SEARCH

SEARCH BY CITATION

Keywords:

  • Histone acetyltransferases;
  • Trrap;
  • Embryonic stem cells;
  • Self-renewal;
  • Heterochromatization

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Chromatin states are believed to play a key role in distinct patterns of gene expression essential for self-renewal and pluripotency of embryonic stem cells (ESCs); however, the genes governing the establishment and propagation of the chromatin signature characteristic of pluripotent cells are poorly understood. Here, we show that conditional deletion of the histone acetyltransferase cofactor Trrap in mouse ESCs triggers unscheduled differentiation associated with loss of histone acetylation, condensation of chromatin into distinct foci (heterochromatization), and uncoupling of H3K4 dimethylation and H3K27 trimethylation. Trrap loss results in downregulation of stemness master genes Nanog, Oct4, and Sox2 and marked upregulation of specific differentiation markers from the three germ layers. Chromatin immunoprecipitation-sequencing analysis of genome-wide binding revealed a significant overlap between Oct4 and Trrap binding in ESCs but not in differentiated mouse embryonic fibroblasts, further supporting a functional interaction between Trrap and Oct4 in the maintenance of stemness. Remarkably, failure to downregulate Trrap prevents differentiation of ESCs, suggesting that downregulation of Trrap may be a critical step guiding transcriptional reprogramming and differentiation of ESCs. These findings establish Trrap as a critical part of the mechanism that restricts differentiation and promotes the maintenance of key features of ESCs. STEM CELLS 2013;31:979–991


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

The histone acetyltransferase (HAT) cofactor transactivation/transformation domain-associated protein (Trrap) has been identified as a common component of several HAT complexes, such as GCN5/PCAF-containing SAGA and Tip60/NuA4, which have been shown to mediate chromatin-based processes, including transcription and DNA repair [1–4]. Several transcription factors have been shown to interact with Trrap/HAT complexes and are potential regulators or mediators of Trrap/HAT chromatin-modifying activities [5–9]. We have previously shown that Trrap plays a critical role in the maintenance of hematopoietic stem cells and that loss of Trrap induces apoptosis through disruption of c-myc- and N-myc-dependent transcription [10]. Furthermore, a recent RNAi screen in mouse embryonic stem cells (ESCs) revealed that Tip60-p400 HAT and TRRAP may be involved in the maintenance of stem cell identity [11]. Consistent with these findings, TRRAP was found to play an important role in maintaining a tumorigenic, stem cell-like state [12], although the role of Trrap/HATs in establishing the distinct chromatin signature required for the unique stem cell properties remains largely unknown.

Deletion of Trrap by gene targeting revealed its essential function in early embryonic development and stem cell clonogenicity [13]. However, studying the precise mechanism by which Trrap and Trrap-containing HAT complexes regulate ESCs has been hampered by the incompatibility of Trrap null mutation with ESC viability [10, 13]. To circumvent this problem, in this study, we generated conditional-knockout (CKO) mouse ESCs that allow inducible inactivation of the Trrap gene after expression of Cre recombinase. Using these cells, we investigated the role of Trrap in the maintenance of key properties of ESCs and identified the mechanism by which Trrap and Trrap-mediated histone acetylation govern the maintenance and propagation of chromatin states and distinct patterns of gene expression essential for self-renewal of pluripotent cells.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

ESC Culture and Trrap Deletion

ESCs that are amenable to inducible deletion of Trrap by Cre recombinase have been described previously [13]. To delete the Trrap gene, 107 ESCs were electroporated at 0.26 kV and 0.5 (×1,000) μF with 20 μg of pMC-Cre and plated on gelatin-coated plates. At 24 hours after electroporation, ganciclovir (Cymevan 500 mg; Roche, France, http://www.roche.com) was added to the culture medium at 2 × 10−6 M for 48 hours for cell selection. To induce ESC differentiation, cells were cultured on 0.1% gelatin-coated plates in medium without leukemia inhibitory factor (LIF) for 4 days. ESCs cultured in differentiating conditions for 2 days were transfected with Flag-tagged TRRAP-expressing vector (Flag-TRRAP) using Lipofectamine 2000 according to the manufacturer's instructions (Invitrogen, France, http://www.invitrogen.com) and then cultured for a further 2 days in differentiating conditions. For the embryoid body formation assay, 2 × 106 ESCs were seeded on nonadherent bacterial dishes in ES medium without LIF and cultured for 10 days. Cells were maintained in suspension and the medium was changed every 2 days. Colony formation assay was carried out as described previously [13].

Alkaline Phosphatase Staining

First, 4 × 105 of the cells that had been electroporated with pMC-Cre and treated with ganciclovir, electroporated with empty vector, or cultured in differentiating conditions were grown on gelatin-coated plates for 72 hours. Cells were then stained for alkaline phosphatase (AP) according to the manufacturer's instructions (AP detection kit, Chemicon International, France, http://www.millipore.com).

Flow Cytometry

To analyze cell cycle profiles, ESCs were harvested by trypsinization, stained with propidium iodide using the Cycletest Plus kit (Becton Dickinson, France, http://www.bd.com), and analyzed with the FACSCalibur system (Becton Dickinson, France, http://www.bd.com). For apoptosis analysis, cells were stained with annexin V and propidium iodide using the Pro-Alert annexin V-FITC apoptosis kit (Clontech, France, http://www.clontech.com) and analyzed with the FACSCalibur system (Becton Dickinson, France, http://www.bd.com).

PCR Analyses

To check the efficiency of Trrap deletion, ESCs were electroporated with pMC-Cre and the DNA extracted was subjected to PCR amplification. Primers used were I2.1 (5′-AGAGTAGAGCGTCATTCGTC-3′) and G7.0 (5′-GACAAAACCAACGACAGAGC-3′).

