D-2-hydroxyglutarate produced by mutant IDH1 perturbs collagen maturation and basement membrane function

Brain-specific Idh1 R132 mutation results in perinatal lethality

Previous in vitro studies have indicated that the mutant protein encoded by IDH1 R132 gains abnormal catalytic activity that allows it to produce D2HG (Dang et al. 2009; Gross et al. 2010; Ward et al. 2010). To determine whether this effect occurs in vivo, we crossed Idh1^LSL/wt mice (Sasaki et al. 2012) with Nestin-Cre transgenic mice (Tronche et al. 1999) to generate Nes-Idh1^R132H/wt (Nes-KI) mice and the corresponding control Nes-Idh1^wt/wt (Nes-WT) animals. Under the control of Nestin, Cre is expressed as early as embryonic day 10.5 (E10.5) in neural stem cells (NSCs), and nearly complete recombination in all CNS cells is achieved by E12.5 (Graus-Porta et al. 2001). We found that Nes-KI mice were born at the expected Mendelian ratio but died shortly after birth (Table 1). Gross examination of whole brains from postnatal day 0 (P0) Nes-KI and Nes-WT mice revealed that expression of the mutant Idh1 protein driven by the Nestin promoter resulted in massive hemorrhage within the cerebral hemispheres and cerebellum (Fig. 1A). Histological analysis confirmed the lack of cerebral development and the presence of large cavities in Nes-KI brains at P0 (Fig. 1B). Even at E15.5, significant cavitation and hemorrhage were already present in Nes-KI embryos (Supplemental Fig. 1A,B), and the laminar nature of the cortex was disturbed (Fig. 1C–E). Cellular proliferation in Nes-KI brains at E15.5 was virtually absent, as determined by BrdU incorporation (Fig. 1F–H). Massive cell death throughout these brains was evidenced by the presence of an increased number of cells positive for cleaved caspase-3, a marker of apoptosis (Fig. 1I–K). Notably, there were no obvious differences between Nes-KI and Nes-WT embryos in NSC distribution at E12.5 (Supplemental Fig. 1C–H).

To extend our results to another brain-specific promoter, we crossed Idh1^LSL/wt mice with GFAP-Cre transgenic mice (Kwon et al. 2001) to generate GFAP-KI mutants and the corresponding GFAP-WT control mice. Previous reports on GFAP-Cre transgenic mice have indicated that Cre is expressed starting at E14.5 but only in the astrocytes of these animals and not in their neurons (Bajenaru et al. 2002). GFAP-KI embryos did not show any significant abnormalities until E15.5, after which ∼60% of embryos exhibited brain hemorrhage (Supplemental Fig. 2A,B). Interestingly, GFAP-KI embryos of the same litter demonstrated variable symptoms (Supplemental Fig. 2B), and, unlike Nes-KI mice, 8.3% of GFAP-KI mice survived until adulthood (Supplemental Table 1). This generally milder phenotype of GFAP-KI embryos may be partly due to the variable mosaic pattern of Cre expression documented among GFAP-Cre mice (see below).

D2HG accumulates to high levels in Nes-KI brains

D2HG accumulates in glioma and AML patients whose malignancies bear IDH1 or IDH2 mutations and inhibits αKG-dependent dioxygenase activity by competing with αKG (Dang et al. 2009; Gross et al. 2010; Ward et al. 2010; Chowdhury et al. 2011; Xu et al. 2011). When we measured αKG and D2HG levels in Nes-KI brain cells, αKG was modestly decreased in the mutant compared with the Nes-WT control, as expected (Fig. 2A). Strikingly, huge amounts of D2HG were detected in the brain cells of both Nes-KI and GFAP-KI mice compared with controls (Fig. 2B; Supplemental Fig. 3A). Significantly, GFAP-KI mice with brain hemorrhage also showed high levels of D2HG, whereas GFAP-KI mice without hemorrhage did not (Supplemental Fig. 3B). Thus, D2HG levels correlate closely with the hemorrhage phenotype.

