Prdm6 drives ductus arteriosus closure by promoting ductus arteriosus smooth muscle cell identity and contractility

Prdm6 interacts with MRTF-A. In an attempt to uncover novel mechanisms that regulate SMC-specific gene expression, we used co-immunoprecipitation/mass spectrometry–based methods to identify MRTF-A binding partners in mouse aortic SMC (AoSMC) lysates. The SMC-selective PR/SET domain-containing protein, Prdm6, was identified in washed MRTF-A immunoprecipitates, as were several known MRTF-A interacting proteins including p300 (12, 30) and importin 9, which facilitates MRTF-A entry into the nucleus (31). (See Supplemental Table 1; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.163454DS1.) Although the lack of a suitable PRDM6 Ab prevented us from examining the interaction between endogenous proteins, both flag-MRTF-A (Figure 1A) and endogenous MRTF-A (Figure 1B) were shown to co-immunoprecipitate with exogenously expressed myc-Prdm6. To further validate and characterize the MRTF-A–Prdm6 interaction, we performed co-immunoprecipitation experiments in COS-7 cells overexpressing a series of MRTF-A deletions. As shown in Figure 1C, myc-Prdm6 interacted with the central region of MRTF-A (aa 109–475) that contains the B1 basic, Q-rich, and SAP domains known to mediate SRF binding, and with an N-terminal fragment of MRTF-A (aa 1–108) that contains the actin-binding REPEL motifs. The MRTF-A transaction domain (aa 480–930) did not interact with Prdm6. Since the formation of large actin-containing binding complexes in cell lysates frequently complicates co-immunoprecipitation results, we used far Western analyses to further test whether the MRTF-A–Prdm6 interaction was direct. As shown in Supplemental Figure 1A, we detected direct binding of Prdm6 only to full-length MRTF-A and the aa 109–475 region (see arrows). In separate experiments, the association of Prdm6 with the N-terminal region of MRTF-A was inhibited by inclusion of cytochalasin D, suggesting that this interaction was actin dependent, and likely nonspecific (Supplemental Figure 1B).

Prdm6 is an SMC-selective MRTF-A binding protein that regulates SMC phenotyFigure 1

Prdm6 is an SMC-selective MRTF-A binding protein that regulates SMC phenotype. (A) COS-7 cells were transfected with flag-Prdm6 with/without myc-MRTF-A. Myc immunoprecipitates were separated on an SDS-PAGE gel, transferred to nitrocellulose, and then probed with anti-flag Ab. n = 2; representative blots shown. (B) 10T1/2 cells were transfected with myc-Prdm6 or empty vector. Myc and control IgG immunoprecipitates were run on SDS-PAGE and probed with an Ab against endogenous MRTF-A. Note that MRTF-A was only detected in IPs from lysates expressing myc-Prdm6 and immunoprecipitated with the anti-myc Ab. n = 2; representative blots shown. (C) COS-7 cells were transfected with myc-Prdm6 and the indicated flag-MRTF-A deletion fragment. Flag immunoprecipitates were run on an SDS-PAGE gel and probed with anti-flag (IP) or anti-myc (Co-IP) Abs. Nonspecific bands for IgG heavy and light chains are marked with arrowheads. n = 2; representative blots shown. (D) Genotype-Tissue Expression (GTEx) consortium data depicting normalized Prdm6 mRNA levels in the indicated human tissues. (E) Rat AoSMCs were treated with siRNAs targeting Prdm6 or GFP for 72 hours. Expression of the indicated genes was analyzed by quantitative reverse transcription PCR and normalized to GAPDH. n = 4; *P < 0.05. TPM, transcripts per million; SMA, SM α-actin.

