the apical and the basal halves is determined bythe boundary between the gastric region and the budding region asdescribed by Campbell and Bode (1983). The grafting techniques thatwe employed were based on the procedures described by Teragawaand Bode (1990, 1991) with the following modifications: (1) the basalhalf of a BrdU/ink-labeled hydra polyp was always grafted to theapical half of a non-BrdU/ink-labeled polyp that had been either HUor HU and drug treated; (2) for any animals that grew buds during thepregrafting treatment, all buds were removed before animals weregrafted; and (3) grafted tissues were held together on a fishing linewith two end pieces of parafilm for 2 hours before any furthertreatment was performed.

Treatment of hydra grafts with reagentsHydra grafts were removed from the fishing lines 2 hours aftergrafting and transferred into microtiter plates (Nunc, Denmark) with1 graft per well. The microtiter plates were chilled to 4°C by placingthem on an ice water bath. The hydra medium in the wells wereremoved by aspiration with a tuberculin syringe and reagent solutionscontaining 5% DMSO were added into each well with 10 µl/well.DMSO was used to introduce reagents between the epithelium andthe mesoglea because it is able to open septate junctions betweenhydra epithelial cells (Fraser et al., 1987; Hans Bode, personal com-munication). The following reagents were tested in grafting treat-ments: bovine serum albumin (BSA; Sigma) at 0.05 mg/ml,fibronectin (Collaborative research Inc. Bedford, MS) at 0.05 mg/ml,GRGDSP (Gly, Arg, Gly, Asp, Ser, Pro) and GRGESP (Gly, Arg,Gly, Glu, Ser, Pro) synthetic peptides (Telios, San Diego, CA) at 0.5mg/ml, and non-immune rabbit serum or polyclonal antibody againsthuman plasma fibronectin (ICN Biomedicals) at 1:10 dilution. DMSOloading of reagents was carried out on ice for 30 minutes before thesolutions were removed and fresh non-DMSO-containing reagentsolutions were added to the wells. This latter step was needed (1) toprevent any leakage of the loaded reagents from the inter-epithelialspace before septate junctions closed and (2) to dilute the residualDMSO in hydra body. Hydra grafts remained on ice for additional 15minutes after DMSO treatment to allow septate junction closure andrecovery. The non-DMSO reagent solutions were changed one moretime to further dilute residual DMSO and hydra grafts were left in thefinal change of non-DMSO reagent solutions at 18°C until 24 hourshad elapsed from the initial time of reagent loading.

Immunocytochemistry and quantitative analysis of in vivoI-cell migrationAt 24 hours after grafting and reagent treatment, hydra grafts wereprocessed for immunohistochemistry staining following the proce-dures described by Teragawa and Bode (1990). Specimens wereexamined with a bright-field light microscope (Nikon) fitted with a

X. Zhang and M. P. Sarras

427Effect of ECM on I-cell migration in hydra

camera lucida attachment. The junction line between the apical andbasal half of the grafts was recognized by the ink labeling of the basalgraft. Migration was considered if a BrdU-positive cell appearedabove the junction line in the apical graft. Because the body wall ofthe hydra body column is composed of an epithelial bilayer, twoepithelial bilayers are compressed together in whole-mount prepara-tions. By focusing through the two epithelial bilayers in hydra whole-mount preparations, all BrdU-positive I-cells in the apical graft wereindividually identified and traced using the camera lucida drawingattachment. Cells in these camera lucida drawings were then countedfor quantitative analysis. In order to avoid experimental error createdby any ambiguity in the ink line, the apical half was divided into tenequal regions. Any BrdU-positive cells that appeared in the regionnext to the ink line in the apical half were not counted as migratedcells. A minimum of three grafts was used per parameter in eachexperiment and all experiments were repeated at least three times. Thetotal number of BrdU-positive cells within each apical graft wascounted and these numbers were normalized with each experiment aspercent of control. Only the percent of control was used when datafrom different experiments was combined for statistical analysis.Based on the combined percent of control, either a student t-test or anANOVA test was used to compare the control and experimentalgroups depending on the number of groups analyzed. A P value ≤0.05was set as the level of significant difference.

