Characterization of microRNA, mRNA, and Protein Expression in Large and Cystic Bovine Follicles

Cystic ovarian disease (COD) is a major contributor to infertility in cattle, with 30% of cows 27 developing ovarian cysts during a given lactation and becoming anovulatory. The cause of COD has 28 remained elusive and is thought to be multifactorial, with angiogenic and genetic contributions proposed. 29 There is an increasing body of work suggesting that microRNAs (miRNAs) may be involved in a number 30 of ovarian-based reproductive disorders. Changes in miRNA expression can affect a multitude of genes 31 and may be important regulators of the dynamic processes involved in each ovarian cycle. We 32 hypothesized that miRNAs are integral to the regulation of critical ovarian processes such as 33 angiogenesis and follicular development and that altered miRNA expression contributes to the onset and 34 progression of ovarian dysfunction and reproductive disorders. Eight miRNAs (miR-15a, -18a, -20a, - 35 21, -29a, -126, -132, Let7a) known to target vascular endothelial growth factor (VEGF), 36 thrombospondin-1 (TSP-1), and play a role in both angiogenesis and folliculogenesis, were selected for 37 analysis. miRNA, mRNA, and protein expression were analyzed in large and cystic bovine follicles using 38 qPCR, western blot, and immunohistochemistry. miR-29a was found to be upregulated in cystic follicles, 39 whereas miR-132 expression was downregulated. VEGF expression increased in cystic follicles at both 40 the transcript and protein level, while no significant differences in TSP-1 expression was observed. Our 41 findings suggest that miR-29a and miR-132 may play an important role in maintaining the balance 42 between VEGF and TSP-1 expression in the ovary, and if disrupted, could potentiate ovarian-based 43 reproductive disorders such as PCOS and cystic ovarian disorder. and imaged using a Bio-Rad ChemiDoc XRS+ system and relative densitometry was determined using Bio-Rad Image Lab software. key cystic the expression patterns of select miRNAs thought be involved in the of ovarian angiogenesis folliculogenesis. demonstrated an increase in miR-29a and decrease in miR-132, along with a significant increase in the mRNA and protein levels of VEGF in granulosa cells, and a significant increase in protein expression levels of VEGFR2, TSP-1, and CD36 in the granulosa and theca cell layers of cystic follicles. These results demonstrate the increased vascular density and pro-angiogenic stimulation of cystic follicles and suggest that miR-29a and miR-132 may be potential mediators of cystic ovarian disease. Together, these results suggest that an increase in pro-angiogenic signalling within the granulosa cells of cystic follicles may not be counteracted with anti-angiogenic signalling or appropriate miRNA regulation, thereby leading to proliferation of vasculature and development of cystic, anovulatory follicles.


INTRODUCTION 48
Reproductive insufficiency results in a significant economic burden to the cattle industry. Cystic 49 ovarian disease (COD), is characterized by the development of large, anovulatory follicles, and is a major 50 hormonal support, as well as facilitate the systemic distribution of steroid hormones produced within the 76 ovary. If angiogenesis throughout the ovarian cycle is disrupted in any way, this can lead to attenuation 77 of follicle growth, interference of ovulation, and affect the development and function of the follicle and 78 corpus luteum [8]. Primordial and primary follicles receive sufficient nutrients and oxygen via passive 79 diffusion from the ovarian stromal vasculature, however for a follicle to grow beyond the primary stage, 80 formation of an individual capillary network is required. This capillary network is initially very thin, 81 roughly structured, and confined to a single layer in the thecal cells, with the granulosa cell layers 82 remaining avascular throughout folliculogenesis [8,9]. There is a large increase in not only total For immunohistochemistry analyses, large and cystic follicles were categorized based on size and 168 then carefully dissected out from the ovarian tissue. The intact follicle was placed in 10% buffered 169 formalin for 96 hours followed by immersion in 70% ethanol until further tissue processing. Tissues were 170 then de-hydrated following standard procedures, embedded in paraffin wax, and sectioned with a rotary 171 microtome. 172