Prdm6 was originally characterized by Davis et al. as a protein enriched in mouse SMCs (21). Interestingly, PRDM6 mRNA levels in humans were highest in blood vessels and other SMC-containing organs (see data from the Genotype-Tissue Expression [GTEx] consortium presented in Figure 1D), providing additional support for PRDM6’s role in SMCs. Although PRDM6 was shown to promote SMC phenotypic modulation, perhaps by interacting with HDACs and/or G9a (21), Gewies et al. detected no significant differences in SMC proliferation, SMC marker gene expression, or SMC investment of vessels in SM22Cre Prdm6fl/fl mouse embryos (22), suggesting that the effects of Prdm6 on SMC phenotype were context dependent. Indeed, PRDM proteins have been shown to interact with multiple chromatin-modifying enzymes and to mediate both transcriptional activation and repression depending upon cellular and tissue context (see ref. 32 for a review). Given that Prdm6 interacted with MRTF-A, a strong transactivator of SMC marker gene expression, we wanted to test whether Prdm6 depletion affected SMC differentiation marker gene expression in our cultured SMC models. As shown in Figure 1E, siRNA-mediated depletion of Prdm6 by approximately 90% in primary rat AoSMCs resulted in a 40% reduction in mRNA levels of several canonical SMC markers (P < 0.05). SMC differentiation marker protein levels were also reduced when primary mouse outflow tract SMCs isolated from Prdm6fl/fl animals were treated with Cre-expressing adenovirus (Supplemental Figure 2).

Prdm6 expression in neural crest–derived SMCs is required for DA closure. Because Prdm6 was shown to be expressed in outflow tract SMCs (21) and because coding mutations in Prdm6 were associated with patent DA in humans (23), we wanted to directly examine the role of Prdm6 in outflow tract development and DA function. We first used RNAscope-based methods to more closely monitor Prdm6 expression during outflow tract maturation starting just after separation of the aortic and pulmonary arteries at E13.5. As shown in Figure 2 and Supplemental Figure 3, using RNAscope probes to PRDM6 resulted in a strong and specific in situ signal in the SMCs of the aortic arch, pulmonary artery, and DA, from E13.5 until just before birth.

Prdm6 is highly expressed in outflow tract SMCs during development.Figure 2

Prdm6 is highly expressed in outflow tract SMCs during development. (A) Schematic of section orientation for DA and outflow tract analysis. (B) RNAscope-based in situ hybridization for Prdm6 shows expression from E13.5 to E17.5 in all outflow tract arteries including the ascending aorta (aAo), descending aorta (dAo), pulmonary artery (PA), and DA. At least 2 pups per time point were examined. Representative images are shown. (C) Costaining of Prdm6 and SM α-actin (SMA) at E18.5 verified that Prdm6 expression is specific to vascular SMCs. Scale bar = 200 μm.

To directly test whether Prdm6 was required for normal neural crest-derived SMC development and function, we bred Prdm6fl/fl mice to a Wnt1Cre2 line that expresses Cre in neural crest cells starting at approximately E10.5 (33). Wnt1Cre2 Prdm6fl/fl mice were born at Mendelian ratios, and as shown in Supplemental Figure 4, were indistinguishable from littermate controls, suggesting that Prdm6 depletion in neural crest cells did not result in embryonic lethality or have gross effects on overall development. In contrast, no Wnt1Cre2 Prdm6fl/fl mice survived until weaning even though 53 of the 266 offspring generated were expected to have this genotype. Based upon our observations during the perinatal period, Wnt1Cre2 Prdm6fl/fl mice died within 2 days of birth, a phenotype similar to that observed in the SM22Cre Prdm6fl/fl model (22) and one that is common in genetically modified mouse models of patent DA (34). We carefully examined DA patency in newly born Wnt1Cre2 Prdm6fl/fl mice (before lethality) by visual scoring of blood within the DA, by outflow tract casting, and by microscopic analysis of paraffin-embedded and frozen sections. As depicted in Figure 3, we observed a patent DA in all Wnt1Cre2 Prdm6fl/fl mice, but not in littermate control animals, which included the following genotypes: Wnt1Cre2 PRDM6wt/fl, Wnt1Cre2 PRDM6wt/wt, PRDM6fl/fl, PRDM6wt/fl, and PRDM6wt/wt. Taken together these data indicated that PRDM6 expression was absolutely required for DA closure, that Cre expression alone or deletion of 1 copy of PRDM6 had no effect on ductus closure, and that the floxed PRDM6 allele behaved in a manner similar to the WT allele.

