A novel Cytochrome P450 26A1 expressing NK cell subset at the mouse maternal‐foetal interface

Abstract Cyp26a1 had important roles in mouse embryo implantation and was highly expressed in some of NK cells at the human maternal‐foetal interface in early pregnancy. However, the regulatory effect of Cyp26a1 on NK cells remains poorly understood. Through qPCR and flow cytometric assays, we found that Cyp26a1 was expressed by mouse uterine NK cells but not spleen NK cells during the peri‐implantation period and there was a group of NK cells that highly expressed Cyp26a1, that is Cyp26a1+NK cell subset. single cell‐population transcriptome sequencing on Cyp26a1+NK and Cyp26a1−NK cell subsets was performed. We found that there were 3957 differentially expressed genes in the Cyp26a1+NK cell subset with a cut‐off of fold change ≥2 and FDR < 0.01, 2509 genes were up‐regulated and 1448 genes were down‐regulated in Cyp26a1+NK cell subset. Moreover, cytokine‐cytokine receptor interaction signalling pathway and natural killer cell–mediated cytotoxicity signalling pathway were enriched according to KEGG pathway enrichment analysis. We further found that the expression of Gzma and Klrg1 was significantly increased and Fcgr4 was significantly decreased when inhibiting Cyp26a1. Our experimental results show that there is a novel NK cell subset of Cyp26a1+NK cells in mouse uterus and Cyp26a1 can regulate the gene expression of Gzma, Klrg1 and Fcgr4 in the Cyp26a1+NK cells.

it into more polar hydroxylated and oxidized derivatives. [7][8][9] Mice of Cyp26a1-null mutants die in the late second trimester of pregnancy with many major morphogenesis defects. 10 Previous research in our laboratory showed that Cyp26a1 was specifically and highly expressed at the maternal-foetal interface in mouse early pregnancy; meanwhile, the pregnancy rate and the number of implantation sites were significantly decreased in Cyp26a1 plasmid-immunized or Cyp26a1-specific antisense oligos treated mice. 11 Further study of our laboratory found that the proportion of NK cells was significantly changed in the uterus in early pregnancy in the above two mouse models. 12 There was a report showing that Cyp26a1 was highly expressed in dNK1 at the human first-trimester maternal-foetal interface. 13 Moreover, Cyp26a1 was highly expressed in dNK cells (decidua NK cells) and lowly expressed in pNK cells (peripheral blood NK cells) of human. 14 However, it has not been reported whether the specific expression of Cyp26a1 in NK cells played a special regulatory role.
In this study, our research discovered that there was a Cyp26a1 + NK cell subset at the mouse maternal-foetal interface in early pregnancy and we performed preliminary research on this NK subset.

| Mice
BALB/c mice of 6-8 weeks were purchased from Beijing Vital River Laboratory Animal Technology Co, Ltd, and 8-12 weeks during the experiment. All animal manipulation procedures were approved by the Institutional Animals Care and Use Committee of the Institute of Zoology, Chinese Academy of Sciences. In the afternoon of the previous day, male mice and female mice were caged in a ratio of 1:1, and the vaginal plugs of female mice were detected in the next morning. The presence of vaginal plugs was recorded as the first day of pregnancy (gd1).

| RNA extraction and quantitative reverse transcription polymerase chain reaction (qPCR)
Total RNA of cells or tissues was extracted according to the TRIzol

| Western blotting
Grind mouse uterine tissue into powder in liquid nitrogen and then added to RIPA Lysis Buffer (C1053, APPLYGEN), simultaneously adding protease inhibitor cocktail (HX1863, huaxingbio). The total protein was extracted according to the instructions of the RIPA Lysis Buffer. The protein concentration was determined by BCA protein quantification kit (23227, Pierce) according to the instructions; then, SDS-PAGE was used to separate protein. The protein on the gel was transferred to the NC membrane. After blocking with 5% skimmed milk, incubating with primary antibody at 4°C overnight and secondary antibody at room temperature 1 hour, and then adding West Pico PLUS chemiluminescent substrate (34579, Thermo Fisher), immunoreactive protein bands were detected and imaging by Genegnome instrument. The primary antibodies used were anti-CYP26A1 (PA5-24602, Invitrogen, 1:1000) and anti-GAPDH (2118, CST, 1:1000).