RNA Extraction 174
RNA extraction was conducted on the large and cystic granulosa cells using the Norgen Biotek 175 Corporation RNA/DNA/Protein Plus Micro Extraction Kit (Norgen Bioteck, Thorold, ON). 300µL of 176 lysis buffer SKP was added to each granulosa cell sample and vortexed for 30 seconds. In order to 177 homogenize the cells further, the lysate was passed 5-10 times through a 25-gauge needle attached 178 syringe until fully homogenized. The lysate was then centrifuged at 8000rpm for 2 minutes in order to 179 pellet cellular debris. Up to 600µL of the lysate supernatant was then transferred into a gDNA 180 Purification Micro Column with collection tube and centrifuged at 8000rpm for 1 minute. Flowthrough 181 was retained and placed on ice for RNA purification. For every 100µL of RNA flowthrough from gDNA 182 extraction, 180µL of 96-100% ethanol was added and mixed by vortexing. Up to 600µL of RNA 183 flowthrough and ethanol was transferred into an RNA/Protein Purification Micro Column with collection 184 tube and centrifuged at 6000rpm for 1 minute. Wash of RNA/Protein purification column was conducted 185 three times by adding 400µL of Wash Solution A to the column and centrifuging at 6000rpm for 1 minute. 186 Following the final wash, the column was spun at 14,000rpm for 2 minutes in order to thoroughly dry 187 the resin. RNA/Protein purification column was then transferred to a 1.7mL elution tube. 50µL of Elution 188 Solution A was added and centrifuged at 2000rpm for 2 minutes followed by 14,000rpm for 1 minute. 189 RNA/Protein column was retained for protein purification procedure. RNA concentration was the 190 quantified using Nanodrop 2000 and stored at -80°C for future use. Following RNA extraction, microRNA cDNA synthesis using reverse transcription was 195 performed using qScript microRNA cDNA synthesis kit (Quanta Biosciences, Beverly, MA). A two-step 196 reaction was performed. In the first step, 2µL of 5x Poly(A) Tailing Buffer, 1µL of Poly(A) Polymerase, 197 and a variable amount of RNA and nuclease free water (up to 7µL and 1µg total RNA) were combined 198 per reaction and incubated in a Bio-Rad T100 Thermal Cycler at 37°C for 60 minutes followed by 70°C 199 for 5 minutes. In the second step, 9µL of microRNA cDNA Reaction Mix and 1µL qScript Reverse 200 Transcriptase was added to the reaction product from step 1 and incubated at 42°C for 20 minutes 201 followed by 85°C for 5 minutes. In addition, a No-Reverse-Transcriptase (NRT) control was conducted, 202 where 1µL of qScript Reverse Transcriptase in the second step was replaced with 1µL of nuclease-free 203 water in order to control for potential genomic DNA contamination. The resulting cDNA product was 204 then stored in -20°C until future use. 205 206 mRNA 207 Following RNA extraction, mRNA cDNA synthesis using reverse transcription was performed 208 using qScript cDNA SuperMix (Quanta Biosciences, Beverly, MA). 4µL of 5x qScript cDNA SuperMix 209 was combined with a variable amount of RNA and RNase/DNase-free water (up to 16µL and 1µg of 210 RNA) per reaction and incubated in a Bio-Rad T100 Thermal Cycler at 25°C for 5 minutes, 42°C for 30 211 minutes, and 85°C for 5 minutes. The resulting cDNA product was then stored in -20°C until future use. 212 213

Reference Gene Selection 214
