MicroRNA Expression in Bovine Cumulus Cells in Relation to Oocyte Quality

Cumulus cells play an essential role during oocyte maturation and the acquisition of fertilizability and developmental competence. Micro(mi)RNAs can post-transcriptionally regulate mRNA expression, and we hypothesized that miRNA profiles in cumulus cells could serve as an indicator of oocyte quality. Cumulus cell biopsies from cumulus−oocyte−complexes that either yielded a blastocyst or failed to cleave after exposure to sperm cells were analyzed for miRNA expression. On average, 332 miRNA species with more than 10 reads and 240 miRNA species with more than 50 reads were identified in cumulus cells; this included nine previously undescribed microRNAs. The most highly expressed miRNAs in cumulus cells were miR-21, members of the let-7 family and miR-155. However, no repeatable differences in miRNA expression between the cumulus cells from oocytes that became blastocysts versus those from non-cleaved oocytes were identified. Further examination of individual cumulus cell samples showed a wide variability in miRNA expression level. We therefore conclude that miRNA expression in cumulus cells cannot be used as an oocyte quality marker.


Introduction
Mammalian oocytes acquire the capacity to become fertilized and acquire developmental competence during their final maturation within the ovary. These oocytes originate from female primordial germ cells that proliferated while migrating towards the gonadal ridges and continue to proliferate until gonadal sex differentiation. The cells are classified as oocytes from the moment they stop dividing and enter meiosis [1]. Meiosis in developing oocytes arrests at the first prophase, and depending on the species, this meiotic arrest can be maintained for decades [2]. As early as this time point, the oocyte acquires competence to develop into an embryo after fertilization [3]. The oocyte only resumes meiosis shortly before ovulation, in response to luteinizing hormone stimulation. After the resumption of meiosis and in addition to chromosome segregation, the oocyte needs to redistribute organelles and proteins in a process known as cytoplasmic maturation [4]. For these processes, the oocyte relies on molecules present within the follicular fluid and on intimate contact with the somatic cells that surround the oocyte, known as cumulus cells. Together, they form the cumulus−oocyte−complex (COC). Tightly-regulated communication between the oocyte and surrounding cumulus cells is important at various stages, including oocyte growth, maturation and fertilization [5][6][7]. Exactly how the cumulus cells contribute to the fertilizability and developmental competence of the oocyte, hereafter referred to as oocyte quality, is not known. From in vitro maturation

Comparison of Individual versus Group Culture
In order to identify miRNA species expressed in cumulus cells that could potentially predict quality of the enclosed oocyte, cumulus biopsies were obtained from COCs after in vitro maturation. Subsequent individual fertilization of COCs and individual embryo culture were performed to follow the progression of each oocyte individually and thereby allow correlation of the cumulus miRNA expression profile to oocyte quality, retrospectively. First, to analyze whether individual embryo culture affected the efficiency of blastocyst formation, embryos were cultured either individually, Non-coding RNA 2017, 3, 12 3 of 15 but with shared culture medium, or in groups of . The percentages of oocytes that cleaved or developed into blastocysts were similar for the two culture conditions (Table 1), demonstrating the validity of the single embryo culture system.

Sequencing of miRNAs in Cumulus Cells
In order to determine whether the miRNA expression profiles of cumulus biopsies from COCs that contained fertilizable and developmentally-competent oocytes (i.e., yielded a blastocyst) differed from those that contained oocytes of poor quality, cells were harvested for small RNA sequencing. After COCs had been matured, a biopsy of the cumulus cells was obtained, and the COCs were cultured individually for fertilization and embryo development.
Cumulus cells surrounding an oocyte that (1) developed into a blastocyst or (2) failed to cleave after exposure to sperm were pooled to ensure sufficient amounts of RNA (each pooled sample contained cumulus cells from 10 to 16 individual COCs). Three pooled samples of cumulus cells harvested from oocytes that did not cleave (referred to as A, B and C) and three pooled samples of cumulus cells that had enclosed oocytes that developed into blastocysts (D, E and F) were sequenced for small RNAs; they yielded an average of 14 million reads ( Figure 1). miRNAs were identified according to entries in miRBase Release 20. A mean of 332 miRNA species had more than 10 reads (Figure 2, blue bars), while 240 miRNA species had more than 50 reads per sample (Figure 2, red bars). The majority of the miRNAs were present in all samples; of the miRNA species identified, 276 with more than 10 counts and 205 with more than 50 counts were present in all samples. No repeatable differences in the numbers of miRNAs identified were present between the two types of sample ( Figure 2).

