Histone deacetylase 1 and 2 regulate Wnt and p53 pathways in the ureteric bud epithelium

Histone deacetylases (HDACs) are a superfamily of enzymes important for modulation of chromatin structure and function via removal of acetyl moieties from lysine residues of histones as well as non-histone proteins. In higher eukaryotes, 18 HDAC isoforms have been identified. Based on sequence homology to yeast HDAC genes, they are divided into four classes: the class I RPD3-like HDACs (1-3 and 8); the class II HDA1-like HDACs (4-7, 9 and 10); Sirtuins 1-7; and the class IV HDAC11 (de Ruijter et al., 2003; Denu and Gottesfeld, 2012).

HDACs were originally assumed to be global transcriptional co-repressors, but it subsequently became clear that HDACs regulate gene expression in a highly selective manner and exhibit both repressive and activating effects. HDAC inhibition or deletion of HDAC genes often results in alterations in a small subset of genes (<10%), and approximately as many are downregulated as are upregulated (Smith, 2008). Mechanistically, HDACs themselves lack intrinsic DNA binding activity and are recruited to target genes through association with various transcriptional complexes (e.g. the Sin3, NuRD, Co-REST and SMRT/N-CoR complexes) (Marks et al., 2003). As such, the specificity of HDACs in gene regulation depends on the partner proteins that they associate with under different pathophysiological conditions. In addition to histones, HDACs can also deacetylate an increasing number of non-histone proteins, including p53, STAT3, Yin Yang transcription factor (YY1), GATA1, E2F1 and Hsp90 (Glozak et al., 2005; Kruse and Gu, 2009; Spange et al., 2009). Deacetylation of these proteins has been shown to affect multiple aspects of their function, such as protein stability, DNA binding affinity and transcriptional activity, adding an extra layer of complexity to the biological roles of HDACs.

In line with recent studies showing that HDACs play both unique and redundant roles in the control of distinctive developmental pathways during embryogenesis, our previous studies revealed that the expression of key developmental renal regulators (e.g. Osr1, Eya1, Pax2/8, Wt1 and Wnt9b) is dependent on intact HDAC activity (Chen et al., 2011; Haberland et al., 2009). Treatment of cultured mouse embryonic kidneys with Scriptaid, a general inhibitor of class I and II HDACs, or MS-275, a selective inhibitor of HDAC1-3, impairs ureteric bud (UB) branching morphogenesis and nephrogenesis, two central processes of metanephric development. Our results also demonstrated that Hdac1-3 are enriched in the undifferentiated metanephric mesenchyme (MM) and UB branches, but are reduced upon differentiation, implicating their crucial role in mouse kidney development.

Among class I and class II HDACs, HDAC1 and HDAC2 are evolutionarily close to each other, sharing a high degree of sequence similarity (∼85% identity at the amino acid level) except for their C-terminal domain. HDAC1 and HDAC2 can form homo- or heterodimers and are found together in almost all nuclear protein complexes, including three well-characterized co-repressor complexes: Sin3, NuRD and Co-REST (Brunmeir et al., 2009). Studies using a variety of genetic models of Hdac1 and Hdac2 indicate that they exert both overlapping and non-redundant functions in different cell types at specific developmental stages (Haberland et al., 2010; Lagger et al., 2002; LeBoeuf et al., 2010; Ma et al., 2012; Montgomery et al., 2007; Ye et al., 2009). In this study, by deleting different combinations of Hdac1 and Hdac2 alleles in the UB cell lineage, we revealed the redundant yet essential cell-autonomous functions of Hdac1 and Hdac2 in the ureteric epithelium during mouse kidney development. Moreover, our findings illustrate the developmental importance of HDAC-mediated control of p53 acetylation.

