ava i l ab l e a t www.sc i enced i r ec t . com

www.e l sev i e r . com/ l oca te / sc r

Stem Cell Research (2011) 6, 168–176

REGULAR ARTICLE

Nonmitogenic survival-enhancing autocrine factorsincluding cyclophilin A contribute todensity-dependent mouse embryonic stem cell growthNikhil Mittal a,1, Joel Voldmanb,⁎

a Department of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139, USAb Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge,MA 02139, USA

Received 1 September 2010; received in revised form 7 October 2010; accepted 11 October 2010Available online 20 October 2010

Abstract An improved understanding of the role of extracellular factors in controlling the embryonic stem cell (ESC)phenotype will aid the development of cell-based therapies. While the role of extracellular factors in controlling thepluripotency and differentiation of embryonic stem cells (ESCs) has been the subject of much investigation, the identity androle of extrinsic factors in modulating ESC growth under conditions supporting self-renewal remain largely unknown. Wedemonstrate that mouse ESC (mESC) growth is density dependent and that one of the mechanisms underlying this phenomenonis the action of survival-enhancing autocrine factors. Proteomic analysis of proteins secreted by mouse ESCs demonstratessignificant levels of cyclophilin A which increases the growth rate of mouse ESCs in a dose-dependent manner. Additionally,inhibition of the cyclophilin A receptor CD147 decreases the growth rate of mESCs. These findings identify cyclophilin A as anovel survival-enhancing autocrine factor in mouse ESC cultures.

© 2010 Elsevier B.V. All rights reserved.

Introduction

Embryonic stem cells (ESCs) are cells derived from the innercell mass (ICM) of preimplantation blastocysts with theability to produce any cell type in the adult animal. WhileESCs can differentiate to form various cell types, they canalso divide to form two daughter ESCs via self-renewal.Although self-renewal encompasses both growth and main-

⁎ Corresponding author. Room 36-824, MIT, 77 MassachusettsAvenue, Cambridge, MA 02139, USA. Fax: +1 617 258 5843.

E-mail address: voldman@mit.edu (J. Voldman).1 Present address: Harvard Stem Cell Institute, Cambridge, MA

02138, USA.

1873-5061/$ – see front matter © 2010 Elsevier B.V. All rights reserveddoi:10.1016/j.scr.2010.10.001

tenance of developmental potential, most research hasfocused on understanding the signaling and transcriptionalnetworks involved in the latter, i.e., maintenance of the ESCpluripotent state (Liu et al., 2007). In contrast, compara-tively few studies have investigated pathways involved inESC growth and death under conditions that have been shownto maintain pluripotency. Studies have shown that theaddition of EGF, ANG-2, Activin, Nodal, LIF, or BMP-4proteins to mouse ESC (mESC) cultures results in improvedgrowth (Heo et al., 2006; Han et al., 2007; Ogawa et al.,2007; Viswanathan et al., 2002; Qi et al., 2004). However,these studies did not include testing for secretion of theseproteins by ESCs; i.e., they did not attempt to fullydemonstrate that these pathways are constitutively activein mouse ESC cultures and that the factors are acting in an

.

169Non-mitogenic survival-enhancing autocrine factors

autocrine fashion. SDF-1 has been shown to act as anautocrine survival factor for mESCs, but only in the absenceof serum (i.e., during serum withdrawal) (Guo et al., 2005).

Because ESCs grow as colonies, juxtacrine signaling couldalso presumably play a growth-supportive role. Todorovaet al. used pharmacological blockers of gap junctionalcommunication and Connexin-43 small-interfering RNA toinhibit signaling via gap junctions in mESCs (Todorova et al.,2008). They demonstrated that gap junctional communica-tion enhances either the growth rate or the attachmentefficiency (or both) of mESCs.

Cell growth is intimately tied to progression through thecell cycle, and several studies have focused on understandingcell cycle control in mESCs. mESC growth in culture ischaracterized by a short cell cycle (11–16 h), primarily owingto a reduction in the duration of the G1 phase as compared tosomatic cells (Orford and Scadden, 2008). This short G1phase is due to a lack of the mitogen-dependent early G1phase of the cell cycle, owing to constitutive cyclin E–CDK2activity (Stead et al., 2002). This constitutive activity ofcyclin E-CDK2 in turn appears to be driven at least partiallyby a constitutively active Ras protein expressed specificallyin ESCs, called ERas, which stimulates PI3K (Takahashi et al.,2003), which is upstream of cyclin E (Lianguzova et al.,2007). Thus, it is unlikely that secreted factors will modulatethe length of the G0/G1 phase of the cell cycle, as they do inseveral other cell types. However, secreted factors couldaffect growth by modulating the rate of apoptosis.

One approach to searching for growth-enhancing secretedmolecules has been to investigate whether proteins secretedby mESCs influence the growth of other types of cells. Guo etal. have shown that mESC-conditioned medium improves thesurvival of hematopoietic progenitor cells (Guo et al., 2006),while Singla et al. showed that mESC-conditioned mediuminhibits apoptosis in H9c2 cardiomyocyte cells in a PI3K-dependent manner (Singla et al., 2008). They determinedthat one of the factors responsible for this effect was TIMP-1.Since these studies show that mESCs secrete factors withantiapoptotic activity for other cell types, we wonderedwhether these (or other mESC-secreted) factors might haveprosurvival and/or mitogenic effects on mESCs themselves.

Figure 1 Density-dependent growth of mESCs. (a) Fold growth ov(b) Colony-forming efficiencies for ABJ1 mESCs plated at differentcells/cm2 have ~15% higher colony-forming efficiency than cells pl

Indeed, although mESCs can be clonally propagated, it iscommonly known that their proliferation is densitydependent.

Here we demonstrate the existence of one or moreautocrine survival factor(s) in mESC cultures. By performingLC/MS-MS analysis of mESC-conditioned medium, we deter-mined that cystatin C and cyclophilin A are two majorconstituents of the mESC secretome. We tested the effectsof these proteins, as well as other proteins known to besecreted by mESCs (such as TIMP-1), and found that onlycyclophilin A improved the growth of mESCs. Additionally,antibody-mediated inhibition of the cyclophilin receptorwith a polyclonal antibody decreased the growth rate ofmESCs. These results demonstrate that cyclophilin A is anautocrine survival factors in mESC cultures.