RNA Extraction, Reverse Transcription, and Quantitative PCR

Total RNA was extracted and purified using the RNeasy mini kit (Qiagen), and reverse transcription was performed according to the manufacturer's instructions (Invitrogen, France, http://www.invitrogen.com). Real-time PCR analysis was performed using Universal ProbeLibrary probes and the FastStart TaqMan Probe Master (Roche, France, http://www.roche.com) and the Stratagene Mx3000P sequence detection system (Agilent Technologies, France, http://www.genomics.agilent.com). The sequences of primers used are listed in supporting information Table S1.

Protein Extraction and Western Blot Analysis

The following antibodies were used: rabbit anti-Trrap (1:500; Santa Cruz, France, http://www.scbt.com), rabbit anti-Oct4 (1:500; Abcam, France, http://www.abcam.com), rabbit anti-Nanog (1:500; Abcamhttp://www.abcam.com), rabbit anti-Sox2 (1:500; Abcamhttp://www.abcam.com), and mouse anti-actin (1:10,000; Santa Cruz). Proteins were visualized with horseradish peroxidase-conjugated anti-rabbit antibody (1:5,000; DAKO, France, http://www.dako.com) and HPRT-conjugated anti-mouse IgG (1:2,000; DAKO, France, http://www.dako.com), followed by use of the ECL chemiluminescence system (Amersham, GE Healthcare Life Sciences, France, http://www.gelifesciences.com).

Chromatin Fractionation

Chromatin fractionation assays were performed as described previously [14]. The following antibodies were used: rabbit anti-histone H3 dimethyl K4 (1:2,000; Abcam, France, http://www.abcam.com), rabbit anti-histone H3 trimethyl K27 (1:3,000; Abcam, France, http://www.abcam.com), rabbit anti-histone H3 acetyl K9 (1:1,000; Upstate), mouse anti-histone H3 trimethyl K9 (1:1,000; Abcam, France, http://www.abcam.com), and rabbit anti-histone H3 (1:1,000; Abcam, France, http://www.abcam.com).

Coimmunoprecipitation

Nonidet P40 lysis buffer containing 150 mM NaCl, 50 mM Tris, and 1% of Nonidet P40 buffer was used for cell lysate and protein extraction. Cell lysates were centrifuged at 14,000 rpm for 15 minutes, and the pellet was suspended in Nonidet P40 lysis buffer. Next, 1 mg of protein was isolated and the volume adjusted to 1 ml for immunoprecipitation (IP) with goat anti-Trrap antibody (Santa Cruz, France, http://www.scbt.com) or without antibody (control). Proteins were precleared with 20 μl of protein A/G agarose (Santa Cruz, France, http://www.scbt.com) for 30 minutes at 4°C on a rotor. Next, 5 μg of goat anti-Trrap antibody was added and incubated for 3 hours at 4°C on a rotor, and 50 μl of protein G Sepharose (Amersham, GE Healthcare Life Sciences, France, http://www.gelifesciences.com) was added and incubated overnight at 4°C on a rotor. Beads were washed three times with Nonidet P40 lysis buffer and suspended in 30 μl of 1× IP buffer boiled for 5 minutes, and supernatants were used in Western blot analysis.

Immunofluorescence Detection

Immunofluorescence staining was carried out as described previously [13]. ESCs were cultured on gelatin-coated coverslips and stained with rabbit anti-acetyl-histone H4 antibody (1:50; Upstate, Millipore, France, http://www.millipore.com) and Cy3-conjugated anti-rabbit IgG (1:200; Sigma, France, http://www.sigmaaldrich.com). Slides were mounted in medium containing 4′,6-Diamidino-2-Phenylindole, Dihydrochloride (DAPI) and analyzed under a fluorescence microscope.

Chromatin Immunoprecipitation Assays

Cells were cross-linked with formaldehyde, and chromatin immunoprecipitation (ChIP) assays were performed as described previously [4], using polyclonal antibodies specific for acetylated histone H4 (Upstate, #06-866, Millipore, France, http://www.millipore.com), Trrap (Santa Cruz, France, http://www.scbt.com), and Oct4, histone H3 trimethyl K27, and histone H3 dimethyl K4 (Abcam, France, http://www.abcam.com). The following ChIP primers were used to analyze protein binding on promoters: Nanog1Fw (5′ACTTCCCACTAGAGATCGCC3′), Nanog1Rev (5′TGTAAGGTGACCCAGACTGG3′), Oct4(1)Fw (5′TAGACGGGTGGGTAAGCAAG3′), Oct4(1)Rev (5′CACAAAGCCTGTTGGCACTG3′), Sox2(2)Fw (5′GAGGATGAAGCACCCTGTTC3′), Sox2(2)Rev (5′GATCACTGGCCACCGATTTC3′), Gata6(1)Fw (GACAGGGCTAGCAGCTCAAC), Gata6(1)Rev (TCACCGAGCCCTAAACAAAC), Gata6(2)Fw (TATTGTCCGCTAGGGCTGAG), Gata6(2)Rev (CTCACCGAGCCCTAAACAAA), Pax3(2)Fw (TCTGCTAGACTCGCACCAAA), and Pax3(2)Rev (GTGAAGGCGAGACGAAAAAG).

ChIP-Seq

Cells were cross-linked with 1% formaldehyde for 10 minutes at room temperature, and ChIP assays were carried out with 500 mg of sonicated chromatin and 5 mg of antibody against Trrap (provided by L. Tora, IGBMC, Illkirch, France), Oct4 (Abcam, #ab18976, France, http://www.abcam.com), and histone AcH4 (Millipore, #06-598, France, http://www.millipore.com). Eluates were reverse cross-linked, and the immunoprecipitated DNA was used for ChIP library preparation according to the manufacturer's protocol (Illumina, France, http://www.illumina.com) followed by sequencing with the Illumina GAII machine following the manufacturer's standard protocol [15]. After sequencing and cluster imaging, quality control and file format conversion were done using FastQC and FastQGroomer, respectively [16], from the Galaxy web-based platform [17]. After results were collapsed to remove duplicate reads, base calling and mapping were done with the short-read aligner Bowtie 1.1.2, allowing a maximum of two mismatches in the seed, and using Mus musculus (mm9) as a built-in index [18]. Model-based analysis of ChIP-seq data (MACS 1.0.1) was used for peak calling of aligned reads, using input data as the control file [19]. Galaxy interval operations (intersect, subtract, and concatenate) were used to calculate the overlap among different conditions. For association of binding regions with potential regulated genes, each ChIP-seq region was assigned to its nearest gene annotated in the refGene table of the RefSeq Genes track, downloaded from the UCSC Table Browser [20].

