The epidermal differentiation-associated Grainyhead gene Get1/Grhl3 also regulates urothelial differentiation

Abstract

Skin and bladder epithelia form effective permeability barriers through the activation of distinct differentiation gene programs. Using a genome-wide gene-expression study, we identified transcriptional regulators whose expression correlates highly with that of differentiation markers in both the bladder and skin, including the Grainyhead factor Get1/Grhl3, which is already known to be important for epidermal barrier formation. In the bladder, Get1 is most highly expressed in the differentiated umbrella cells and its mutation in mice leads to a defective bladder epithelial barrier formation due to the failure of apical membrane specialization. Genes encoding components of the specialized urothelial membrane, the uroplakins, were downregulated in Get1^−/− mice. At least one of these genes, uroplakin II, is a direct target of Get1. The urothelial-specific activation of the uroplakin II gene is due to selective binding of Get1 to the uroplakin II promoter in urothelial cells, which is most likely regulated by histone modifications. These results show a crucial role for Get1 in urothelial differentiation and barrier formation.

Introduction

One of the main functions of the urothelium is to form an effective permeability barrier, preventing the access of microorganisms and selectively controlling the movement of water and solutes between urine and the tissue. To achieve this function, urothelial cells undergo progressive differentiation steps from embryonic day (E) 12 to 18. The urothelium is considered to represent one of the tightest and most impermeable barriers in the body (Lewis and de Moura, 1982; Negrete et al, 1996). The mouse urinary bladder, which initially develops from the cloaca at E12, is fully formed at E14, when it can be isolated as a discrete unit (Staack et al, 2003). At E15, the endoderm-derived urothelium is a three-layered structure, composed of basal, intermediate, and superficial layers; basal cells are thought to progressively move towards the surface, differentiating first into intermediate cells and then into superficial cells. When fully differentiated, the superficial cells are referred to as umbrella cells. At E15, the apical membrane of the superficial cells is characterized by microvilli. As these cells differentiate, they markedly enlarge and the apical-surface microvilli are replaced by rigid ridges. At E18, a distinctive pattern of ridges is formed on the apical surface of umbrella cells, interspersed with the asymmetric unit membrane (AUM) plaques, which occupy most of the surface (Jezernik et al, 2000; Erman et al, 2006). The major protein components of the AUM plaques are the uroplakins (Upk), an evolutionarily conserved family (Garcia-Espana et al, 2006) of at least five integral membrane proteins: UpkIa, UpkIb, UpkII, UpkIIIa, and UpkIIIb (Wu et al, 1994; Sun et al, 1996; Deng et al, 2002; Min et al, 2006). Disruption of the UpkII and UpkIIIa genes in mice leads to impaired umbrella-cell development and compromised urothelial permeability barrier function (Hu et al, 2000, 2002; Kong et al, 2004).

Although the morphological and biochemical features of urothelial differentiation and barrier formation are well characterized, the transcriptional mechanisms that coordinate this process have just begun to be identified. In bladder tissue derived from mouse embryonic stem cells, Foxa1 and Foxa2 have been suggested as potential transcription factors involved in differentiation of urothelium (Oottamasathien et al, 2007). In an in vitro differentiation model of human urothelial cells, FOXA1 and IRF-1 have been identified as PPARγ-induced intermediary transcription factors necessary for urothelial differentiation (Varley et al, 2009). In recent years, aspects of the transcriptional mechanisms underlying this process in skin have been elucidated. Among the transcription factors important for terminal differentiation and barrier formation in epidermis are Klf4 (Segre et al, 1999) and Grainyhead transcription factor Get1/Grhl3 (Ting et al, 2005; Yu et al, 2006), which indirectly or directly activate many genes responsible for epidermal barrier formation (Dai and Segre, 2004; Koster and Roop, 2007). Grainyhead transcription factors have an evolutionarily conserved role in the regulation of epidermial barrier formation. Drosophila Grainyhead is required for normal cuticle strength and wound healing (Bray and Kafatos, 1991; Mace et al, 2005). In addition to regulating the mammalian epidermis differentiation program, Get1 has a role in keratinocyte migration, which is a crucial aspect of the wound-healing mechanism (Hislop et al, 2008; Yu et al, 2008).

We reasoned that a genome-wide characterization of the bladder differentiation program, and a comparison with the well-characterized skin-differentiation program, might provide further insights into this problem. Similar to the urothelium, one of the primary roles of the stratified epidermis is to form an effective permeability barrier. But in contrast to the urothelium, a different gene program is activated to mediate differentiation of the epidermis; the epidermal barrier depends primarily on protein components of the cornified envelope and the composition of lipids between cells of the cornified layer (Segre, 2003). Common to both epithelia, cell–cell adhesion, in part mediated by tight-junction proteins, is important for effective barriers (Acharya et al, 2004; Niessen, 2007). In this paper, we define and compare the gene-expression programs in the developing skin and bladder, and identify Get1 as an important regulator of urothelial terminal differentiation and barrier formation.

Results

Identification of a shared transcriptional regulatory program in bladder and skin development