Results

Growth and colony-forming efficiency of mESCs inserum-containing medium are density dependent

We first examined whether mESC growth is density depen-dent, which is a simple indicator (with caveats, elaboratedbelow) of the existence of autocrine growth-supportivefactors in a culture (Lauffenburger and Cozens, 1989).Throughout, we differentiate between growth (which is therate of change of cell number in the culture) andproliferation (travel through the cell cycle, i.e., mitosis);the rate of cell growth is equal to the rate of cellproliferation minus the rate of cell death. We plated ABJ1mESCs at various densities and measured the increase in cellnumber at the ends of Day 1 (24 h after plating) and Day 2.We observed density-dependent growth during both timeperiods (Fig. 1a, Day 1, P=5.7e-6, Day 2, increasing region(250–2500 cells/cm2), P=0.003). Separately, we determinedthat the lag phase for mESC culture is less than 12 h (Fig. S1).Thus the density-dependent fold increase in cell number overDay 2 is solely due to differences in the growth rate, and notdue to variations in the length of the lag phase. We observedsimilar trends for a second independently derived cell line

er Day 1 and Day 2, for ABJ1 mESCs plated at various densities.densities, assessed at the end of Day 2. Cells plated at 3000

ated at 300 cells/cm2 (P=0.004).

170 N. Mittal, J. Voldman

(Fig. S2, CCE mESCs). The average doubling time on Day 2 forABJ1 mESCs was ~11 h, similar to that measured for D3mESCs by Stead et al. (2002). The decrease in growthobserved on Day 2 for densities greater than 5000 cells/cm2

is likely due to a combination of nutrient depletion and theaccumulation of metabolites (Fig. S3). To determine thecontribution of density-dependent colony-forming efficien-cies on the observed density dependence of growth, wemeasured the colony-forming efficiency of cells plated attwo different densities using a microfluidic device thatenabled the plating of single cells at defined densities(Supplemental Information). The colony-forming efficiencieswe obtained (~40–60%) are comparable to those measuredby others (Schratt et al., 2001). Importantly, we found thatthe colony-forming efficiency was ~15% higher at the higherplating density (Fig. 1b, P=0.004; Fig. S2), in agreement withthe fold growth data. These results suggest that an autocrinesurvival factor may exist in mESC cultures.

mESC cultures contain one or more autocrinegrowth-supportive large molecules

While a density-dependent growth increase is usuallyassociated with the presence of an autocrine growth-supportive factor in the culture (Lauffenburger and Cozens,1989), such an effect could also be produced by the density-dependent depletion of a growth inhibitor from the medium.Since the contents of serum are not defined, it is not possibleto rule out the existence of a growth-inhibiting molecule asone of its constituents. To distinguish between these twomechanisms, we used a complementary approach of usingconditioned medium (CM) from culture in a serum-freemedium to determine if cell-secreted molecules are growthsupportive, along with assays of the constituents of theserum-free medium to determine if they are growthinhibitory. We focused on large molecules (N3 kDa) as theyare the most common signaling molecules, and because we

Figure 2 Effects of secreted proteins and medium proteins on mEgrown in conditioned medium versus control. CM increased proliferatgrowth over Day 2, relative to control, in media where different comresulted in no significant decrease in proliferation. Removal of tranentirely, caused significant growth reduction (P=7e-4 and 1e-3, rresulted in no change in fold growth.

could dialyze the conditioned media with fresh media toequilibrate any changes in the small molecule fraction of themedia (e.g., due to nutrient depletion).

As a first step, we grew cells in a serum-free mediumformulation (N2B27+LIF+BMP-4) developed by Ying et al.(2003) specifically for expanding mESCs while maintainingthem in a pluripotent state. We observed density-dependentgrowth in this medium as well (Fig. S4). Next, we collectedconditioned medium from cultures, dialyzed it against freshmedia using a 3-kDa membrane, and compared growth ofcells in dialyzed CM against control medium that was treatedidentically except that it was never cultured with cells. Weobserved that the large-molecule fraction of CM (N 3 kDa)prepared using this medium was growth supportive on Day 1and Day 2, resulting in a ~15% increase in growth (relative tocontrol) on Day 1, and a ~25% increase in growth on Day 2(Fig. 2a).

The serum-free medium (N2B27) consists of a 1:1 mix ofN2 medium and Neurobasal/B-27 medium, to which leukemiainhibitory factor (LIF) and bone morphogenetic protein 4(BMP-4) are added to prevent the differentiation of themESCs. There are a total of 7 large molecules in the finalmedium—bovine serum albumin (BSA), transferrin, insulin,superoxide dismutase 1 (SOD-1), catalase, LIF, and BMP-4(Ying et al., 2003; Brewer et al., 1993) (SupplementaryInformation). Although these molecules have been shown tobe growth supportive for several other cell types (Viswanathanet al., 2002; Qi et al., 2004; Keenan et al., 2006; Fattmanet al., 2003; Michiels et al., 1994), we examined whethertheymight be growth inhibitory for mESCs. While the contentsof B-27 are known (Brewer et al., 1993), the relative amountsare proprietary (Invitrogen), precluding the ability to removethe protein components on an individual basis. Instead weremoved the protein components of N2 (BSA, transferrin, andinsulin) individually, and then removed the high MW compo-nents of the B-27 formulation entirely. When we removed theBSA or insulin from N2, we observed no significant change in

SC growth. (a) Fold growth over Day 1 and Day 2 for ABJ1 mESCsion by ~15% on Day 1 (P=.02) and ~25% on Day 2 (P=5e-4). (b) Foldponents were removed. Removal of BSA or insulin (Ins) from N2sferrin from N2 (Trans), or of the high MW components of B-27espectively). Reduction of LIF or BMP-4 concentrations by 50%