DNA Fluorescence In Situ Hybridization Analysis

ESCs grown on gelatin-coated glass coverslips were fixed (4% paraformaldehyde in phosphate buffered saline (PBS) for 15 minutes), washed three times (with PBS, 5 minutes each), and permeabilized (0.5% Triton X-100 in PBS for 5 minutes). A 5′-biotinylated DNA probe (Invitrogen, France, http://www.invitrogen.com) recognizing the mouse major satellite repeat (MSR) (GenBank accession number X06899) was applied overnight in a hybridization solution (50% formamide, 2× SSC, 10% dextran sulfate, and 1 mg/ml tRNA) at 37°C. Probe sequences were the following: MSR: 5′-CTCGCCATATTTCACGTCCTAAAGTGTG ATTTCTC-3′; scrambled: 5′-TCTACGTTACCATCTCAGTGCGTATCGTTCTATTCA-3′.

Fluorescence Recovery After Photobleaching Analysis

H2B-GFP plasmid (pBOS-H2B-GFP; BD Pharmingen, France, http://www.bdbiosciences.com) was used for electroporation of ESCs, and fluorescence recovery after photobleaching (FRAP) analysis was performed 3 days later. A confocal microscope (Carl Zeiss, France, http://microscopy.zeiss.com) with a 30 mW argon/neon laser at 65% output and 10–70 iterations was used for the analysis. Half of the nucleus was bleached, including euchromatin and heterochromatin, and images were collected every 5 seconds for 5 minutes. Image analysis was performed with LSM software (Carl Zeiss, France, http://microscopy.zeiss.com).

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Trrap Deletion in Conditional-Knockout ESCs Is Compatible with Cell Viability and Fails to Trigger Apoptosis

We have previously shown that homozygous mutation of Trrap is incompatible with ESC growth [13] and that inducible deletion of Trrap in bone marrow in vivo induces apoptosis of hematopoietic stem cells [10]. To examine whether Trrap is also required for the maintenance of ESCs and to study a mechanism by which Trrap may regulate stem cell features, we used a Trrap conditional-knockout mouse ESC line (Trrapfloxed/targeted; Trrapf/t) in which the Trrap targeted and floxed allele can be deleted by transient expression of Cre recombinase (Fig. 1A). The efficiency of Trrap deletion mediated by electroporation of Trrapf/t ESCs with Cre-containing plasmid (pMC-Cre) was verified by PCR and Western blot analyses, which revealed that virtually all Trrap floxed alleles and more than 90% of Trrap protein disappeared within 48 hours after transfection (Fig. 1B), indicating an efficient depletion of Trrap in ESCs (CKO). Hereafter, Trrapf/t ESCs electroporated with pMC-Cre are designated as Trrap-CKO cells. Trrapf/t ESCs electroporated with empty vector (Co) or grown in differentiating conditions without LIF (Co(−LIF)) were used as controls.

thumbnail image

Figure 1. Trrap deletion in conditional-knockout embryonic stem cells (ESCs) is compatible with cell viability and fails to trigger apoptosis. (A): Strategy used for Trrap deletion in Trrap conditional-knockout ESCs. Trrapf/t ESCs were electroporated with pMC-Cre (CKO) or empty vector (Co). To ensure that all cells that were nontransfected or failed to undergo Cre-mediated recombination were eliminated from the culture, at 24 hours after electroporation, ganciclovir was added to the culture medium of pMC-Cre-electroporated cells for 48 hours. Partial structure of the Trrap targeted and floxed alleles is shown. LoxP sites are represented by empty triangles, and exons are indicated by boxes (the exon flanked by LoxP sites is indicated by a black box). (B): DNA and proteins were extracted from Co and CKO ESCs and subjected to PCR amplification and Western blotting, respectively. DNA subjected to PCR revealed Trrap floxed (F), targeted (T), and deleted (Δ) alleles. Asterisk (*) denotes a nonspecific cross-reacting antigen used as a loading control for Western blot analysis. (C): Percentage of colonies in Co and CKO ESC cultures as well as in ESCs cultured under differentiating conditions (without LIF) for 4 days (Co(−LIF)). Data presented are the mean percentage of colonies (±SD) and are representative of triplicate experiments. (D): Fluorescence-activated cell sorting (FACS) analysis of apoptosis in Trrap-CKO ESCs. Trrapf/t ESCs were electroporated with pMC-Cre (CKO) or empty vector (Co) as in (A), and the annexin V-positive fraction was analyzed by flow cytometry. Percentages of annexin V-positive (apoptotic) cells are shown. (E): Quantification of annexin V-positive cells as in (D). Data presented are means (±SD) and are representative of triplicate experiments. (F): Cell cycle profile of Trrap-deficient ESCs. Trrap-CKO and control (Co) ESCs were stained with propidium iodide and analyzed by FACS. Abbreviations: Co, control; CKO, conditional-knockout; LIF, leukemia inhibitory factor.

Download figure to PowerPoint

Whereas colony formation was not affected in Trrap-containing cells (Trrapf/t cells transfected with empty vector; Co), we found that Trrap-CKO cells failed to form stem cell-like colonies (Fig. 1C, and data not shown). These results suggest that Trrap-deficient cells completely lose the capacity to form typical stem cell-like colonies, a phenotype similar to that of ESCs grown without LIF (Co(−LIF)) (Fig. 1C). To examine whether the lack of colonies in Trrap-deficient cells is caused by induction of cell death, we analyzed the apoptotic fraction of Trrap-containing and Trrap-deficient ESCs by flow cytometry after staining with FITC-conjugated annexin V. The apoptotic fractions of Trrap-deficient and Trrap-containing cells were not significantly different (Fig. 1D, 1E). In addition, the cell cycle profile of Trrap-deficient cells was similar to that of Trrap-containing controls, although a moderate decrease in S-phase fraction was observed in Trrap-deficient cells (Fig. 1F). These results indicate that increased apoptosis and cell cycle block are unlikely to be the cause of the loss of colony-forming potential in Trrap-deficient ESCs.

