Abstract

   Decidual CD4^+ T (dCD4 T) cells are crucial for the maternal-fetal
   immune tolerance required for a healthy pregnancy outcome. However,
   their molecular and functional characteristics are not well elucidated.
   In this study, we performed the first analysis of transcriptional and
   alternative splicing (AS) landscapes for paired decidual and peripheral
   blood CD4^+ T (pCD4 T) cells in human early pregnancy using high
   throughput mRNA sequencing. Our data showed that dCD4 T cells are
   endowed with a unique transcriptional signature when compared to pCD4 T
   cells: dCD4 T cells upregulate 1,695 genes enriched in immune system
   process whereas downregulate 1,011 genes mainly related to mRNA
   catabolic process and the ribosome. Moreover, dCD4 T cells were
   observed to be at M phase, and show increased activation,
   proliferation, and cytokine production, as well as display an
   effector-memory phenotype and a heterogenous nature containing Th1,
   Th17, and Treg cell subsets. However, dCD4 T cells undergo a comparable
   number of upregulated and downregulated AS events, both of which are
   enriched in the genes related to cellular metabolic process. And the
   changes at the AS event level do not reflect measurable differences at
   the gene expression level in dCD4 T cells. Collectively, our findings
   provide a comprehensive portrait of the unique transcriptional
   signature and AS profile of CD4^+ T cells in human decidua and help us
   gain more understanding of the functional characteristic of these cells
   during early pregnancy.

   Keywords: decidual CD4^+ T cells, early human pregnancy, transcriptome,
   alternative splicing, high-throughput mRNA sequencing

Introduction

   The maternal-fetal interface (MFI) is regarded as the interface between
   the extraembryonic tissues of developing conceptus and the uterine
   mucosa, and the immune cells at the MFI are the maternal immune cells
   that populate the decidua ([45]1–[46]3). During the first trimester of
   healthy pregnancy, natural killer (NK) cells, macrophages, and T cells
   are the most abundant leukocytes in human decidua, whereas dendritic
   cells (DCs), B cells and NKT cells are rare ([47]3). Since these
   decidual leukocytes are reported to play a key role in facilitating
   tolerance to the semi-allogeneic fetus and therefore maintaining a
   successful pregnancy, a deeper understanding of the molecular and
   functional features of these cells will provide novel insights into the
   pathogenesis of pregnancy-related complications and poor postnatal
   health ([48]3–[49]5). In first-trimester human decidua, approximately
   70% of leukocytes are decidual NK (dNK) cells, which display a
   phenotype of CD56^brightCD16^dim and play important roles in promoting
   neo-angiognesis and trophoblast invasion, remodeling of spiral
   arteries, and directing placentation ([50]6–[51]9). Whole-genome
   transcriptome analysis revealed that dNK cells upregulate expression of
   many genes related to inhibitory receptors, growth factors, cell cycle,
   and cytokines/chemokines, but downregulate expression of the genes
   involved in activating receptors and costimulatory factors as compared
   with peripheral blood NK cells ([52]10, [53]11).

   Unlike dNK cells, decidual T cells are less abundant and do not have a
   trophic function. However, they are thought to play rather important
   roles in immune regulation and allograft tolerance at the MFI ([54]3,
   [55]12, [56]13). In early human pregnancy, about 10–20% of decidual
   leukocytes are CD3^+ TCRαβ^+ T cells, and approximately 30–45% of them
   are CD4^+ T cells, that display an activated/memory CD25^dim cell
   surface phenotype ([57]3, [58]14–[59]16). More than 20 years ago, Saito
   and colleagues observed that in the first trimester of human normal
   pregnancy, decidual CD4^+ T (dCD4 T) cells express the
   T-cell-activation antigens CD69, HLA-DR, interleukin-2 receptor-alpha
   (IL-2Rα), and IL-2Rβ at a significantly higher level than do peripheral
   blood CD4^+ T (pCD4 T) cells, indicating that dCD4 T cells are
   regionally activated at an early stage of pregnancy ([60]17).
   Furthermore, recent studies showed that dCD4 T cells in early pregnancy
   express higher levels of the T regulatory (Treg)-cell markers CD25 and
   FOXP3, the proliferation-associated antigen Ki-67, programmed cell
   death-1 (PDCD1, also called PD-1), and T-cell immunoglobulin mucin-3
   (Tim-3) than their peripheral blood counterparts ([61]13,
   [62]18–[63]20). However, the complexity of dCD4 T cells has not been
   well elucidated, and a thorough understanding of the molecular and
   functional features of these cells would shed light on their eventual
   roles during early pregnancy.

   Upon encountering antigens presented by antigen-presenting cells (APCs)
   and being driven by a set of transcriptional regulators and cytokines,
   naive CD4^+ T helper (Th) cells are able to differentiate into distinct
   effector subsets, including Th1, Th2, Th17, and Treg cells ([64]5,
   [65]21, [66]22). Th1, Th2, and Th17 cells are characterized by their
   production of IFN-γ, IL-4, and IL-17, respectively. Their
   differentiation is controlled by the following lineage-specific
   “master” transcription factors: T-box-binding transcription factor
   (T-bet) for Th1, GATA-binding protein 3 (GATA3) for Th2, and
   RAR-related orphan receptor gamma (RORγ) for Th17 cells; whereas Treg
   cells are defined and driven by expressing the forkhead transcription
   factor FOXP3 ([67]5, [68]13, [69]21, [70]22). Although growing
   evidences suggest that Treg cells in the decidua contribute to
   maternal-fetal immune tolerance and pregnancy maintenance, the role and
   necessity of other decidual Th-cell subsets in a normal pregnancy is
   still largely unknown or remains controversial ([71]3, [72]16, [73]23,
   [74]24).

