Optimization of sperm RNA processing for developmental research

Recent studies have demonstrated the significance of sperm RNA function as a transporter of important information directing the course of life. To determine the message contained in sperm RNA, it is necessary to optimize transcriptomic research tools. The current study was performed to optimize the processing of sperm RNA from sample storage to quantitative real-time PCR and assess the corresponding method with to evaluate male fertility and its representative markers, equatorin (EQTN) and peroxiredoxin (PRDX). Following successive steps of the Minimum Information for Publication of Quantitative Real-Time PCR Experiments guidelines, several options were compared using boar spermatozoa. To evaluate the optimized procedures, the relationship between mRNA expression of EQTN and PRDX in spermatozoa and the fertility (litter size) of 20 individual boars was assessed. Unexpectedly, DNase treatment during RNA isolation had the deleterious effect by decreasing the RNA concentration by 56% and eliminating the correlation between EQTN and PRDX4 mRNA expression and male fertility. Moreover, when sperm RNA was processed using the corresponding method, the results showed the highest exon sequence expression, male fertility prediction power, and consistency. This optimized protocol for predicting male fertility can be used to study the transport of messages directing the life course from spermatozoon to offspring.

. The RT-qPCR data of isolated RNA with DNase from − 80 °C stored after RNAlater treatment was excluded from further statistical analysis because its melting temperature differed in melting curve analysis and showed a decreased amplification peak compared to the other samples ( Fig. 2A). The average Cq value of exon primers between oligo dT and random hexamers showed a significant difference (Fig. 2B). The GAPDH RT-qPCR data showed that the three methods (1. isolated RNA from fresh spermatozoa with DNase. 2. isolated RNA from 4 °C stored spermatozoa without RNAlater and DNase treatment. 3. isolated RNA from− 80 °C stored spermatozoa without RNAlater and DNase treatment.) in oligo dT cDNA and only one method (isolated RNA from 4 °C stored spermatozoa without RNAlater and with DNase treatment) in random hexamer cDNA showed a lower Cq value in the exon primer compared to using both intron and junction primers (Fig. 2C,D). The sizes of the PCR products were 110, 112, and 194 base pairs in the exon, intron, and junction, respectively ( Fig. 2E and Supplementary Fig. S4 online).
correlation between Rt-qpcR data and male fertility. To evaluate the correlation of RT-qPCR data for each genes and male fertility, the relative expression of genes in 20 individual boar spermatozoa were examined using optimized procedures. According to cDNA synthesis and RT-qPCR data for the GAPDH primer, the oligo dT primer in cDNA synthesis and exon primer of the gene in RT-qPCR were adequate for measuring the expression of mRNA. To rule out the effects of genomic DNA (gDNA), intron primers for each marker were also checked with RT-qPCR. Total cDNAs of 20 individual boar spermatozoa were prepared and the mRNA expression of the EQTN and PRDX4 genes was evaluated. EQTN and PRDX4 expression levels were significantly correlated with litter size when RNA was isolated without DNase and when the Cq value from exon primer RT-qPCR was considered as mRNA expression (rEQTN = − 0.341 and rPRDX4 = 0.321; Fig. 3A,B). When the Cq value of the intron primer was considered to rule out the gDNA factor, only EQTN was significantly correlated with litter size (rEQTN = − 0.374 and rPRDX = 0.267; Fig. 3C,D). In contrast to the data obtained from isolated RNAs without DNase, the relative expression of both EQTN and PRDX4 in DNase-treated RNA showed no correlation with litter size (rEQTN = 0.013 and rPRDX = 0.124; Fig. 3E,F). Additionally, the Cq value from the Assessment of male fertility predicting power of mRnA markers. To assess the male fertility predicting power of mRNA markers when processed with optimized protocols, ROC curve analysis was performed. The overall accuracies of EQTN and PRDX4 in non-DNase-treated samples were the same (60%; Fig. 4A,B). When the gDNA factor was considered in EQTN relative expression, all predictive values were decreased (sensitivity, specificity, negative predictive value, positive predictive value, and overall accuracy were 59.52%, 61.11%, 39.29%, 78.13%, and 60.00% before gDNA consideration and 48.48%, 50.00%, 26.09%, 72.73%, and 48.89% after gDNA consideration, respectively; Fig. 4A). Interestingly, although the cDNA templates were from same samples, the coefficient of variation and quartile deviation showed different patterns according to DNase treatment and gDNA factor consideration (Fig. 4C). www.nature.com/scientificreports/

