An Alternative Retinoic Acid-responsive Stra6 Promoter Regulated in Response to Retinol Deficiency*

Background: Stra6 is a trans-membrane retinol transporter involved in retinoic acid (RA) signaling. Results: The epigenetic signature of the Stra6 gene reveals an RA-responsive element. Conclusion: An intragenic RARE drives RA-responsive expression of two different Stra6 isoforms. Significance: The novel, shorter Stra6 transcript may encode a functionally different retinol transporter. Cellular uptake of vitamin A (retinol) is essential for many biological functions. The Stra6 protein binds the serum retinol-binding protein, RBP4, and acts in conjunction with the enzyme lecithin:retinol acyltransferase to facilitate retinol uptake in some cell types. We show that in embryonic stem (ES) cells and in some tissues, the Stra6 gene encodes two distinct mRNAs transcribed from two different promoters. Whereas both are all-trans-retinoic acid (RA)-responsive in ES cells, the downstream promoter contains a half-site RA response element (RARE) and drives an ∼13-fold, RA-associated increase in luciferase reporter activity. We employed CRISPR-Cas9 genome editing to show that the endogenous RARE is required for RA-induced transcription of both Stra6 isoforms. We further demonstrate that in ES cells, 1) both RARγ and RXRα are present at the Stra6 RARE; 2) RA increases co-activator p300 (KAT3B) binding and histone H3 Lys-27 acetylation at both promoters; 3) RA decreases Suz12 levels and histone H3 Lys-27 trimethylation epigenetic marks at both promoters; and 4) these epigenetic changes are diminished in the absence of RARγ. In the brains of WT mice, both the longer and the shorter Stra6 transcript (Stra6L and Stra6S, respectively) are highly expressed, whereas these transcripts are found only at low levels in RARγ−/− mice. In the brains of vitamin A-deficient mice, both Stra6L and Stra6S levels are decreased. In contrast, in the vitamin A-deficient kidneys, the Stra6L levels are greatly increased, whereas Stra6S levels are decreased. Our data show that kidneys respond to retinol deficiency by differential Stra6 promoter usage, which may play a role in the retention of retinol when vitamin A is low.

Mutations in STRA6 can result in the Matthew-Wood syndrome, which consists of severe microphthalmia, pulmonary agenesis, bilateral diaphragmatic eventration, duodenal stenosis, pancreatic malformations, and growth retardation (13,20,21). Some mutations in STRA6 in humans involve either a homozygous insertion/deletion in exon 2 or a homozygous insertion in exon 7, predicting a premature stop codon (20). Several STRA6 mutations associated with human disease have been shown to limit or abolish retinol uptake into cells (23,24). Three groups have independently reported marked ocular defects in Stra6 null mice (25)(26)(27). Genetic ablation of Stra6 results in a reduced retinoid content in the retinal pigment epithelium and neurosensory retina (greater than a 95% reduction in retinyl esters) with consequently fewer cone photoreceptor cells and diminished cone b-wave amplitude (25). Under the evaluated conditions, knock-out of Stra6 did not impair the physiological functions of retinoids in tissues other than the eye (26). Under vitamin A-deficient conditions, Stra6 was reported to facilitate redistribution from storage tissue (e.g. liver and lungs) to the eye (27), possibly in conjunction with the recently identified RBPR2, another RBP4 receptor (4). Ablation of Stra6 can also protect animals from the insulin-resistant state induced by feeding a high fat, high sucrose diet (28), which may be a result of impaired Jak/Stat signaling (29). Why humans with certain mutations in STRA6 exhibit a more severe morphological phenotype (25)(26)(27) than the Stra6 knock-out mice is currently not understood.
Vitamin A acts through its biologically active metabolite, alltrans-retinoic acid (RA), and the retinoic acid receptors (RARs) to regulate large numbers of genes at the transcriptional level in various cell types (30 -32). On a molecular level, RA mediates major changes in epigenetic marks on specific target genes (33)(34)(35)(36)(37). We recently showed that Stra6 transcripts are increased by RA treatment of embryonic stem (ES) cells and that this induction depends on RAR␥ (38).
There is a great interest in the cellular uptake of vitamin A, yet the mechanisms by which the Stra6 gene is regulated at the transcriptional level are not clearly defined. In this report, we demonstrate that the Stra6 gene possesses two different promoters, both of which are induced by RA through a single intragenic RARE. We characterize the Stra6 retinoic acid-responsive DNA element (RARE) and demonstrate that RXR␣/RAR␥ heterodimers directly associate with this cis-acting element of Stra6 to activate RA-dependent transcription. Finally, we identify the differential regulation of the two Stra6 promoters in the kidneys of vitamin A-deficient animals, which points to physiologically distinct functions of the two Stra6 isoforms.