Figure 1.
Total number of micro RNA (miRNA) reads in each sample. The total number of miRNA sized reads for each sample of cumulus cells pooled from 10 to 16 bovine cumulus-oocyte complexes; on average, 14 million reads were obtained per sample, as indicated by the black line. The red dotted line marks five million reads, the amount considered sufficient for meaningful analysis. A, B, C = cumulus pieces surrounding oocytes that failed to cleave after exposure to sperm; D, E, F = cumulus pieces surrounding oocytes that developed into a blastocyst after fertilization.
Non-coding RNA 2017, 3, 12 4 of 15 miRNAs were identified according to entries in miRBase Release 20. A mean of 332 miRNA species had more than 10 reads (Figure 2, blue bars), while 240 miRNA species had more than 50 reads per sample (Figure 2, red bars). The majority of the miRNAs were present in all samples; of the miRNA species identified, 276 with more than 10 counts and 205 with more than 50 counts were present in all samples. No repeatable differences in the numbers of miRNAs identified were present between the two types of sample ( Figure 2). In short, cumulus cells harvested from the COCs of oocytes that failed to cleave after exposure to sperm and those that yielded a blastocyst showed a similar expression of miRNAs. Of all of the identified miRNAs, miR-21-5p was the most abundant (>2,000,000 reads, Table 2), but the number of reads was similar between the two groups. Different members of the let-7 family, miR-155 and miR-99a-5p also gave more than 100,000 reads in each group (Table 2).  In short, cumulus cells harvested from the COCs of oocytes that failed to cleave after exposure to sperm and those that yielded a blastocyst showed a similar expression of miRNAs. Of all of the identified miRNAs, miR-21-5p was the most abundant (>2,000,000 reads, Table 2), but the number of reads was similar between the two groups. Different members of the let-7 family, miR-155 and miR-99a-5p also gave more than 100,000 reads in each group (Table 2).
When comparing the miRNAs expressed in cumulus cells from around an oocyte that developed into a blastocyst after fertilization with those surrounding an oocyte that had not cleaved, a high variability between the different pooled samples and no repeatable differences between the groups were observed (Figure 3). Samples A and B (cumulus cells enclosing oocytes that did not cleave after exposure to sperm) and Samples F and D (cumulus cells surrounding oocytes that developed into blastocysts) grouped together after hierarchical clustering. By contrast, Samples C (cumulus cells enclosing oocytes that did not cleave) and E (cumulus cells surrounding oocytes that developed into blastocysts) exhibited a markedly different miRNA expression profile; this was also apparent after principal component analysis ( Figure 4). In general, within the tested groups, the expression of miRNAs was variable, which presumably contributed to the absence of differences in the expression profiles between the groups. The list of the 20 most abundantly-expressed miRNAs in bovine cumulus cells. NC = cumulus pieces enclosing oocytes that failed to cleave after exposure to sperm; Blast = cumulus pieces enclosing oocytes that developed into a blastocyst after fertilization.    The PCA was performed using the 50 miRNAs that had the largest variation across all samples. The features have been shifted to be zero centered (i.e., the mean value across samples was shifted to zero) and scaled to have unit variance (i.e., variance across samples was scaled to one) before the analysis. The groups do not cluster, not least because Samples B and E differ greatly from the rest for the primary component and from each other for the secondary component. Red circles = cumulus pieces surrounding oocytes that failed to cleave after exposure to sperm; blue circles = cumulus pieces surrounding oocytes that developed into blastocysts after fertilization.