In order to delete the Hdac1 and Hdac2 genes specifically from the UB lineage, we crossed Hdac1^flox/flox, Hdac2^flox/flox and Hoxb7-CreEGFP transgenic mice (Montgomery et al., 2007; Zhao et al., 2004). Previous studies have shown that Hoxb7-directed GFP expression is observed in the Wolffian duct at embryonic day (E) 10.0 and its derivatives, the UB and its branches, but not in the MM lineage (Zhao et al., 2004). To test the efficacy of Hoxb7-driven Cre-mediated excision, we examined the expression of Hdac1 and Hdac2 proteins by immunohistochemistry in wild-type and mutant kidney tissues at E13.5. Consistent with our previous report, Hdac1 and Hdac2 are expressed in both the UB and MM cells in wild-type mice (Fig. 1A,E). By contrast, in UB^{Hdac1,2−/−} mice (Hoxb7-Cre^tg/+; Hdac1^flox/flox; Hdac2^flox/flox), Hdac1 and Hdac2 are not detected in the UB cells but are maintained in the surrounding mesenchymal cells (Fig. 1B,F). In accordance with the key functions of Hdac1 and Hdac2 in histone deacetylation, the acetylation levels of histone H3 (lysine 9 and 14, or lysine 9 specifically) and H4 (lysine 5, 8, 12 and 16) are substantially increased in the ureteric cells of UB^{Hdac1,2−/−} kidneys (Fig. 1I-N). Collectively, these results demonstrate efficient deletion of Hdac1 and Hdac2 from the UB branches. It is worth noting that knockout of Hdac1 and Hdac2 in oocytes leads to no apparent change in histone H3K9 acetylation (Ma et al., 2012). Thus, our results suggest that Hdac1 and Hdac2 might have different histone residue specificity in different cell types at specific developmental stages.

Our results revealed that mice with no more than three deleted alleles of Hdac1 and Hdac2 exhibit no significant abnormalities in kidney development (supplementary material Fig. S1); moreover, these mice survive to adulthood without any overt abnormalities in growth or development. By contrast, concurrent deletion of all four alleles of Hdac1 and Hdac2 results in early postnatal lethality by 2-4 weeks of age (supplementary material Fig. S2). Histological analysis of kidney tissue from UB^{Hdac1,2−/−} mice at postnatal day (P) 0 showed absence of the nephrogenic zone, lack of cortico-medullary patterning, and the formation of multiple epithelial cysts (Fig. 2A-F). In line with the histological observations, immunofluorescence staining demonstrated that the UB^{Hdac1,2−/−} neonates completely lack Six2-positive and Pax2-positive cells (Fig. 2G-J).

To begin to define the embryological mechanisms leading to this phenotype, we monitored UB morphogenesis in a real-time manner in vivo and in vitro, taking advantage of the GFP fluorescence in the UB tissue driven by the Hoxb7 promoter. UB^{Hdac1,2−/−} mice exhibited attenuated UB branching as early as E13.5 (Fig. 3A), and started to show degeneration of the UB tissue after 1-2 days in culture (Fig. 3B). Previous studies have shown that Hdac1 and Hdac2 regulate apoptosis and proliferation in a wide range of cells. To examine whether increased apoptosis and decreased proliferation contribute to the observed defects in UB branching morphogenesis, we examined the status of cell apoptosis and proliferation at E13.5, when the abnormal UB branching is first noted. Quantification of active caspase 3 (aCasp3)-positive cells revealed a significant increase in the number of apoptotic UB cells in UB^{Hdac1,2−/−} versus wild-type kidneys at E13.5 (Fig. 4A-E). Consistently, staining of phospho-histone γH2AX, a marker of DNA double-strand breaks, showed that UB^{Hdac1,2−/−} kidneys exhibited increased DNA damage in UB cells (Fig. 4F-I). Analysis of phospho-histone H3 (pH3), a marker of mitosis, revealed that the rate of cell proliferation was decreased by 37.5% in UB^{Hdac1,2−/−} relative to wild-type kidneys (Fig. 4J-N). This is consistent with the known pro-proliferative functions of Hdac1 and Hdac2. Taken together, these results indicated that Hdac1 and Hdac2 are crucial for UB cell growth, survival and branching morphogenesis.

To further elucidate the developmental pathways regulated by Hdac1 and Hdac2, we carried out a genome-wide microarray analysis on RNA samples extracted from wild-type and UB^{Hdac1,2−/−} kidneys at E13.5. The raw and analyzed data have been deposited in the NCBI Gene Expression Omnibus (GEO) under accession number GSE35432. The results revealed that 496/41,000 probes (∼1.2%) are significantly altered in UB^{Hdac1,2−/−} kidneys (by ≥1.4-fold, P<0.05, n=4), of which 226 transcripts (0.55%) were upregulated (range 1.4- to 7.5-fold) and 270 (0.66%) downregulated (range 1.4- to 3.8-fold) (Fig. 5A; see GSE35432).