EXPERIMENTAL PROCEDURES
Cell Culture and Retinoic Acid Treatment of ES Cells-The WT (CCE) and RAR␥ Ϫ/Ϫ ES cell lines were derived and cultured as described (33). RA (Sigma) was added to the cells 24 h after plating (1 M final concentration), and ethanol (EtOH, 0.1%) served as a vehicle control. For time course evaluations, ES cells were plated in gelatin-coated tissue culture dishes and treated with RA (1 M) at various time points up to 72 h prior to harvesting.
Vitamin A-deficient Mice-WT C57Bl/6 and LRAT null C57Bl/6 mice were given vitamin A-sufficient (VAS) or vitamin A-deficient (VAD) chow for 10 weeks, as described (6). The treatment of the mice was approved by the Institutional Animal Care and Use Committee at Weill Cornell. Tissues were harvested, RNA was isolated, and semiquantitative RT-PCR was performed as described (6). The brains were dissected without the eyes. Serum and liver retinoid levels were measured by HPLC as described (6), demonstrating that the mice on the VAD chow were vitamin A-deficient.
RNA Isolation and Reverse Transcription-Total RNA was isolated from F9 and ES cells using TRIzol reagent (Invitrogen), and RNA was quantitated by optical density at 260 nm. The RNA (1 g) was reverse transcribed (Quanta Biosciences, Gaithersburg, MD) and then diluted 1:10 with H 2 O. The cDNA obtained was diluted 10-fold, and 3 l of this cDNA was utilized for PCRs.
Generation of cDNA, Semiquantitative, and Real-time PCR-Real-time PCR was performed using SYBR Green Supermix (Quanta Biosciences) in a 15-l reaction containing reaction mix (1ϫ), a 0.2 M concentration of each primer, and 3 l of cDNA template. The reactions were run on a Bio-Rad MyiQ TM single color real-time PCR detection system (Bio-Rad). Amplification in the linear range was demonstrated by a serial dilution of cDNA from RA-treated wild type (WT) cells included in each reaction (1:1, 1:5, 1:10, 1:50, 1:100, and 1:500). Reactions with H 2 O and template without reverse transcriptase, respectively, served as negative controls for primer-dimer and for amplification of residual genomic DNA. All real-time PCR primers were designed to span intronic regions. Primer sequences are listed (Table 1). Each expression analysis was performed at least three times (e.g. nՆ3, independently propagated cells, experiment repeated three times). Within each PCR analysis, samples were run in triplicate. PCR products were verified by DNA sequencing.
Mapping of Transcriptional Start Sites-5Ј-Rapid amplification of cDNA ends (RACE) was performed with the Roche Applied Science 5Ј/3Ј RACE kit using mStra6(Ϫ)J for cDNA synthesis and mStra6(Ϫ)B for nested PCR amplification. The PCR products were purified using a PCR purification kit (Qiagen), and the eluted DNA was recovered by ligation into the pGEM-T easy vector (Promega, WI). The ligated products were recovered by bacterial transformation. Single colonies were picked, and plasmid DNA was isolated. Insert-containing plasmids were identified by restriction digestion and sequenced using T7(ϩ) and SP6(ϩ) primers.
Constructing and Assaying Stra6 Reporter Plasmids-Firefly luciferase reporter constructs were designed by cloning PCR fragments of various Stra6 genomic regions into the pGL3 reporter plasmid. In brief, PCR fragments were SpeI/SalI-digested and cloned into NheI/XhoI sites of the pGL3 basic plasmid (Promega, WI). The Stra6 reporter constructs and their genomic coordinates are shown in Table 2. Point mutations were introduced into the pGL3 mStra6 -0.8 construct using Stratagene QuikChange (see Table 1 for primer sequences). The ES cells were transiently transfected with different luciferase reporter constructs using Lipofectamine LTX (Invitrogen). The pRL-TK reporter plasmid, Renilla luciferase-thymidine kinase (Promega, WI), served as a control for transfection efficiency. WT ES cells were transfected and then cultured in DME containing 1.0 M RA for 8 or 24 h, as indicated. Firefly and Renilla luciferase activities were sequentially measured with a luminometer using the Dual-Luciferase reporter assay system (Promega).  Increase in luciferase activity upon 24-h RA treatment (relative to vehicle-treated cells). Shown are genomic coordinates for Stra6 reporter elements and the RARE (mm9). Note that pGL3 Stra6-P L /P S is a composite reporter construct that contains two Stra6 genomic regions (one upstream and one downstream of the luciferase coding region). These constructs are further described in the legend to Fig. 2A Genome Editing and Screening-Double-stranded DNA fragments targeting the Stra6-RARE (Table 1) were cloned into the pX330 vector (39). Integrity of the insert was confirmed by sequencing, and the construct was transfected into WT ES cells. Upon overnight recovery, the transfected cells were reseeded at low density (2 ϫ 10 3 cells/15-cm dish) to obtain independent clonal lines. The colonies were screened by PCR (mStra6(ϩ)P and mStra6-R(Ϫ)H primers) followed by MboI restriction digest. The PCR bands with modified MboI sites were sequenced to map the exact mutations introduced at the Stra6-RARE. Cell lines with single and double mutations of the Stra6-RARE were assayed in three independent experiments for transcriptional induction of Stra6 by RA.