| Preparation of mouse uterine single cell suspension
Referring to the method in the previous article with minor modifications, 15 the main steps were as follows. After the mice were sacrificed, the uterine tissues were cut out, washed once with 1xPBS, shredded in digestion mixture and digested at 37°C in a shaker for 30 minutes. The digestion mixture was prepared with RPMI Media 1640 and contained 1 mg/mL collagenase IV (C5138, Sigma-Aldrich), 0.3 mg/mL hyaluronidase (H3506, Sigma-Aldrich) and 8% FBS. After digestion, digestive fluid was removed by centrifuge, 1×PBS was added into tissue and cell mixture and incubate at 37°C in a shaker for 15 minutes, and then filter with a 400 mesh sieve to obtain a single cell suspension of uterus.

| Preparation of mouse spleen single cell suspension
Refer to the previously mentioned with small modifications. 16 The main method was as follows. A fresh spleen of mouse was put into 1×PBS and shredded to small pieces. Then through filter with a 100 mesh sieve. Added 3 volumes red blood cell lysate (40401ES60, YEASEN) to cell suspension and centrifuge for 5min, then discard the supernatant. Add 1×PBS to resuspend the cells and filter with a 400 mesh sieve. Spleen single cell suspension was obtained after washed again with 1×PBS.

| Immunohistochemistry
The operation method of paraffin section was the same as 2.9 Immunohistofluorescence, until the sections were incubated with rabbit anti-Cyp26a1 antibody (ab151968, abcam, 1:100) overnight.
The frozen sections (5 μm) of mouse (BALB/c strain) uterus of gd6 were mounted on 3-Aminopropyl-Triethoxysilane-coated slides. The frozen sections were treated with 0.3% Triton X-100 for 10 minutes and blocked in 5% BSA at room temperature for 1 hour and after the embedding agent was washed. Then, the frozen sections were incubated with goat anti-Nkp46 antibody (AF2225, R&D Systems, 1:50) overnight. The next day, the 3% H 2 O 2 was used to treat both of the paraffin sections and frozen sections for 15 minutes to inhibit endogenous peroxidase activity. After washing thoroughly in PBS, all the sections were incubated with HRP-conjugated secondary antibody (1:200) at room temperature for 1 hour. The colour was developed with diaminobenzidine tetrahydrochloride (DAB) kit (ZLI-9017, ZSGB-BIO). Images were gotten by the Nikon ECLIPSE Ni-U microscope. The purity of RNA was tested using NanoDrop 2000 micro-spectrophotometer. The concentration and integrity of RNA were tested using Agilent 2100 Bioanalyzer and Agilent RNA 6000 Pico Kit.

| Single cell-population transcriptome sequencing
Transcriptome library was constructed following SMART-seq2 protocol, and transcriptome sequencing data were obtained based on the illumina NovaSeq 6000 sequencing technology platform.
Sequencing reads were aligned to the mouse genomes (mm10) by using the HISAT2 (tophat) with the default parameters. Total read counts for each protein-coding gene were extracted using HTSeq (version 0.6.0) and then loaded into R package DESeq2 to calculate differentially expressed genes with cut-off of fold change ≥ 2 and FDR < 0.01. The GO and KEGG analyses of differentially expressed genes were performed using R package clusterProfiler.

| New Cyp26a1 + NK cell subset
3.1.1 | Expression pattern of Cyp26a1 in mouse uterine tissue during the peri-implantation period qPCR and Western blotting experiments were conducted to detect Cyp26a1 expression in mouse uterine tissue during the peri-implantation period. The results showed that Cyp26a1 was specifically and dynamically expressed in the peri-implantation in mouse uterine tissue, which was lowly expressed on gd4 before embryo implantation, highly expressed on gd5 and gd6 after embryo implantation, and then down-regulated after gd6 ( Figure 1A-B).

