Distinct functions and regulation of epithelial progesterone receptor in the mouse cervix, vagina, and uterus

While the function of progesterone receptor (PR) has been studied in the mouse vagina and uterus, its regulation and function in the cervix has not been described. We selectively deleted epithelial PR in the female reproductive tracts using the Cre/LoxP recombination system. We found that epithelial PR was required for induction of apoptosis and suppression of cell proliferation by progesterone (P4) in the cervical and vaginal epithelium. We also found that epithelial PR was dispensable for P4 to suppress apoptosis and proliferation in the uterine epithelium. PR is encoded by the Pgr gene, which is regulated by estrogen receptor α (ERα) in the female reproductive tracts. Using knock−in mouse models expressing ERα mutants, we determined that the DNA−binding domain (DBD) and AF2 domain of ERα were required for upregulation of Pgr in the cervix and vagina as well as the uterine stroma. The ERα AF1 domain was required for upregulation of Pgr in the vaginal stroma and epithelium and cervical epithelium, but not in the uterine and cervical stroma. ERα DBD, AF1, and AF2 were required for suppression of Pgr in the uterine epithelium, which was mediated by stromal ERα. Epithelial ERα was responsible for upregulation of epithelial Pgr in the cervix and vagina. Our results indicate that regulation and functions of epithelial PR are different in the cervix, vagina, and uterus.


INTRODUCTION
Progesterone (P 4 ) and estradiol (E 2 ) are major ovarian steroid hormones crucial for the development and homeostasis of the female reproductive tract (i.e., uterus, cervix, and vagina) [1]. P 4 and E 2 function through progesterone receptor (PR) and estrogen receptor (ER), respectively. PR and ER are ligand−dependent transcription factors belonging to the nuclear receptor superfamily [1]. ERα and ERβ are encoded by Esr1 and Esr2, respectively [2]. ERα is the major ER subtype in the female reproductive tract of mice and humans [2][3][4]. ERα is composed of several distinct functional domains, N−terminal activation function 1 (AF1), DNA−binding domain (DBD), a hinge region, and C−terminal ligand− binding domain containing the AF2. ERα activates its target gene expression by binding to estrogen response elements (EREs) through its DBD (classical mechanism) or other transcription factor (e.g., AP-1 and Sp-1) binding sites through protein−protein interactions (tethering mechanism) [2]. The Pgr gene coding for PR is a well− known ERα target gene in the female reproductive tract [1,5]; however, the mechanism of Pgr activation by ERα is poorly understood. Reporter assays using isolated regulatory sequences of the PGR gene have implicated both classical and tethering mechanism in the transcriptional activation of PGR in MCF7 breast cancer cells [6][7][8][9]. However, it is unclear whether it is relevant to the normal female reproductive tract and whether both pathways are important in the natural chromatin context.
During the female sexual cycle in humans and rodents, an E 2 surge promotes and a P 4 surge inhibits epithelial cell proliferation in the reproductive tracts [10]. The uterine and vaginal epithelia consist of columnar www.impactjournals.com/oncotarget and squamous epithelial cells, respectively. The cervical epithelium, however, is composed of columnar and stratified squamous epithelial cells [11]. The hormone− mediated regulation of epithelial cell proliferation in the reproductive tracts involves crosstalk between the stromal and epithelial compartment [12]. Epithelial ERα is dispensable for E 2 −induced proliferation of uterine columnar epithelial cells [13]. Stromal ERα is required for squamous epithelial cell proliferation in the cervix and vagina [14]. Epithelial PR is dispensable for P 4 −induced suppression of epithelial cell proliferation in the uterine tissue recombinants derived from neonatal mice [15]. P 4 suppresses apoptosis in the uterine epithelium, which is mediated by stromal PR [16]. While PR is expressed in the cervical stroma and epithelium [17], knowledge on the function of PR and effects of P 4 in the cervix is limited.
Using knock−in mice expressing mutant ERα lacking activities of DBD, AF1, or AF2, we show that all three domains are required for upregulation of Pgr in the cervical epithelium and vagina. We also show that AF1 is dispensable for upregulation of Pgr in the cervical and uterine stroma. Unlike the uterus, P 4 promotes epithelial apoptosis in the cervix and vagina. P 4 inhibits E 2 −induced cervical and vaginal squamous cell proliferation. Using epithelial Pgr−deficient mice, we demonstrate that epithelial PR is required for P 4 −induced apoptosis and suppression of cell proliferation in the cervical and vaginal epithelium. Our results are the first to show different functions and regulation of epithelial PR in the cervix, vagina, and uterus under the same hormonal condition.