171Non-mitogenic survival-enhancing autocrine factors

the growth of mESCs over 2 days (in N2B27+LIF+BMP-4,Fig. 2b), indicating that unless there is a substantial decreasein the concentration of these proteins (these proteins arepresent in B-27 as well (Brewer et al., 1993)), they are neithergrowth supportive nor inhibitory. When we removed transfer-rin from N2 we found that the growth rate of mESCs decreasedby ~15% (Pb0.05), demonstrating that transferrin is growthsupportive for mESCs. When we removed the high MWcomponents of B-27 altogether, there was an ~60% decreaseinmESC growth (Pb0.05), indicating that B-27, which containsBSA, transferrin, insulin, SOD-1, and catalase (Brewer et al.,1993), is strongly growth supportive, perhaps due to thesynergistic action of some of these proteins. Separately, wealso examined the impact of catalase supplementation on thegrowth of mESCs in N2, and observed a growth-supportiveeffect (Fig. S5). Finally, although complete removal of LIF andBMP-4 from the medium would induce differentiation, weassayed whether reduction of LIF or BMP-4 concentrations by50% would affect growth, which it did not. Other studies haveshown that near-complete depletion of LIF (Viswanathanet al., 2002) or BMP-4 (Qi et al., 2004) does lead to a reductioninmESC survival, indicating that they are growth-supportive aswell. Together these results demonstrate that the large (N3kDa) molecules present in N2B27+LIF+BMP-4 have either no

Figure 3 (a) Cell cycle distribution for cells plated at 1000 cells/cm(dead) cells in the culture at the end of Day 2 for cells grown in condcells that detached on Day 2, for two plating densities (P=0.01). (d) Tin conditioned medium (CM) versus control (P=2e-5).

effect or a supportive effect on growth, and thus are notgrowth inhibitory.

To summarize, the large-molecule fraction of CM isgrowth supportive and there are no large growth-inhibitorymolecules in the original medium. Thus, the growthsupportive effect of the CM must be due to the productionof a large growth-supportive factor by the cells used tocondition the medium. This further implies that the density-dependent growth observed in Fig. 1 is at least partially dueto the presence of a N3-kDa growth-supportive autocrinemolecule.

The autocrine growth factor(s) in mESC cultures is(are) not mitogenic, but improve(s) survival

The growth rate of a culture can be increased via twomechanisms—a reduction in the length of the cell cycle(“mitogenic activity,” i.e., increased proliferation rate), ora reduction in the rate of cell death. We wanted todetermine whether the enhanced growth of mESCs atparticular densities was due to increased mitogenic activityand/or improved survival. If a mitogenic factor was beingproduced by the cells, one would expect that the G1 phasewould be shorter (because S and G2/M are typically fixed),

2 and 10 000 cells/cm2. (b) The percentage of YO-PRO-1-positiveitioned medium (CM) versus control (P=0.01). (c) The fraction ofhe fraction of unattached cells at the end of Day 2 for cells grown

172 N. Mittal, J. Voldman

and hence the proportion of cells in G1 would decrease.Using flow cytometry, we measured the proportion of cells indifferent phases of the cell cycle at two densities at the endof Day 1, and observed no difference (Fig. 3a). Addition of CMalso did not affect the cell-cycle distribution (Fig. S6). Thisindicates that the growth enhancement at higher densities isnot due to increased mitogenic activity but is instead likelydue to a decrease in the rate of cell death. The lack ofmitogenic activity is perhaps not surprising, given that ESCsappear to lack the mitogen-dependent early G1 phase (Steadet al., 2002).

Since the colony-forming efficiency improves with in-creasing density (Fig. 1b), this suggests that there is indeedimproved survival at the higher densities. To investigate thisfurther, wemeasured the fraction of dead cells in the cultureat the end of Day 2. We focused on Day 2 because differencein the death rate on Day 1 could also be due to differences inthe attachment efficiency. We added either CM or controlmedium to cells on Day 2; at the end of Day 2 we stained cellsusing YO-PRO-1 iodide which selectively stains dead cells.We observed that when grown in CM, the fraction of stainedcells was only 62±10% of the corresponding fraction for cellsgrown in control medium (P=0.01, Fig. 3b). This resultconfirms that mESC-secreted autocrine factors promotesurvival.

Interestingly, we observed that most cells that stainedpositively for YO-PRO-1 exhibited an unattached, roundedmorphology (142/143 cells or 99%, Fig. S7), and conversely 95%(142/149) of unattached cells stained positively for YO-PRO-1iodide, indicating a compromised membrane. Similarly, 97%of unattached cells showed the presence of DNA nicks viaTUNEL assay (Fig. S8). Thus death and loss of attachmentseem to occur almost simultaneously in mESC cultures. Thisobservation enables an alternate method for measuring therate of cell death in these cultures—via counting of thefraction of unattached cells in the culture. We counted thefraction of cells that were detached at the ends of Day 1 andDay 2, at 1000 and 5000 cells/cm2, the latter being the densitythat shows the highest growth on Day 2 under serum-freeconditions (Fig. S4, ABJ1 cells). From this, we calculated thefraction of cells that detached on Day 2, and indeed found adensity-dependent decrease in the fraction of these cells

Figure 4 (a) Image of a portion of a silver-stained gel resulting f(serum-free) medium. The arrows indicate the location of two additionmESCs cultured with different concentrations of cyclophilin A (addedependent manner (P=0.03). (c) Fold growth on Day 2 ofmESCs after a(2 μg/ml), showing significantly decreased growth on addition of the C

(Fig. 3c). Additionally, cells grown in CM on Day 2 showed a~50% reduction in the percentage of unattached cells (Fig. 3d),again confirming that the autocrine factor(s) decrease the rateof cell death in mESC cultures.

Cyclophilin A is an autocrine growth factor inmESC cultures

We next sought to identify candidate autocrine factors thatmight be responsible for the observed growth supportiveeffects. Cystatin C, TIMP-1, osteopontin, and clusterin havebeen identified (via multiplexed ELISA) as proteins secretedfrom mESCs grown without LIF for 2 days (Singla andMcDonald, 2007) (at concentrations ofN1 ng/ml), whileTIMP-1 has also been found in the CM of mESCs grown with LIF(Guo et al., 2006). Additionally, transcripts for osteopontinand clusterin have been detected in mESCs grown in mediumcontaining LIF (Ivanova et al., 2002). FGF-4 has beenseparately identified as being secreted by mESCs, but doesnot affect the survival or cycling rate of mESCs (Wilder et al.,1997). We examined the effect on mESC growth ofsupplementing the media with recombinant versions ofeach of these proteins at levels likely to be present in theculture and observed either no effect or a growth inhibitoryeffect (Supplemental Information).

