Different Cre systems induce differential microRNA landscapes and abnormalities in the female reproductive tracts of Dgcr8 conditional knockout mice

Abstract Objectives The female reproductive tract comprises several different cell types. Using three representative Cre systems, we comparatively analysed the phenotypes of Dgcr8 conditional knockout (cKO) mice to understand the function of Dgcr8, involved in canonical microRNA biogenesis, in the female reproductive tract. Materials and Methods Dgcr8 f/f mice were crossed with Ltf icre/+, Amhr2 cre/+ or PR cre/+ mice to produce mice deficient in Dgcr8 in epithelial (Dgcr8 ed/ed), mesenchymal (Dgcr8 md/md) and all the compartments (Dgcr8 td/td) in the female reproductive tract. Reproductive phenotypes were evaluated in Dgcr8 cKO mice. Uteri and/or oviducts were used for small RNA‐seq, mRNA‐seq, real‐time RT‐PCR, and/or morphologic and histological analyses. Result Dgcr8 ed/ed mice did not exhibit any distinct defects, whereas Dgcr8 md/md mice showed sub‐fertility and oviductal smooth muscle deformities. Dgcr8 td/td mice were infertile due to anovulation and acute inflammation in the female reproductive tract and suffered from an atrophic uterus with myometrial defects. The microRNAs and mRNAs related to immune modulation and/or smooth muscle growth were systemically altered in the Dgcr8 td/td uterus. Expression profiles of dysregulated microRNAs and mRNAs in the Dgcr8 td/td uterus were different from those in other genotypes in a Cre‐dependent manner. Conclusions Dgcr8 deficiency with different Cre systems induces overlapping but distinct phenotypes as well as the profiles of microRNAs and their target mRNAs in the female reproductive tract, suggesting the importance of selecting the appropriate Cre driver to investigate the genes of interest.


MicroRNAs (miRNAs) are evolutionarily conserved small non-coding
RNAs that function in RNA silencing and post-transcriptional regulation of gene expression. miRNAs also regulate various cellular pathways necessary for the development and proper functions of organs, such as the oviduct and uterus, in the female reproductive tract. 1,2 DGCR8 is an RNA-binding protein that works with DROSHA to produce precursor miRNA in the nucleus, while DICER generates mature miRNAs and endogenous small interfering RNAs in the cytoplasm. 3 To study the function of miRNAs, especially canonical miRNAs in the female reproductive tract, we generated Dgcr8 f/f ; progesterone receptor (PR) cre/+ (Dgcr8 td/td ) mice and reported that canonical miRNAs are essential for uterine morphogenesis and physiology, including natural immune modulation. 4 PR cre/+ (PR-Cre) mice have been mostly used to study uterine biology during pregnancy and various diseases. However, PR-Cre inactivates genes not only in the female reproductive tract but also in other progesterone-responsive organs, including the ovary, pituitary gland and mammary gland. 5 In the uterus, PR is spatiotemporally expressed in all the major uterine compartments: myometrium, stroma, and epithelium. Furthermore, PR is also expressed in immune cells, such as natural killer (NK) cells, 6 macrophages, 7 dendritic cells 8 and T cells, 9 suggesting that PR-Cre may affect various immune cells as well as all the major uterine cells in a spatiotemporal manner.
In addition to PR-Cre, other Cre mice with unique purposes are currently available for conditionally inactivating gene(s) of interest in the female reproductive tract, especially in the uterus. Anti-mullerian hormone receptor type 2 (Amhr2)-Cre mice are mainly used to target genes in stromal and myometrial compartments of the uterus and oviduct as well as of the ovary. 10 Temporally, Cre action starts from midgestational embryo development (embryonic day 12.5) under the control of the Amhr2 promoter in Amhr2-Cre mice. Recently, other Cre mice, such as lactoferrin (Ltf)-iCre, small proline-rich protein 2f (Sprr2f)-Cre, and Wnt family member 7a (Wnt7a)-Cre, were generated to target genes in the epithelial compartment. The spatiotemporal actions of each Cre on the uterine epithelium are unique. Although Wnt7a-Cre is expressed throughout the epithelium of the prenatal Müllerian tract, 11 Ltf and Sprr2f are not expressed in the immature mouse uterus, but robustly expressed in the uterine epithelium of adult mice. 12,13 Ltf and Sprr2f, well-known oestrogen-responsive genes, are expressed not only in the uterine but also in the oviductal epithelium after puberty. 13 Collectively, the spatiotemporal modes of each Cre system may provide diverse reproductive phenotypes. Thus, insights into the mode(s) of actions of multiple Cre systems are required to precisely delineate the functions of genes of interest in a cell type-specific manner in the female reproductive tract.