Histological features of the female reproductive tract lacking epithelial PR
To interrogate a role of epithelial PR in responses of the female reproductive tract to P 4 , we made use of Wnt7a−Cre/Pgr f/f mice (referred to as Pgr ed/ed hereafter; ed, epithelial deletion). In these mice, PR expression was ablated in the epithelium, but not stroma, of the cervix, vagina, and uterus ( Figure 1A). To characterize epithelial PR functions under the same hormonal condition, we treated ovariectomized mice with E 2 and P 4 for 7 days. Seven−day treatment with E 2 was required for the entire cervical epithelia to reach the full thickness (Supplementary Figure S1). Regions of the uterus, cervix, and vagina that were analyzed throughout the study are shown in Figure 1B. The cervical epithelia of vehicle− treated Pgr f/f and Pgr ed/ed mice were thin (13.2 μm in average) and composed of 2−3 cell layers ( Figure 1C). E 2 increased the height of the cervical epithelium in Pgr f/f (65.8 ± 4.2 μm) and Pgr ed/ed mice (66.6 ± 2.9 μm) ( Figure  1C). The cervical epithelium height in E 2 +P 4 −treated Pgr f/f (57.5 ± 5.0 μm) and Pgr ed/ed mice (60.7 ± 4.6 μm) was not significantly different from genotype−matching mice treated with E 2 alone. The nucleus/cytoplasm ratio (0.66 ± 0.03) of cervical suprabasal cells in E 2 −treated Pgr f/f mice was significantly smaller than the ratio (1.09 ± 0.06) in E 2 +P 4 −treated mice (P = 0.05). A similar difference was observed in Pgr ed/ed mice [0.66 ± 0.02 (E 2 ) vs. 1.03 ± 0.01 (E 2 +P 4 ); P = 0.05]. Hydropic degeneration was prominent in the cervix of E 2 +P 4 −treated Pgr f/f but not Pgr ed/ed mice ( Figure 1C). E 2 induced hyperplasia and keratinization in the vaginal epithelium of Pgr f/f and Pgr ed/ed mice ( Figure  1D). In the vaginal epithelium of E 2 +P 4 −treated Pgr f/f mice, keratinization was absent and hydropic degeneration was observed ( Figure 1D). Also found was infiltration of K14−negative non−epithelial cells (Supplementary Figure S2), which are likely neutrophils [18]. These P 4 − induced histologic changes in the vaginal epithelium were absent in E 2 +P 4 −treated Pgr ed/ed mice ( Figure 1D). E 2 induced hyperplasia in the uterine epithelium, which was reversed by P 4 in both Pgr f/f and Pgr ed/ed mice ( Figure  1E). Phenotypes described in this study are summarized in Table 1. P 4 fails to inhibit epithelial cell proliferation in the cervix and vagina, but not uterus of Pgr ed/ed mice P 4 inhibits epithelial cell proliferation in the female reproductive tract [15,19]. We sought to determine whether epithelial PR is required for this effect. We analyzed expression of cell proliferation marker Ki67 to measure cell proliferation. Less than 1% of cervical epithelial cells were proliferative in vehicle−treated Pgr f/f and Pgr ed/ed mice (Figure 2A-2B). E 2 −induced cervical epithelial cell proliferation was not different between the two genotypes. Compared to genotype−matching mice treated with E 2 alone, cervical epithelial cell proliferation www.impactjournals.com/oncotarget in the basal layer was significantly decreased in E 2 +P 4 − treated Pgr f/f but not Pgr ed/ed mice (Figure 2A Table 1). In the uteri of both Pgr f/f and Pgr ed/ed mice, E 2 −induced epithelial cell proliferation was significantly inhibited by P 4 ( Figure 2E-2F). In both genotypes, however, stromal cell proliferation was increased in the uteri of E 2 +P 4 −treated mice compared to E 2 −treated mice ( Figure 2F). Similar results were obtained in the columnar epithelium of the cervix (data not shown). These results indicate that, under the same hormonal condition, epithelial PR is required for P 4 −induced suppression of cell proliferation only in the cervical and vaginal squamous epithelium.

