Peroxiredoxin 4 as potential fertility marker in boars

Background: High levels of reactive oxygen species are toxic to spermatozoa as they induce DNA damage, which leads to male infertility. Therefore, reduction of oxidative stress using reducing agents, such as antioxidant enzymes, is required. Peroxiredoxins (Prdxs) are known to play a critical role in the regulation of male fertility as antioxidant enzymes. Although several studies have suggested a close association between Prdxs and male fertility, few studies have explored the ecacy of Prdxs to predict fertility. Therefore, the current study was designed to discover the most sensitive biomarkers among the Prdxs with six isoforms, and furthermore to determine whether Prdxs are more suitable fertility markers for boar spermatozoa compared to conventional semen analysis. Results: The mRNA levels of PRDXs and several sperm parameters were examined in spermatozoa collected from 20 boars with different litter sizes. Our study showed that there was a signicant positive correlation between the litter size and the levels of PRDX 4 among all isoforms in spermatozoa. Subsequently, a regression analysis using a combination of markers was conducted to increase ecacy for fertility prediction. According to the data, PRDX4 had the highest ecacy compared to other combination models. The prediction accuracy of male fertility was further evaluated through receiver operating characteristic curve analysis, which showed that PRDX 4 could predict the litter size with a high overall accuracy of 95%. Conclusions: As fertility biomarkers, genomic markers might be more accurate for predicting male fertility compared to conventional semen analysis. Our ndings indicate that PRDX 4 might be a potential biomarker for diagnosing male fertility and subsequently improving reproductive ecacy in boars.


Background
It is widely known that high levels of reactive oxygen species (ROS) are toxic to spermatozoa owing to DNA damage, which leads to apoptotic-like changes in these cells (1)(2)(3)(4)(5). This affects sperm motility, morphology, and function, all of which affect male fertility. Compared to other cell types, spermatozoa are particularly sensitive to ROS due to their restricted antioxidation ability (1-4, 6, 7). For this reason, substantial research has been focused on antioxidants to suppress oxidative stress in recent years. Dandekar et al. reported that antioxidant enzymes, such as superoxide dismutase and glutathione peroxidase, protect spermatozoa from ROS-induced pathologies (8). Another recent study demonstrated that catalase exerts a protective role to restore sperm quality (9,10). In agreement with these ndings, some studies have reported the differential expression of antioxidant enzymes between fertile and infertile men (11)(12)(13). Consistent with these evidences, our previous proteomic study demonstrated that the sperm proteome changes during capacitation, which is one of the fundamental processes that spermatozoa undergo to acquire the ability to fertilize an egg (14). Furthermore, this analysis revealed that peroxiredoxin (Prdx) 5 is highly expressed in capacitated spermatozoa. Previously, Prdxs have been reported to be vital antioxidant enzymes in spermatozoa and one of the most common antioxidant enzymes that counter reactive oxygen species (ROS) in almost all cell types (2,(14)(15)(16)(17). Similarly, another proteomic study demonstrated that the levels of Prdxs are reduced in spermatozoa of infertile male patients (13). In addition, an extensive literature research revealed that the abnormal function of Prdxs affects sperm motility, capacitation, mitochondrial membrane potential, and DNA integrity (2,(14)(15)(16)(17).
Thus, the levels or activities of Prdxs in spermatozoa may serve as a potential target for diagnosing male fertility. While a large body of knowledge supports the close association between Prdxs and male fertility, the likelihood of Prdxs to predict fertility is still lacking, particularly in a clinical setting. In addition, investigation of particular isotypes of PRDX that could speci cally de ne fertility levels is also of great importance.
Many limitations are encountered when conducting research on humans to diagnose and predict diseases, such as infertility, due to various reasons (i.e., a variety of etiologies, unexplained etiologies, limited number of reproductive trials, sample size, ethical issues, etc.). Therefore, many studies have been conducted on swine samples due to their advantages as a biomedical model. Various genetic traits related to fertility have been well established in swine, which is thus a very effective biomedical model (18)(19)(20)(21). Moreover, the fertilizing ability of a human male can only be evaluated with one woman, whereas pigs can be used as a better fertility model as arti cial insemination (AI) allows one to obtain a variety of fertility outcomes by breeding with many females with different genetic traits.
Therefore, the goals of our study were ) to better elucidate the role of each PRDX in male fertility, for which, the expression of PRDXs was measured and compared with clinical fertility data for pig, ) to elucidate the e cacy of PRDXs as biomarkers for predicting the male fertility, and iii) to develop the most useful biomarker models using a combination of PRDX isoforms and conventional methods, as alternatives to independent approaches for predicting male fertility.