To determine what other proteins might be secreted bymESCs, we performed a gel analysis of mESC-conditionedmedium followed by LC/MS-MS. We detected two bands inthe CM lane of the gel that were not present in the control(Fig. 4a, Supplemental Information). The lower MW band wasfound to consist primarily of profilin-1 (UniProt MW 15 kDa),cystatin C (UniProt MW 15.5 kDa), and thioredoxin (UniProtMW 12kDa). The presence of profilin-1 is somewhatunexpected as it is an intracellular protein, although astudy on proteins secreted by neurospheres found profilin-2,also an intracellular protein, in the CM (Kato et al., 2006). Amajor component of the higher MW band was found to becyclophilin A (CypA), also known as peptidyl-prolyl isomeraseA (Ppia).

We next sought to determine whether supplementingmedia with any of these factors would increase survival. As

rom SDS-PAGE of medium conditioned by ABJ1 cells and controlal bands observed in the ABJ1 CM lane. (b) Fold growth on Day 2 ofd on Day 2), showing that growth increases in a concentration-ddition of no antibody, goat IgG, or a polyclonal antibody to CD147D147 Ab (P(CD147, normal IgG)=4e-6, P(CD147, PBS)=2e-8).

173Non-mitogenic survival-enhancing autocrine factors

noted above, we found that cystatin C was not growthsupportive for mESCs (Supplementary Information). CypAsecretion has been shown to be induced by oxidative stress(Jin et al., 2000; Seko et al., 2004), and CypA has been shownto protect various types of cells against cell death inducedby such stress (Jin et al., 2000; Boulos et al., 2007).Additionally, CypA-/- embryos show slightly reduced fertility(Colgan et al., 2004), indicating that it may play a protectiverole in embryos, and CypA protein is expressed in mouse ESCs(Colgan et al., 2000). Given this information, we decided toinvestigate the effects of CypA on mESC growth. Based on therelative intensity of staining of the bands in the gel, weestimated the concentration of CypA in the CM at ~100 ng/ml. When we added recombinant CypA to N2B27+LIF+BMP-4on Day 2, we observed a 9% increase in growth at 100 ng/ml,and a significant 21% increase at 200 ng/ml for ABJ1 mESCs(P=0.03, Fig. 4b). This increased growth on CypA additionwas confirmed in an independently derived cell line (Fig.S9b, D3 mESCs).

We next sought to block the cycophilin A receptor.Yurchenko et al. have identified CD147 (also known asBasigin and/or Emmprin) as a receptor for CypA (Yurchenkoet al., 2002). In particular, they determined that the proline180 and glycine 181 residues in the extracellular domain ofCD147 are critical for signaling induced by extracellularCypA. mESCs have been found to express the CypA receptorCD147 (Intoh et al., 2009), with expression decreasing as thecells differentiate. We also detected the presence of theCD147 protein in mESCs (Fig. S10). No monoclonal antibodyagainst mouse CD147 exists that has been shown tospecifically target only the CypA binding site on mouseCD147. However, when we added a polyclonal anti-CD147antibody on Day 2 to mESC cultures plated at 5000 cells/cm2,we did observe a decrease in the growth rate of ABJ1 mESCsby ~20% relative to normal goat IgG (Fig. 4c, Pb0.05). Theobserved decrease in growth on addition of the receptorantibody is comparable to the increase in growth producedby the action of mESC CM (Fig. 2a). Similar results wereobtained for D3 mESCs (Fig. S9). Together, these resultsdemonstrate that cyclophilin A is an autocrine survival factorin mESC cultures.

While previous studies have shown that CypA-CD147signaling in other cell types can lead to activation of theErk pathway (Boulos et al., 2007; Yurchenko et al., 2002),studies in mESCs have suggested that inhibition of the PI3k/Akt pathway, but not the Erk pathway, leads to changes incell cycle (Lianguzova et al., 2007) and rate of apoptosis(Gross et al., 2005). However, we did not detect activationor inhibition of Akt on addition of CypA to mESC cultures(Fig. S11), suggesting that CypA may improve survival inmESCs via an Akt-independent mechanism.

Discussion

Identification of the extrinsic factors required for the growthof self-renewing mESCs is interesting for two reasons: tocomplement efforts focused on understanding the extrinsicfactors required for self-renewal (Ying et al., 2003), andbecause improving the growth of ESCs in culture would aid intheir expansion for therapeutic purposes. That mESCs can beclonally propagated suggests that there may be no extrinsic

factors required for their growth, although autocrine loopscan operate at the single-cell level. The observation thatmESC growth is density dependent suggests that, even if notrequired, autocrine factors may promote the growth ofmESCs.

One potential approach for identifying such factors wouldbe to examine the in vivo compartment from which ESCs arederived, namely the inner cell mass. Indeed, withinreasonable limits, mouse, bovine, and porcine embryosshow improved survival at higher densities (Lane andGardner, 1992; Khurana and Niemann, 2000; Stokes et al.,2005), suggesting the presence of soluble signals, whichappear to act via the PI3-K pathway (O'Neill, 2008).Additionally, ICM cells have been found to undergo apoptosisin vivo; ~10–20% of cells in the mouse and human ICM areundergoing apoptosis on Day 5 postfertilization (Hardy,1997). Again, the percentage of dead cells has been shownto be inversely correlated with the total number of cells inthe blastocyst, suggesting that antiapoptotic factors mayplay a role in modulating this process. However, ESCs maynot be identical to cells of the ICM (Tang et al., 2010).