Candidate reference genes for miRNA qPCR analysis were selected from previous literature on 215 miRNA reference gene analyses [28,29]. A total of 6 candidates were chosen: miR16-5p, miR106a-5p, 216 Let-7a-5p, miR93-5p, miR191-5p, and U6. In addition, candidate reference genes were selected from 217 previous literature on Bos Taurus mRNA reference gene analyses [30]. A total of 8 candidates were 218 chosen: ACTB, GAPDH, UBA52, RPS18, RPL19, HPRT1, H3F3B, and YWHAZ. qPCR was performed 219 as described below and reference gene stability was assessed using three different reference gene 220 software systems: geNorm, Normfinder, and Bestkeeper. included miR15a-5p, miR-18a-5p, miR20a-5p, miR21-5p, miR-29a-3p, miR126-5p, miR132-3p, and 228 Let-7a-5p, with miR16-5p and miR106a-5p used as reference genes (Quanta Biosciences, Beverly, MA). 229 All primer efficiencies were calculated by performing a standard curve using a 2:1 dilution of a pooled 230 reaction sample from 50ng/µL to 0.78ng/µL. Only primer efficiencies between 90-110% were accepted. 231 MiRNA primer sequences and efficiencies can be found in Supplementary Table 1. 232 For miRNA expression analysis, a master mix was created using 5µL of PerfeCTa SYBR Green 233 Supermix (5x), 0.2µL PerfeCTa Universal PCR Primer (10µM), 0.2µL Quanta Biosciences miRNA 234 specific primer (10µM), and 2.6µL of RNase/DNase free water in each reaction well. The master mix 235 was calculated based on the number of wells required for analysis. 8µL of the master mix was then added 236 to the wells followed by 2µL of diluted cDNA (1.5ng/µL per reaction); resulting in a reaction total of 237 10µL per well. Signal detection was acquired using a two-step protocol: 95°C for 2 minutes, followed 238 by 44 repeated cycles of 94°C for 30 seconds and 60°C for 30 seconds, ending with a melt curve 239 acquisition from 65-95°C. The Bio-Rad CFX96 Real-Time PCR system was used for the qPCR analysis. 240 Relative quantity of miRNA targets was log-transformed in BioRad CFX Manager 3.0 and normalized 241 to the relative quantity of the reference genes miR16-5p and miR106a-5p across all samples. 242 243 mRNA 244 mRNA targets of interest included VEGFA, and TSP-1, with RPS18 and RPL19 used as reference 245 genes (Lab Services, University of Guelph). All mRNA primers were either designed using Primer-246 BLAST NCBI or sequences were selected from previous literature. Primer efficiencies were calculated 247 by performing a standard curve using a 2:1 dilution of a pooled reaction sample from 50ng/µL to 248 0.78ng/µL. Only primer efficiencies between 90-110% were accepted. mRNA primer sequences for Bos 249 Taurus can be found in Supplementary Table 2. 250 mRNA expression profiles were performed using the Bio-Rad CFX96 Real-Time PCR system. 251 A master mix was created using 5µL SensiFAST SYBR No-ROX (FroggaBio, Scientiic Solutions, 252 Toronto, ON), 1µL forward primer (1.25µM), 1µL reverse primer (1.25µM), and 2µL RNase/DNase free 253 water. The master mix was calculated based on the number of wells required for analysis. 9µL of the 254 master mix was then added to the wells followed by 1µL of diluted cDNA (5ng/µL per reaction); 255 resulting in a reaction total of 10µL per well. Signal detection was acquired using a two-step protocol: 256 95°C for 30 seconds, followed by 39 repeated cycles of 95°C for 5 seconds and 60°C for 5 seconds, 257 ending with a melt curve acquisition from 65-95°C. Relative quantity of gene targets was log-transformed 258 in BioRad CFX Manager 3.0 and normalized to the relative quantity of the reference genes RPS18 and 259 RPL19 across all samples. Etobicoke, ON). Signals were detected and imaged using a Bio-Rad ChemiDoc XRS+ system and 287 relative densitometry was determined using Bio-Rad Image Lab software.  Table  306 3); however, incubation time was kept constant for both large and cystic follicles. Slides were then 307 counterstained with Carazzi's Hematoxylin, and mounted on glass coverslips. 308 309

Image Analysis 310