Wnt1Cre2 Prdm6fl/fl mice exhibit patent DA.Figure 3

Wnt1Cre2 Prdm6fl/fl mice exhibit patent DA. (A) Intracardiac Microfil injection (top) and histologic examination (bottom) of DA closure in Wnt1Cre2 Prdm6fl/fl and littermate control mice at P1. The DA in these images is marked by a red arrowhead. At least 6 pups per group were examined. Representative images are shown. (B) Histological analysis of DA closure in E18.5 embryos isolated from dams treated for 4 hours with indomethacin (20 mg/kg). As marked by blue arrowheads, note that indomethacin treatment resulted in premature DA closure in utero in littermate control but not Wnt1Cre2 Prdm6fl/fl mice. Scale bar = 200 μm. At least 7 fetuses per group were analyzed. Representative images shown. dAo, descending aorta; PA, pulmonary artery.

Functional DA closure in mice is initiated within minutes after birth and is mediated by severe vessel constriction. Full anatomic closure takes place over a longer period and involves intimal thickening and eventual fibrotic remodeling of the DA into a remnant structure known as the ligamentum arteriosum (35). Based upon the timing of lethality in our model and the fact that we never observed even partial ductus closure in Wnt1Cre2 Prdm6fl/fl mice, we hypothesized that Prdm6 expression in SMC was required for the contractile pathways that mediate DA vasoconstriction. To begin to test this idea, we stimulated ductus closure in E18.5 embryos by treating pregnant dams for 4 hours with indomethacin (20 mg/kg), a nonsteroidal antiinflammatory drug used to treat patent DA in humans because it inhibits the production of prostaglandin E2 (PGE2), which helps maintain ductus patency by stimulating ductus SMC relaxation. While this procedure resulted in premature DA closure in all littermate controls, it had no effect on DA closure in Wnt1Cre2 Prdm6fl/fl fetuses (Figure 3B). We also subjected E18.5 newborns (delivered by cesarean section, C-section) to 100% oxygen, which is known to stimulate ductus SMC contraction by mechanisms not yet fully described (see refs. 36, 37 for reviews). As with indomethacin treatment, oxygen exposure for 1 hour did not stimulate DA closure in Wnt1Cre2 Prdm6fl/fl animals (data not shown). Importantly, we observed functional DA closure of WT mice within 30 minutes of the C-section procedure even under ambient air conditions.

Neural crest cell investment of the outflow tract is unaffected by Prdm6 depletion. In our model, the Wnt1 promoter drives Cre expression specifically in neural crest cells relatively early during development. Since cardiac neural crest cells delaminate from the developing neural tube and migrate into the pharyngeal arches, where they condense and eventually differentiate into the SMCs that layer the outflow tract and aortic arch arteries, including the DA, the defects in DA closure observed could have been due to failure of neural crest cells to correctly populate these vessels. To address this possibility, we crossed our Wnt1Cre2 Prdm6fl/fl mice to the ROSA26LacZ reporter strain, which allowed us to perform lineage tracing of WT and Prdm6-deficient neural crest cells during development. As shown in Figure 4A, X-gal staining of outflow tract and aortic arch arteries from Wnt1Cre2 Prdm6fl/fl ROSA26LacZ pups at P1 was consistent with normal neural crest cell migration and was not detectably different from littermate controls (Wnt1Cre2 ROSA26LacZ). To further examine the effects of Prdm6 depletion on outflow tract morphology and neural crest cell identity, we stained frozen sections from P1 pups with anti-LacZ and anti–SM α-actin Abs. As shown in Figure 4, B and C, LacZ and SM α-actin expression in the outflow tract and DA of Wnt1Cre2 Prdm6fl/fl ROSA26LacZ mice was similar to that of genetic littermate controls. Importantly, virtually all cells within the medial layers of the outflow tract vessels including the DA costained for LacZ and SM α-actin, and we observed no visible immunofluorescence in any sections with secondary Ab alone (see Supplemental Figure 5). Taken together these data suggest that Prdm6 depletion in Wnt1-expressing cells did not affect neural crest delamination or migration or have observable effects on the investment of the outflow tract arteries and DA with SM α-actin–expressing, neural crest cell–derived SMC.