Quantitative real-time reverse transcription-PCR (qRT-PCR) from Pooled Cumulus Complex Samples
Although miRNA expression did not differ significantly between the two groups, expression of miRNAs for which the abundance appeared to differ according to the heat map ( Figure 3) was further examined by qRT-PCR. The selected miRNAs had low or moderate total reads at sequencing, and to compensate for the anticipated low expression levels, pooled samples of cumulus cells from individually-cultured COCs were analyzed. The expression of miR-214, miR-424 and 2284ab was below the level of detection by qRT-PCR, and indeed, the read numbers for these miRNAs were around or below 100 (Table 3). For miR-342, miR-2478 and miR424-3p, expression levels (Table 3) were similar in the cumulus cells surrounding an oocyte that gave rise to a blastocyst to those in cumulus cells around oocytes that did not cleave ( Figure 5). The PCA was performed using the 50 miRNAs that had the largest variation across all samples. The features have been shifted to be zero centered (i.e., the mean value across samples was shifted to zero) and scaled to have unit variance (i.e., variance across samples was scaled to one) before the analysis. The groups do not cluster, not least because Samples B and E differ greatly from the rest for the primary component and from each other for the secondary component. Red circles = cumulus pieces surrounding oocytes that failed to cleave after exposure to sperm; blue circles = cumulus pieces surrounding oocytes that developed into blastocysts after fertilization.

Quantitative Real-Time Reverse Transcription-PCR (qRT-PCR) from Pooled Cumulus Complex Samples
Although miRNA expression did not differ significantly between the two groups, expression of miRNAs for which the abundance appeared to differ according to the heat map ( Figure 3) was further examined by qRT-PCR. The selected miRNAs had low or moderate total reads at sequencing, and to compensate for the anticipated low expression levels, pooled samples of cumulus cells from individually-cultured COCs were analyzed. The expression of miR-214, miR-424 and 2284ab was below the level of detection by qRT-PCR, and indeed, the read numbers for these miRNAs were around or below 100 (Table 3). For miR-342, miR-2478 and miR424-3p, expression levels (Table 3) were similar in the cumulus cells surrounding an oocyte that gave rise to a blastocyst to those in cumulus cells around oocytes that did not cleave ( Figure 5).

qRT-PCR from Individual Cumulus Complex Samples
For both the sequencing and the initial qRT-PCR analysis, pooled samples of cumulus cells from individually-cultured COCs were used. However, since oocyte quality is likely to be dependent on many different factors acting at different times and on different pathways, pooling of the samples might obscure miRNAs that are involved in regulating oocyte competence. Therefore, expression levels of candidate miRNAs were determined using qRT-PCR in cumulus cells from individual COCs. The selection of miRNAs examined was based on the sequencing results; four of the selected

qRT-PCR from Individual Cumulus Complex Samples
For both the sequencing and the initial qRT-PCR analysis, pooled samples of cumulus cells from individually-cultured COCs were used. However, since oocyte quality is likely to be dependent on many different factors acting at different times and on different pathways, pooling of the samples might obscure miRNAs that are involved in regulating oocyte competence. Therefore, expression levels of candidate miRNAs were determined using qRT-PCR in cumulus cells from individual COCs. The selection of miRNAs examined was based on the sequencing results; four of the selected miRNAs had been highly expressed and showed a moderate difference in expression between 'non-cleaved' COCs and COCs that gave rise to a blastocyst, namely let-7g, let-7i, miR-222 and miR-218. For all four miRNAs, a high variability in miRNA expression between the twelve individual samples within a group was observed ( Figure 6).

qRT-PCR from Individual Cumulus Complex Samples
For both the sequencing and the initial qRT-PCR analysis, pooled samples of cumulus cells from individually-cultured COCs were used. However, since oocyte quality is likely to be dependent on many different factors acting at different times and on different pathways, pooling of the samples might obscure miRNAs that are involved in regulating oocyte competence. Therefore, expression levels of candidate miRNAs were determined using qRT-PCR in cumulus cells from individual COCs. The selection of miRNAs examined was based on the sequencing results; four of the selected miRNAs had been highly expressed and showed a moderate difference in expression between 'non-cleaved' COCs and COCs that gave rise to a blastocyst, namely let-7g, let-7i, miR-222 and miR-218. For all four miRNAs, a high variability in miRNA expression between the twelve individual samples within a group was observed ( Figure 6).

