Comparative Evaluation of Various Human Feeders for Prolonged Undifferentiated Growth of Human Embryonic Stem Cells



Human embryonic stem (hES) cells are typically derived and serially propagated on inactivated murine embryonic fibroblast (MEF) feeders. The use of MEFs and other components of animal origin in the culture media for hES cell support substantially elevates the risk of contaminating these cell lines with infectious agents of animal origin thereby severely limiting their potential for clinical application. We have previously shown that it is possible to derive and establish new hES cell lines in a xeno-free culture system using human fetal muscle fibroblast feeders. In this report, we have comparatively evaluated a panel of 11 different human adult, fetal, and neonatal feeders for hES cell support and have ranked them as supportive and non-supportive. We report that two adult skin fibroblast cell lines established in-house from abdominal skin biopsies supported prolonged undifferentiated hES cell growth for over 30 weekly passages in culture. Furthermore, hES cell lines cultured on adult skin fibroblast feeders retain hES cell morphology and remain pluripotent. Also, differences in feeder support exist between human cell types and sources. The use of human adult skin feeders is convenient for hES cell support given the ease of obtaining skin biopsies.


Pluripotent human embryonic stem (hES) cells promise to revolutionize the future of reparative medicine through the development of cell and tissue replacement therapies. hES cells were first isolated in 1994 [1, 2] while the first continuous hES lines were established in 1998 [3]. Our understanding of hES cell biology is still in its infancy, but this is likely to change once hES cell lines become widely and routinely available to all researchers. Nevertheless, hES cell research needs to be pursued aggressively if we are to quickly realize the full therapeutic potential of reparative hES cell therapy. Specifically, advances must be made in culture techniques to obtain purer and safer populations of functionally normal undifferentiated hES and hES-directed cell types. All the current 78 hES cell lines registered with the National Institute of Health (NIH) have been derived and propagated by direct contact on murine embryonic fibroblast (MEF)s and in the presence of culture medium containing animal-based ingredients. The development and refinement of xeno-free culture systems that avoid the risk of hES cell contamination with adventitious agents and that maintain pure undifferentiated hES cell populations which are amenable to the massive scale-up of hES cell numbers are thus important. Also, functional genomics data need to be generated for a better understanding of the genetic pathways that regulate self-renewal and differentiation. The overcoming of these bottlenecks will contribute to bringing hES cell biology faster from bench to bedside. hES cell lines are commonly derived and propagated in vitro with defined culture media containing 20% calf serum, bovine insulin, and porcine transferrin or with commercially available Knockout (KO) culture media containing bovine serum albumin and supplemented with basic fibroblast growth factor (bFGF) [35]. hES cell lines are also feeder-cell dependent and require either inactivated early passaged MEF feeders [3, 4] or inactivated human fibroblast feeders [6] for attachment, and nourishment, and to keep them undifferentiated. Xu et al. [5] have also shown that undifferentiated hES colonies can be successfully maintained and propagated on noncellular matrices (laminin and Matrigel) for over 100 population doublings in the presence of MEF-conditioned culture media, but disappointingly these matrices are also of animal origin. We recently derived an hES cell line on human feeders in the absence of xenoproteins and showed that human feeders were superior to feeder-free matrices for prolonged undifferentiated hES cell growth [6]. We showed that hES cell lines can be propagated with very low spontaneous differentiation and for prolonged periods on human fetal muscle (FM), human fetal skin (FS), and human adult fallopian tubal fibroblast feeders (AFT) with defined culture media containing 20% fetal calf serum (H1 media) or with commercially available KO culture media. hES colonies on these human fibroblast feeders are morphologically thinner, flatter, have straight defined boundaries giving the colony an angular shape and have longer interpassage intervals of 8 to 9 days as compared to typical rounder, thicker colonies on MEF feeders with 7-day passage intervals [6]. We have also demonstrated that new hES cell lines can be established in xeno-free conditions with prolonged undifferentiated growth for at least 10 passages in vitro when human serum is used to replace fetal calf serum in the make-up of hES culture media [6]. However, we observed that prolonged use of human serum beyond the tenth passage in our culture media leads to increasing differentiation rates. This may be attributed to batch-to-batch variation in the serum samples we obtained commercially and from patients or the presence of certain molecules in human sera which induce and accelerate differentiation. Heat inactivation of human sera did not appear to have a beneficial effect in reducing differentiation rates. In contrast, two hES cell lines, HES-3 and HES-4 (proprietary hES cell lines from ES Cell International; Singapore; have now been propagated on FM feeders for over 50 passages (150 population doublings) with no noticeable increase in differentiation when either KO media or standard H1 culture media containing 20% fetal calf serum was used. To examine if there were differences in support between feeder sources, we have evaluated a panel of in-house derived and commercial human fetal and adult fibroblast feeders for their ability to support prolonged undifferentiated hES cell growth.