Prdm6 depletion does not affect neural crest cell investment of the outflowFigure 4

Prdm6 depletion does not affect neural crest cell investment of the outflow tract arteries. (A) X-gal staining of outflow tract arteries in Wnt1Cre2 ROSA26LacZ and Wnt1Cre2 Prdm6fl/fl ROSA26LacZ pups at P1. At least 4 pups per group were examined. Representative images are shown. (B) Immunofluorescence staining for SM α-actin (SMA) and LacZ in P1 outflow tract sections from littermate control and neural crest–specific Prdm6-deficient mice. At least 4 pups per group were examined. Representative images are shown. Scale bar = 200 μm. (C) The boxed area in panel B is shown at higher magnification. SMA (green), LacZ (cyan), CD31 (red), and DAPI (blue). Scale bar = 200 μm. def, deficient.

Prdm6 depletion inhibits DA contractility. To better examine the functional consequences of Prdm6 deletion on outflow tract vascular function, DA and ascending aorta segments were isolated from Wnt1Cre2 Prdm6fl/fl and littermate controls at E18.5 and then cannulated for pressure myography as previously described (3841). In brief, vessel segments were exposed to stepwise increases in intralumenal pressure followed by various contractile stimuli while computer-assisted video microscopy was used to record vessel diameter (see Figure 5A). When compared with littermate controls, DA segments isolated from Wnt1Cre2 Prdm6fl/fl mice were wider in diameter under basal (5 mmHg) and working pressures (20 mmHg) and failed to exhibit a myogenic (i.e., distension-induced) contractile response (Figure 5, B and C). Ascending aorta diameter was similar between Wnt1Cre2 Prdm6fl/fl and littermate controls under these conditions, and neither control nor Prdm6-deficient ascending aorta segments exhibited a myogenic response (Figure 5D).

Prdm6 depletion reduces DA tone and contractile responses.Figure 5

Prdm6 depletion reduces DA tone and contractile responses. (A) Visualization of the DA vessel segment isolated for myography experiments. (B) Representative diameter tracings for littermate control and Prdm6-deficient DAs exposed to increasing hydrostatic pressure, KCl, oxygen, and the thromboxane (TxA2) agonist, U46619 (100 nM). Please see Methods for more details. Summary graphs for the indicated exposures in DA (C) and ascending aorta (D) vessel segments. Note that Prdm6 deficiency did not affect the function of ascending aorta segments. *P < 0.05 for pressure and concentration response curves compared with baseline values (ANOVA) or U46619 response (t test); †P < 0.05 between Wnt1Cre2 Prdm6fl/fl and control response curves (2-way ANOVA).

To test whether Prdm6 deficiency inhibited depolarization-dependent contractility, vessel segments were exposed to increasing concentrations of KCl that are known to activate voltage-dependent calcium channels. While 50 mM KCl resulted in an 80% decrease in the diameter of DA segments isolated from littermate controls, it had virtually no effect on DA segments isolated from Wnt1Cre2 Prdm6fl/fl mouse embryos (Figure 5C). Based upon extensive studies demonstrating that increasing oxygen concentrations promote DA constriction, we also tested whether Prdm6 deficiency affected this contractile mechanism. Time-limited exposures to increasing oxygen concentrations in the circulating buffer resulted in significant constriction of control DA segments as expected (Figure 5C), but this response was completely absent in Prdm6-depleted DA segments. Vessel segments (under oxygenated conditions) were also exposed to the thromboxane receptor agonist, U46619, which potently constricts most blood vessels through GPCR-coupled signaling pathways. Importantly, U46619 treatment of DA segments isolated from Prdm6-deficient mice resulted in a 40% reduction in vessel diameter, providing reassurance that our Prdm6-deleted vessel preparations were viable and could generate a significant contractile response under at least some conditions. However, as above, U46619-induced constriction of Prdm6-deficient DA segments was significantly less than that measured in DA segments isolated from littermate controls (–42.2% vs. –76.5%, P < 0.05). Of interest, Prdm6 depletion had no effect on U46619-induced contraction in ascending aorta segments or on the relatively modest constriction induced in aortic segments by KCl and oxygen exposure (Figure 5D). Taken together, these data indicate that DAs from Wnt1Cre2 Prdm6fl/fl mice have reduced tone and innate contractility and that these properties are similar to those measured in the ascending aorta at this gestational time point (E18.5).