   In eukaryotic organisms, alternative splicing (AS) is a crucial and
   ubiquitous mechanism that regulates gene expression and generates
   transcript/protein diversity by producing two or more distinct mRNA
   transcripts from the same precursor mRNA (pre-mRNA) using different
   splice sites ([75]25, [76]26). Nearly 75% of all human genes and 94% of
   human multiexonic genes undergo AS, with a bias toward genes expressed
   in the nervous and immune systems, and toward developmental
   stage-specific and tissue-specific expression ([77]27–[78]29).
   Moreover, AS plays a crucial role in shaping the T-cell response to
   stimulation ([79]30), and in the process of CD4^+ T-cell activation
   ([80]31) and differentiation ([81]32). In CD4^+ T cells, many genes
   with known functions in immunobiology have been reported to undergo AS
   changes during immune responses, including genes coding for cell
   surface and adapter proteins (e.g., CD45, CD96, CTLA-4),
   transcriptional regulators (e.g., GATA3, FOXP3, HIF1α), RNA-processing
   (e.g., AUF-1, CELF2, EIF4G2) and intracellular signaling/transport
   molecules (e.g., AKAP9, CLK2, MAP3K7), and cytokines (e.g., IL4, TIR8)
   ([82]25, [83]30, [84]33, [85]34). As yet, little is known about the
   genome-wide AS profile of CD4^+ T cells in human decidua as compared to
   those in the peripheral blood.

   In this study, we compared the transcriptional profile and AS landscape
   between paired decidual and peripheral CD4^+ T cells from healthy women
   at the first trimester of pregnancy using high-throughput mRNA
   sequencing (mRNA-Seq), which can generate quantitative measurements of
   gene expression and AS events, as well as provide greater accuracy and
   sensitivity than microarrays ([86]35, [87]36). Our findings are the
   first to show that dCD4 T cells upregulate genes related to immune
   system process and downregulate genes involved in mRNA catabolic
   process, whereas AS is not a major contributor to these changes. Our
   results thus provide new insights into the molecular traits of CD4^+ T
   cells in human decidua during early pregnancy.

Materials and Methods

Subjects

   This study was approved and performed in compliance with the Medical
   Ethics Committee of the International Peace Maternity and Child Health
   Hospital of China Welfare Institute (Shanghai, China). Twelve healthy
   pregnant women with no history of spontaneous abortion, preterm labor
   or preeclampsia in their previous pregnancies were recruited for this
   study. Their blood samples and decidual tissues were obtained while
   they were undergoing elective surgical abortion at the department of
   Obstetrics and Gynecology in the International Peace Maternity and
   Child Health Hospital of China Welfare Institute affiliated to Shanghai
   Jiao Tong University School of Medicine. Informed consent was obtained
   from all participants in written form before enrollment. Venous blood
   was collected in EDTA-anticoagulant tubes (BD Vacutainer, USA)
   immediately before termination of pregnancy, and the autologous
   decidual tissues were collected and stored in sterile ice-cold
   phosphate-buffered saline (PBS). Samples from three pregnant women
   (mean age, 26 years; range, 22–28 years) with mean Gestational Day 50
   (range, Days 44–58) were used for mRNA-Seq analysis; those from other
   five women (mean age, 30 years; range, 22–39 years) with mean
   Gestational Day 45 (range, Days 38–50) were used for validation of the
   mRNA-Seq data by flow cytometry staining, whereas samples from the last
   four women (mean age 34 years, range 30–39; mean Gestational Day 45,
   range 43–50) were applied for determination of CD4^+ T-cell functional
   status (Figure S1 in Supplementary Material).

Separation of Mononuclear Cells

   Separation of decidual mononuclear cells (DMCs) from decidual tissues
   was performed according to the procedure of non-enzymatic leukocytes
   separation, as described previously ([88]5, [89]13, [90]37–[91]39).
   Briefly, vacuum-aspirated abortion tissues were washed in sterile PBS;
   the decidual tissue was separated macroscopically from the fetal tissue
   and placenta, and then cut into small pieces (<1 mm^3) using ocular
   scissors and filtered through a 74-µm nylon mesh filter to obtain a
   single cell suspension. Both DMCs and peripheral blood mononuclear
   cells (PBMCs) were isolated and purified by density gradient
   centrifugation using Lymphoprep™ (AS1114546, Axis-shield).

Monoclonal Antibodies (mAbs) and Reagents

   Fluorescein-conjugated anti-human mAbs, including anti-CD3 FITC (clone:
   UCHT1), anti-CD4 V500 (RPA-T4), anti-CD8 APC/Cy7 (SK1), anti-CD56 PE
   (B159), and anti-CD279 (PD-1) PE (EH12.1, also known as EH12) were
   purchased from BD Pharmingen. Meanwhile, the other anti-human mAbs
   including anti-CD3 APC (HIT3a), anti-CD8a PerCP/Cy5.5 (HIT8a),
   anti-CD38 FITC (HIT2), anti-CD63 FITC (H5C6), anti-CD69 APC (FN50),
   anti-CD122 (IL-2Rβ) APC (TU27), anti-CD183 (CXCR3) PE/Cy7 (G025H7),
   anti-CD183 (CXCR3) APC (G025H7), anti-CD192 (CCR2) APC/Cy7 (K036C2),
   anti-CD196 (CCR6) PE (G034E3), anti-CD197 (CCR7) PE (G043H7),
   anti-CD276 (B7-H3) PE (DCN.70), anti-CD45RO FITC (UCHL1), anti-CD197
   (CCR7) PE/Cy7 (G043H7), anti-IFN-γ APC/Cy7 (4S.B3), anti-IL-4 PE/Cy7
   (MP4-25D2), anti-IL-17A FITC (BL168), and anti-FOXP3 PE (206D) were
   purchased from BioLegend.

Flow Cytometry

   Cell surface and intracellular staining were performed as previously
   described ([92]5, [93]40–[94]42). Surface staining of the isolated
   mononuclear cells (PBMCs and DMCs) was realized by incubating the cells
   directly with different cocktails of anti-human mAbs in 100 µL PBS
   containing 3% (v/v) fetal bovine serum at room temperature for 30 min.
   For intracellular detection of IFN-γ, IL-17A, IL-4, and Foxp3, the
   isolated mononuclear cells were stimulated with 81 nM
   phorbol-12-myristate-13-acetate (PMA) and 1.34 µM ionomycin for 4 h in
   the presence of brefeldin (10.6 µM) and monensin (2 µM) (eBiosciences,
   USA); cells were stained first for surface markers and then for
   intracellular cytokines (IFN-γ, IL-17A, IL-4) and Foxp3 by using Foxp3
   staining buffers (eBioscience, USA) following the manufacturer’s
   recommendation. The immunostained cells were collected and analyzed on
   a BD FACS Canto II flow cytometer (BD Biosciences, USA), and data were
   processed using the FlowJo 7.6.1 software.