Materials and Methods

Feeder Layers

In-House Derived Feeders

Seven pathologically normal in-house derived human fibroblast cell lines were evaluated for their ability to support hES cell growth. Human FM, human FS and human AFT primary cultures were established as previously described [6, 7]. Human adult muscle (AM), human adult skin (AS), human adult glandular endometrium (AGE), and human adult stromal endometrium (ASE) biopsies were aseptically collected in a sterile balanced salt solution directly from the abdominal region or pre-menopausal inner lining of the uterus from consenting patients. AS samples were provided from two different patients, and two separate AS fibroblast cell lines were established. AM samples were skeletal muscle from the same patients who donated skin samples. AGE and ASE primary explants were obtained from one and the same patient, and this particular patient was different from the patients supplying AS and AM. Muscle, skin and endometrium explants (1 mm3) were placed inside 25 cm2 tissue culture flasks (Nunclon; Roskilde; Copenhagen, Denmark) and separately grown in the presence of human feeder establishment medium or Chang's medium (Irvine Scientific; Santa Ana, CA; at 37°C in a 5% CO2 in air atmosphere. Human feeder establishment medium (xeno-free) contained 50% Dulbecco's modified Eagle's medium (DMEM), 50% human serum, 1X antimycotics and 2 mmol/l L-glutamine (Invitrogen; Carlsbad, CA; The flasks were left undisturbed for 3 days after which a small aliquot of fresh human feeder establishment medium or Chang's medium was used to top up the existing media every 48 hours. After 10 to 15 days, confluent primary monolayers were established. All primary cultures were serially passaged mechanically using a rubber policeman or with 0.05% trypsin-EDTA (Invitrogen) in human feeder maintenance medium or Chang's medium and then cryopreserved. Human feeder maintenance medium (xeno-free) contained 90% high-glucose DMEM, 10% fetal calf serum or 10% human serum, 2 mmol/l L-glutamine, 50 IU/ml penicillin and 50 μg/ml streptomycin (Invitrogen). Ethical approval was obtained, and the patients donating material were negative for HIV 1 and 2 and hepatitis B virus. For the various experiments, frozen human feeders were thawed and expanded in human feeder maintenance medium or Chang's medium. All fibroblast feeders were expanded and evaluated at the 4th, 6th and 8th passages.

Commercial Feeders

Human FS (D551/CCL-10, American Type Culture Collection [ATCC]), human fetal lung (MRC-5 and WI-38, National Institute for Biological Standards and Control), and neonatal foreskin (CCL-2552, ATCC) were purchased frozen, thawed and expanded in the same human feeder maintenance medium as the in-house derived feeders for this study. Expanded passages were also frozen for future studies. All these commercial feeders had fibroblast morphology in vitro and were pathologically and karyotypically normal. MEF feeders B-83 and B-84 (Institute of Reproduction and Development, Monash University; Victoria, Australia; were used as controls. All commercial feeders including MEFs were used at the same passages as the in-house derived feeders.

Propagation of hES Cells on Human Feeders

Feeder Layer Preparation

Confluent monolayers of fibroblast cells grown in human feeder maintenance medium were treated with mitomycin-C (Sigma; Chicago, IL; for 2.5 hours as previously described [4, 6]. The monolayer was washed with phosphate-buffered saline, (PBS; Invitrogen) three times and dispersed mechanically or with trypsin to produce a single-cell suspension of feeder cells and plated on plastic single-well tissue culture dishes (Becton-Dickinson; Franklin Lakes, NJ; In each 1 ml dish, 180,000-220,000 mitomycin-C-treated feeder cells were plated. The optimum seeding density seemed to be 185,000 cells per dish. Culture dishes were pre-coated with 0.1% weight/volume gelatin (Sigma) to promote feeder attachment. All inactivated feeder monolayers for the in-house derived and commercial human feeders and the MEF controls were plated at the same density.