Follicle sections stained by immunohistochemistry were imaged by bright-field microscopy using 311 a Nikon Eclipse E600 microscope with a QImaging camera. Follicles were imaged at 200x total 312 magnification and analyzed for cytoplasmic factors VEGFA, VEGFR2, TSP-1, CD36, IGF1, and IGF2. 313 In order to identify the most stable reference genes for miRNA qPCR analyses, candidate 330 reference genes were selected from previous literature [28,29]. A total of six candidate reference genes 331 (U6, let7a-5p, miR16-5p, miR106-5p, miR93-5p, and miR191-5p) were tested across twelve bovine 332 granulosa cell samples. To determine the most steadily expressed miRNAs, three different software 333 programs were used: geNorm, Normfinder, and Bestkeeper. respectively ( Figure 3B). In addition, the pairwise variation (V value) for RPS18 and RPL19 was 0.071 365 ( Figure 3C), well below the minimum value of 0.15. Normfinder determined UBA52 to be the most 366 stable with a stability value of 0.061 ( Figure 3D). Finally, Bestkeeper analysis indicated UBA52 to be 367 the most stable as it had the lowest standard error, highest R value and lowest p-value ( Figure 3F). 368 Ultimately, RPS18 and RPL19 were chosen as the most stable reference genes in bovine granulosa cells 369 as they demonstrated stability across all three software program analyses, a high degree of correlation 370 between M value and stability value ( Figure 3E), and a stable pairwise variation ( Figure 3C). 371 372

MicroRNA Expression in Granulosa Cells of Large and Cystic Bovine Follicles 373
Large and cystic bovine follicles were aspirated to collect follicular fluid and granulosa cells for 374 expression analyses. Expression of eight miRNAs, previously shown to be involved in angiogenesis and 375 folliculogenesis as well as have the ability to target VEGF and TSP-1, were analyzed using qPCR. 376 miRNA expression was normalized to miR16-5p and miR106a-5p. Expression of miR29a-3p was found 377 to be increased in cystic follicles as compared to large (p<0.05; Figure 4E), while expression of miR132-378 3p was significantly reduced in cystic follicles as compared to large (p<0.05; Figure 4G). No other 379 significant changes in miRNA expression was found between large and cystic follicles. Granulosa cells aspirated from large and cystic ovarian follicles were collected and total RNA 383 was extracted for qPCR analysis. VEGFA and TSP-1 expression was normalized to RPS18 and RPL19, 384 as determined through reference gene software analyses. VEGFA was found to be increased in cystic 385 follicles as compared to large follicles (p<0.05; Figure 5C), however no significant difference in TSP-1 386 expression was observed ( Figure 5D). 387 388

Protein Expression in Granulosa and Theca Cells of Large and Cystic Bovine Follicles 389
Western blot analysis was used to evaluate expression of VEGF in granulosa cells collected from 390 follicular fluid aspirates of large and cystic bovine ovarian follicles. Levels of VEGF were found to be 391 significantly increased in cystic follicles as compared to large (p<0.05; Figure 5A, B). 392 Protein expression was also analyzed using immunohistochemistry in order to evaluate 393 expression of additional vascular factors VEGF, VEGFR2, TSP-1, and CD36, as well as mitogenic 394 factors IGF1 and IGF2 in the granulosa and theca cell layers of large and cystic bovine ovarian follicles. 395 The pro-angiogenic factor, VEGF, was expressed at high levels in both the large and cystic follicles, with 396 no statistically significant differences observed ( Figure 6). In contrast, the receptor of VEGF, VEGFR2, 397 was found to be enhanced in cystic follicles as compared to large (p<0.05; Figure 6). Interestingly, the 398 anti-angiogenic factor TSP-1 (p<0.01; Figure 6), along with its receptor, CD36 (p<0.001; Figure 6), were 399 both found to be upregulated in cystic follicles as compared to large follicles. Finally, mitogenic factors 400 IGF1 and IGF2, both involved in cellular proliferation and inhibition of apoptosis, had high levels of 401 expression in both large and cystic follicles, however no significant differences were observed between 402 groups ( Figure 6).  reproductive disorder that causes infertility and significant economic burden to dairy and beef industries. 