We have used an alternate approach for identifying ESCgrowth/survival factors, which is to identify factors secreteddirectly from mESCs, in vitro. While two previous studieshave examined the mESC secretome (Guo et al., 2006; Singlaand McDonald, 2007), they both used ELISA-basedapproaches, which are limited to assaying factors for whichantibodies are available. Instead, we used mass spectrom-etry to identify secreted factors present in the media thatcould be candidate autocrine molecules. From this list offactors, we identified cyclophilin A as an autocrine ligand inmESC cultures. It is possible that other factors in addition toCypA act in an autocrine fashion to affect the growth ofmESCs. The approach taken in this study helps guide thediscovery of future factors. For instance, the lack of changein the cell cycle distribution across cell density or with CM(Fig. 3, Fig. S6) makes it likely that additional autocrinegrowth factors will also be survival factors, rather thanmitogens.

Cyclophilin A was isolated in 1984 as a protein with highaffinity for the immunosuppressive drug cyclosporin A(Handschumacher et al., 1984) (hence the name cyclo-philinA). Additionally, most cyclophilins possess peptidyl-prolylisomerase activity, i.e., they help in the folding of proteins(Wang and Heitman, 2005). While the above two functionsare associated with intracellular CypA, subsequent studiesfound that CypA is also present extracellularly (Sherry et al.,1992). While CypA appears to be expressed ubiquitously(Wang and Heitman, 2005), recently it has been found thatCypA is overexpressed in several cancers (Li et al., 2006;Yang et al., 2007; Li et al., 2008), suggesting that it may playa role in carcinogenesis.

It has been shown that some of the actions of extracellularCypA are mediated through its binding to CD147 (Yurchenkoet al., 2002; Li et al., 2006), which is also highly expressed insome cancers (Riethdorf et al., 2006). However, it is notclear whether the growth-enhancing effect of cyclophilin A ismediated by CD147 alone. One group reported that a CD147antibody blocked the progrowth effect of CypA in humanpancreatic cancer cells (Li et al., 2006), but not in humanvascular smooth muscle cells (Yang et al., 2005), suggestingthat CypA may act through more than just CD147. With

174 N. Mittal, J. Voldman

regard to its expression in mESCs, CD147 has been found tobe expressed in the inner cell mass of mouse blastocysts(Igakura et al., 1998), and it has been recently found to beexpressed more strongly in mESCs than in their differentiat-ed progeny (Intoh et al., 2009).

Interestingly, CypA also appears to be produced by humanembryonic stem cells (hESCs; (Bendall et al., 2009),supplemental data), and the CD147 precursor protein wasdetected in the plasma membrane of hESCs via massspectrometry (Dormeyer et al., 2008), raising the possibilitythat CypA plays a protective role in hESC cultures as well.This may have useful implications for improving the cloningefficiency of hESCs, which can be as low as 1% (Ware et al.,2006). Additionally, if CypA-CD147 signaling is important forthe survival of hESCs and other human stem cells it may beimportant for the survival of human cancer stem cells aswell.

Materials and methods

Cell culture and staining

ABJ1 (with Oct4-GFP reporter, kindly provided by GeorgeDaley's lab), CCE, and D3 mESCs were cultured in TCPS dishes(Nunc) coated with 0.1% gelatin (Millipore) in a 37 °Chumidified environment with 7.5% CO2. For maintenance ofmESCs, we fed cells daily and passaged every other day at 10000 cells/cm2.

Serum-containing media consisted of DMEM (Invitrogen)supplemented with 15% ES-qualified fetal bovine serum(Invitrogen), 4 mM L-glutamine (Invitrogen), 1 mM nonessen-tial amino acids (Invitrogen), 50 U/mL penicillin, 50 μg/mLstreptomycin (Invitrogen), 100 μMβ-mercaptoethanol (Sigma-Aldrich), and 1000 U/ml leukemia inhibitory factor (LIF,Millipore). Serum-free medium consisted of N2B27 supple-mented with LIF and BMP-4 (Ying et al., 2003). N2B27 is a 1:1mixture of N2 medium and B27 medium: the N2 mediumconsists of 25 μg/ml insulin, 100 μg/ml transferrin, 20 nMprogesterone, 20 nM sodium selenite, 100 μM putrescine (allfrom Sigma-Aldrich), and 50 μg/ml BSA (Invitrogen), in DMEM/F12 (Invitrogen); B27 medium was made by adding B27supplement (Invitrogen) to Neurobasal medium (Invitrogen).N2B27was supplementedwith 1000U/ml LIF (Millipore) and 10ng/ml rhBMP-4 (R&D Systems). Cells in serum-free mediumwere dissociated from the surface with TrypLE Select(Invitrogen) instead of 0.25% trypsin-EDTA (used for serum-containing media). Cells were allowed to adapt to serum-freemedium by culturing them for 10–15 passages in this mediumprior to performing experiments in this medium. Experimentswere performed between 3 and 7 passages after thawing cells.

Recombinant mouse (rm) cystatin C, rmTIMP-1, rmOsteo-pontin and rmClusterin, rhFGF-4, and rhcyclophilin A werepurchased from R&D Systems. Human cyclophilin A (CypA)shares a 96% sequence similarity with mouse CypA, whilehuman FGF-4 shares an 80% sequence similarity with mouseFGF-4. Goat anti-mouse CD147 polyclonal antibody and goatIgG control were purchased from Santa Cruz Biotechnology.

Staining with YO-PRO-1 iodide (Invitrogen) was performedfor 15 min at 37 °C, at 2 μM final concentration. TUNEL assayreagents were purchased from Roche and staining wasperformed according to the manufacturer's protocols.

Growth and colony-forming efficiency measurementsfor various densities

Growth versus density experiments were performed in 12-well plates (Nunc, Delta surface), in which the wells werelarge enough to allow for uniform seeding. Cells werecounted using a Coulter counter (Beckman Z2). Measuredcell numbers were adjusted by subtracting backgroundcounts from control media with trypsin/TrypLE but notcontaining cells. To allow for the accumulation of autocrinefactors and to avoid a potential second lag phase associatedwith cells having to adapt to fresh medium, we did not feedcells during the 48-h assay period.