Animal and semen collection
Yorkshire boar (range = 13 to 25 months) semen was collected using the gloved-hand technique twice per week. Collected semen was transferred onto ice from the Grand-Grandparents Farm (Sunjin Co., Danyang, Korea) to the laboratory within 2 h(22). This experiment was performed in two stages. ) Semen samples were collected from 10 boars at the Grand-Grandparents Farm (Sunjin Co., Danyang, Korea). Depending on boar litter sizes, samples were categorized into high (n = 5, average = 13.62 ± 0.25) and low-litter size groups (n = 5, average = 11.34 ± 0.32). Each semen sample was centrifuged at 500 ×g for 20 min using a  (14,23). To evaluate sperm functional parameters between the two groups, sperm motility, motion kinematics, sperm capacitation status, and mRNA and protein levels of Prdx4 were evaluated. ) Unlike the rst trial, we randomly collected 20 individual boar semen samples of unknown fertility. Correlation analysis and quality assessment were performed as described for the previous stage.
Combined H33258/chlortetracycline uorescence (H33258/CTC) assessment H33258/CTC assessment was conducted to evaluate the capacitation/acrosome status of spermatozoa (2,23). Brie y, 135 μL of each sample was incubated with 15 μL of H33258 solution for 10 min at room temperature. Then, 250 μL of 2% polyvinylpyrrolidone in Dulbecco's phosphate-buffered saline (DPBS) was added and centrifuged at 100 ×g for 2.5 min to remove the surplus dye. After that, the cell pellet was resuspended in 100 μL of DPBS and 100 μL of CTC solution. A microphot-FXA microscope was used to observe the capacitation status of spermatozoa using ultraviolet BP 340-380/LP 425 and BP 450-490/LP 515 excitation/emission lters for H33258 and CTC, respectively. The capacitation status was classi ed into non-capacitated (F pattern), capacitated (B pattern), and acrosome-reacted (AR pattern), as reported previously (2,27). At least 400 spermatozoa were counted per slide.
RNA extraction, cDNA synthesis, and quantitative real-time PCR (qRT-PCR) PureLink TM RNA Mini Kit (Invitrogen, Carlsbad, CA, USA) and TRIzol (Invitrogen) were used for RNA extraction from each spermatozoa group. Brie y, the prepared sperm pellets were lysed using the lysis buffer with 2% 2-mercaptoethanol (Sigma-Aldrich). Next, 200 μL of chloroform (Sigma-Aldrich) was added to samples and incubated at 20 °C for 5 min. Samples were then centrifuged at 12,000 ×g for 25 min. After centrifugation, 500 μL of the upper layer was moved into a new RNase-free tube along with an equal volume of 100% ethanol. The mixture was then transferred into a spin cartridge (Invitrogen) and centrifuged at 12,000 ×g for 15 s. Finally, to isolate the RNA, 20 μL of nuclease-free water was added to the spin cartridge. RNA concentration and 260/280 ratio were evaluated using a NanoDrop spectrophotometer (NanoDrop Technologies, Wilmington, USA). cDNA synthesis was performed using High-Capacity cDNA Reverse Transcription Kits (Applied Biosystems, Foster City, CA, USA) following the manufacturer's instructions. Finally, PrimerSelect TM software (DNASTAR, Madison, WI, USA) was used to design primers for PRDX 1-6 and GAPDH (Supplementary table. 1). All primers were designed according to the Reference genome Sscrofa11.1 Primary Assembly. qRT-PCR was performed, and the results were analyzed using ABI PRISM 7500 (Applied Biosystems) and the 2 −ΔΔCt method (28).
Western blot analysis of total Prdx 4 and thiol-oxidized Prdx 4 Western blot analysis of total Prdx 4 and thiol-oxidized Prdx 4 in bovine spermatozoa was performed. Brie y, each sample was washed three times with DPBS and centrifuged at 10,000 × g for 10 min. The supernatant was removed, and sperm pellets were resuspended in Laemmli sample buffer (63 mm Tris, 10% glycerol, 10% sodium dodecyl sulphate, 5% bromophenol blue) with (reducing condition) or without (non-reducing condition) 100 mM DTT and incubated at room temperature for 10 min. After incubation, samples were centrifuged at 10,000 × g for 10 min, and cell pellets were boiled at 100°C for 3 min. Samples were resolved by SDS-PAGE using a 12% mini-gel system (Amersham, Piscataway, NJ, USA), and separated proteins were transferred to a polyvinylidene uoride membrane (Amersham). The membrane was blocked for 1 h at room temperature with blocking agent (3%; Amersham). To detect Prdx 4 under both reducing and non-reducing conditions, the membrane was incubated overnight with a rabbit polyclonal anti-peroxiredoxin 4 antibody (1: 2,000; Abcam) overnight at 4°C (16). Membranes were then incubated with horseradish peroxidase (HRP)-conjugated goat anti-rabbit IgG (Abcam) diluted 1:5000 for 1 h at room temperature. α-tubulin was used as an internal control (detected with a mouse monoclonal anti-α-tubulin antibody, 1:10000; Abcam) for 2 h at room temperature. Membranes were washed three times with PBS-T, and protein-antibody complexes were visualized using enhanced chemiluminescence.