To analyze whether certain pathways or biological processes are especially sensitive to the loss of Hdac1 and Hdac2 in the UB cells, Ingenuity Pathway Analysis (IPA) was performed on the differentially expressed transcripts. This analysis indicated that the most significantly enriched genes participate in: (1) cell morphology; (2) cellular growth and proliferation; (3) cellular development; (4) cell death and survival and (5) cellular movement (Fig. 5B). The most affected canonical pathways include: (1) basal cell carcinoma signaling; (2) Wnt/β-catenin signaling; (3) sonic hedgehog signaling; (4) human embryonic stem cell pluripotency and (5) tight junction signaling (Fig. 5C). The complete list of genes for each category and pathway is shown in supplementary material Tables S1 and S2.

Further analysis using the Biological Networks Gene Ontology (BiNGO) tool revealed that many genes involved in: (1) tube development; (2) Wnt receptor signaling pathway; (3) ureteric bud development; (4) cell-cell adhesion; (5) kidney development and (6) positive regulation of cell proliferation are downregulated in UB^{Hdac1,2−/−} kidneys (Table 1). Interestingly, we found decreased expression of a group of cytokeratins, including Krt7, Krt8, Krt18, Krt19 and Krt23 (Table 2), which is consistent with our immunostaining results using a pan-cytokeratin antibody (Fig. 1I-N). Cytokeratins, the largest intermediate filament protein group, are recognized as peptide fingerprints for the classification of epithelial cells and are implicated in cytoplasmic organization and cellular communication. In embryonic kidneys, Krt8 and Krt18 are specifically expressed in the ureteric epithelium, whereas Krt23 is specifically expressed in the UB tips.

We validated the microarray results by quantitative real-time PCR (qPCR) and in situ hybridization (ISH) of known developmental regulators in the E13.5 kidney. No change was observed in the mRNA expression of Bmyc, cRet, Wnt11, Etv4, Emx2 and Six2 between UB^{Hdac1,2−/−} and wild-type kidneys (note that Six2 was downregulated later at P0; Fig. 1). By contrast, Axin2, Shh, Tcf7, Wnt4 and Lef1 were downregulated, and the stromal gene Meis1 was upregulated, confirming our microarray results (Fig. 5D and Fig. 6). It should be noted that qPCR and especially ISH demonstrated a significant decrease of Wnt7b and Wnt9b, whereas the microarray failed to detect this decrease. This is likely to be a sensitivity issue: microarray of whole kidney RNA is not a sensitive approach to detect focal changes in gene expression in the ureteric epithelium.

β-catenin is a multifunctional protein that plays a crucial role in UB branching morphogenesis as well as nephrogenesis. One role of β-catenin is to coordinate cell-cell adhesion through the formation of adherens junctions with E-cadherin and α-catenin. Another function of β-catenin is to regulate gene transcription, primarily through interactions with the Tcf/Lef family of transcription factors. This function is under the control of the canonical Wnt signaling pathway. Either deletion or overexpression of β-catenin in UB cells leads to kidney hypoplasia (Bridgewater et al., 2008, 2011; Marose et al., 2008). Hdac1 and Hdac2 have been reported to regulate β-catenin in epidermal progenitor cells and oligodendrocytes. In epidermal progenitor cells, Hdac1 and Hdac2 are required for the elevation of β-catenin during hair follicle fate acquisition. Epidermis that is deficient of Hdac1 and Hdac2 displays uniform, low-level expression of β-catenin (LeBoeuf et al., 2010). Conversely, deletion of Hdac1 and Hdac2 in oligodendrocytes results in the stabilization and nuclear translocation of β-catenin, leading to activation of the canonical Wnt/β-catenin pathway (Ye et al., 2009). Our immunofluorescence results showed that β-catenin is dramatically decreased in the UB cells of UB^{Hdac1,2−/−} kidneys at E13.5 and thereafter (Fig. 7). These data are consistent with our observation made above that Wnt signaling is repressed in UB^{Hdac1,2−/−} kidneys.

To exclude the possibility that we missed an initial activation of Wnt signaling upon loss of Hdac1 and Hdac2 in our animal model, we examined the acute response of cultured embryonic kidneys (E13.5) to HDAC inhibitor, using Axin2, a direct target of canonical Wnt signaling, as a read-out. As indicated by ISH, Axin2 is expressed at a high level in the UB cells and at a low level in the surrounding mesenchymal cells (Fig. 8A). As expected, Axin2 is rapidly induced by the GSK3β inhibitor LiCl, which stabilizes β-catenin and thus activates the canonical Wnt signaling pathway (Fig. 8A). We found that a 6-h treatment with HDAC inhibitor substantially decreases the mRNA level of Axin2 with or without LiCl (Fig. 8B,C). By contrast, expression of cRet is not altered (Fig. 8D). These results suggest that the repression of canonical Wnt signaling in the UB cells of UB^{Hdac1,2−/−} kidneys is likely to be a direct rather than secondary effect. As discussed below and shown in supplementary material Table S5, several genes in the Wnt pathway are regulated by Hdac1 and Hdac2 and are also directly bound by p53 in the developing kidney, linking HDAC-Wnt to p53.