let-7g
Relative expression  Not-cleaved Blastocyst Finally, analysis of the sequencing data indicated 178 novel putative miRNAs. Not all of these miRNAs were detected in the cumulus cell samples; however, nine novel miRNA species were detected with a mean of >10 reads, indicating that the miRNAs are indeed generated and expressed (Table 4). However, the expression levels of these novel miRNAs did not differ between the non-cleaved and blastocyst groups. Table 4. Novel expressed miRNAs, predicted and expressed miRNAs identified from read data based on read count and secondary structure according to the miRPara (miRNA prediction software tool) classification score. The description of chromosomal location is shown with Chr (chromosome), start and stop.

Discussion
In this study, the miRNA expression profiles of cumulus cells from developmentally-competent COCs were compared with those from around oocytes did not cleave after exposure to sperm. As far as we know, this is the first study to examine whether cumulus cell miRNA expression can be used to predict the quality of oocyte before fertilization. In our study, samples of cumulus cells from individually-fertilized and cultured COCs were used and retrospectively assigned to post-fertilization developmental competence groups.
For identification of miRNAs expressed in cumulus cells, next generation sequencing was used. From both groups, three pooled samples of cumulus cells, each consisting of cells from around 10 COCs, were analyzed. Cumulus cells surrounding an oocyte that did not cleave after sperm exposure or developed into a blastocyst were chosen, to have the two most different stages to see a change in miRNA expression. However, no specific miRNA species were identified that could be correlated to oocyte quality, suggesting that miRNA expression in cumulus cells is not predictive of oocyte quality. Since our aim was to identify competent oocytes before fertilization, we decided to compare miRNA expression patterns in cumulus cells that did not cleave after exposure to sperm, versus those that developed into a blastocyst after exposure to sperm. This did not allow us to distinguish between those oocytes that were fertilizable and those that were not. For the identification of cumulus markers that would predict embryo developmental competence, comparison between cells originating from embryos that remained at the two-cell stage versus those that become a blastocyst would be useful, but this would always require fertilization and the generation of embryos. Oocyte fertilizability and developmental competence are dependent on various factors, and pooling of the cumulus pieces might have obscured any influence of individual miRNAs. In this respect, we further showed that the expression levels of several selected miRNAs varied greatly between cumulus complexes from individual COCs; at least for these miRNAs, the variation in expression was similar for competent and incompetent COCs. Oocytes from slaughterhouse ovaries were used in this study, which represent a heterogeneous population. Possibly, a more homogeneous oocyte population, for example originating from one breed, would reveal specific miRNAs. Our data indicate however that these could not be used as markers in a heterogeneous population, such as the human population.
A study of human oocytes reported a change in expression level between the germinal vesicle and the metaphase II stage for 15 of 722 miRNAs expressed during maturation [37], suggesting specific roles during maturation and early embryo development. It would be interesting to determine which miRNA expression levels change the most in bovine cumulus cell complexes during maturation. On a similar note, the difference in miRNA expression patterns between outer cumulus and inner cumulus, i.e., the cells furthest from or closest to the oocyte, might shed more light on the function of miRNA in cumulus cells.
Our sequencing results showed that miR-21-5p is the most abundant miRNA in bovine cumulus cells, followed by various members of the let-7 family and miR-155. In human cumulus cells, members of the let-7 family and miR-21 are also the most abundant miRNAs; however, their functions remain unclear [14]. An anti-apoptotic effect was proposed for miR-21 in mouse granulosa cells; however, it was not clear how this effect was achieved, because no change in apoptotic protein expression was observed after miR-21 knockdown [39]. A relatively low expression of miR-21 was detected in bovine oocytes, but the expression increased dramatically during early embryonic development [36]; similarly, a 25-fold increase in miR-21 expression was reported in pig cumulus cells during maturation [40]. A potential target of miR-21 is programmed cell death protein 4 (PDCD4), and interaction between miR-21 and PDCD4 has been reported in pig oocytes [40]. Whether miR-21 has a similar function in bovine cumulus cells remains to be determined.
Analyzing miRNA expression in oocytes and cumulus cells using a PCR array showed a negative fold change of miR-155 expression between bovine oocytes and cumulus cells, indicating higher expression in cumulus cells than in oocytes. Interestingly, this fold change increased during maturation, indicating that either the expression levels in the cumulus cells decreased or the level in the oocyte increased, or both [33]. miR-155 was also found to be abundantly expressed in exosomes recovered from bovine follicular fluid, and the expression level was higher in fluid from follicles that contained a growing oocyte compared to those with a fully-grown oocyte [41]. Based on the high expression levels of miR-155 in cumulus cells, it seems likely that this miRNA is produced by the cumulus cells themselves, although it cannot be excluded that cumulus cells take up miR-155 from the follicular fluid.
High expression levels of various let-7 family members have been reported in cumulus cells; among the most abundant let-7 family members were let-7f, let-7i and let-7g. Let-7 was also abundant in human and mouse cumulus cells [42]. Interestingly, let-7 miRNAs are negatively associated with pluripotency [43], although they are highly expressed in oocytes [35,44]. This indicates a high degree of tissue specificity of miRNA expression and underlines the likelihood that they are involved in regulating different processes at specific time points.
Although target prediction methodologies are improving, computational prediction of miRNA targets needs to be experimentally validated since predicted target sequences can give rise to false positives. In this respect, examination of candidate gene expression after overexpressing or downregulation of miRNAs appears essential, preferentially combined with a demonstration of an interaction between the miRNA and its target mRNA, for instance using luciferase reporter constructs [45].
Overall, the results of this study suggest that the quality of an oocyte within a COC before fertilization cannot be predicted by the miRNA expression palette of the cumulus cells. However, the high and dynamically changing expression levels of various miRNA species suggest that they are important for cumulus cell function. Whether cumulus cell miRNAs are transported into the oocyte remains to be investigated. The identification of novel miRNAs expressed by cumulus cells emphasizes that much remains to be discovered about these regulatory molecules.