Discussion
Despite the critical contribution of sperm RNA to male fertility 26,41 , embryo development 42 , epigenetic inheritance for acquired traits included in the paternal genome 43 , and health 44 , the lack of optimization of transcriptomic research tools prevents a thorough understanding of sperm biology. Therefore, we optimized the processing of sperm RNA from sample storage to RT-qPCR and evaluated the male fertility predicting power of the optimized protocol.
The significance of the porcine model on biomedical research studying a spectrum of human diseases, including obesity, arthritis, cardiovascular disease, and skin and eye conditions are already well known 45,46 . In addition, Several studies emphasize the particular importance of the porcine model in xenotransplantation 47 . In the same vein, the porcine model can also be useful in the study of human male infertility. What makes this possible is the similarity between porcine genome and the human genome and well-organized breeding data from AI 35,48 . Furthermore, many researh groups are conducting comprehensive trascriptomic analysis in boar spermatozoa [49][50][51] . In the present study, a porcine model was used to evaluate the fertility prediction of sperm RNA.
First, we compared RNAlater treatment to the snap-freezing method in LN 2 for sample storage step. Many recent studies reported that the effect of RNAlater on isolated RNA depends on the cell type, storage period, and research application [52][53][54][55] . To the best of our knowledge, this is the first study to compare the use of RNAlater and snap-freezing for spermatozoa. RNAlater treatment was time-consuming and costly, without beneficially affecting sperm RNA. Thus, treatment with RNAlater was not used in further steps. www.nature.com/scientificreports/ For RNA isolation of spermatozoa, a two-step lysis was used. Spermatozoa chromatin is consists of the enriched disulfide bonds, which can only be broken down by detergents (Sodium dodecyl dulfate or sodium lauryl sulfate) and reducing agents (e.g., dithiothreitol or β-mercaptoethanol). To successfully isolate the large number of sperm RNAs in sperm nuclei 21,22 , the sperm head should be fully lysed. Trizol, which contains neither detergents nor reducing agents, cannot lysis sperm head. For the separation of sperm RNA from a cell lysate, chloroform was used. Chloroform is a common and key reagent used for isolation of sperm RNA, both by Trizol-modified methods but also for commercially available RNA extraction kits during phase separation 5,56 .
Generally, DNase is used to rule out gDNA contamination in isolated RNA. However, DNase has harmful effect not only in gDNA but also in concentrations of RNA and small RNAs 57,58 . In this study, DNase affected RNA concentration and quality in RNAlater-treated samples. Although the DNase alone had no significant effect on RNA quality during the optimization procedure, the exon Cq value of DNase-treated samples was similar with the intron value. Based on these studies, we suggested that DNase may induce the degradation of small RNA in spermatozoa during RNA isolation. This could lead to biased interpretation of sperm transcriptomic study. Therefore, later researchers need to reconsider the use of DNase during sperm RNA study.
For the cDNA synthesis step, the effects of oligo dT and random hexamer primers were compared. The size of the PCR product with the junction primer for GAPDH was 194 base pairs, suggesting that most amplicons were from gDNA. Because of this, the Cq value of the exon primer should be significantly lower than that of the intron primer and junction primer for detecting mRNA. In case of fresh sperm sample, DNase treated RNA showed this corresponding expression pattern. The oligo dT primer cDNA, from 4 to − 80 °C snap-frozen storage and without DNase-treated isolated sperm RNA samples also showed this corresponding expression pattern. Thus, snap-frozen samples in LN 2 , with or without DNase RNA isolation and Oligo dT cDNA synthesis, were selected as the optimized sperm processing conditions for predicting male fertility.
To assess the male fertility predicting power of mRNA markers using the optimized protocols, a correlation test was conducted to evaluate the mRNA expression of EQTN and PRDX4 could represent the male fertility data or not. EQTN and PRDX4 exon expression levels were negatively and positively correlated with litter size in isolated RNA without DNase, respectively. In contrast, no correlation between the exon expression of EQTN and PRDX4 from isolated RNA with DNA was observed. As previously noticed, in the isolated RNA with DNase treatment, the expression of the exon sequence was similar or lower than the expression of the intron sequence. In the other word, the portion of RNA in the sample was too low to indicate the correlation between gene expression and male fertility. Laurell et al. described a method for ruling out the gDNA factor in cDNA samples using an optimized gDNA-specific ValidPrime assay and gDNA reference sample 59 . Similarly, we directly evaluated the expression of gene intron sequences and attempted to exclude gDNA contamination. Only EQTN was correlated with male fertility when the gDNA factor was considered. To clarify the effect of gDNA factor consideration in male fertility prediction value the cut-off value of EQTN relative expression was settled at fixed litter size (13). The decreased male fertility prediction value in EQTN expression was a major limitation when considering the gDNA factor. Moreover, the losing correlation on PRDX4 and decrease in predictive value of EQTN was corresponded with coefficient of variation and quartile deviation.
In summary, we optimized the processing of sperm RNA for screening male fertility from sample storage to RT-qPCR. The optimized condition involved snap-freezing in LN 2 , without DNase RNA isolation, oligo dT cDNA synthesis, and RT-qPCR with exon primers. The highest exon sequence expression, male fertility prediction power, and consistency were obtained using this method. The mRNA level obtained using the optimized method showed better male fertility predicting power and consistency of data. This study will shed light on medical and industrial use of sperm RNA to evaluate infertility, animal health, and offspring phenotypes. Moreover, the optimized protocol can be used to elucidate the underlying mechanisms of life course research from spermatozoa to the next generation.