ROS levels are reduced in Nes-KI brains

To elucidate the mechanisms underlying the brain hemorrhage observed in Idh1-KI mice, we first investigated the regulation of ROS in the mutant brain cells. Since the mutated IDH1 protein loses catalytic activity for the normal αKG- and NADPH-generating reaction and gains abnormal enzymatic reactivity that produces D2HG and consumes NADPH, it is plausible that cells expressing the mutated IDH1 enzyme would show an increase in their intracellular NADP⁺/NADPH ratio. As NADPH is a cofactor in the recycling of anti-oxidant molecules such as glutathione (GSH) and ascorbate, it has also been assumed that such an increase in the NADP⁺/NADPH ratio would lead to a corresponding increase in intracellular ROS. Indeed, total brain cells from E14.5 Nes-KI embryos showed a slight but significant increase in their intracellular NADP⁺/NADPH ratio compared with Nes-WT controls (Fig. 3A), consistent with a loss of conventional NADPH production and enhanced NADPH consumption. In addition, intracellular levels of reduced GSH and ascorbate were decreased in Nes-KI cells (Fig. 3B,C), implying that the observed reduction in NADPH attenuates the recycling of GSH and ascorbate. However, to our surprise, intracellular ROS levels measured using the CM-H₂DCFDA fluorescent probe were not elevated in total Nes-KI brain cells but instead were dramatically decreased (Fig. 3D). One possible explanation for this reduction of intracellular ROS may be due to enhanced catalase activity, but not protein expression, detected in the mutant cells (Fig. 3E,F).

We then examined intracellular ROS levels in various subpopulations of embryonic brain cells. Moderate ROS accumulation in NSCs reportedly promotes their proliferation and differentiation (Le Belle et al. 2011), so we wondered whether a drop in intracellular ROS would alter NSC numbers and perhaps account for the brain hemorrhage observed in Idh1-KI mice. For our analysis, we monitored the expression of Prominin-1 (CD133 in humans) as a marker of NSCs, and A2B5 as a marker of glial-restricted precursor cells (GPCs). The double-negative (DN) subpopulation that lacks both of these markers includes neural-restricted precursor cells, terminally differentiated neurons and glial cells, and other cell types such as endothelial cells. We found that, compared with Nes-WT controls, ROS levels were reduced in the Prominin-1⁺ (NSC), Prominin-1⁻A2B5⁺ (GPC), and Prominin-1⁻A2B5⁻ (DN) cell populations in Nes-KI brains (Fig. 3G–I). However, the proportions of these three subsets among total Nes-KI brain cells were normal and consistent with the pattern of Nestin expression in Nes-KI brains (Fig. 3J; Supplemental Fig. 1D,G). Taken together, these results indicate that the increased cell death and reduced proliferation in the mutant brains cannot be due to ROS accumulation and suggest that the Idh1 R132 mutation does not influence NSC proliferation or differentiation, at least during the earliest stages of embryonic brain development.

In addition to its effects on ROS levels, D2HG reportedly increases histone methylation because it inhibits the JHDMs required for cell differentiation (Lu et al. 2012). However, we found no differences in histone H3 methylation between the E14.5 control and Nes-KI NSCs or GPCs (Fig. 3K). On the other hand, Nes-KI NSCs and GPCs both showed reduced levels of 5-hydroxymethylcytosine (5hmC) compared with controls (Fig. 3L–N), presumably due to D2HG-mediated inhibition of TET hydroxylases. Precisely how such altered epigenetic modifications could contribute to the phenotypes of Idh1-KI mice remains to be clarified.

