Identification of Potential Pluripotency Determinants for Human Embryonic Stem Cells Following Proteomic Analysis of Human and Mouse Fibroblast Conditioned Media Andrew B. J. Prowse,† Leon R. McQuade,‡ Katherine J. Bryant,§ Helder Marcal,†,| and Peter P. Gray*,†,⊥ School of Biotechnology and Biomolecular Sciences, The University of New South Wales, Sydney, NSW, 2052, Australia, Australian Proteome Analysis Facility, Macquarie University, Sydney, NSW, 2109, Australia, Faculty of Medicine, The University of New South Wales, Sydney, NSW, 2052, Australia, The Graduate School of Biomedical Engineering, The University of New South Wales, Sydney, NSW, 2052, Australia, and The Australian Institute for Bioengineering and Nanotechnology, The University of Queensland, Brisbane, QLD, 4072, Australia Received April 22, 2007
The unique pluripotential characteristic of human embryonic stem cells heralds their use in fields such as medicine, biotechnology, biopharmaceuticals, and developmental biology. However, the current availability of sufficient quantities of embryonic stem cells for such applications is limited, and generating sufficient numbers for downstream therapeutic applications is a key concern. In the absence of feeder layers or their conditioned media, human embryonic stem cells readily differentiate to form embryoid bodies, indicating that trophic factors secreted by the feeder layers are required for long-term proliferation and maintenance of pluripotency. Adding further complexity to the elucidation of the factors required for the maintenance of pluripotency is the variability of different fibroblast feeder layers (of mouse or human origin) to effectively support human embryonic stem cells. Currently, the deficiency of knowledge concerning the exact identity of factors within the pathways for self-renewal illustrates that a number of factors may be required to support pluripotent, undifferentiated growth of human embryonic stem cells. This study utilized a proteomic analysis (multidimensional chromatography coupled to tandem mass spectrometry) to isolate and identify proteins in the conditioned media of three mitotically inactivated fibroblast lines (human fetal, human neonatal, and mouse embryonic fibroblasts) used to support the undifferentiated growth of human embryonic stem cells. One-hundred seventy-five unique proteins were identified between the three cell lines using a e1% false positive rate of identification. These proteins were organized into 17 categories. The differentiation and growth factor and extracellular matrix and remodeling categories contained proteins from many of the key pathways already implicated in the maintenance of human embryonic stem cell pluripotency including the Wnt, BMP/TGF-β1, Activin/Inhibin, and insulin-like growth factor-1 pathways. The conditioned media of fibroblast feeder layers is a complex system, and this study assists in narrowing potential candidates responsible for the support of undifferentiated human embryonic stem cells. Keywords: fibroblast conditioned media • proteomic analysis • human embryonic stem cells • pluripotency • transforming growth factor beta • undifferentiated
Introduction Human embryonic stem (hES) cells have the capacity for prolonged self-renewal and the ability to produce at least one * To whom correspondence should be addressed. Prof. Peter P. Gray, Director, Australian Institute for Bioengineering and Nanotechnology, Level 6, Building 80 Queensland Bioscience Precinct, The University of Queensland, Brisbane, Qld, 4072, Australia; Telephone, +61-7-33462600; Fax, +61-7-33462101; E-mail,
[email protected]. † School of Biotechnology and Biomolecular Sciences. ‡ Macquarie University. § Faculty of Medicine. | The Graduate School of Biomedical Engineering. ⊥ The University of Queensland.