Purification and RNA Isolation of CD4^+ T Cells

   The decidual and peripheral blood CD4^+ (dCD4 and pCD4) T cells were
   purified from the DMCs and PBMCs, respectively, by sorting on a FACS
   Aria II (BD Biosciences, USA) based on the surface expression of CD56,
   CD3, and CD4 markers (CD56^−CD3^+CD4^+), always achieving a purity
   greater than 95%. Total RNA of freshly sorted CD4^+ T cells was
   extracted using the Trizol Reagent (Invitrogen, USA) and was purified
   by RNeasy Micro Kit (Cat No.: 74004, Qiagen, Germany). RNA purity and
   concentration were determined using a NanoPhotometer^®
   spectrophotometer (IMPLEN, USA) and Qubit^® RNA Assay Kit in Qubit^®
   2.0 Flurometer (Life Technologies, USA), respectively. The RNA
   integrity was evaluated by the RNA Nano 6000 Assay Kit of the Agilent
   Bioanalyzer 2100 system (Agilent Technologies, USA) ([95]43).

Library Preparation, Clustering, and mRNA-Seq

   Library preparation, clustering, and mRNA-seq were performed as
   previously described ([96]43). We used 200 ng of RNA per sample (three
   samples of both pCD4 and dCD4 T cells) as input material for the RNA
   sample preparations. cDNA libraries were constructed using the
   NEBNext^® Ultra™ RNA Library Prep Kit for Illumina^® (NEB, USA),
   according to the manufacturer’s instructions. The products were
   purified using the AMPure XP system and library quality was determined
   on the Agilent Bioanalyzer 2100 system. We carried out clustering of
   the index-coded samples on a cBot Cluster Generation System using a
   HiSeq 2500 PE Cluster Kit (Illumia, USA), following the manufacturer’s
   recommendations. After cluster generation, the prepared libraries were
   sequenced on an Illumina Hiseq 2500 platform and 125 bp paired-end
   reads were produced.

Sequence Alignment and Gene Expression Analysis

   Sequenced reads were aligned to the human reference genome (hg19
   version) using the STAR software package ([97]44). Exons from all
   isoforms of a gene were merged to create one meta-gene. The number of
   reads falling in the exons of this meta-gene was counted using
   HTSeq-count, and differential expression analysis was performed using
   DESeq ([98]45). Differences in gene expression with a P-value < 0.05
   (paired test) were considered significant. The complete data have been
   deposited in NCBI Gene Expression Omnibus with accession number
   [99]GSE97395.

AS Analysis

   rMATS (v3.2.1 beta) was used to identify and assess differential
   expression of AS events ([100]46). We used the hg19 genome as our
   reference genome, and transcripts annotation from the NCBI Reference
   Sequence ([101]47) for AS events annotation. The “exon inclusion level”
   (Ψ) value was estimated by the percentage of the density of exon
   inclusion reads among the sum of the densities of exon inclusion reads
   plus exon skipping reads, as described previously ([102]46, [103]48).
   Differential AS events were accepted if they could achieve the
   threshold of |ΔΨ| > 0.05 and a false discovery rate (FDR) of < 0.05,
   which were divided into two sets: upregulated AS sets were those with
   an FDR < 0.05 and a ΔΨ > 0.05, and downregulated AS sets were those
   with an FDR < 0.05 and a ΔΨ < −0.05.

Heatmap, Gene Ontology Annotation, KEGG Pathway, and Gene Set Enrichment
Analysis (GSEA)

   Heatmap analysis of differential genes was performed using R.
   Functional annotations, including Gene Ontology (GO) annotation and
   Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment
   analysis of interesting genes, were carried out using the DAVID
   Bioinformatics Resources ([104]http://david.ncifcrf.gov/home.jsp).
   P-values < 0.05 were considered as the significance threshold. GSEA was
   performed by using the GSEA-P software, MSigDB 1.0, as described
   previously ([105]49).

Results

Distinct Transcriptional Signatures between Human dCD4 and pCD4 T Cells

   To elicit the genome-wide transcriptional and AS differences between
   pCD4 and dCD4 T cells, three healthy women at the first trimester of
   normal pregnancy were recruited, and fluorescence-activated cell
   sorting (FACS) was performed to isolate and purify CD4^+ T cells
   (CD56^−CD3^+CD4^+) from the decidua and autologous peripheral blood
   (Figures S1 and S2 in Supplementary Material). Polyadenylated
   [poly(A)+] RNA samples of both cell populations from each donors were
   sequenced using Illumina Hiseq 2500 platform, which yielded a total of
   213 million 2 × 125-bp paired-end reads, ranging from 30 to 40 million
   reads per sample (Figure [106]1A). After aligning all sequenced reads
   to the human reference genome, between 88 and 94% of reads per sample
   were mapped uniquely on the hg19 reference (Figure [107]1A). Multi-loci
   mapping and unmapped reads were excluded from further analyses.

Figure 1.

   [108]Figure 1
   [109]Open in a new tab

   Human dCD4 T cells show a distinct transcriptional signature and
   upregulate genes related to immune system process as compared with
   autologous pCD4 T cells. Three healthy women at the first trimester of
   normal pregnancy were recruited and their dCD4 and pCD4 T cells were
   isolated by fluorescence-activated cell sorting (FACS). (A) Summary of
   mRNA-Seq data for the purified dCD4 and pCD4 T cells. RNA samples of
   paired dCD4 and pCD4 T cells from three individuals were sequenced on
   the Illumina Hiseq 2500 platform, yielding approximately 30–40 million
   2 × 125-bp paired-end reads per sample, which were then mapped to the
   human reference genome (hg19 version). (B) Number and percentage of the
   differentially expressed genes (P < 0.05, paired test) that were
   upregulated or downregulated in dCD4 T cells versus their pCD4 T
   counterparts. (C) Heatmap of differentially expressed genes. Each line
   represents one gene. Each column represents one sample. Different
   colors represent the expression levels (from blue to red: increased
   expression). (D) Functional enrichment of the upregulated genes in dCD4
   T cells by GO annotation of biological process. The top 20 GO terms are
   listed. (E,F) GSEA plots of GO categories including immune response
   (E), immune system process, adaptive immune response, immune effector
   process, and regulation of immune system process (F) in dCD4 versus
   pCD4 T cells. dCD4 T, decidual CD4^+ T; pCD4 T, peripheral blood CD4^+
   T; No., number; %, percentage; GO, Gene Ontology; GSEA, gene set
   enrichment analysis; Nom, Nominal; NES, Normalized Enrichment Score.