D2HG stabilizes Hif1α proteins in Nes-KI mice

Xu et al. (2011) previously demonstrated that D2HG blocks the hydroxylation of proline residues in HIF1α proteins, but Koivunen et al. (2012) recently presented evidence suggesting that D2HG can activate PHDs. To resolve this controversy, we assessed Hif1α expression in our Nes-KI mutants. We detected higher levels of HIF1α protein in Nes-KI embryos compared with controls (Fig. 4A), but the same was not true for Hif1a mRNA (Fig. 4B,C, left). Moreover, the increased stabilization of Hif1α protein was observed even in E14.5 Nes-KI brains showing little or no hemorrhage. When we examined the mRNA expression of Hif1α target genes in Nes-KI embryos, we found that transcription of the vascular endothelial growth factor (Vegf) gene was preferentially up-regulated in E13.5–E16.5 Nes-KI brains compared with Nes-WT controls (Fig. 4B,C). Transcription levels of the Hif1α target genes glucose transporter-1 (Glut-1) and phosphoglycerate kinase 1 (Pgk1) were also moderately up-regulated at E15.5–E16.5 (Fig. 4C). On the other hand, the expression of angiopoietin-1 (Ang1), which is not a Hif1α target gene, was unaffected. Thus, mutation of Idh1 is associated with stabilized Hif1α and up-regulated Hif1α-dependent transcription. We therefore concur with Xu et al. (2011) that the D2HG associated with the Idh1 mutation blocks PHD-mediated modification of HIF1α protein and thus promotes its transactivation activity.

Collagen maturation is impaired in Nes-KI mice

The impaired proline hydroxylation of Hif1α in Nes-KI brains prompted us to explore whether other processes involving prolyl hydroxylases were affected in our mutants. Prolines in collagen proteins are hydroxylated by collagen prolyl-4-hydroxylases I, II, and III (C-P4H I–III), and lysines in collagen proteins are hydroxylated by the procollagen-lysine 2-oxoglutarate 5-dioxygenases 1, 2, and 3 (Plod1–3) (Kivirikko and Myllylä 1985; Myllyharju and Kivirikko 2004). Increased amounts of hydroxyproline (Hyp) and hydroxylysine (Hyl) residues in a mature collagen protein reinforce its helical structure and promote its thermal stability (Kivirikko and Myllylä 1985; Myllyharju and Kivirikko 2004). Type IV collagen is a key BM component and promotes interactions between astrocytes and endothelial cells (Kalluri 2003). We hypothesized that the D2HG accumulation in Nes-KI mice might attenuate collagen modification by αKG-dependent enzymes and thus disrupt type IV collagen maturation, leading to BM breakdown. To investigate this possibility, we first analyzed levels of soluble collagen in Nes-KI and Nes-WT brains, since an increased proportion of soluble collagen indicates a lack of maturation of this protein. We recovered the fraction of cell extract proteins soluble in Triton X-100 and performed immunoblotting to detect type IV collagen. Soluble type IV collagens α1 and α2 were dramatically increased in tissues of Nes-KI mutants compared with Nes-WT controls (Fig. 5A). However, when we evaluated type IV collagen mRNAs in the brains of these embryos, four out of six Col4a mRNAs were expressed at comparable levels in mutant and control cells (Supplemental Fig. 4A). Thus, the increase in soluble type IV collagen proteins in Nes-KI brain cells is most likely due to the accumulation of immature, nonhelical forms of these molecules. Consistent with this hypothesis, the migration in a 5% SDS gel of the band representing the type IV collagen protein in the Nes-KI mutant was faster than the band in the Nes-WT extract (Fig. 5A), implying that Hyl-mediated glycosylation was impaired. Similar phenotypes have been observed in tissues of mice deficient for P4ha1 (the α subunit of C-P4H I) or Plod3 (Rautavuoma et al. 2004; Holster et al. 2007). Thus, we propose that the excess D2HG present due to the activity of the mutated Idh1 enzyme impairs Hyp- and Hyl-mediated modification of collagen proteins, preventing their maturation.

To confirm the physiological significance of these in vitro findings, we performed immunofluorescence staining of type IV collagen to analyze its distribution between blood vessels and perivascular astrocytes. Intense, continuous staining of blood vessel type IV collagen was present in Nes-WT E15.5 brains, whereas only faint, dispersed staining was seen in the mutant (Fig. 5B). These data further support our hypothesis that D2HG attenuates type IV collagen maturation and thereby impairs BM structure and function.