To gain insights into the common and unique transcriptional regulatory programs during bladder and skin differentiation, we profiled global gene expression in whole mouse bladder and dorsal skin at E14.5, E16.5, and E18.5. At E18.5, a total of 1540 transcriptional regulators are expressed in both the bladder and skin, whereas there are 264 and 532 transcriptional regulators uniquely expressed in the bladder and skin, respectively (Figure 1A). To enrich for transcriptional regulators important for developmental regulation, we narrowed these lists down to those that show significant change in expression during these developmental time points. Over these developmental time points, there are 114 transcriptional regulators that are differentially expressed in both the bladder and skin, whereas there are 29 and 425 transcriptional regulators differentially expressed uniquely in the bladder and skin, respectively. To further enrich for transcriptional regulators likely to be important for terminal differentiation, we correlated the differentially expressed transcriptional regulators with the expression of differentiation markers in the epithelia of the bladder and skin. Uroplakin genes, which encode terminal differentiation products of the bladder epithelium, show a gradual increase over the time course of bladder development (Figure 1B and Supplementary Figure S1A and B). In contrast, many genes that encode components of the epidermal barrier, including loricrin, filaggrin, involucrin, periplakin, and envoplakin, show a marked increase from E14.5 to E16.5, with a slight decrease at E18.5 during skin development (Figure 1B and Supplementary Figure S1A). There are 15 transcriptional regulators whose expression correlates highly with the expression of differentiation markers in both tissues (Figure 1C and Supplementary Figure S1C), while there are 17 (Figure 1A and D, and Supplementary Figure S1C) and 322 (Figure 1A and Supplementary Table S1) transcriptional regulators whose expression correlates uniquely with differentiation markers in the bladder and skin, respectively. The transcriptional regulators that show specific correlation with differentiation markers in each tissue are candidates for playing important roles in the differentiation process in the bladder (Figure 1D) and skin (Supplementary Table S1). Consistent with this idea, Foxna1, Skn1a, and Dlx3, which have been implicated in epidermal differentiation (Brissette et al, 1996; Andersen et al, 1997; Morasso et al, 1999; Baxter and Brissette, 2002), are all found among the factors in Supplementary Table S1. Interestingly, Get1, a known regulator of epidermal terminal differentiation (Ting et al, 2005; Yu et al, 2006), is among the transcriptional regulators showing expression highly correlated to differentiation markers in both bladder and skin epithelium (Figure 1C and Supplementary Figure S1C). We were also intrigued to find that E14.5 dorsal skin has a more similar global gene-expression profile to the bladder samples from all time points than to E16.5 and E18.5 dorsal skin (Supplementary Figure S2A and B). The similarity of E14.5 dorsal skin to the bladder is partially due to the high expression of simple epithelial keratins (e.g. Krt8, Krt18, and Krt19) and low expression of late epidermal differentiation genes (e.g. small proline rich-like 3, desmoplakin, and corneodesmosin) in E14.5 skin (Supplementary Figure S2C). E16.5 and E18.5 dorsal skin are also molecularly distinct because they undergo marked gene-expression changes during hair-follicle morphogenesis.

Figure 1.

Shared transcriptional regulatory programs of bladder and skin differentiation. (A) Global analysis of common and unique transcriptional regulators regulated during bladder and skin differentiation at E14.5, E16.5, and E18.5. TR, transcription regulator. (B) The expression patterns of terminal differentiation markers of bladder epithelium and dorsal skin at the indicated embryonic age. (C) Heat map of transcriptional regulators with expression highly correlated to terminal differentiation markers of the bladder and dorsal skin. Note the Grhl3 (Get1) in bold. (D) Heat map of transcriptional regulators that are expressed in the bladder but not in dorsal skin, and that are highly correlated to bladder terminal differentiation markers. Note that Foxq1 has multiple probe sets. For panels C and D, expression levels are normalized across time points and indicated by the colorimetric ratio scale.

To gain insights into the biological processes during bladder development, we clustered the time-course profiles of the set of genes (4101 probe sets) that show significant change in expression during the time course. Functional annotation analysis identified significantly enriched Gene Ontology biological process categories within these clusters (Supplementary Figure S3A and B). We found that the expression of differentiation genes increased over the time course, whereas genes encoding cell-cycle regulators and cellular metabolic processes are highly enriched in clusters, showing decreasing expression during bladder development. These findings are consistent with decreasing epithelial proliferation as differentiation progresses. Interestingly, these same clusters contain a large number of genes encoding chromatin-modifying factors and other transcriptional regulators, suggesting more active chromatin modifications in early bladder development, perhaps related to much higher proliferation rate during this time period.

Get1 is highly expressed in umbrella cells of the mouse urothelium

Grainyhead transcription factors are regulators of epidermal barrier function and wound healing in species as divergent as Drosophila and mice (Mace et al, 2005; Ting et al, 2005). In mice, one of three Grainyhead genes, Get1, has an important role in regulation of epidermal terminal differentiation. The striking correlation of Get1 expression with bladder epithelial differentiation markers (Figure 1C and Supplementary Figure S1C) suggests a role in bladder epithelial differentiation as well. This finding is of interest because the bladder epithelium is endodermally derived and is characterized by a differentiation program that is strikingly divergent from that of the ectodermally derived epidermis. Get1 transcripts are highly expressed in bladder epithelium at E15.5 when the differentiation program is initiated (Kudryavtseva et al, 2003; Auden et al, 2006). Get1 protein is specifically and strongly expressed in the developing (Figure 2A) and mature (Figure 2B) umbrella cells of the bladder epithelium at E16.5 and E18.5. This staining is specific because Get1^−/− bladder epithelium shows no staining (Figure 2C). In addition, Get1 is strongly expressed in intermediate and superficial layers of the ureter epithelium (data not shown). The high expression of Get1 in the most-differentiated cells of the bladder epithelium suggests a possible role in the terminal differentiation program and barrier formation in the bladder.

Figure 2.

Get1 is required for bladder epithelium umbrella-cell development. (A–C) WT mouse bladder epithelial sections from E16.5 (A) and E18.5 (B) were immunostained with Get1 antibody, whereas Get1 immunostaining in the E18.5 Get1^−/− bladder was used as negative control (C). Arrowheads point to umbrella cells. Get1 is specifically expressed in the developing and mature umbrella cells of the urothelium. (D, E) Urinary bladder (Bl) and kidneys (K) at E18.5 from WT (D) and Get1^−/− mice (E). The Get1^−/− urinary bladder is enlarged, whereas the kidneys appear normal. (F–I) Hematoxylin–eosin staining of the bladder epithelium from WT (F, H) and Get1^−/− (G, I) mice at E16.5 (F, G) and E18.5 (H, I). Arrowheads point to the umbrella cells in WT mice at E16.5 and E18.5. Arrows point to the surface-layer cells in Get1^−/− urothelium at E18.5. The dashed lines indicate the border between the epithelium and mesenchyme. Umbrella cells are absent from the Get1^−/− bladder epithelium. Scale bar: (A–C) and (F–I), 12.5 μM.

Get1 is required for formation of normal umbrella cells in the bladder urothelium

To investigate the potential role of Get1 in terminal differentiation of the urothelium, we used Get1^−/− mice previously made in our laboratory (Yu et al, 2006). Get1^−/− mice, which die at birth, have neural tube defects, an open eye phenotype, and an epidermal barrier defect (Ting et al, 2003, 2005; Yu et al, 2006, 2008). In addition, Get1^−/− mice have enlarged urinary bladders (Figure 2D and E); this phenotype is fully penetrant at E18.5. To determine the role of Get1 in urothelial terminal differentiation, we examined the urinary bladder histology of WT and Get1^−/− mice at E16.5 and E18.5, the time period during which the apical cell membranes of umbrella cells form the characteristic rigid ridges (Erman et al, 2006). At E16.5, the most superficial layer of the urothelium is composed of differentiating umbrella cells with a rugged apical surface (Figure 2F). In contrast, in the Get1^−/− bladder, the superficial epithelial cells are more round and have a smooth surface (Figure 2G). At E18.5, the superficial umbrella cells are fully formed with the characteristic rugged apical plasma membrane in WT mice (Figure 2H), whereas the Get1^−/− urothelium contains smaller cells with a smooth apical cell surface (Figure 2I), suggesting that Get1 is required for the formation of normal umbrella cells.