Serial Passaging of hES Cell Cultures

HES-3 and HES-4 (ES Cell International) of different ethnic backgrounds were grown on inactivated FM, FS, AFT, AS, AM, AGE, ASE (in-house); D551, MRC-5, WI-38, CCL-2552 (commercial) and MEF B-83, B-84 (controls) feeder layers with H1 medium or KO hES media. H1 medium contained 80% high-glucose DMEM, 20% hyclone defined fetal calf serum (Hyclone; Logan UT;, 2 mmol/l L-glutamine, 50 IU/ml penicillin, and 50 μg/ml streptomycin, 1X non-essential amino acids, 1X insulin-transferrin-selenium G supplement, and 0.1 mmol/l α-mercaptoethanol (Invitrogen). KO medium (xeno-free) contained 80% KO-DMEM, 20% KO serum replacement (Invitrogen), 4-8 ng/ml bFGF (Sigma) and similar supplements as H1 medium. Evaluation of all culture systems was conducted in tandem, and cultures were inspected daily. The process of mechanical cutting and dispase (Sigma) digestion for passaging small clumps of hES cells to fresh feeders was performed regularly for all types of feeders at approximately 7- to 8-day intervals before the onset of hES cell differentiation using established protocols [4, 6]. Solid dispase (0.17 g) (Sigma) was dissolved in 10 ml H1 or KO medium and dissected colonies incubated with this solution for 40-60 seconds to aid in feeder detachment. The growth characteristics of hES cell colonies in terms of shape, thickness, fragility, and extent of differentiation were carefully recorded at low and high magnifications. hES cell lines grown on FM, AS and D551 human feeders were successfully cryopreserved using the protocols developed by Reubinoff et al. [8].

hES Cell Characterization

Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR) Analysis

Total RNA was extracted from hES cells with Trizol™ reagent (Invitrogen) following the manufacturer's protocol. First-strand synthesis was performed using the SuperScript™ first-strand synthesis system for RT-PCR (Invitrogen). One μl of first-strand reaction was used for each 50 μl PCR reaction together with 50 pmol of forward and reverse primers. Primer data: ACTB (X00351) primers (sense: 5′-TGGCACC ACACCTTCTACAATGAGC-3′; antisense: 5′-GCACAGC TTCTCCTTAATGTCACGC-3′) [9]. FGF4 (M17446) primers (sense: 5′-CTACAACGCTACGAGTCCTACA-3′; antisense: 5′-GTTGCACCAGAAAAGTCAGAGTTG-3′) [10]. HESX1 (AF059734) primers (sense: 5′-GGATTTCAT TCCCTAGCGTGG-3′; antisense: 5′-GTGATTCTCTATG GGACCTTTTC-3′). TERT (AF015950) primers (sense 5′-CGGAAGAGTGTCTGGAGCAA-3′; antisense: 5′-GGATG AAGCGGAGTCTGGA-3′) [5]. Pit oct unc domain class 5 transcription factor 1 (POU5F1) (Z11899) primers (sense: 5′-CGRGAAGCTGGAGAAGGAGAAGCTG-3′; antisense 5′-AAGGGCCGCAGCTTACACATGTTC-3′) [9]. REX1 (AK056719) primers (sense: 5′-GCGTACGCAAATTAAA GTCCAGA-3′; antisense 5′-CAGCATCCTAAACAGCTC GCAGAAT-3′) [10]. PCR cycles consisted of an initial denaturation step at 94°C for 5 minutes followed by 35 amplification cycles of 30 seconds 94°C denaturation, 30 seconds 54°C annealing and 60 seconds 72°C extension. A final extension at 72°C for 5 minutes was included. Annealing temperature for the human telomerase reverse transcriptase (hTERT reaction) and Rex-1 primers was 60°C and 56°C respectively. Ten μl of the PCR product were run on a 1% agarose gel stained with ethidium bromide (EtBr), and visualized with a UV table. Negative controls for each target gene were performed in separate tubes with templates prepared from a first-strand reaction where RT was omitted.