3796
Journal of Proteome Research 2007, 6, 3796-3807
Published on Web 07/27/2007
type of differentiated progeny. Mammalian embryonic stem cells are derived from the inner cell mass of the preimplantation blastocyst and being pluripotent have the capacity to differentiate into cell types representative of all germ lineages both in vivo and in vitro. Concomitantly this characteristic heralds the use of hES cells in fields such as medicine, biotechnology, biopharmaceuticals, and developmental biology. However, the current availability of sufficient quantities of hES cells for such applications is limited. Pluripotent, undifferentiated mouse embryonic stem (mES) cells were initially established and maintained by co-culturing 10.1021/pr0702262 CCC: $37.00
2007 American Chemical Society
research articles
Pluripotency Determinants for hES
on mitotically inactivated mouse embryonic fibroblast (MEF) feeder layers and have successfully been cultured in both feeder and feeder-free conditions by addition to the media of the IL-6 related cytokine, leukemia inhibitory factor (LIF).1 Activating an intracellular signaling cascade involving Janus kinase (JAK) and a latent signal transducer and activator of transcription, STAT3, LIF is able to induce self-renewal of mES cells over numerous passages. Conversely, prolonged self-renewal of hES cells is not effected by the addition of LIF to the culture media, despite the presence of LIF cell surface receptors, gp130 receptors, and a functional JAK/STAT3 pathway.2,3 The canonical Wnt signaling pathway is implicated in the process of selfrenewal in hES cells4 though how this pathway interacts with the transcription factors, Oct4, Nanog, and Sox2 known to mediate ES self-renewal, requires elucidation.5 In addition, basic fibroblast growth factor (bFGF; FGF2) has been shown to be capable of maintaining hES cells and is used routinely in hES cell culture media.6 In the absence of feeder layers, hES and mES cells readily differentiate to form embryoid bodies, clearly indicating that trophic factors secreted by the feeder layers, or their presence as a matrix, are required for long-term proliferation and maintenance of pluripotency. This was most clearly demonstrated as researchers successfully commenced the maintenance and pluripotent proliferation of hES cells using media previously conditioned by mouse embryonic fibroblasts (MEFCM) and an extracellular matrix such as Matrigel (BD Biosciences) for cellular attachment.7 The identification of ligand receptors demonstrated that the addition of FGF2,8 Activin A9 and Noggin10 to undifferentiated hES cell cultures may actually inhibit or repress differentiation down specific germ lineages rather than induce self-renewal. Conversely, the use of some of these factors are employed to direct differentiating hES cells down particular germ lineages,11,12 indicating that their use for undifferentiated hES cell culturing needs to be understood in a dose-dependent manner. Recent refinement to culturing systems have minimized the risk of xenotransmission from non-human sources and have enabled undifferentiated propagation of hES cells by the use of human marrow stromal cells,13 adult fallopian tubal cells,14 human neonatal fibroblasts (HNF),15 and human fetal fibroblasts (HFF).16 A comparative study on the ability of 13 human and mouse feeder layers to support hES cells revealed a gradation. Three frequently used fibroblast lines (HFF, MEF, and HNF) were ranked first, fourth, and seventh, respectively, in their support of undifferentiated proliferation,17 adding further complexity to the elucidation of factors required for undifferentiated proliferation. Transcriptome analysis of feeder layers18 and the proteomic analysis of fibroblast conditioned media of both mouse and human origin19-21 in conjunction with cytokine signaling studies5 are continuing to provide an insight into the key factors required for the undifferentiated proliferation of pluripotent ES cells. These studies revealed a varied network of transcripts and proteins with diverse known functions of intracellular and extracellular origin. Therefore, it is apparent at this point in time, in the absence of an identified biochemical pathway for self-renewal, that a number of factors may be required to support pluripotent, undifferentiated growth of hES cells in a defined animal-free media.22 This study utilized a proteomic analysis to isolate and identify the proteins in the conditioned media of three mitotically inactivated fibroblast lines, two of human origin (HFF and