To further determine the effect of Get1 on umbrella cell maturation in the bladder, we examined the ultrastructure of umbrella cells from E18.5 WT and Get1^−/− mice. Using scanning electron microscopy, we found that superficial cells of the WT urothelium are about 50 μm in diameter and are demarcated by a tight-junctional ring (Figure 3A). Their surface appears pleated with the characteristic ridges (Figure 3B) as described previously (Erman et al, 2006). In contrast, superficial cells of the Get1^−/− urothelium are much smaller, less than 10 μm in diameter (Figure 3C), and the apical surface is covered by small microvilli rather than ridges (Figure 3D). This appearance is characteristic for WT superficial epithelium, earlier in development at E15.5 (Erman et al, 2006). As described previously using transmission electron microscopy, the apical membrane of WT apical umbrella cells is scalloped in appearance and the cells contain abundant fusiform cytoplasmic vesicles (Figure 3E), which have an elongated form and are thought to be responsible for delivering uroplakins to the surface. In contrast, the superficial cells of the Get1^−/− urothelium are round, with a much smoother apical membrane containing microvilli. In addition, the cytoplasmic vesicles, which have a discoid rather than a fusiform appearance (Figure 3F), are less abundant (Figure 3G) in Get1^−/− superficial cells. Together, these results show that the Get1 gene is critical for normal umbrella-cell formation in the bladder epithelium (Figure 3H).

Figure 3.

Get1 is required for apical cell-membrane specialization in the bladder epithelium. (A–D) Scanning electron microscopy of the bladder epithelial surface of WT (A, B) and Get1^−/− (C, D) mice at E18.5. Red arrows mark the outline of superficial cells. Yellow arrowheads point to hinge regions on the plicated surface of WT mice. Yellow arrows point to microvilli in the Get1^−/− urothelium. The superficial cells of the Get1^−/− urothelium are much smaller than in the WT, and their surface contains microvilli rather than ridges. (E, F) Transmission electron microscopy of WT (E) and Get1^−/− (F) bladder epithelium. Arrowheads in (E) point to fusiform vesicles. Arrowheads in (F) point to discoidal vesicles. Arrows in (F) point to microvilli. The Get1^−/− urothelium lacks urothelial apical plaques, and uroplakin-containing fusiform vesicles are much less prominent than in the WT. (G) Quantification of vesicle number in bladder superficial cells. The Y-axis shows the vesicle number per μm². The results represent the mean and s.e.m. for three WT and three Get1^−/− embryos. (H) Model for the role of Get1 in urothelial cell differentiation.

Get1 is required for permeability barrier formation and terminal differentiation in bladder epithelium

Terminal differentiation of umbrella cells is critical for the establishment of the urinary bladder barrier. Therefore, to investigate the effect of Get1 on bladder permeability barrier function, we tested the ability of the bladder epithelium to exclude methylene blue as described previously (Hu et al, 2000). Compared with WT bladders, methylene blue easily penetrated into the E18.5 Get1^−/− bladder wall (Figure 4A–D), suggesting that in Get1^−/− bladders, permeability barrier function is impaired. This finding is consistent with the impaired formation of umbrella cells. Impaired umbrella-cell differentiation was further supported by analysis of keratin (K) expression (Erman et al, 2006). At E16.5, K18 is normally expressed in the subapical region of the developing umbrella cells of the bladder epithelium (Figure 4E). In contrast, K18 is not detected in Get1^−/− bladder epithelium (Figure 4F). K20 is another urothelial differentiation marker that is expressed in umbrella cells at a later stage (Figure 4G). In contrast, K20 is strikingly downregulated in Get1^−/− urothelium at E18.5 (Figure 4H). At both E16.5 and E18.5, K6 is expressed at a low level in the basal and intermediate layers, but not in the umbrella-cell layer of the WT urothelium (Figure 4I and K). In contrast, K6, a marker for wounded epithelia, is markedly upregulated throughout the Get1^−/− uroepithelium at both stages (Figure 4J and L). As K6 expression is sometimes associated with hyperproliferation, we evaluated mitotic activity in the bladder epithelium. The mitotic cell marker phospho-histone H3 is unchanged in Get1^−/− urothelium, suggesting that Get1 does not affect urothelial cell proliferation (Figure 4M–O). Together, these data indicate that Get1 selectively regulates the urothelial terminal differentiation program.

Figure 4.

Get1 is required for effective permeability-barrier formation and terminal differentiation in mouse bladder epithelium. (A, B) Gross pictures of methylene-blue penetration assay of WT (A) and Get1^−/− (B) urinary bladders. (C, D) Post-sections of WT (C) and Get1^−/− (D) urinary bladders after methylene-blue assay. Arrows in (C) and (D) point to the bladder epithelium. The methylene-blue dye infiltrates the bladder wall of Get1^−/− mice, indicating impaired permeability barrier. (E–N) Immunostaining showing expression of K18 (E, F), K20 (G, H), K6 (I–L) and phospho-histone 3 (M, N) in WT (E, G, I, K, M) and Get1^−/− (F, H, J, L, N) bladder epithelium at the indicated stages. Arrows in (I) and (K) point to normal umbrella cells. Expression of the umbrella-cell markers K18 and K20 is decreased, indicating impaired differentiation. The expression of K6 is markedly upregulated and extends to the surface in Get1^−/− bladder epithelium, where it is normally excluded. (O) Quantification of phospho-histone-3-positive cells in WT and Get1^−/− bladder epithelium. The number of phospho-histone-3-positive cells is normal, indicating that the cell proliferation is unchanged in the Get1^−/− bladder epithelium. Scale bar: (C, D) 50 μM; (E–N) 12.5 μM.