420 The cause of COD is still unknown, however experts have suggested that it may comprise a variety of 421 factors including environment, genetics, endocrine and angiogenic factors [1]. As cows do not develop 422 cysts with every lactation, or during every ovarian cycle, a change in gene expression has been suggested 423 as a possible mediator and contributor to the disease [3]. As miRNAs are key regulators of gene 424 expression, they may be integral to the gene regulatory mechanisms involved in COD and thereby 425 regulate key angiogenic factors such as VEGF and TSP-1 that are essential for follicle development, 426 blood vessel growth, and ovarian function [11,12]. In addition, this study also aimed to characterize the 427 mRNA and protein expression patterns of several vascular and mitogenic growth factors along with their 428 receptors in cystic follicles [1,31]. Characterizing the expression patterns in cystic follicles can lead to a 429 better understanding of the role that miRNAs, angiogenic factors, and mitogenic factors potentially play 430 in COD. In cystic follicles, our study, as well as others, have demonstrated an increase in VEGF 445 expression, therefore one would hypothesize that miRNAs known to target VEGF would potentially be 446 downregulated, thereby reducing the inhibitory effect on VEGF. Interestingly, this study observed a 447 significant increase in miR-29a along with a significant increase in VEGF expression. This upregulation 448 of miR-29a may be present due to the enhanced VEGF expression in the cystic follicles, thereby acting 449 as a compensatory mechanism in order to try to decrease the proliferation of endothelial cells. In addition, 450 miR-29a may be targeting other factors involved in regulating blood vessel formation or folliculogenesis, 451 such as TSP-2 or IGF1. During follicle growth, the IGF family works in synergy with gonadotropins 452 FSH and LH, in order to regulate proliferation and differentiation of granulosa and theca cells [23]. IGF1 453 has also been shown to initiate follicle growth, suppress granulosa cell apoptosis, and has been suggested 454 as a potential mediator of accelerated preantral follicle growth in PCOS patients [31,37]. This suggests 455 that the enhanced expression of miR-29a could be acting to downregulate IGF1 in order to reduce 456 proliferation and induce apoptosis of granulosa cells in these cystic follicles, although this study does not 457 present mRNA transcript or protein expression levels of IGF1 to further confirm this theory. Overall, the 458 regulatory role that miR-29a has in the context of follicle growth has not been fully elucidated. This study 459 shows that miR-29a is upregulated in cystic follicles, therefore may be playing a role in enhancing the 460 proliferation, decreased estradiol production, and increased VEGF signalling. miR-132 has also been 470 shown to promote angiogenesis, where the use of miR-132 inhibitors results in anti-angiogenic effects 471 [42]. The regulatory effect that miR-132 has within the ovary and specifically in cystic follicles is of 472 great interest. Based on this study and the current literature, miR-132 appears to be a great candidate for 473 future functional studies as it is evident that it plays an important role in follicle and blood vessel 474

development. 475
Previous work in our lab has uncovered an inverse relationship between the expression of VEGF 476 and TSP-1 throughout folliculogenesis. Using bovine follicles ranging in size small (<5mm), medium 477 In order to do this, RNA was extracted from granulosa cells of large and cystic bovine follicles and qPCR 484 was conducted. As hypothesized, VEGF mRNA expression was significantly upregulated in cystic 485 follicles as compared to large, while TSP-1 expression did not differ, although it was expressed at very 486 low levels. In addition to mRNA expression, protein was also collected from granulosa cells of large and 487 cystic follicles and western blots for protein quantification were performed. Similar to mRNA expression, 488 VEGF protein expression was significantly upregulated in cystic follicles. Previous data has shown that 489 In order to assess IGF1 and IGF2 expression, as well as angiogenic factors VEGF, VEGFR2, 511 TSP-1, and CD36, immunohistochemistry (IHC) was conducted on the theca and granulosa cell layers 512 of large and cystic follicles. As mentioned previously, VEGF increases with follicle maturity, as does its 513 receptor VEGFR2 [14], therefore it was hypothesized that these factors would be further increased in 514 cystic follicles. While no significant difference in VEGF expression in granulosa and theca cells was 515 detected, a significant increase in VEGFR2 was observed. These results correlate with previous studies, 516 again suggesting that VEGF and the VEGF signalling pathway plays an essential role in the development 517 of cystic follicles due to enhanced blood vessel development within the thecal cell layer [11,12,14]. 