Ultrastructural analysis of DMSO-loaded hydraTo examine the effect of DMSO loading on the attachment of epithe-lial cells to the extracellular matrix, scanning (SEM) and transmission(TEM) electron microscopy was performed. Hydra were DMSOloaded with BSA or test molecules such as fibronectin and thenprocessed for SEM or TEM at various time points (0-24 hours aftertreatment) as previously described by Sarras et al. (1993). For SEManalysis, hydra were freeze-fractured (Sarras et al., 1993) to allow theepithelial bilayer to be observed in either a transverse or coronalplane.

RESULTS

Perturbation of mesoglea collagen cross-linking byβ-APN inhibited I-cell migrationThe hydra grafting method utilized in this study is illustratedin Fig. 1. With this method, BrdU-positive cells appearing inthe host hydra (apical half of the grafts) are solely due tomigration of cells from the donor hydra (basal half of the graft).Illustrations of these grafts and the appearance of migratedcells in the apical half of the grafts is shown in Figs. 2 and 3.The total number of BrdU-positive cells within the apical halfwas used for quantitative analysis of cell migration. Under theBrdU-labeling conditions used, the present study focused on I-cell migration. As shown in Fig. 4, I-cell migration was sig-nificantly reduced in host (apical) grafts treated with β-APN.Because β-APN treatment reduced but did not totally inhibit I-cell migration, the order in which hydra were treated with HUand β-APN did not significantly alter the final results. Incontrast, no significant differences were observed betweencontrol and β-xyloside-treated groups. These results indicatedthe relative importance of collagen cross linking versus pro-teoglycan processing during I-cell migration in hydra.

In addition to the total number of migrated cells, the distanceof cell migration was also analyzed. Apical grafts were dividedinto four zones longitudinally and BrdU-positive cells withineach zone were counted and plotted. To determine if the

absolute distance of cell migration was affected, migrated cellswithin each specific zone were compared between control andexperimental groups. The results indicated that, although thetotal number of migrated cells was reduced in the β-APN-treated apical half, cells were still able to migrate to all fourregions (most proximal to most distal). Therefore, β-APNtreatment did not reduce the maximal distance cells could beseen to migrate, but it did reduce the number of cells that couldeffectively migrate.

Fibronectin, antibody to fibronectin and RGDSsynthetic peptide inhibited I-cell migration in hydragraftsBrief exposure to DMSO has been shown to open the septatejunctions temporarily between hydra epithelial cells (Fraser etal., 1987; Hans Bode, personal communication). Thisprocedure was used in the present study to introduce macro-molecules between epithelial cells so that their effect on in vivoI-cell migration could be analyzed. As shown in Fig. 5, I-cellmigration was inhibited in hydra grafts treated with fibronectin,RGD peptide and antibody to fibronectin as compared to their

A

B

A

B

BrdU treated

Graft

HU or drugtreated

DMSO loadreagents

Localize withantiBrdU andanalyze withcameralucida

Fig. 1. Illustration of the in vivo I-cell migration assay whichinvolves hydra grafting techniques. Group A hydra were treated with0.01 M hydroxyurea (HU) for 5 days to eliminate their I-cellpopulation. A subgroup of hydra polyps were double treated with0.05 mM β-APN or 0.01 mM β-xyloside for 15 days to perturb theirmesoglea structure and then with HU for 5 days. A second subgroupof hydra were treated with this order reversed (i.e. 5 days HUtreatment followed by 15 days of β-APN or β-xyloside treatment).Group B hydra were labeled with 1 mM BrdU containing 1% Indiaink at 12 hours and 1 hour prior to grafting. The basal half of BrdU-labeled polyps was always grafted with the apical half of HU-treatedor double-treated polyps and the grafts were kept in hydra mediumfor 24 hours. The determination of the cutting line separating apicaland basal grafts is described in ‘Materials and Methods’. In aseparate set of experiments, non-β-APN- or non-β-xyloside-treatedhydra grafts were treated on ice for 30 minutes with reagents(fibronectin, synthetic peptides or antibodies) containing 5% DMSOat 2 hours after grafting. After washes in DMSO-free reagentsolutions, hydra grafts were incubated a total of 24 hours in eachreagent solution. All grafts were processed for immunocytochemicalstaining using BrdU antibody and viewed with a light microscopecontaining a camera lucida attachment.