Data Analysis and Statistics-Data from at least three independent experiments were analyzed using one-way analysis of variance in the expression and ChIP analyses. The S.E. was determined for each of the data sets (at least three biological, indepen- The promoter regions P L and P S correspond to the P RefSeq and to the novel promoter proximal to the Stra6 RARE, respectively. Note that the heat map is aligned with the schematics in D below. C, detection of Stra6 transcripts from P L and P S promoters as well as total Stra6 transcript levels in untreated and RA-treated WT and RAR␥ Ϫ/Ϫ ES cells by RT-PCR (top). The primers utilized are specified by letters. RAR␥ transcript levels and 36B4 (loading control) are also shown. Bottom, mapping the 5Ј-end of RA-induced Stra6 transcripts by 5Ј-RACE. D, Stra6 genomic structure (top), reporter constructs (middle), and promoter diagram (bottom). Alternative transcripts and Stra6 genomic regions are depicted at the top. The RefSeq promoter (P L ) and the novel RA-responsive promoter (P S ) identified are shown here. Squares, exons (open, untranslated). Stra6 reporter constructs were evaluated for RA induction of P L and P S (-fold induction by RA is noted to the right). The alternative Stra6 transcripts correspond to NM_001162475.1 (6.1), NM_009291.2 (6.2), NM_001162476.1 (6.3), NM_001162479.1 (6.4), and a novel short isoform (6.5; GenBank TM number AK092227.1). Stra6 promoter regions (P L and P S ) and exon usage of the 5Ј-mRNAs are shown. The locations of primer binding sites are specified by single letters below or above the arrowheads (forward and reverse primers, respectively). Error bars, S.E. FEBRUARY 13, 2015 • VOLUME 290 • NUMBER 7 dent repeats, each in triplicate, plotted as error bars in the graphs), and analysis of variance values of p Ͻ 0.05 among compared samples were assigned statistical significance.

The Increase in Stra6 Transcripts in RA-treated ES Cells Is
Associated with Extensive Epigenetic Changes-We employed WT and RAR␥ Ϫ/Ϫ ES cells (38) to delineate the dynamics of the increase in Stra6 transcripts associated with RA treatment of murine ES cells in culture. We demonstrated that there is a 24-fold increase in Stra6 transcripts 48 h after RA addition in WT ES cells and that this increase does not occur in the RAR␥ Ϫ/Ϫ ES cells (Fig. 1A). We next determined the epigenetic changes that occur along the entire Stra6 genomic region in response to RA treatment of ES cells. Transcriptional activation is strongly associated with Lys-9/14 acetylation (H3K9/14ac) and Lys-4 trimethylation (H3K4me3) of histone 3 (43). We therefore assessed the dynamics of these epigenetic marks in RA-treated WT and RAR␥ Ϫ/Ϫ ES cells using ChIP-chip analysis ( Fig. 1B). At the RefSeq promoter, which we named Promoter Long (P L ), the increase in H3K9/14ac was much greater in WT than in RAR␥ Ϫ/Ϫ cells ( Fig. 1B). At a region ϳ12 kb downstream of the P L (RefSeq) promoter, we again observed an RAdependent increase in the H3K9/14ac epigenetic marks in WT but not in RAR␥ Ϫ/Ϫ ES cells (Fig. 1B). Because this may reflect a downstream promoter driving expression of a shorter transcript, we named this region Promoter Short (P S ). The H3K4me3 mark was present in both WT and RAR␥ Ϫ/Ϫ cells at P L but not at P S (Fig. 1B). We detected increased Stra6 transcript levels after RA addition only in the WT cells. Thus, the presence of the H3K9/14ac mark correlated with transcriptional activation, whereas the increase in the H3K4me3 mark specifically at P L was insufficient to induce transcription because the increase in the H3K4me3 mark occurred even in the RAR␥ Ϫ/Ϫ cells (Fig.  1B). In contrast, H3K27me3, a repressive histone modification introduced by the Polycomb repressive complex 2 (PRC2), was decreased at both Stra6 promoters P L and P S in response to RA treatment in the WT but not in the RAR␥ Ϫ/Ϫ ES cells (Fig. 1B). The dynamic changes in these three key histone epigenetic marks revealed two distinct, genomic regions of the Stra6 gene, which in WT ES cells are remodeled in response to RA.