At least two studies have shown that Hdac1 and Hdac2 are responsible for deacetylation of p53 (Trp53) in epidermal progenitor cells (LeBoeuf et al., 2010) and oocytes (Ma et al., 2012). Acetylation is a key determinant of p53 function by increasing protein stability, DNA binding affinity and transcriptional activity (Brooks and Gu, 2011; Meek and Anderson, 2009). Previous studies in our laboratory demonstrated that p53 is a key renal development regulator that controls cell proliferation, differentiation and apoptosis pathways (Hilliard et al., 2011, 2014; Saifudeen et al., 2002, 2009). Gain-of-function experiments showed that excessive p53 levels mediated by deletion of Mdm2 in the UB or MM led to bilateral renal dysplasia, which could be rescued by concurrent removal of p53 (Hilliard et al., 2011, 2014). We therefore asked whether Hdac1 and Hdac2 modulate p53 acetylation as well as its function in UB cells. Immunofluorescence revealed marked increases in the levels of p53 acetylated at lysine 370, 379 and 383 (corresponding to lysine 373, 382 and 386 in human, respectively) in the ureteric epithelium of UB^{Hdac1,2−/−} kidneys at E13.5 (Fig. 9A-D). As anticipated, we also observed an accumulation of total p53 protein in the ureteric epithelium (Fig. 9E,F).

Considering the observed upregulation of p53 in Hdac1/Hdac2-deficient UB cells, we hypothesized that p53 mediates the renal defects observed in UB^{Hdac1,2−/−} mice, and employed a genetic rescue approach that involved crossing the Hoxb7-Cre^tg/+; Hdac1^flox/flox; Hdac2^flox/flox mice to p53^−/− mice. Gross and histological analyses of the kidney at P0 revealed that removal of p53 on the UB^{Hdac1,2−/−} background partially rescued kidney development in a gene dosage-dependent manner (Fig. 10). Remarkably, Mendelian proportions of UB^{Hdac1,2−/−}; p53^+/− and UB^{Hdac1,2−/−}; p53^−/− mice were retrieved at P30, indicating that elimination of one allele of p53 is sufficient to rescue the early postnatal lethality of UB^{Hdac1,2−/−} mice (Table 3). Thus, although a proportion of UB^{Hdac1,2−/−}; p53^+/– and UB^{Hdac1,2−/−}; p53^–/– mice exhibited no overt histological rescue, a measurable improvement in renal survival must occur in these mice that allows their survival for at least 30 days. Remarkably, two of the 11 UB^{Hdac1,2−/−}; p53^+/− mice were still alive and fertile at 13 months of age and their kidneys exhibited relatively normal histology (supplementary material Fig. S3A-D). These results suggest that hyperacetylation of p53 upon loss of Hdac1 and Hdac2 is one of the important mediators of the congenital renal dysgenesis observed in UB^{Hdac1,2−/−} mice.

To begin to identify the potential targets of p53 leading to the renal dysgenesis, we cross-referenced our UB^{Hdac1,2−/−} microarray data with p53 ChIP-seq in E15.5 mouse kidneys (Li et al., 2013). Among the 496 altered genes in E13.5 UB^{Hdac1,2−/−} kidneys, 121 are bound by p53 in their promoter regions (supplementary material Table S3). The Hdac1/2-regulated and p53-bound genes include Lef1, Axin2, Tcf7 (Wnt signaling), Ptch1 (Shh signaling), several pro-apoptosis and cell cycle regulatory genes, kinesin family members (important for ciliary function), calbindin and keratin genes. IPA revealed that the top five molecular and cellular functions regulated by these genes are: (1) cell death and survival; (2) molecular transport; (3) cellular growth and proliferation; (4) cellular development and (5) cell cycle (supplementary material Table S4); and the top five affected canonical pathways are: (1) Vdr/Rxr activation; (2) neuroprotective role of Thop1 in Alzheimer's disease; (3) Wnt/β-catenin signaling; (4) protein kinase A signaling and (5) sonic hedgehog signaling (supplementary material Table S5 and Fig. S4). Hypergeometric distribution analysis demonstrated that cell death and survival genes are highly over-represented (51 of 121 genes, P=0.00037).