Deletion of the Get1 gene affects multiple genes in the urothelial terminal differentiation program

To gain insights into the molecular mechanisms of bladder defects in Get1^−/− mice, we carried out a microarray analysis of RNA from the bladders of three Get1^−/− and three Get1^+/− E18.5 mice. Statistical differential-expression analysis identified 622 differentially expressed probe sets (P<0.00138, false discovery rate within 10%) in the Get1^−/− mice (Supplementary Figure S4). We then filtered this list based on at least 1.5-fold up- or downregulation, which resulted in 226 probe sets. Out of these 226 probe sets, 66 probe sets were regulated during the normal bladder differentiation program as described previously (Figure 1 and Supplementary Figure S3): 20 downregulated, 46 upregulated. Intriguingly, the downregulated genes in Get1^−/− bladders (Figure 5A) are all in clusters 1–3 (increasing expression during wild-type bladder differentiation; Supplementary Figure S3), suggesting that Get1 is likely to have a function in activating the expression of a battery of urothelial terminal differentiation genes. The uroplakins are assembled into heterodimers that form planar crystals, and these crystals eventually form the AUM plaques, which are crucial components in bladder barrier function (Sun et al, 1996; Sun, 2006). Transcripts of all members of the uroplakin family are significantly downregulated in Get1^−/− bladders (Figure 5C and Supplementary Table S2). Among these genes, UpkII is strikingly downregulated 8.5-fold. Another markedly downregulated gene is Snx31, which is involved in endosomal recycling (Cullen, 2008) and could possibly have a function in plaque formation. In contrast, the upregulated genes are distributed among all the clusters (Figure 5B and C), and include genes associated with injury response, such as keratin 6a.

Figure 5.

Identification of significantly differentially expressed genes in E18.5 Get1^−/− bladder. (A) Downregulated and (B) upregulated bladder differentiation-related genes. Expression levels from microarrays are normalized across time points and indicated by the colorimetric ratio scale. The fold and P-value of significant differential expression, as well as the cluster membership of the WT time course, for each gene are indicated. Note that UpkII (in bold) is significantly downregulated, and that several genes have multiple probe sets. (C) Quantitative RT–PCR validation of selective genes that are differentially regulated in Get1^−/− E18.5 bladders. Expression for each gene is normalized to 1 for WT. Results represent the mean and s.e.m. for three replicates.

Cell adhesion is a common mechanism contributing to barrier formation both in the epidermis and in the bladder epithelium. The deletion of Get1 leads to defective cell–cell adhesion in the epidermis because of the downregulation of several cell-adhesion molecules (Yu et al, 2006). Interestingly, tight-junction structures are less prominent and there is abnormal separation of superficial cells in the Get1^−/− bladder (Figure 6A and B). Some tight-junction transcripts, such as Cldn5, Cldn6, Cldn8, Cldn9 and Cldn11, are downregulated in the Get1^−/− bladder epithelium, whereas others, such as Cldn2, Cldn3, and Cldn7, are upregulated (Figure 6C and Supplementary Figure S5). Transcripts for Cldn1, Cldn4, Cldn23, and Ocln, which are downregulated in Get1^−/− epidermis, are not altered in the Get1^−/− bladder. Therefore, Get1 seems to regulate different cell–cell adhesion molecules in the bladder epithelium and epidermis.

Figure 6.

Adhesion abnormalities in superficial urothelial cells in the Get1^−/− bladders. (A, B) Transmission electron microscopy images of WT (A) and Get1 mutant (B) bladders at E18.5. The white arrows in panels A and B point to tight junctions. The red arrowheads in panel A point to normal cell–cell adhesions. The red arrowheads in panel B point to separated cell–cell adhesions. There are abundant fusiform vesicles in panel A, whereas only a few discoid vesicles are observed in panel B. Vesicles are indicated by the stars. Scale bar: 0.5 μM. The results are based on three WT and three Get1^−/− E18.5 embryos. (C) The statistically significant differentially expressed cell-adhesion genes in Get1^−/− bladder and skin. NC indicates no significant change; NS indicates not ‘present' in the microarray data set.

The uroplakin II gene is a direct target of Get1 in bladder epithelium

All major uroplakin (UpkIa, UpkIb, UpkII, UpkIIIa, and UpkIIIb) transcripts were downregulated in Get1^−/− urothelium at both E16.5 and E18.5 stages (Figures 5, 7A and B). Similarly, at least three major uroplakin proteins (UpkIb, UpkII, and UpkIIIa) that are preferentially expressed in umbrella cells were downregulated in the Get1^−/− bladder epithelium at both E16.5 and E18.5 (Figure 7C–N). These results are consistent with a model whereby Get1 promotes the differentiation of the Upk-expressing umbrella cells, and suggests the possibility that Get1 might directly activate uroplakin genes.

Figure 7.

Get1 is required for normal uroplakin expression in the bladder epithelium. (A, B) Semi-quantitative qRT–PCR analysis of UpkIa, UpkIb, UpkII, UpkIIIa, and UpkIIIb in WT and Get1^−/− bladder epithelium at E16.5 (A) and E18.5 (B). Expression of all uroplakin transcripts is decreased in the Get1^−/− bladders. (C–N) Immunostaining showing expression of UpkIa (C–F), UpkII (G–J), and UpkIIIa (K–N) in WT (C, E, G, I, K, M) and Get1^−/− (D, F, H, J, L, N) bladder epithelium at the indicated stages. The bladder epithelial expression of UpkIa, UpkII, and UpkIIIa proteins is decreased in Get1^−/− mice, based on four WT and four Get1^−/− mice. Scale bar: (C–N) 12.5 μM.

To start investigating whether Get1 is a direct regulator of uroplakin genes, we used MatInspector (Cartharius et al, 2005) and BEARR (Batch Extraction and Analysis of cis-Regulatory Regions) (Vega et al, 2004) to identify high-affinity Get1 DNA-binding sites in a genomic region from 50 kb upstream to 5 kb downstream of the coding region of each of the uroplakin genes. Several high-affinity sites were identified in all five uroplakin genes, and some of these sites were conserved between human and mouse (Supplementary Figure S6A). One of these conserved Get1-binding sites resided in an evolutionarily conserved region in the UpkII gene, located in the region from −105 to −112 bp upstream of the transcription start site of the human UpkII gene (Figure 8A). This binding site is functional because the expression of an UpkII reporter plasmid in RT4 human bladder cells, which express endogenous Get1, is highly dependent on this site (Figure 8B). Furthermore, in chromatin immunoprecipitation (ChIP) experiments, Get1 binds to this region of the UpkII promoter in RT4 cells. This binding is specific as Get1 does not bind to another region downstream of the UpkII promoter that lacks high-affinity Get1 DNA-binding sites (Figure 8C). Together, these experiments suggest that Get1 directly regulates the expression of the UpkII gene in bladder epithelial cells.