Loss of Trrap Triggers Unscheduled Differentiation of ESCs

To examine whether Trrap is required for the maintenance of stem cell-like properties and to elucidate a possible mechanism by which Trrap may regulate the self-renewal and differentiation fates of ESCs, we analyzed morphological changes and AP activity, a marker of undifferentiated ESCs that is commonly used to study the self-renewal capacity of ESCs [21], in Trrap-deficient and Trrap-containing ESCs. Remarkably, upon depletion of Trrap, the colony morphology of ESCs was lost and cells exhibited a flattened and elongated morphology, had reduced cell-to-cell contact, and grew in monolayers (Fig. 2A), thus showing a growth pattern similar to that of ESCs grown without LIF (Fig. 2A, 2B). This sharply contrasts with the normal ESC growth pattern, in which small, spherical cells grow in dense, well-delimited three-dimensional colonies (Fig. 2A). Moreover, AP staining revealed a dramatic loss of staining in Trrap-deficient cells, indicative of differentiation, an observation similar to that for the cells grown without LIF (Fig. 2C, 2D). These results show that depletion of Trrap leads to unscheduled differentiation of ESCs, consistent with the essential role of Trrap in the maintenance of stem cell properties and prevention of differentiation.

thumbnail image

Figure 2. Loss of Trrap triggers unscheduled differentiation of embryonic stem cells (ESCs). (A): Colony morphology of Trrap-deficient ESCs. Trrapf/t ESCs were electroporated with pMC-Cre (CKO) or empty vector (Co) or incubated without LIF (Co(−LIF)) and photographed with bright-field and dark-field filters at ×40 magnification. Representative live images of different ESC cultures are shown. (B): Percentage of colonies with different morphologies (undifferentiated, mixed, and differentiated) with phase-contrast microscopy. Data presented are mean percentage of colonies with different morphologies (±SD) and are representative of triplicate experiments. (C):Trrapf/t ESCs electroporated with pMC-Cre (CKO) or empty vector (Co) or incubated without LIF (Co(−LIF)) were fixed and stained for alkaline phosphatase (AP) activity (purple staining, magnification ×40). (D): Quantification of AP-positive colonies as in (C). Data presented are means (±SD) and are representative of triplicate experiments. Abbreviations: Co, control; CKO, conditional-knockout; LIF, leukemia inhibitory factor.

Download figure to PowerPoint

Trrap Deletion Induces Heterochromatization and Impairs Chromatin Dynamics of ESCs

As a component of HAT complexes, Trrap participates in remodeling chromatin structure and function. Therefore, we next examined the global chromatin organization and dynamics of architectural proteins in ESCs after Trrap depletion. For this, MSR (a marker of chromocenters) signals were revealed by fluorescence in situ hybridization (FISH) in Trrap-containing and Trrap-deficient ESCs. In Trrap-containing cells, MSR signals appeared as large, poorly defined regions, characteristic of chromatin of undifferentiated ESCs (Fig. 3A). In contrast, in Trrap-deficient ESCs, MSR signals were confined to numerous discrete foci with well-defined borders, a finding consistent with the global chromatin reorganization observed in differentiated cells induced by LIF withdrawal (Fig. 3A). These observations suggest that Trrap may be involved in the chromatin-based mechanism that controls self-renewal and pluripotency of ESCs.

thumbnail image

Figure 3. Trrap deletion induces heterochromatization and impairs chromatin dynamics of embryonic stem cells (ESCs). (A): DNA FISH analysis of mouse MSRs (red signal, left panels) in Trrap-CKO and Co ESCs and differentiating ESCs (grown without LIF). As a control, a scrambled 5′-biotinylated FISH probe of similar length was used and yielded no detectable signal (data not shown). Cells were counterstained with DAPI (DNA, blue signal). (B): Loss of Trrap impairs chromatin dynamics of ESCs. Trrapf/t ESCs transfected with empty vector (Co) or pMC-Cre (CKO) or incubated without LIF (Co(−LIF)) for 24 hours were transfected with H2B-GFP plasmid and subjected to fluorescence recovery after photobleaching (FRAP) analysis 48 hours later. Diagrams of representative recovery of fluorescence are shown. (C): Mean (±SD) of the FRAP of at least five cells per condition was used to infer a fitted curve. Percentage of initial fluorescence signal is depicted on the y-axis and time in seconds on the x-axis. The three curves shown are significantly different (p < .05). Abbreviations: Co, control; CKO, conditional-knockout; DAPI, 4′,6-Diamidino-2-Phenylindole, Dihydrochloride,; LIF, leukemia inhibitory factor; ROI, region of interest.

Download figure to PowerPoint

The chromatin of ESCs undergoes substantial spatial rearrangements during the early stages of stem cell differentiation [22]. Therefore, we next assessed whether the observed changes in global cell morphology and chromatin organization in Trrap-deficient ESCs were associated with hyperdynamic chromatin properties, a hallmark of pluripotent ESCs. For this, we used a FRAP assay of the core histone H2B in Trrap-deficient and Trrap-containing ESCs. Half of the nucleus was bleached and recovery of fluorescence was registered for at least 5 minutes. Whereas the recovery signal reached 50% of the initial intensity after 250 seconds in Trrap-containing cells, it reached only 10% in Trrap-deficient cells, a level similar to that in differentiated cells induced by LIF withdrawal (Fig. 3B, 3C). These results indicate that the effects on chromatin dynamics induced by Trrap depletion are likely to precede differentiation and thus cannot be explained as merely the induction of differentiation. Together, these results show that Trrap plays an essential role in the maintenance of a characteristic chromatin signature (open hyperdynamic chromatin state) that may govern the propagation of distinct gene expression patterns and the identity of ESCs.

