Bovine tissues were obtained from the local slaughterhouse. Tissue collection was in accordance with German, British and European laws and provisions for ethical regulation. Cumulus-oocyte complexes (COCs) were collected and categorized exactly as in Janowski et al. (2012). For oocyte RNA preparation, COCs were mechanically denuded using a glass pipette followed by a 2 min trypsin incubation (1x Trypsin-EDTA solution; T4174; Sigma, Deisenhofen, Germany) to remove residual cumulus cells, then frozen at −80°C. Bovine myometrial cells were prepared from excised uteri exactly as in Heng et al. (2008). The isolated myometrial cells were plated at 6,000 cells/well in a 48-well plate and cultured at 37°C with 5% CO₂ for 5 days before stimulation. Bovine ovaries without an obvious corpus luteum were selected for the isolation of granulosa and theca interna cells from mid-phase antral follicles (ca. 6–10 mm diameter) using standard procedures (Holtorf et al., 1989; Bathgate et al., 1999). Follicular fluid, including COCs and granulosa cells, was aspirated from the follicles, and collected in granulosa culture medium (DMEM/F-12 (1:1) with 1% bovine fetal serum, 1% antibiotic-antimycotic (ABAM), 2 mM L-glutamine). After manually removing the COCs, the granulosa cells were harvested by centrifugation (150 g, 10 min) and rinsed once with culture medium as above. Four to six millimeters follicles were obtained by dissection. The theca interna layer of mid-phase antral follicles was peeled off and digested with collagenase IV (1 mg/ml) in Medium 199, including trypsin inhibitor (100 μg/ml) at 38°C for 1 h. The isolated theca cells were collected and washed in Medium 199 with 1% ABAM. After an isotonic shock to eliminate red blood cells, the theca cells were seeded at 20,000 cells/well in a 48-well plate in theca culture medium (McCoy's 5A medium, including 1% ABAM, 2 mM L-glutamine, 10 mM HEPES, 0.1% BSA, 10 ng/ml bovine insulin, 5 μg/ml bovine transferrin, and 5 ng/ml sodium selenite) and incubated at 38°C with 5% CO₂ for ~40 h before stimulation, as indicated. All chemicals and media were from Sigma. Primary bovine mammary epithelial cells (MECs) were prepared according to Yang et al. (2006) and were a generous gift from Professor Hans-Martin Seyfert (Dummerstorf).

In a separate approach, follicular fluids from medium-sized (diameter 3–6 mm) and large (diameter 6–10 mm) antral follicles of freshly collected slaughterhouse ovaries were individually aspirated, flushed with an additional 100 μl of Dulbecco's phosphate buffered saline (DPBS; BioWhittaker, Lonza, Switzerland), and collected into separate tubes. After removal of COCs and brief centrifugation to separate other cells, the supernatants were analyzed for progesterone and estradiol using established in-house radioimmunoassays (Christenson et al., 2013) and for bovine INSL3 by highly specific time-resolved fluorescent immunoassay (TRFIA; Anand-Ivell et al., 2011). Because of inaccuracies in measuring individual antral fluid volumes, hormones were first calculated and compared as total amount per follicle, or as estradiol/progesterone ratio. These individual values were then converted to mean concentrations per follicle by applying average volumes for 3–6 mm follicles of 35 μl and for 6–10 mm follicles of 180 μl after accounting for somatic cells and COCs.

COCs were collected from 3 to 6 mm medium-sized bovine follicles, as above, and subjected to brilliant cresyl blue (BCB) staining to indicate developmental competence, as previously described (Janowski et al., 2012). For in vitro embryo production, COCs were pooled, washed in DPBS, and then subjected to in vitro maturation exactly as described previously (Janowski et al., 2012). The COCs were then fertilized in vitro using frozen-thawed bovine semen, presumptive zygotes manually denuded, and resulting embryos cultured at 38.5°C in 5%CO₂, 5% O₂, and 90% N₂ up to blastocyst stage (day 8; Janowski et al., 2012).

Total RNA from endometrial biopsies, myometrial, granulosa, and theca cells, as well as oocytes and embryos was prepared using the RNAqueous-micro total RNA isolation kit (Ambion, Thermo-Fischer, Darmstadt, Germany) according to the manufacturer's instructions, and then single strand cDNA was prepared following the procedure described in Heng et al. (2008). Semi-quantitative RT-PCR was carried out using the primers listed in Supplementary Table 1, designed using the Primer3 software applied to sequences XM_002694415 (bRXFP1) and NM_001304277 (bRXFP2). 0.25 μM of each primer pair was used to amplify 1 μl of cDNA using EmeraldAMP MAX HS PCR master mix (50% of the reaction volume) (Takara-Clontech, Heidelberg, Germany). For each reaction, DNA was denatured at 95°C for 45 s, annealing was carried out at the temperature indicated in the figure legends, with elongation at 72°C for 1 min, for a total of 35 cycles. As control for the integrity of the RNA, all samples were also validated for the expression of the housekeeping gene bovine rp27a (not shown).