Colony-forming efficiency was measured using a micro-fluidic device that allows plating of cells at defined densities(Supplemental Information). The initial number of cells andthe number of colonies at the end of Day 2 were countedusing phase-contrast microscopy (Zeiss Axiovert 200). Datashown are from three experiments. As above, cells were notfed during the 48-h assay period.

Conditioned medium

We prepared conditioned medium by collecting mediumfrom a dish and filtering it through a 0.25-μm filter (with asyringe) to remove floating cellular debris. The control wasincubated for the same length of time in a similar dish as thecell culture, and filtered in an identical manner. We thendialyzed equal amounts by weight of CM and control using a3-kDa centrifugal filter unit (Millipore), replenished theN3-kDa fraction with freshb3-kDa nutrient components, andagain checked that both media had the same final weight.We also added LIF (1000 U/ml) and BMP-4 (10 ng/ml) tocompensate for any potential depletion during culture, toensure that cells would be maintained in a self-renewingstate. For Day 1 growth/death assessment, CM was collectedat the end of Day 1 from cells plated at 15 000 cells/cm2. ForDay 2 growth/death assessment, CM was collected at the endof Day 2 from cells that were plated at 30 000 cells/cm2, andthen fed at 24 h with fresh medium. The CM was applied tocells plated at 1000–2000 cells/cm2. Experiments wereperformed three times, in triplicate each time. Data fromdifferent days were normalized using the control data.

Cell cycle analysis

Cells were detached using TrypLE (Invitrogen), permeabi-lized with 0.1% Triton X, and stained with 50 μg/mlpropidium iodide (Sigma) in the presence of 500 μg/mlRibonuclease A (Sigma). The samples were then analyzedusing a FACScan cytometer (Becton Dickinson). The dataobtained were analyzed using ModFit LT (Verity SoftwareHouse) to determine the percentages of cells in the variousphases of the cell cycle.

Statistical analysis of density-dependent growth

ANOVA was performed on fold growth data for Day 1 to obtainthe presented P value. For comparison of growth for Day 2,we used ANOVA on Day 2 growth data for the first threedensities shown, i.e. 250, 1000, and 2250 cells/cm2, which

175Non-mitogenic survival-enhancing autocrine factors

corresponds to the increasing region of the growth curve forthis period of time. Since cells at this density show nodifference in fold growth at the end of Day 1, it is enough tojust compare the growth at the end of Day 2 to make astatement about growth on Day 2. Adjusting for the Day 1growth led to similar results (Fig. S2).

SDS-PAGE and mass spectrometric analyses

Cells were plated at 30 000 cells/cm2, fed on Day 1, and CMwas collected at the end of Day 2. Control medium (RM) wassimilarly incubated, but without cells. Four milliliters of CMand RM was concentrated to approximately 100 μl using a 3-kDa centrifugal filter unit (Millipore). CM and RM were mixed1:1 with Laemmli buffer (Bio-Rad), heated at 95 °C for 5 min,and run on 12 or 18% polyacrylamide gels (Bio-Rad) at 15 mA.Gels were stained using a silver stain (SilverQuest, Invitro-gen) for visualization. For mass spectrometric analyses, 15ml CM and RM was concentrated using a 3-kDa centrifugalfilter unit (Millipore) first, followed by drying in a SpeedVacto approximately 15 μl. Following SDS-PAGE, gels werestained using Coomassie blue (SimplyBlue, Invitrogen).Appropriate bands were analyzed using LC/MS-MS at theBiopolymers Laboratory at the Koch Institute for IntegrativeCancer Research (MIT).

Cell detachment analysis

Cell culture supernatant (including detached cells) wasgently triturated once while in the well plate and thencounted using a Coulter counter (Beckman Z2, 10–20 μMsize range). To measure the disintegration rate of theunattached cells we incubated culture supernatant (con-taining unattached cells) for 24 h and counted the numberof cells at the beginning and end of the incubation. Wefound that detached cells disintegrated with a half-life of10.5±0.9 h (Supplemental Information). Thus the numberof cells that detached during Day 2 was determined asfollows: number of cells detached at the end of Day 2 (i.e.,over Days 1 and 2) – 0.21×number of cells detached at theend of Day 1.

Supplementary material related to this article can befound online at doi:10.1016/j.scr.2010.10.001.

Acknowledgments

We thank members of the Voldman lab, especially KatarinaBlagovic, for their suggestions and comments. We also thankthe Biopolymers Laboratory at the Koch Institute forIntegrative Cancer Research (MIT) for performing the massspectroscopic analysis presented in this work. This work wassupported by the NIH (EB007278).

References

Bendall, S.C., Hughes, C., Campbell, J.L., Stewart, M.H., Pittock,P., Liu, S., Bonneil, E., Thibault, P., Bhatia, M., Lajoie, G.A.,2009. An enhanced mass spectrometry approach reveals humanembryonic stem cell growth factors in culture. Mol. Cell.Proteomics 8, 421–432.

Boulos, S., Meloni, B.P., Arthur, P.G., Majda, B., Bojarski, C.,Knuckey, N.W., 2007. Evidence that intracellular cyclophilin Aand cyclophilin A/CD147 receptor-mediated ERK1/2 signallingcan protect neurons against in vitro oxidative and ischemicinjury. Neurobiol. Dis. 25, 54–64.

Brewer, G.J., Torricelli, J.R., Evege, E.K., Price, P.J., 1993.Optimized survival of hippocampal-neurons in B27-supplementedNeurobasal(Tm), a new serum-free medium combination. J.Neurosci. Res. 35, 567–576.

Colgan, J., Asmal, M., Luban, J., 2000. Isolation, characteriza-tion and targeted disruption of mouse Ppia: cyclophilin A isnot essential for mammalian cell viability. Genomics 68,167–178.

Colgan, J., Asmal, M., Neagu, M., Yu, B., Schneidkraut, J., Lee, Y.,Sokolskaja, E., Andreotti, A., Luban, J., 2004. Cyclophilin Aregulates TCR signal strength in CD4(+) T cells via a proline-directed conformational switch in Itk. Immunity 21, 189–201.