| Animals and genotyping of Dgcr8 cKO mice
All mice used in this study were maintained in accordance with the policies of the Institutional Animal Care and Use Committee (IACUC170174) of CHA University. Dgcr8 f/f mice were initially generated and provided by Dr Elaine Fuchs. 14 PR-Cre, Amhr2-Cre and Ltf-iCre mice were generously provided by Dr Francesco DeMayo, 5 Dr Richard Behringer 10 and Dr Sudhansu K. Dey, 12 respectively.
Genotyping PCR was performed using specific primers (Table S1) and genomic DNA extracts from mouse tail biopsies.

| Fertility analysis and preimplantation embryo culture
Dgcr8 td/td mice have an anestrous cycle; therefore, to induce ovulation, Dgcr8 td/td mice were administered intraperitoneal (IP) injections of 5 IU PMSG (Sigma-Aldrich, St. Louis, MO, USA) for 48 hours, followed by IP injections of 5 IU hCG (Sigma-Aldrich). Dgcr8 td/td mice were then bred with wild-type fertile males, and pregnancy was evaluated by the presence of a vaginal plug the next morning. The other Dgcr8 cKO mice were naturally mated. The 2-cell embryos and/or fragmented oocytes were flushed from the oviducts on day 2 of pregnancy (Day 2).

| RNA extraction, reverse transcription-PCR (RT-PCR) and real-time RT-PCR
Total RNA was extracted from the mouse uterus using TRIzol Reagent (Invitrogen Life Technologies, San Diego, CA, USA) according to the manufacturer's protocols. First-strand cDNA was synthesized from 1 μg of total RNA using M-MLV reverse transcriptase (Promega, Madison, WI, USA) and RNasin Ribonuclease Inhibitor (Promega). For quantification of expression levels, real-time RT-PCR was performed using iQ™ SYBR Green Supermix (Bio-Rad, Hercules, CA, USA) on a BIO-RAD iCycler as previously described. 4 The synthesized cDNA was utilized for PCR and real-time RT-PCR with specific primers (Table S1).

| Tissue collection and histological analysis
Female reproductive organs were dissected and fixed in 4% paraformaldehyde (PFA) for histology or snap-frozen for RNA and/or protein preparation. Tissues were embedded in paraplast (Leica Biosystems, St. Louis LLC, Diemen, the Netherlands). Sections were cut at 5 μm and stained with haematoxylin and eosin (H&E) (Sigma-Aldrich), and observed by light microscopy.

| Immunofluorescence
Sections were subjected to antigen retrieval in 10 mM sodium citrate buffer (pH 6.0) for 20 min. Non-specific staining was blocked using

| Library preparation for small RNA sequencing (RNA-seq) and miRNA expression analysis
For control and test RNAs, the construction of the library was performed using the NEBNext Multiplex Small RNA Library Prep kit (New England BioLabs, Inc, USA) according to the manufacturer's instructions. Briefly, for library construction, total RNA from each sample was used 1 μg to ligate the adapters, and then, cDNA was synthesized using reverse transcriptase with adaptor-specific primers. To quantify the miRNA expression levels, total RNA (1 μg) was converted to cDNA, and real-time RT-PCR (50 ng of cDNA) was performed following the protocol from the HB miR Multi Assay Kit (Heim Biotek, Gyeonggi-do, Korea).