518 519 Interestingly, significant increases in both TSP-1 and CD36 in cystic follicles as compared to 520 large were observed, which contrasts the mRNA expression data showing no change in TSP-1 521 expression. One important difference between the qPCR analyses compared to the IHC analyses is that 522 the former analysed only granulosa cell expression, while the latter analysed both granulosa and theca 523 cells. The thecal cell layer is highly vascularized in antral follicles as well as cystic follicles, while the 524 granulosa cell layer remains avascular [8,9]. As such, the IHC analysis includes the protein expression conditions. As such, two or more reference genes that have been experimentally determined is considered 548 optimal for qPCR analyses [48]. 549 For miRNA and mRNA expression analyses using qPCR, six and eight candidate housekeeping 550 genes were selected to analyze, respectively, on large and cystic bovine granulosa cells. Reference gene 551 stability was assessed using various software systems, geNorm, Normfinder, and Bestkeeper. These 552 software systems use various algorithms to determine which reference gene has the most stable 553 expression across all samples. In addition, some software will assess combinations of reference genes in 554 order to determine which sets of genes, when averaged together, would be optimal for qPCR 555 normalization, thereby adhering to the MIQE guidelines. This analysis determined that miR-16-5p and 556 miR-106a-5p together are the most stable reference genes to use for miRNA expression analysis of our 557 sample set. Interestingly, the most commonly used reference gene for miRNA expression analysis, U6, 558 was shown to be the least stable reference gene, with great variability across this sample set. Additionally, 559 RPS18 and RPL19 were shown to be the most stable mRNA transcripts and therefore chosen as reference 560 genes for mRNA expression analyses. Overall, these results demonstrate the importance of analyzing 561 reference gene stability before proceeding with qPCR analysis, as improper reference gene selection can 562 result in dramatically altered results. 563 564 CONCLUSION 565 In conclusion, this study aimed to characterize the mRNA and protein expression profiles of key 566 angiogenic and mitogenic factors in cystic follicles as well as analyze the expression patterns of select 567 miRNAs thought to be heavily involved in the regulation of ovarian angiogenesis and folliculogenesis. 568 This study demonstrated an increase in miR-29a and decrease in miR-132, along with a significant 569 increase in the mRNA and protein levels of VEGF in granulosa cells, and a significant increase in protein 570 expression levels of VEGFR2, TSP-1, and CD36 in the granulosa and theca cell layers of cystic follicles. 571 These results demonstrate the increased vascular density and pro-angiogenic stimulation of cystic 572 follicles and suggest that miR-29a and miR-132 may be potential mediators of cystic ovarian disease. 573 Together, these results suggest that an increase in pro-angiogenic signalling within the granulosa cells of 574 cystic follicles may not be counteracted with anti-angiogenic signalling or appropriate miRNA 575 regulation, thereby leading to proliferation of vasculature and development of cystic, anovulatory 576

Availability of Data and Material 586
The data generated during the current study are available from the corresponding author on reasonable 587 request. 588 589