The Stra6 Gene Encodes Two Transcripts Expressed from Two Distinct Promoters-The data from our ChIP-chip experiments (Fig. 1B) pointed to a second regulatory region in the Stra6 intragenic region, termed P S . A short form of Stra6, which encodes an amino-terminally truncated Stra6 protein, has been identified in human samples (GenBank TM AK092227.1). We hypothesized that the P S epigenetic "hotspot" located 12 kb downstream of the P L (RefSeq) reflects transcriptional initiation of a short Stra6 isoform (Stra6 S ) expressed in mouse ES cells in response to RA.
We designed primers specific for the long and the putative short isoform, respectively, and evaluated the transcript levels of these two Stra6 isoforms in WT and RAR␥ Ϫ/Ϫ ES cells in response to RA treatment. The transcript levels of both isoforms were dramatically increased by RA treatment of WT cells, whereas no transcripts were detected in the RAR␥ Ϫ/Ϫ ES cells (Fig. 1C). As a positive control for the induction of Stra6 by RA, we employed pan-primers to evaluate the total Stra6 transcript levels (all isoforms) (Fig. 1C). To confirm the genotype of the RAR␥ knock-out ES cells, we evaluated the RAR␥ transcript levels in the WT and the RAR␥ Ϫ/Ϫ ES cells (Fig. 1C). The 36B4 transcript levels were used as a reference gene (Fig. 1C).
We next determined that the Stra6 short isoform was generated by a downstream promoter. We used 5Ј-RACE to identify transcriptional start sites (TSSs) of Stra6 transcription in WT ES cells before and after treatment with RA. We identified two TSSs ϳ560 and ϳ300 bp upstream of the Stra6(Ϫ)B primer binding site (Fig. 1C, right). The 5Ј-RACE PCR products are slightly larger than the expected sizes of the long and the short isoform of Stra6 (521 and 261 bp) due to the addition of a 5Ј-linker (41 bp). Thus, our data show that the two TSSs give rise to the long and a short isoforms of Stra6, both of which increased in WT but not in RAR␥ Ϫ/Ϫ ES cells after RA addition (Fig. 1C). We conclude that in WT ES cells, a long and a short Stra6 transcript are generated from two different promoters, the P L and the P S promoter, respectively (shown in Fig. 1B, red bars at the bottom). The levels of both transcripts increase upon RA treatment of WT ES cells (Fig. 1C). The Stra6 transcripts originating from the P L and the P S promoters are designated Stra6 L (Stra6.1) and Stra6 S (Stra6.5), respectively (Fig. 1D). Transcript levels of Stra6.2, Stra6.3, and Stra6.4 were below detection in ES cells (data not shown).
Transcriptional Activation Assays Reveal an RARE Proximal to the Promoter of the Stra6 S Isoform-We next cloned putative promoter regions of Stra6 upstream of the luciferase reporter gene in order to identify the genomic sequences involved in the induction of Stra6 by RA (Fig. 1D). We performed transient transfection assays in WT ES cells and evaluated conserved regions proximal to P L (0.6 kb) and P S (1.7 and 1.0 kb) for RAresponsive transcriptional activity (1 M RA, 24 h). In order to evaluate potential enhancer activity of P S , we included a construct in which P S is located downstream of the P L TSS, thereby modeling transcription of the long Stra6 isoform. We found that P S was induced to a greater extent than P L by RA (ϳ5-fold versus 2-fold, respectively), whereas P S failed to enhance the promoter activity of P L (1.3-fold) (Fig. 1D). These transient transfection assays confirmed the presence of an unannotated RA-inducible promoter (P S ), which is located ϳ12 kb downstream of the RefSeq promoter (P L ) identified in the NCBI database (Fig. 1, B and D).
We next cloned various regions proximal to the Stra6 P S upstream of the reporter luciferase gene and performed transient transfection assays in WT and RAR␥ Ϫ/Ϫ ES cells. The cells were cultured in either the absence or presence of RA for 8 or 24 h (Fig. 2). We utilized a variety of deletion constructs in the 1.7-kb region proximal to P S to demonstrate that the greatest increase in transcriptional activation after RA treatment of WT ES cells occurred when a small, 286-bp region was included (e.g. Stra6-0.9, Stra6-0.8, bracketed in Fig. 2C). The majority of the RA inducibility was lost when we deleted this region (Stra6-0.6, Stra6-0.5) (Fig. 2B). In the RAR␥ Ϫ/Ϫ ES cells, the Stra6-0.8 construct did not result in RA induction of the luciferase reporter (Fig. 2B), indicating that this RAresponsive region requires RAR␥ for transcriptional activity.