Early Differentiation Markers (Endoderm, Mesoderm, Ectoderm)

Pre-designed Assays-on-Demand™ TaqMan™ probes and primer pairs for POU5F1 (NM 002701.1), the endoderm early differentiation marker alpha fetoprotein (AFP) (NM 001134), the mesoderm early differentiation marker bone morphogenetic protein 4 (BMP4) (NM 13085), and the ectoderm early differentiation marker neurogenic differentiation 1 (NEUROD1) (NM 002500) were obtained from Applied Biosystems Incorporated (ABI) (Foster City, CA; Total RNA was extracted from undifferentiated and differentiated hES colonies with Trizol™ reagent and reverse transcribed using the SuperScript II™ first-strand synthesis system. Total RNA was quantified using a ND-1000 spectrophotometer (NanoDrop Technologies; Rockland, DE; and contaminating genomic DNA removed with DNA-free™ reagent (Ambion; Austin, TX; before reverse transcription. Real-time quantitative RT-PCR (qRT-PCR) analysis was conducted using the ABI PRISM® 7000 Sequence Detection System. After an initial incubation step for 2 minutes at 50°C and denaturation for 10 minutes at 95°C, qRT-PCR was carried out using 40 cycles of PCR ( 95°C for 15 seconds, 60°C for 60 seconds). Equal amounts of input RNA were used for all qRT-PCR reactions, reactions were performed in triplicate, and 18S recombinant RNA levels served as internal controls.


hES colonies were incubated with 50 μg/ml colcemid solution (Invitrogen) for 1.5 hours at 37°C and in a 5% CO2 in air atmosphere. Cells were trypsinized, washed with PBS (Invitrogen), and pellets resuspended and incubated with 0.075M potassium chloride (KCl for 30 minutes at 37°C).

Following treatment with hypotonic solution, cells were fixed with 3:1 methanol:glacial acetic acid three times and dropped onto pre-cleaned chilled glass slides. Chromosome spreads were Giemsa banded and photographed. At least 20 metaphase spreads and five banded karyotypes were evaluated for chromosomal rearrangements.

Teratoma Formation in Severely Combined Immunodeficient (SCID) Mice

Morphologically undifferentiated regions of HES-3 and HES-4 colonies grown on AS, FM, and D551 human feeders for 20 passages were mechanically dissected into clumps of about 300 to 400 cells each. About 2-5 × 106 cells were injected with a sterile 25G needle into the thigh muscle of SCID mice. Two SCID mice were injected for teratoma formation for each feeder. The mice were sacrificed 8-12 weeks later, and tumors dissected and fixed in 4% paraformaldehyde. Tumors were embedded in paraffin and examined histologically after hematoxylin and eosin staining.

SSEA-3, SSEA-4, Tra-1-60 and Tra-1-81 Cell Surface Undifferentiation Markers

For immunoflourescence demonstration of stem cell surface markers Tra-1-60 and Tra-1-81, hES colonies were fixed in 4-well slide flasks (BD) with 100% ethanol for 20 minutes. For SSEA-3 and SSEA-4 staining, hES colonies were fixed in 4% paraformaldehyde for 30 minutes. The sources of the monoclonal antibodies for the detection of the markers were as follows: SSEA-3 (MC-631) and SSEA-4 (MC-813-70), Development Studies Hybridoma Bank (Iowa City, IA); Tra-1-60 and Tra-1-81, gifts from Dr. Peter Andrews, University of Sheffield. Primary antibodies were diluted in PBS (Invitrogen) and blocking performed with 2mg/ml bovine serum albumin (BSA; Sigma) in 1X PBS (Invitrogen) at 4°C for 15 minutes. Antibody localization was performed using rabbit anti-mouse immunoglobulin secondary antibody conjugated to flourescein isothiocyanate (Sigma). hES cells were routinely tested for markers of undifferentiation every 15 passages.

Alkaline Phosphatase

Alkaline phosphatase activity was detected with the Vector Red Alkaline Phosphatase Substrate Kit I (Vector Labs, Inc., Burlingame, CA; and viewed with rhodamine excitation and emission filters. hES cells were routinely tested for alkaline phosphatase activity every 15 passages.

Fluorescein Isothiocyanate (FITC) Flow Cytometry

For fluorescence-activated cell sorting (FACS) analysis, entire hES colonies were carefully dissected around the colony periphery and trypsinized into single cell suspension. Single cells were washed with PBS (Invitrogen), blocked with 2 mg/ml BSA (Sigma) in 1X PBS (Invitrogen) at 40°C for 30 minutes and incubated on ice with 100 μl of diluted primary and secondary FITC-conjugated antibodies. FACS was performed on a Coulter Epics Elite sorter according to 530 nm band pass filter for detection of FITC-conjugated antibodies. Data analysis was performed on WinMDI v 2.8 software. Ten thousand events were collected for each sample. Forward and side scatter plots were used to exclude dead cells and cell debris from the histogram plot analysis. At least two independent experiments were performed for each cell line studied.