All PCR products were analyzed by gel electrophoresis on a 2% agarose gel including 1% ethidium bromide, with visualization using a GelDoc XR system (Bio-Rad, Gladesville, Australia). For PCR product sequencing, the DNA in each gel band was extracted using the MinElute Gel Extraction Kit (Qiagen, Hilden, Germany), and sequenced (Eurofins-MWG, Ebersberg, Germany).

Preliminary studies, prior to the publication of the bovine genome project, had employed heterologous RT-PCR primers, derived from human sequence information, together with RNA from selected bovine tissues. Initial sequence information obtained was subsequently fully confirmed by the bovine genome project, which is used here for reference. Having confirmed by RT-PCR the correct sequences for the full-length functional bRXFP1 and bRXFP2 transcripts, expression constructs were generated, whereby the signal peptides of the two receptor sequences were replaced by the signal peptide-encoding sequence from the bovine prolactin gene, together with an intermediate short FLAG sequence, exactly as used previously for the expression of the human RXFP1 and RXFP2, and other G-protein coupled receptors (Hsu et al., 2000, 2002). The two final bovine constructs were generated from assembled multiple oligonucleotides (custom gene synthesis by GeneArt, Regensburg, Germany), their sequences confirmed, and inserted into the vector pcDNA3.1/Zeo(+) (Invitrogen) for expression.

For transfection, the Neon electroporation system (Invitrogen) was used: the bRXFP1 or bRXFP2 expression plasmids were mixed with HEK 293T cells in a ratio of ~0.2 μg DNA/50,000 cells/well (48-well plate), and pulsed at 1,300 V for 30 ms to transfect the DNA into the cells. Plasmids were allowed to express for 24 h with the transfected cells cultured in DMEM/F12 (1:1) at 37°C with 5% CO₂. In parallel, HEK 293T cells which had been permanently transfected with human RXFP1 or human RXFP2 (Heng et al., 2008) were plated at 50,000 cells/well in a 48-well plate, and cultured overnight in DMEM/F12 (1:1) with 10% bovine fetal serum, 2 mM L-glutamine and 1% Penicillin-Streptomycin at 37°C with 5% CO₂.

For the transient and permanently transfected HEK 293T cells and the primary myometrial (6,000 cells/well) and theca cells (20,000 cells/well), after the initial pre-culture period (as above) medium was changed and supplemented or not with 1 mM IBMX (3-isobutyl-1-methylxanthine; Sigma) as indicated (30 min for myometrial cells, 10 min for theca cells). Following this pre-incubation, cells were supplemented or not with various peptides, also as indicated in the figure legends, and incubation continued for a further 20 min. Total cAMP was extracted from the combined cells and culture medium from each well and measured by time-resolved fluorescent immunoassay (TRFIA) as described in Anand-Ivell et al. (2007). cAMP production by the transfected HEK 293 cells was measured directly (range1–243 pmol/ml); cAMP from primary myometrial and theca cells made use of the more sensitive TRFIA employing sample acetylation (detection range 0.04–9.72 pmol/ml).

All data were analyzed and plotted using GraphPad Prism 6. The EC50 values were calculated by non-linear regression of log (agonist) against response (three parameter). All column data were analyzed by ANOVA followed by Neuman-Keuls post-hoc test.

At the outset of this project bRXFP1 and bRXFP2 sequence information for the bovine in the NCBI genome database had been missing, incomplete, and in part incorrect, based as it was on partial homologies to other species and application of exon-finding algorithms. RT-PCR strategies were therefore applied using oligonucleotide primers which appeared to be homologous between different species, and using various bovine tissues as a source of potential receptor transcripts. The latest release of the NCBI database (Jan. 2017) now includes complete putative transcript sequences (excluding 5′UTRs) for bRXFP1 (XM_610789.7 and XM_002694415.3) and bRXFP2 (NM_001304277.1), based, however, still on predictions using exon-finding algorithms, homologies to known RXFP1 and RXFP2 sequences in other species, and validated by global RNA-Seq sequence information. Our independent RT-PCR analyses together with genomic BLAST confirmed these two sequences at the expressed transcript level and their corresponding exon-intron structures (Figures 2, 3).

Additionally, our experience from working with RXFP1 and RXFP2 in other species has highlighted the in vivo generation of alternatively spliced transcripts for both genes, particularly involving the short 5′ exons representing individual leucine-rich repeats (Anand-Ivell et al., 2006; Heng et al., 2008). The RT-PCR matrix developed to assess the bovine transcripts was designed to identify possible splice variants in expressing cells and tissues (Figures 2, 3). Indeed, in the bovine both genes exhibited multiple splice variants, in which one or more exons are missing, as well as alternative novel exons (Figures 2, 3). It is to be noted that none of the deleted exons, led to a frame-shift or in-frame stop codons. In contrast, where there appears to be the introduction of a novel exon (bRXFP1 exon 10a; bRXFP2 exons 5a and 11a) this indeed leads to an in-frame stop codon in each case. Only where multiple PCR products for a given cell type or tissue consistently indicate the expression of a full-length, putatively functional transcript can we declare that bRXFP1 or bRXFP2 may be functionally expressed in that tissue. Although the RNA-Seq data included in the NCBI database as validation of a particular sequence also suggests the alternative loss of similar exons, the short-sequence nature of RNA-Seq is not sufficient to absolutely discriminate entire transcript structures as implied by the listed so-called RXFP1 “splice variants” (accession numbers XM_010813810-12, XM_010823315-17, XM_005217632, XM_005217635-36, XM_005194978-79, XM_005194981-82).