Figure 8.

Get1 directly regulates Uroplakin II expression and is dependent on histone modifications. (A) Identification of an evolutionarily conserved Get1 DNA-binding site (DBS) in the UpkII promoter. Location of the binding site upstream of the transcriptional start site is indicated for both human and mouse UpkII genes. (B) WT and mutated −2800 UpkII luciferase-reporter constructs were transfected into RT4 bladder cells. Mutation of the Get1-binding site resulted in decreased UpkII promoter activity. (C) Chromatin immunoprecipitation (ChIP) assays carried out on RT4 cells using a Get1 antibody or IgG as a control, followed by PCR amplification of a region of the human UpkII promoter containing the Get1 site indicated in (A, upper panel) and a non-specific downstream (DS) region of the UpkII gene as a negative control (lower panel). Input DNA is unprecipitated chromatin from the IgG sample. Get1 binds specifically to the promoter region of the UpkII gene. (D, E) Quantitative real-time PCR of Get1 (D) and UpkII (E) transcripts in undifferentiated (NHEK-U) and differentiated (NHEK-D) normal human epidermal keratinocytes, and RT4 cells. Standard error of the mean and mean fold differences in expression were calculated using three replicates normalized to RPLPO. (F) Get1 and UpkII protein expression in RT4 cells and NHEK-D detected by immunofluorescence staining. Although Get1 protein is expressed in both cell types, UpkII is selectively expressed in RT4 cells. Nuclear DAPI staining is red. (G) ChIP assays carried out on NHEK-D and RT4 cells using antibodies against RNA polymerase II, Get1, H3K9Ac, and H3K27me3, followed by PCR amplification of a 193-bp region of the human UpkII promoter containing the Get1 site in (A). Get1 and RNA polymerase-II binding, and histone modifications were quantified using quantitative PCR (upper) and also analyzed by direct inspection on agarose gels (lower). Black bars indicate binding in RT4 cells and gray bars indicate binding in NHEK-D cells. Get1 binding to the UpkII promoter is associated with increased RNA polymerase II binding and histone modifications associated with active gene expression. TSS indicates transcription start site. (H) Quantitative real-time PCR of UpkII in NHEK-D cells with and without TSA treatment. TSA treatment increased the expression of UpkII in human keratinocytes.

Epigenetic mechanisms for the cell-specific activation of the uroplakin II gene

Although the Get1 gene is expressed in both the epidermis and bladder epithelia, the UpkII gene is selectively expressed in the bladder epithelium. This suggests that other mechanisms may have a function in the selective activation of UpkII in the bladder epithelium. To address this question, we used normal human epidermal keratinocytes (NHEK) and the bladder epithelial cell line RT4. Like the epidermis, differentiated NHEK cells express Get1 but not UpkII, whereas RT4 cells express Get1 and UpkII (Figure 8D–F), as does the bladder. Using ChIP assays, we found that RNA polymerase binding and Get1 binding to the UpkII promoter is more prominent in RT4 cells than in NHEK (Figure 8G). Furthermore, H3K9 acetylation, a marker of open chromatin, is increased on the UpkII promoter in RT4 cells, whereas H3K27 trimethylation, a marker of transcriptional repression, is more prominent in the NHEK cells (Figure 8G). Consistent with a role for histone deacetylation in repression of the UpkII gene in NHEK cells, treatment of these cells with the histone-deacetylation complex (HDAC) inhibitor trichostatin A (TSA) resulted in a 30-fold increase in the expression of UpkII in NHEK cells (Figure 8H). Together, these experiments suggest that the ability of Get1 to selectively activate the UpkII gene in uroepithelial cells is regulated at the level of DNA binding, which is most likely controlled by histone modifications and chromatin structure.

Discussion

In this paper, we carried out a genome-wide time-course gene-expression study of skin and bladder development to identify unique and common transcription programs. The Grainyhead transcription factor Get1 is among a small subset of transcription factors showing expression that correlates with differentiation-marker expression in both the skin and bladder. Gene deletion shows that Get1 is critical for apical-membrane specialization of the urothelium and terminal differentiation of umbrella cells, as well as for barrier formation. A broad range of genes associated with urothelial differentiation are affected by the Get1 deletion. Among the genes that were strikingly downregulated in Get1^−/− mice are the prominent components of the specialized membrane, the uroplakins. The UpkII gene was identified as a direct target of Get1. The target-gene specificity in bladder epithelial cells is due to selective binding of Get1 to the UpkII promoter, which is mostly probably regulated by histone modification. These results suggest that Get1 has analogous roles in regulation of terminal differentiation and barrier formation in different epithelia in the mouse by regulating distinct downstream targets.

Unique and shared transcriptional regulatory mechanism in bladder and skin development

The terminal differentiation programs in bladder and skin epithelium are distinct. In the mouse, the epidermis starts stratifying at E14.5, ultimately producing morphologically distinct layers, which are clearly identified at E18.5. The basal layer contains proliferating cells that undergo successive differentiation steps through the spinous, granular, and cornified layers as they move towards the surface (Candi et al, 2005). The cornified layer is composed of dead cells that contain a cellular envelope linked to a cytoskeleton of significant strength. Thus, in the epidermis, multiple cytoplasmic structural proteins, many encoded by the epidermal differentiation complex, as well as lipid-producing enzymes are critical for complete differentiation and barrier formation (Dai and Segre, 2004; Candi et al, 2005; Koster and Roop, 2007). In contrast, in the bladder epithelium, the uroplakin membrane proteins are critical for barrier formation (Hu et al, 2000; Kong et al, 2004). Furthermore, after the initial formation, cell renewal is much slower in the bladder epithelium than in the epidermis (Sun, 2006). In contrast to these differences, the barriers of the two epithelia have cell–cell adhesion molecules in common, in particular tight-junction complex mechanisms (Acharya et al, 2004; Niessen, 2007).

