Stem Cells
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

First published online November 8, 2007
Stem Cells Vol. 26 No. 2 February 2008, pp. 387 -400
doi:10.1634/stemcells.2007-0599; www.StemCells.com
© 2008 AlphaMed Press

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental Data
Right arrow All Versions of this Article:
2007-0599v1
2007-0599v2
26/2/387    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Reprints/Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Arrell, D. K.
Right arrow Articles by Terzic, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Arrell, D. K.
Right arrow Articles by Terzic, A.

EMBRYONIC STEM CELLS

Cardioinductive Network Guiding Stem Cell Differentiation Revealed by Proteomic Cartography of Tumor Necrosis Factor {alpha}-Primed Endodermal Secretome

D. Kent Arrell, Nicolas J. Niederländer, Randolph S. Faustino, Atta Behfar, Andre Terzic

Marriott Heart Disease Research Program, Division of Cardiovascular Diseases, Departments of Medicine, Molecular Pharmacology and Experimental Therapeutics, and Medical Genetics, Mayo Clinic College of Medicine, Rochester, Minnesota, USA

Key Words. Embryonic stem cell • Cardiopoiesis • Endoderm • Secretome • Proteomics • Two-dimensional gel electrophoresis Multidimensional liquid chromatography-tandem mass spectrometry • Network biology • Systems biology

Correspondence: Correspondence: Andre Terzic, M.D., Ph.D., Mayo Clinic, 200 First Street SW, Rochester, Minnesota 55905, USA. Telephone: 507-284-2747; Fax: 507-266-9936; e-mail: terzic.andre{at}mayo.edu

Received on July 25, 2007; accepted for publication on October 24, 2007.

First published online in STEM CELLS EXPRESS  November 8, 2008.

   ABSTRACT
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 Conclusion
 Disclosure of Potential...
 Acknowledgments
 References
 
In the developing embryo, instructive guidance from the ventral endoderm secures cardiac program induction within the anterolateral mesoderm. Endoderm-guided cardiogenesis, however, has yet to be resolved at the proteome level. Here, through cardiopoietic priming of the endoderm with the reprogramming cytokine tumor necrosis factor {alpha} (TNF{alpha}), candidate effectors of embryonic stem cell cardiac differentiation were delineated by comparative proteomics. Differential two-dimensional gel electrophoretic mapping revealed that more than 75% of protein species increased >1.5-fold in the TNF{alpha}-primed versus unprimed endodermal secretome. Protein spot identification by linear ion trap quadrupole (LTQ) tandem mass spectrometry (MS/MS) and validation by shotgun LTQ-Fourier transform MS/MS following multidimensional chromatography mapped 99 unique proteins from 153 spot assignments. A definitive set of 48 secretome proteins was deduced by iterative bioinformatic screening using algorithms for detection of canonical and noncanonical indices of secretion. Protein-protein interaction analysis, in conjunction with respective expression level changes, revealed a nonstochastic TNF{alpha}-centric secretome network with a scale-free hierarchical architecture. Cardiovascular development was the primary developmental function of the resolved TNF{alpha}-anchored network. Functional cooperativity of the derived cardioinductive network was validated through direct application of the TNF{alpha}-primed secretome on embryonic stem cells, potentiating cardiac commitment and sarcomerogenesis. Conversely, inhibition of primary network hubs negated the procardiogenic effects of TNF{alpha} priming. Thus, proteomic cartography establishes a systems biology framework for the endodermal secretome network guiding stem cell cardiopoiesis.

Disclosure of potential conflicts of interest is found at the end of this article.


   INTRODUCTION
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 Conclusion
 Disclosure of Potential...
 Acknowledgments
 References
 
Morphogenesis of specific regions within the mesodermal layer of a developing embryo gives rise to the future heart [1, 2]. Although anatomically pinpointed to the anterolateral mesoderm, heart-forming regions cannot generate myocardial tissue without inductive stimuli [3, 4]. Tissue recombination experiments have established that cardioinductive signaling is largely extramesodermal in origin, arising primarily from the adjacent ventral endoderm [5–8]. Cardiogenic instruction induces sequential expression of cardiac-specific transcription factors, including the homeobox protein NK-2 homolog E (Nkx2.5), which serves as an early molecular marker of cardiac fate, and myocyte-specific enhancer factor 2C (MEF2C), which further promotes cardiomyocyte differentiation [9, 10]. Cardiac specification can also be achieved in vitro, through embryonic stem cell (ESC)-based derivation of embryoid bodies (EBs) that recapitulate the obligatory endoderm-mesoderm interaction [11, 12]. Moreover, isolated visceral endoderm-like cells in coculture promote cardiac differentiation of pluripotent stem cells [13]. Exposure to endoderm-conditioned medium is both necessary and sufficient to drive cardiogenesis, underscoring the importance of released soluble factors [14]. Despite success in generating cardiomyocytes outside an embryo [14], which has potential for regenerative application [15], endodermal secretome constituents guiding cardiogenesis have yet to be resolved.

The emerging concept of a secretome refers to secreted proteins and the machinery involved in the secretion process [16]. Secreted proteins generally contain signature amino-terminal signal peptides that guide translocation across the endoplasmic reticulum (ER) for transport via the Golgi apparatus to the plasma membrane, where subsequent membrane fusion releases extracellular proteins [17, 18]. This process has been termed classic or ER/Golgi-dependent protein secretion [19]. A number of secreted proteins, however, including specific cytokines, growth factors, and extracellular matrix proteins, do not follow canonical secretion processes. These proteins lack the standard signal peptide and typical ER/Golgi-dependent post-translational modifications (PTMs), they are not detectable within the ER or Golgi, and their secretion is unaltered by classic inhibitors of ER/Golgi-dependent transport [20]. Comprehensive secretome categorization therefore requires designation of both canonically and noncanonically secreted proteins, in addition to proteins engaged in their secretion.

Here, candidate effectors of cardiac differentiation within the endodermal secretome were delineated by comparative proteomics with and without priming by the reprogramming cytokine tumor necrosis factor {alpha} (TNF{alpha}), recently identified as a potent cardiogenic inducer [14]. Two-dimensional gel electrophoresis (2DE) in tandem with multidimensional shotgun chromatography, followed by nanoelectrospray liquid chromatography-tandem mass spectrometry, resolved the TNF{alpha}-induced paracrine response, exposing a cardioinductive network within the endodermal secretome. Definitive secretome proteins were stratified by application of bioinformatic algorithms to identify canonical and noncanonical indices of secretion. Functional synergism was demonstrated through secretome-mediated promotion of ESC cardiac commitment and responsiveness to targeted disruption of network hubs. Proteomic mapping of differentially primed endodermal secretomes reveals in this way a systems biology framework guiding stem cell cardiac differentiation.


   MATERIALS AND METHODS
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 Conclusion
 Disclosure of Potential...
 Acknowledgments
 References
 
Embryonic Stem Cells and Embryoid Bodies
Murine CGR8 ESCs were differentiated into three-layer EBs through the hanging drop method, and cardiac sarcomerogenesis was tracked by laser confocal microscopy using an anti-{alpha}-actinin antibody (1:1,000; Sigma-Aldrich, St. Louis, http://www.sigmaaldrich.com) [21–23]. In a subset of experiments, the endodermal layer was eliminated from EBs through generation of hanging drops in the presence of recombinant leukemia inhibitory factor (100 U/µl) once differentiation had been initiated. Levels of transforming growth factor β-1 (TGFβ-1), a stimulator of cardiac differentiation [21], were determined using the BD Biosciences (San Diego, http://www.bdbiosciences.com) enzyme-linked immunosorbent assay kit, with measurements at 450 nm performed in a photometric microplate reader. TGFβ-1 signaling pathway activation was monitored by confocal examination of phosphorylated small mothers against decapentaplegic 3 (phosphorylated Smad3 [P-Smad3]) localization using an anti-P-Smad3 antibody (1:2,000; obtained from Dr. E. Leof, Mayo Clinic), and nuclei were visualized by 4',6-diamidino-2-phenylindole (DAPI) counterstaining [21, 22].

Endoderm-Derived Secretome
Murine visceral endoderm-like cells, derived from an F9 embryonal carcinoma cell line, were cultured in a 1:1 mixture of Ham's F12 medium and high-glucose Dulbecco's modified Eagle's medium with 7.5% fetal calf serum, penicillin, streptomycin, GlutaMAX, nonessential amino acids (Invitrogen, Carlsbad, CA, http://www.invitrogen.com), and β-mercaptoethanol (Sigma-Aldrich). A total of 20 dishes (10 ml of media per dish) were cultured to 80% confluence and washed multiple times with phosphate-buffered saline prior to culturing in serum-free Glasgow's minimum essential medium (Mediatech, Manassas, VA, http://www.cellgro.com) in the absence (10 dishes) or presence (10 dishes) of TNF{alpha} (30 ng/ml). Conditioned medium was harvested after 24 hours, with media from each treatment pooled into two groups of five dishes each (two 50-ml aliquots). A protease inhibitor cocktail (Complete; Roche Diagnostics, Basel, Switzerland, http://www.roche-applied-science.com) was added to the pooled media, and the four 50-ml samples were centrifuged for 10 minutes at 2,000g, decanted, centrifuged again, and passed through a 0.2-µm filter. Total protein was quantified by Bio-Rad (Hercules, CA, http://www.bio-rad.com) protein assay prior to and following concentration by filter centrifugation at 3,000g through 5-kDa molecular weight cut-off Amicon Ultra-15 filters (Millipore, Billerica, MA, http://www.millipore.com).

