Fine-tuned adaptation of embryo-endometrium pairs at implantation revealed by gene regulatory networks Tailored conceptus-maternal communication at implantation

Interactions between embryo and endometrium at implantation are critical for the progression and the issue of pregnancy. These reciprocal actions involve exchange of paracrine signals that govern implantation and placentation. However, it remains unknown how these interactions between the conceptus and the endometrium are coordinated at the level of an individual pregnancy. Under the hypothesis that gene expression of endometrium is dependent on gene expression of extraembryonic tissues, we performed an integrative analysis of transcriptome profiles of paired conceptuses and endometria obtained from pregnancies initiated by artificial insemination. We quantified strong dependence (|r|>0.95, eFDR<0.01) in transcript abundance of genes expressed in the extraembryonic tissues and genes expressed in the endometrium. The profiles of connectivity revealed distinct co-expression patterns of extraembryonic tissues with caruncular and intercaruncular areas of the endometrium. Notably, a subset of highly co-expressed genes between conceptus (n=229) and caruncular areas of the endometrium (n=218, r>0.9999, eFDR<0.001) revealed a blueprint of gene expression specific to each pregnancy. Functional analyses of genes co-expressed between conceptus and endometrium revealed significantly enriched functional modules with critical contribution for implantation and placentation, including “in utero embryonic development”, “placenta development” and “regulation of transcription”. Functional modules were remarkably specific to caruncular or intercaruncular areas of the endometrium. The quantitative and functional association between genes expressed in conceptus and endometrium emphasize a coordinated communication between these two entities in mammals. To our knowledge, we provide first evidence that implantation in mammalian pregnancy relies on the ability of the conceptus and the endometrium to develop a fine-tuned adaptive response characteristic of each pregnancy.


40
In mammals, pregnancy recognition requires a tightly synchronized exchange of signals between the 41 competent embryo and the receptive endometrium. The initiation of this signaling is triggered by key factors 42 produced by the conceptus (1, 2) which are translated by the endometrial cells into actions that will condition 43 the trajectory of embryo development as well as progeny phenotype. In mammalian species, including 44 human, rodents and ruminants, the delicate balance in embryo-maternal communication is affected by the 45 way the embryos are generated (natural mating, artificial insemination, in vitro fertilization somatic cell 46 nuclear transfer) and by the sensor-driver properties of the endometrium defined by intrinsic maternal 47 factors (i.e.: maternal metabolism, ageing) and environmental perturbations (i.e.: pathogens, nutrition) (3-48 5). The concept of sensor property applied to the mammalian endometrium was first proposed in a pioneer 49 paper as was suggested the notion of endometrial plasticity (6). This property was recently confirmed in 50 vitro with an aberrant responsiveness of human endometrial stromal cultured cells in the context of recurrent 51 pregnancy loss (7). Nevertheless, it remains unaddressed whether the mammalian endometrium is able to 52 develop an adaptive embryo-tailored response in a normal pregnancy.

53
In mammalian reproduction, sheep and cattle are research models that have relevantly contributed key 54 insights in the understanding of molecular and physiological pregnancy-associated mechanisms, including 55 the deciphering of embryo-endometrium interactions (8,9). In the bovine species, by gestation days 7-8, 56 the blastocyst enters the uterine lumen. After hatching by days 8-9, the outer monolayer of trophectoderm 57 cells establishes direct contact with the luminal epithelium of the endometrium (10). On gestation days 12-58 13, the blastocyst is ovoid in shape (~2-5 mm) and transitions into a tubular shape by days 14-15. By day 59 ~15, rapidly proliferating trophectoderm cells of the extra-embryonic tissues synthesize and release IFNT 60 (11)(12)(13)(14)(15), which is the major pregnancy recognition signal in ruminants (1,9,16,17). The disrupted release 61 of the oxytocin-dependent pulses of prostaglandin F2 alpha (18) allows maintenance of progesterone 62 production by a functional corpus luteum (18), which is critical for the establishment and progression of 63 pregnancy (1,4,9,12,14,15,19). IFNT actions include induction of numerous classical and non-classical

72
The cross-talk between the conceptus and the endometrium is associated with the expression and 73 regulation of a wealth of genes in each entity (24,25). The nature of the conceptus modifies gene 74 expression of the endometrium in cattle (6,26,27) and decidualizing human endometrial stromal cells (28).

75
Similarly, the endometrium from dams with different fertility potentials (29) or metabolic status (30) 76 influences the gene expression of the conceptus. Despite the growing evidence of the interactions between 77 conceptus and endometrium at the level of gene regulation, the pathways and the functions that result from 78 this interaction have yet to be unveiled. Furthermore, the lack of integrated analysis between paired 79 conceptus and endometrium has made it challenging to advance our understanding of the functional 80 interactions between these two entities in normal pregnancies.