Cell Cycle/DNA Profile Analysis

hES cells in single-cell suspensions of about 6 × 105 - 1.5 × 106 cells/ml in PBS (Invitrogen) were fixed in 70% ethanol on ice for 1.5 hours. The cell suspension was then passed through a 60 μm filter. One hundred μl of propidium iodine stain (Beckman Coulter; Fullerton, CA; were added to 300 μl of filtered hES cell suspension and incubated at room temperature for 30 minutes. The samples were analyzed as described above.


Growth Characteristics of Human Feeder Cells

Primary cultures of AGE and AS cells had an epitheliod morphology while the rest of the feeders were long spindle-shaped fibroblasts. At first passage and thereafter the AGE cells continued to remain epithelioid while the AS epithelioid cells transformed to fibroblasts. The AFT fibroblasts were larger cells compared to the rest of the feeders. AGE, ASM, and MRC-5 fibroblasts did not survive the mitomycin C treatment well with many dead unattached cells after exposure. Xeno-free human feeder establishment medium and Chang's medium supported the growth of all human feeders.

Behavior of hES Cells on Human Feeders

The behavior of human feeders in terms of support and nonsupport of hES cells is summarized and ranked in Table 1. The most supportive human feeders, FM, AS and D551, were superior to the conventional MEF (controls).

Table Table 1.. Growth characteristics of HES-3 and HES-4 on supportive and non-supportive human feeders
Feeder typeFeeder rankingFeeder passagehES line supportedLatest hES passageApproximate % of hES cells stained with Tra-1-60Approximate % of completely undifferentiated colonies
Human fetal      
1 FM14-16HES-3 HES-450+ 50+80 8090 85
2 D551/CCL-10 (skin)210-15HES-3 HES-425+ 25+86 8090 80
3 FS84-8HES-3 HES-420+ 20+— —40 50
Human adult      
4 MRC-5 (lung)Non-supportive14HES-3 HES-41+ 1+— —10 10
5 WI-38 (lung)Non-supportive18HES-3 HES-42+ 2+— —20 20
6 AS34-16HES-3 HES-430+ 30+70 7075 70
7 AFT64-8HES-3 HES-420+ 20+— —55 55
8 AM94-8HES-3 HES-49+ 11+— —40 45
9 AGENon-supportive4HES-3 HES-40+ 1+— —0 0
10 ASENon-supportive4HES-3 HES-40+ 0+— —0 0
Human neonatal      
11 CCL-2552 (foreskin)713HES-3 HES-42+ 2+— —55 55
12 B-83 (MEF)44-6HES-3 HES-440+ 40+— —60 75
13 B-84 (MEF)54-6HES-3 HES-440+ 40+— —55 65

FM fibroblasts were clearly superior to the other in-house derived and commercial fibroblast feeders and ranked number 1 (Table 1). Consistently good hES cell colony growth was obtained with FM fibroblasts (Fig. 1A) with 85% to 90% of completely undifferentiated colonies in 8 days compared to 55% to 75% on MEF controls (Table 1). Colonies on FM fibroblasts were also thicker than colonies on other human feeder layers. The commercially available D551 fetal skin fibroblast cell line and in-house derived human AS fibroblasts supported prolonged undifferentiated hES cell growth as well (Fig. 1B, C; Table 1). Colonies on D551 showed 80% to 90% of hES colonies with complete undifferentiation in 8 days compared to 70% to 75% for AS feeders in 8 days (Figs. 1B, C and F, Table 1). These values were also much higher than MEF controls. Although AFT, foreskin, FS and AM (ranks 6-9) also supported hES cell growth, they were inferior to MEF controls (40% to 55% versus 55% to 75%). The two AS cell lines established in-house from skin samples taken from two different patients supported hES cell growth equally well. HES-3 and HES-4 have been propagated and expanded on FM for over 50 passages, on D551 for over 25 passages, and on AS for over 30 passages. H1 and KO medium (Invitrogen) supplemented with 4-8 ng/ml of bFGF supported hES cell growth on these feeders equally well, but hES cell colonies were found to be thicker and more compact when KO medium was used.

Figure Figure 1..