Defective collagen maturation in Nes-KI mice is associated with an ER stress response

C-P4H I–III and Plod1–3 are localized in the ER. When the activities of these enzymes are suppressed, newly synthesized collagens are not properly modified, leading to the accumulation of unfolded proteins in the ER (Marutani et al. 2004; Rautavuoma et al. 2004; Holster et al. 2007). Because the presence of these abnormal proteins in bulk triggers an ER stress response that leads to apoptosis (Ron and Walter 2007), we hypothesized that such a response might be induced by the impaired collagen maturation in Nes-KI mutants. To test this idea, we used quantitative RT–PCR (qRT–PCR) to examine the mRNA levels of several ER stress-responsive genes in the brains of Nes-KI and Nes-WT embryos at E14.5–E16.5. Specifically, we assayed ER stress-responsive transcription factors (Atf4 and Chop), ER chaperones (Hsp47, Herp, Edem1, Grp78/Bip, Grp94, and p58IPK), an ER stress sensor protein (Oasis), and other ER stress-related proteins (Gadd34, PDI, and Erol1). Of these 12 genes, three (Chop, Atf4, and Herp) were significantly up-regulated in E14.5 Nes-KI brains, whereas seven (Chop, Atf4, Hsp47, Herp, Edem1, Oasis, and Gadd34) were up-regulated by E15.5–E16.5 (Fig. 5C,D; Supplemental Fig. 4B). This elevation of ER stress-responsive gene expression is most likely responsible for the increased apoptosis observed in Nes-KI brains.

GFAP-KI mice do not develop glioma

To determine whether the Idh1 R132H mutation on its own can drive glioma formation, we monitored the health and longevity of our live-born GFAP-KI mice (8.3% of the total) (Supplemental Table 1). Surviving GFAP-KI mice showed a markedly shorter life span than GFAP-WT mice (Supplemental Fig. 5A) without observable glioma formation. These results suggest that, in the brain, expression of the mutated Idh1 enzyme alone is insufficient for gliomagenesis. Alternatively, glioma may be formed at a later time point, but this may be complicated by the reduced life span of the mutant mice.

Intriguingly, many surviving GFAP-KI mice exhibited splenomegaly (39%) and/or liver tumors (21%) or other abnormalities (Supplemental Fig. 5B–D). GFAP-KI mice with splenomegaly showed a modest increase in total white blood cells and reticulocytes and a decrease in red blood cells but no alteration to platelet numbers (Supplemental Fig. 5E). To determine the cause of these non-brain-related phenotypes, we verified the expression pattern of GFAP-Cre with Rosa26-LSL-TdTomato reporter mice (Madisen et al. 2010). GFAP-Cre-mediated expression of TdTomato protein was observed not only in the brains, but also in a wide variety of other tissues. In most mutants, Cre was expressed in a mosaic pattern unique to each individual, but some mutants showed high Cre expression throughout their whole bodies (Supplemental Fig. 6). Thus, it is most likely that the non-brain phenotypes in GFAP-KI mice arose due to leaky expression of Cre in hematopoietic cells. It has been previously shown that Tet2 knockout mice also exhibit defects in hematopoietic stem cell homeostasis but show no brain abnormalities (Ko et al. 2011; Li et al. 2011; Moran-Crusio et al. 2011; Quivoron et al. 2011). This similarity in phenotypes between GFAP-KI mice and Tet2 knockout mice suggests both that hematopoietic cells are probably more sensitive to D2HG than brain cells and that these phenotypes may partly depend on D2HG-mediated inhibition of Tet2. We also examined the phenotypes of mice in which the Idh1-KI mutation is driven by Cre under the control of the myeloid lineage-specific promoter LysM (Sasaki et al. 2012). These animals have hematopoietic phenotypes similar, but not identical, to those of GFAP-KI mice, presumably reflecting a different Cre expression pattern under the two different promoters in various hematopoietic lineages.

An obvious way to enhance the chance of glioma formation in GFAP-KI mice was to cross these animals to mutants with a deletion of the tumor suppressor p53. However, due to the leaky expression of GFAP-Cre, both control GFAP-WT/p53^flox/flox mice and GFAP-KI/p53^flox/flox double-mutant mice developed the same broad spectrum of tumors (Supplemental Fig. 7). It remains a formidable challenge to generate a live Idh1-KI mouse that expresses mutant Idh1 protein only in the brain, can be crossed into a tumor-prone background, and can thus be used to identify the other factors that combine with altered Idh1 activity to drive glioma formation.