81
Here, we hypothesized that gene expression of extraembryonic tissue is not independent from gene 82 expression of endometrium. In the present study, we carried out an integrative analysis of transcriptome 83 profiles of paired conceptuses and endometria at the onset of implantation aiming at the identification of 84 regulatory pathways that have coordinated expression between the conceptus and endometrium in normal 85 pregnancies. Surprisingly, our results show that at gestation day 18 in cattle, several hundred genes have 86 an expression profile in conceptus and caruncular areas of the endometrium that is unique to each 87 pregnancy. Analyses of genes co-expressed between the conceptus and the paired-associated 88 endometrium revealed significantly enriched functional modules with critical contribution for implantation 89 and placentation. Our data provide evidence that successful implantation in mammalian pregnancy relies 90 on the ability of the endometrium to elicit a fine-tuned adaptive response to the conceptus.

93
We analyzed the RNA-sequencing data that consisted of samples collected from five cattle pregnancies 94 terminated at gestation day 18 (GSE74152 (26)). The conceptus was dissected, and transcriptome data 95 was generated for extraembryonic tissue; whereas the endometrium was dissected into caruncular (gland-96 free) and intercaruncular (containing endometrial glands) areas, and transcriptome data was generated 97 from both regions of the endometrium (Fig 1A). Therefore, the dataset analyzed was comprised of three 98 samples collected from each pregnancy: extraembryonic, caruncular, and intercaruncular tissues ( Fig 1B).

116
The associated expression between two genes can be assessed by correlative metrics (31) within (32,33) 117 or between tissues (33, 34). Thus, we calculated Pearson's coefficient of correlation (r (35)) to test whether 118 there is association between the transcript abundance of genes expressed in extraembryonic tissue and 119 endometrium (caruncular or intercaruncular tissues). We reasoned that under a null hypothesis, the

123
The distribution of correlation coefficients for all pairs of genes expressed in extraembryonic and caruncular 124 tissues averaged 0.13 (Fig 2A), and the equivalent distribution obtained for all pairs of genes expressed in 125 extraembryonic and intercaruncular tissues averaged 0.03 ( Fig 2B). Both distributions deviated significantly 126 from a distribution obtained from shuffled data that disrupted the pairing of the conceptus and endometrium 127 (P < 2.2 -16 , S1 Fig). We calculated the empirical FDR (eFDR) and noted that absolute correlation coefficients 128 in both distributions were highly significant when greater than 0.95 (eFDR < 0.007, S2 Fig, S1 Table). Of

142
Notably, all 9548 genes expressed in extraembryonic tissue where positively (r > 0.95) and negatively (-143 0.95 > r) correlated with genes expressed in caruncular or intercaruncular tissues ( Fig 1C). Eighty percent 144 and 95% of the genes expressed in caruncle tissues were negatively and positively correlated with genes 145 expressed in extraembryonic tissue. Similarly, 83% and 88% of the genes expressed in intercaruncular 146 tissues were negatively and positively correlated with genes expressed in extraembryonic tissue ( Fig 1C).

147
The distribution of degrees of connectivity for significant correlations (|r| > 0.95, eFDR < 0.01) between 148 extraembryonic and caruncular tissues was not equivalent to the distribution observed between 149 extraembryonic and intercaruncular tissues (P < 2.2 -16 ). The genes expressed in extraembryonic tissue 150 were significantly correlated with 295 genes expressed in caruncular tissues on average (median = 101).

157
We then examined if genes co-expressed in extraembryonic tissue and endometrium have expression 158 patterns that are unique to pregnancies. We identified 229 and 218 genes expressed in extraembryonic 159 and caruncular tissues, respectively (|r| > 0.9999, eFDR < 0.001, S1 Table), whose expression profiles 160 produced equivalent dendrograms for extraembryonic and caruncular tissues independently (P = 0.008,  Fig 2F). We did not identify groups of genes co-167 expressed in extraembryonic and intercaruncular tissues capable of producing dendrograms that mirrored 168 each other. These results demonstrate that genes highly co-expressed between extraembryonic and 169 caruncular tissues form a signature that independently distinguishes pregnancies in an equivalent manner.