| Cell culture, transfection and luciferase assay
293T cells were grown in high glucose DMEM supplemented with 10% FBS, penicillin at 37°C, and 5% CO 2 . 293T cells were transiently

| Statistical analysis
All values represent the mean ± standard deviation. Statistical analyses were performed using the unpaired Student's t tests and P < 0.05 was considered statistically significant for more than 3 groups.

| Multiple Cre systems effectively delete Dgcr8 in the female reproductive tract
The uterine cell type-specific actions of the three representative Cre systems used in this study were summarized based on previous reports ( Figure 1A). To understand the temporal activity of Cre drivers in the female reproductive tract, the expression profiles of Amhr2, Pgr and Ltf were first examined during postnatal days (PND), oestrous cycle and early pregnancy ( Figure 1B-D and Figure   S1). RT-PCR results showed that Amhr2 is highly expressed in the developing uterus at PND 0, but maintained at undetectable levels during the oestrous cycle and early pregnancy. Pgr expression was very low at PND 0 and 3 but increased after PND 7 in the uterus.
Ltf mRNAs were detected in the mouse uterus at PND 28 (4 weeks of age). During the oestrous cycle, Pgr and Ltf showed stage-specific expression patterns. During early pregnancy, Ltf expression with a peak level on Day 1 gradually decreased onward, whereas Pgr expression was low on Days 1 and 2, followed by substantial increases on Days 3-5.
To validate the actions of each Cre system in the female reproductive tract of Dgcr8 cKO mice, Dgcr8 f/f mice were crossed with Ltf-iCre, Amhr2-Cre and PR-Cre mice to produce Dgcr8 ed/ed (epithelium-specific), Dgrc8 md/md (mesenchyme-specific) and Dgrc8 td/td (all the major uterine cell types) mice, respectively. Cre-mediated deletion of exon 3 of the Dgcr8 allele produced a 262 bp PCR product (white arrowhead), whereas a floxed allele resulted in a 1085 bp product (black arrowhead) ( Figure 1E). Consistent with a previous report 12, 16 that Ltf is expressed in the epithelium of the oviduct and uterus and in some immune cells, PCR results using genomic DNA of tissues from Dgcr8 ed/ed mice showed a deletion of exon 3 in the oviduct and uterus, but not in other tissues. Dgcr8 md/md mice showed deletion of Dgcr8 only in the ovary, oviduct and uterus, among all the tested tissues. PR-Cre activity was detected not only in the female reproductive tract but also in the pituitary of Dgcr8 td/ td mice.

| Dgcr8 cKO mice crossed with different Cre systems show a distinct spectrum of fertility
To compare the fertility of Dgcr8 cKO female mice with different Cre systems, they were mated with mature fertile males for 8-10 weeks ( Figure 1F). Dgcr8 td/td female mice never produced any litter, as we previously reported. 4 However, Dgcr8 md/md mice were sub-fertile and Dgcr8 ed/ed mice were normal with respect to the number of pups produced. By monitoring oestrous cycles with daily vaginal smears over a 2-week period, we observed that Dgcr8 td/td female mice were anestrus, whereas Dgcr8 md/md and Dgcr8 ed/ed mice exhibited regular 4-5 day oestrous cyclicity similar to that of Dgcr8 f/f control mice ( Figure 1G). 3.3 | Dgcr8 deficiency in the oviduct affects quality of ovulated oocytes followed by fertilization in Dgcr8 md/md mice To explore the underlying causes of sub-fertility in Dgcr8 md/md mice, we examined in detail the reproductive phenotypes during early pregnancy. Since Dgcr8 td/td mice are anovulatory due to pituitary defects, the numbers and fertilization rates of ovulated metaphase II (MII) oocytes were examined only in Dgcr8 md/md and Dgcr8 ed/ed mice.
The number of ovulated MII oocytes was not statistically different between genotypes, although there was a moderate reduction in Dgcr8 ed/ed and Dgcr8 md/md mice ( Figure 2A). However, the fertilization rate was significantly reduced in oocytes from Dgcr8 md/md mice ( Figure 2B). When 2-cell embryos harvested from the Dgcr8 md/md oviduct were cultured in vitro, they developed to the blastocyst stage similar to that of Dgcr8 f/f and Dgcr8 ed/ed mice ( Figure 2C). We then investigated the quality and quantity of the oocyte right after ovulation at post-hCG 16 hours. As shown in Figure 2D, the quantity and quality of oocytes were similar to those of Dgcr8 f/f mice immediately after ovulation. This is consistent with the fact that zona pellucida remnants and degenerated oocytes were often observed from the oviducts of Dgcr8 md/md mice on Day 2 ( Figure 2E-F). Furthermore, histological observation of the ovaries of Dgcr8 md/md mice indicated that ovulation was not affected in these mice ( Figure 2G).