Prdm6 expression is required for enrichment of a DA-selective gene program. To help deduce the mechanisms by which Prdm6 affected DA contractility, we performed bulk RNA-Seq analysis on DA tissue isolated from Wnt1Cre2 Prdm6fl/fl and littermate control mice at E18.5 (before DA closure). We also isolated mRNA from the ascending aorta to better define the differences between these 2 neural crest–derived SMC populations and to test whether Prdm6 had vessel-specific effects. Importantly, we sequenced at least 5 samples for each genotype and tissue, which increased the statistical power of subsequent comparisons. As summarized in the principal component analysis and hierarchical clustering in Figure 6, A and B, we saw excellent agreement in expression profiles between samples, allowing us to draw several important conclusions. First, even though DA and ascending AoSMCs originated from similar neural crest cell progenitors, we observed significant differences in gene expression patterns between these tissues. Interestingly, mRNA levels for SRF and nearly all the SRF-dependent SMC differentiation marker genes (SM MHC, SM α-actin, SM γ-actin, Cnn1, SM22, and MLCK/telokin) were significantly higher in DA samples, indicating that strong SMC identity may be critical for DA function.

Prdm6 depletion inhibits a ductus-selective gene program.Figure 6

Prdm6 depletion inhibits a ductus-selective gene program. (A) Principal component analysis of RNA-Seq data from ascending aorta and DA samples isolated from Wnt1Cre2 Prdm6fl/fl and littermate control mice at E18.5. (B) Two-dimensional hierarchical clustering of the top 1,000 genes differentially expressed in ascending aorta and DA samples from control and Prdm6-deficient mice. Potentially relevant Prdm6-dependent genes are listed on the right, and full gene lists for each cluster are available in Supplemental Table 2. (C) Volcano plot demonstrating differential gene expression between control and Prdm6-deficient DA samples. Genes colored in red are considered as significantly different (Padj < 0.01, logFC > 0.5, or logFC < –0.5). (D) Gene ontology analysis of genes significantly up- or downregulated by Prdm6 depletion in DA samples. logFC, log fold-change.

Second, as illustrated by clusters 3 and 4 in Figure 6B and the volcano plot in Figure 6C, the depletion of Prdm6 in neural crest cells resulted in a significant shift in DA gene expression patterns, especially in those genes that distinguish the DA from the ascending aorta. Of the 519 genes that were more highly expressed in the DA by at least 1.5-fold, 319 of those were significantly (P < 0.01) downregulated by Prdm6 depletion. Similarly, of the 399 genes that exhibited lower expression in the DA (less than 70% versus aorta), 228 of those were significantly upregulated by Prdm6 depletion. Gene Ontology analysis of the most significant differentially expressed genes between control and Wnt1Cre2 Prdm6fl/fl mice (Padj < 0.01 and logFC > 0.5 or logFC < –0.5) revealed that the SMC differentiation, SMC proliferation, and muscle contraction gene programs were downregulated in the Prdm6-deficient DA while those related to nerve function and extracellular matrix expression and organization were upregulated (Figure 6D). Third, Prdm6 depletion altered the expression level of many genes previously implicated in DA closure (see ref. 34 for a review), including the PGE2 receptor, EP4; the transcription factors Tfap2b, myocardin, Foxc1, and Hand2; the Notch signaling components, Jag1 and Notch3; SM MHC; integrin linked kinase; and fibulin1. Finally, when coupled with our demonstration that Prdm6 inhibited DA contractility, our RNA-Seq data expand the list of candidate genes that may regulate DA function to include those genes that are involved with voltage-dependent excitation contraction coupling and oxygen sensing.