A) Undifferentiated HES-4 colony (40 passages), on FM 5 × magnification.B) Undifferentiated HES-3 colony (20 passages) on D551, 5 × magnification. C) Undifferentiated HES-4 colony (30 passages) on AS, 5 × magnification. D) Differentiating hES colony on MRC-5, 10 × magnification. E) Undifferentiated HES-3 colony (50 passages) on B-83 (control), 5 × magnification. F) Undifferentiated HES-4 colony (30 passages) on AS, 40 × magnification. G) Differentiating hES cells with enlarged cytoplasm on MRC-5, 20 × magnification. H) Differentiating hES cells with enlarged cytoplasm on AGE, 40 × magnification. Bars: (A-C) 200 μm; (D) 100 μm; (G) 50 μm; (F, H) 30 μm.

AGE, ASE, MRC-5 and WI-38 feeders could not support hES cell growth even before the first passage (Table 1). Colonies on these feeders were very thin and with more differentiation thus making passaging very difficult. The hES cells on these feeders lost their compact, high nuclear:cytoplasm ratio rapidly and adopted an enlarged morphology with more cytoplasm (Figs. 1D, G and H). Colonies on AM, FS and AFT feeders were slightly thicker and had less differentiation but were also found to be unsuitable for the prolonged support of hES cells (Table 1). AM feeders could only support hES cell growth for 11 passages while FS and AFT feeders were able to support hES cell growth for up to 20 passages. The foreskin feeder CCL 2552 provided reasonable hES cell support but was not as optimal as FM, AS or D551 in terms of differentiation rates and thickness of colonies (Table 1).

Characterization of hES Cells Grown on Human Feeder Layers

Characterization studies showed that HES-3 and HES-4 on AS (30 passages) and D551 (25 passages) fibroblast feeders all tested positive for the expression of POU5F1, FGF4, HESX1 [11], TERT and REX1 [9] by RT-PCR (Fig. 2), displayed normal karyotypes (Fig. 3), expressed SSEA-3, SSEA-4, Tra-1-60 and Tra-1-81 cell surface markers (Fig. 4), and tested positive for alkaline phosphatase activity (Fig. 4) similar to FM and MEF controls. Histology of teratomas produced in SCID mice revealed the presence of tissues from all three germ layers (Fig. 5) confirming the pluripotent nature of both hES cell lines grown on human feeders.

Figure Figure 2..

RT-PCR analysis of AS supported HES-3.Lane 1, Fermentas 100 bp ladder; Lane 2, ÁCTB (400 bp band); Lane 3, POU5F1 (247 bp band); Lane 4, FGF4 (369 bp band); Lane 5, HESX1 (308 bp band); Lane 6, REX1 (305 bp band); Lane 7, TERT (145 bp band). Bands below 100 bp are unspecific background products.

Figure Figure 3..

A normal karyotype of hES cells that have been expanded on AS feeders for 30 passages.

Figure Figure 4..

Characterization of hES colonies on AS feeders.A) Surface marker expression of SSEA-4, 10 × magnification; B) SSEA-4, 20 × magnification; C) Tra-1-60, 5 × magnification; D) Tra-1-60, 40 × magnification; E) Tra-1-81, 10 × magnification; F) alkaline phosphatase, 5 × magnification. Bars: (A, E) 100 μm; (B) 50 μm; (C, F) 200 μm; (D) 20 μm.

Figure Figure 5..

Teratomas in SCID mice injected with HES-3 cells supported by AS.A) Cartilage, 10 × magnification; (B) cartilage and glands, 5 × magnification; C) neural rosettes, 5 × magnification,; D) cystic spaces and cartilage, 5 × magnification; E) tubular structures, 20 × magnification; F) developing gut, 20 × magnification. Bars: (A) 100 μm; (B-D) 200 μm; (E, F) 50 μm. c = cartilage, gl = glands, nr = neural rosettes, seconds = cystic spaces, tb = tubular structures, dg = developing gut.

FITC flow cytometry percentages (Tra-1-60 antibody staining) for entire HES-3 and HES-4 populations on FM, AS, and MEF feeders (similar passages) are summarized in Table 1, and a representative FACS histogram for Tra-1-60 staining is presented in Figure 6. Preliminary cell cycle/DNA profile analyses (propidium iodide staining) of HES-3 and HES-4 supported by MEF and FM feeders indicated that >10% of the entire hES cell population was in the sub-G1 peak. A smaller percentage of the entire hES cell population was in the G1 phase compared to the FM and WI-38 populations (Table 2). Similar values for hES cell death were obtained for AS, FM, and D551 feeders.