428

respective controls (BSA, RGE peptide andnon-immune serum).

Attachment of epithelial cells to theextracellular matrix was not disruptedwith the drugs or reagents used inthese studiesTo determine if the inhibition of I-cellmigration could be a secondary effect resultingfrom a disruption of attachment of epithelialcells to the mesoglea, ultrastructural studieswere performed. It was previously shown thattreatment of hydra with β-APN or β-xylosidealtered the structure of hydra extracellularmatrix (mesoglea) but did not disrupt theattachment of epithelial cells to the mesoglea.This analysis was expanded in the currentstudy to evaluate the effect of the DMSO-loading procedure on the attachment of epithe-lial cells to the mesoglea. As shown in Fig. 6,at the concentrations used in these studies,DMSO loading of molecules such asfibronectin (0.05 mg/ml) did not cause a dis-ruption of the attachment of epithelial cells tothe extracellular matrix as evaluated by SEMor TEM analysis.

DISCUSSION

Among all hydra cell types, interstitial cells arethe fastest dividing subgroup of cells with a S-phase of about 12 hours (Campbell and David,1974). This feature gave us the opportunityspecifically to analyze I-cell migration withinhydra grafts. Previous studies have shown thattreatment of hydra with HU for 5 days depletesthe majority of their I-cell population (Bode etal., 1976) and, therefore, when the apicalhalves of these hydra are grafted to the basalhalves of normal hydra, an I-cell populationgradient is created. This gradient results in ahigher number of cells migrating from basalhalf to the apical half as compared to hydragrafts formed between non-HU-treated polyps(Teragawa and Bode, 1990). The combinationof this HU-induced cell migration pattern andthe BrdU labeling of I-cells granted us anunique model system to determine if perturba-tion of ECM structure could affect I-cellmigration. It should be noted that while othercell types do migrate in hydra (e.g. nemato-cytes), the current study only focused on I-cellmigration because of the techniques employed.

In regard to our pharmacological studies, β-APN has been shown to interfere with collagencross linking by inhibiting lysyl oxidase (Pageand Benditt, 1972; Wilmarth and Froines,1992). Previous studies have established thatβ-APN does affect mesoglea structure (Sarraset al., 1993) and collagen cross linking in hydra

X. Zhang and M. P. Sarras

Fig. 2. Whole-mount preparations of control (A,C,D) and experimental (B) graftsstained with antibody to BrdU are shown. An experimental graft DMSO-loaded withfibronectin (0.05 mg/ml) is shown in B while the control shown represents a graftedDMSO-loaded with BSA at the same concentration. At low magnification, the juncturepoint between the apical (the host half for cell migration) and basal (donor half whichwas injected with BrdU and ink-labelled) halves of the graft are indicated by the arrows(shown in A and B). BrdU-labelled nuclei of I-cells, which had migrated from the basalto the apical half, are indicated by the arrowheads in A and B. At higher magnification,BrdU-labelled I-cell nuclei in the apical half could be distinguished from one another(C). These I-cells reside in the epithelial bilayer of the tubular body column of hydra.By focusing through the two epithelial bilayers, which are compressed together inwhole-mount preparation, I-cells can be identified and the total number that hadmigrated into the apical half of the graft counted. As viewed at the periphery of thegraft using phase-contrast microscopy (D), some I-cells (arrowheads) are seen injuxtaposition to the extracellular matrix (arrows), which appears as a clear line at thebase of the ectodermal cell layer in these preparations. Bar for A and B, 125 µm. Barfor C and D, 25 µm.