Two-Dimensional Gel Electrophoresis and Image Analysis
Endoderm secretome proteins, normalized to initial medium volume, were resolved by immobilized pH gradient (IPG) 2DE following addition to isoelectric focusing (IEF) buffer (7 M urea, 2 M thiourea, 2% [wt/vol] 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonic acid [CHAPS], 15 mg/ml DeStreak reagent [GE Healthcare, Little Chalfont, U.K., http://www.gehealthcare.com], 1 x Bio-Rad ampholytes). IEF was carried out in actively rehydrated pH 3–10 or 4–7 IPG Ready Strips (Bio-Rad) and passively rehydrated pH 6–11 Immobiline DryStrips (Amersham Biosciences, Piscataway, NJ, http://www.amersham.com) [24]. Focused IPG strips were resolved by 15% and 7.5% SDS-polyacrylamide gel electrophoresis using a Protean II XL system (Bio-Rad). Gels were then silver-stained and digitized at 400 dots per inch for spot image analysis using PDQuest version 7.3.1 (Bio-Rad). Spot intensities were normalized against protein molecular mass standard intensities and ppm-scaled, and missing values were provided with a nominal nonzero estimate for fold change determination. Fold change was calculated as the treatment ratio [mean TNF{alpha}]:[mean untreated] for protein spots that increased or the negative inverse of this ratio for proteins that decreased. For proteins identified in more than one spot, the sum of ppm values for all spots was used to determine a weighted average treatment ratio. Proteins were considered increased or decreased for changes >1.5-fold. Protein species of interest were isolated, destained, and prepared for liquid chromatography-tandem mass spectrometry (LC-MS/MS) [25].

Shotgun Protein Digestion and Cation Exchange Fractionation
TNF{alpha}-primed secretome proteins were concomitantly prepared for multidimensional LC-MS/MS [26]. After concentration, proteins were sequentially digested in solution with endoproteinase Lys-C (enzyme:substrate ratio, 1:125; Sigma-Aldrich) at 37°C for 16 hours, followed by trypsin (enzyme:substrate ratio, 1:100; Promega, Madison, WI, http://www.promega.com) at 37°C for 16 hours. Peptides were desalted by trapping onto a Peptide MacroTrap (Michrom Bioresources, Auburn, CA, http://www.michrom.com), batch-eluted with 60% acetonitrile, 0.1% formic acid, and dried in a SpeedVac (Savant Instruments, Thermo Fisher Scientific Inc., Waltham, MA, http://www.thermofisher.com). Peptides were then reconstituted in 50 µl of buffer A (5 mM KH2PO4, pH 3.0, 10% acetonitrile), loaded onto a polysulfoethyl aspartamide column (2 x 50 mm, 5 µm/300 Å; Michrom Bioresources), and subjected to strong cation exchange (SCX) chromatography. Fractions were collected at 1-minute intervals (150 µl/minute), with peptides eluted over a 50-minute gradient by increasing KCl using buffer B (5 mM KH2PO4, pH 3.0, 10% acetonitrile, 500 mM KCl), beginning with 0% buffer B from 0 to 15 minutes, increasing linearly to 20% buffer B over 20 minutes, increasing linearly to 80% buffer B over the next 10 minutes, held at 80% buffer B for 2 minutes, and dropping to 0% buffer B over the final 3 minutes, with continuous monitoring for protein by dual-channel UV absorbance at 214 and 280 nm, followed by drying in a SpeedVac prior to LC-MS/MS [14].

Nanoelectrospray Linear Ion Trap Tandem Mass Spectrometry
Tryptic peptides obtained by parallel analyses (2DE or SCX) were subjected to nanoelectrospray LC-MS/MS for protein identification. Peptides from two-dimensional (2D) spots were reconstituted in 0.1% trifluoroacetic acid and trap-injected onto a 75-µm x 5-cm ProteoPep C18 PicoFrit nanoflow column (New Objective, Woburn, MA, http://www.newobjective.com). Chromatography was performed using 0.2% formic acid in both solvent A (98% water, 2% acetonitrile) and solvent B (80% acetonitrile, 10% isopropanol, 10% water), with peptides eluted over a 40-minute gradient from 5% solvent B to 45% solvent B using a Michrom Paradigm MG4 nano-high-performance liquid chromatography (nanoHPLC) system (Michrom Bioresources) coupled to a ThermoFinnigan linear ion trap quadrupole (LTQ) mass spectrometer (Thermo Fisher Scientific Inc.). Scanning for eluting peptide ions was carried out between 400 and 1,400 m/z, switching to MS/MS collision-induced dissociation (CID) mode on ions exceeding an intensity of 1,000. Alternatively, SCX fraction peptides were reconstituted in 0.1% trifluoroacetic acid, trap-injected onto a 75-µm x 10-cm ProteoPep C18 nanoflow column, and eluted with a 75-minute gradient from 5% to 45% solvent B using a splitless Eksigent nanoHPLC system (Sciex; MDS Nordion, Ottawa, http://www.mds.nordion.com) coupled to a ThermoFinnigan LTQ Fourier transform (FT) mass spectrometer (Thermo Fisher Scientific Inc.). Peptide ions were scanned in FT mode between 375 and 1,600 m/z, switching to LTQ MS selected ion monitoring (SIM) mode on ions exceeding an intensity of 3,000 and to MS/MS CID mode on ions exceeding SIM mode intensities of 500 [14].

Peptide and Protein Identification of Mass Spectrometric Data
Raw MS/MS data from 2D gel samples were converted to .dta files using Bioworks 3.1 (ThermoElectron), and merged files were correlated to theoretical tryptic fragments in Swiss-Prot (version 47.1; indexed Aug 2, 2005, comprising 188,752 sequences) using Mascot version 2.1 [27]. Searches were conducted on all species, with up to two missed cleavages and mass tolerances of ±1.2 Da for precursor ions and ±0.6 Da for MS/MS product ions, allowing for the following differential modifications: protein N-terminal acetylation, methionine oxidation, and cysteine carbamidomethylation. Peptides were accepted as positive matches based on probability (p < .05). Positive matches to nonmouse species lacking Swiss-Prot mouse homologs were screened against the TrEMBL database (indexed Jan 23, 2007, comprising 3,633,676 sequences). SCX fraction LC-MS/MS .dta files were correlated to mouse and mammalian entries from Swiss-Prot (version 47.1) and NCBInr (indexed Aug 5, 2005, comprising 2,739,666 sequences) using the integrated TurboSEQUEST/Bioworks 3.1 platform [28]. Matching peptides were filtered by satisfying ≥4 parameters: cross correlation (XCorr); ≥2.0; delta correlation ({Delta}Cn); ≥0.100; preliminary score (Sp); ≥500.0; preliminary score rank (rSp) ≤5; percentage of ions ≥30. These filtering criteria, in tandem with a reverse decoy database, yielded a false-positive rate of 4.6% for the SCX LC-MS/MS analysis [29]. Mascot (version 2.1) was used to search SCX data against SwissProt and NCBInr databases, with peptides accepted as positive matches based on probability (p < .05). MS/MS spectra from 2D gel spots were matched to multiple unique peptides or, in the case of proteins identified on the basis of a single peptide match, by manual spectrum inspection. Single peptide hits were considered matches if the Mascot score indicated identity or extensive homology (p < .05), with detected fragment ions from the MS/MS spectrum above baseline noise, demonstrable continuity in b- or y-ion series, and proline residues yielding intense y-ions [26]. Six additional cases were included as likely matches, despite lower scores, on the basis of three additional criteria: (a) positive detection by shotgun analysis (six of six); (b) presence at the expected relative molecular mass (Mr) or isoelectric point (pI) or both (five of six); and/or (c) positive identification of the identical protein in an adjacent spot (four of six). Protein identities were assigned by the minimum number of proteins required to account for all positive peptide matches [30]. Confidence in protein assignments was supported by assessment of similarity of observed pI and Mr based on 2D gel position to that predicted (ExPASy pI/Mr tool; http://us.expasy.org/tools/pi_tool.html), taking into consideration expected Mr and/or pI shifts resulting from Swiss-Prot-curated protein PTMs. Proteins identified from 2DE were inspected for independent confirmation by LTQ-FT multidimensional LC-MS/MS shotgun analysis.

Bioinformatic Secretome Classification
Identified proteins were assessed in silico for inclusion in the secretome, defined as secreted proteins and the machinery involved in their secretion [16]. Protein FASTA sequences were screened for canonical amino-terminal ER/Golgi-dependent secretion signal peptides using SignalP 3.0 hidden Markov matrix (HMM) scoring [31]. FASTA sequences were also screened by SecretomeP 2.0 mammalian neural network scoring for proteins secreted independently of signal peptide-mediated transport [32]. Additional secreted (empirical) or secretion machinery (accessory) proteins were included on the basis of reported evidence.