| Deletion of Dgcr8 affects the uterine architecture with Cre-specific distinct spectrum
To examine whether Dgcr8 deficiency affects uterine development, we investigated the uteri of all the Dgcr8 cKO mice at 4 and 9 weeks of ages. The gross morphology and uterine weight to body weight ratio of 4-week-old Dgcr8 td/td mice were different from all the other genotypes. However, at 9 weeks of age, these defects were observed not only in Dgcr8 td/td mice, but also in Dgcr8 md/ md mice ( Figure 4A-B). Interestingly, a severely atrophic myometrium was observed only in adult Dgcr8 td/td mice ( Figure 4C-D). In general, gross histology and real-time RT-PCR analysis and/or immunostaining of cell type-specific markers showed that Dgcr8 ed/ ed and Dgcr8 md/md uteri were similar to those of Dgcr8 f/f mice

| mRNAs that control immune responses and negatively regulate smooth muscle cell proliferation were systemically upregulated in the uteri of Dgcr8 cKO mice
We then performed small RNA-seq and mRNA-seq for uteri of 4-week-old Dgcr8 td/td mice to elucidate the molecular mechanisms underlying the severe uterine phenotypes ( Figure 5). Uteri from 4-week-old Dgcr8 td/td mice were chosen because they showed the onset of multiple uterine defects, and PR was also expressed in all the major uterine cell types at this stage. 4 In mRNA-seq data, 573 and 424 genes with 1.5-fold cut-off values were upregulated and downregulated in Dgcr8 td/td mice, respectively ( Figure S2 and Table   S2). GSEA analyses showed the systemic upregulation of the gene sets associated with 'immune response', including leucocyte proliferation, migration, chemotaxis and blood vessel dilation ( Figure 5A, C). In addition, gene sets associated with 'negative regulation of smooth muscle cell proliferation and development' were upregulated ( Figure 5A, C). These results are consistent with the phenotypes in Dgcr8 td/td mice, such as acute inflammatory infiltration of immune cells in the female reproductive tract and severe atrophy in uterine smooth muscle (Figures 4 and 6). In contrast, gene sets involved in ribosome biogenesis, RNA methylation (RNA stability) 17 and negative regulation of T cell-mediated immunity were mainly downregulated in Dgcr8 td/td mice ( Figure 5B, D). Small RNA-seq data showed that 1035 out of the 1976 mouse miRNAs were detected in the uterus. However, only 32 and 27 miRNAs were down-and upregulated, respectively, with a 1.5-fold cut-off value in Dgcr8 td/td mice ( Figure S3 and Table S3-S4).

| Dgcr8 cKO mice provide distinct and overlapped target profiles in a Cre-dependent manner
To identify the direct target mRNAs of differentially expressed miR-NAs (DEMs) in the uteri of Dgcr8 td/td mice, we obtained a dataset of potential target genes of DEMs in Dgcr8 td/td mice using miRmap and compared this dataset with mRNA-seq data. We found that 208 upregulated and 132 downregulated genes were overlapped between both datasets ( Figure 6A). Since miR-149-5p, miR-29c-3p and miR-446b-3p miRNAs are representative of miRNAs with a Crespecific unique expression in the uterus, these miRNAs and their target mRNAs associated with 'immune response' and/or 'negative regulation of smooth muscle cell proliferation and development' were further evaluated ( Figure 6B). When luciferase constructs that included the 3' UTR of Cxcl12, Agtr2, Itga9 or Tspan2 mRNAs were co-transfected with miRNA mimics, the luciferase activity was significantly reduced ( Figure 6C). This suggests that the target mRNAs that control the immune response are directly regulated by the miRNAs.
We further examined and compared the expression profiles of these miRNAs and their target genes between Dgcr8 cKO mice  Figure 6G,H). Interestingly, the acute immune infiltration into the organs in the female reproductive tract was persistently observed in the Dgcr8 td/td uterus, but not in other genotypes ( Figure S4).