Based on the above analyses, we further examined how deletion of p53 affected UB cell survival in UB^{Hdac1,2−/−} mice. Immunostaining results demonstrated that UB^{Hdac1,2−/−}; p53^–/– mice had a similar increase in γH2AX-positive UB cells, whereas three of four UB^{Hdac1,2−/−}; p53^–/– kidneys showed fewer aCasp3-positive UB cells, when compared with UB^{Hdac1,2−/−} kidneys at E14.5 (4.73±0.47% UB^{Hdac1,2−/−} versus 2.73±0.35% UB^{Hdac1,2−/−}; p53^–/– apoptotic ureteric cells, P=0.0249) (Fig. 10C,D).

To test more directly the cell-autonomous role of p53 hyperacetylation upon loss of HDAC activities, we treated HCT116 p53 (TP53)^+/+ and p53^−/− isogenic human colon cancer cells with the class I-specific HDAC inhibitor MS-275, which preferentially inhibits HDAC1, 2 and 3. Light microscopy analysis revealed that MS-275 induces notable growth arrest and cell death in both p53^+/+ and p53^−/− cells, although p53^−/− cells are clearly more resistant to MS-275 than p53^+/+ cells (supplementary material Fig. S5A).

Next, we quantified the status of the cell cycle and of apoptosis using flow cytometry. We found that the anti-proliferative effects of MS-275 are not dependent on p53 function in HCT116 cells. MS-275 induced G1 and G2 phase arrest in both cell lines, with a decrease in S-phase cells to similar degrees (supplementary material Fig. S5B,D and Table S6), consistent with a previous report that p53 is dispensable for cell cycle arrest in mouse embryonic fibroblasts deficient for Hdac1 and Hdac2 (Wilting et al., 2010). By contrast, p53^–/– cells underwent significantly less apoptosis than p53^+/+ cells (3.52% versus 6.18%, respectively) after a 48-h treatment with 10 µM MS-275 (supplementary material Fig. S5C,E), indicating that MS-275 induces apoptosis partially through p53 in HCT116 cells. Western blot analyses confirmed that MS-275 effectively increased the level of acetylated histone H3 and p53 in p53^+/+ cells and the level of acetylated histone H3 in p53^–/– cells (supplementary material Fig. S5F). Moreover, the total level of p53 is also increased in p53^+/+ cells (supplementary material Fig. S5F). Cleavage of poly (ADP-ribose) polymerase 1 (PARP1) by caspases is a hallmark of apoptosis; it promotes apoptosis by preventing DNA repair under normal conditions or stress. When we compared the HCT116 cell lines, we identified a strong induction of cleaved PARP1 in p53^+/+ cells but a much weaker induction in p53^–/– cells (supplementary material Fig. S5F). These findings are consistent with a recent report from Sonnemann et al. (2014,), showing that HDAC inhibition in HCT116 cells is partially dependent on p53.

This study investigated the functions of Hdac1 and Hdac2 in the ureteric epithelium during kidney development. Our results demonstrate that: (1) Hdac1 and Hdac2 perform redundant yet essential functions in UB branching morphogenesis; (2) Hdac1 and Hdac2 are required for the canonical Wnt signaling pathway in kidney development; and (3) Hdac1 and Hdac2 are required to suppress hyperacetylation of p53 in the ureteric epithelium, which partially accounts for the UB^{Hdac1,2−/−} phenotype.

Kidney development is dependent on reciprocal epithelial-mesenchymal interactions between the UB and the MM, which drive iterative branching of the UB and subsequent induction of nephrons. Our previous studies showed that pharmacological inhibition of class I and II HDACs by Scriptaid, or of Hdac1, 2 and 3 by MS-275, causes disruptions of the reciprocal signaling pathways, leading to growth arrest and apoptosis. As many HDAC isoforms (e.g. Hdac1, 2 and 3) are expressed in the UB and MM, it was important to determine the distinct functions of individual enzymes in specific cell lineages. In this study, we set out to elucidate the developmental roles of Hdac1 and Hdac2 in the ureteric epithelium using a conditional knockout strategy. Although global deletion of either Hdac1 or Hdac2 leads to a lethal phenotype (Lagger et al., 2002; Montgomery et al., 2007), demonstrating the unique roles of these two enzymes in early embryogenesis, redundancy of Hdac1 and Hdac2 function has been found in many somatic cell types (e.g. oligodendrocytes, Schwann cells, cardiomyocytes, basal cells and adipocytes) (Haberland et al., 2010; Jacob et al., 2011; LeBoeuf et al., 2010; Montgomery et al., 2007; Ye et al., 2009). In line with the previous studies, our results reveal that one allele of either Hdac1 or Hdac2 is sufficient to support normal kidney development, whereas loss of all four alleles leads to renal hypodysplasia and early postnatal lethality.