Epithelial PR is required for proper differentiation of cervical and vaginal squamous epithelial cells
Histological features observed in the cervical and vaginal epithelium (see Figure 1C-1D) suggested that P 4 regulates differentiation of squamous cells. Thus, we carried out Alcian blue staining and immunohistochemistry for cytokeratin 10 (K10), squamous differentiation marker [20]. Alcian blue staining is typically used to detect cervical acidic mucins, which are expressed by differentiated cells [21]. As expected, E 2 induced expression of K10 in the suprabasal layer in the cervical epithelia of Pgr f/f and Pgr ed/ed mice ( Figure 3A). K10 expression was reduced in the cervical epithelia of E 2 +P 4 − treated Pgr f/f and Pgr ed/ed mice, indicating that epithelial PR is dispensable for K10 suppression by P 4 . Alcian blue staining was prominent in the cervix of Pgr f/f mice treated with E 2 +P 4 , but absent in identically treated Pgr ed/ed mice ( Figure 3B). This result indicates that P 4 −promoted Alcian blue staining is dependent upon epithelial PR. P 4 inhibited E 2 −induced K10 expression and increased Alcian blue staining in the vaginal epithelium of Pgr f/f but not Pgr ed/ed mice ( Figure 3C-3D and Table 1). These results indicate that epithelial PR is required for P 4 −mediated regulation of squamous cell differentiation in the vagina to the greater extent than in the cervix.

Figure 2: Epithelial PR is required for P 4 to inhibit epithelial cell proliferation in the cervix and vagina, but not uterus.
Ovariectomized mice were treated with female sex hormones as indicated. Paraffin sections of the cervix A., vagina C., and uterus E. were stained for Ki67 (green). Nuclei were stained with Hoechst 33342 (blue). Scale bars, 30 μm. Results were quantified as described in Materials & Methods. Quantified results of the cervix B., vagina D., and uterus F. are shown as mean ± S.E.M (n = 3-5 per group). * P ≤ 0.01, ** P ≤ 0.05. www.impactjournals.com/oncotarget Epithelial PR plays distinct roles in epithelial apoptosis in the lower and upper reproductive tracts P 4 inhibits apoptosis in the uterine epithelium [16]. We carried out TUNEL assays to determine the role of P 4 and epithelial PR in regulation of apoptosis in the cervix, vagina, and uterus. A few TUNEL−positive apoptotic cells were found in the cervical epithelium of vehicle− and E 2 − treated Pgr f/f and Pgr ed/ed mice ( Figure 4A-4B). Compared to E 2 −treated mice, the apoptotic index was increased in the cervical epithelium of E 2 +P 4 −treated Pgr f/f but not Pgr ed/ed mice ( Figure 4A-4B). Similar results were obtained in the vaginal epithelium ( Figure 4C-4D and Table 1). E 2 induced apoptosis in the uterine epithelium regardless of the genotype, which was inhibited by P 4 in both Pgr f/f and Pgr ed/ed mice ( Figure 4E-4F and Table 1). These results indicate that epithelial PR is required for P 4 −induced apoptosis in the epithelium of lower reproductive tracts (i.e., cervix and vagina), but dispensable for P 4 to inhibit apoptosis in the uterine epithelium.