Protein Interaction Network Analysis
Proteins fulfilling secretome criteria were submitted, with fold-change ratios, for network analysis by Ingenuity Pathways Knowledge Base (Ingenuity Systems, Redwood City, CA, http://www.ingenuity.com) to identify functional networks within the TNF{alpha}-stimulated secretome. The composite network and subnetworks were depicted using Cytoscape 2.3.2 [33], and topological properties were characterized by the Cytoscape module, Network Analyzer (http://med.bioinf.mpi-inf.mpg.de/netanalyzer/index.php). Computed properties included the following: (a) node degree (k), the number of edges, or links, connected to each node; (b) average degree (kaverage) = 2l/n, where l is the total number of edges and n the total nodes; (iii) node degree distribution (P[k]), the probability that a specified node has k links when k is known for each node, defined as P[k] = X[k]/n, where X[k] is the number of nodes with degree k and n is the total number of network nodes [34, 35]; and (iv) clustering coefficient distribution (C[k]), for all nodes with k neighbors, where C[k] = 2L/[ki(ki – 1)], indicating neighborhood interconnectiveness, when L is the number of links present among neighbors of node i with degree ki [36]. These parameters enable discrimination among random, scale-free, and hierarchical network topographies on the basis of the P[k] versus k and C[k] versus k relationships [37]. Since P[k] versus k follows either a normal or power law distribution [37], the Anderson-Darling normality test [38] was performed on P[k] data, identifying the distribution as non-Gaussian. The P[k] versus k power law relationship was then calculated using a cumulative distribution function [39] to determine {gamma} in the power law distribution (P[k] ~ k–{gamma}), by Equation 1:


Formula 1

1
where {gamma} is power law exponent, n is the number of network nodes, xi is the node degree, and xmin minimum node degree within the network, with statistical error {sigma} [39] for Equation 1 defined by Equation 2:


Formula 2

2
The C[k] ~ k–{gamma} power law distribution was plotted on a log-log scale, with {gamma} calculated by least squares regression.

Embryonic Stem Cell-Based Cardiopoiesis
The efficacy of the endodermal secretome to drive cardiac specification was monitored in monolayer, over 12 days, with ESCs initially cultured at 100 cells per cm2. Cardiopoietic stem cell proliferation and purity was assessed by ArrayScan high-throughput multichannel fluorescence automated microscopy (Cellomics, Pittsburgh, http://www.cellomics.com), using anti-{alpha}-actinin antibody (1:1,000) to track sarcomerogenesis [14]. Nuclear translocation of cardiac transcription factors Nkx2.5 and MEF2C was monitored by confocal microscopy detection of antibodies (anti-Nkx2.5, 1:300 [Santa Cruz Biotechnology Inc., Santa Cruz, CA, http://www.scbt.com]; anti-MEF2C, 1:400 [Cell Signaling Technology, Beverly, MA, http://www.cellsignal.com]), using DAPI for nuclear counterstaining, and normalized to background using Metamorph (Molecular Devices Corp., Union City, CA, http://www.moleculardevices.com) [14, 25]. Network perturbation was carried out by inhibiting primary hubs, with cardiac differentiation monitored by epifluorescence using anti-{alpha}-actinin antibody and live microscopy [14].

Statistics
Comparison between groups was performed using a standard t test of variables with 95% confidence intervals. Data are expressed as mean ± SE.


   RESULTS
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 Conclusion
 Disclosure of Potential...
 Acknowledgments
 References
 
Embryoid Body Cardioinductive Priming by TNF{alpha}
ESCs form cardiac tissue arising from the mesodermal layer of differentiating EBs [21, 22]. Here, cardiac specification in EBs was demonstrated by immunohistochemical detection of the cardiac sarcomeric protein {alpha}-actinin (Fig. 1A, upper panel). Application of the reprogramming cytokine TNF{alpha} (30 ng/ml) promoted cardiac specification and sarcomerogenesis (Fig. 1A, lower panel). TNF{alpha} increased up to 2.5-fold the proportion of cells that underwent cardiac differentiation, with maximal cardiogenic effect attained after addition of the cytokine between day 2 and day 5 of differentiation (Fig. 1B), a time frame that coincides with endoderm-mediated mesodermal cardiac differentiation [14]. Moreover, the levels of TGFβ-1, a prototypic paracrine stimulator of cardiogenesis [21], were doubled following TNF{alpha} priming, as demonstrated in extracellular media of differentiating EBs (Fig. 1C). To determine whether a specific EB layer was the primary target of TNF{alpha} action, extracellular TGFβ-1 levels were also monitored in EBs made devoid of endoderm by treatment with leukemia inhibitory factor [40] and in a visceral endoderm-like cell line. Although TNF{alpha} priming was without significant effect on TGFβ-1 levels arising from endoderm-devoid ectoderm/mesoderm containing EBs, direct stimulation by the cytokine of isolated endoderm induced a surge in TGFβ-1 secretion, mimicking the response of intact EBs (Fig. 1C). Thus, TNF{alpha} promotes mesodermal cardiac differentiation by enhancing endoderm-dependent cardiogenic signaling.


Figure 1
View larger version (62K):
[in this window]
[in a new window]

  		Figure 1. TNF{alpha} potentiates mesodermal cardiac differentiation within EBs in an endoderm-dependent paracrine fashion and modulates the composition and activity of the Endo secretome. (A): Immunofluorescent detection of the cardiac sarcomeric protein {alpha}-actinin in EBs at day 9 of differentiation, indicating that the sarcomeric content in untreated EBs (upper panel) was increased by TNF{alpha} treatment (lower panel). Low-magnification epifluorescence microscopy of EBs is shown (upper insets; scale bars = 500 µm). High-magnification confocal microscopy indicates the presence of normal sarcomeric architecture (lower insets; scale bars = 20 µm). (B): Percentage increase in sarcomerogenesis following administration of TNF{alpha} during EB differentiation, determined at day 9 of differentiation in comparison with untreated controls. (C): Enzyme-linked immunosorbent assay quantification of TGFβ-1, a prototypic stimulator of cardiac differentiation, from conditioned medium of untreated and TNF{alpha}-treated EBs (EB), Ect/Mes, and isolated Endo cells. (D): Representative silver-stained two-dimensional gels (pH 3–10, 15% SDS-polyacrylamide gel electrophoresis) of proteins concentrated from 5 ml of unprimed (left) and TNF{alpha}-primed (right) endoderm-conditioned medium. Significantly more protein species were detected in gels of TNF{alpha}-treated (1,368 ± 19.1; n = 2 per treatment) versus untreated (1,070 ± 30.4; n = 2 per treatment) samples, and although most proteins increased in abundance, specific examples of downregulated proteins were also detected (insets [Da, Db]). Numbered spots (Da) correspond to identified proteins listed in Figure 3. Protein molecular mass markers are indicated. (E): PDQuest quantitative image analysis resolved 1,516 distinct protein species (supplemental online Table 1), binned by extent of TNF{alpha}-induced fold change and graphed by percentage of protein species per bin. (F): Activation of the Smad signaling pathway, indicative of nuclear transduction mediated by TGFβ-1-dependent signaling, in recipient ESCs cultured in monolayer, measured by immunofluorescent detection of intranuclear phosphorylated Smad3 (P-Smad3) after exposure to unprimed (–) and TNF{alpha}-primed (+) endoderm-conditioned medium. Nuclei were counterstained with 4,6-diamidino-2-phenylindole. *, p < .05 versus unprimed. Abbreviations: AU, arbitrary units; Ect/Mes, EBs devoid of endoderm; Endo, endodermal; P-Smad, phosphorylated Smad3; TGF, transforming growth factor; TNF{alpha}, tumor necrosis factor {alpha}.

 
TNF{alpha} Priming Transforms the Endodermal Secretome
To dissect the action of the cytokine on endodermal paracrine signaling [14], comparative proteomic analysis was here extended on conditioned medium of visceral endoderm-like cells cultured for 24 hours with and without TNF{alpha} (30 ng/ml) priming. The protein concentration of medium from cytokine-primed cells was 83% greater than that of the secretome of unstimulated visceral endoderm-like cells, 23.84 ± 2.15 versus 13.05 ± 0.75 µg/ml (p = .021), a difference evident in broad pH range 2D gels of volume-normalized protein retentate (Fig. 1D). Significantly more protein species were detected in replicate gels of TNF{alpha}-treated (1,381 and 1,354) versus untreated (1,048 and 1,091) samples (p = .0036). Densitometric quantification of relative spot intensities for the 1,516 unique protein species resolved at pH 3–10 revealed a distribution in the extent of protein change (Fig. 1E; supplemental online Table 1). Although 13% of protein species (197 spots) changed <1.5-fold in abundance, >75% (1,148 spots) increased at least 1.5-fold in cytokine-primed versus unprimed endodermal medium. Concordant with the near doubling of total protein, the spot intensity distribution peaked with 40% of all proteins (606 spots) increasing 1.5–5 fold (Fig. 1E). In contrast, 11.3% of protein species (171 spots) decreased >1.5-fold (Fig. 1E), including a number that were highly downregulated (e.g., Fig. 1D, insets 1Da, 1Db). In total, 482 protein species were significantly altered (p < .05), 442 of these increasing >1.5-fold, 33 decreasing >1.5-fold, and 7 changing <1.5-fold, indicating that TNF{alpha} elicited distinct rearrangement of the secretome rather than indiscriminate protein upregulation. Moreover, TNF{alpha} priming led to functional changes within the secretome, with the ability to alter signal transduction activity in recipient cells. Specifically, the abundance of phosphorylated Smad3, demonstrative of nuclear signal transduction activated by TGFβ-1 signaling [14, 21], was doubled in ESCs cultured with cytokine-stimulated versus unstimulated endodermal-conditioned medium (Fig. 1F). Thus, enhanced endodermal paracrine activity was associated with TNF{alpha}-mediated transformation of the secreted pool of proteins.