To help support our RNA-Seq data, we used RNAscope approaches to monitor the expression of Tfapb2 and EP4 during outflow tract development in Wnt1Cre2 Prdm6fl/fl and littermate control mice. As shown in Figure 7 and Supplemental Figure 6, our RNAscope methods resulted in specific in situ signal that was fairly selective for ductus SMCs and was mostly abolished by Prdm6 depletion. Although the increased expression of Cnn1 protein in the DA was also reduced by PRDM6 depletion (Figure 7C), we had difficulty detecting differences in SM α-actin and SM MHC protein expression by immunofluorescence. Although we cannot rule out discrepancies between mRNA and protein levels, this was most likely due to the limits of immunofluorescence quantification when examining highly expressed proteins.

Characterization of EP4, Tfap2b, and CNN1 expression in littermate controlFigure 7

Characterization of EP4, Tfap2b, and CNN1 expression in littermate control and Prdm6-deficient neural crest–derived SMCs. RNAscope-based in situ hybridization of EP4 (A) and Tfap2b (B) expression in outflow tract arteries at the indicated developmental time points. At least 3 pups per group per time point were examined by these methods. Representative images are shown. (C) Immunofluorescence staining of CNN1 in outflow tract arteries at E15.5 and E17.5. At least 3 pups per group per time point were examined. Representative images are shown. Scale bar = 200 μm. For all panels note the Prdm6-dependent expression of each gene in the DA (white arrowheads).

Characterization of Prdm6 DNA binding by ChIP-Seq. Regardless of whether Prdm6 functions as a direct histone methyltransferase, data from the current and previous studies indicate the Prdm6 has substantial effects on gene expression, suggesting that it interacts with DNA directly (perhaps through its 4 Zn fingers) or is recruited as part of a larger DNA binding complex. To further identify the mechanisms by which Prdm6 controls gene expression, we performed ChIP-Seq experiments in outflow tract SMCs isolated from E17.5 mouse embryos to identify Prdm6 binding sites within the SMC genome. Since suitable Abs for Prdm6 are not available, we used lentivirus to express flag-Prdm6 in our cells. Our final data set included only ChIP peaks detected in 2 separate immunoprecipitations from the same sample. We observed nearly 14,000 Prdm6 ChIP peaks that were associated with approximately 5,200 genes. Of the approximately 1,500 genes that were differentially expressed in Prdm6-deficient ductus tissue, about half were associated with Prdm6 binding (Figure 8B). Although these data indicate that Prdm6 affects the transcription of these genes, it also suggests that Prdm6’s effects on gene expression are modified by additional transcription and/or chromatin signals. Importantly, Prdm6 binding was detected within or near several genes known to regulate ductus function, including EP4, endothelin-1, Jag1, Connexin 40, and the transcription factors Myocd, Foxc1, and Twist1 (Figure 8A). The Prdm6 gene itself contained 4 Prdm6 binding sites, perhaps suggesting feedback regulation of its activity. Somewhat surprisingly we did not observe many Prdm6 binding sites within the SMC differentiation marker genes, though 1 strong peak was present at the SRF-dependent enhancer within the SM α-actin first intron that we have previously characterized (42).

ChIP-Seq analysis of Prdm6 binding in mouse outflow tract SMCs.Figure 8

ChIP-Seq analysis of Prdm6 binding in mouse outflow tract SMCs. (A) Schematic of Prdm6 ChIP-Seq binding data for the indicated genes. (B) Venn diagram illustrating overlap of genes that bind Prdm6 (green circle) and those shown to be differentially expressed in ductus samples from littermate control and Wnt1Cre2 Prdm6fl/fl mice (red circle). (C) Characterization of Prdm6 binding by gene region. (D) The top 2 overrepresented cis binding elements in the Prdm6 ChIP-Seq data set.

A breakdown of Prdm6 binding by gene component (Figure 8C) demonstrated that Prdm6 was frequently associated with transcription start sites, suggesting that it plays a role in transcription initiation. However, Prdm6 binding was even more prevalent downstream of the transcription start site (TSS), especially in early introns, raising the possibility that it has additional roles in transcription maintenance or processing. Bioinformatic comparison of the Prdm6 binding regions revealed several overrepresented sequences that could reflect direct Prdm6 binding or its interaction with additional transcription factors (Figure 8D). Of interest was a TTTC/AT sequence that was identified as a potential Prdm6 binding site by Schmitges et al., who performed a large-scale ChIP-Seq screen on 78 Zn finger–containing proteins overexpressed in HEK293T cells including Prdm6 (43). Consensus sequences for AP1 and TEAD transcription factors were also detected by this comparison.