Consistent with these distinct differentiation programs, we found that 6.4 and 60.6% of the transcriptional regulators unique to the bladder and skin, respectively, were highly correlated with the expression of differentiation genes in each organ, whereas this was true for less than 1% of the transcriptional regulators common to the bladder and skin (Figure 1A). These data suggest that transcriptional programs specific for each organ are important for terminal differentiation. At the same time, our data also provide evidence for shared transcriptional mechanisms between these divergent epithelia, and Get1 is among the small number of transcriptional regulators showing expression that correlates with the differentiation programs in both organs (Figure 1C). Although previous work has established the role of Get1 in the regulation of epidermal differentiation and barrier formation in diverse species (Bray and Kafatos, 1991; Wilanowski et al, 2002; Venkatesan et al, 2003; Mace et al, 2005; Ting et al, 2005; Yu et al, 2006), this study shows that Get1 has a similar role in barrier formation in an internal epithelium. Interestingly, Get1 is expressed in other internal epithelia, including the gastrointestinal track (Kudryavtseva et al, 2003; Auden et al, 2006). In Get1^−/− mice, the forestomach shows evidence of impaired differentiation and the intestine is dilated with blood in the lumen (Supplementary Figure S7A–D) (Yu et al, 2006), suggesting the possibility that Get1 might have a more general role in maintaining internal epithelial integrity.

Get1 regulates uroplakin gene expression, umbrella-cell differentiation, and barrier formation

The most striking urothelial abnormality of the Get1^−/− mice is defective formation of umbrella cells, suggesting that Get1 is critical for the transition from intermediate to umbrella cells (Figure 3H). Decreased expression in the Get1^−/− bladder of multiple genes that normally show urothelial differentiation-linked expression is also consistent with this conclusion (Figure 5). Thus, Grainyhead-like transcription factor Get1 is critical for the promotion of terminal differentiation of the bladder epithelium.

Among the differentially expressed urothelial differentiation genes, decreased uroplakin expression (Figure 7) is likely to be critical for this defect in Get1^−/− mice. Uroplakins are the major subunits of the AUM plaques in the apical plasma membrane, which are generated after the formation of uroplakin heterodimers (UpkIa–II and UpkIb–III) in the endoplasmic reticulum and planar crystals in the post-Golgi compartment (Sun, 2006). Our results show that all uroplakins are moderately downregulated in Get1^−/− urothelium at the mRNA level and strikingly at the protein level. Low expression of uroplakins leads to failure of plaque formation and loss of the typical rugged apical surface (Kong et al, 2004). In the Get1^−/− urothelium, the normal plicated surface is replaced by numerous microvilli at the apical surface (Figure 3C and D). Therefore, the failure of AUM plaque assembly is probably the main cause of defective urinary bladder barrier formation in Get1^−/− mice. Consistent with this idea, deletion of UpkII or UpkIIIa resulted in defective bladder barrier formation because AUM plaques were not well assembled in the umbrella-cell apical membrane (Hu et al, 2000; Kong et al, 2004). Both UpkII^−/− and UpkIIIa^−/− urothelium showed a similar phenotype as the Get1^−/− urothelium. The superficial cells in UpkII^−/− mice are smaller and devoid of rigid plaques at the apical membrane (Kong et al, 2004) secondary to the failed assembly of UpkIa/II heterodimers. The UpkIIIa^−/− urothelium also had smaller superficial cells lacking mature fusiform vesicles, and the apical surface was covered with microvilli because of the assembly failure of the UpkIb–III heterodimer (Hu et al, 2000). The phenotypic similarity between Get1^−/− on the one hand and UpkII^−/− and UpkIIIa^−/− urothelium on the other supports the idea that Get1 affects barrier formation by regulating uroplakins and AUM assembly. However, Get1 regulates additional terminal differentiation genes and these are likely to contribute to the phenotype as well.

Previous studies using in vitro differentiation systems have indicated that the nuclear receptor PPARγ is important for bladder epithelial differentiation. Activation of PPARγ induces the expression of UpkII and UPKIa and regulates the components of tight junction in bladder epithelial cells (Varley et al, 2004, 2006). Recent studies indicate that PPARγ acts upstream of transcription factors, FOXA1 and IRF-1, which may directly regulate uroplakin gene expression (Varley et al, 2009). Consistent with this idea, PPARγ shows increasing expression from E14.5 to E18.5 during bladder development (Figure 2). The effect of PPARγ gene deletion on bladder development has not been studied. In the epidermis, PPARγ is expressed at a relatively low level (Dahten et al, 2007), and PPARγ^−/− mice did not show any abnormality (Mao-Qiang et al, 2004). Therefore, although PPARγ might be important for normal bladder epithelial differentiation, it seems to act differently from Get1.

The cell-specific regulation of Get1 on UpkII is dependent on histone modification

Our data indicate that Get1 transcriptionally regulates the expression of UpkII gene by directly binding to a high-affinity binding site in the UpkII promoter (Figure 8). Get1 might directly regulate other uroplakin genes as well, because several Get1-binding sites are found around all uroplakin genes (Supplementary Figure S6A and B). In addition to UpkII, the UpkIb and UpkIIIa genes contain conserved sites between human and mouse that bind Get1 in RT4 urothelial cells (Supplementary Figure S6C). Although the bladder epithelia of Get1 and UpkII knockout mice show overlapping differentiation phenotypes, only the UpkII knockout mice show upregulation of other uroplakin genes, presumably because of compensatory mechanisms (Kong et al, 2004). In contrast, in Get1^−/− mice, all uroplakins are downregulated, which further supports the primary role of Get1 in uroplakin gene regulation.

As UpkII is not expressed in the differentiated epidermis, we considered the reasons for selective actions of Get1 on the UpkII promoter in bladder epithelial cells. One possible model is that UpkII activation depends on a combinatorial action of Get1 with other transcription factors that are selectively expressed in bladder epithelium. Under this model, Get1 binds to the UpkII promoter in epidermal cells but is unable to activate the gene because the accessory factor is missing. An alternative model is that the UpkII gene is modified by chromatin alterations in the epidermis, thus preventing binding of Get1 in these cells. We tested these models by studying the binding of Get1 to the UpkII promoter in urothelial cells and differentiated epidermal keratinocytes. We found that Get1 binds selectively to the UpkII promoter in the urothelial cells, and that this binding is associated with binding of RNA polymerase (Figure 8G). In bladder epithelial cells, this region of the UpkII promoter has an enrichment of histone H3K9 acetylation, which has been shown to correlate with actively transcribed genes (Pokholok et al, 2005). In contrast, in epidermal keratinocytes, the UpkII gene has a marked decrease of Get1 and RNA polymerase binding, as well as decreased histone H3K9 acetylation. Instead, there is increased histone H3K27 trimethylation, a well-studied chromatin modification associated with silent genes (Zhao et al, 2007). Furthermore, differentiated epidermal keratinocytes treated with TSA, a potent Class I and II HDAC inhibitor, showed an increase in UpkII expression (Figure 8H), supporting the idea that the presence of an H3K9-acetylation mark is correlated with expression of UpkII. Therefore, our results suggest that epigenetic mechanisms could allow Get1 to selectively activate different gene targets in a cell-specific manner (Figure 9).