| Expression pattern of Cyp26a1 in mouse uterine NK cells
The results of qPCR on CD45 + CD122 + CD3 − NK cells isolated from the uterus and spleen on gd5 and qPCR on CD45 + CD122 + CD3 -NK cells isolated from the uterus on gd4-gd7 showed that Cyp26a1 was highly expressed in NK cells in the uterus and lowly expressed in NK cells in the spleen ( Figure 1C), in addition, the expression trend of Cyp26a1 in NK cells in the uterus appeared a dynamic change during the peri-implantation period, which is similar to Cyp26a1 expression pattern in uterine tissue ( Figure 1D).

| Grouping and verification of NK cells expressing Cyp26a1
In order to explore whether NK cells expressing Cyp26a1 could form a subset, we used anti-Cyp26a1 antibody to label uterine NK cells, and combined with Alexa Fluor 488 fluorescent secondary antibody to perform flow cytometric assays, what surprised us was that NK cells in the uterus on gd4-gd7 can be divided into two cell subsets of CD45 + CD122 + CD3 − Cyp26a1 + NK (Cyp26a1 + NK) and Figure 1E). The gating strategy and complete data were shown in Figure S1. The ratio of Cyp26a1 + NK was high on gd4 and then showed a downward trend ( Figure 1F). Meanwhile, we also found that dynamic change of CD45 + CD122 + CD3 -NK cells was similar the expression pattern of There were obvious green signals of Cyp26a1 on the Cyp26a1 + NK cell subset, while the Cyp26a1 − NK cell subset had no detectable green signals ( Figure 1G).

| Localization of Cyp26a1 + NK cells
In order to study the location of Cyp26a1 + NK cells in the uterus, the immunohistochemical experiment of Cyp26a1 and NKp46 localization on mouse gd6 uterine tissue using anti-Cyp26a1 or anti-NKp46 antibodies was performed ( Figure S2B). Meanwhile, Immunohistofluorescence experiments using anti-Cyp26a1 and anti-NKp46 antibodies co-stained were also performed to detect the co-localization of Cyp26a1 and NKp46 on mouse gd6 uterine tissue ( Figure 2 and Figure S2A).

| Single cell-population transcriptome sequencing and analysis of Cyp26a1 + NK and Cyp26a1 -NK cell subsets in uterus on gd5
To investigate the function of Cyp26a1 + NK cells, single cell-population transcriptome sequencing was performed on Cyp26a1 + NK and in Cyp26a1 + NK cell subset, and 1448 genes were down-regulated (Table S2).
Furthermore, we performed GO and KEGG analyses. According to the results of GO analysis, we found that the enriched biological processes include T cell activation, myeloid leucocyte activation and negative regulation of immune system process ( Figure 4A).
KEGG analysis showed that cytokine-cytokine receptor interaction, Fc gamma R-mediated phagocytosis and natural killer cellmediated cytotoxicity were enriched ( Figure 4B-C). Pf4 and Ccl12 were highly expressed in Cyp26a1 + NK cell subset, and Tnfrsf9 and Fasl were lowly expressed in Cyp26a1 + NK cell subset ( Figure 4D).

| Verification of single cell-population transcriptome sequencing
Cyp26a1 and several NK cell-related genes were chosen for qPCR experiments to verify the transcriptome sequencing results in Cyp26a1 + NK cells and Cyp26a1 − NK cells isolated from the mouse uterus on gd5. The results of qPCR showed that Cyp26a1 and Cx3cr1 mRNA were highly expressed in the Cyp26a1 + NK cell subset; however, FasL, Gzma, Klrd1, Klrg1 and prf1 were lowly expressed in the Cyp26a1 + NK cell subset ( Figure 5A). Furthermore, qPCR results were largely consistent with the single cell-population transcriptome sequencing results ( Figure 5B).