   For mRNA-Seq datasets, the read count is demonstrated to be linearly
   related and represent a good approximation to the abundance of the
   target gene ([110]45, [111]50). In the present study, the number of
   reads was counted using HTSeq-count and differential expression
   analysis was performed using DESeq ([112]45). Data showed that there
   were 2,706 differentially expressed genes between dCD4 and pCD4 T
   cells, with most of them (62.6%, 1,695 genes) being upregulated and the
   remaining 1,011 genes (37.4%) being downregulated in dCD4 T cells as
   compared with autologous pCD4 T cells (Figures [113]1B,C; Table S1 in
   Supplementary Material).

dCD4 T Cells Upregulate the Molecules Related to Immune System Process

   Functional enrichment of the upregulated genes in dCD4 T cells by GO
   annotation of Biological Process (BP) revealed that these genes are
   most significantly enriched in the category of immune system process
   (P = 4.97E−53), followed by response to stimulus/stress, regulation of
   immune system process, immune response, regulation of response to
   stimulus, positive regulation of immune system process, cell activation
   and the defense response categories (Figure [114]1D). GSEA has been
   used widely as a knowledge-based approach to interpret genome-wide
   expression profiles ([115]49). GSEA using GO categories of BP showed
   that the gene set of immune response was most positively enriched in
   dCD4 versus pCD4 T cells (Normalized Enrichment Score = 2.2944),
   followed by immune system process, defense response, response to
   external stimulus, adaptive immune response, immune effector process,
   regulation of immune system process, and inflammatory response (Figures
   [116]1E,F; Figure S3 in Supplementary Material). In addition, KEGG
   pathway and GSEA analyses revealed that the upregulated genes in dCD4 T
   cells were dramatically enriched in the terms related to antigen
   processing and presentation (P = 2.06E−15 with KEGG analysis;
   NES = 2.6970 with GSEA), allograft rejection, lysosome, cytokine, and
   cytokine receptor interaction, Toll-like receptor signaling pathway,
   and p53 and MAPK signaling pathways (Figure S4 in Supplementary
   Material).

   In order to validate the mRNA-Seq data, we selected 10 immune response
   associated molecules and performed flow cytometry staining in five
   other human samples at the first trimester of normal pregnancy (Figure
   S1 in Supplementary Material). Being almost completely consistent with
   the mRNA-Seq data, we observed that human dCD4 T cells highly
   upregulate CCR2, CCR6, CXCR3, CD38, CD69, CD63, CD122 (encoded by
   IL2RB), CD276, and PD-1 (encoded by PDCD1), but downregulate CCR7, as
   compared with autologous pCD4 T cells (Figure [117]2; Figure S5 in
   Supplementary Material).

Figure 2.

   [118]Figure 2
   [119]Open in a new tab

   Validation of mRNA-Seq data by flow cytometry staining.
   (A,C,E,G,I,K,M,O,Q) Comparison of the gene expression (measured as the
   read count) of CCR2, CCR6, CCR7, CXCR3, CD69, CD63, IL2RB, CD276, and
   PDCD1 between paired pCD4 and dCD4 T cells. Each symbol reflects a
   sample and each line reflects samples from the same individual (n = 3
   per group). (B,D,F,H,J,L,N,P,R) Representative flow cytometric
   histograms (left) and cumulative data (right) illustrating the
   comparison of molecular expression of indicated proteins between paired
   pCD4 and dCD4 T cells (n = 5 per group). The cells were gated in CD4^+
   T cells and the geometric MFI values were calculated using FlowJo 7.6.1
   software. pCD4 T, peripheral blood CD4^+ T; dCD4 T, decidual CD4^+ T;
   %, percentage; MFI, mean fluorescent intensity.

   Taken together, these data showed that human dCD4 T cells upregulate
   genes mainly related to immune system process, suggesting that these
   genes might underline an enhanced immune function of dCD4 T cells
   during human early pregnancy.

dCD4 T Cells Downregulate Genes Related to mRNA Catabolic Process and the
Ribosome

   In contrast, multiple other categories including nuclear-transcribed
   mRNA catabolic process (P = 9.00E−26 with GO annotation), mRNA
   catabolic process (GO P = 1.41E−25), nuclear-transcribed mRNA catabolic
   process (GO P = 1.85E−24), protein RNA complex assembly (NES = −1.81
   with GSEA), ribonucleoprotein complex biogenesis and assembly (GSEA
   NES = −1.77), establishment and/or maintenance of chromatin
   architecture (GSEA NES = −1.72) and the ribosome (P = 5.22E−28 with
   KEGG analysis; GSEA NES = −3.05), were significantly enriched in the
   downregulated genes in dCD4 T versus pCD4 T cells by GO annotation,
   KEGG pathway and/or GSEA analyses (Figure [120]3). These results
   indicated that human dCD4 T cells mainly downregulate the genes related
   to mRNA catabolic process and the ribosome.

Figure 3.

   [121]Figure 3
   [122]Open in a new tab

   Functional enrichment analyses for genes downregulated in dCD4 versus
   pCD4 T cells. (A,B) The top 20 GO and KEGG terms enriched for
   downregulated genes in dCD4 versus pCD4 T cells. (C) GSEA plots of
   GO/KEGG categories including ribonucleoprotein complex biogenesis and
   assembly, establishment and/or maintenance of chromatin architecture,
   and ribosome. GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and
   Genomes; GSEA, gene set enrichment analysis; Nom, Nominal; NES,
   Normalized Enrichment Score.

Enhanced Functionality and Subset Complexity of dCD4 T Cells

   To further characterize the phenotype and functionality of human dCD4 T
   cells, we performed GSEA analysis for the GO categories involved in
   cell cycle, activation, proliferation, and cytokine production.
   Interestingly, we found that the gene sets related to the following
   categories: M phase of mitotic cell cycle, mitotic sister chromatin
   segregation, M phase (Figure [123]4A; Figure S6A in Supplementary
   Material); lymphocyte activation, T cell activation, regulation of
   lymphocyte activation (Figure [124]4B; Figure S6B in Supplementary
   Material); cell proliferation, (positive) regulation of cell
   proliferation (Figure [125]4C; Figure S6C in Supplementary Material);
   and cytokine production, cytokine secretion and cytokine biosynthetic
   process (Figure [126]4D; Figure S6D in Supplementary Material), were
   remarkably and positively enriched in dCD4 versus pCD4 T cells,
   indicating that human dCD4 T cells stay in M phase, and show increased
   activation and proliferation, as well as have an enhanced functionality
   represented by cytokines production.