Genome-wide profiling revealed that changes in gene expression are highly specific in UB^{Hdac1,2−/−} mice at E13.5. Only 226 transcripts (0.55%) were upregulated and 270 transcripts (0.66%) were downregulated (using a cut-off of 1.4-fold, P<0.05). This is consistent with the increasing evidence showing that Hdac1 and Hdac2 selectively control specific gene expression programs in different tissues. For example, deletion of Hdac1 and Hdac2 in the heart resulted in dysregulation of only ∼1.6% of the transcriptome (Montgomery et al., 2007). Given the apparent cell apoptosis and growth arrest observed, it is not surprising that genes regulating cell death and survival and cellular growth and proliferation are among the most notably dysregulated functional categories. More interestingly, we found that the canonical Wnt signaling pathway is among the most affected pathways in UB^{Hdac1,2−/−} kidneys.

The canonical Wnt signaling pathway controls both UB branching morphogenesis and nephrogenesis. Specifically, Wnt7b is expressed in the UB trunk and is essential for the establishment of a cortico-medullary axis during later stages of kidney development. In Wnt7b mutant mice, cortical epithelial development is normal but the medullary zone fails to form (Yu et al., 2009). Wnt9b, another member of the Wnt family secreted by the UB trunk epithelium, plays an indispensable role in the activation of Wnt4, which is both necessary and sufficient for the induction of renal vesicles from renal progenitor cells. Wnt9b is also important for the proliferation/renewal of the renal progenitor cells. Genetic deletion of Wnt9b in mice results in failure of nephron induction and premature exhaustion of the progenitor pool (Carroll et al., 2005; Karner et al., 2011). Our results show significant decreases in Wnt7b, Wnt9b and Wnt4 expression in UB^{Hdac1,2−/−} kidneys at E13.5, associated with the downregulation of a number of Wnt target genes, such as Axin2, Tcf7 and Lef1. Consistent with these findings, immunofluorescence results demonstrate that membrane-associated β-catenin is greatly decreased in the UB cells of UB^{Hdac1,2−/−} kidneys. β-catenin exists in the cell in two pools: membrane associated and cytosolic. On the membrane, as a component of adherens junctions (AJs), β-catenin links cadherins to the cytoskeleton, whereas in the cytoplasm β-catenin is an essential mediator of the Wnt signaling pathway (Perez-Moreno and Fuchs, 2006). Although the interplay between the two pools remains to be fully elucidated, time-lapse microscopy using photoactivatable GFP-tagged β-catenin has revealed that the membrane-associated pool of β-catenin can be internalized together with E-cadherin, accumulates at the perinuclear endocytic recycling compartment (ERC) upon AJ dissociation, and can be translocated into the nucleus upon Wnt pathway activation (Kam and Quaranta, 2009). Therefore, it is conceivable that decreased levels of membrane-associated β-catenin reduce the availability of cytosolic β-catenin and thus lead to the decrease in Wnt signaling in UB^{Hdac1,2−/−} kidneys. Moreover, we observed that a 6-h treatment of cultured embryonic kidney with HDAC inhibitor leads to a prompt reduction of Axin2, a direct target of Wnt, supporting the idea that Hdac1 and Hdac2 function in the activation of Wnt signaling in the embryonic kidney.

It is also of note that Shh, a hedgehog homolog, is one of the most suppressed genes in UB^{Hdac1,2−/−} kidneys. During kidney development, Shh is initially expressed in the distal part of the UB, and then in the urothelium. Shh acts as a paracrine signal to promote mesenchymal cell proliferation, and regulates the pattern of mesenchymal differentiation. Conditional deletion of Shh in the UB results in renal hypoplasia, hydronephrosis and hydroureter in newborn pups (Yu et al., 2002). However, such a phenotype is not observed in UB^{Hdac1,2−/−} mice. It is likely to be masked by the more severe dysplastic phenotype of UB^{Hdac1,2−/−} kidneys, or the residual level of Shh is still sufficient to support mesenchymal cell proliferation and differentiation. Taken together, our results demonstrate that Hdac1 and Hdac2 play a crucial role in maintaining essential ureteric genes for renal growth and differentiation.