Dormeyer, W., van Hoof, D., Braam, S.R., Heck, A.J.R., Mummery,C.L., Krijgsveld, J., 2008. Plasma membrane proteomics ofhuman embryonic stem cells and human embryonal carcinomacells. J. Proteome Res. 7, 2936–2951.

Fattman, C.L., Schaefer, L.M., Oury, T.D., 2003. Extracellularsuperoxide dismutase in biology and medicine. Free Radic. Biol.Med. 35, 236–256.

Gross, V.S., Hess, M., Cooper, G.M., 2005. Mouse embryonic stemcells and preimplantation embryos require signaling through thephosphatidylinositol 3-kinase pathway to suppress apoptosis.Mol. Reprod. Dev. 70, 324–332.

Guo, Y., Hangoc, G., Bian, H., Pelus, L.M., Broxmeyer, H.E., 2005.SDF-1/CXCL12 enhances survival and chemotaxis of murineembryonic stem cells and production of primitive and definitivehematopoietic progenitor cells. Stem Cells 23, 1324–1332.

Guo, Y., Graham-Evans, B., Broxmeyer, H.E., 2006. Murineembryonic stem cells secrete cytokines/growth modulators thatenhance cell survival/anti-apoptosis and stimulate colony for-mation of murine hematopoietic progenitor cells. Stem Cells 24,850–856.

Han, H.J., Han, J.Y., Heo, J.S., Lee, S.H., Lee, M.Y., Kim, Y.H.,2007. ANG II-stimulated DNA synthesis is mediated by ANG IIreceptor-dependent Ca2+/PKC as well as EGF receptor-dependentPI3K/Akt/mTOR/p70S6K1 signal pathways in mouse embryonicstem cells. J. Cell. Physiol. 211, 618–629.

Handschumacher, R.E., Harding, M.W., Rice, J., Drugge, R.J., 1984.Cyclophilin—a specific cytosolic binding-protein for cyclosporin-A. Science 226, 544–547.

Hardy, K., 1997. Cell death in the mammalian blastocyst. Mol. Hum.Reprod. 3, 919–925.

Heo, J.S., Lee, Y.J., Han, H.J., 2006. EGF stimulates proliferationof mouse embryonic stem cells: involvement of Ca2+ influx andp44/42 MAPKs. Am. J. Physiol. Cell Physiol. 290, C123–C133.

Igakura, T., Kadomatsu, K., Kaname, T., Muramatsu, H., Fan, Q.W.,Miyauchi, T., Toyama, Y., Kuno, N., Yuasa, S., Takahashi, M.,Senda, T., Taguchi, O., Yamamura, K., Arimura, K., Muramatsu,T., 1998. A null mutation in basigin, an immunoglobulinsuperfamily member, indicates its important roles in peri-implantation development and spermatogenesis. Dev. Biol.194, 152–165.

Intoh, A., Kurisaki, A., Yamanaka, Y., Hirano, H., Fukuda, H.,Sugino, H., Asashima, M., 2009. Proteomic analysis of membraneproteins expressed specifically in pluripotent murine embryonicstem cells. Proteomics 9, 126–137.

Ivanova, N.B., Dimos, J.T., Schaniel, C., Hackney, J.A., Moore, K.A.,Lemischka, I.R., 2002. A stem cell molecular signature. Science298, 601–604.

Jin, Z.G., Melaragno, M.G., Liao, D.F., Yan, C., Haendeler, J., Suh,Y.A., Lambeth, J.D., Berk, B.C., 2000. Cyclophilin A is a secretedgrowth factor induced by oxidative stress. Circ. Res. 87,789–796.

176 N. Mittal, J. Voldman

Kato, T., Heike, T., Okawa, K., Haruyama, M., Shiraishi, K.,Yoshimoto, M., Nagato, M., Shibata, M., Kumada, T., Yamanaka,Y., Hattori, H., Nakahata, T., 2006. A neurosphere-derivedfactor, cystatin C, supports differentiation of ES cells into neuralstem cells. Proc. Natl Acad. Sci. USA 103, 6019–6024.

Keenan, J., Pearson, D., Clynes, M., 2006. The role of recombinantproteins in the development of serum-free media. Cytotechnol-ogy 50, 49–56.

Khurana, N.K., Niemann, H., 2000. Effects of oocyte quality, oxygentension, embryo density, cumulus cells and energy substrates oncleavage and morula/blastocyst formation of bovine embryos.Theriogenology 54, 741–756.

Lane, M., Gardner, D.K., 1992. Effect of incubation volume andembryo density on the development and viability of mouseembryos in vitro. Hum. Reprod. 7, 558–562.

Lauffenburger, D., Cozens, C., 1989. Regulation of mammalian-cellgrowth by autocrine growth-factors—analysis of consequencesfor inoculum cell-density effects. Biotechnol. Bioeng. 33,1365–1378.

Li, M., Zhai, Q.H., Bharadwaj, U., Wang, H., Li, F., Fisher, W.E.,Chen, C.Y., Yao, Q.Z., 2006. Cyclophilin A is overexpressed inhuman pancreatic cancer cells and stimulates cell proliferationthrough CD147. Cancer 106, 2284–2294.

Li, Z.Y., Zhao, X., Bai, S.J., Wang, Z., Chen, L.J., Wei, Y.Q., Huang,C.H., 2008. Proteomics identification of cyclophilin a as apotential prognostic factor and therapeutic target in endome-trial carcinoma. Mol. Cell. Proteomics 7, 1810–1823.

Lianguzova, M.S., Chuykin, I.A., Nordheim, A., Pospelov, V.A., 2007.Phosphoinositide 3-kinase inhibitor LY294002 but not serumwithdrawal suppresses proliferation of murine embryonic stemcells. Cell Biol. Int. 31, 330–337.

Liu, N., Lu, M., Tian, X.M., Han, Z.C., 2007. Molecular mechanismsinvolved in self-renewal and pluripotency of embryonic stemcells. J. Cell. Physiol. 211, 279–286.

Michiels, C., Raes, M., Toussaint, O., Remacle, J., 1994. Importanceof Se-glutathione peroxidase, catalase, and Cu/Zn-Sod for cell-survival against oxidative stress. Free Radic. Biol. Med. 17,235–248.