We analyzed the DNA sequence of the RA-responsive region (Fig. 2C, bracketed) and identified a conserved RARE halfsite, ATGACC (boldface type indicating conserved nucleotides), located in this intronic region (Fig. 2C, double-boxed  sequence). When the RARE half-site was mutated, the RAresponsiveness of the Stra6-0.8 reporter construct was abrogated (Fig. 2B). The Stra6-m2 contains a triple mutation in which the ATGAC (double-boxed in Fig. 2C) is mutated to AACTC. We also tested another RARE half-site (CTGAA; boxed in Fig. 2C). Again, we introduced a triple mutation (CTGAA to CACTA), but this mutation did not affect the RA-responsiveness of the Stra6 -0.8 reporter construct (Fig.  2B). Thus, the ATGAC half-site is responsible for the RAresponsiveness of the P S Stra6 promoter ( Fig. 2A).
In summary, we found that a 0.8-kb region proximal to P S was sufficient to drive a large increase (ϳ13-fold) in luciferase reporter gene expression in response to RA (Fig. 2) and that this increase required RAR␥. We further identified an RARE halfsite, which, when mutated, abolished the RA-responsiveness of the reporter construct. Mapping the RARE of Stra6 using reporter assays. A, schematic representation of the Stra6 genomic region and reporter constructs. Promoters for the long and the novel short forms of Stra6 transcripts are designated as P L and P S , respectively. The arrow indicates the epigenetic "hotspot" observed in the heat map (Fig. 1). R, conserved RARE half-site. A gray bar below the schematic reporter construct marks the RA-responsive region shown in C. B, Stra6 reporter assays in WT and RAR␥ Ϫ/Ϫ ES cells upon RA treatment. Deletion analysis of the RA-responsive Stra6-1.7 construct is shown at the top. Point-mutated Stra6-0.8 constructs (m1 and m2), the P L construct, and Stra6-0.8 construct in RAR␥ Ϫ/Ϫ ES cells (boxed) are shown at the bottom. Stra6-m1 and Stra6-m2 indicate mutations in RARE half-sites (see details below). All luciferase assays were performed at least three times, starting with fresh cells. *, p Ͻ 0.05 relative to vehicle control. C, key elements of the Stra6 short isoform proximal to the essential region identified through the reporter assays (bracketed): the transcriptional start site (i.e. the 5Ј-end of the transcript) (Ͼ), the conserved RARE half-site with the inverted repeat shown in boldface type (double boxed, m2, TGA 3 ACT), the non-essential RARE half-site (boxed, m1, TGA 3 ACT), and the putative start codon (boldface type). Exon sequences are depicted in capital letters with gray shades indicating exonic regions present in both Stra6 S and Stra6 L . The ‚, splice acceptor site for splicing of the longer isoforms (in which case the upstream region of this exon is spliced out). OE, splice acceptor site common for the short (6.5) and the longer (6.1) Stra6 isoforms. D, diagram of the screening strategy and sequences of CRISPR modified ES clones. The CRISPR targeting sequence is underlined, and the site of double-stranded breakage is indicated by a gray triangle. The genomic sequences of three clones are depicted (one biallelic and two monoallelic mutations). Note that only the mutated allele is shown. E, PCR screening and expression of Stra6 isoforms in ES cell lines with mutated RAREs. F, transcriptional induction of Stra6 in ES cell lines with mutated RAREs. The induction of Stra6 by RA was impaired in cell lines with monoallelic mutations and strongly impaired in the cell line with biallelic mutations in the Stra6 RARE (*, p Ͻ 0.05 relative to RA-treated WT cells). Stra6 transcript levels were normalized to the 36B4 reference gene. Error bars, S.E.

The Stra6 RARE Is Required for Induction of Endogenous
Stra6 by RA in ES Cells-We next wanted to address the role of the endogenous RARE in RA-induced transcription of Stra6. We employed the CRISPR-Cas9 technology to introduce a targeted deletion of the Stra6 RARE (Fig. 2D) in WT ES cells. We evaluated independent clones for RA-responsive transcription of Stra6 and found that ablating the Stra6 RARE resulted in impaired induction of both Stra6 S and Stra6 L (Fig. 2E). The cell lines harboring monoallelic or biallelic deletion of the Stra6 RARE exhibited ϳ45 and 95% reduction, respectively, in the total Stra6 transcript levels (all isoforms) relative to RA-stimulated WT ES cells (Fig. 2F).