Applying the matrix of PCR primers indicated in Figures 2, 3, various cells and tissues of the female reproductive system were interrogated for the expression of bRXFP1 (Figures 4, 5) and bRXFP2 (Figures 6, 7). Not all successful PCR primer combinations are illustrated. bRXFP1 is consistently shown to be expressed in the endometrium, myometrial cells, and theca cells, though not in granulosa cells, nor in oocytes and early embryos (not shown). Arrowheads indicate the correctly sized and sequenced PCR products; alternatively sized products, evident in most samples, and mostly as quantitatively minor bands, proved on sequencing to represent splice variants. Myometrial cells are shown in detail in Figure 5, whereby the dominant PCR product always appears to represents the full-length bRXFP1 transcript.

In contrast, bRXFP2 is expressed as correctly sized PCR products (arrowheads) only in theca cells and oocytes, and weakly in myometrial cells. Other sized products, for example, in early embryos, proved upon sequencing to be exclusively splice variants. Detailed RT-PCR analysis for bovine oocytes is shown in Figure 7, where the major product represents the correct full-length transcript, and minor bands represent splice variants.

RT-PCR analysis was additionally undertaken in all cell and tissue samples from the female reproductive system for transcripts representing bovine RLN3, since potentially this could be a local ligand for RXFP1 in these tissues. For no sample could any specific RLN3-related transcripts be identified (not shown).

INSL3 concentration was measured in follicular fluids from either medium-sized (diameter 3–6 mm) or large (diameter 6–10 mm) antral follicles from freshly collected ovaries. Total INSL3 per follicle was very variable at 13.2 ± 13.8 ng (n = 115) and 24.5 ± 43.1 ng (n = 17), respectively (not significantly different). Assuming approximate antral fluid volumes (also taking into account approximate cell mass) for each category of 35 and 180 μl, respectively, this implies in vivo concentrations of INSL3 of 376 and 136 ng/ml in bovine follicular fluid, which approximately corresponds to the concentration of INSL3 observed in rat testicular interstitial fluid (ca. 400 ng/ml; ref). There was no association between individual INSL3 amount per follicle and either estradiol or progesterone concentration, nor their ratio (data not shown). Nor was there any significant difference in INSL3 content per follicle and the ability of individual oocytes to stain with BCB or not.

From the RNA analysis, two cell types appeared to express both bRXFP1 and bRXFP2 as apparently full-length functional transcripts. To confirm this, primary myometrial cell cultures were stimulated with either porcine RLN or bovine INSL3, measuring acute cAMP production in the presence of the phosphodiesterase inhibitor IBMX as endpoint. For both peptides a significant stimulation was observed at all concentrations applied (Figure 8A). For cultured bovine theca cells, INSL3 (Figure 8C), but neither human nor porcine relaxin (Figure 8B), was able significantly to stimulate cAMP production in a dose-dependent manner. The fold-increase in cAMP observed was similar to that seen for human primary cell types (Anand-Ivell et al., 2007; Heng et al., 2008).

Full-length transcripts for bRXFP1 and bRXFP2 were synthesized, replacing the natural leader sequences with that of the bovine prolactin gene, as previously described for the human equivalent receptors (Hsu et al., 2002), and subcloned into the expression vector pcDNA3.1/Zeo (+). These plasmids were transiently transfected into HEK293T cells and acutely stimulated with various putative ligands as indicated in Figures 9A,B. For bRXFP1 only porcine RLN, recombinant human RLN2 and human RLN3 showed any stimulation within the range of concentration applied (Figure 9A). For bRXFP2 only INSL3 was effective at low concentrations, though porcine RLN, rhRLN, but not RLN3, could stimulate at very high concentration (Figure 9B). The amounts of cAMP generated were similar to previously published studies for the transfected human receptors (e.g., Hsu et al., 2002; Heng et al., 2008). In parallel, similar stimulations were carried out using permanently transfected cells expressing human RXFP1 (hRXFP1) and hRXFP2 (Table 1). Whereas, hRXFP1 indicated EC50s for porcine and human RLN of <1 nM, the corresponding values for bRXFP1 were 10.8 and 48.7 nM, respectively. Similarly for RLN3, the EC50 value differed by almost one order of magnitude (Table 1). For RXFP2, in contrast, EC50 values were largely comparable between human and bovine receptors (Table 1). For its cognate ligand INSL3, whether human or bovine, the EC50s were consistently 0.6–0.7 nM, and for RLN of different sources EC50 values ranged from 10 to 50 nM. hRLN3 was unable to stimulate either hRXFP2 or bRXFP2 (Table 1).

YD was supported in part by Immundiagnostik GmbH, Germany, who however had no influence on any aspect of the research. The other 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.