HNF), and one of mouse (MEF). The aim was to identify proteins common to all three lines, common to the two lines that provide better hES cell support (HFF and MEF), and unique to those providing the best support (HFF), as a means of identifying proteins that underpin the maintenance of undifferentiated proliferation of hES cells. A significant limitation in proteomic analysis is the difficulty in observing low copy number proteins such as transcription factors, considered biologically significant in the maintenance of hES cell pluripotency. Thus, our approach involved identification of peptides using a two-step LC separation utilizing strong cation exchange (SCX) and reverse-phase sequentialstep elution followed by peptide sequence analysis using tandem MS.23
Experimental Section Establishment and Culturing of Fibroblasts. A human fetal fibroblast line (HFF01) was established from dermal tissue derived from the therapeutic termination of an early second trimester pregnancy. The human neonatal foreskin line, HNF02, previously reported was used in this study.21 Informed maternal consent and Human Research Ethics Committee (The University of New South Wales, UNSW) approval was granted for the use of human fetal tissue (HREC 02247) and human neonatal tissue (HREC 03055). Mouse embryonic fibroblasts (MEF) were established from 14.5dpc embryos of 129sv mice under consent from the Animal Care and Ethics Committee, UNSW (ACEC 04/120B). Primary cultures were established in T25 tissue culture flasks24-26 and then expanded into T75 flasks (Nalge Nunc International, Rochester, NY). Cultures were maintained in Dulbecco’s modified Eagle’s media containing 25 mM glucose (DMEM-high glucose) supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine, 25 U/mL penicillin, 25 mg/mL streptomycin (all Invitrogen/Gibco, Carlsbad, CA), and maintained at 37 °C; 5% CO2. Confluent, HFF01, HNF02, and MEF fibroblasts at passages eight, six, and four, respectively, were mitotically inactivated with mitomycin C (Sigma; St Louis, MO) at a final concentration of 10 mg/mL and incubated for 2.5 h at 37 °C; 5% CO2. Passage numbers for collection of conditioned media (CM) were chosen based on those shown to give best support of ESI-hES3 from previous studies (unpublished data). The cells in each flask were washed gently three times with PBS before addition of 30 mL of DMEM containing 1% insulin, transferrin, and selenium (ITS-A; Gibco). After 18 h incubation at 37 °C, 5% CO2, the media was filtered using a 0.2 µm Ministart sterile filter (Sartorius, Goettingen, Germany), and a protease inhibitor cocktail (Sigma) containing leupeptin hemisulphate, aprotinin, AEBSF, E-64, pepstatin, and bestatin was added according to the manufacturer’s recommendations. The culture supernatant was stored at -80 °C. Two-Dimensional Liquid Chromatography and MS-MS Analysis. Protein concentration and LC methods essentially followed those previously described.21 Trypsin digested peptides were separated by on-line SCX and C18 nanoLC using an Ultimate HPLC, Switchos, and Famos autosampler system (LCPackings, Amsterdam, Netherlands). Conditioned media (90 mL) was TCA-precipitated and the resulting pellet (∼25 µg) was digested with 5 µg trypsin in 50 µL, 20 mM NH4HCO3 for 14 h at 37 °C. Peptides (10 µL) were diluted in 90 µL 0.1% v/v formic acid and loaded onto a SCX microtrap (168 mm; Michrom BioreJournal of Proteome Research • Vol. 6, No. 9, 2007 3797
research articles
Prowse et al.
sources, Auburn, CA) at 20 µL/min. Peptides were eluted using 20 µL volumes of ammonium acetate in a stepwise gradient range of 5-1000 mM. The unbound load fraction and each salt step were concentrated and desalted using a micro C18 precolumn (500 µm × 2 mm; Michrom Bioresources) with H2O: CH3CN (98:2, 0.1% formic acid) at 20 µL/min. After a 10 min wash, the precolumn was switched (Switchos) into line with a fritless analytical C18 column (75 µm × 12 cm) and peptides eluted using a linear gradient of H2O:CH3CN (95:5, 0.1% formic acid-buffer A) to H2O:CH3CN (40:60, 0.1% formic acid-buffer B) at 200 nL/min over 30 min. High voltage (2300 V) was applied at the beginning of the column through a low volume tee (Upchurch Scientific, Oak Harbor, WA) and the outlet positioned 0.1 cm from the orifice of an API Qstar Pulsar I hybrid tandem mass spectrometer (Applied Biosystems). Positive ions were generated by electrospray and the Qstar operated in information dependent acquisition mode. An MS survey scan was acquired (m/z 3501700, 0.75 s) and the two largest multiply charged ions (counts > 20, charge state g 2 and e 4) sequentially selected by Q1 for MS-MS analysis. Tandem mass spectra were accumulated for 2 s (m/z 65-2000). Database Searching. Processing scripts generated data suitable for submission to the database search program MASCOT (Mascot Distiller, Version 2.1.0, http://www.matrixscience.com) using the Rodentia, Bos taurus, and Homo sapiens taxonomies. The data sets were searched against both NCBI (http:// www.ncbi.nlm.nih.gov, latest search date 17/8/06 with 151 