Ogawa, K., Saito, A., Matsui, H., Suzuki, H., Ohtsuka, S., Shimosato,D., Morishita, Y., Watabe, T., Niwa, H., Miyazono, K., 2007.Activin-nodal signaling is involved in propagation of mouseembryonic stem cells. J. Cell Sci. 120, 55–65.

O'Neill, C., 2008. Phosphatidylinositol 3-kinase signaling in mamma-lian preimplantation embryo development. Reproduction 136,147–156.

Orford, K.W., Scadden, D.T., 2008. Deconstructing stem cell self-renewal: genetic insights into cell-cycle regulation. Nat. Rev.Genet. 9, 115–128.

Qi, X.X., Li, T.G., Hao, J., Hu, J., Wang, J., Simmons, H., Miura, S.,Mishina, Y., Zhao, G.Q., 2004. BMP4 supports self-renewal ofembryonic stem cells by inhibiting mitogen-activated proteinkinase pathways. Proc. Natl Acad. Sci. USA 101, 6027–6032.

Riethdorf, S., Reimers, N., Assmann, V., Kornfeld, J.W., Terrac-ciano, L., Sauter, G., Pantel, K., 2006. High incidence ofEMMPRIN expression in human tumors. Int. J. Cancer 119,1800–1810.

Schratt, G., Weinhold, B., Lundberg, A.S., Schuck, S., Berger, J.,Schwarz, H., Weinberg, R.A., Ruther, U., Nordheim, A., 2001.Serum response factor is required for immediate-early geneactivation yet is dispensable for proliferation of embryonic stemcells. Mol. Cell. Biol. 21, 2933–2943.

Seko, Y., Fujimura, T., Taka, H., Mineki, R., Murayama, K., Nagai,R., 2004. Hypoxia followed by reoxygenation induces secretionof cyclophilin A from cultured rat cardiac myocytes. Biochem.Biophys. Res. Commun. 317, 162–168.

Sherry, B., Yarlett, N., Strupp, A., Cerami, A., 1992. Identificationof cyclophilin as a proinflammatory secretory product oflipopolysaccharide-activated macrophages. Proc. Natl Acad.Sci. USA 89, 3511–3515.

Singla, D.K., McDonald, D.E., 2007. Factors released from embryonicstem cells inhibit apoptosis of H9c2 cells. Am. J. Physiol. HeartCirc. Physiol. 293, H1590–H1595.

Singla, D.K., Singla, R.D., McDonald, D.E., 2008. Factors releasedfrom embryonic stem cells inhibit apoptosis in H9c2 cells throughPI3K/Akt but not ERK pathway. Am. J. Physiol. Heart Circ.Physiol. 295, H907–H913.

Stead, E., White, J., Faast, R., Conn, S., Goldstone, S., Rathjen, J.,Dhingra, U., Rathjen, P., Walker, D., Dalton, S., 2002.Pluripotent cell division cycles are driven by ectopic Cdk2, cyclinA/E and E2F activities. Oncogene 21, 8320–8333.

Stokes, P.J., Abeydeera, L.R., Leese, H.J., 2005. Development ofporcine embryos in vivo and in vitro; evidence for embryo 'crosstalk' in vitro. Dev. Biol. 284, 62–71.

Takahashi, K., Mitsui, K., Yamanaka, S., 2003. Role of ERas inpromoting tumour-like properties in mouse embryonic stemcells. Nature 423, 541–545.

Tang, F.C., Barbacioru, C., Bao, S.Q., Lee, C., Nordman, E., Wang,X.H., Lao, K.Q., Surani, M.A., 2010. Tracing the derivation ofembryonic stem cells from the inner cell mass by single-cell RNA-Seq analysis. Cell Stem Cell 6, 468–478.

Todorova, M.G., Soria, B., Quesada, I., 2008. Gap junctionalintercellular communication is required to maintain embryonicstem cells in a non-differentiated and proliferative state. J. Cell.Physiol. 214, 354–362.

Viswanathan, S., Benatar, T., Rose-John, S., Lauffenburger, D.A.,Zandstra, P.W., 2002. Ligand/receptor signaling threshold (LIST)model accounts for gp130-mediated embryonic stem cell self-renewal responses to LIF and HIL-6. Stem Cells 20, 119–138.

Wang, P., Heitman, J., 2005. The cyclophilins. Genome Biol. 6.Ware, C.B., Nelson, A.M., Blau, C.A., 2006. A comparison of NIH-

approved human ESC lines. Stem Cells 24, 2677–2684.Wilder, P.J., Kelly, D., Brigman, K., Peterson, C.L., Nowling, T.,

Gao, Q.S., McComb, R.D., Capecchi, M.R., Rizzino, A., 1997.Inactivation of the FGF-4 gene in embryonic stem cells alters thegrowth and/or the survival of their early differentiated progeny.Dev. Biol. 192, 614–629.

Yang, H., Li, M., Chai, H., Yan, S.Y., Lin, P., Lumsden, A.B., Yao, Q.Z., Chen, C.Y., 2005. Effects of cyclophilin A on cell proliferationand gene expressions in human vascular smooth muscle cells andendothelial cells. J. Surg. Res. 123, 312–319.

Yang, H.R., Cben, H., Yang, J.C., Qiao, S.Y., Zhao, S.Y., Yu, L.,2007. Cyclophilin A is upregulated in small cell lung cancer andactivates ERK1/2 signal. Biochem. Biophys. Res. Commun. 361,763–767.

Ying, Q.L., Nichols, J., Chambers, I., Smith, A., 2003. BMP induction ofId proteins suppresses differentiation and sustains embryonic stemcell self-renewal in collaboration with STAT3. Cell 115, 281–292.

Yurchenko, V., Zybarth, G., O'Connor, M., Dai, W.W., Franchin, G.,Hao, T., Guo, H.M., Hung, H.C., Toole, B., Gallay, P., Sherry, B.,Bukrinsky, M., 2002. Active site residues of cyclophilin A arecrucial for its signaling activity via CD147. J. Biol. Chem. 277,22959–22965.