429Effect of ECM on I-cell migration in hydra

(Sarras et al., 1991b). In addition, previous in vitro studies haveshown that, when used as a substratum, collagens (especiallytype IV collagen) promote migration of various cell types suchas bronchial epithelial cells (Rickard et al., 1993), neural crestcells (Perris et al., 1991) and hydra nematocytes (Agosti andStidwill, 1990). Under in vivo conditions, type I, III and IVcollagen were shown to appear along the neural crest migratorypathways during development of chick embryos (Perris et al.,1991). This indicated a role for these ECM components inmediating neural crest cell migration during embryonic devel-opment. Therefore, in the current study, the reduction of I-cell

migration in β-APN-treated grafts can be interpreted as theresult of alterations in mesoglea structure related to the per-turbation of collagen processing. β-xyloside has been shownto interfere with the addition of glucosylaminoglycan (GAG)chains to proteoglycan core proteins (Lelongt et al., 1988).When added to culture medium, this reagent inhibited themigration of primary mesenchyme cells in sea urchin embryos(Lane and Solursh, 1988). While β-xyloside treatment doesinhibit hydra head regeneration (Sarras et al., 1991b) and mor-phogenesis of hydra cell aggregates (Sarras et al., 1993), itstreatment had no affect on I-cell migration in hydra grafts. Wecan therefore eliminate alterations in cell migratory patterns asthe basis for β-xyloside’s inhibitory effect on general mor-phogenesis in hydra. Overall, our results indicate that, in hydra,I-cell migration is more sensitive to alterations in collagenstructure than alterations in proteoglycan structure.

Under in vitro conditions, fibronectin has been shown topromote cell adhesion and cell migration in neural crest cells(Perris et al., 1989; Dufour et al., 1988), keratinocytes (Sarretet al., 1992) and smooth muscle cells (Naito et al., 1992).Hydra nematocytes have been shown to bind to a fibronectin-coated substratum and this binding is known to be RGD

Fig. 3. Two sample camera lucida drawings of a control graft (A) anda graft treated with β-APN (B). For quantitative analysis of I-cellmigration, the apical half and the juncture point with the basal half wasdrawn for each graft. The junction line between the two graft halvescould be identified under the microscope by the ink particle labeling ofthe basal half. This juncture point is indicated by the arrowheads in thecamera lucida drawings shown in this figure. The asterisk (*) indicatestentacles of the apical half of the graft. Bar, 180 µm.

Treatment of apical halves

Analysis of total number of BrdU positive cellscounted within the apical halves

Per

cent

of c

ontr

ol (

HU

gro

up)

Fig. 4. The effect of drugs that perturb ECM structure on in vivo I-cell migration. Hydra polyps used for apical grafts were treated witheither 0.05 mM β-APN or 0.01 mM β-xyloside for 15 days and with0.01 M HU for 5 days (HU+β-APN or HU+β-XYL). For details, see‘Materials and Methods’. Asterisks indicate groups statisticallydifferent from controls with P≤0.05.

Per

cent

of c

ontr

ol g

roup

Per

cent

of c

ontr

ol g

roup

Analysis of total number of BrdU positive cellscounted within the apical halves

Treatment of grafts (mg/ml)

Treatment of apical grafts (dilution)

Fig. 5. The effect of DMSO-loading with fibronectin (FN), syntheticpeptides (RGD or RGE) and antibody to fibronectin (Anti-FN) on I-cell migration. (A) FN and RGD peptide significantly reduced I-cellmigration as compared to BSA or RGE controls. (B) Anti-FN alsosignificantly inhibited I-cell migration as compared to non-immuneserum (Non-Immune) controls. I-cell migration was analyzed asdescribed in ‘Materials and Methods’ and in the figure legend of Fig.3. Asterisks indicate groups significantly different from controls(P≤0.05).

430 X. Zhang and M. P. Sarras

Fig. 6. Ultrastructural analysis of the effect of the DMSO-loading procedure on the attachment of epithelial cells to the extracellular matrix inhydra. Control (A,C,E; BSA, 0.05 mg/ml) and experimental (B,D,F; fibronectin, 0.05 mg/ml) specimens are shown. As shown by SEM at low(A,B) and intermediate (C,D) magnification, neither BSA or fibronectin at the concentrations used in this study resulted in any apparentdisruption of the attachment of ectodermal (Ec) or endodermal epithelial cells to the hydra extracellular matrix (arrowheads indicate theattachment sites of epithelial cells to the hydra mesoglea). The attachment of epithelial cells to the extracellular matrix was confirmed by TEManalysis of BSA- (E) and fibronectin- (F) treated specimens. The close association of the epithelial plasma membrane to the mesoglea (M) isindicated by the arrowheads in E and F. Bar for A and B, 10 µm. Bar for C and D, 2.5 µm. Bar for E and F, 233 nm.