Arti cial insemination (AI)
We randomly selected 20 Yorkshire boars with unknown fertility from the Grand-Grandparents Farm to evaluate the correlation between litter size and sperm parameters or PRDX expression levels in individual samples. Selected semen was diluted to 30 × 10 6 sperm cells/100 mL with Beltsville Thawing Solution for AI (22,23). Diluted boar semen samples were inseminated by an experienced arti cial inseminator into sows (n = 568).

Quality assessment of markers
To determine the accuracy of predicting boar fertility based on the litter size, we evaluated the following ve major parameters during the screening test: sensitivity, speci city, positive predictive value (PPV), negative predictive value (NPV), and overall accuracy, which were determined according to the cut-off value associated with receiver operating curve (ROC) analysis as described previously (23,27,29,30).
Brie y, cut-off value was decided by the predicted probability that the individual semen appertained to [litter size ≥ 12] according to the marker(s) in logistic regression. Sensitivity, which is the true positive results, is calculated as the number of boars with predicted probability ≥0.5 to the number of boars with litter size ≥12 piglets. Speci city, which is the true negative results, is calculated as the proportion of the number of boars with predicted probability <0.5 to the number of boars with a litter size <12 piglets. The positive predictive value was decided as the proportion of the actual number of boars that had a litter size ≥12 piglets to the number of boars with a predicted probability ≥0.5. The negative predictive value was decided as the proportion of the actual number of boars that had a litter size <12 piglets to the number of boars with a predicted probability <0.5.

Marker combination and screening test
To evaluate multiple gene combinations, backward multiple regression analysis was performed. Backward multiple regression analysis conducted with a full model (all isoforms of PRDXs, HYP, and protein levels of Prdx 4) and at every single stage gradually removes variables from regression model to discover a reduced model that principal explains the data. We then calculated the combined biomarker score, depending on the following formulae:

Statistical analysis
The data were analyzed using SPSS (v. 25.0; Chicago, IL). Pearson correlation coe cients were evaluated to determine the correlation between motility, motion kinematics, capacitation status, litter size, and PRDX expression levels. Backward multiple linear regression was used to identify the optimal combination conditions. The combined scores were then calculated according to the regression formulae described above. Receiver operating characteristic (ROC) curve analysis was performed to evaluate the ability of individual parameters for predicting the litter size. cut-off value was decided by the predicted probability that the individual semen appertained to [litter size ≥ 12] according to the marker(s) in logistic regression. (23,(30)(31)(32). To predict the litter size using ROC curves, data were analyzed by Student's two-tailed t-test.
Differences between high and low-litter size groups were considered signi cant at p values less than 0.05. Data are presented as mean ± SEM.