ERα DBD and AF2 are required for upregulation of Pgr in the cervix and vagina
ERα regulates transcription of Pgr in the vagina and uterus [5]. We found that E 2 increased the Pgr mRNA level significantly in the cervix of Esr1 +/+ but not Esr1 −/− mice ( Figure 5A). Similarly, PR protein was readily detected in cervical tissue extracts and paraffin sections derived from Esr1 +/+ but not Esr1 −/− mice ( Figure 5B-5C). These results demonstrate that ERα activates transcription of Pgr in the cervix. To understand a mechanism of transcriptional activation of Pgr by ERα, we made use of knock−in mouse models expressing ERα mutant defective for DBD, AF1, or AF2 ( Figure 5D). The AA allele expresses an ERα mutant (E207A/G208A) that lacks the ERE−binding activity [22]. The AF2 allele expresses ERα harboring L543A/L544A substitution mutations in the helix 12, which abrogates the AF2 function [23]. The neo allele has a neo cassette in the exon 2, which is designed to abrogate expression of ERα [24]. This allele, however, expresses truncated ERα proteins lacking the AF1 domain due to cryptic splicing events [25,26]. PR expression was Reproductive tracts were isolated from female mice that were ovariectomized and then treated as indicated. Representative images of each group are shown (n = 3). A. Epithelial PR is dispensable for suppression of K10 expression by P 4 in the cervical epithelium. Cervical sections were stained for K10 (green) and nuclei were stained with Hoechst 33342 (blue). Dotted lines separate epithelium (ep) and stroma (st). Scale bar, 50 μm. B. Alcian blue staining is increased by P 4 in an epithelial PR−dependent manner in the cervical epithelium. Cervical sections were stained with Alcian blue. Nuclei were counterstained with Nuclear Fast Red. Scale bar, 50 μm. C. Epithelial PR is necessary for suppression of K10 expression by P 4 in the vaginal epithelium. Vaginal sections were stained for K10 (green) and nuclei were stained with Hoechst 33342 (blue). Scale bar, 50 μm. D. Alcian blue staining is increased by P 4 in an epithelial PR−dependent manner in the vaginal epithelium. Vaginal sections were subjected to Alcian blue staining. Nuclei were counterstained with Nuclear Fast Red. Scale bar, 50 μm. www.impactjournals.com/oncotarget barely detectable in cervical tissue extracts obtained from Esr1 AA/− and Esr1 AF2/AF2 mice ( Figure 5B). Similarly, E 2 failed to induce PR expression in the cervix of Esr1 AA/− and Esr1 AF2/AF2 mice ( Figure 5C). Albeit reduced compared to Esr1 +/+ mice, the PR level was increased in cervical tissue extracts from Esr1 neo/neo mice compared to Esr1 −/− , Esr1 AA/− , and Esr1 AF2/AF2 mice ( Figure 5B). Upregulation of PR expression in the cervix of Esr1 neo/neo mice was mostly  Upregulation of PR expression is absent in the cervix of Esr1 AA/− and Esr1 AF2/AF2 mice. Ovary−intact mice of indicated genotypes (8-12 weeks of age) were treated with E 2 for 3 days to synchronize mice at the estrus. Expression of PR in the cervix was determined by immunoblot. Pgr −/− tissue extracts were used as negative control. GAPDH was used as loading control. Results were taken from separate gels indicated by black boxes. Intervening lanes in a same gel were removed, which was indicated by vertical white lines. Representative results from more than 2 mice per genotype are shown. C. E 2 −induced PR expression is absent in the cervix of Esr1 AA/− and Esr1 AF2/AF2 mice. Ovariectomized mice (n = 3 per group) were treated with vehicle or E 2 for 7 days. Cervical tissue sections were stained for PR (green). Nuclei were stained with Hoechst 33342 (blue).  Figure S4), indicating that ERβ is not required for upregulation of Pgr. To determine whether stroma−epithelium cross talk is involved in upregulation of Pgr by E 2 −ERα, we used Wnt7a−Cre/Esr1 f/f (referred to as Esr1 ed/ed hereafter). ERα expression was efficiently ablated in epithelial but not stromal cells in the cervix and vagina of Esr1 ed/ed mice ( Figure 5E). While E 2 induced PR expression in the cervical stroma of Esr1 f/f and Esr1 ed/ed mice, E 2 failed to do so in the cervical epithelium of Esr1 ed/ed mice ( Figure  5F). Similar results were obtained from the vagina ( Figure  5G). These results suggest that, in the cervix and vagina, E 2 −induced upregulation of Pgr is cell−autonomous and mediated commonly by ERα DBD (i.e., ERE−binding) and AF2.

ERα DBD and AF2 are required for upregulation of stromal Pgr and downregulation of epithelial Pgr in the uterus
Consistent with previously published results [13], expression of ERα was lost in the epithelium but retained in the stroma of the Esr1 ed/ed uteri ( Figure 6A). While stromal PR expression was increased by E 2 in the Esr1 f/f and Esr1 ed/ed uteri, E 2 decreased the PR levels in the uterine epithelium in both genotypes ( Figure 6B). This result confirmed the paracrine mechanism of Pgr downregulation by ERα in the uterine epithelium [5,27]. Stromal upregulation and epithelial downregulation of Pgr by E 2 were absent in the uterus of Esr1 −/− , Esr1 AA/− , and Esr1 AF2/AF2 mice ( Figure 6C). As reported previously [5], E 2 activated stromal Pgr expression but failed to downregulate epithelial Pgr in the uterus of Esr1 neo/ neo mice ( Figure 6C). In the Esr2 −/− uterus, E 2 increased and decreased stromal and epithelial PR expression, respectively (Supplementary Figure S4), indicating that ERβ is not required for regulation of Pgr in the uterus. These results indicate that the function of ERα DBD and AF2 in stromal cells is required for stromal upregulation and epithelial downregulation of Pgr in the uterus.