Endodermal Secretome Cartography
To determine specific protein alterations induced by TNF{alpha} priming, a discriminatory analysis of primed versus unprimed endoderm secretome was carried out by ion trap MS/MS of in-gel tryptic digests from resolved proteins. Assignments were obtained for 153 nonredundant 2D gel spots resolved at various pH gradients and acrylamide concentrations (Fig. 2A–2D; supplemental online Table 2) by matching derived spectra (e.g., Fig. 2E) either to multiple unique peptides per protein or, for proteins identified by a single peptide match, by manual inspection of the MS/MS spectrum. A total of 99 unique proteins accounted for all spots, with several proteins located at multiple positions and some spots comprising more than one protein. Two independent measures were used to validate protein identities: (a) uniformity between predicted Mr/pI values and those observed within the 2D gels, and (b) detection by complementary 2D LC-MS/MS shotgun analysis (supplemental online Table 2). Every identified protein was confirmed by one or both of the two additional methods, with more than 80% verified by either approach alone (84 of 99 by shotgun, 83 of 99 by Mr/pI uniformity), whereas two-thirds (68 of 99) were authenticated by the increased stringency of both measures (Fig. 3; supplemental online Tables 3 and 4).


Figure 2
View larger version (71K):
[in this window]
[in a new window]

  		Figure 2. Endodermal secretome cartography. The two-dimensional (2D) gel positions of proteins identified by linear ion trap quadrupole (LTQ) liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis of tryptic peptides extracted from resolved 2D gel spots were mapped onto gel images of TNF{alpha}-primed secretome resolved under various isoelectric focusing (pH) and SDS-polyacrylamide gel electrophoresis (%) conditions: (A): pH 4–7, 7.5%; (B, C): pH 3–10, 15%; (D): pH 6–11, 15%. Identified spots are circled and numbered, with the corresponding identities of definitive secretome proteins listed in Figure 3, and the remaining identified proteins listed in supplemental online Table 3. Spots circled in blue are duplicates from panels in Figures 1D (1–3), 2A (32–40), and 2C (146). Spots 134a and 134b in (D) were resolved at pH 6–11 from a single pH 3–10 spot (134) in (C). Insets in (B) enlarge a region to indicate identified spots, and insets in (C) enlarge a region to indicate the in-gel position of TNF{alpha} in TNF{alpha}-primed (upper panel) versus unprimed (lower panel) secretomes. Unnumbered circled spots were not identified but may be post-translationally modified forms of 118 and 119. (E): A representative LTQ LC-MS/MS product ion spectrum, obtained for spot 118 and modified from BioWorks 3.1 to indicate detected b-ions in red and y-ions in blue. The corresponding peptide sequence (inset) was identified as TNF{alpha} peptide 112–123. Abbreviations: a.a., amino acids; Mox, oxidized methionine; TNF, tumor necrosis factor.

 


Figure 3
View larger version (64K):
[in this window]
[in a new window]

  		Figure 3. Definitive endodermal secretome. Iterative bioinformatic screening was used to assess identified proteins for inclusion as definitive members of the secretome. Proteins with predicted amino-terminal signal peptides required for canonical endoplasmic reticulum/Golgi-dependent secretion were classified as canonical secretome proteins (yellow) on the basis of SignalP 3.0 HMM scoring (cutoff, >0.9). Proteins predicted to be secreted by measures independent of signal peptide transport that lacked signal peptides were categorized as noncanonical (blue) on the basis of SecretomeP 2.0 NN scoring (cutoff, >0.6). Note that many canonically secreted proteins also received high SecretomeP scores. Remaining proteins were then investigated for reported evidence of secretion or of function as secretion machinery components and categorized as empirical (orange) or accessory (green), respectively. Protein names are listed with their corresponding spot numbers to locate their two-dimensional (2D) gel position(s) in Figure 2. The Mascot score and number of unique peptides detected for identification of reported proteins (2DE score) are indicated. Proteins validated independently by 2D LC-tandem mass spectrometry (LC-MS/MS) and/or Mr/pI uniformity (for precursor or mature forms) are indicated by number of unique peptides detected during 2D LC-MS/MS and/or checked in the Mass/pI uniformity column, as applicable. Complete MS/MS data for all proteins is outlined in supplemental online Table 2. Fold change was calculated as described in Materials and Methods. *, TrEMBL accession numbers for two proteins lacking Swiss-Prot mouse entries that were positively identified from the UniProt TrEMBL database. **, For proteins detected in more than one spot, only the maximum score and corresponding number of unique peptides are reported. Abbreviations: BMP, bone morphogenetic protein; 2DE, two-dimensional gel electrophoresis; HMM, hidden Markov matrix; LC, liquid chromatography; NN, neural network; TNF, tumor necrosis factor.

 
Identified proteins were stratified by iterative bioinformatic screening using established criteria to restrict the search for bona fide secretome components. Using the SignalP 3.0 algorithm for detection of amino-terminal signal peptides indicative of canonical ER/Golgi-dependent secretion [31], 28 proteins were categorized as "canonical" secretome proteins (Fig. 3). A separate algorithm, SecretomeP 2.0, predicting proteins secreted by measures independent of the presence of a canonical signal peptide [32], identified an additional 13 proteins secreted through a "noncanonical" process (Fig. 3). Consistent with the definition of secretome [16], seven additional proteins were included on the basis of reported "empirical" evidence of secretion or as "accessory" secretion machinery components (Fig. 3). Among the definitive secretome, independent validation measures were consistent with those of the overall group, as 42 of 48 proteins (88%) were also identified by 2D LC-MS/MS, 38 of 48 (79%) by Mr/pI uniformity, and 32 of 48 (67%) by both measures (Fig. 3). In addition, nearly 70% of these proteins were modulated by TNF{alpha} priming, with quantitation of biological replicates indicating that 32 increased 1.5-fold or more, 1 was downregulated >4-fold, and the remaining 15 were unchanged, differing <1.5-fold up or down (Fig. 3). Thus, this complementary protein identification strategy, combined with informatic stratification criteria, categorically mapped a definitive set of 48 secretome proteins within the TNF{alpha}-primed endodermal-conditioned medium.

Endodermal Secretome Network Assembly and Topography
Network analysis revealed 105 nodes linked by 334 interactions or edges, integrating 47 of the 48 secretome proteins in an organized assemblage (Fig. 4). TNF{alpha}, with 74 paired interactions, had the highest number of directly connected neighbors, suggesting that the composite network was a nonstochastic derivation of TNF{alpha} priming. The nonrandomness of network topology was corroborated by derivation of interrelationships between the network variables node degree (k), node degree distribution (P[k]), and clustering coefficient distribution (C[k]). A normal distribution for P[k] versus k, indicative of a random Erdös-Rényi network [41], was ruled out by the Anderson-Darling normality test (p < .0001). Instead, the network possessed scale-free topology (Fig. 4, lower left), following a power law distribution where P[k]~k–{gamma} [34] with {gamma} = 1.781 ± 0.076. Moreover, the log-log plot of C[k] versus k also exhibited a power law distribution with high correlation (R2 = 0.9417), where C[k]~k–0.8041 (Fig. 4, lower left inset), indicating hierarchical modularity within the network [37]. In the network, kaverage = 6.038, and only 17/105 nodes had k>kaverage, typifying hierarchical networks where small proportions of nodes possess numerous connections [34, 35, 37]. TNF{alpha} (k = 74) and TGFβ-1 (k = 42) exhibited greatest connectivity, followed by interferon-{gamma}, k = 41; p53, k = 34; Hras, k = 31; and interleukin (IL)-4, k = 30. Thus, the derived secretome forms a nonstochastic network with scale-free, hierarchical architecture.


Figure 4
View larger version (71K):
[in this window]
[in a new window]

  		Figure 4. Endodermal secretome protein-protein interaction composite network. Definitive secretome proteins (colored nodes) were submitted to Ingenuity Pathways Analysis as focus nodes, of which 47 possessed curated interactions, generating a TNF{alpha}-centric, 105-protein interaction network. Nodes are listed by gene name for simplicity, with focus proteins indicated in parentheses in Figure 3 with their corresponding protein names. Node colors indicate the extent of TNF{alpha}-induced expression level change (legend) and correspond to bin coloring in Figure 1E, whereas network nodes not detected during proteomic analysis are unfilled. A plot of degree distribution (P[k]) versus degree (k) followed a power law distribution (lower left), indicating a scale-free, nonstochastic network architecture. A log-log plot of clustering coefficient distribution (C[k]) versus k (lower left inset) indicated that connectivity within protein neighborhoods also followed a power law distribution consistent with hierarchical tendencies. The most highly connected nodes, which serve as hubs within the network, are noted: Tnf, Tgfb1, Ifng, Tp53, Hras, and Il4. Protein names of additional nodes (in white) can be accessed online through the Mouse Genome Informatics portal (http://www.informatics.jax.org). Abbreviations: Hras, Harvey rat sarcoma virus oncogene 1; Ifng, interferon-{gamma}; Il4, interleukin-4; Tgfb1, transforming growth factor β-1; Tnf, tumor necrosis factor {alpha}; Tp53, transformation-related protein 53.

 
A defining feature of the composite network was the identity and rank order of its three formative modules (Fig. 5). Cardiovascular system growth and development was the most prominent subnetwork, containing 22 focus proteins (Fig. 5A). Correlating with the composite, this subnetwork demonstrated a TNF{alpha}-focused structure and contained elements of the IL-6 (nodes: Tnf, Col1a1) and integrin signaling pathways (Col1a1, Col1a2, Col4a3), with additional signaling pathways linked to TNF{alpha} (Fig. 5A). Cellular assembly and organization, established around 13 focus proteins, was the second most prominent subnetwork, supported by the phosphatidylinositol-3-kinase/protein kinase B (Akt) (Akt2, Tp53, Chuk, Ywhae), wingless-type (Wnt)/β-catenin (Akt2, Tgfb1, Tp53), and nuclear factor-{kappa}B (Akt2, Chuk, Ube2v1) pathways (Fig. 5B). Finally, the cell-to-cell signaling and interaction subnetwork, built around a core of 12 focus proteins, was supported by the TGFβ-1 signaling cascade (Hras, Smad2) and additional growth factor and signaling pathways linked to Hras, as well as interferon-{gamma} (Ifng) and IL-4 (Il4) signaling pathways (Fig. 5C). Thus, the endodermal secretome network displayed TNF{alpha}-centric behavior and was composed of subnetworks involved in cardiovascular system development, cellular assembly and organization, and cell-to-cell signaling and interaction.