The Stra6 Promoters, P S and P L , Show Differential Epigenetic Responses to RA in WT Versus RAR␥ Ϫ/Ϫ ES Cells-The consensus RARE element is composed of two direct repeats (RGKTSAN 2/5 RGKTSA). In contrast, the RARE half-site identified in the reporter assays is part of an inverted repeat, (gaaT-GACCTctttAGATCAaca, half-sites in capital letters; see the legend to Fig. 2 for details). We therefore evaluated the RAR␥ and RXR␣ association with the Stra6 P S and P L regions by ChIP. We show that RAR␥ and RXR␣ associate with the P S promoter region in the WT cells but not in the RAR␥ Ϫ/Ϫ ES cells (Fig. 3, A and B). H3K27ac, generally a mark of enhancers/transcriptional activation (44,45), occurred at higher levels at both the P S and P L regions following RA addition to the WT cells but not the RAR␥ Ϫ/Ϫ cells (Fig. 3C). The acetylation of H3K27 correlated with coactivator protein p300 association (Fig. 3D), which indicates a role for p300 in introducing this histone modification (46).
We detected the H3K27me3 epigenetic mark in both WT and RAR␥ Ϫ/Ϫ ES cells; this mark showed a dramatic decrease upon RA addition at both the Stra6 P S and P L regions in WT, whereas higher levels were observed in the RAR␥ Ϫ/Ϫ cells relative to WT (Fig. 3E). The trimethylation of H3K27 correlated with the levels of the Polycomb group protein Suz12 (Fig. 3D), a key component of the Polycomb repressive complex 2 that trimethylates H3K27 (47).
Finally, we evaluated the levels of PolII at the Stra6 P S and P L regions. We demonstrate that at both the Stra6 P S and P L regions, the PolII association increased greatly in WT cells upon RA addition (Fig. 3G). This increase was not seen in the RAR␥ Ϫ/Ϫ ES cells (Fig. 3G). The IgG ChIP, a negative control, demonstrates that the background levels of nonspecific ChIP signals do not change in response to RA (Fig. 3H).
Thus, from our ChIP-chip and ChIP-qPCR assays, we conclude that among the evaluated histone marks, the greatest difference between Stra6 promoters P L and P S relates to the H3K4me3 mark. Specifically, the H3K4me3 levels increased at P L in an RA-dependent manner in both WT and RAR␥ Ϫ/Ϫ ES

TABLE 3 Transcription factors and epigenetic marks associated with Stra6 promoters in ES cells
The

ChIP-qPCR
PolII-CTD RA-dependent increase in WT but not in RAR␥ Ϫ/Ϫ RA-dependent increase in WT but not in RAR␥ Ϫ/Ϫ ChIP-qPCR cells, whereas at P S , the H3K4me3 levels did not change in response to RA (Fig. 1B). Importantly, whereas we detected no RAR␥ and RXR␣ association with P L , both receptors showed strong association with the P S proximal promoter region (Fig. 3,  A and B). The epigenetic signatures of Stra6 P S and P L are summarized in Table 3. The Stra6 Short Isoform mRNA Is Highly Expressed in Brain and Is Reduced in Vitamin A Deficiency-We next asked if the short Stra6 isoform is expressed in animals and, if so, in what tissues. We determined the levels of Stra6 S and Stra6 L transcripts by RT-PCR in various tissues of WT mice (Fig. 4A). We detected Stra6 L transcripts in the heart, intestine, and brain (Fig. 4A). Importantly, we also observed high levels of the Stra6 S transcripts in the brain, uterus, kidney, and heart (Fig. 4A). We assessed the RAR␥ transcript levels in these tissues (Fig. 4A) and found that the Stra6 S and Stra6 L transcripts are present in a subset of the tissues that are positive for RAR␥ transcripts. This supports a role for RAR␥ in regulating the transcription of both Stra6 isoforms.
Vitamin A-deficient Mice Display Aberrant Stra6 Transcript Levels in Brain and Kidneys-We investigated whether the Stra6 S and Stra6 L transcript levels were differentially regulated in response to vitamin A deficiency. We tested the effects of vitamin A deficiency on Stra6 transcript levels by placing WT and LRAT Ϫ/Ϫ C57Bl/6 male and female mice on a vitamin A-deficient diet for 10 weeks and compared these mice with controls fed a normal (vitamin A-sufficient) chow diet. We previously showed that LRAT Ϫ/Ϫ mice become VAD more rapidly because they are unable to store retinol in the ester form in the liver and other tissues that act as storage depots for retinyl esters (6). We found greatly increased Stra6 L transcript levels in kidneys of WT and LRAT Ϫ/Ϫ mice that were deprived of vitamin A (Fig. 4B) and decreased levels of Stra6 S transcripts in the kidneys of LRAT Ϫ/Ϫ mice relative to the WT mice. This difference was enhanced by vitamin A deficiency, which almost abolished the Stra6 S transcript levels in LRAT Ϫ/Ϫ mice (Fig. 4B). The RAR␥ transcript levels in the kidneys did not change in response to vitamin A deficiency (Fig. 4B), suggesting that the changes in Stra6 transcript levels are not caused by altered levels of RAR␥. In vitamin A-sufficient conditions (WT-VAS) the Stra6 S is the isoform predominantly expressed in the kidney, but with increasing severity of vitamin A depletion (either genetic, dietary, or combined), we observed a shift from Stra6 S to Stra6 L expression (Fig. 4B). We conclude that the Stra6 gene exhibits isoform-specific expression (e.g. differential promoter usage) in response to the levels of retinol and/or retinol metabolites in the kidney.