| D ISCUSS I ON
The uterus is a complex organ that consists of three major tissue compartments: myometrium, stroma and epithelium, with dynamic changes in various immune cells during the reproductive cycle. In this aspect, PR is a good Cre driver given its expression in all the major uterine cell types, and thus, PR-Cre mice have been exploited for gene deletion studies in all the major uterine compartments. 5,18 Recently, we also demonstrated that canonical miRNAs are essential for uterine physiology and fertility using Dgcr8 td/td mice. 4 However, PR-Cre mice may display compound phenotypes in multiple cell types. Thus, comparative analyses of all the three Dgcr8 cKO mice in this study will improve the understanding of the spatiotemporal action of each Cre system in the female reproductive tract. When Dgcr8 is deleted by Amhr2-Cre, defects in the oviduct and uterus were observed before and after puberty, respectively (Figures 3   and 4). These results suggest that phenotypes are exposed much later than the onset of Cre expression and/or follow a tissue-specific expression in Dgcr8 md/md mice. Unlike Dgcr8 td/td mice that suffer from infertility and uterine deformities, Dgcr8 ed/ed and Dgcr8 md/ md mice showed regular oestrous cycles, showed normal architecture, and produced pups, suggesting that the deletion of Dgcr8 in epithelial and mesenchymal cells does not significantly deteriorate fertility (Figure 1). However, considering that both PR and Amhr2 are expressed in the stroma and myometrium in the female reproductive tract, 19 it is noteworthy that Dgcr8 td/td , but not Dgcr8 md/md mice exhibited acute inflammation in all the tissues of the female reproductive tract (Figure 6 and Figure S4). These results suggest that immune cells deficient in canonical miRNAs may affect uterine development and physiology. In fact, PR is present in a variety of cell types, including immune cells, such as NK cells, 6 macrophages, 7 dendritic cells 8 and T cells, 9 and non-immune cells, such as neuronal cells. However, it was also reported that Ltf is expressed in some immune cells, especially neutrophils and macrophages. 12,16 Thus, deletion of Dgcr8 using immune-cell-specific Cre, such as macrophagespecific LyzM-Cre, 20 may provide clues to decipher these puzzled phenotypes that are observed in Dgcr8 td/td mice but not in others.