Figure 4.

   [127]Figure 4
   [128]Open in a new tab

   dCD4 T cells stay in M phase and show increased activation,
   proliferation, and cytokine production, as well as contain Th1, Th17,
   and Treg cell subsets and display an effector-memory phenotype. (A–D)
   GSEA plots of GO categories, including M phase of mitotic cell cycle,
   mitotic sister chromatin segregation (A); lymphocyte activation, T cell
   activation (B); cell proliferation, positive regulation of cell
   proliferation (C); and cytokine production and cytokine secretion (D).
   (E,G,I,L) Comparison of the gene expression (measured as the read
   count) of IFNG, IL17A, IL4, and FOXP3 between in paired pCD4 and dCD4 T
   cells at rest. Each symbol reflects a sample and each line reflects the
   samples from an individual (n = 3 per group). (F,H,J,K) Comparison of
   the expression of IFN-γ, IL-17A, IL-4, and Foxp3 between paired pCD4
   and dCD4 T cells as determined by intracellular staining upon
   stimulation with phorbol-12-myristate-13-acetate and ionomycin in the
   presence of brefeldin and monensin. Similar results were obtained from
   four individuals at the first trimester of normal pregnancy. (M–O)
   Representative flow cytometric plots (M), bar graphs (N) and pie charts
   (O) displaying the proportions of native (T[N], CD45RO^−CCR7^+),
   effector (T[E], CD45RO^−CCR7^−), T[CM] (CD45RO^+CCR7^+), and T[EM]
   (CD45RO^+CCR7^−) cells in paired pCD4 and dCD4 T cells. Each symbol
   reflects a sample and each line reflects the samples from the same
   person (n = 4 per group). pCD4 T, peripheral blood CD4^+ T; dCD4 T,
   decidual CD4^+ T; GO, Gene Ontology; GSEA, gene set enrichment
   analysis; Nom, Nominal; NES, Normalized Enrichment Score; UD,
   undetected.

   Furthermore, we evaluated the transcript and protein abundances of the
   representative cytokines and “master” transcription factors for the
   Th1, Th2, Th17, and Treg cell commitment, in resting or activated dCD4
   and pCD4 T cells. Data showed that dCD4 T cells had a higher mRNA
   expression level of IFNG, IL17A, RORC, and FOXP3 at rest, as well as
   produced more IFN-γ, IL-17A, and Foxp3 upon stimulation with PMA and
   ionomycin as determined by intracellular staining; in contrast, the
   mRNA expression of IL4, IL5, IL13, and GATA3 at rest, together with
   IL-4 secretion after stimulation, were at an extremely low level in
   both pCD4 and dCD4 T cells or not different between these cells
   (Figures [129]4E–L; Figures S7A,B in Supplementary Material). These
   data suggested that human dCD4 T cells are a heterogeneous population
   containing Th1, Th17, and Treg cell subsets.

   Memory is the hallmark of adaptive immune response, and memory T cells
   are divided into at least two distinct subsets: central memory T
   (T[CM]) and effector memory T (T[EM]) cells, based on their different
   effector functions and homing capacities ([130]42, [131]51, [132]52).
   T[EM] cells are the first responders capable of migrating into inflamed
   tissues and possess immediate effector functions, whereas T[CM] cells
   can home to lymphoid organs where they readily proliferate and produce
   more secondary effectors ([133]42, [134]51). Consistent with previous
   studies ([135]53, [136]54), we observed that human dCD4 T cells
   contained a higher percentage of CD45RO^+ cells, which are regarded as
   memory T cells, as compared with pCD4 T cells (Figure S7C in
   Supplementary Material). Moreover, when we compared the proportions of
   native (T[N], CD45RO^−CCR7^+), effector (T[E], CD45RO^−CCR7^−), T[CM]
   (CD45RO^+CCR7^+), and T[EM] (CD45RO^+CCR7^−) cells between pCD4 and
   dCD4 T cells using flow cytometry staining ([137]51), we found that
   dCD4 cells significantly increased the proportion of T[EM] cells but
   decreased T[N] cells, revealing that human dCD4 T cells mainly consist
   of T[EM] cells whereas T[N] cells are almost absent (Figures
   [138]4M–O).

   Collectively, these results showed that human dCD4 T cells during early
   pregnancy are endowed with enhanced activation, high proliferation
   potential, and elevated functionality in terms of cytokine production,
   as well as with a complex nature containing Th1, Th17, and Treg cell
   subsets and displaying an effector-memory phenotype.

Genes in dCD4 T Cells Undergo a Comparable Number of Upregulated and
Downregulated AS Events

   Alternative splicing is an important mechanism involved in shaping CD4
   T-cell activation, differentiation, and immune response to stimulation
   ([139]30–[140]33). Here, we applied the rMATS (v3.2.1 beta) paired
   model to identify and analyze the differentially expressed AS events
   using the splice junction counts as the input ([141]46). Five basic and
   generally recognized AS modes were investigated, including skipped exon
   (SE), mutually exclusion exons (MXE), alternative 5′ splice site
   (A5SS), alternative 3′ splice site (A3SS), and retained intron (RI). A
   total of 127,147 AS events, belonging to 10,281 genes, were found in
   the dCD4 and pCD4 T cells, with 512 genes showing evidence of all five
   AS types (Figures [142]5A,B). SE was the most common type of AS event,
   accounting for 46.5% (59,154 SE events in 9,204 genes) of all splicing
   events; in contrast, RI was the least common (1.8%, 2,284 RI events in
   1,158 genes) (Figures [143]5A,B).

Figure 5.