Figure 5
View larger version (61K):
[in this window]
[in a new window]

  		Figure 5. Endodermal secretome functional subnetworks. Resident subnetworks within the composite are functionally distinguishable, as the 47 secretome focus nodes can be segregated into three functional subnetworks. (A): The cardiovascular growth and development subnetwork contains 22 focus proteins, the highest number of interconnected secretome components of any subnetwork, and is also TNF{alpha}-centric, comprising elements of the interleukin (IL)-6 and integrin signaling pathways. (B): The cellular assembly and organization subnetwork included 13 focus proteins, comprising components of the phosphatidylinositol-3-kinase/Akt, Wnt/β-catenin, and nuclear factor-{kappa}B signaling pathways. (C): The cell-to-cell signaling and interaction subnetwork including 12 focus proteins integrated constituents of the transforming growth factor β-1, interferon, and IL-4 signaling pathways. Node colors indicate the extent of TNF{alpha}-induced expression level change (legend) and correspond to colors used in Figures 1E and 4, whereas network nodes not detected during proteomic analysis are unfilled. Nodes are listed by gene name for simplicity, with focus proteins indicated in parentheses in Figure 3 with their corresponding protein names, whereas the remaining protein nodes (in white) can be accessed online through the Mouse Genome Informatics portal. Abbreviation: TNF{alpha}, tumor necrosis factor {alpha}.

 
Primacy of Cardiovascular Developmental Function and Network Robustness
Exclusion of TNF{alpha} from network generation demoted the primary ranking of cardiovascular development, substantiating the role of TNF{alpha} in facilitating cardiogenesis (Fig. 6A). Indeed, the resolved elements of the TNF{alpha}-primed secretome perform specialized functions permissive of cardiovascular development, including homeostasis regulation (e.g., superoxide dismutase 1, thioredoxin 1, and ubiquitin) [42–45], extracellular matrix organization (e.g., galectin-1, gelsolin, nidogen 2, and perlecan) [46–50], and Ca2+-dependent signaling (e.g., calmodulin and calreticulin) [51–54], and include proteins involved in growth factor activation and signaling (e.g., cyclophilin A, macrophage migration inhibitory factor [MIF], procollagen C-proteinase [bone morphogenetic protein 1] enhancer, and perlecan) [55–61] (Fig. 6B). Thus, the TNF{alpha}-dependent prominence of cardiovascular development suggested by ontological stratification correlates with TNF{alpha}-mediated upregulation of proteins supporting cardiogenic processes.


Figure 6
View larger version (67K):
[in this window]
[in a new window]

  		Figure 6. Network functions, robustness, and sensitivity to targeted hub attack. (A): Developmental functions associated with the secretome network as ranked by significance, determined by Ingenuity Pathways Knowledge Base, when TNF{alpha} is included (upper histogram) or excluded (lower histogram) as a focus protein for network analysis. (B): Enlarged regions of two-dimensional gel spots (left) indicating examples of identified secretome protein expression levels before (upper 2D) and after (lower 2D) TNF{alpha} treatment and depicted by three-dimensional surface profiles of unique identified spots (adjacent panels). Relative levels of expression in all biological replicates (n = 2 per treatment) are shown in the histograms to the right, in arbitrary units, with the lowest-expressing replicate normalized to 1.0 and ordered by increasing expression abundance. In all cases, expression was greater in TNF{alpha}-treated samples. (C): Immunofluorescent detection of cardiac {alpha}-actinin at day 9 of differentiation using low-magnification epifluorescence microscopy to indicate EB sarcomeric content, which is increased by TNF{alpha} treatment (lower left) relative to untreated EBs (upper left). Targeted perturbation of the two primary network hubs, TNF{alpha} and transforming growth factor β-1 (TGFβ-1), by addition of either the TNF{alpha} inhibitor infliximab at 150 ng/ml (middle panels) or the TGFβ-1 inhibitor LAP at 25 ng/ml (right panels) did not obviate cardiac differentiation (upper panels) but did reduce the procardiogenic effects of TNF{alpha} stimulation (lower panels). (D): Independent quantitative analysis of the treatments in (C), determined by video microscopy evaluation of beating area as a percentage of total EB area (n = 20 EBs assessed per treatment). *, p < .05 versus unprimed EBs; **, p < .05 versus EBs treated with either TNF{alpha} or infliximab. Abbreviations: BMP, bone morphogenetic protein; LAP, latency-associated peptide; PCP, procollagen proteinase; SOD, superoxide dismutase; TNF{alpha}, tumor necrosis factor {alpha}.

 
The two most highly connected hubs, TNF{alpha} and TGFβ-1, were selectively antagonized to assess network robustness [62]. The TNF{alpha}-induced increase in cardiac {alpha}-actinin-positive cells (Fig. 6C, left upper and lower panels) was blunted by the TNF{alpha} inhibitor infliximab (at 150 ng/ml), which alone did not prevent cardiogenesis (Fig. 6C, middle upper and lower panels). Similarly, inhibition of TGFβ-1 by addition of 25 ng/ml latency-associated peptide (LAP) negated the procardiogenic effect of TNF{alpha} without disrupting basal levels of cardiac differentiation (Fig. 6C, right upper and lower panels). Consistent with {alpha}-actinin measurements, infliximab elicited a significant reduction in, and LAP precluded, the TNF{alpha}-induced increase in cardiac beating area (Fig. 6D). Thus, the responsiveness to targeted node attack indicated the importance of identified primary network hubs in promoting cardiac differentiation.

Functional Validation of TNF{alpha} Derived Network in Securing Embryoid Body-Independent Cardiopoiesis
ESCs can be guided toward cardiogenic differentiation, yet this process is inefficient outside a differentiating embryo or EB [14]. To validate the cardioinductive potential and functional cooperativity of TNF{alpha}-primed network constituents, cardiac specification, or cardiopoiesis [14, 63], was followed in ESC monolayers (Fig. 7). In contrast to the limited aptitude of unprimed endoderm secretome in committing pluripotent stem cells to the cardiac fate, TNF{alpha}-primed secretome fostered nuclear translocation of cardiac transcription factors (Fig. 7A, 7B), critical for initiation of cardiogenesis. Early in the differentiation process, at day 4, stem cells treated with TNF{alpha}-primed endoderm secretome displayed four- and sixfold increases in nuclear localization of cardiac transcription factors Nkx2.5 and MEF2C, respectively (Fig. 7A, 7B). Over time, TNF{alpha} priming significantly increased the overall efficacy of the differentiation process, securing, at day 7, a threefold increase in cardiac progenitors (Fig. 7C), an intermediate phenotype in the continuum of stem cell-based cardiogenesis [14]. By day 12, TNF{alpha} priming translated into a threefold greater yield of stem cell-derived cardiomyocytes, demonstrating definitive expression and sarcomeric organization of the myofibrillar protein {alpha}-actinin (Fig. 7D). Thus, TNF{alpha} priming of the endoderm secretome potentiated cardiac commitment and sarcomerogenesis of ESCs, validating the cardioinductive properties and synergistic actions of the cytokine-primed endoderm-derived protein network.


Figure 7
View larger version (58K):
[in this window]
[in a new window]

  		Figure 7. TNF{alpha} priming potentiates endodermal secretome-mediated ESC cardiac specification. The cardiogenic potential of the TNF{alpha}-primed secretome was validated by 12 D monitoring of ESC cardiac specification by immunofluorescent detection of cardiac-specific proteins in cell monolayers exposed to unprimed (left images, [–] in histogram) versus TNF{alpha}-primed endoderm-conditioned medium (right images, [+] in histogram). At D4, nuclear translocation of the cardiac-specific transcription factors Nkx2.5 (A) and MEF2C (B) was four- and sixfold greater, respectively, in cells treated with TNF{alpha}-primed secretome (n = 30 nuclei each). Nuclear fluorescence intensity was measured by cross-sectional densitometry (yellow bars). (C): By D7, treatment with TNF{alpha}-primed secretome translated into threefold greater abundance of cardiopoietic (cardiac precursor) stem cells, as determined by the number of cells containing nuclear MEF2C (n = 48 wells of a 96-well enzyme-linked immunosorbent assay [ELISA] plate measured for each treatment). (D): By completion of differentiation at D12, the TNF{alpha}-primed secretome mediated the production of threefold more stem cell-derived cardiomyocytes than did unprimed secretome (n = 48 wells of a 96-well ELISA plate measured for each treatment). In all panels, nuclei were counterstained with DAPI. *, p < .05 versus unprimed. Abbreviations: AU, arbitrary units; D, day; DAPI, 4',6-diamidino-2-phenylindole; MEF2C, myocyte-specific enhancer factor 2C; TNF{alpha}, tumor necrosis factor{alpha}.