In the brains of vitamin A-deficient animals, the levels of both long and short Stra6 isoforms are decreased (Fig. 4C), whereas the RAR␤ and RAR␥ transcript levels did not change in response to vitamin A deficiency in WT and LRAT Ϫ/Ϫ mice (Fig. 4C). Also, we detected decreased levels of both the long and short Stra6 isoforms in the brains of RAR␥ null mice on a chow diet that contained vitamin A (Fig. 4D). Thus, RAR␥ is necessary for maintaining the normal levels of both Stra6 S and Stra6 L transcripts in the brain.

DISCUSSION
Stra6 is involved in numerous cell signaling pathways and mediates the uptake of retinol from holo-RBP (2). Thus, there is increasing interest in the regulation of Stra6 expression, the focus of this research. Murine Stra6 was originally cloned by Bouillet et al. (12), who determined that a genomic region 4.5 kb upstream of the RefSeq promoter does not recapitulate the RA inducible transcription of Stra6. We have mapped a Stra6 RARE to a region 12 kb downstream of the RefSeq promoter, and we demonstrate here that two different transcripts are generated from the Stra6 gene in cultured ES cells and in various murine tissues. In WT ES cells, RA induces transcription of both Stra6 S and Stra6 L through a single RARE (Figs. 1 and 2). Similarly, in the brain, Stra6 L and Stra6 S levels both decrease in response to vitamin A deprivation, but intriguingly, in the kidney, the Stra6 S and Stra6 L transcripts are regulated in an opposing fashion (Fig. 4).
Consistent with the increase in Stra6 L and Stra6 S transcript levels in RA-treated WT ES cells (Fig. 1), we observed RA-dependent increases in PolII levels at both Stra6 L and Stra6 S promoters (Fig. 3). In contrast, RAR␥ and RXR␣ showed strong association only with the Stra6 S promoter (Fig. 3B). The absence of RAREs proximal to Stra6-P L is supported by our analysis of published ChIP-seq data (48) and is consistent with previous findings (12). We employed CRISPR-Cas9 technology to demonstrate that in ES cells, the Stra6 RARE, which is located proximal to the Stra6 S promoter, is required for RAresponsive transcription of both Stra6 S and Stra6 L (Fig. 2E).
We focused on the epigenetic changes associated with the transcriptional activation of the Stra6 gene by RA. We found that both Stra6 L and Stra6 S promoters are transcriptionally activated by RA, as evident by an RA-dependent increase in histone acetylation (H3K9/14ac, H3K27ac, and p300) and Polycomb depletion (H3K27me3 and Suz12). Importantly, these marks exhibited no changes when RAR␥ Ϫ/Ϫ ES cells were treated with RA (Fig. 3), demonstrating the requirement for RAR␥ in RA-dependent Polycomb removal and histone acetylation of the Stra6 genomic region and in the transcriptional activation of Stra6 in ES cells.
In contrast to histone acetylation, which increased at both Stra6 P L and P S , we observed an RA-associated increase in H3K4me3 levels at the Stra6 L promoter only; notably, the H3K4me3 levels increased even in RAR␥ Ϫ/Ϫ ES cells (Fig. 1B). The RA-associated increase in H3K4me3 at the Stra6 L promoter is consequently independent of RAR␥, and this modifi-cation alone is insufficient to induce transcriptional activation of Stra6. LRAT and Meis1 display similar transcriptional regulation; RAR␥ is required for the RA-dependent increase in H3K9/14ac marks, whereas the RA-induced increase in H3K4me3 takes place even in RAR␥ Ϫ/Ϫ ES cells (38). In contrast, the Cyp26a1 promoter exhibited lower levels of both H3K4me3 and H3K9/14ac marks in RA-treated RAR␥ Ϫ/Ϫ relative to WT ES cells (38). The H3K4me3 modification is deposited by mammalian homologs of the Drosophila Trithorax, termed myeloid/lymphoid or mixed-lineage leukemia (MLL) proteins, and in particular, the MLL2 complex plays an important role in differentiation of ES cells (49).