Results
Correlation of conventional semen parameters, sperm capacitation status, and expression of PRDXs with selected high-and low boar litter size To nd a sensitive marker for litter-size prediction, conventional semen analysis (including sperm motility, motion kinematics, and capacitation status analyses) and analysis of the mRNA levels of PRDXs were performed and results were compared between the high-(n=5) and low-litter size (n=5) groups. Our results showed that among PRDXs, only PRDX 4 had signi cantly higher expression in the high-litter size group (Table 1, Figure 1B, p < 0.05). However, no signi cant association was found for other parameters between the two groups ( Table 1).
Correlation of conventional semen parameters, sperm capacitation status, and PRDX expression levels with 20 boars with unknown litter sizes Next, Pearson correlation analysis was conducted to demonstrate the correlation between different parameters and individual boars (n=20) with different litter sizes for increasing the accuracy of prediction. Only the mRNA levels of PRDX 4 positively correlated with the litter size (R = 0.603, p < 0.05, Table 2). Although a signi cant correlation was observed between PRDXs, there was no correlation between the levels of PRDXs and the litter size, except for the level of PRDX 4 (p < 0.05, Table 2). In addition, when sperm parameters were compared to the litter size, only HYP spermatozoa showed a signi cantly positive correlation with the litter size (R = 0.506, p < 0.05) ( Table 2). Simultaneously, we performed western blotting under reducing (total Prdx 4) and non-reducing (thiol-oxidized Prdx 4) conditions to analyze the thiol-oxidation ratio of Prdx 4. However, we did not see any band for the nonreducing condition (thiol-oxidized Prdx4). This is seen as Prdx4 from all samples had already been oxidized by excessive levels of ROS, as shown in a previously published study (16). Therefore, we analyzed the correlation between Prdx4 protein levels and individual boar samples (n=20) with different litter-sizes under reducing condition. Interestingly, the protein levels of Prdx 4 showed a signi cant negative correlation with the litter size, in contrast with the mRNA levels of PRDX 4 ( Figure 2, Table 2, R = -0.539, p < 0.0 Quality estimation of single/multiple markers for predicting litter size For the screening test, we analyzed the following ve key parameters: sensitivity, speci city, NPV, PPV, and overall accuracy. Before performing the quality assessment, we set the cut-off values for mRNA and protein levels of Prdx 4 and HYP as 0.71, 2.2939, and 11.82, respectively, based on the ROC curves (Table  4). Subsequently, to predict the average litter size, we used the cut-off values of mRNA and protein levels of PRDX 4 and HYP. Figure 1D shows that the ability of PRDX 4 mRNA levels to predict high-litter size facilitated increasing the average litter size signi cantly (from 11.38 to 12.93) (p < 0.05). In addition, selecting boars with predicted high litter size based on HYP marker signi cantly increased the average litter size (from 11.94 to 12.87) ( Figure 1C, p < 0.01). In contrast, the ability of Prdx 4 protein levels to predict low-litter size facilitated increasing the average litter size signi cantly (from 12.09 to 13).
Even though there was no correlation between the litter size and the levels of PRDX isoforms, except for the level of PRDX 4, we tried to discover markers that are more sensitive by combining PRDX isoforms with the protein levels of Prdx 4 to increase diagnostic accuracy. Backward multiple linear regression was performed using the, HYP level, mRNA expression levels of all PRDXs, and protein levels of Prdx 4; six types of models were found to be signi cantly related to the litter size (p < 0.05, Table 4). Subsequently, quality estimations (sensitivity, speci city, NPV, PPV, and overall accuracy) were conducted. Although the six types of models have much higher diagnostic accuracy compared to HYP and Prdx 4 protein levels, none of the combination models had higher diagnostic accuracy than the mRNA levels of PRDX 4 ( Table  4).