   [144]Figure 5
   [145]Open in a new tab

   dCD4 T cells undergo a comparable number of upregulated and
   downregulated AS events. (A) Summary of the numbers of five types of AS
   events, differentially expressed events (with upregulated and
   downregulated events) and the genes they belong to. Column 1 shows the
   five most common types of AS events that were investigated. Columns 2
   and 3 show the initial number of AS events found in the combined
   samples of pCD4 and dCD4 T cells with their frequency among total
   events (column 2), as well as the number of genes these events belong
   to (column 3). Columns 4 and 5 show the number of differentially
   expressed events (FDR < 0.05 with |ΔΨ| > 0.05 between samples) with
   upregulated (FDR < 0.05 with ΔΨ > 0.05) and downregulated (FDR < 0.05
   with ΔΨ < −0.05) AS events in dCD4 T cells versus their autologous pCD4
   T counterparts (column 4), as well as the number of genes that these
   significant events belong to (column 5). (B) Venn diagram of genes
   showing the initial SE, MXE, A5SS, A3SS, and RI in the combined samples
   (pCD4 and dCD4 T cells). (C,D) Venn diagram of genes showing
   upregulated (C) or down-regulated (D) SE, MXE, A5SS, A3SS, and RI
   events in dCD4 T cells with respect to pCD4 T cells. AS, alternative
   splicing; SE, skipped exon; MXE, mutually exclusion exons; A5SS,
   alternative 5′ splice site; A3SS, alternative 3′ splice site; RI,
   retained intron; %, percentage; No., Number; Sig., Significant; up,
   upregulated; down, downregulated; FDR, false discovery rate.

   Interestingly, compared with pCD4 T cells, comparable numbers of
   upregulated (at the threshold of FDR < 0.05 and ΔΨ > 0.05) and
   downregulated (FDR < 0.05 and ΔΨ < −0.05) splicing events were observed
   in human dCD4 T cells (upregulated events versus downregulated events):
   2,378 versus 2,083 of SE, 1,809 versus 1,647 of MXE, 397 versus 404 of
   A5SS, 511 versus 503 of A3SS, 96 versus 97 of RI, and 5,191 versus
   4,737 of total AS events (Figure [146]5A; Tables S2–S6 in Supplementary
   Material). These differential AS events also occurred in a comparable
   number of genes (genes showing upregulated events versus downregulated
   events): 1,535 versus 1,438 for SE, 904 versus 886 for MXE, 347 versus
   364 for A5SS, 452 versus 439 for A3SS, 88 versus 91 for RI, and 2,607
   versus 2,547 for all AS events (Figures [147]5A,C,D). In addition, the
   numbers of genes undergoing SE-, MXE-, A5SS-, A3SS-, or RI-upregulation
   specifically, or of genes undergoing two or more types’ upregulation
   simultaneously, were comparable to those undergoing downregulation
   specifically or simultaneously, in dCD4 versus pCD4 T cells (Figures
   [148]5C,D).

Both Upregulated and Downregulated AS Events in dCD4 T Cells Are Enriched in
the Genes Related to Cellular Metabolic Process

   Functional enrichment analysis of the genes undergoing differential AS
   events in dCD4 T cells (at the threshold of |ΔΨ| > 0.05 and FDR < 0.05,
   compared with pCD4 T cells; including all five AS types) highlighted
   the categories related to cellular metabolic process (P = 4.12E−43 with
   GO annotation), cellular nitrogen compound metabolic process
   (P = 3.85E−40), cellular macromolecule metabolic process
   (P = 1.18E−38), ubiquitin mediated proteolysis (P = 4.00E−10 with KEGG
   analysis), valine, leucine, and isoleucine degradation (P = 7.69E−09),
   and the spliceosome (P = 5.95E−07) (Figure S8 in Supplementary
   Material). Furthermore, enrichment analysis of genes containing
   upregulated AS events (at the threshold of ΔΨ > 0.05 and FDR < 0.05,
   compared with pCD4 T cells; including all five AS types) highlighted
   the categories related to cellular metabolic process (P = 2.78E−28 with
   GO annotation), cellular nitrogen compound metabolic process
   (P = 1.08E−27), cellular macromolecule metabolic process
   (P = 6.77E−25), ubiquitin mediated proteolysis (P = 2.07E−05 with KEGG
   analysis), RNA degradation (P = 2.54E−05), and the spliceosome
   (P = 2.60E−06), which were also significantly enriched for the genes
   containing downregulated AS events (at the threshold of ΔΨ < −0.05 and
   FDR < 0.05; P = 4.94E−25, 3.21E−24, 9.60E−26, 9.64E−10, 5.47E−04, and
   1.68E−04, respectively) in human dCD4 T cells (Figure [149]6).
   Additionally, enrichment analysis of the genes undergoing only
   upregulated or only downregulated AS events, or the genes showing
   evidence of both upregulated and downregulated AS events
   simultaneously, also highlighted the term related to cellular metabolic
   process (Figures S9–S10 in Supplementary Material). Similar results
   were observed when dividing AS events into SE, MXE, A5SS, A3SS, and RI,
   respectively (Figures S11–S20 in Supplementary Material). Therefore,
   these data revealed that both upregulated and downregulated AS events
   in dCD4 T cells are remarkably enriched in the genes related to
   cellular metabolic process.

Figure 6.

   [150]Figure 6
   [151]Open in a new tab

   Functional enrichment analysis of the genes undergoing upregulated or
   downregulated AS events in dCD4 T cells. (A–D) The top 20 GO (A,C) and
   KEGG (B,D) terms enriched for genes undergoing upregulated [(A,B),
   FDR < 0.05 with ΔΨ > 0.05 between samples] or downregulated [(C,D),
   FDR < 0.05 with ΔΨ < −0.05 between samples] AS events (SE, MXE, A5SS,
   A3SS, and RI are combined together) in dCD4 versus pCD4 T cells. AS,
   alternative splicing; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of
   Genes and Genomes; FDR, false discovery rate; SE, skipped exon; MXE,
   mutually exclusion exons; A5SS, alternative 5′ splice site; A3SS,
   alternative 3′ splice site; RI, retained intron.

Changes at the AS Event Level Do Not Imply Measurable Differences at the Gene
Expression Level in dCD4 T Cells

   Finally, the overlap between the sets of genes undergoing differential
   splicing-level and expression-level changes in dCD4 T cells was
   observed to be very low: 748 genes in the overlap versus 2,706 genes
   showing expression-level changes and 3,959 genes showing splicing-level
   changes (Figure [152]7A). Meanwhile, the overlap between sets of genes
   undergoing splicing-level and expression-level upregulation or
   downregulation in dCD4 T cells was also very low: 122 genes in the
   overlap versus 1,695 genes showing expression-level upregulation and
   2,607 genes showing splicing-level upregulation; 117 genes in the
   overlap versus 1,011 genes showing expression-level downregulation and
   2,547 genes showing splicing-level downregulation (Figure [153]7B).
   These results indicated that changes at the AS event level do not
   result in measurable differences at the gene expression level in human
   dCD4 T cells.

Figure 7.