916 Rodentia and 143 745 Homo sapien protein sequences) and Swiss-Prot databases (http://www.expasy.org, release 50.4, 25/ 7/06 with 17 299 Rodentia and 14 106 Homo sapien protein sequences) to help eliminate redundancies and identify proteins that only appear in one of the databases. An initial search was conducted using the following criteria; missed cleavages ) 0; peptide mass tolerance ) (100 ppm and peptide MS/MS tolerance ) (0.2 Da and no ion score cutoff. Proteins were then filtered according to (a) overall protein score generated by the Mascot search engine (g32), (b) individual peptide scores (g30), (c) a continuous series of y ions for individual peptides, and (d) a search for peptide redundancy. To achieve a false positive rate of e1%, total peptide sequence data for each cell line was searched against an equivalent human or mouse scrambled database. False positive rates were determined according to previously published methods27 (eq 1). %Fal ) 2[nscr/(nscr + nreal)]
(1)
where %Fal ) estimated false positive rate, nscr ) number of peptides identified from the scrambled database, and nreal ) number of peptides identified from the real database. Protein IDs were considered if they contained at least 1 unique peptide satisfying the following criteria: missed cleavages ) 0; peptide mass tolerance ) (50 ppm; peptide MS/MS tolerance ) (0.2 Da; and a Mascot ion score cutoff of 54.5, 47, and 53 for HFF01, HNF02, and MEF, respectively. If a protein was identified by one peptide, the original search data was rechecked for additional matching peptides to strengthen the protein ID. Failing inclusion in the list with these criteria, MS/ MS spectra of the single peptide were analyzed (see Supporting Information). CM from each fibroblast line was analyzed at least twice and data combined for Mascot searching. Peptide Redundancy and Protein Families. Proteins in the same family group (e.g., collagens) were identified with distinct 3798
Journal of Proteome Research • Vol. 6, No. 9, 2007
IDs only when matched to at least one peptide (based on sequence) unique to that member of the family or a combination of peptides contained by no other member. Proteins not satisfying these requirements have been identified in the Supporting Information (Supplementary Tables 1-6). Peptides were also searched individually against the appropriate taxonomic database using the NCBI protein-protein Basic Local Alignment Search Tool (http://www.ncbi.nlm.nih.gov/BLAST/) to identify peptide matches to any other protein, regardless of family.
Results and Discussion Each fibroblast cell line used in this study was chosen due to its ability to support the undifferentiated growth of hES3 (unpublished data). Stem cell colonies grown on each of the cell lines displayed characteristic undifferentiated morphology and were positive for the hES cell pluripotency markers SSEA-4 and Tra-1-60 via immunofluorescence and Oct-4 via RT-PCR. Determination of False Positive Rates. The positive peptide identifications from both the real and scrambled databases of either Rodentia or Homo sapien were used to calculate false positive rates of e1%27 (see Supporting Information Table 7). The final search parameters used for analysis achieved false positive rates of 0.53% (HFF01), 0.91% (HNF02), and 0.49% (MEF) a reduction of approximately 30% for each cell line from the original search parameters. Alterations in the ion score cutoff achieved the greatest reduction in scrambled peptides giving an acceptable rate of e1%. Applying the appropriate cutoff resulted in a significant reduction of proteins identified. As an example, for HFF01, using the strict search parameters with an ion score cutoff of 45, the false positive rate was 2.5%, and the number of positive protein identifications was 176 (Supporting Information Table 7). To obtain e1%, the ion score cutoff was increased to 54.5, resulting in 116 positive protein identifications, a reduction of 60 proteins. Although maintaining a rate of e1% increases confidence in the final proteins identified, we consider that this may be unnecessarily excluding proteins worthy of inclusion. Considering the current paucity of information concerning determinants involved in the maintenance of hES cell pluripotency, we believe that inclusion of a more extensive list will allow fellow researchers the opportunity to interpret our data in an independent manner and substantiate the findings through independent research efforts. To facilitate this, we have included as Supporting Information an extended list of all proteins from each cell line identified using an ion score cutoff of 30 and a reduced peptide mass tolerance of (100 ppm (Supplementary Tables 1-3). Using a false positive rate of e1%, between 20 and 40 proteins from each cell line were identified by