DGCR8 is required for the biogenesis of canonical miRNAs,
whereas DICER is indispensable in the production of both canonical and non-canonical miRNAs. Thus, it is expected that Dicer and Dgcr8 cKO mice would show overlapping phenotypes in the female reproductive tract. In fact, Dgcr8 md/md mice showed similar but less severe abnormalities compared to Dicer md/md mice. For example, Dgcr8 md/ md and Dicer md/md mice showed abnormally short oviducts, whereas Dgcr8 td/td and Dicer td/td mice did not display morphologic deformities in the oviduct (Figure 7). While embryos were trapped in the oviduct that often harboured cysts during early pregnancy in Dicer md/ md mice, 21-23 but not in Dgcr8 md/md mice (Figure 2), the increased Dgcr8 md/md mice had a normal uterine structure (Figure 4). In contrast, when PR-Cre was used, Dgcr8 td/td and Dicer td/td mice showed similar uterine phenotypes, such as decreased number of glands and atrophic stroma. 24 In addition, considering that Dgcr8 is an essential factor for the biogenesis of canonical miRNAs, it was quite surprising that only 32 miRNAs with a 1.5-fold change were downregulated in the uteri of Dgcr8 td/td mice ( Figure 6). Intriguingly, only a handful of miRNAs were downregulated in the uteri of Dicer td/td mice as well. 24 PR-Cre, Amhr2-Cre and Ltf-iCre mice have been employed to understand the functions of other genes important for uterine biology. For example, each Cre mouse was individually used to examine the function of Pten, a well-known tumour suppressor gene, in tumorigenesis of endometrial cancer. [25][26][27][28] Pten td/td mice showed rapid development of endometrial cancer with full penetration, whereas Pten md/md mice failed to initiate tumorigenesis. As Amhr2 is not expressed in the epithelial compartment of the uterus, it is reasonable that Pten deletion in the stroma and myometrium could not provoke endometrial cancer. 25,26 However, Pten ed/ed mice developed atypical epithelial hyperplasia but did not develop endometrial cancer, 27 suggesting that Pten signalling in the stroma restrains epithelial cell transformation from hyperplasia to carcinoma. Deletion of Tsc1, a direct inhibitor of mTORC1, in the female reproductive tract sterilized both Tsc1 td/td and Tsc1 md/md female mice, resulting from oviductal hyperplasia, retention of embryos in the oviduct, and implantation failure. 19 However, embryo development was disrupted in Tsc1 td/td , but not in Tsc1 md/md mice. 19 Collectively, these reports, as well as this study, suggest that selection of the Cre system leads to a differential spectrum of phenotypes in tissues with multiple cell types in the female reproductive tract.
Previously, we demonstrated that DGCR8-dependent canonical microRNAs are essential for uterine development and physiological processes such as proper immune modulation, reproductive cycle and steroid hormone responsiveness in mice. 4 Especially, we observed that an excessive influx of immune cells occurs in the ovary, oviduct and uterus of Dgcr8 td/td mice on Day 2. 4 In addition, percentage of CD45-positive cells were increased in the uterus of 4-week-old Dgcr8 td/td mice ( Figure 6G, H). These are consistent with the results of small RNA-and mRNA-seq, which revealed that miRNAs and their potential target mRNAs involved in immune responses were dysregulated in the uteri of Dgcr8 td/td mice. However, acute inflammation in Dgcr8 td/td mice was not observed in Dgcr8 md/ md and Dgcr8 ed/ed mice ( Figure 6 and Figure S4), suggesting that the profiles of dysregulated miRNAs could depend on the Cre system. Although many miRNAs are known to regulate immune responses, such as Toll-like receptor (TLR) signalling, 29 the functions of these miRNAs in the uterus are poorly understood. Luciferase assays validated that miRNAs directly inhibit Cxcl12, Itga9, Agtr2 and Tspan2 ( Figure 6C). ITGA9 plays a very important role in neutrophil migration, 30 and CXCL12 induces cell migration via the CXCR4/CXCR7 complex. 31-33 TSPAN2 induces M2 polarization in microglia, and AGTR2 has vasodilating and blood pressure-reducing effects. 34,35 Another unique phenotype observed in Dgcr8 td/td uteri was that the inner circular smooth muscle became atrophic at the adult stage ( Figure 4). The role of miRNAs in the proliferation of uterine smooth muscle cells is unknown, but Npr3 and Tpm1, potential target genes of miR-149-5p and miR-29c-3p, are known to negatively regulate smooth muscle cell proliferation. NPR3 increases F I G U R E 7 Phenotypic comparison of Dgcr8 cKO and Dicer cKO mice with three Cre systems. ━, Normal; ↓, Decrease; N/A, Not available endothelial cell proliferation and inhibits vascular smooth muscle growth via ERK 1/2 phosphorylation. 36 In addition, TPM1 inhibits vascular smooth muscle proliferation and migration progression via the HIF-1α/ miR-21 / TPM1 pathway. 37 Although the expression of miR-21 was not reduced in Dgcr8 td/td mice, it is thought that the expression could be sufficiently regulated by other miRNAs.
Thus, it is assumed that the downregulation of miRNAs maintains the increased levels of their target mRNAs, subsequently inducing leucocyte migration and differentiation, blood vessel dilation, and/ or suppression of smooth muscle proliferation. Collectively, differential phenotypes and landscapes of miRNAs and mRNAs in Dgcr8 cKO mice with different Cre systems suggest that selection of Cre is critical to understanding the function of the gene of interest in the female reproductive tract in a spatiotemporal manner.

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.