Our data from WT mice on VAS versus VAD diets show differences in the regulation of Stra6 L and Stra6 S transcripts in the kidneys (Fig. 4B). When retinol is supplied daily in the diet (Fig. 4B, VAS), only the Stra6 S transcript is expressed. However, when retinol is not present in the diet for 10 weeks, the LRAT Ϫ/Ϫ mice become vitamin A-deficient (6), and only the Stra6 L transcript is expressed in the kidneys (Fig. 4B, VAD, LRAT Ϫ/Ϫ ). In WT mice that are on a VAD diet for 10 weeks and are only partially vitamin A-deficient (6), both the Stra6 L and Stra6 S transcripts are expressed (Fig. 4B, Wildtype, VAD). These data show that Stra6 S mRNA is expressed in kidneys when retinol is present in the body but that as retinol stores are depleted in the livers of WT mice during the 10 weeks on the vitamin A-deficient diet, expression of the Stra6 L transcript is activated in the kidneys although retinol is still present in the WT mice. Our data strongly suggest that when circulating retinol is abundant (from dietary uptake), Stra6 S is the predominant isoform expressed in the murine kidneys. In contrast, as retinol stores are depleted as a result of dietary vitamin A deficiency, expression of the Stra6 L isoform increases in the kidneys. The differential regulation of the two Stra6 transcript isoforms in the kidney in response to vitamin A deprivation points to physiologically distinct functions of the different isoforms in the regulation of retinoid homeostasis.
In the brain, unlike the kidney, both the Stra6 L and Stra6 S transcripts are reduced in both WT and LRAT Ϫ/Ϫ mice on a vitamin A-deficient diet (Fig. 4C). Thus, in the brain, we do not observe the major switch from the Stra6 S to Stra6 L transcripts during vitamin A deficiency that we detected in the kidneys (Fig. 4, B versus C). Furthermore, we also show that both the Stra6 L and Stra6 S transcript levels are reduced in the brains of RAR␥ Ϫ/Ϫ mice compared with the WT mice (Fig. 4D). This is in marked contrast to the elevated levels of Stra6 observed in RAtreated F9 RAR␣ null cells (35) and in the testicular tubules of RAR␣ null mice (12). Consequently, whereas RAR␥ plays a direct role in activating Stra6 expression, RAR␣ may function to repress transcription of Stra6. Whether loss of RAR␣ results in elevated levels of one or multiple Stra6 isoforms is not known. The presence of RAR␥ transcripts in each of the tissues where either Stra6 S or Stra6 L transcripts are present (uterus, kidney, heart, brain, and intestine) further supports the dependence of Stra6 expression on RAR␥.
Stra6 expression is directly regulated by RA (Fig. 3A) and responds to vitamin A availability in vivo (Fig. 4). However, because we observed elevated levels of Stra6 L in the kidneys of vitamin A-deficient animals, we conclude that Stra6 L can be expressed in a manner independent of RA. This may reflect an additional layer of regulation whereby the kidneys are able to increase Stra6 L levels when faced with vitamin A scarcity.
Our studies identify a shorter Stra6.5 isoform (Fig. 1), which encodes a putative amino-terminally truncated Stra6 protein (532 amino acids). In contrast, Stra6.1, Stra6.2, Stra6.3, and Stra6.4 isoforms all encode identical, larger Stra6 proteins (670 amino acids, NP_001155947.1). The putative Stra6 S protein may share functional properties with the B-chain of human RBPR2, whose protein coding sequence is split onto two separate genes, one encoding the amino-terminal A-chain and one the carboxyl-terminal B-chain, respectively (4). It is interesting to note that RBP4 interactions and retinol uptake are impaired by insertions at residues 16, 84, and 133 in the amino-terminal region of the larger Stra6 protein (50), a region that is absent from Stra6 S (and from the B-chain of RBP2). In contrast, mutations at residues 323, 357, and 548, shared by both Stra6 isoforms, completely abolished the function of the larger Stra6 protein (50). It is tempting to speculate that the carboxyl-terminal region of Stra6 has a structural role in RBP4 interaction, whereas the amino-terminal region of Stra6 may serve a regulatory function, possibly by distinguishing between apo-and holo-RBP4 (unliganded and liganded, respectively). Importantly, that distinction could be involved in regulating the direction of retinol transport (22). This model suggests that decreasing Stra6 S levels in the kidneys may be a physiological response to vitamin A deprivation to reduce the loss of vitamin A to the urine. Further studies will be needed to characterize the functional differences between the Stra6 proteins encoded by the Stra6 L and Stra6 S transcripts, but our data indicate that the kidneys respond to vitamin A deficiency by differential Stra6 promoter usage, which may play a major role in the retention of retinol when vitamin A is scarce.