Trrap Regulates Expression and Function of the Stemness Master Genes

Because the stemness-related genes Nanog, Oct4, and Sox2 are major players involved in the maintenance of self-renewal and pluripotency of ESCs [23], and because some of these genes associate with chromatin-modifying complexes in ESCs [24], we next analyzed expression of these stemness master genes in Trrap-deficient and Trrap-containing ESCs. We found that the mRNA and protein levels of Nanog, Oct4, and Sox2 were strikingly lower in Trrap depleted cells than those in Trrap-containing cells (Fig. 4A, 4B). Levels of all three proteins decreased in Trrap-deficient cells and Trrap-containing cells cultured without LIF (Fig. 4B). Moreover, transcript levels of Tcf3, a gene known to repress Nanog and Oct4 expression [25, 26], were higher in Trrap-deficient ESCs than in Trrap-containing control cells, a finding consistent with increased expression of Tcf3 in ESCs grown without LIF (Fig. 4C). These results demonstrate that Trrap deletion results in a dramatic downregulation of the stemness master genes Nanog, Oct4, and Sox2. Therefore, unscheduled differentiation of Trrap-deficient ESCs may be a consequence of downregulation of stemness master genes in the absence of Trrap.

thumbnail image

Figure 4. Loss of Trrap induces changes in expression of stemness master genes and lineage-specific genes. (A, B):Trrapf/t embryonic stem cells (ESCs) were mock-transfected (Co), transfected with pMC-Cre (CKO), or incubated without LIF (Co(−LIF)) and 3 days later, expression of stemness genes was analyzed by qRT-PCR (A) and Western blotting (B). Gapdh was used as a loading control for qRT-PCR assays. Equal loading in Western blot analysis was verified by anti-actin antibodies. (C):Trrapf/t ESCs were transfected as in (A) and expression of Tcf3, a Nanog target gene, was analyzed by qRT-PCR. (D):Trrapf/t ESCs were transfected as in (A) and expression of differentiation genes was analyzed by qRT-PCR. Gapdh was used as a loading control. Bars represent means (±SD) for triplicate samples and are representative of three independent experiments. Abbreviations: Co, control; CKO, conditional-knockout; LIF, leukemia inhibitory factor.

Download figure to PowerPoint

To assess whether the effects of changes in morphology are associated with expression of developmental and differentiation genes, we next analyzed expression of differentiation marker genes. Our analysis revealed that specific differentiation markers from the three germ layers (ectoderm, mesoderm, and endoderm; Gata4, Gata6, Tbx5, and Pax3) and from trophectoderm (Eomes and Cdx2) were strongly upregulated in Trrap-deficient cells compared with Trrap-containing cells, whereas the expression of Isl1 was unchanged (Fig. 4D). These results show that Trrap deletion in ESCs leads to overexpression of differentiation genes, suggesting that Trrap may maintain pluripotency by mediating expression of pluripotency master genes and by repressing differentiation genes.

Loss of Trrap Results in Abrogation of Histone Acetylation and Cross-Influences Repressive Histone Marks

Because Trrap-containing HAT complexes play key roles in chromatin states, we next investigated whether unscheduled differentiation of Trrap-deficient ESCs are linked to changes in chromatin modifications. To this end, we analyzed levels of active (AcH4, 2meH3K4, and AcH3K9) and repressive (3meH3K27 and 3meH3K9) histone marks in Trrap-containing and Trrap-deficient ESCs as well as in ESCs grown without LIF, by immunofluorescence and Western blotting (using the chromatin-bound fraction obtained by chromatin fractionation). Loss of Trrap resulted in a dramatic reduction in histone H4 and histone H3K9 acetylation in ESCs, consistent with a critical role of Trrap in the integrity and activity of HAT complexes (Fig. 5A, 5B). In sharp contrast, Trrap depletion resulted in a striking increase in H3K27 trimethylation (3meH3K27, Fig. 5A), the histone mark associated with the repressive state, suggesting that loss of Trrap and histone acetylation may trigger Polycomb-mediated H3K27 trimethylation. Interestingly, Trrap depletion was also accompanied by an appreciable decrease in H3K4 dimethylation (2meH3K4) (Fig. 5A, and data not shown), the histone mark associated with the active chromatin state, suggesting a role for Trrap and/or histone acetylation in the activity of H3K4 methyltransferase. Therefore, local changes in histone modifications associated with the promoters of pluripotency genes are consistent with global changes in histone modifications in Trrap-deficient ESCs.

thumbnail image

Figure 5. Loss of Trrap induces changes in global chromatin organization by global and specific epigenetic changes. (A): Global changes in histone modifications in embryonic stem cells (ESCs) lacking Trrap. Trrapf/t ESCs were transfected with empty vector (Co), transfected with pMC-Cre (CKO), or incubated without LIF (Co(−LIF)), and 3 days later, levels of indicated histone modifications were analyzed by Western blotting using the corresponding antibodies. Equal loading was verified by anti-histone H3 antibodies. (B): Analysis of acetylated histone in Trrap-deficient ESCs. Trrapf/t ESCs were transfected with empty vector (Co), transfected with pMC-Cre (CKO), or grown without LIF (Co(−LIF)), and 3 days later, cells were stained for acetyl-histone H4 levels and counterstained with DAPI. (C): Occupancy of Trrap at the promoters of stemness master genes in ESCs. Left panel: Chromatin immunoprecipitation (ChIP) assay of Trrap and Oct4 occupancy at the Nanog, Sox2, and Oct4 promoters was performed using the chromatin from Trrap-containing (Trrapf/t) ESCs immunoprecipitated with anti-Trrap and anti-Oct4 antibodies. Right panel: ChIP assay of Trrap occupancy at the Gata6 and Pax2 promoters was performed using the chromatin from Trrap-containing ESCs immunoprecipitated with anti-Trrap antibodies. Duplicate samples (S1 and S2) were used for the ChIP assay. Two independent primer pairs (#1 and #2) were used to examine Trrap binding at the Gata6 promoter. (D): Co-IP of endogenous Trrap and Oct4 proteins in ESCs. Total cell extracts (1 mg) from Trrapf/t ESCs were immunoprecipitated with or without anti-TRRAP antibody and probed with antibodies against Oct4, Trrap, and c-Myc. For input, 50 μg of proteins was used. (E–G): ChIP analysis of histone modifications at the promoters of stemness master genes in Trrap-deficient ESCs. Trrapf/t ESCs were transfected with empty vector (Co) or pMC-Cre (CKO), and at 72 hours after transfection and 48 hours after ganciclovir treatment, the chromatin immunoprecipitated by anti-acetyl-histone H4, anti-2meH3K4, and anti-3meH3K27 was analyzed by PCR with primers specific for the promoters of Nanog (E), Oct4 (F), and Sox2 (G). Quantification of levels of histone marks was performed by densitometric analysis and normalized to input. Abbreviations: Co, control; CKO, conditional-knockout; DAPI, 4′,6-Diamidino-2-Phenylindole, Dihydrochloride; LIF, leukemia inhibitory factor.