Transcription mechanisms that control Prdm6 expression. The fact that noncoding polymorphisms within the Prdm6 gene were associated with both cardiovascular disease and Prdm6 expression in arteries (see Table 1) strongly suggests that proper control of Prdm6 levels is important for normal SMC function. Although Prdm6 exhibits SMC-selective expression in mice (21), and relatively strong expression in human blood vessels (Figure 1D), nothing is known about the mechanisms that drive these expression patterns. To begin to analyze the transcription mechanisms that regulate Prdm6 expression, we took advantage of several genome-wide data sets that we previously generated to characterize chromatin structure and transcription factor binding in HuAoSMCs (16, 19). The Int3.1 region (red box in Figure 9A) was of particular interest for several reasons. It contained a highly conserved 325 bp sequence that included binding motifs for transcription factors known to regulate SMC-specific gene expression (see Figure 9C). It contained a DNase I hypersensitive open chromatin region that was found to be SMC selective when compared to ENCODE data from 7 other non-SMC cell types. It was marked by histone modifications known to be associated with regulatory regions (H3K4 methylation and H3K27 acetylation). It was shown to bind SRF and RBPJ in ChIP-Seq assays. And it contained a genetic variation (rs17149944) that was one of several that define a linkage disequilibrium block associated with blood pressure and Prdm6 expression.

Identification of regulatory elements and genetic variations that control PFigure 9

Identification of regulatory elements and genetic variations that control Prdm6 expression in SMCs. (A) Schematic illustrating the genome-wide data sets used to prioritize our search for regulatory elements that control the SMC-selective expression of Prdm6. (B) The indicated Prdm6 region was cloned into the appropriate luciferase reporter plasmid and then transfected into HuBrSMCs and mouse ECs. Luciferase activity was measured at 48 hours and is expressed relative to the appropriate empty vector. n ≥ 3 for all experimental groups. *P < 0.05 versus promoter less (t test); #P < 0.05 versus ECs (t test). (C) DNA sequence of the 325 bp conserved region within the Int3.1 enhancer. (D) PCR mutagenesis was used to generate the indicated mutations in the context of the highly active In3.1 conserved regulatory region. Luciferase activity was measured at 48 hours and is expressed relative to the activity of the WT Int3.1 conserved construct. n = 5 for all groups. *P < 0.05 versus WT (t test). (E) Targeted ChIP assays measuring SRF, RBPJ, and TEAD1 binding to the endogenous Int3.1 region. PCR primers used for these experiments are shown in red in C. n = 3 for all groups. *P < 0.05 versus IgG (t test). (F) Int3.1 conserved-luciferase was transfected into ECs and SMCs with/without myocardin. Luciferase activity was measured at 48 hours and is expressed relative to Int3.1 conserved luciferase activity in the presence of empty expression vector. n = 5 for all groups. *P < 0.05 versus plus empty vector (t test). (G) PCR mutagenesis was used to generate an allelic series for the rs17149944 variation within the context of Int3.1-luciferase construct. Note that the presence of the minor A allele significantly reduced the activity of the Int3.1 enhancer. n = 5 per group. *P < 0.05 versus WT (t test).

Table 1

SNPs within the Prdm6 third intron that associate with cardiovascular disease and Prdm6 expression