Figure 9.

Model of UpkII regulation in the urothelium and epidermis. In the urothelium, the H3K9Ac mark is present and Get1 binds to the promoter and activates transcription of the UpkII gene. In contrast, in the epidermis, where the H3K9-acetylation mark is absent and an H3K27me3-repressive mark is enriched, Get1 does not bind as efficiently to the UpkII gene. We speculate that for some epidermal-specific genes (gene A) the reverse may true; Get1 activates selectively in epidermal cells. For yet other genes that are expressed in both the epidermis and urothelium (gene B), Get1 may bind and activate in both cell types.

In conclusion, we have shown that the Grainyhead-like transcription factor Get1 has a critical function in regulating terminal differentiation and barrier function of the bladder epithelium, in addition to its previously described role in epidermal differentiation. During evolution, distinct gene sets involved in barrier formation in at least two different epithelia have acquired Get1-binding sites, thus bringing these genes under control of Get1. Epigenetic mechanisms could allow selective regulation of the appropriate set of genes for each epithelium, and we speculate that such mechanisms might be widely used to allow cell-specific activity of a developmental transcription factor.

Materials and methods

Microarray analysis

Mouse backskin and bladder were dissected from WT and Get1^−/− mice at E14.5, E16.5, and E18.5. RNA was isolated and microarray experiments were carried out as described previously (Yu et al, 2006). Both tissues contain both epithelial and mesenchymal components. The GC-robust multichip average method was used for background correction of the microarray data. Statistically differentially expressed transcriptional regulators during bladder and skin differentiation were identified by one-way ANOVA (P<0.01, false discovery rate within 10%) of the respective time-course profiles. Expression profiles of representative terminal differentiation markers of the bladder (uroplakins) and skin (claudin 4, plakophilin 3, periplakin, and envoplakin) were used to calculate Pearson product moment correlation coefficients. Correlation coefficients of at least 0.8 were used to define transcriptional regulators with expression highly correlated to terminal differentiation markers. Global similarity in gene expression between the dorsal skin and bladder at E14.5, E16.5, and E18.5 was determined by averaging the Pearson product moment correlation coefficients across all probe sets for each pair of sample types (tissue/stage). For example, the dorsal skin at E14.5 and bladder at E14.5 has an average correlation coefficient of 0.92 across all probe sets. Statistically differentially expressed genes (4101 probe sets) during bladder differentiation were identified by one-way ANOVA across time points (P<0.01, false discovery rate within 10%). After averaging the gene expression values across replicates for each time point, the time-course profiles were normalized to mean 0 and s.d. 1 for k-means clustering using Euclidean distance. Enrichment of functional categories (Gene Ontology Biological Process) within each cluster was found using the Functional Annotation Tool in DAVID 2008 (Dennis et al, 2003). As there are many redundant categories, only representative categories are shown in Figure 2. Transcriptional regulators were defined as probe sets with annotations of Gene Ontology Biological Process ‘regulation of transcription, DNA-dependent' (GO ID 0006355) and Gene Ontology Cellular Component ‘nucleus' (GO ID 0005634). Chromatin modifiers were defined as probe sets with annotations of Gene Ontology Biological Process ‘establishment and/or maintenance of chromatin architecture' (GO ID 0006325). In order to account for the gene expression variability due to different scan dates in the Get1 wild-type and knockout microarrays, we removed the scan date batch effect using Partek Genomics Suite. Statistically differentially expressed genes (622 probe sets) between wild type and knockout were identified by one-way ANOVA across time points (P<0.00138, false discovery rate within 10%). This list was then filtered based on at least 1.5-fold up- or downregulation, resulting in 226 highly and significantly differentially expressed probe sets. The intersection between these genes and the normal bladder differentiation-regulated genes is 66 probe sets. Note that the microarray data have been submitted to http://www.ncbi.nlm.nih.gov/geo.

Histology, immunostaining, and in situ hybridization

Embryonic mouse urinary bladders were fixed in 10% formalin and paraffin embedded. Following sectioning, 5-μm sections were stained with hematoxylin and eosin. For immunohistochemistry staining, antigen retrieval was carried out by heating slides to 95°C for 10 min in 0.01-M citrate buffer (pH6) in a microwave oven. The sections were then immunostained by the ABC peroxidase method (Vector) using diaminobenzidine as the enzyme substrate and hematoxylin as a counterstain. We used the following primary antibodies: rabbit anti-mouse Get1 (Yu et al, 2008) and ARP33196 (Aviva Systems Biology); mouse monoclonal anti-human keratin 18 (Dako); mouse monoclonal anti-human cytokeratin 20 (Dako); rabbit anti-mouse keratin 6 (Covance); rabbit anti-phospho-histone H3 (Cell Signaling Technology); rabbit anti-mouse UpkIa (Santa Cruz Biotechnology); rabbit anti-mouse UpkII (Santa Cruz Biotechnology); and rabbit anti-mouse UpkIIIa (Santa Cruz Biotechnology). All the staining results were based on studies on at least three WT and three KO mice.

Scanning and transmission electron microscopy

For scanning electron microscopy, five WT and four Get1^−/− bladders were dissected from E18.5 embryos and fixed in 10% formalin overnight. After a PBS wash, tissues were postfixed in 1% osmium for 1 h and dehydrated through graded-ethanol series and HMDS. After gold coating, tissues were observed with a scanning electron microscope as described previously. For transmission electron microscopy, three WT and three Get1^−/− bladders were dissected from E18.5 embryos and fixed by immersion in 4% paraformaldehyde and 1% glutaraldehye in 0.1 M PBS (pH 7.4) at room temperature, followed by a PBS wash. The tissues were then postfixed with 0.2% osmium and dehydrated through graded-ethanol series, and embedded in Agar 100 resin. Ultrathin sections were contrasted with uranyl acetate and lead citrate, and examined on a transmission electron microscope as described previously (Yu et al, 2008).