Download figure to PowerPoint

To elucidate the precise mechanism by which Trrap regulates the expression of stemness genes and to determine whether global changes in histone modifications in Trrap-deficient ESCs are also present at the promoters of specific genes, we performed ChIP analysis of histone modifications at the Nanog, Oct4, and Sox2 promoters using specific antibodies against Trrap and the AcH4, 2meH3K4, and 3meH3K27 histone marks. ChIP analysis revealed that in ESCs, Trrap is associated with the Nanog, Oct4, and Sox2 promoters (Fig. 5C, left panel), but not with the Gata6 and Pax3 promoters (Fig. 5C, right panel). The association of Trrap with the promoters of stemness-related genes was correlated with the presence of Oct4, which is known to bind to the Nanog and Sox2 promoters as well as to its own promoter [27] (Fig. 5C). Consistent with this finding, co-IP experiments revealed that in ESCs Trrap interacts with Oct4 and c-Myc, a Trrap-interacting transcription factor (Fig. 5D). Furthermore, ChIP analysis also revealed that AcH4 levels were significantly lower at the Nanog, Oct4, and Sox2 promoters in Trrap-deficient ESCs than in Trrap-containing cells, whereas 2meH3K4 levels were only moderately lower at the Nanog and Oct4 promoters (Fig. 5E–5G). In contrast, 3meH3K27 levels were markedly higher at all three promoters (Fig. 5E–5G). Therefore, Trrap is essential for the maintenance of active versus repressive methyl marks at the promoters of the stemness master genes in ESCs.

Failure to Downregulate Trrap Prevents Differentiation of ESCs

Because striking changes in histone modifications of ESCs lacking Trrap are associated with unscheduled differentiation, we next tested whether Trrap protein levels are regulated during differentiation of ESCs. We triggered the differentiation of ESCs by withdrawing LIF and monitored Trrap protein levels. Both total protein levels and the chromatin-bound fraction of Trrap were markedly lower after differentiation of pluripotent ESCs (Fig. 6A, 6B), suggesting that downregulation of Trrap may be part of a cellular mechanism guiding reconfiguration of chromatin and the transcriptional program during differentiation of pluripotent cells.

thumbnail image

Figure 6. Failure to downregulate Trrap prevents embryonic stem cell (ESC) differentiation triggered by LIF withdrawal. (A): Trrap protein levels are downregulated during differentiation of ESCs. Trrap-containing (Trrapf/t) ESCs were grown with (Co) or without LIF (Co(−LIF)) for 96 hours, and total cell lysates were prepared and Trrap protein was revealed by anti-Trrap antibodies. Asterisk (*) denotes a nonspecific cross-reacting antigen used as a loading control. (B): Binding of Trrap to chromatin during differentiation. Trrap-containing (Trrapf/t) ESCs were grown in self-renewing (+LIF) and differentiating (−LIF) conditions for 4 days, and Trrap protein levels bound to chromatin were analyzed by Western blotting using acid-extracted histones and anti-Trrap antibody. Trrap protein levels were also analyzed in EB induced by growing ESCs under differentiating conditions on low-attachment plates and without LIF for 4 days. Equal loading was verified by Ponceau S staining (lower panel). (C): RT-PCR analysis of Trrap mRNA levels in differentiating ESCs after overexpression of exogenous TRRAP. Trrapf/t ESCs grown with (Co) or without LIF (Co(−LIF)) for 48 hours were mock-electroporated (−TRRAP) or electroporated with TRRAP expression vector (+TRRAP), and 48 hours later, levels of TRRAP were analyzed by RT-PCR. Gapdh was used as a loading control. (D–F): ESCs in which differentiation had been triggered by LIF withdrawal were electroporated with TRRAP expression vector as in (A) and stem cell features were evaluated by morphology of colonies and AP staining (D, E) and global chromatin organization (F). Percentage of ES-like colonies and AP-positive colonies was determined by scoring at least 100 colonies in each culture (E). Abbreviations: Co, control; AP, alkaline phosphatase; DAPI, 4′,6-Diamidino-2-Phenylindole, Dihydrochloride; EB, embryoid body; LIF, leukemia inhibitory factor.

Download figure to PowerPoint

To further investigate the importance of Trrap downregulation during differentiation and the potential coupling between downregulation of Trrap and differentiation of ESCs, we ectopically expressed Trrap in differentiating ESCs (Fig. 6C) and monitored its effect on induced ESC differentiation. To this end, ESCs grown without LIF were electroporated with human TRRAP expression vector or empty vector (control), and both endogenous and exogenous gene expression were verified by RT-PCR (Fig. 6C). We found that forced expression of TRRAP in ESCs grown without LIF almost completely blocked their differentiation, as judged by the presence of characteristic spherical three-dimensional colonies positive for AP staining in ESC cultures overexpressing TRRAP (Fig. 6D, 6E). Furthermore, FISH staining revealed that these colonies contain cells that exhibit large MSR signals with poorly defined borders (Fig. 6F), indicative of undifferentiated ESCs. These observations show that failure to downregulate Trrap prevents differentiation of ESCs, suggesting that downregulation of Trrap may be important for differentiation of ESCs.