The regulatory regions depicted in green at the bottom of Figure 9A were PCR-amplified from human genomic DNA, subcloned into the appropriate luciferase reporter vectors, and then tested for transcriptional activity in SMCs and endothelial cells (ECs) (to examine cell type specificity). As shown in Figure 9B, the Prdm6 TSS drove high luciferase expression in human bronchial SMCs (HuBrSMCs), but exhibited relatively similar activity in ECs, suggesting that it functions more as a basal promoter. When subcloned upstream of the proximal SV40 promoter, Int2 functioned as a strong repressor, while Int3.2 had little activity. In contrast, the Int3.1 region had significant SMC-selective enhancer activity (~10-fold) within the same context (Figure 9B). An Int3.1 fragment that contained only the 325 bp conserved region exhibited remarkable SMC-selective activity (nearly 70-fold), strongly supporting its role in the SMC-selective expression of Prdm6. Importantly, individual mutations to the consensus SRF, RBPJ, and TEAD binding sites within the conserved Int3.1 region (Figure 9C) significantly inhibited its activity (Figure 9D), and targeted ChIP assays detected binding of each of these factors to the Int3.1 region within the endogenous Prdm6 gene (Figure 9E). Providing additional support for the importance of SRF, overexpression of myocardin significantly increased the activity of the Int3.1 conserved region (Figure 9F).

As shown in Table 1, polymorphisms within an approximately 35 kb region of the Prdm6 third intron (see Figure 9A) have been associated with cardiovascular disease and blood pressure. All 6 are in high linkage disequilibrium (r2 > 0.78), and the risk allele at each variant has been associated with a similar decrease in Prdm6 expression in human arteries. Thus, any one or more of these could be the causal variant at this important locus. Closer examination of the chromatin environment surrounding these SNPs revealed that only 2 were within regions that might exhibit regulatory activity. Since rs17149944 (but not rs2287696) was within a region that exhibited positive transcriptional activity in luciferase assays (Int3.1), we hypothesized that it was the causal variant and that the minor allele at this locus reduced Prdm6 expression by inhibiting the activity of the Int3.1 enhancer. In support of this idea, mutation of the major rs17149944 allele (G to A) inhibited the activity of the Int3.1 enhancer by over 50% (Figure 9G).

SMC-specific depletion of Prdm6 in adult mice did not affect BP or the development of hypertension. The early lethality observed upon deletion of Prdm6 globally or in SM22- or Wnt1-expressing cells has made it difficult to determine whether Prdm6 expression in vascular SMCs plays a role in blood pressure regulation in adult animals. To directly test this, we crossed the Prdm6fl/fl mice with a well-characterized tamoxifen-inducible Cre line driven by the SM MHC promoter (SMMHCCreERT2) (44). In brief, following telemeter implantation and equilibration, mice were injected with tamoxifen (or corn oil) for 5 consecutive days, and blood pressure was monitored continuously over the next several weeks by radio telemetry. As shown in Figure 10, we did not detect significant differences between tamoxifen- and vehicle-treated mice, suggesting that Prdm6 depletion does not affect baseline blood pressure. Since many phenotypes in genetically modified mice are only revealed after significant stress, we challenged mice with increasing doses of the NO synthase inhibitor, l-NAME (50 mg/L, 150 mg/L, or 450 mg/L in drinking water), to induce hypertension. Although l-NAME treatment resulted in an 18 mmHg rise in blood pressure, we did not observe significant differences between the tamoxifen- and vehicle-treated groups.

Prdm6 depletion in adult SMCs does not affect blood pressure or l-NAME–induFigure 10

Prdm6 depletion in adult SMCs does not affect blood pressure or l-NAME–induced hypertension. Following telemeter implantation and equilibration, SMMHCCreERT2 Prdm6fl/fl mice were treated with 100 mg/kg tamoxifen (n = 3) or corn oil (n = 3) by oral gavage for 5 consecutive days. l-NAME was added to drinking water at increasing doses (50 mg/L, 150 mg/L, 450 mg/L) 10 days later as indicated. (A) Blood pressure (mean arterial pressure, MAP) was monitored continuously by radio telemetry over the entire experiment. All blood pressure measurements are presented as averages over 24-hour periods. (B) PCR detection of floxed allele recombination in aorta samples from oil- and tamoxifen-treated SMMHCCreERT2 Prdm6fl/fl mice. (C) Quantitative PCR–based measurement of Prdm6 mRNA depletion in aorta samples from oil- and tamoxifen-treated SMMHCCreERT2 Prdm6fl/fl mice. n = 3–5, *P < 0.05 (t test).

留言 (0)

沒有登入
gif