Figure Figure 6..

A) Representative FACS histogram analysis of Tra-1-60 stained single HES-3 cells from FM feeders and (B) AS feeders.M1 indicates the proportion of cells stained positive with the Tra-1-60 antibody.

Table Table 2.. DNA profile comparison between hES and feeder cells
  1. a

    *HES-3 and HES-4 were grown on FM feeders

Cell Line*% of total cell population (after gating)

Morphologically undifferentiated 50P HES-3 colonies supported by FM, D551, and AS feeders were found to express only BMP4 and NEUROD1 markers of differentiation at very low levels as compared to hES colonies supported by AGE, MRC-5 feeders, and high density differentiated 14-day-old hES cultures. AFP expression was undetectable in hES colonies supported by FM, AS and D551 while NEUROD1 expression was detectable only at cycle 39 by qRT-PCR (Table 3).

Table Table 3.. Differentiation marker expression in hES-3 colonies (50P) cultured on various human feeders
  • a

    aA total of 40 PCR cycles was performed for each reaction.

  • b

    bnd = not detected after 40 PCR cycles

  • c

    Ct = threshold cycle; SD = standard deviation

 Mean ± SD Ct values for early markers of differentiation in hES-3 colonies cultured on human feedersa
 Supportive human feedersNon-supportive human feedersDifferentiated hES colonies
Gene7 days on AS7 days on D5517 days on FM7 days on MRC-57 days on AGE14 days on D551 (high density culture)
POU5F1 (OCT 3/4)25.3 ± 0.0124.8 ± 0.0126.0 ± 0.0627.0 ± 0.1827.0 ± 0.0125.4 ± 0.01
AFPndbndnd30.0 ± 0.0130.2 ± 0.0131.7 ± 0.01
BMP432.0 ± 0.0131.8 ± 0.4432.2 ± 0.0729.1 ± 0.1228.6 ± 0.0130.7 ± 0.01
NEUROD139.4 ± 0.0138.6 ± 0.5039.2 ± 0.4135.3 ± 0.0233.0 ± 0.0136.0 ± 0.01


This study demonstrates that not all human feeders support hES growth equally well. This is also the first report of an adult fibroblast cell line (AS) that is capable of supporting prolonged undifferentiated hES cell growth. However, our results conform to the general view that fetal or embryonic tissues perform better in vitro than adult tissues because FM and D551 (fetal skin) were the most successful in supporting growth of undifferentiated hES cells. But not all fetal feeders of the same cell type appear to perform equally well. The commercial FS (D551) was superior to the in-house derived FS, and this may have been due to differences in the sources of fetal skin biopsies, types of culture media used, and methods of derivation. On the same token, not all mouse embryonic fibroblast cell lines support hES cell growth. B-83 and B-84 appear to support hES cell growth quite well while Sim's Thioguanine-Ouabain-resistant fibroblasts do not [12]. Serially expanded FM and AS feeders were found to support hES cell growth very well as late as even the sixteenth passage. This is striking in comparison to MEF feeders which perform optimally between the fourth to sixth passages. A major advantage of FM over AS as a feeder is its potentially low bioburden because FM is not anatomically exposed to the external environment and hence to adventitious agents.

However, AS feeder cell lines were easy to establish and expand, and irrespective of patient they supported hES cell growth suggesting their consistency as a feeder. Additionally, AS fibroblasts may be preferred because they are also easily accessible from any part of the body through a simple noninvasive skin biopsy taken from any donor. It is important to note that both AS primary cultures started with an epithelioid morphology confirming their epidermal origin rather than dermis. We are therefore unsure if AS of dermal origin will support hES cells equally well. Interestingly, the in-house AM cell line which was nonsupportive was obtained from one of the patients who also donated an AS sample. It thus appears that there are differences in support between cell types from the same donor and that the ability of a feeder to support hES growth could be cell type specific rather than patient specific.

Fong et al. [13] showed that the 30 to 40 inner cell mass (ICM) cells from good quality human blastocysts are held together tightly with many junctional complexes, and the ICM has to be placed as a single clump on inactivated feeders for successful derivation of hES cell lines. Dissociating the ICM into single cells and then seeding them on the feeder compromises successful derivation of hES cell lines. Thus the social nature of the ICM cells and the presence of tight junctional complexes between them alludes to the fact that feeder cells may be essential for the derivation of new hES cell lines but may not be necessary for subsequent propagation of hES cells. AS feeders should also be able to support the derivation of new hES lines similar to FM shown in our previous studies [6], but we did not attempt to derive a new cell line on AS feeders because of the difficulty in obtaining human embryos.