Idh1 mutation and ROS regulation

The conversion of intracellular NADPH to NADP⁺ supplies vital reducing power for the generation of anti-oxidant molecules such as reduced GSH and ascorbic acid (Finkel and Holbrook 2000). As expected, we detected an increase in the NADP⁺/NADPH ratio and a decrease in GSH and reduced ascorbic acid in Idh1-KI cells. However, although an elevated NADP⁺/NADPH ratio has been assumed to induce ROS accumulation, we found dramatically lower ROS levels and increased catalase activity in Idh1-KI cells. Thus, excessive oxidative stress is not the cause of the brain hemorrhage in Idh1-KI mice. The normal Idh1 protein is localized in both the cytosol and the peroxisome, where catalase resides. Peroxisomes are organelles with essential metabolic functions that affect the generation and decomposition of ROS (Geisbrecht and Gould 1999). It is possible that aberrant metabolism in the peroxisomes of Idh1-KI mice results in the observed dysregulation of anti-oxidants. In addition to changes in catalase activation, alterations in the level of NADPH oxidases, which are enzymes that generate superoxide from oxygen and NADPH (Ying 2008), may also contribute to the observed phenotype, although it is technically challenging to experimentally interrogate this idea.

Idh1 mutation and brain hemorrhage

Brain hemorrhage is an often fatal complication of premature birth, vascular malformations, and brain disorders. The mechanisms that cause brain hemorrhage may be specific to this organ because of the close connection between brain endothelial cells and perivascular astrocytes (Ballabh et al. 2004). Two lines of evidence support our view that D2HG accumulation is directly linked to brain hemorrhage.

Idh1 mutation and tumorigenesis

Due to the embryonic lethality of Nes-KI mice and the leakiness of Cre expression in GFAP-KI mice, we cannot state unequivocally that Idh1 mutations cause brain tumors in vivo. However, none of our live-born GFAP-KI mice developed spontaneous brain tumors, implying that genetic alterations in addition to Idh1 mutation are required for tumor initiation in mice. Nevertheless, our findings clearly demonstrate that Idh1 mutation induces cellular and microenvironmental changes that have been linked to tumor promotion. Although ROS possess mutagenic/oncogenic potential, reduced ROS levels have been previously shown to promote tumor growth and progression (at least in certain contexts) (Finkel and Holbrook 2000). BM breakdown may therefore contribute to tumor growth and progression (Tate and Aghi 2009). In addition, D2HG is associated with increased Hif1α activity and Vegf expression, which can facilitate tumor angiogenesis. Notably, glioma patients with IDH1 mutations that are treated with VEGF(R)-blocking agents such as bevacizumab and sunitinib malate exhibit prolonged survival and delayed tumor recurrence compared with patients treated with cetuximab, a monoclonal antibody that blocks the epidermal growth factor receptor (EGFR) (Lv et al. 2011).

Our findings are consistent with a previous human study in which mutations in the coding sequences of genes encoding collagens and other extracellular matrix-related proteins were identified in 13 out of 21 GBM tumor samples (Supplemental Table 2; Parsons et al. 2008). A different study of 91 GBM tumor samples discovered additional mutations in exons of four collagen genes (The Cancer Genome Atlas Research Network 2008). Most recently, 152 mutations of 28 collagen genes were identified in 316 ovarian tumors (The Cancer Genome Atlas Research Network 2011). Importantly, GBM patients with BM-related gene mutations rarely carry IDH1 mutations (Supplemental Table 2). Moreover, IDH1 mutations occur in chondromas (Amary et al. 2011a,b; Pansuriya et al. 2011), and these neoplasms show alterations to collagen distribution upon progression to malignancy (Ueda et al. 1990). These data collectively indicate that normal collagen maturation is important for suppressing cancer and suggest that IDH1 mutation leading to D2HG production severely impairs collagen functions in glioma cells such that tumor progression is facilitated.