We next analyzed binding of ERα to the Pgr gene using the previously published ERα ChIP−seq data [28,29]. Approximately 200 kb upstream and 100 kb downstream of the Pgr transcription start site (TSS) were analyzed for the binding of wild−type ERα. While there were two E 2 −independent ERα−binding peaks at the 3' end of the Pgr gene, E 2 enhanced ERα binding to the intragenic regions and −72 kb and −155 kb cluster in the Esr1 +/+ uterus ( Figure 7A). E 2 also increased RNA polII binding to the Pgr gene in the same tissue. While the −155 kb cluster was closer to AK054106, E 2 did not regulate this gene (data not shown). Notably, we identified predicted half or full EREs in all ERα−binding sites ( Figure 7A). Consistently, in the uterus of Esr1 AA/− mice, both E 2 − dependent and −independent ERα binding to the Pgr locus were not observed ( Figure 7A). Concordantly, E 2 −induced recruitment of RNA polII was absent in the Esr1 AA/− uterus. We confirmed by ChIP followed by qPCR that E 2 induced enrichment of ERα binding to the −62 kb region (7.1-fold) and intron 3-4 (6.2-fold) of Pgr in the uterus of Esr1 +/+ but not Esr1 AA/− mice ( Figure 7B). These results suggest that, in the uterus, ERα activates transcription of Pgr mainly by binding to EREs. DISCUSSION P 4 inhibits E 2 −promoted epithelial cell proliferation in the murine female reproductive tract [19,30]. We showed that, under the same hormonal condition, epithelial PR was necessary for P 4 to suppress E 2 −induced epithelial cell proliferation in the cervix and vagina, but dispensable in the uterus (Figure 2). It is shown that epithelial PR is required for P 4 −mediated inhibition of uterine epithelial cell proliferation [30]. This discrepancy may be due to differences in treatment design (P 4 for 3 days and then E 2 +P 4 for 1 day vs. E 2 +P 4 for 7 days), E 2 doses (50 ng vs. 1 μg), mouse genetic background (B6.SJL.129 mixed vs. FVB) and/or housing environment. We note that 7−day treatment with E 2 and P 4 might be pharmacologic rather than physiologic. Nonetheless, we believe our results are biologically relevant because P 4 −mediated inhibition of uterine epithelial proliferation during pregnancy depends on Hand2 transcription factor [31]. Expression of Hand2 depends on P 4 and PR and is restricted in the uterine stroma [31]. P 4 inhibited E 2 −induced apoptosis in the uterus of both Pgr f/f and Pgr ed/ed mice ( Figure 4E-4F). It is consistent with that P 4 prevents epithelial cell apoptosis in the tissue recombinant composed of Pgr +/+ stroma and Pgr −/− epithelium [16]. E 2 induces apoptosis in the uterine epithelium of Esr1 f/f and Esr1 ed/ed mice [13]. These results suggest that stromal ERα promotes and stromal PR inhibits apoptosis of uterine epithelial cells. One possibility is that stromal PR activates expression of a paracrine factor(s) that promotes cell survival. PR interacts with ERα and modulates the function of ERα in breast cancer cells [32]. Another possibility is that PR interacts with ERα in uterine stromal cells and inhibits ERα−mediated expression of a pro−apoptotic secretory factor(s). In the rodent uterus, E 2 −induced hyperplasia is eliminated by apoptosis [33]. Perhaps, E 2 −induced apoptosis is to balance out E 2 −induced hyperproliferation of uterine epithelial cells. Consistent with this idea, the apoptotic index was correlated with the proliferative index in the uterine epithelium ( Figure 2F and 4F). Unlike the uterus, P 4 promoted apoptosis in the cervical and vaginal epithelium, which was dependent on epithelial PR ( Figure  4A-4D). Epithelial PR was necessary for P 4 to increase the ratio of nucleus to cytoplasm and suppress expression of K10 in the vagina but not cervix (Table 1). These results reveal that epithelial PR is differentially required for P 4 responses of epithelial cells in the cervix, vagina, and uterus.