Relocation of Trrap and Oct4 on Selected Promoters During Differentiation of ESCs

To gain insight into the mechanism by which Trrap mediates stem cell features and differentiation, we next analyzed the binding patterns of Trrap by ChIP followed by massively parallel DNA sequencing (ChIP-seq) in ESCs and differentiated cells (mouse embryonic fibroblasts [MEFs]). To further explore a relationship between Trrap and Oct4 occupancy during differentiation, we also performed ChIP-seq (using anti-Oct4 antibodies) to identify genes associated with Trrap and Oct4 binding in ESCs and MEFs. Because histone acetylation is mediated by Trrap-containing HAT complexes, ChIP-seq analysis of histone AcH4 presence was also performed to map the AcH4 profile and to identify Trrap/AcH4-associated genes.

Antibodies against Trrap, Oct4, and pan-acetyl-histone H4 were used in the ChIP-seq experiment. In ESCs, an important proportion of ChIP-seq peaks (1,962 of 4,659; 42%) were Trrap peaks and slightly lower fractions were Oct4 and AcH4 peaks (32% and 26%, respectively) (Fig. 7A). Interestingly, in ESCs, there was a significant overlap between Oct4 and Trrap peaks (506 peaks; 11%), between Oct4 and AcH4 peaks (6%), and between Trrap and AcH4 peaks (8%) (Fig. 7B). In addition, in ESCs 223 peaks were common to Trrap, Oct4, and AcH4 (5%). We found that, compared with ESCs, MEFs had a sharply decreased proportion of Oct4 peaks (from 32% to 4%), a similarly decreased fraction of Trrap peaks (from 42% to 8%), and a concurrently increased fraction of AcH4 peaks (from 26% to 88%) (Fig. 7A). In addition, in MEFs, a relatively small fraction of peaks were common to Trrap, Oct4, and AcH4, or pairs of these (0%–3%). Because a large number of common Trrap/Oct4 loci in ESCs could potentially explain a role for Trrap in sustaining stemness, we analyzed these peaks in more detail. We found that an important proportion of peaks (11%) corresponded to noncoding RNA loci (n = 54), whereas the remaining peaks matched 238 unique coding genes. Importantly, pathway analysis of these 238 genes revealed an enrichment in morphogenesis and developmental pathways. Some examples are Epha7, Tbx4, Smad4, Mycn, Ctnna1, and Foxa2.

thumbnail image

Figure 7. Relationship between genome-wide occupancy of Trrap and Oct4 and histone H4 acetylation in ESCs and MEFs. (A): Distribution of Trrap, Oct4, and AcH4 ChIP-seq peaks in ESCs and MEFs. (B): Venn diagrams showing a significant overlap between Trrap, Oct4, and AcH4 in ESCs and a relatively small fraction of common peaks of Trrap, Oct4, and AcH4 in MEFs. Note a sharp decrease in Oct4 and Trrap peaks and a concurrent increase in AcH4 peaks in MEFs compared with ESCs. Abbreviations: ESCs, embryonic stem cells; MEFs, mouse embryonic fibroblasts.

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

In this study, we examined the role of Trrap, an essential cofactor of chromatin-modifying complexes with HAT activity, whose function has been linked to gene transcription and other chromatin-based processes [28]. We found that Trrap plays a critical role in maintaining ESC identity through the maintenance of histone acetylation states and the chromatin signature that dictates the transcriptional program characteristic of pluripotent cells. Trrap deletion in ESCs resulted in unscheduled differentiation of these cells as judged by morphological, biochemical, and gene expression markers. Our results are consistent with recent studies indicating that the Tip60-p400 HAT complex plays a role in maintaining the stem-like state of ESCs [11] and brain tumor-initiating cells [12] and that the histone deacetylase (HDAC) Mbd, a component of the HDAC NuRD complex, is important in the differentiation of stem cells [29]. We further identified Trrap as part of the mechanism that restricts differentiation and promotes the maintenance of stem cell identity (self-renewal and pluripotency). These findings underscore the importance of a tight regulation of histone acetylation status, which prevents the shift in the balance between self-renewal and differentiation [30].

We found that Trrap deletion results in unscheduled differentiation in the absence of cell death and cell cycle block, perhaps surprisingly considering that the loss of Trrap in adult stem cells (hematopoietic stem cells) triggers cell death [10]. These findings indicate that Trrap participates in the maintenance of embryonic and somatic stem cells through distinct mechanisms. The fact that ESCs lacking Trrap failed to show the severe cell cycle block, characteristic of Trrap-deficient MEFs [13], may be explained by the well-known differences in cell cycle control between these cell types, including the absence of specific checkpoints in ESCs [31]. Therefore, Trrap may be one of the rare molecules that participate in distinct mechanisms involved in maintenance of ESCs and adult stem cells. Morphological changes observed in ESCs upon Trrap depletion were paralleled by a decrease in H4 acetylation and H3K4 dimethylation, which are both associated with transcriptionally active euchromatin, and a dramatic increase in H3K27 trimethylation, an epigenetic marker of silenced heterochromatin. These striking changes in histone modifications reflect a global shift from euchromatin to heterochromatin, consistent with a role of Trrap in the maintenance of a global open conformation, a hallmark of ESCs [22, 32, 33].

Whereas loss of H4 acetylation upon Trrap deletion is consistent with the essential role of Trrap in HAT complexes [28], the concurrent loss of histone acetylation and H3K4 dimethylation suggests that Trrap is required for the function of H3K4 methyltransferase in stem cells. Trrap may mediate the expression and/or activity of H3K4 methyltransferase. Alternatively, Trrap-mediated histone acetylation may positively cross-influence H3K4 methylation by serving as docking sites for these methyltransferases, a mechanism described for retention of bromodomains on gene promoters [34]. Although it remains to be determined whether Trrap mediates the activity of H3K4 methyltransferase directly or indirectly (through histone acetylation), the interactions between histone acetylation and methylation as well as between acetylation and phosphorylation with both positive and negative cross-influences have been described [35, 36].