 

   DISCUSSION
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
Conclusion
 Disclosure of Potential...
 Acknowledgments
 References
 
Paracrine signals of endodermal origin are necessary and sufficient for cardiogenic specification of ESCs, yet this cardioinductive milieu remains largely undefined. The composition of the endoderm-released subproteome was resolved here by tandem proteomic approaches and stratified by bioinformatic iteration to define canonical and noncanonical secretome components. Endoderm priming with the reprogramming cytokine TNF{alpha} exposed a nonstochastic, hierarchical secretome network. Constituent subnetworks indicated the prevalence of cardiovascular development within the TNF{alpha}-centric network, verified in silico and validated in vitro by targeted hub perturbation. TNF{alpha} priming of the endoderm secretome translated into enhanced acquisition of the cardiac phenotype from pluripotent stem cells, independent of embryogenesis. These findings provide a proteomic foundation from which to develop a bona fide systems understanding of guided developmental cardiogenesis.

Proteomic Strategy
Secretome analysis is inherently limited by low protein concentration and obfuscation by protein contaminants from serum or of intracellular origin. Here, endodermal cells were cultured serum-free, with the secretome harvested within 24 hours to minimize the potential for cell death [64]. To facilitate protein detection by 2DE, the endodermal secretome was concentrated 170- to 200-fold by filter centrifugation. Two-dimensional gel resolution was optimized by substituting 2-hydroxyethyl disulfide for dithiothreitol in IEF buffer for all pH gradients, and identified proteins were independently verified by 2D LC-MS/MS, Mr/pI uniformity, or both. Shotgun 2D LC-MS/MS proved beneficial for confirmation of proteins matched by only 1–2 peptides during the 2DE analysis. Application of this multiple validation strategy restricted the number of positive matches and, in combination with the sample preparation to reduce artifacts, maximized confidence in the accuracy of the acquired secretome.

Secretome Categorization
The secretome is composed of secreted proteins, as well as the protein machinery involved in the secretion process [16]. The standard criterion for eukaryotic protein secretion is amino-terminal signal peptide-mediated transport [19, 20], defined herein as canonical. From the identified endodermal proteins, 28 were categorized as canonically secreted by SignalP 3.0 HMM scoring [31]. Although numerous alternative algorithms exist, the identical list of 28 proteins was obtained when using the three-algorithm combination found to provide maximum sensitivity and specificity for signal peptide detection [65]. This restrictive classification, however, excludes proteins secreted through unconventional means. Eukaryotes are postulated to exploit still-undefined secretion pathways [19, 20]. Indeed, several mechanisms have been described for prokaryotes [66]. Certain intracellular proteins are commonly detected in secretome analyses across multiple cytotypes [67–69], with potential for extracellular moonlighting functions [70, 71]. A priori prediction of nonclassic protein secretion [32], revealed 13 additional noncanonical components of the secretome. Seven proteins that did not fulfill criteria for being either canonical or noncanonical but have been experimentally confirmed to be secreted [55, 72–74] or to contribute to the secretion process [72, 75, 76] were categorized as empirical or accessory members of the secretome. Thus, comprehensive categorization of secreted proteins stratified the endodermal secretome, identifying a set of 48 components.

Network Analysis
Although comprising diverse functional categories, the 47 protein members of the endodermal secretome clustered into an integrated, nonrandom 105-node network with definitive architectural features. The resolved network fulfilled criteria of nonstochastic topography based on collective indices of network connectedness, including node degree (k), degree distribution (P[k]), and clustering coefficient distribution (C[k]) [37]. Whereas random Erdös-Rényi networks possess a normal distribution for P[k] versus k [41], the log-log plots of P[k] versus k and C[k] versus k, described here, fit power law distributions indicative of a scale-free network with hierarchical architecture [34, 35, 37]. Moreover, with a {gamma} exponent of 0.8041, the C[k] versus k relationship approximates an ideal scaling of C[k]~k–1, a vital feature of hierarchical modularity and a characteristic extension of scale-free architecture [37]. Deconvolution of the composite into discrete, nonoverlapping subnetworks provided evidence of modularity, as groups of nodes cluster into functional neighborhoods [77]. Based on functional annotation, nearly half of the endodermal secretome nodes mapped to the cardiovascular growth and development category, with remaining nodes mapping cellular assembly and organization or cell-to-cell signaling and interaction. The primacy of cardiovascular growth and development exposes the proteomic substrate underlying procardiogenic properties of the endoderm [13, 14]. The major hub of the composite network and primary subnetwork was TNF{alpha}, with 74 independent protein links. The centrality of this reprogramming cytokine provides a proteomic foundation for the role of TNF{alpha} as a potent inducer of endoderm-mediated cardioinduction through activation of specific signaling pathways [14]. Indeed, targeted hub attack verified the importance of the cytokine to the primacy of cardiovascular development within the assembled network. A number of proteins found to be upregulated by TNF{alpha} support developmental processes through cellular functions, including homeostasis regulation, extracellular matrix organization, Ca2+-dependent signaling, and growth factor activation [42–61]. Other identified network proteins that support cardiogenic processes (e.g., tissue inhibitor of metalloproteinases-2, matrix metalloproteinase-2, and secreted protein acidic and rich in cysteine) may further contribute to the cardioinductive outcome [78–81], despite being unchanged in abundance following TNF{alpha} priming. Corroboration of TNF{alpha}-centric network robustness [62] was demonstrated through inhibition of the primary hubs, TNF{alpha} and TGFβ-1. This blunted the cardiogenic aptitude of TNF{alpha} priming, indicating the significance of these hubs in the cardiogenic process, consistent with TNF{alpha} and TGFβ-1 serving as major stimulators of cardiac differentiation [14, 21]. Of note, a basal level of cardiac differentiation, leading to initiation of contractility, proceeded even without TNF{alpha} or TGFβ-1 priming. Thus, the constitutive set of network proteins contributes to cardiogenesis, with targeted hub manipulation modulating outcome.

Secretome Functional Biology
Validation of network functional cooperativity was obtained experimentally through direct application of the TNF{alpha}-primed endodermal secretome on naïve ESCs. Unprimed secretome provided a milieu conducive to cardiac differentiation, yet the ESC-derived cardiomyocyte yield was limited. TNF{alpha} priming rendered the collective secretome highly cardiogenic, potentiating nuclear translocation of cardiac transcription factors MEF2C and Nkx2.5 early in differentiation, ultimately tripling the yield of cardiac progeny. Endodermal cells have been shown to stimulate both embryonic carcinoma cells and ESC cardiac differentiation during coculture, but detailed aspects of the cardioinductive determinants remained uncharacterized [13, 82]. The present results indicate that the endodermal secretome alone, in the absence of coculture, is sufficient to mediate cardiac differentiation. TNF{alpha} was found, moreover, to stimulate the cardiogenic aptitude of the endodermal secretome. Indeed, in vivo heart-restricted overexpression of this reprogramming cytokine promotes cardiac specification of implanted ESCs [14]. Microarray data indicated that TGFβ-activated kinase 1 and p38 mitogen-activated protein (MAP) kinase were upregulated in endodermal cells upon TNF{alpha} priming, and pharmacological inhibition of p38 confirmed its importance for cardioinduction [14]. This class of MAP kinase is also involved in secretory processes [83, 84], providing a link to the upregulated secretome proteins. These in vivo and in vitro observations extend our knowledge of the endodermal composition and of a dynamic role for TNF{alpha} in potentiating paracrine effects on cardiac differentiation.


   CONCLUSION
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Conclusion
Disclosure of Potential...
 Acknowledgments
 References
 
Proteomic dissection applied here and translated by a multimodal systems biology approach provides an objective framework [85–88] underlying cytokine-primed, endoderm-guided cardiogenesis. Tandem mass spectrometric investigation of the constitutive endodermal secretome, resolved by complementary 2DE and 2D LC, revealed a set of paracrine effectors assembled in a nonstochastic protein interaction network. Network function was validated by manipulation of primary hubs, modulating the effectiveness of cardiac phenotype acquisition in naïve ESCs. Thus, proteomic network resolution establishes a malleable scaffold amenable to systems interrogation of the guided transition from pluripotency to lineage specification.


   DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Conclusion
 Disclosure of Potential...
Acknowledgments
 References
 
The authors indicate no potential conflicts of interest.


   ACKNOWLEDGMENTS
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Conclusion
 Disclosure of Potential...
 Acknowledgments
References
 
We thank David Muddiman and the Mayo Proteomics Research Center staff, particularly Benjamin Madden, Kenneth Johnson, and Christopher Mason, for expert guidance in mass spectrometry and bioinformatic analysis. Support was provided by the NIH, Marriott Heart Disease Research Program, Marriott Foundation, Ted Nash Long Life Foundation, Asper Foundation, and Mayo Clinic. D.K.A. was supported by the Heart and Stroke Foundation of Canada.