   [154]Figure 7
   [155]Open in a new tab

   Changes at the AS event level do not imply measurable differences in
   the gene expression level in dCD4 T cells. (A) Venn diagram of genes
   undergoing significant splicing-level and expression-level changes in
   dCD4 T cells with respect to pCD4 T cells. (B) Venn diagram of genes
   undergoing splicing-level upregulation and down-regulation, as well as
   expression-level upregulation and downregulation, in dCD4 T cells with
   respect to pCD4 T cells. AS, alternative splicing; up, upregulated;
   down, downregulated.

Discussion

   T cells possess a particular capacity to capture membrane molecules
   from APCs through a process termed trogocytosis ([156]55–[157]58). Zhou
   et al. revealed that presentation of captured peptide-MHC-II ligands by
   CD4^+ T cells has a stimulatory effect on naive T cells, and these
   cells with acquired peptide-MHC-II molecules become effector CD4^+ T
   cells that manifest a better recall response ([158]56). In the present
   study, human dCD4 T cells in early pregnancy were found to
   significantly upregulate the genes involved in immune response (Figure
   [159]1) and antigen processing and presentation (Figure S4 in
   Supplementary Material), with an upregulation of most MHC-II molecules
   including HLA-DMA, DMB, DOA, DPA1, DPB1, DQA1, DQA2, DQB1, DQB2, DRA,
   DRB1, DRB5, and DRB6 (Figure S21 in Supplementary Material), suggesting
   that human dCD4 T cells have acquired MHC-II molecules and become
   effector cells that provide a better immune response than pCD4 T cells
   during early pregnancy.

   Since the first trimester of pregnancy is believed to be a
   pro-inflammatory phase, the presence of immune cells in the decidua,
   which are active, functional and carefully controlled rather than being
   suppressed, is important for a healthy pregnancy outcome ([160]59,
   [161]60). Indeed, human dCD4 T cells were reported to express multiple
   T-cell-activation surface markers during early pregnancy, such as the
   early activation antigen CD69, late activation antigen HLA-DR and very
   late antigen 1 (also known as ITGA1 and CD49a), indicating that these
   cells are regionally activated ([162]17, [163]61–[164]66). Consistent
   with that, our data from genome-wide transcriptional profiling showed
   that the genes related to T-cell activation are highly enriched, with
   most of them being upregulated in dCD4 T cells (Figure [165]4B; Figure
   S6B in Supplementary Material), such as CD25, CD38, CD69, CD122, ITGA1
   (CD49a), LAT2, CD40LG, and HLA-DRs (Figure [166]2; Figure S21 in
   Supplementary Material), suggesting that human dCD4 T cells are more
   active than their counterparts in the peripheral blood. In addition,
   dCD4 T cells expressed a high level of proliferation-associated gene
   MKI67 (encoding antigen Ki-67) (Figure S21G in Supplementary Material),
   and were found to be at M phase and show increased proliferation and
   functionality in terms of cytokine production (Figure [167]4). However,
   the mechanisms beyond this enhanced profile of human dCD4 T cells
   during early pregnancy remain poorly understood and require further
   investigation.

   Herein, human dCD4 T cells were observed to highly upregulate both
   transcript and protein levels of chemokine receptors CXCR3 and CCR6
   (Figure [168]2), which are preferentially expressed on Th1 and
   Th17 cells, respectively ([169]67, [170]68). CXCR3^+CCR6^+CD4^+ cells
   were reported to be enriched in conventional Th1 and Th17 cells, as
   well as Th1/17 cells that are characterized by their capacity to
   co-produce IFN-γ and IL-17, with conventional Th1 cells being
   dominantly enriched ([171]69). Our co-expression analysis revealed that
   both CXCR3^+CCR6^+ and CXCR3^+CCR6^− cells are increased in dCD4 versus
   pCD4 T cells (Figure S22 in Supplementary Material), indicating that
   dCD4 T cells probably contain more Th1, Th17, and/or Th1/17 cells.
   Indeed, we found that human dCD4 T cells can produce more IFN-γ,
   IL-17A, and even Foxp3, but not IL-4, upon stimulation with PMA and
   ionomycin (Figures [172]4E–M). However, the cells co-producing
   IFN-γ/IL-17A (IFN-γ^+IL-17A^+, Th1/17 cells), IFN-γ/IL-4
   (IFN-γ^+IL-4^+), IFN-γ/Foxp3 (IFN-γ^+Foxp3^+), IL-17A/IL-4
   (IL-17A^+IL-4^+), IL-17A/Foxp3 (IL-17A^+Foxp3^+), or IL-4/Foxp3
   (IL-4^+Foxp3^+), were very few or undetectable in both pCD4 and dCD4 T
   cells (Figure S22 in Supplementary Material). Therefore, our data
   showed that human dCD4 T cells are a heterogeneous population
   containing conventional Th1, Th17, and Treg cell subsets, whereas the
   cells co-producing IFN-γ, IL-17A, or Foxp3 are almost absent. These
   findings are suggestive of a potential role of these subsets in a
   successful pregnancy.

   At present, a large body of evidence has demonstrated that decidual
   Treg cells contribute to creating a local tolerant microenvironment and
   maintaining a successful pregnancy; however, the role and importance of
   other Th subsets in the decidua is still largely unknown or remains
   controversial ([173]3, [174]16, [175]23, [176]24, [177]70). Consistent
   with our results (Figures [178]4E–L), it was reported that IFN-γ and
   IL-17, secreted by human decidual Th1 and Th17 cells, play an important
   role in vascular remodeling ([179]71) and trophoblast invasion
   ([180]23), and that Th2 cytokines are not prerequisite to maintain a
   normal pregnancy. Nonetheless, a study using chemokine receptor
   expression profiles showed that while Tregs cells are significantly
   enriched in the decidua of early pregnant women, Th17 cells are nearly
   absent ([181]13). Moreover, our results are in contrast with the
   hypothesis that there is a suppressed Th1- and Th17-type immunity
   associated with a predominant Th2-type immunity during normal pregnancy
   ([182]60, [183]71–[184]73). Whether the presence of Th1 and Th17 cells
   in decidua is required or not for a healthy pregnancy outcome needs
   further investigation.