| Changes of NK cell function-related molecules after inhibiting Cyp26a1
To evaluate the regulatory effect of Cyp26a1 on NK cell functionrelated molecules, we isolated CD45 + lymphocytes from the uterus on gd5, and then treated with the Cyp26a1 inhibitor R115866 or Cyp26a1 antibody. Gzma and Klrg1 were up-regulated in both Cyp26a1 inhibitors R115866-treated and Cyp26a1 antibody-treated group when compared with the control (DMSO or IgG) group ( Figure 6A-B, Figure 6D-E). At the same time, we also found that after treating CD45 + lymphocytes with the Cyp26a1 antibody, the expression of Fcgr4 was down-regulated relative to the control group ( Figure 6F). However, Cyp26a1 inhibitor R115866 treatment did not cause changes in Fcgr4 in comparison with the control group ( Figure 6C).

| D ISCUSS I ON
We found that Cyp26a1 was dynamically and specifically expressed in the mouse uterine tissue during the peri-implantation period, and the expression level increased remarkably at D5 and then gradually declined ( Figure 1A-B). These observations were consistent with previous research in our laboratory. 11,20,21 We also found that Cyp26a1 was expressed in mouse uterine NK cells, but not in spleen NK cells ( Figure 1C), which was consistent with the report in human NK cells. 13,14 There were a variety of NK cell subsets at the human maternal-foetal interface, and different NK subsets had different gene expression profiles and functions. 13,22 Previous studies in our laboratory had shown that Cyp26a1 could affect pregnancy outcomes by regulating immune cells, such as regulating Th17 cells, affecting the iDC/mDC ratio and the ratio of NK cells. 12,23,24 In this study, we discovered that there was a    Data were means ± SEM from five biological replicates, * represents P < .05, ** represents P < .01, *** represents P < .001 and **** represents P < .0001 (two-tailed unpaired Student's t test) of total CD45 + lymphocytes and the genes we tested were highly expressed in NK cells. Figure 1A, Figure 1B and Figure 1D showed the expression of Cyp26a1 was low on gd4, and Figure 1E and Figure 1F showed that the population of Cyp26a1 + NK cell was the high. This may be due to the following reasons: Figure 1A and Figure 1B detects the expression of Cyp26a1 in all cells of the uterus. Figure 1D detects the expression of Cyp26a1 in total NK cells, Figure 1E and Figure 1F shows the proportion of Cyp26a1 + NK cells in CD45 + CD3cells and total NK cells, respectively. NK cells account for a few proportion of uterine cells. The expression of Cyp26a1 in uterine tissue mainly comes from glandular epithelial and luminal epithelial cells. 11,21 There were several drawbacks in this study. The actual function of Cyp26a1 + NK cells was unclear. Our finding that Cyp26a1 regulates the function of NK cells through Gzma, Klrg1 and Fcgr4 needs to be verified in vivo.
In conclusion, our data illustrated that there was a subset of Cyp26a1 + NK cells at the mouse maternal-foetal interface and this group of cells has low expression of NK cell toxicity-related genes.
We speculate that Cyp26a1 has a modulating effect on NK cells during the peri-implantation period in mouse and this effect may be exercised through Gzma, Klrg1 and Fcgr4 (Figure 7).

ACK N OWLED G EM ENTS
We thank PhD Duo Peng of Harvard TH Chan School of Public Health for revision on the manuscript. We thank PhD Guang-zhe Ge of Beijing Institute of Genomics for help in transcriptome sequencing analysis. This work is funded by National Key R&D Program of China (no. 2016YFC1000905) .

CO N FLI C T O F I NTE R E S T
The authors confirm that there are no conflicts of interest. Funding acquisition (lead); Methodology (supporting); Resources (lead).

DATA AVA I L A B I L I T Y S TAT E M E N T
We confirm that the data in our paper can be used.