431Effect of ECM on I-cell migration in hydra

dependent (Ziegler and Stidwill, 1991). Although nematocytescan bind to fibronectin, studies by Agosti and Stidwill (1990)have shown that mature nematocytes migrate poorly on thisECM component. The current study revealed that under in vivoconditions, I-cell migration can be inhibited by macromole-cules that can compete with cell-fibronectin interactions (e.g.intact fibronectin, RGD peptide or antibody to fibronectin).

The inhibitory effect of fibronectin and antibody tofibronectin on I-cell migration may seem contradictory. Anumber of interpretations can be proposed for this result. Forexample, these results could result from a competition betweenendogenous mesoglea fibronectin and exogenously addedsoluble fibronectin. In this case, it can be proposed that normalI-cell migration is dependent on the interaction of cell surfacefibronectin receptors with fibronectin molecules that areinsoluble and bound within the three-dimensional structure ofthe hydra ECM. The presence of hydra cell surface receptorsfor fibronectin is supported by in vitro cell adhesion studies byZiegler and Stidwill (1992). Exogenously applied solublefibronectin would compete for I-cell ECM receptors andinterfere with normal contact guidance mechanism andtherefore result in an inhibition of cell migration. In contrast,antibody to fibronectin could mask endogenously boundfibronectin and thereby interfere with the ability of I-cell ECMreceptors to bind to fibronectin in the ECM. These proposalsare supported by the fact that RGD peptide could also inhibitI-cell migration. The RGD amino acid sequence is known tobe a cell binding domain for fibronectin and can bind tointegrin receptors (see review by Akiyama et al., 1990; Hynes,1992). When studied in an in vitro assay with isolated hydranematocytes, this peptide was shown to inhibit nematocytebinding to fibronectin (Ziegler and Stidwill, 1992). Theseobservations can now be extended to the in vivo situation inthe case of migrating I-cells. The inhibitory affect of RGDpeptide on I-cell migration appeared to be specific since theinactive peptide RGE had no effect on I-cell migration.

Although the total number of migrating cells is reduced aftertreatment with β-APN, fibronectin, anti-fibronectin antibody orRGDS synthetic peptides, the maximal distance of I-cellmigration, however, was not affected. One explanation for thisresult is that the migration of particular subpopulations of I-cells was significantly inhibited by these reagents while othersubpopulations of I-cells were only marginally affected or werenot affected at all. While the immunocytochemical proceduresutilized in this study present I-cells as a morphologicallyhomogeneous group, they are in fact a heterogeneous cell pop-ulation composed of multipotent stem cells and various celllineage precursor cells (Heimfeld and Bode, 1986a,b). Thesesubpopulations of cells could be selectively sensitive to theperturbing reagents used due to the expression of specific cellsurface receptors for ECM components.

Several mechanisms have been proposed to explain cellmigration in hydra. These mechanisms include (1) the mechan-ical forces resulting from tissue changes during contraction andexpansion of polyps (Teragawa and Bode, 1990), (2) cell-cellinteractions involved in contact guidance (Campbell andMarcum, 1980) and (3) external cues such as chemotacticsignaling that may be related to the head activator gradientalong the longitudinal axis of the organism (Teragawa andBode, 1991). In addition to these proposed mechanisms, ourdata indicate the potential role cell-ECM interactions during in

