Transcriptome analysis reveals differences in mechanisms regulating cessation of luteal function in pregnant and non-pregnant dogs

Abstract

Background: In the domestic dog, corpora lutea (CL) are the only source of progesterone (P4), both in pregnant and non-pregnant cycles because there is no placental steroidogenesis. The absence of an endogenous luteolysin in absence of pregnancy results in long-lasting physiological pseudopregnancy, strongly contrasting with the acute luteolysis observed prepartum. The underlying biological mechanisms and the involvement of P4 signalling remain, however, not fully understood. Therefore, here, next-generation sequencing (RNA-Seq) was performed on CL from the late luteal phase and compared with normally luteolyzing CL collected at the prepartum P4 decrease.

Results: The contrast "luteal regression over luteolysis" yielded 1595 differentially expressed genes (DEG). The CL in late luteal regression were predominantly associated with functional terms linked to extracellular matrix (p = 5.52e-05). Other terms related to transcriptional activity (p = 2.45e-04), and steroid hormone signalling (p = 2.29e-04), which were more highly represented in late regression than during luteolysis. The prepartum luteolysis was associated with immune inflammatory responses (p = 2.87e-14), including acute-phase reaction (p = 4.10e-06). Immune system-related events were also more highly represented in CL derived from normal luteolysis (p = 7.02e-04), compared with those from dogs in which luteolysis was induced with an antigestagen (1480 DEG in total). Additionally, the withdrawal of P4 at mid-gestation resulted in 92 DEG; over-represented terms enriched in antigestagen-treated dogs were related to the inflammatory response (p = 0.005) or response to IL1 (p = 7.29e-05). Terms related to proliferation, e.g., centrosome organization (p = 0.002) and steroid metabolic processes (p = 0.001), prevailed at mid-gestation. Thereby, our results revealed the nature of luteotropic effects of P4 within canine CL. It appears that, even though they result in diminished steroidogenic output, the effect of antigestagens is more related to the withdrawal of P4 support than to the PGF2alpha-related inflammatory reaction observed at physiological parturition.

Conclusions: We report the differential gene expression associated with maintenance and cessation of luteal function in pregnant and non-pregnant dogs. Based on the differentially expressed genes, we indicate functional pathways and gene networks that are potentially involved in the underlying endocrine and molecular mechanisms. This study establishes future research directions that may be helpful in understanding some of the clinical conditions, such as luteal insufficiency, associated with negative pregnancy outcome in dogs.

Keywords: CL; Dog (Canis familiaris); Induced abortion; Luteal regression; Prepartum luteolysis.

Fig. 1

Representative heatmaps showing the RNA-Seq analysis of differentially expressed genes (DEG) of two contrasts: a “luteal regression over luteolysis”, and b “mid-gestation over luteolysis”. For each gene the red color indicates high expression relative to the average expression of the gene, while the blue color indicates low expression. a 1595 DEG were detected for the contrast “luteal regression over luteolysis”, 1054 genes were less and 541 were more highly expressed in luteal regression (p < 0.01, FDR < 0.1), b 1745 DEG were found in the contrast “mid-gestation over luteolysis”, 1055 were less and 690 were more expressed in mid-gestation (P < 0.01, FDR < 0.1). The main functional terms overrepresented in each of the groups are listed (details, including statistics are presented in the text). The entire list of DEG identified genes is provided as Additional files 1 and 2

Fig. 2

Representative heatmaps showing the RNA-Seq analysis of differentially expressed genes (DEG) of two contrasts: a “mid-gestation over antigestagen”, and b “luteolysis over antigestagen”, are presented. For each gene the red color indicates high expression relative to the average expression of the gene, while the blue color indicates low expression. a 492 genes were found for the contrast “mid-gestation over antigestagen” based on the p < 0.01 threshold (shown). Of these, 92 DEG passed the FDR < 0.1 selection (adjusted p-value), of which 38 genes were more and 54 genes were less expressed at mid-gestation. These genes were used for further downstream analyses. b 1480 DEG were identified in the contrast “luteolysis over antigestagen”, 793 were more and 687 were less expressed in the antigestagen group compared to the luteolysis group (p < 0.01, FDR < 0.1). The main functional terms overrepresented in each of the groups are listed (details, including statistics are presented in the text). The entire list of DEG identified genes is provided as Additional files 3 and 4

