Characterization of highly proliferative decidual precursor cells during the window of implantation in human endometrium

Pregnancy depends on the wholesale transformation of the endometrium, a process driven by differentiation of endometrial stromal cells (EnSC) into specialist decidual cells. Upon embryo implantation, decidual cells impart the tissue plasticity needed to accommodate a rapidly growing conceptus and invading placenta, although the underlying mechanisms are unclear. Here we characterize a discrete population of highly proliferative mesenchymal cells (hPMC) in midluteal human endometrium, coinciding with the window of embryo implantation. Single-cell transcriptomics demonstrated that hPMC express genes involved in chemotaxis and vascular transmigration. Although distinct from resident EnSC, hPMC also express genes encoding pivotal decidual transcription factors and markers, most prominently prolactin. We further show that hPMC are enriched around spiral arterioles, scattered throughout the stroma, and occasionally present in glandular and luminal epithelium. The abundance of hPMC correlated with the in vitro colony-forming unit activity of midluteal endometrium and, conversely, clonogenic cells in culture express a gene signature partially conserved in hPMC. Cross-referencing of single-cell RNA-sequencing data sets indicated that hPMC differentiate into a recently discovered decidual subpopulation in early pregnancy. Finally, we demonstrate that recurrent pregnancy loss is associated with hPMC depletion. Collectively, our findings characterize midluteal hPMC as novel decidual precursors that are likely derived from circulating bone marrow-derived mesenchymal stem/stromal cells and integral to decidual plasticity in pregnancy. Significance statement Transformation of cycling endometrium into the decidua of pregnancy requires extensive tissue remodeling. Perturbations in this process lead to breakdown of the maternal-fetal interface and miscarriage. Here we report the characterization of decidual precursor cells during the window of implantation. We demonstrate that decidual precursors are clonogenic and primed for exponential growth. They likely originate from bone marrow-derived MSC and give rise to a distinct decidual subpopulation in pregnancy. Recurrent pregnancy loss is associated with loss of decidual precursor cells prior to conception, raising the possibility that they can be harnessed for the prevention of pregnancy disorders, including miscarriages and preterm labor.


77
The uterine mucosa -the endometrium -is a highly regenerative tissue capable of adopting 78 different physiological states during the reproductive years [1,2]. In the midluteal phase of the 79 menstrual cycle, the endometrium starts remodeling intensively in response to elevated  containing granule exocytosis [6,8]. Perturbations that lead to an excessive or persistent pro-104 senescent decidual response presents the embryo with a uterine mucosa that is easily to 105 invade but also prone to breakdown. Clinically, a pro-senescent decidual response is 106 associated with rapid conceptions (i.e. 'super-fertility') and recurrent pregnancy loss [8,16,17].

119
Based on single-cell transcriptomics, we recently described the presence of a discrete 120 population of highly proliferative mesenchymal cells (hPMC) in midluteal endometrium [8].

121
Here, we report on the characterization of these cells. Our findings indicate that hPMC are the 122 human equivalent of bone marrow-derived decidual precursor cells in mice. We provide 123 evidence that hPMC impart clonogenicity of midluteal endometrium and give rise to a distinct 124 subpopulation of decidual cells at the maternal-fetal interface in pregnancy. We further 125 demonstrate that a lack of these novel human decidual precursor cells is associated with 126 recurrent pregnancy loss.   Table S1.

141
Endometrial tissue processing, primary cell culture, and CFU assay

142
Processing of endometrial biopsies, isolation of sushi domain containing 2 (SUSD2)-positive 143 cells using magnetic-activated cell sorting, and culturing of primary endometrial cells were 144 carried out as described in our published step-by-step protocol [21]. CFU assays were also 145 performed as described previously [22]. Briefly, EnSC were seeded at a clonal density of 30 146 cells/cm 2 on 1 mg/ml fibronectin-coated 6-well plates and cultured in 10% DMEM/F12 medium 147 supplemented with basic fibroblast growth factor (bFGF; 10 ng/ml; Merck Millipore, Watford, 148 UK) for 10 days with a 50 % media change on day 7. Colonies were monitored microscopically 149 to ensure that they were derived from single cells. After 10 days, colonies were washed with 150 PBS and fixed in 4% formaldehyde for 10 min at room temperature (RT) before staining with

217
Mann-Whitney U test, student t-test, and one-way ANOVA on ranks (Kruskal-Wallis) test were 218 used as appropriate. Spearman rank-order correlation coefficient was calculated to measure 219 the strength and direction of association between two variables. P < 0.05 was considered 220 statistically significant.

289
We reasoned that the time-dependent increase in anillin + cells in the stroma could be  was also quantified (Fig. S3) [36]. The samples were selected to coincide with the implantation 296 window (LH+7 and LH+9) or the late-secretory phase of the cycle (LH+11). Anillin + CD56 + cells 297 were noticeably absent in all but one sample at LH+7 but their abundance increased markedly 298 upon transition to the late-secretory phase in parallel with anillin + CD163 + macrophages (Fig.   299   2D). The reverse pattern was observed for anillin + CD34 + cells. At the start of the implantation 300 window (LH+7), 61% of anillin + cells did not express IC markers, but this dropped to 32% by 301 the late-secretory phase of the cycle (Fig. 2E).

319
We reasoned that if hPMC are derived from circulating progenitor cells and account for 320 endometrial clonicity during the implantation window, the transcriptomic profiles of clonogenic 321 cells in culture should exhibit less variability when compared to cultured resident PVC or EnSC 322 that have been exposed to iterative menstrual cycles in vivo. To test this hypothesis, we used 323 magnetic-activated cell sorting to isolate SUSD2 + PVC and SUSD2 -EnSC from 3 different 324 biopsies and subjected both cell fractions to standard cultures as well as CFU assays (Fig. 4 Table S4). Conversely, differences in gene expression between 349 clonal and resident cells in vitro were also partially maintained between hPMC and EnSC in 350 vivo (P = 6.1 ´ 10 -19 ) (Fig. 4C). Mining of the data provided further evidence that hPMC are the 351 progenitors of colony-forming cells in culture. As anticipated, proliferation-associated genes 352 were higher expressed in eMSC/TA cells when compared to cultured EnSC or PVC (Fig. 4D),

364
We reported previously that recurrent pregnancy loss, defined here as 3 or more miscarriages, 365 is associated with loss of eMSC and TA cells, as assessed by CFU assays on freshly isolated 366 PVC and EnSC, respectively, from midluteal biopsies [14]. Hence, we examined if this 367 prevalent reproductive disorder is also associated with loss of anillin + hPMC in vivo. To mitigate 368 against interference from anillin + IC, the analysis was restricted to endometrial biopsies 369 obtained on LH+7, a timepoint when anillin + CD56 + cells are virtually absent (Fig. 2D). A total 370 of 30 biopsies from RPL patients (n = 15) and control subjects (n = 15) were processed for 371 immunofluorescence microscopy to quantify anillin + cells in the stromal compartment as well 372 anillin + cells co-expressing CD163 or CD34. While the percentage of anillin + CD163 + cells did 373 not differ between the two clinical groups (Fig. 5A), a significant increase in anillin + CD34 + cells 374 was observed in RPL patients when compared to control subjects (P < 0.05, Mann-Whitney 375 test; Fig. 5B). Conversely, RPL was associated with a reciprocal reduction in hPMC, i.e. anillin + 376 cells that do not express IC markers (P < 0.05, Mann-Whitney U test; Fig. 5C). The abundance 377 of hPMC population was noticeably more variable in tissue samples from RPL patients when 378 compared to control subjects (Fig. 5C)