vivo I-cell migration. The inhibition of I-cell migrationobserved in this study could reflect a direct interaction of I-cells with the ECM or could result from alterations in thenormal attachment of epithelial cells with the ECM, which thencauses a secondary inhibitory effect on I-cell migration. In thelatter case, altered epithelial attachment to the ECM wouldresult in a perturbation of cell-cell interactions and/or chemo-tactic signaling systems that normally occur during I-cellmigration. The ultrastructural analyses performed in this studyhowever, indicate that, under the conditions used, no disrup-tion in the attachment of epithelial cells with the mesogleacould be observed. While this in itself does not excludepotential secondary inhibitory effects, taken in concert, all ofthe data presented in the current study is consistent with adirect interaction of I-cells with ECM components. As a finalnote regarding epithelial-ECM interactions in hydra, it shouldbe noted that DMSO loading of higher concentrations offibronectin (e.g. 0.1 mg/ml) can cause a rapid dissociation ofhydra cells in the adult organism (data not shown). Thissuggests that, while epithelial-fibronectin interactions may bea component of epithelial-ECM attachment in hydra, theseinteractions were not affected by the concentration of reagentsused in the present study. Although the exact mechanismsunderlying in vivo I-cell migration in hydra are not yet clear,the current study and others do point to the presence of specificECM cell surface receptors within the different hydra celltypes. In this regard, Ziegler and Stidwill (1992) have isolatedintegrin-like plasma membrane proteins from nematocyteswith binding affinity for fibronectin. Further studies will berequired to identify the full spectrum of ECM cell surfacebinding proteins among the different hydra cell types and todetermine their respective roles in the process of patternformation in this organism.

The authors wish to thank Dr Hans R. Bode for his continuoussupport and Drs Lynne Littlefield and Hiroshi Shimizi for their helpand suggestions regarding hydra grafting and quantitative analysis ofin vivo I-cell migration. The authors also wish to thank the technicalsupport of Jacquelyn K. Huff in regard to the ultrastructural studiespresented in this article. The studies described in this article weresupported by funds provided by NIH (RR06500) and the InternationalJuvenile Diabetes Foundation Inc.

REFERENCES

Agosti, C. G. and Stidwill, R. P. (1991). In vitro migration of hydranematocytes: the influence of the natural extracellular matrix (the mesoglea),of collagen type IV and type I, laminin, and fibronectin on cell attachment,migration parameters, and on patterns of cytoskeletal proteins. Cell Motilityand the Cytoskeleton. 20, 215-227

Akiyama, S. K., Nagata, K. and Yamada, K. M. (1990). Cell surfacereceptors for extracellular matrix components. Biochemica et Biophysica1031, 91-110.

Bode, H. R., Flick, K. M. and Smith, G. S. (1976). Regulation of interstitialcell differentiation in Hydra attenuata. I. Homeostatic control of interstitialcell population size. J. Cell Sci. 20, 29-46.

Campbell, R. D. and David, C. N. (1974). Cell cycle kinetics and developmentof Hydra attenuata. II. Interstitial cells. J. Cell Sci. 16, 349-358.

Campbell, R. D. and Marcum, B. A. (1980). Nematocyte migration in hydra:Evidence for contact guidance in vivo. J. Cell Sci. 41, 33-51

Campbell, R. D. and Bode, H. R. (1983). Terminology for morphology andcell types. In Hydra: Research Methods. (ed. H. M. Lenhoff), pp 5-14. NewYork: Plenum Press.

Day, R. M. and Lenhoff, H. M. (1981). Hydra mesoglea: a model for

432

investigating epithelial cell-basement membrane interactions. Science 211,291-294.

Dufour, S., Duband, J., Humphries, M. J., Obara, M., Yamada, K. M. andThiery, J. P. (1988). Attachment, spreading an locomotion of avian neuralcrest cells are mediated by multiple adhesion sites on fibronectin molecules.EMBO J. 7, 2661-2671

Fraser, S. E., Green, C. R., Bode, H. R. and Gilula, N. B. (1987). SelectiveDisruption of Gap Junctional Communication Interferes with a PatterningProcess in Hydra. Science 237, 49--55

Hay, E. D. (1991). Collagen and other matrix glycoproteins in embryogenesis.In Cell Biology of Extracellular Matrix, 2nd edition. (ed. E. D. Hay), pp 419-462, New York: Plenum Press.

Heimfeld, S. and Bode, H. R. (1986a). Growth regulation of the interstitial cellpopulation in hydra. III. interstitial cell density does not control stem cellproliferation. Dev. Biol. 116, 51-58.