Fig. 3

Venn diagram showing the intersection of differentially expressed genes (DEG) in the contrast “luteal regression over luteolysis” compared with the contrast “mid-gestation over antigestagen”. Lists of DEG (p < 0.01, FDR < 0.1) presented in Additional files 1 and 3 were used. Three analyses are presented: with up- and down-regulated genes (regardless of log2 ratio), or with upregulated genes (log2 > 1), or with downregulated genes (log2 < −1). When full sets of genes were used (up and down regulated), 11 genes were found overlapping in both contrasts, 1584 genes were found only in the contrast “luteal regression over luteolysis” and 81 genes were found in “mid-gestation over antigestagen”. Adding log2 ratio > 1 to the threshold, 2 over-represented genes overlapped in both contrasts, and for log2 ratio < −1, 5 downregulated genes were found

Fig. 4

Venn diagram showing the intersection of differentially expressed genes (DEG) in the contrast “luteolysis over antigestagen” compared with the contrast “mid-gestation over antigestagen”. Lists of DEG (p < 0.01, FDR < 0.1) presented in Additional files 3 and 4 were used. Three analyses are presented: with up- and down-regulated genes (regardless of log2 ratio), or with upregulated genes (log2 > 1), or with downregulated genes (log2 < −1). When full sets of genes were used (up and down regulated), 36 genes were identified in both contrasts, 1444 genes were found in “luteolysis over antigestagen” only, and 56 genes in “mid-gestation over antigestagen”. Setting the log2 ratio > 1, 4 genes commonly over-represented in both comparisons were found, and with log2 ratio < −1, 17 commonly downregulated genes were detected

Fig. 5

Cytoscape analysis of GO for the contrast “luteal regression over luteolysis” is presented. The functional terms overrepresented in each of the groups are shown. As input differentially expressed genes (DEG) were used (threshold was set at p < 0.01, FDR < 0.1). The redundant and non-informative terms were removed, and the resulting network was manually rearranged. For each network the size of the node implies the number of genes, while the color intensity denotes the level of enrichment (see legend to illustration). Functional networks, which were more highly represented in the luteal regression group (a), refer predominantly to matrix remodeling, to the steroid hormone signaling pathway and to cAMP-mediated signaling. Networks more highly represented during prepartum luteolysis (b) were associated to immune system, inflammatory response and regulation of apoptotic signaling

Fig. 6

Cytoscape analysis of GO for the contrast “luteolysis over antigestagen” is presented. The functional terms overrepresented in each of the groups are shown. As input differentially expressed genes (DEG) were used (threshold was set at p < 0.01, FDR < 0.1). The redundant and non-informative terms were removed, and the resulting network was manually rearranged. For each network the size of the node implies the number of genes, while the color intensity denotes the level of enrichment (see legend to illustration). a The more highly represented functional networks for the prepartum luteolysis group include, e.g., response to wounding, defense response, positive regulation of cellular process, cytoskeleton organization and apoptotic signaling, as well as cell death. b The functional networks more represented in the antigestagen-treated group refer, e.g., to the negative regulation of transcription and networks relating to response to TGFβ

Fig. 7

Expression of selected target genes for each of the investigated contrasts, as determined by Real Time (TaqMan) PCR. a “luteal regression over luteolysis” (t-test: blue = luteal regression > luteolysis; red = luteolysis > luteal regression), b “mid-gestation over luteolysis” (t-test: blue = mid-gestation > luteolysis; red = luteolysis > mid-gestation; black = unchanged), c “mid-gestation over antigestagen-treatment” (t-test: blue = mid-gestation > antigestagen; red = antigestagen > mid-gestation), and d “antigestagen-treatment over luteolysis” (t-test: red = luteolysis > antigestagen; blue = antigestagen > luteolysis; black = unchanged). Due to the uneven distribution of data, logarithmic transformation was performed and the results are presented as geometric means (Xg) + − geometric standard deviation (SD). An unpaired, two-tailed Student’s t-test was applied; p < 0.05 was considered as statistically significant. RGE = relative gene expression

Fig. 8

A cumulative presentation of major conclusions drawn from the present study is shown. Thus, luteal regression in non-pregnant bitches appears to be a slowly ongoing, passive degenerative process leading to corpus albicans formation, associated with structural remodelling processes in the absence of an acute inflammatory reaction. The latter, i.e., acute inflammation, is observed during prepartum luteolysis, and is most probably caused by PGF2alpha of utero-placental origin. The antigestagen-mediated effects seem to relate primarily to inhibition of the luteotropic function of progesterone (P4), rather than to the inflammatory reaction evoked by its withdrawal