Methylene-blue penetration assay

We used a modification of a previously described method by Hu et al (2000). Four WT and four Get1^−/− bladders were dissected from E18.5 embryos. A total of 20 μl of 0.1% methylene blue (Sigma–Aldrich) in 0.9% NaCl was injected to the bladder from the uretery side. The dye was removed 20 min later, and the bladders were extensively washed five times with 0.9% NaCl, after which the bladders were photographed by a Nikon Coolpix 995 camera. The post-sectioning was based on four WT and three Get1^−/−mice.

RT–PCR analysis

Semi-quantitative RT–PCR was carried out with cDNA generated using the High Capacity cDNA Archive Kit (Applied Biosystems), using total RNA prepared from the urinary bladder of WT and Get-1^−/− embryos using Trizol Reagent (Invitrogen). Reactions were sampled after 25, 28, and 30 cycles at different PCR conditions to monitor product accumulation. We used the following primers: UpkIa forward: TCGGTACATGATCCTCACG; UpkIa reverse: GTCCGACGACAGCACAGTGG; UpkIb forward: GCCTCTTCTGCTTGTCCGTTC; UpkIb reverse: CATCGTTATTCTCCACCCG; UpkII forward: CCACACTGCCTGTCCAGACCT; UpkII reverse: GGTTTGTCACCTGATATGCG; UpkIIIa forward: TATGGTCAGATCCCATCCG; UpkIIIa reverse: CTTGCTGGAGAACACCTCTGC; UpkIIIb forward: AGATCCCGACCTCCCACAG; UpkIIIb reverse: AAGTTATGACGATCATACAGCCG; Get1 forward: GAATTACAAGTCTGTGCCACCA; and Get1 reverse: ATTTGCTGACCTTTTTCTGAGC. Quantitative RT–PCR was carried out using the following TaqMan Gene Expression Assays (Applied Biosystems): UpkII (Hs00171854_m1, Mm00447665_m1), Get 1(Grhl3, Hs00297962_m1, Mm01193339_m1), UpkIa (Mm01176593_m1), UpkIb (Mm00769504_m1), UpkIIIa (Mm00452321_m1), UpkIIIb (Mm00452321_m1), Nfix (Mm00477796_m1), Smarcd3 (Mm00491850_m1), Snx31 (Mm00508591_m1), Agr2 (Mm00507853_m1), Fmo5 (Mm00515805), Yeats4 (Mm00518302_m1), Hadc2 (Mm00515108_m1), Cldn2 (Mm00516703_s1), Cldn5 (Mm00727012_s1), Sprr1a (Mm01962902_s1), Reg3g (Mm00441127_m1), human RPLP0 and mouse GAPDH as endogenous controls.

Cell culture

RT4 cells were maintained in McCoy's 5A medium (Gibco) supplemented with 5% penicillin–streptomycin and 10% FBS. Neonatal human epidermal keratinocytes were purchased from Cascade and grown according to the manufacturer's instructions in EpiLife Medium (Cascade) supplemented with EDGS (Cascade) and 0.6 mM CaCl₂. Cells were grown for 5 days in a medium supplemented with a final concentration of 1.8 mM CaCl₂ to induce differentiation. For TSA experiments, NHEK cells were seeded into six-well plates, differentiated for 5 days, and then treated with 200 ng of TSA (Cell Signaling Technology). Cells were collected 40 h after treatment.

Site-directed mutagenesis and transient transfection reporter assays

A 2.8-kb mouse UpkII promoter sequence was cloned into the pGL3-basic luciferase reporter vector (Promega). The Get1 core-binding site was mutated to AATTCGTCT by site-directed mutagenesis, using the QuikChange^® Site-Directed Mutagenesis Kit (Strategene) as per the manufacturer's instructions. RT4 cells were seeded into six-well plates 1 day before transfection. Luciferase reporter constructs (1 μg) were transfected by Expressfect (Denville Sci. Inc.) as per the manufacturer's instructions. To control for transfection efficiency, transfections were normalized to Renilla luciferase vector as described previously (Yu et al, 2006). Luciferase activity was measured 1 day after transfection. All experiments were carried out at least three times, each time in triplicate.

Gel mobility shift assays

Get1 protein expression and gel mobility shift assays were carried out as described previously (Yu et al, 2006). The sequences of oligonucleotides for binding were: consensus: TCCTGTTAAACCGGTTTTTCTAGT; consensus mut: CAGTGAGCAGAGATCACGCCATTG; UpkIa (7.5^*): CTTGCACAAACCAGTTCTGGTCCC; UpkIb (7.9^*): CCAG GAGGAACAGGTTCATTTTCA; UpkIIIa (7.1^*): TCTTCCTAAACCGGGTGTGGTGGC; UpkIIIa (7.5^*): CAATGGCAAACCAGTTTTGCCCTG; UpkIIIb (7.5^*A): CTGCGTGAAACCAGTTACTGGTCT; UpkIIIb (7.5^*B): GAGGAGAGAACCAGTTCTGGCAAG; human UpkIb: ACCAGTTGAACTGGTTATACATTC; human UpkIIIa: GACAGACAAACCAGTTTACTGCTG.

Chromatin immunoprecipitation assays

For the chromatin immunoprecipitation (ChIP) assays, we followed previously described methods (Rabinovich et al, 2008) with the following modifications. We used the following antibodies for immunoprecipitations: RNA polymerase II (Covance Cat#MMS-126R), IgG (Sigma Cat#15006-10MG), H3K9Ac (AbCam Cat# ab4441), H3K27me3 (AbCam Cat# ab6002), and Get1/Grhl3 (Aviva Cat# ARP33196). Primers used for amplification of ChIP DNA were: RNApolII forward, AGATGAAACCGTTGTCCAAACT; RNApolII reverse, AGGTTACGGCAGTTTGTCTCTC; UpkII-33/-226 forward, GGGTG GCTGATTTTGTCCT; UpkII-33/-226 reverse, GTGATTTCAAACCCCACCT; UpkIIDS forward, TGGCCTCGTCTTTGTCTTTT; UpkIIDS reverse, CATGATGAAACCCCGTCTCT; UpkIb forward, GGGTGAGGCAGAACATCACT; UpkIb reverse, TGGAGGGTTCATATCACAAAAG; UpkIIIa forward, CTGTTCAGGGGACAGGAAAG; UpkIIIa reverse, CCCAGGAGAAGTCCTGTGTC; DHFR forward, CTGATGTCCAGGAGGAGAAAAGG; DHFR reverse, AGCCCGACAATGTCAAGGACTG.

Supplementary Material

Supplementary Figures

Supplementary Table S1

Supplementary Table S2

Review Process File