Another factor associated with tumorigenesis in general is altered epigenetic modification, and dramatic epigenetic changes are observed in tumor cells in glioma and AML patients with IDH1/2 mutations (Figueroa et al. 2010; Noushmehr et al. 2010). Although we did not detect any significant changes in histone H3 methylation of proteins in very early stage Nes-KI embryos, 5hmC was greatly reduced in both NSCs and GPCs of these mutants (Fig. 3K–N). We therefore speculate that a key tumorigenic aspect of IDH1 mutation is D2HG-mediated alteration of epigenetic modification that also favors tumor promotion.

In conclusion, our in vivo study has demonstrated that brain-specific mutation of Idh1 is embryonic-lethal in mice due to brain hemorrhage. In brain cells isolated from these mutants, D2HG produced by the activity of the abnormal Idh1 enzyme may function as an oncometabolite that induces HIF target gene transactivation, disrupts collagen maturation, and impairs BM structure. Since GFAP-Cre-driven Idh1 mutant mice are not suitable for the glioma model, future studies are needed to establish the tumorigenic roles of Idh1 mutation by using inducible Cre systems.

Mice

Idh1 (Sasaki et al. 2012), Nestin-Cre (Tronche et al. 1999), GFAP-Cre (Kwon et al. 2001), Rosa-LSL-TdTomato (Madisen et al. 2010), and p53^flox (Jonkers et al. 2001) mice have been described previously. All animals were treated in accordance with the NIH Guide for Care and Use of Laboratory Animals as approved by the Ontario Cancer Institute Animal Care Committee.

Mouse tissue preparation

For total brain cell preparations, brains were dissected from mouse embryos and dissociated to create single-cell suspensions via enzymatic degradation using a neural tissue dissociation kit (Miltenyi Biotec) according to the manufacturer's protocol. Briefly, whole embryonic brains were minced, a prewarmed enzyme mix was added to the tissue pieces, and the mixture was incubated for 15 min at 37°C. The tissue was further mechanically dissociated by trituration, incubated for 20 min at 37°C, and passed through a 70-μm cell strainer. To isolate NSCs, GPCs, and DN cells, we applied the extracts to MACS prominin-1- or A2B5-coupled microbeads (Miltenyi Biotec). After two washes in PBS(−), cells were processed for various assays as described below.

Metabolite determinations

For D2HG measurement, metabolites were extracted using 80% aqueous methanol as previously described (Lu et al. 2006). Briefly, 30–50 mg of embryonic brain was immersed in 80% methanol at −80°C and homogenized. Extracts were centrifuged at 13,000 rpm for 15 min at 4°C, and supernatants were dried at room temperature and stored at −80°C. Extracts were subjected to ion-paired reverse-phase LC coupled to negative mode electrospray triple-quadrupole mass spectrometry using multiple reaction monitoring. Integrated elution peaks were compared with metabolite standard curves for absolute quantification. For intracellular ROS measurement, cells (1 × 10⁵) were incubated with 100 μM CM-H₂-DCFDA (Invitrogen) for 30 min at 37°C. Data were collected on a FACSCalibur flow cytometer (Becton Dickinson). For NADP⁺/NADPH ratios and αKG measurements, cells (2 × 10⁶) were processed using the Fluoro NADP/NADPH detection kit (Cell Technology) or the α-Ketoglutarate Assay kit (BioVision), respectively. For GSH/GSSG ratios, 2 × 10⁶ (for GSH) and 1 × 10⁶ (for GSSG) cells were lysed by two freeze/thaw cycles and assessed using the Bioxytech GSH/GSSG-412 kit (OxisResearch). For ascorbic acid, cells (4 × 10⁵) were evaluated using the EnzyChrom Ascorbic Acid Assay kit (BioAssay Systems). For catalase activity, cells (1 × 10⁶) were assessed using the Catalase Assay kit (Sigma-Aldrich). For all kits, the manufacturers' protocols were followed.

qRT–PCR

Total RNA was extracted and purified using the RNeasy kit (Qiagen) according to the manufacturer's protocol. Purified RNA was treated with DNaseI and reverse-transcribed using the iScript cDNA Synthesis kit (Bio-Rad). cDNAs were subjected to qRT–PCR using SYBR Green PCR Master Mix (Applied Biosystems). Primers used for qRT–PCR are listed in Supplemental Table 3.