The mechanism by which loss of Trrap triggers an increase in H3K27 trimethylation and loss of H3K4 dimethylation is unclear. Loss of Trrap and histone acetylation may trigger Polycomb-mediated trimethylation of H3K27, although whether Trrap plays a role in the regulation of Polycomb remains to be determined. The coupling of H3K4 and H3K27 methylation marks is believed to be important in the establishment of the bivalent chromatin domains, the structures that are important for marking key developmental genes in ESCs [37]. Loss of histone acetylation and increase in H3K27 methylation in cells lacking Trrap are associated with condensation of chromatin into distinct foci (heterochromatization) and may promote silencing of the transcriptionally active genome of ESCs. Therefore, the role of Trrap and Trrap-mediated histone acetylation in stem cells may be to impede the formation of heterochromatin, thus keeping stem cells in an open chromatin state. It is believed that virtually all genes in the ES genome are accessible and that competition between different genes exists [23]; therefore, the transcriptional potential of the genome of Trrap-deficient ESCs is likely to be restricted. Increased expression of differentiation-specific genes (such as Gata6) in differentiating Trrap-deficient ESCs suggests that Trrap also participates in the repression of differentiation-specific genes in stem cells. Because Trrap was not associated with the promoters of Gata6 and Pax3, Trrap may inhibit the differentiation genes indirectly, by stimulating the expression of gene repressors, although the possibility that Trrap protein acts as a direct repressor of differentiation genes could not be ruled out. Therefore, further studies are needed to establish the precise mechanism underlying Trrap-mediated repression of the differentiation genes in ESCs.

Our genome-wide binding analysis of Trrap and Oct4 revealed many common Trrap/Oct4 noncoding RNA and coding loci in ESCs, among which there was enrichment in morphogenesis and developmental pathways. Importantly, ChIP-seq analysis revealed a distinct distribution of Trrap and Oct4 in ESCs and MEFs. A significant co-occupancy of Trrap and Oct4 in ESCs is consistent with a physical and functional interaction between Oct4 chromatin-modifying complexes in the maintenance of stemness [24], and the striking relocalization of Trrap and Oct4 binding in MEFs may reflect the need for chromatin reconfiguration with differentiation. It is to be noted that although the overlap between Oct4, Trrap, and AcH4 peaks is significantly higher in ESCs, the loss of overlap in MEFs may be, at least in part, due to a lower number of Oct4 and Trrap peaks. Our finding that the number of AcH4 peaks is significantly greater in MEFs than in ESCs may be explained by differentiation-specific compartmentalization of chromatin into distinct active regions, as opposed to hyperdynamic homogeneously acetylated chromatin in ESCs.

On the basis of our findings, we propose that Trrap may restrict differentiation and maintain stem cell properties by simultaneous activation of many genes, likely through the default pluripotent transcription factor network [23] and by repression of distinct differentiation genes (supporting information Fig. S1). Because some cancers are believed to follow a stem cell model in which the differentiation of cancer stem cells may generate functional and phenotypic heterogeneity [38], it remains to be tested whether the Trrap-mediated mechanism controlling self-renewal and differentiation of ESCs is shared by cancer stem cells. Of note, TRRAP was found to play a role in maintaining a tumorigenic, stem cell-like state [12], and cancer genome studies have uncovered recurrent mutations in TRRAP [39, 40]. Establishing how HATs and histone modifications control key features of normal stem cells and putative cancer-initiating cells may prove beneficial for a better understanding molecular mechanisms underlying tumor development.

CONCLUSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

The results demonstrate that Trrap cofactor of HATs is essential to maintain ESC characteristics and its loss induces unscheduled differentiation of ESCs. In addition, conditional Trapp knock out altered ESCs specific chromatin features inducing heterochromatization and loss of histone dynamics followed by a dramatic downregulation of the stemness master genes Nanog, Oct4, and Sox2 coupled with an overexpression of differentiation genes of the three germ layers. Moreover, loss of Trrap induced a dramatic decrease in acetylation and the histone mark associated with the active chromatin states (H3K4 dimethylation) associated with a striking increase in H3K27 trimethylation, the histone mark associated with the repressive state thus demonstrating that Trrap is essential for the maintenance of active versus repressive methyl marks at the promoters of the stemness master genes in ESCs. Along with this line we demonstrate that failure to downregulate Trrap prevent ECSs from differentiating when cultured in differentiating conditions. ChIP-seq helps us understand the relationship between Trrap and Oct4 occupancy during differentiation showing a significant overlap between Oct4 and Trrap binding in ESCs but not in differentiated MEFs, supporting a functional interaction between Trrap and Oct4 in the maintenance of stemness. In conclusion we establish Trrap as an essential player of the mechanism that restricts differentiation and promotes the maintenance of key features of ESCs.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

We thank Laszlo Tora for his generous gift of the anti-TRRAP antibodies and Michael Cole for the TRRAP expression vector. We are grateful to Karen Müller and John Daniel for editing the manuscript. CS is supported by a fellowship from the Association pour la Recherche sur le Cancer (ARC; France). TV is supported by a Ph.D. fellowship from the Ligue Nationale contre le Cancer (LNCC; France). This work was supported by grants from ARC, the Association for International Cancer Research (AICR; U.K.), the Ligue Nationale contre le Cancer (LNCC, France), and the Institut National du Cancer (INCA, France) (to Z.H.).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. MATERIALS AND METHODS
  5. RESULTS
  6. DISCUSSION
  7. CONCLUSION
  8. Acknowledgements
  9. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
  10. REFERENCES
  11. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
sc-12-0918_sm_SupplTable1.pdf63KSupplementary Table 1
sc-12-0918_sm_SupplTable2.xls115KSupplementary Table 2
sc-12-0918_sm_SupplFigure1.jpg2628KSupplementary Figure 1

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.