   REFERENCES
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Conclusion
 Disclosure of Potential...
 Acknowledgments
 References
 

  1. Srivastava D, Olson EN. A genetic blueprint for cardiac development. Nature 2000;407:221–226.[CrossRef][Medline]

  2. Buckingham M, Meilhac S, Zaffran S. Building the mammalian heart from two sources of myocardial cells. Nat Rev Genet 2005;6:826–835.[CrossRef][Medline]

  3. Foley A, Mercola M. Heart induction: Embryology to cardiomyocyte regeneration. Trends Cardiovasc Med 2004;14:121–125.[CrossRef][Medline]

  4. Olson EN. Gene regulatory networks in the evolution and development of the heart. Science 2006;313:1922–1927.[Abstract/Free Full Text]

  5. Schultheiss TM, Xydas S, Lassar AB. Induction of avian cardiac myogenesis by anterior endoderm. Development 1995;121:4203–4214.[Abstract]

  6. Olson EN, Srivastava D. Molecular pathways controlling heart development. Science 1996;272:671–676.[Abstract]

  7. Lough J, Sugi Y. Endoderm and heart development. Dev Dyn 2000;217:327–342.[CrossRef][Medline]

  8. Harvey RP. Patterning the vertebrate heart. Nat Rev Genet 2002;3:544–556.[CrossRef][Medline]

  9. Lin Q, Schwarz J, Bucana C et al. Control of mouse cardiac morphogenesis and myogenesis by transcription factor MEF2C. Science 1997;276:1404–1407.[Abstract/Free Full Text]

  10. Garry DJ, Olson EN. A common progenitor at the heart of development. Cell 2006;127:1101–1104.[CrossRef][Medline]

  11. Thomson JA, Itskovitz-Eldor J, Shapiro SS et al. Embryonic stem cell lines derived from human blastocysts. Science 1998;282:1145–1147.[Abstract/Free Full Text]

  12. Smith AG. Embryo-derived stem cells: Of mice and men. Annu Rev Cell Dev Biol 2001;17:435–462.[CrossRef][Medline]

  13. Mummery C, Ward-van Oostwaard D, Doevendans P et al. Differentiation of human embryonic stem cells to cardiomyocytes: Role of coculture with visceral endoderm-like cells. Circulation 2003;107:2733–2740.[Abstract/Free Full Text]

  14. Behfar A, Perez-Terzic C, Faustino RS et al. Cardiopoietic programming of embryonic stem cells for tumor-free heart repair. J Exp Med 2007;204:405–420.[Abstract/Free Full Text]

  15. Hodgson DM, Behfar A, Zingman LV et al. Stable benefit of embryonic stem cell therapy in myocardial infarction. Am J Physiol Heart Circ Physiol 2004;287:H471–H479.[Abstract/Free Full Text]

  16. Tjalsma H, Bolhuis A, Jongbloed JDH et al. Signal peptide-dependent protein transport in Bacillus subtilis: A genome-based survey of the secretome. Microbiol Mol Biol Rev 2000;64:515–547.[Abstract/Free Full Text]

  17. Walter P, Gilmore R, Blobel G. Protein translocation across the endoplasmic reticulum. Cell 1984;38:5–8.[Medline]

  18. Mellman I, Warren G. The road taken: Past and future foundations of membrane traffic. Cell 2000;100:99–112.[CrossRef][Medline]

  19. Nickel W. The mystery of nonclassic protein secretion. A current view on cargo proteins and potential export routes. Eur J Biochem 2003;270:2109–2119.[Medline]

  20. Nickel W. Unconventional secretory routes: Direct protein export across the plasma membrane of mammalian cells. Traffic 2005;6:607–614.[CrossRef][Medline]

  21. Behfar A, Zingman LV, Hodgson DM et al. Stem cell differentiation requires a paracrine pathway in the heart. FASEB J 2002;16:1558–1566.[Abstract/Free Full Text]

  22. Perez-Terzic C, Behfar A, Mery A et al. Structural adaptation of the nuclear pore complex in stem cell-derived cardiomyocytes. Circ Res 2003;92:444–452.[Abstract/Free Full Text]

  23. Chung S, Dzeja PP, Faustino RS et al. Mitochondrial oxidative metabolism is required for the cardiac differentiation of stem cells. Nat Clin Pract Cardiovasc Med 2007;4:S60–S67.[CrossRef][Medline]

  24. Arrell DK, Elliott ST, Kane LA et al. Proteomic analysis of pharmacological preconditioning. Novel protein targets converge to mitochondrial metabolism pathways. Circ Res 2006;99:706–714.[Abstract/Free Full Text]

  25. Perez-Terzic C, Faustino RS, Boorsma BJ et al. Stem cell transformation into cardiac phenotype guided by directed remodeling of nuclear transport machinery. Nat Clin Prac Cardiovasc Med 2007;4:S1–S9.[CrossRef]

  26. Link AJ, Eng J, Schieltz DM et al. Direct analysis of protein complexes using mass spectrometry. Nat Biotechnol 1999;17:676–682.[CrossRef][Medline]

  27. Perkins DN, Pappin DJC, Creasy DM et al. Probability-based protein identification by searching sequence databases using mass spectrometry data. Electrophoresis 1999;20:3551–3567.[CrossRef][Medline]

  28. Ducret A, Van Oostveen I, Eng JK et al. High throughput protein characterization by automated reverse-phase chromatography/electrospray tandem mass spectrometry. Protein Sci 1998;7:706–719.[Medline]

  29. Elias JE, Gygi SP. Target-decoy search strategy for increased confidence in large-scale protein identifications by mass spectrometry. Nat Methods 2007;4:207–214.[CrossRef][Medline]

  30. Nesvizhskii AI, Aebersold R. Interpretation of shotgun proteomic data: The protein inference problem. Mol Cell Proteomics 2005;4:1419–1440.[Abstract/Free Full Text]

  31. Bendtsen JD, Nielsen H, von Heijne G et al. Improved prediction of signal peptides: SignalP 3.0. J Mol Biol 2004;340:783–795.[CrossRef][Medline]

  32. Bendtsen JD, Jensen LJ, Blom N et al. Feature based prediction of non-classic and leaderless protein secretion. Protein Eng Des Sel 2004;17:349–356.[Abstract/Free Full Text]

  33. Shannon P, Markiel A, Ozier O et al. Cytoscape: A software environment for integrated models of biomolecular interaction networks. Genome Res 2003;13:2498–2504.[Abstract/Free Full Text]

  34. Barabási A-L, Albert R. Emergence of scaling in random networks. Science 1999;286:509–512.[Abstract/Free Full Text]

  35. Jeong H, Tombor B, Albert R et al. The large-scale organization of metabolic networks. Nature 2000;407:651–654.[CrossRef][Medline]

  36. Watts DJ, Strogatz SH. Collective dynamics of ‘small-world’ networks. Nature 1998;393:440–442.[CrossRef][Medline]

  37. Barabási A-L, Oltvai ZN. Network biology: Understanding the cell's functional organization. Nat Rev Genet 2004;5:101–113.[CrossRef][Medline]

  38. Anderson TW, Darling DA. Asymptotic theory of certain "goodness-of-fit" criteria based on stochastic processes. Ann Math Stat 1952;23:193–212.[CrossRef]

  39. Newman MEJ. The structure and function of complex networks. SIAM Rev 2003;45:167–256.[CrossRef]

  40. Bader A, Gruss A, Hollrigl A et al. Paracrine promotion of cardiomyogenesis in embryoid bodies by LIF modulated endoderm. Differentiation 2001;68:31–43.[CrossRef][Medline]

  41. Erdös P, Rényi A. On the evolution of random graphs. Publ Math Inst Hung Acad Sci 1960;5:17–61.

  42. Corti S, Locatelli F, Donadoni C et al. Wild-type bone marrow cells ameliorate the phenotype of SOD1–G93A ALS mice and contribute to CNS, heart and skeletal muscle tissues. Brain 2004;127:2518–2532.[Abstract/Free Full Text]

  43. Kobayashi-Miura M, Shioji K, Hoshino Y et al. Oxygen sensing and redox signalling: The role of thioredoxin in embryonic development and cardiac diseases. Am J Physiol Heart Circ Physiol 2007;292:H2040–H2050.[Abstract/Free Full Text]

  44. Tetzlaff MT, Yu W, Li M et al. Defective cardiovascular development and elevated cyclin E and Notch proteins in mice lacking the Fbw7 F-box protein. Proc Natl Acad Sci U S A 2004;101:3338–3345.[Abstract/Free Full Text]

  45. Ko JA, Kimura Y, Matsuura K et al. PDZRN3 (LNX3, SEMCAP3) is required for the differentiation of C2C12 myoblasts into myotubes. J Cell Sci 2006;119:5106–5113.[Abstract/Free Full Text]

  46. Goldring K, Jones GE, Thiagarajah R et al. The effect of galectin-1 on the differentiation of fibroblasts and myoblasts in vitro. J Cell Sci 2002;115:355–366.[Abstract/Free Full Text]

  47. Chan J, O'Donoghue K, Gavina M et al. Galectin-1 induces skeletal muscle differentiation in human fetal mesenchymal stem cells and increases muscle regeneration. STEM CELLS 2006;24:1879–1891.[Abstract/Free Full Text]

  48. Plageman TF Jr, Yutzey KE. Microarray analysis of Tbx5-induced genes expressed in the developing heart. Dev Dyn 2006;235:2868–2880.[CrossRef][Medline]

  49. Bader BL, Smyth N, Nedbal S et al. Compound genetic ablation of nidogen 1 and 2 causes basement membrane defects and perinatal lethality in mice. Mol Cell Biol 2005;25:6846–6856.[Abstract/Free Full Text]

  50. Costell M, Carmona R, Gustafsson E et al. Hyperplastic conotruncal endocardial cushions and transposition of great arteries in perlecan-null mice. Circ Res 2002;91:158–164.[Abstract/Free Full Text]

  51. Houston DS, Carson CW, Esmon CT. Endothelial cells and extracellular calmodulin inhibit monocyte tumor necrosis factor release and augment neutrophil elastase release. J Biol Chem 1997;272:11778–11785.[Abstract/Free Full Text]

  52. Berger I, Bieniossek C, Schaffitzel C et al. Direct interaction of Ca2+/calmodulin inhibits histone deacetylase 5 repressor core binding to myocyte enhancer factor 2. J Biol Chem 2003;278:17625–17635.[Abstract/Free Full Text]

  53. Li J, Pucéat M, Perez-Terzic C et al. Calreticulin reveals a critical Ca2+ checkpoint in cardiac myofibrillogenesis. J Cell Biol 2002;158:103–113.[Abstract/Free Full Text]