FACS analysis consistently revealed that a greater percentage of the entire HES-3 and HES-4 population on FM feeders stained positive for the Tra-1-60 antibody as compared to similar passaged HES-3 and HES-4 populations on AS feeders. This could suggest that FM feeders are superior to AS feeders in maintaining the undifferentiated hES cell phenotype. Preliminary cell cycle/DNA profile analysis of HES-3 and HES-4 supported by MEF and FM feeders revealed that a large proportion (>10%) of the hES population was in the sub-G1 peak (Table 2). The sub-G1 peak could represent an apoptotic cell population and some mechanically damaged cells. This figure is very much higher than the sub-G1 populations of normal human FM and W1-38 fibroblasts and also higher than spontaneous apoptosis levels associated with most cultured human cell lines. Though an increase in the number of apoptotic cells in the hES cell population may reflect suboptimal culture conditions, cells harvested for these experiments were from healthy hES cell cultures with little differentiation and grown under very stringent conditions. Sathananthan et al. [14] reported the presence of “blebs” on the surface of hES cells in their ultrastructural studies by electron microscopy. Plasma membrane blebs are believed to be typical of cells that are undergoing apoptosis [14]. A recent publication on the transcription profile of a murine ES cell line reported the upregulation of a pro-apoptotic Bcl-2 antagonist gene [15]. We have also detected that numerous apoptosis-associated mRNA transcripts are upregulated in hES cells (unpublished data).

The role of apoptosis in development is well established, and stem cells have the ability to self-renew and differentiate. Our results are compatible with these views and could possibly reflect a bona fide physiological mechanism in progress rather than cell death as a result of suboptimal culture conditions. Furthermore, when we compared the DNA profiles of two normal human fibroblast feeder cell lines (WI-38 and FM) with the profiles of two hES cell lines (HES 3 and HES 4), we found that hES cells like murine ES cells have a smaller proportion of the entire cell population in the G1 stage of the cell cycle.

In the present study, FM, D551 and AS human feeders were able to support two different hES cell lines of different sex and ethnic origin (HES-3 and HES-4) with low spontaneous differentiation (Tables 1, 3, Table 3.) suggesting the consistency of support of these human feeders irrespective of hES line. Transcription profiling and proteomic approaches to determine the differences between supportive and nonsupportive human feeders may help identify important groups of growth factors responsible for such support. Such data could also possibly help in the development of feeder-free hES culture media in the future and even help identify regulatory pathways involved in the maintenance of the pluripotent state. It may also be important to examine the effects of mitomycin C treatment and gamma irradiation on feeder gene expression.

Many more new hES lines which have had no prior contact with xenoproteins or xenosupport systems need to be established for future therapeutic application. It is therefore timely and appropriate now to formulate a “gold standard” that is free of adventitious agents for the culture and establishment of new fully undifferentiated hES cell lines. Further refinements in the entire hES derivation and propagation protocols may thus be very essential. Of particular importance would be the evaluation of human complement instead of guinea pig complement and the screening of human antibodies against the human blastocyst trophectoderm raised in rabbits for immunosurgery protocols, or the evaluation of other mechanical methods such as microdissection [16] or infrared laser for isolating ICM from blastocysts. New serum-free culture media formulations containing human-based ingredients need to be evaluated alongside the chosen human feeder support systems. The conventional commercial KO serum replacement (Invitrogen) sold for research purposes contains products of animal origin and is thus undesirable [12]. It would also be preferable to use in vitro fertilization rather than intracytoplasmic injection (ICSI) generated blastocysts because ovine/bovine hyaluronidase is traditionally used in the denuding of oocytes for the ICSI procedure in assisted conception programs.

It may be possible to also derive and propagate fibroblast human feeders in defined serum-free culture media containing recombinant human serum albumin as a protein source instead of human serum thus overcoming the variability of human serum. It may also be necessary that new human feeder cell lines for hES cell support be derived in current good manufacturing practice conditions. All these important issues need to be considered in developing new and safer hES cell lines that can then be used for producing clinically applicable cells or tissues for transplantation therapy.


This study was supported by a grant from Embryonic Stem Cell International Pte. Ltd.