Materials and Methods
All chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA) unless otherwise indicated.

Collection of Cumulus−Oocyte−Complexes, In Vitro Maturation and Fertilization
Cumulus−oocyte−complexes were collected and cultured as described previously [46]. COCs were matured for 23 h in groups of 60-70 in NaHCO 3 -buffered M199 (Gibco BRL, Paisley, U.K.) supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin (Gibco BRL) and 0.05 IU follicle stimulating hormone/mL (Organon, Oss, The Netherlands) at 39 • C in a humidified atmosphere of 5% CO 2 -in-air. After maturation, small pieces of the cumulus complexes were retrieved using a narrow-bore Pasteur pipette. These cell clumps were transferred to lysis buffer (Exiqon, Vedbaek, Denmark) containing 10 µL/mL β-mercaptoethanol, labelled and stored at −80 • C until further use. Subsequent individual fertilization and embryo culture enabled retrospective linking of the cumulus pieces to the quality of the oocyte that they had enclosed.
In vitro fertilization was performed as described previously [47]. In short, Percoll-washed sperm cells from a bull of proven fertility were added to fertilization medium supplemented with 1.8 IU/mL heparin, 20 µM D-penicillamine, 10 µM hypotaurine and 1 µM epinephrine at a final concentration of 1 × 10 6 sperm cells/mL. Terasaki microwell plates (Nalge Nunc International, Rochester, NY, USA) were used to allow single oocyte fertilization, and one oocyte per well was incubated in 10 µL of fertilization medium and covered with light mineral oil (Irvine Scientific, Santa Ana, CA, USA). After 18-22 h at 39 • C in a humidified atmosphere containing 5% CO 2 and 7% O 2 , presumptive zygotes from single oocyte fertilization were denuded by repeated aspiration through a narrow pipette; presumptive zygotes from group culture were denuded by vortexing for 3 min; all denuded zygotes were transferred to synthetic oviductal fluid (SOF) for further culture [46]. On day 5 of culture, cleaved embryos were transferred to a new well containing fresh SOF and cultured until day 8.