  54. Lynch JM, Chilibeck K, Qui Y et al. Assembling pieces of the cardiac puzzle; calreticulin and calcium-dependent pathways in cardiac development, health, and disease. Trends Cardiovasc Med 2006;16:65–69.[CrossRef][Medline]

  55. Jin ZG, Melaragno MG, Liao DF et al. Cyclophilin A is a secreted growth factor induced by oxidative stress. Circ Res 2000;87:789–796.[Abstract/Free Full Text]

  56. Hong F, Lee J, Song JW et al. Cyclosporin A blocks muscle differentiation by inducing oxidative stress and inhibiting the peptidyl-prolyl-cis-trans isomerase activity of cyclophilin A: Cyclophilin A protects myoblasts from cyclosporin A-induced cytotoxicity. FASEB J 2002;16:1633–1635.[Abstract/Free Full Text]

  57. Suzuki M, Takamura Y, Maeno M et al. Xenopus laevis macrophage migration inhibitory factor is essential for axis formation and neural development. J Biol Chem 2004;279:21406–21414.[Abstract/Free Full Text]

  58. Ge G, Greenspan DS. BMP1 controls TGFbeta1 activation via cleavage of latent TGFbeta-binding protein. J Cell Biol 2006;175:111–120.[Abstract/Free Full Text]

  59. Ge G, Greenspan DS. Developmental roles of the BMP1/TLD metalloproteinases. Birth Defects Res C Embryo Today 2006;78:47–68.[CrossRef][Medline]

  60. Aviezer D, Hecht D, Safran M et al. Perlecan, basal lamina proteoglycan, promotes basic fibroblast growth factor-receptor binding, mitogenesis, and angiogenesis. Cell 1994;79:1005–1013.[CrossRef][Medline]

  61. Ibrahimi OA, Zhang F, Hrstka SC et al. Kinetic model for FGF, FGFR, and proteoglycan signal transduction complex assembly. Biochemistry 2004;43:4724–4730.[CrossRef][Medline]

  62. Albert R, Jeong H, Barabási A-L. Error and attack tolerance of complex networks. Nature 2000;406:378–382.[CrossRef][Medline]

  63. Behfar A, Terzic A. Derivation of a cardiopoietic population from human mesenchymal stem cells yields cardiac progeny. Nat Clin Pract Cardiovasc Med 2006;3:S78–S82.[CrossRef][Medline]

  64. Mbeunkui F, Fodstad O, Pannell LK. Secretory protein enrichment and analysis: An optimized approach applied on cancer cell lines using 2D LC-MS/MS. J Proteome Res 2006;5:899–906.[CrossRef][Medline]

  65. Klee EW, Ellis LB. Evaluating eukaryotic secreted protein prediction. BMC Bioinformatics 2005;6:256.[CrossRef][Medline]

  66. Tjalsma H, Antelmann H, Jongbloed JD et al. Proteomics of protein secretion by Bacillus subtilis: Separating the "secrets" of the secretome. Microbiol Mol Biol Rev 2004;68:207–233.[Abstract/Free Full Text]

  67. Grønborg M, Kristiansen TZ, Iwahori A et al. Biomarker discovery from pancreatic cancer secretome using a differential proteomic approach. Mol Cell Proteomics 2006;5:157–171.[Abstract/Free Full Text]

  68. Pellitteri-Hahn MC, Warren MC, Didier DN et al. Improved mass spectrometric proteomic profiling of the secretome of rat vascular endothelial cells. J Proteome Res 2006;5:2861–2864.[CrossRef][Medline]

  69. Khwaja FW, Svoboda P, Reed M et al. Proteomic identification of the wt-p53-regulated tumor cell secretome. Oncogene 2006;25:7650–7661.[CrossRef][Medline]

  70. Jeffery CJ. Moonlighting proteins: Old proteins learning new tricks. Trends Genet 2003;19:415–417.[CrossRef][Medline]

  71. Jeffery CJ. Mass spectrometry and the search for moonlighting proteins. Mass Spectrom Rev 2005;24:772–782.[CrossRef][Medline]

  72. Carreño FR, Goñi CN, Castro LM et al. 14-3-3 epsilon modulates the stimulated secretion of endopeptidase 24.15. J Neurochem 2005;93:10–25.[CrossRef][Medline]

  73. Seelenmeyer C, Wegehingel S, Tews I et al. Cell surface counter receptors are essential components of the unconventional export machinery of galectin-1. J Cell Biol 2005;171:373–381.[Abstract/Free Full Text]

  74. Fujimoto N, Akimoto Y, Suzuki T et al. Identification of prostatic-secreted proteins in mice by mass spectrometric analysis and evaluation of lobe-specific and androgen-dependent mRNA expression. J Endocrinol 2006;190:793–803.[Abstract/Free Full Text]

  75. Okazaki K, Niki I, Iino S et al. A role of calcyclin, a Ca2+-binding protein, on the Ca2+-dependent insulin release from the pancreatic beta cell. J Biol Chem 1994;269:6149–6152.[Abstract/Free Full Text]

  76. Temesvari L, Zhang L, Fodera B et al. Inactivation of lmpA, encoding a LIMPII-related endosomal protein, suppresses the internalization and endosomal trafficking defects in profilin-null mutants. Mol Biol Cell 2000;11:2019–2031.[Abstract/Free Full Text]

  77. Ravasz E, Somera AL, Mongru DA et al. Hierarchical organization of modularity in metabolic networks. Science 2002;297:1551–1555.[Abstract/Free Full Text]

  78. Alexander SM, Jackson KJ, Bushnell KM et al. Spatial and temporal expression of the 72-kDa type IV collagenase (MMP-2) correlates with development and differentiation of valves in the embryonic avian heart. Dev Dyn 1997;209:261–268.[CrossRef][Medline]

  79. Brauer PR, Cai DH. Expression of tissue inhibitor of metalloproteinases (TIMPs) during early cardiac development. Mech Dev 2002;113:175–179.[CrossRef][Medline]

  80. Linask KK, Han M, Cai DH et al. Cardiac morphogenesis: Matrix metalloproteinase coordination of cellular mechanisms underlying heart tube formation and directionality of looping. Dev Dyn 2005;233:739–753.[CrossRef][Medline]

  81. Stary M, Pasteiner W, Summer A et al. Parietal endoderm secreted SPARC promotes early cardiomyogenesis in vitro. Exp Cell Res 2005;310:331–343.[CrossRef][Medline]

  82. Mummery CL, Ward D, van den Brink CE et al. Cardiomyocyte differentiation of mouse and human embryonic stem cells. J Anat 2002;200:233–242.[CrossRef][Medline]

  83. Kang YS, Park YG, Kim BK et al. Angiotensin II stimulates the synthesis of vascular endothelial growth factor through the p38 mitogen activated protein kinase pathway in cultured mouse podocytes. J Mol Endocrinol 2006;36:377–388.[Abstract/Free Full Text]

  84. Frossi B, Rivera J, Hirsch E et al. Selective activation of Fyn/PI3K and p38 MAPK regulates IL-4 production in BMMC under nontoxic stress condition. J Immunol 2007;178:2549–2555.[Abstract/Free Full Text]

  85. Butland G, Peregrin-Alvarez JM, Li J et al. Interaction network containing conserved and essential protein complexes in Escherichia coli. Nature 2005;433:531–537.[CrossRef][Medline]

  86. Schadt EE, Lum PY. Systems biology approaches to metabolic and cardiovascular disorders. Reverse engineering gene networks to identify key drivers of complex disease phenotypes. J Lipid Res 2006;47:2601–2613.[Abstract/Free Full Text]

  87. Arrell DK, Niederländer NJ, Perez-Terzic C et al. Embryonic stem cell cardiac differentiation: A proteomic perspective. Adv Mol Med 2006;2:149–156.

  88. Arrell DK, Niederländer NJ, Perez-Terzic C et al. Pharmacoproteomics: Advancing the efficacy and safety of regenerative therapeutics. Clin Pharmacol Ther 2007;82:316–319.[CrossRef][Medline]




This article has been cited by other articles:


Home page
 Stem CellsHome page
S. Yamada, T. J. Nelson, R. J. Crespo-Diaz, C. Perez-Terzic, X.-K. Liu, T. Miki, S. Seino, A. Behfar, and A. Terzic
Embryonic Stem Cell Therapy of Heart Failure in Genetic Cardiomyopathy
Stem Cells, October 1, 2008; 26(10): 2644 - 2653.
[Abstract] [Full Text] [PDF]

Home page
 Circ. Res.Home page
M. S. Penn and A. A. Mangi
Genetic Enhancement of Stem Cell Engraftment, Survival, and Efficacy
Circ. Res., June 20, 2008; 102(12): 1471 - 1482.
[Abstract] [Full Text] [PDF]

Home page
 Stem CellsHome page
T. J. Nelson, R. S. Faustino, A. Chiriac, R. Crespo-Diaz, A. Behfar, and A. Terzic
CXCR4+/FLK-1+ Biomarkers Select a Cardiopoietic Lineage from Embryonic Stem Cells
Stem Cells, June 1, 2008; 26(6): 1464 - 1473.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental Data
Right arrow All Versions of this Article:
2007-0599v1
2007-0599v2
26/2/387    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow Reprints/Permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Arrell, D. K.
Right arrow Articles by Terzic, A.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Arrell, D. K.
Right arrow Articles by Terzic, A.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
STEM CELLS THE ONCOLOGIST CME ALPHAMED PRESS JOURNALS