It should be noted that, although it might be difficult to completely dissociate the direct effects of HDAC inactivation from the secondary effects of cell loss due to apoptosis, there are several clues that support the notion that many of the gene expression changes are specific. First, ISH and microarrays were performed at E13.5, prior to the observed degeneration of the UB tree. Several genes, such as Tcf7, Wnt7b, Wnt9b, Axin2 and Shh, were specifically altered. Second, if the effects of Hdac1/2 inactivation on gene expression were generalized or non-specific due to cell death, we would have expected a much more substantial change in the number of genes altered, as opposed to the few percent of differentially expressed genes observed in the study. Third, our results cannot be fully explained by excessive cell death of UB tip cells since the expression of cRet, Wnt11, Bmyc and Etv4, four specific markers of UB tips, is unaltered in UB^{Hdac1,2−/−} mice.

Although lysine acetylation was originally discovered in histones, these are not the only proteins that can be acetylated. p53 was the first non-histone protein found to be regulated by acetylation (Gu and Roeder, 1997; Luo et al., 2000). The acetylation of multiple lysine residues in the C-terminus (K305, K370, K372, K373, K381, K382 and K386) and DNA-binding domain (K164) of human p53 is significantly enhanced in response to stress and required for its stabilization, transcriptional activation and transcription-independent function in apoptosis (Brooks and Gu, 2011). p53 in which all eight lysine residues are substituted (8KR) loses the ability to mediate cell cycle arrest and apoptosis (Tang et al., 2008). There is also evidence that HDAC1 is recruited to p53 by an MDM2-containing protein complex, and HDAC1-mediated deacetylation of p53 is required for degradation of p53 (Ito et al., 2002).

Consistent with two recent studies showing that double knockout of Hdac1 and Hdac2 enhances p53 acetylation in mice epidermal progenitor cells and oocytes, our results demonstrated that loss of Hdac1 and Hdac2 in the ureteric epithelium leads to marked hyperacetylation and accumulation of p53. However, to our knowledge, the present study is the first to show rescue of organogenesis, albeit partial, upon concomitant germline deletion of p53 in vivo. This finding is in line with our previous report that excessive p53 accumulation in the UB or renal progenitor cells (mediated by deletion of Mdm2) causes bilateral renal dysplasia, while concomitant elimination of p53 rescues the renal phenotype and animal survival. With regards to the underlying cellular mechanisms, our immunostaining results demonstrate that p53 is partially responsible for UB cell apoptosis in UB^{Hdac1,2−/−} kidneys at E14.5. In support of these results, bioinformatics analysis revealed that 121/496 dysregulated genes in UB^{Hdac1,2−/−} kidneys are bound by p53 in their promoter regions and are thereby potentially regulated by p53. Moreover, over 40% of these genes are highly enriched in the category of cell death and survival. As to the apparent paradox of longer survival in the triple UB^{Hdac1,2−/−}; p53^−/− null mutants yet no significant morphological rescue, we suggest that ‘histological’ and ‘functional’ rescues might not be identical in this setting. Loss of p53 might have improved functional but not structural variables, e.g. glomerular filtration or tubular transport functions. This might account for the overall longer survival of UB^{Hdac1,2−/−}; p53^+/− or UB^{Hdac1,2−/−}; p53^−/− mice, which did not show structural rescue.

Together, these results suggest that Hdac1 and Hdac2 are required to suppress the hyperacetylation and activation of p53 in the UB lineage, thus protecting cells from apoptosis in renal development. Further studies using p53 lysine-specific mutant mice are needed to address the question of whether the pro-apoptotic function of p53 is dependent on its acetylation during kidney development. In summary, our data reveal that Hdac1 and Hdac2 play an essential role in cell proliferation, survival and transcriptional regulation in the ureteric epithelium during kidney development, and indicate that this occurs, in part, through p53-mediated pathways.

Mice bearing conditional null alleles of Hdac1 and Hdac2 (Hdac1^flox/flox and Hdac2^flox/flox) were obtained from the laboratory of Eric Olson (Montgomery et al., 2007) and were crossed to Hoxb7-CreEGFP transgenic mice (Hoxb7-Cre^tg/+) (Zhao et al., 2004) to delete the Hdac1 and Hdac2 genes, singly or in combination, specifically in the UB epithelium. For the rescue experiment, p53^–/– mice (Jackson Laboratory) were first crossed to the Hdac1^flox/flox; Hdac2^flox/flox mice, and then further crossed to Hoxb7-CreEGFP transgenic mice to remove p53 on the UB^{Hdac1,2−/−} background.