RNA Isolation
Total RNA was isolated using the miCURY RNA isolation kit (Exiqon), according to the manufacturer's protocol. In short, samples were lysed with 350 µL lysis buffer containing 10 µL/mL β-mercaptoethanol and shearing stress was applied by aspiration through a 23-gauge needle, followed by the addition of 200 µL 100% ethanol. The samples were loaded onto the columns provided, washed by centrifugation, and the RNA was eluted using 50 µL elution buffer.
An additional concentration step was performed on cumulus cell samples from individual COCs. To do this, the RNeasy MinElute Cleanup Kit (Qiagen, Hilden, Germany) was used, and the RNA was isolated as per the manufacturer's instructions. In brief, 350 µL RLT buffer and 675 µL 100% ethanol were added to 100 µL of sample. This solution was transferred to an RNeasy MinElute spin column, and after three centrifugation and washing steps, the RNA was eluted using 17 µL RNase-free water.

Library Preparation and Next Generation Sequencing
Library preparation and next generation sequencing was performed by Exiqon. The concentration of total RNA from pooled cumulus samples (10)(11)(12)(13)(14)(15)(16) varied between 10 and 20 ng/µL. To obtain a minimum of 100 ng total RNA, 45 µL of each sample were converted into miRNA next generation sequencing libraries using a NEBNEXT kit (New England Biolabs, Ipswich, MA, USA) as per the manufacturer's instructions. Adaptors were ligated to the ends of the sequences, and libraries were generated using 3 and 5 rapid amplification of cDNA ends (RACE)-like protocols. Libraries were purified after 15-cycle pre-PCR using QiaQuick columns (Qiagen) and size fractioned on a LabChip XT (Caliper, Hopkinton, MA, USA). Bands representing 15-40-bp fragments were excised with an automated gel cutter, evaluated using a Bioanalyzer DNA 2100 chip and quantified by qPCR. Samples were normalized and pooled in equimolar concentrations [48] and sequenced using v3 sequencing (Illumina Next Seq 500 system, San Diego, CA, USA). Eight-level quality score-binning was used, enabling a more compact storage of raw sequences [49].
The miRNA sequencing data have been deposited in NCBI's Gene Expression Omnibus database and are accessible through Geo series Accession Number GSE94771.

First-Strand Synthesis and qRT-PCR
The universal cDNA synthesis kit II (Exiqon) was used to perform first-strand synthesis, with 10 µL template RNA in a total volume of 20 µL as per the manufacturer's instructions. Samples were incubated for 1 h at 42 • C followed by 5 min at 95 • C and then immediately cooled on ice.
PCR was performed in a total volume of 15 µL with 1 µL cDNA, 1.5 µL locked nucleic acid (LNA) primer pair (Exiqon) and 7.5 µL PCR master mix (Bio-Rad, Hercules, CA, USA). Quantitative real-time PCR was performed using a CFX Connect™ Real-Time PCR detection system (Bio-Rad). The program started with 10 min at 95 • C followed by 40 cycles each of 10 s at 95 • C and 60 s at 60 • C. Melting curves were plotted after each cycle series.
A standard curve of a 3-fold dilution series was obtained by plotting the log of the starting amount against the cycle threshold value of the dilution series. Different potential reference miRNAs were tested, and in accordance with GeNorm guidelines [50], the three small RNAs with the most stable M-value were considered optimal and adequate for normalization (Appendix A: Figure A1). The geometric means of expression for the reference miRNAs miR-26a, miR-191 and let-7a were used for normalization of the target miRNAs.

Statistical Analysis
Comparisons between culture conditions and the resulting cleavage and blastocyst production rates were performed using Student's t-tests. Threshold cycle (Ct)-values of possible reference miRNAs were analyzed using GeNorm [50].