one peptide only. To increase the confidence for these protein identifications, Supporting Information has been included outlining further peptide IDs from a search criterion with an ion score cutoff of 30 and a reduced peptide mass tolerance of (100 ppm (Supporting Information, Tables 4-6). No other search criteria were altered. Following the revised search there remained proteins with a single matching peptide and in accordance with the guidelines on proteomics standards (http://www.mcponline.org/ misc/ParisReport_Final.shtml) annotated MS/MS spectra have been included as Supporting Information for these peptides. Protein Families. There were a number of proteins identified in all three lists that were members of certain protein families including the collagens, actins, tubulins, and heat shock
research articles
Pluripotency Determinants for hES Table 1. Percentage Distribution of Proteins Based on Category and Cell Type category breakdowna
all three lines
HFF and MEF
HFF and HNF
MEF and HNF
unique
total/ category
total/ category (%)
Catalytic Chaperones and Heat Shock Cytoskeletal Degradation and Turnover Differentiation and Growth Factors Extracellular Matrix and Remodelling Immunity ITS Membrane Metabolism Miscellaneous Nuclear Proteins Protein Synthesis and PTM Regulatory Repair and Protection Signal Transduction Transport and Binding Total/Cell Line Comparison Total/Cell Line Comparison (%)
0 3 7 1 1 13 0 0 0 6 0 0 0 1 0 0 2 34 19.43
0 1 1 0 0 3 0 0 0 1 0 2 1 1 0 0 0 10 5.71
0 3 1 0 3 5 0 0 0 2 0 1 2 0 0 0 0 17 9.71
0 1 1 0 0 2 0 0 0 0 0 0 0 0 0 0 0 4 2.29
9 6 25 4 7 15 5 1 4 7 2 2 8 3 6 2 4 110 62.86
9 14 35 5 11 38 5 1 4 16 2 5 11 5 6 2 6 175 100
5.14 8.00 20.00 2.86 6.29 21.71 2.86 0.57 2.29 9.14 1.14 2.86 6.29 2.86 3.43 1.14 3.43 100
a
Proteins distributed into categories were those identified under criteria achieving a e1% false positive rate.
proteins. Determining the uniqueness of these proteins was dependent on the identification of at least one peptide, or a combination of peptides, not present in any other family members. As an example, heat shock protein 90-R and heat shock protein 90-β were identified by 4 and 5 peptides, respectively. Both proteins shared the identified peptides ADLINNLGTIAK, EDQTEYLEER, and SLTNDWEDHLAVK; however, for 90-R, the extra peptide HLEINPDHSIIETLR was unique. Similarly, 90-β had two extra peptides, ELISNASDALDK and HSQFIGYPITLYLEK, distinguishing it from 90-R and other members of the family. Ambiguity also arose due to listings in NCBI of the same gene under different names. Cross referencing with Swiss-Prot entries assisted in eliminating some of these. Following this procedure protein members of the same family were identified as unique. Proteins remained, however, that were not distinguishable from other family members (e.g., some members of the tubulin and actin families) based on peptide IDs. These have been included in the list with notes pertaining to ambiguity (see Tables 1-6 in Supporting Information). Identification of Bovine Proteins. Despite washing all cell lines three times with PBS prior to the addition of serum-free media, traces of FBS complicated protein identification. The lists of peptides therefore were also searched against the Bos taurus taxonomy. Proteins were identified as originating from FBS when Mascot scores and peptides identified using the Bos taurus taxonomy exceeded those identified when searched against the respective mouse or human databases. FBS proteins identified included albumin, R-1 Antiproteinase and Antithrombin III (see Supporting Information Tables 1-3 for a full list of proteins from HFF01, HNF02, and MEF). Peptides identified as originating from FBS proteins also matched human or mouse proteins at a significant Mascot score, although at a lower score than against the bovine databases. These matches were removed using the stringent parameters applied for a e1% false positive rate. Distribution of Proteins Based on Category and Comparison of HFF, MEF, and HNF Cell Lines. One-hundred seventyfive different proteins were identified from the three cell lines with a e1% false positive rate. Identified proteins were classified according to functional groups as shown in Table 1. The greatest percentage of proteins (21.7%) were classified within
Figure 1. Distribution of proteins identified in HFF, HNF, and MEF cell lines. The total number of proteins identified using the criteria for a e1% false positive rate of identification was 175. Proteins are distributed based on matches in one, two, or all three cell lines.
the extracellular matrix and remodelling category. Proteins with roles in the formation and regulation of the cytoskeleton accounted for 20% of the total proteins identified, whereas proteins identified in the other 15 categories represented