Immunoblotting

For detection of most proteins, brain cell lysates were prepared in lysis buffer (20 mM Tris-HCl at pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.1% [v/v] 2-mercaptoethanol, 0.1% [v/v] Triton X-100, 1× protease inhibitor cocktail [Roche]), electrophoretically separated on denaturing SDS polyacrylamide gels, and transferred to a PVDF membrane (Roche). For histone protein detection, E14.5 NSCs and GPCs were suspended in extraction buffer (PBS[–] containing 0.5% [v/v] Triton X100, 1× protease inhibitor cocktail [Roche]) for 10 min on ice. HCl was added to 0.2 M followed by incubation overnight at 4°C. After centrifugation to recover the supernatant, proteins were diluted in 1× NuPAGE LDS sample buffer (Invitrogen) supplemented with 50 mM DTT and electrophoresed as above. For collagen protein detection, E14.5–E15.5 whole embryonic brains were homogenized in homogenization buffer (20 mM Tris-HCl at pH 7.4, 0.2 M NaCl, 0.1 M glycine, 0.1% [v/v] Triton X-100, 50 μM DTT, 1× protease inhibitor cocktail [Roche]). Proteins were separated on a 5% Tris-glycine SDS-PAGE gel and transferred to a PVDF membrane.

Proteins fixed on PVDF membranes were visualized using antibodies to type IV collagen (600-401-106, Rockland), HIF1α (10006421, Cayman Chemical), H3K4me3 (ab1012, Abcam), H3K9me3 (ab8898, Abcam), H3K9me2 (ab1220, Abcam), H3R17me2 (8284, Abcam), H3K36me3 (ab9050, Abcam), histone H3 (ab10799, Abcam), H3K27me3 (07-449, Millipore), H3K79me2 (9757, Cell Signaling Technologies), catalase (ab1877, Abcam), or β-actin (A2066, Sigma-Aldrich). Primary antibodies were detected using HRP-conjugated secondary antibodies to mouse IgG or rabbit IgG (GE Healthcare). Immunocomplexes were visualized using ECL + reagents (GE Healthcare).

Quantitative analysis of 5hmC

Detection of 5hmC was performed as previously described (Ko et al. 2010). Briefly, genomic DNA was treated with sodium bisulfite using the EpiTect bisulfite kit (Qiagen). DNA samples were denatured, and twofold serial dilutions were spotted onto a nitrocellulose membrane in an assembled Bio-Dot apparatus (Bio-Rad). The membrane was incubated with anti-cytosine 5-methylenesulfonate antibody. After incubating with HRP-conjugated anti-rabbit IgG secondary antibody, the membrane was visualized by enhanced chemoluminescence.

Histology

Tissue samples were fixed in 10% buffered formalin or 4% paraformaldehyde (PFA) at 4°C and embedded in paraffin. Sections (5 μm) were stained with Harris hematoxylin and eosin according to standard protocols. For cell proliferation assays, pregnant females were injected intraperitoneally with 0.1 mg BrdU (Roche) per gram body weight 45 min before dissection to isolate embryos. Paraffin sections of fixed embryos were processed using the BrdU In Situ Detection kit (Boehringer Mannheim) to detect BrdU. For immunostaining, sections were dewaxed, rehydrated, and microwaved in 10 mM sodium citrate. Slides for peroxidase reactions were treated with 3% hydrogen peroxide in PBS to kill endogenous peroxidase activity. Slides were incubated overnight at 4°C in primary antibody, washed, and incubated in secondary antibody for 1 h at room temperature. Slides for peroxidase reactions were treated with ABC reagent (Vector Laboratories) for 1 h and washed in PBS. Primary antibodies recognizing cleaved caspase-3 (9661, Cell Signaling) or type IV collagen (600-401-106, Rockland) were visualized using secondary antibodies as above. Microscopic observations and images were acquired using a Leica DM6000 or Leica DMI6000 instrument.