Heimfeld, S. and Bode, H. R. (1986b). Growth regulation of the interstitial cellpopulation in hydra. IV. Control of nerve cell and nematocyte differentiationby amplification of non-stem interstitial cells. Dev. Biol. 116, 59-68.

Hynes, R. O. (1992). Integrins: Versatility, Modulation, and Signaling in celladhesion. Cell 69, 11-25.

Lane, M. C. and Solursh, M. (1988). Dependence of sea urchin primarymesenchyme cell migration on Xyloside- and sulfate-sensitive cell surface-associated components. Dev. Biol. 127, 78-87.

Lelongt, B., Markino, H., Dalecki, T. M. and Kanwar, Y. S. (1988). Role ofproteoglycans in renal development. Dev. Biol. 128, 256-276

Naito, M., Funaki, C., Hayashi, T., Yamada, K., Asai, K., Yoshimine, N.and Kuzuya, F. (1992). Substrate-bound fibrinogen, fibrin and other cellattachment-promoting proteins as a scaffold for cultured vascular smoothmuscle cells. Atherosclerosis 96, 227-234.

Page, P. C. and Benditt, E. P. (1972). Diseases of connective and vasculartissues. IV. The molecular basis of lathyrism. Lab. Invest. 26, 22-26.

Perris, R., Paulsson, M. and Bronner-Fraser, M. (1989). Molecularmechanisms of avian neural crest cell migration on fibronectin and laminin.Dev. Biol. 136, 222-238.

Perris, R., Krotoski, D. and Bronner-Fraser, M. (1991). Collagens in neuralcrest development: distribution in vivo and migration-promoting ability invitro. Development 113, 969-984.

Rickard, K. A., Taylor, J., Rennard, S. I. and Spurzem, J. R. (1993).Migration of bovine bronchial epithelial cells to extracellular matrixcomponents. Am. J. Respri. Cell Mol. Biol. 8, 63-68.

Sarras, M. P. Jr., Madden, M. E., Zhang, X., Gunwar, S., Huff, J. K. andHudson, B. G. (1991a). Extracellular matrix (mesoglea) of Hydra vulgaris. I.Isolation and characterization. Dev. Biol. 148, 481-494

Sarras, M. P. Jr., Meador, D. and Zhang, X. (1991b). Extracellular matrix(mesoglea) of Hydra vulgaris. II. Influence of collagen and proteoglycancomponents on head regeneration. Dev. Biol. 148, 495-500

Sarras, M. P. Jr., Zhang, X., Huff, J. K., Accavitti, M. A., St.John, P. L. andAbrahamson, D. R. (1993). Extracellular matrix (mesoglea) of Hydravulgaris. III. Formation and function during morphogenesis of hydra cellaggregates. Dev. Biol. 157, 383-398.

Sarret, Y., Stamm, C., Jullien, D. and Schmitt, D. (1992). Keratinocytemigration is partially supported by the cell-binding domain of fibronectin andis RGDS-dependent. J. Invest. Dermatol. 99, 656-659.

Teragawa, C. K. and Bode H. R. (1990). Spatial and temporal patterns ofinterstitial cell migration in hydra vulgaris. Dev. Biol. 138, 63-81.

Teragawa C. K. and Bode H. R. (1991). A head signal influences apicalmigration of interstitial cells in hydra vulgaris. Dev. Biol. 147, 293-302.

Wilmarth, K. R. and Froines, J. R. (1992). In vitro and in vivo inhibition oflysyl oxidase by aminopropionitriles. J. Toxicol. Environ. Health. 37, 411-423.

Wood, R. L. and Kuda, A. M. (1980). Formation of junctions in regeneratinghydra: Septate junctions. J. Ultrastruct. Res. 70, 104-117

Ziegler, U. and Stidwill, R. P. (1992). The attachment of nematocytes from theprimitive invertebrate Hydra to fibronectin is specific and RGD-dependent.Exp. Cell Res. 202, 281-286.

(Accepted 6 November 1993)

X. Zhang and M. P. Sarras