Materials and methods
All procedures involving animals were approved by the Institutional Animal Care and Use Committee of Chung-Ang University (Approval No. 2017-00018) and performed according to the corresponding guidelines. All methods were performed in accordance with the relevant guidelines and regulations. Sperm sample preparation. Yorkshire boar semen was provided by Sunjin Co. grand-grand parents farm (Danyang, Korea). The sexually mature (11-23 months) and body weight over 90 kg boars were used to collect semen sample. For the laboratory work, the semen samples were collected once in summer season (June-July). The ejaculates of boar were obtained by gloved-hand technique 60 and collected semen were diluted to a density of 30 × 10 6 sperm cells/mL in 100 mL of Beltsville thawing solution 37,61 . During sample transport, semen was maintained at 17 °C in an ice box 61. After arrival in the laboratory, each semen sample was washed with discontinuous 35 and 70% Percoll (Sigma-Aldrich Co, St. Louis, MO, USA) gradient to eliminate non-sperm contents from the semen 62 . Thereafter, sperm pellets were snap-frozen in LN 2 (− 196 °C) or incubated in RNAlater (12 × 10 7 sperm cells/mL; Invitrogen, Carlsbad, CA, USA) overnight in 4 °C. Both samples were stored in 4 °C or− 80 °C for 7 days before RNA isolation. To minimize the effect of individual variation, semen samples from three boars were combined and used for method optimization with GAPDH gene. Moreover, to determine whether the optimized method can be used for accurately predicting fertility phenotype of boar spermatozoa, the mRNA expression level of semen from 20 different boars were evaluated.
computer-assisted sperm analysis. Sperm motility and motion kinematics of Percoll separated spermatozoa was checked by Computer-assisted sperm analysis system (SAIS-Plus ver.10.1; Medical Supply, Seoul, Korea). Percoll separated spermatozoa was resuspended in 1 mL of mTCM199 media and 10 μL of resuspended  Rt-qpcR. The cDNA was mixed with SYBR Green PCR master mix (Applied Biosystems, Foster city, CA, USA) and amplified on a 7,500 fast real-time PCR system (Applied Biosystems). As a positive control and negative control, cDNA from porcine testis RNA and no template and reverse transcriptase cDNA was used, respectively. All primers were designed based on Reference genome Sscrofa11.1 Primary Assembly. For method optimization, we compared the exon, intron, and exon-exon junction primers of GAPDH, a reference gene for RT-qPCR in boar spermatozoa (Table 2) 63 . The EQTN and PRDX4 genes were selected as target markers of male fertility 37,39 . Exon and intron primers of selected markers were designed; and for reference gene, GAPDH was used ( Table 2). The amplification efficiency of primers was tested with calibration curve 64 . Briefly, the cDNA was diluted into 1, 2.5, 5, 10, and 25X (250, 100, 50, 25, and 10 ng) in each well and Cq value of studied genes were analyzed with amount of cDNA ( Supplementary Fig. S5 online). The total reaction volume was 20 µL (100 ng of cDNA). The cycling parameters were 95 °C (10 min) followed by 40  fertility data acquisition. Fertility data from randomly selected 20 Yorkshire boars were gathered by AI using their semen. Semen samples were collected after sexual maturation and diluted as described above. AI was performed by professional technicians from Sunjin Co. twice per estrus 32,65,66 . The inseminated sows were managed at 20 ± 5 °C with air circulation, 2:1 light/dark cycle, and adequate feed for pregnant sow during the experi- www.nature.com/scientificreports/ mental period. AI was performed for 1 year to avoid the seasonal fluctuation and boar semen was inseminated to average 17.25 ± 1.98 sows (total 360 trial). The average number of born piglets from AI results using boar semen was considered as the fertility outcome.

RnA isolation and cDnA synthesis. Sperm
Assessment of male fertility predicting power of mRnA. The correlation between fertility and mRNA expression of EQTN and PRDX4 from 20 individual boar semen sample were analyzed with linear regression in all treatments. To estimate the male fertility predicting power of mRNA markers, receiver operating characteristic (ROC) curve analysis was utilized to set the cut-off value and assess the four key parameters i.e. sensitivity, specificity, positive predictive value, and negative predictive value of each marker 67 . The cut-off values of relative expression in all treatments were decided based on a fixed litter size (13) to compare each treatment. Sensitivity is the percentage of boars that were true-positive when tested for mRNA expression. Specificity is the percentage of boars that tested as true-negative. The positive predictive value is the percentage of boars that tested as positive and had a true-positive litter size. The negative predictive value is the percentage of boars that tested as negative or had a true-negative litter size.