Using mouse genetic models, we determined that ERE binding (i.e., classical mechanism) of ERα is required for upregulation of Pgr in the cervix, vagina, and uterus ( Figure 5B-5D, 6C, 7 and Supplementary Figure S3). The ERα DBD mutant (E207A/G208A) has acquired the capability to bind hormone response elements such as progesterone response element [29]. We do not believe that the gain−of−function compromises our conclusion because this ERα mutant did not bind to the Pgr locus (Figure 7). It is possible that the ChIP− seq analyses might have failed to identify weak ERα− binding sites; thus, we cannot rule out a possibility that the ERα tethering mechanism also contributes to transcription of Pgr in the female reproductive tract. Knock−in mice expressing an ERα mutant defective for the tethering mechanism would be required to test this. Reporter assays have shown that AP-1 and Sp-1 sites at the promoter downstream are required for ERα−mediated PGR activation in MCF7 breast cancer cells [7,8]. In the uterus, strong E 2 −induced DBD−dependent ERα binding occurred mostly at far upstream ( > 60 kb) of the Pgr gene ( Figure 7A). Genome−wide enrichment of ERα binding to promoter−distal regions has been identified [28,34]. ERE−dependent long−range activation of PGR by ERα has been demonstrated in MCF7 cells [6]. It is likely that ERα binding to Pgr in the uterus occurs in stromal cells because epithelial ERα was not required for upregulation of stromal Pgr and downregulation of epithelial Pgr ( Figure 6B). We postulate that similar long−range regulation by ERα is involved in transcriptional activation of Pgr in the cervix and vagina. The AF2 domain was also required for upregulation of Pgr by ERα in the cervix, vagina, and uterus ( Figure 5B-5D, 6C and Supplementary Figure S3). Upregulation of Pgr was evident in the cervical and uterine stroma, but not vagina of Esr1 neo/ neo mice ( Figure 5C, 6C and Supplementary Figure S3). In tissue recombinants derived from Esr1 neo/neo mice, PR expression is induced by E 2 in the uterine but not vaginal stroma [5]. These results indicate that (1) ERα DBD and AF2 are commonly required for activation of Pgr in the female reproductive tracts and (2) AF1 is required for upregulation of Pgr only in the vaginal and cervical epithelium and vaginal stroma. Differential requirement of ERα domains for activation of Pgr may be due to distinct chromatin structures in different tissues and cell types, which may require different coactivators [2,35]. E 2 suppressed expression of Pgr in the uterine epithelium via stromal ERα ( Figure 6B), which is in agreement with other findings [5,13]. Downregulation of Pgr in the uterine epithelium involves transcriptional repression [5,27]. It is unlikely that the non−cell autonomous downregulation of Pgr is through a cell− cell contact because epithelial and stromal cells are physically separated by the basement membrane. Thus, we postulate that a paracrine factor(s) produced by stromal cells is responsible for repression of Pgr in the uterine epithelium. ERα DBD, AF1, and AF2 were required for downregulation of Pgr in the uterine epithelium ( Figure  6C), further suggesting that ERα DBD, AF1, and AF2 are necessary for regulating expression of such a factor(s). The AP-1 site at the +745 bp region of PGR has been implicated in repression of PGR in MCF7 cells [9]. It is plausible that a paracrine factor activates AP-1, which in turn suppresses transcription of Pgr in the uterine epithelium.
Our results demonstrate distinct functions of epithelial PR in the cervix, vagina, and uterus, suggesting that PR regulates unique sets of genes and pathways in these tissues. Warranted are further studies to identify PR target genes and pathways in the stroma and epithelium of these tissues. Our results also demonstrate unique mechanisms of Pgr regulation in the cervix, vagina, and uterus. Together, our results underscore the complexity of function and regulation of Pgr in the female reproductive tracts.

Ethics statement
Investigation has been conducted in accordance with the ethical standards and according to national and international guidelines. All procedures were carried out according to animal protocols approved by the University of Houston Institutional Animal Care and Use Committee.