Kidneys were fixed in 10% buffered formalin, embedded in paraffin, and sectioned at 4 μm. Histology analyses were performed by standard periodic acid-Schiff (PAS) staining and Hematoxylin and Eosin (H&E) staining. Immunofluorescence was performed as previously described (Chen et al., 2011). Primary antibodies and working concentrations are listed in supplementary material Table S7. The peroxidase-based Vectastain ABC Elite Kit (Vector Laboratories) was used for DAB detection.

Embryonic kidneys were aseptically micro-dissected from timed-pregnant mice and cultured on polycarbonate Transwell filters (0.4 μm pore size, Corning Costar) over medium [DMEM/F-12 containing 10% fetal bovine serum (FBS)] at 37°C and 5% CO₂.

Kidneys were harvested at E13.5 and stored in RNAlater RNA stabilization reagent (Qiagen). Wild-type and UB^{Hdac1,2−/−} kidneys were each divided into four random pools (n=4). Total RNA was then isolated using the RNeasy Mini Kit (Qiagen).

Microarray analysis was performed according to established protocols (Schanstra et al., 2007). Briefly, fluorescently labeled cRNA was generated from 0.5 μg total RNA in each reaction using the Agilent Fluorescent Direct Label Kit and 1.0 mM Cyanine 3′- or 5′-labeled dCTP (PerkinElmer). Hybridization was performed using the Oligonucleotide Microarray Hybridization and In Situ Hybridization Plus Kit (Agilent). The labeled cRNA was hybridized to Agilent 44K whole mouse genome oligonucleotide microarray (containing ∼41,000 probes) as previously described (Schanstra et al., 2007). The arrays were scanned using a dual-laser DNA microarray scanner (Agilent). The data were then extracted from images using Feature Extraction software 6.1 (Agilent). Microarray data are available at GEO under accession number GSE35432.

MultiExperiment Viewer v4.9 software was used to generate lists of genes differentially expressed between wild-type and UB^{Hdac1,2−/−} kidneys, using P≤0.05 and a minimum 1.4-fold change in gene expression. Genes were classified according to their function using IPA software and BiNGO classification systems as previously described (Chen et al., 2011).

Quantitative real-time PCR (qPCR) was performed using the One-Step Brilliant Quantitative RT-PCR Master Mix Kit (Applied Biosystems). Real-time PCR reaction mix contained 200 nm forward primer, 200 nm reverse primer, and 60 ng total RNA. Relative levels of mRNA were normalized to Gapdh. The primers used for qPCR are listed in supplementary material Table S8.

ISH was performed using digoxigenin-labeled antisense probes on kidney tissue fixed with 4% paraformaldehyde as previously described (Chen et al., 2011).

HCT116 p53^+/+ and p53^–/– cells were obtained from Dr Hua Lu (Tulane University, New Orleans, LA, USA, and Johns Hopkins University Cell Center, Baltimore, MD, USA). Cells were grown in high-glucose DMEM with stable glutamine supplemented with 10% FBS and 10 mg/ml antibiotics (penicillin and streptomycin), under 5% CO₂ and saturated moisture. Cells were treated with vehicle control DMSO or 10 µM MS-275 for 24, 48 or 72 h.

Cells were treated with vehicle control DMSO or 10 µM MS-275 for 48 h. At the time of harvesting, cells were digested with 0.05% trypsin and resuspended in phosphate-buffered saline (PBS). For cell cycle analysis, 2×10⁵ cells were stained with 50 µg/ml propidium iodide (PI) using the Coulter DNA Prep Reagents Kit (Beckman Coulter). For evaluation of the early stages of apoptosis, 2×10⁵ cells were stained with Annexin V-FITC and PI using the Annexin V-FITC Apoptosis Detection Kit (Sigma). Cells were then analyzed using a Beckman Coulter FC500 flow cytometer and CXP software (Beckman Coulter) for acquisition and analysis. The cell cycle distribution was further analyzed by ModFit LT 3.0 software (Verity Software House).

Whole-cell proteins were extracted using RIPA buffer (Invitrogen) and nuclear proteins were extracted using the Nuclear Extract Kit (Active Motif). Primary antibodies and working concentrations are listed in supplementary material Table S7.

We thank Dr Eric Olson for the Hdac1^flox/flox and Hdac2^flox/flox mice; Dr Carlton Bates for the Hoxb7-CreEGFP transgenic mice; and Dr Hua Lu for providing HCT116 cell lines. We acknowledge the support of the Tulane Renal and Hypertension Center of Excellence, Center for Stem Cell Research & Regenerative Medicine and the Center for Gene Therapy.