   T-bet is the “master” transcription regulator that directs Th1-cell
   lineage commitment through transactivation of IFN-γ ([185]74). However,
   previous studies showed that rapid IFN-γ production by memory CD4^+ T
   cells occurred in the absence of T-bet expression ([186]75); the level
   of T-bet peaked only and briefly when Ifng^+ CD4 T[E] expanded
   ([187]76). Moreover, in the memory phase, only a small percentage of
   Ifng^+ T cells maintained a T-bet^hi phenotype ([188]76); and
   downregulation of T-bet was required for the development of
   tissue-resident memory T cells in epithelial sites ([189]77).
   Consistent with these findings, we observed that when human dCD4 T
   cells produced a higher level of IFN-γ and displayed a memory phenotype
   (Figure [190]4), there was no difference in TBX21 (encoding T-bet)
   expression between pCD4 and dCD4 T cells (Figure S23 in Supplementary
   Material).

   In eukaryotic organisms, most pre-mRNAs tend to undergo AS, which is a
   ubiquitous and vital mechanism to control gene expression at the co-
   and post-transcriptional level, resulting in production of multiple
   distinct mRNA transcripts and proteins with diverse cellular functions
   ([191]25, [192]78). Blencowe and colleagues performed the first
   analysis of AS complexity in human tissues using high-throughput
   mRNA-Seq datasets (consisting of 17–32 million 32-bp-long reads),
   estimating that there are approximately 100,000 AS events in major
   human tissues, with an intermediate- to high-abundance ([193]35). After
   aligning all the sequenced reads (approximately 30–40 million
   2 × 125-bp paired-end-long reads) to the human reference genome (hg19
   version), between 88 and 93% of reads per sample were uniquely mapped
   (Figure [194]1A), and a total of 127,147 AS events, belonging to 10,281
   genes, were also found in human dCD4 and pCD4 T cells during early
   pregnancy (Figure [195]5A). Furthermore, our data revealed that SE is
   the most common type of AS event (46.5%), whereas RI is the least
   common (1.8%) in human dCD4 and pCD4 T cells (Figure [196]5A). These
   results are consistent with the observations in human hematopoietic
   progenitor cells ([197]26), human embryonic kidney 293 T and Ramos B
   cells ([198]79), and human epithelial (PC3E) and mesenchymal (GS689)
   prostate cancer cell lines ([199]46); however, these results differ
   from the findings in Physcomitrella ([200]80), Arabidopsis ([201]81),
   rice ([202]82), filamentous fungus Trichoderma longibrachiatum
   ([203]83) and Verticillium dahliae ([204]84), where RI is the most
   prevalent mechanism, and also differ from the observation in bovine
   longissimus muscle ([205]36) where A3SS is the primary AS mode.

   Using quantitative AS microarray profiling, recent studies have shown
   that the majority of the genes undergoing differential AS-level changes
   are different from the set of genes showing differential
   expression-level changes during Jurkat T-cell activation ([206]33), and
   many more genes are affected through AS modulation than being affected
   by expression-level modulation in CD4^+ T cells upon CD28
   co-stimulation ([207]85). Similarly, the overlap between the sets of
   genes undergoing differential splicing-level and expression-level
   changes in dCD4 T cells was observed to be very low in our study
   (Figure [208]7A). Functional enrichment analysis revealed that the set
   of genes with changes in both expression and AS levels (748 genes in
   the overlap) is enriched in immune system process, which is distinct
   from the observation in the genes undergoing only AS-level change
   (3,211 gene, enriched in cellular metabolic process) but similar with
   those genes showing only expression-level change (1,958 genes, enriched
   in immune system process) (Figure S24 in Supplementary Material). Since
   the upregulated genes in dCD4 T cells were also significantly enriched
   in immune system process (Figure [209]1), we speculated that AS might
   play a key role in regulating the expression of these upregulated
   genes. However, our data showed that the overlap between sets of genes
   undergoing splicing-level and expression-level upregulation was very
   low: 122 genes in the overlap versus 1,695 genes showing
   expression-level upregulation and 2,607 genes showing splicing-level
   upregulation (Figure [210]7B). Moreover, although dCD4 T cells
   expressed a higher level of IFNG, CXCR3, RORC, IL17A, CCR6, FOXP3,
   IL-10, PDCD1, and CCR7 (Figures [211]2 and [212]4; Figures S5, S7, and
   S23 in Supplementary Material), which are important genes for CD4^+
   T-cell differentiation, function or recruitment, the total and
   differentially expressed AS events in these genes were surprisingly few
   (IFNG, CXCR3, RORC, PDCD1, CCR7) or undetectable (IL17A, IL-10) in both
   pCD4 and dCD4 T cells, or not different between these cell populations
   (CCR6, FOXP3) (Table S7 in Supplementary Material). These data
   indicated that AS does not make a major contribution to regulate gene
   expression of these key molecules, and that changes at the AS event
   level do not imply measurable differences at the gene expression level
   in human dCD4 T cells.

   In summary, we have examined the transcriptional and AS signatures of
   paired dCD4 pCD4 T cells from healthy women at the first trimester of
   normal pregnancy. Our data revealed that dCD4 and pCD4 T cells show a
   distinct transcriptional signature: dCD4 T cells upregulate genes
   involved in the immune system process, but downregulate genes related
   to mRNA catabolic process and the ribosome; human dCD4 T cells stay in
   M phase, and show increased activation, proliferation, and cytokine
   production, as well as contain Th1, Th17, and Treg cell subsets and
   display an effector-memory phenotype. However, dCD4 T cells undergo a
   comparable number of upregulated and downregulated AS events, both of
   which are enriched in the genes related to cellular metabolic process.
   Moreover, the changes at the AS event level do not reflect measurable
   differences at the gene expression level in dCD4 T cells. This study
   thus provides a comprehensive framework of the transcriptional and AS
   landscapes of human dCD4 T cells, which deepens our understanding of
   the characteristic and functionality of these cells during early
   pregnancy.

Ethics Statement

   This study was approved and performed in compliance with the Medical
   Ethics Committee of the International Peace Maternity and Child Health
   Hospital of China Welfare Institute (Shanghai, China). Informed consent
   was obtained from all participants in written form before enrollment.

Author Contributions

   WZ and ZL designed and performed most of the experiments and drafted
   the manuscript. XL, SZ, YZ, XM, TY, FT, X-RL, and JF performed a part
   of the experiments. AK and WZ edited the manuscript. YL supervised the
   study and proofread the manuscript.

Conflict of Interest Statement

   The authors declare that the research was conducted in the absence of
   any commercial or financial relationships that could be construed as a
   potential conflict of interest.

Acknowledgments