170
Visualization of co-expressed networks in extraembryonic tissue and endometrium

171
Our analysis was not an exhaustive evaluation of all potential co-expression networks that exist between 172 conceptus and endometrium. Thus, we developed a web interface for dynamic and interactive data 173 visualization based on the co-expression analysis conducted in the present study (36,37) 174 (https://biaselab.shinyapps.io/eet_endo/). The public access to this web application allows a user to

184
The heatmap resultant of clustering the two datasets (extraembryonic and caruncular tissues) showed the 185 formation of an organized co-expression network between the genes expressed in extraembryonic and 186 caruncular tissues (Fig 3A). We identified 36 clusters formed by the genes expressed in extraembryonic 187 tissue that presented enrichment for several biological processes (FDR < 0.2, Fig 3B), where we identified 188 several genes expressed in extraembryonic tissue significantly co-expressed with genes expressed in 189 caruncular tissues (see S1 Data for a list of genes). For instance, 142 genes associated with regulation of

206
for several Biological processes (Fig 3A, S2 Data). Among the genes forming significant co-expression with 207 extraembryonic tissue, we identified 96 genes in cluster 3 associated with "intracellular protein transport", 208 as well as 111 and four genes associated with regulation of transcription in clusters 4 and 5, respectively.

209
Notably, ten genes on cluster 15 were associated with "defense response to virus", and the annotated 210 genes are known to be stimulated by interferon-tau (IFIT1, IFIT3, IFIT5, ISG15, MX1, MX2,  in caruncle (Fig 3C). Second, ten genes associated with "transmembrane transport" in extraembryonic 219 tissue form positive co-expression connections ( ̅ 1 = 0.97, n = 22) with 12 genes associated with "regulation 220 of transcription, DNA-templated" expressed in caruncle (Fig 3D). These results are coherent with a co-221 expression between genes expressed in extraembryonic and caruncular tissues, with functional 222 implications to conceptus attachment and implantation.

224
The independent clustering of the correlation coefficients obtained from the genes expressed in

228
Interestingly, there were 85 and 27 genes associated with "mRNA processing" and "stem cell population 229 maintenance", respectively on cluster 3. On cluster five, we identified 12 genes associated with "negative 230 regulation of cell proliferation" and seven genes associated with "regulation of receptor activity". On cluster  with "oxidation-reduction process", "cell redox homeostasis", "electron transport chain", and "tricarboxylic 250 acid cycle". Cluster four contained 63 genes associated with "regulation of transcription", and cluster seven 251 contained 11 genes associated with "fatty acid beta-oxidation".

252
The

293
Among the genes expressed in caruncular or intercaruncular tissues that were co-expressed with 294 extraembryonic tissues, it was noticeable that several genes were associated with regulation of gene 295 expression. This finding is in line with former publications reporting that the regulatory network needed for 296 endometrial remodeling (48) during attachment is conceptus-dependent. In the caruncular tissue, we 297 specifically identified 15 genes associated with "defense response to virus", of which eight genes had their 298 expression modulated by interferon-tau, produced by the trophoblast between gestation days 9 and 25 (49).

299
This result provide additional knowledge on the biological actions of interferon-tau and other conceptus-300 originated signaling on the remodeling of the caruncle (50).

319
The analyses carried out in this study have provided novel insights into the molecular contribution of

341
All analytical procedures were carried out in R software (36). The files and codes for full reproducibility of 342 the results are listed on the S1 Code.

343
Data analyzed and estimation of gene expression levels

344
The appropriated approval from institutional committees of ethical oversight for animal use in research was 345 obtained as reported previously (26). Briefly, all five cattle gestations were initiated by artificial insemination 346 using semen from a single bull, and later terminated on gestation day 18 for sample collection. We analyzed 347 RNA-seq generated from samples obtained from cattle gestations interrupted at day 18 (n = 5, GSE74152).

350
The reads were aligned to the bovine genome (Bos taurus, UMD 3.1) using STAR aligner (56)

357
Three samples were collected from the same pregnancy, thus the data structure ( Fig 1B) allowed us to 358 quantify the association between genes expressed in extraembryonic tissue and endometrium (caruncular 359 and intercaruncular tissues). We utilized Pearson's coefficient of correlation due to its sensitivity to

366
Testing the resemblance of two distance matrices

367
We calculated distance matrices for extraembryonic tissue and caruncle passed on the Pearson's 368 coefficient of correlation of the expressed genes within tissues. The correlation matrix was subtracted from 369 one to obtain a distance matrix which was used as input for clustering using the method 'complete'. We 370 used the Mantel statistic test implemented in the 'mantel' package to assess the correlation between the 371 two dissimilarity matrices. The significance of the Mantel statistic was assessed by a permutation approach.

372
Clustering of samples, heat maps, and network visualization