Discussion
Male infertility in both humans and animals is a major global problem that can lead to huge economic losses (29,33). Many studies have been conducted to overcome male infertility, and 30-80% of infertile men exhibit spermatozoa damaged by ROS, which might be one of the cause of male infertility (1,34,35). Thus, to increase the accuracy of predicting male fertility, several studies have been conducted to discover an e cient antioxidant enzyme as a fertility-related biomarker (29,(36)(37)(38). The use of such biomarkers has greatly improved the diagnostic accuracy of male fertility. Therefore, we investigated whether the expression levels of PRDXs could be utilized commercially along with conventional semen analysis for predicting male fertility.
First, we investigated whether the litter size can be determined by conventional semen analysis. Our results showed that there was no signi cant difference between the high-(average litter size = 13.62 ± 0.25) and low-litter size group (average litter size = 11.34 ± 0.32) in terms of conventional semen parameters, such as sperm motility, motion kinematics, and capacitation status. These results corroborated those of Kwon et al. and further supported the limitation of conventional semen analysis for diagnostic accuracy (29). We then investigated the expression levels of PRDX as a potential biomarker for diagnosing male fertility in both high-and low-litter size groups.
Despite the fact that various isoforms of Prdx seem to affect spermatozoa, our results showed that out of the six isoforms, only PRDX 4 is highly expressed in the high-litter size group. Based on this nding, we speculated that each isoform should be examined individually in order to identify the potential of different isoforms as biomarkers for a given species. Simultaneously, to gain further insight regarding correlation between Prdx 4 protein levels and the litter size, western blotting was conducted.
Correlation analysis was performed for each parameter, using the clinical fertility data of 20 randomly selected Yorkshire boars. Corroborating with the above result, among the six isoforms of PRDXs, only PRDX 4 exhibited a signi cant positive correlation with litter size (r = 0.603). In contrast to PRDX 4 mRNA levels, total Prdx 4 protein levels negatively correlated with the litter size (r = -0.539). This inverse pattern between mRNA and protein levels of PRDX 4 might be explained by several factors, including regulation of translation and protein stability (46)(47)(48). Although di cult to explain, understanding the basis of this inverse correlation will be critical to explaining the relevance of PRDX 4 transcript and protein markers. Our proteomic results corroborated those of Gong et al., who found that oxidized Prdx 1 and 6 are highly expressed in infertile men (13). Although oxidized bands could not be quanti ed, this could be explained to be a result of increased Prdx 4 di-sul de bond formation due to high ROS levels as were found in O'Flaherty's study (16). This consensus in results might be due to the highly quali ed redox ability of Prdx 4. Previous studies had reported that after losing its redox activity, Prdx 4 disrupts sperm physiology and function, resulting in cell death(49, 50). Although the biochemical signi cance of Prdx 4 in male fertility is not completely elucidated, it is tempting to speculate that Prdx 4 plays a considerable role in sperm quality and male fertility in swine compared to those in other mammals.
In contrast to several studies that found that inactive Prdxs led to loss of sperm motility and motion kinematics(2, 15), we found that Prdxs did not affect sperm motility and other motion parameters. This might suggest that the correlation between Prdxs and litter size is directly related through mechanisms other than reduction of sperm motility. Although there is no signi cant difference in HYP between highand low-litter size groups, we observed that HYP also had a signi cant positive correlation with the litter size (r = 0.506). Evaluation of HYP is considered a powerful tool in conventional semen analysis. To penetrate the zona pellucida of oocytes and execute successful fertilization, spermatozoa must acquire a special movement called HYP(51). Even though the Pearson correlation coe cient of HYP is relatively lower than that of Prdx 4, HYP might still be a representative biomarker among conventional semen parameters. Although HYP is a phenomenon that occurs after capacitation in the uterus, it is di cult to conclude whether it is pre-capacitated, as our results showed no alteration of capacitation status and acrosome reaction status. However, this is also a very interesting result and exploration of its implications requires further research.
Next, we performed quality assessment to determine the accuracy of these biomarkers using ROC curves. Our results revealed that the mRNA levels of PRDX 4 are a superior male fertility biomarker compared to HYP and PRDX4 protein levels. In addition, multiple linear regression analysis was performed with HYP, protein levels of Prdx 4, and the mRNA expression levels of all isoforms of PRDXs to increase the diagnostic accuracy of male fertility. However, none of the combination models showed a higher diagnostic accuracy than the mRNA levels of PRDX 4. Eventually, we evaluated the accuracy of this biomarker to predict litter size of eld fertility data from 20 randomly selected pigs. As a biomarker to predict or diagnose the male fertility, the mRNA levels of PRDX 4 were associated with a tremendous increase in the litter size (1.55 litter size/insemination). However, HYP and Prdx 4 protein expression were only associated with a slight increase in the litter size (0.93 and 0.91 litter size/insemination).

Conclusions
To the best of our knowledge, this is the rst study to comprehensively elucidate the role of PRDX 4 in male fertility. Our study demonstrated that PRDX 4 plays the most signi cant role in male fertility and is the most sensitive biomarker for the prediction of male fertility in swine. Therefore, we speculated that PRDX 4 has a good potential as a potential biomarker to increase litter size in swine. However, our research may have two limitations. i) Although several studies demonstrated that Prdxs are closely related with male fertility (15,44,52,53), our study showed that other isotypes of PRDX did not exhibit any correlation with male fertility. For other PRDX isotypes, the correlation with a wider range of litter sizes may have been signi cant; however, no signi cant correlation was observed for the narrow range of litter size (10.3-14.2) used in the current study. ii) Even though each isoform of Prdxs is functional and can be interactive, we did not analyze all isoforms of Prdx at the protein level. Recent studies reported that lack of Prdx 6 in mice affected male fertility, reduced the e cacy of in vitro fertilization, and blocked development of early embryos before the blastocyst stage. Therefore, further studies should be conducted using large-scale eld trials to identify whether the protein levels of other isoforms of Prdx are associated with the litter size, in order to identify more sensitive and accurate markers to predict male fertility. Availability of data and materials: The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations
Ethics declarations: All procedures were conducted according to the ethical guidelines for treatment of animals and were approved by the Institutional Animal Care and Use Committee (IACUC) of Chung-Ang University, Seoul, Republic of Korea (approval number: 2017-00018).
Consent for publication: Not applicable.
Competing interests: The authors declare that they have no competing interests.
48. Moritz CP, Muhlhaus T, Tenzer S, Schulenborg T, Friauf E. Poor transcript-protein correlation in the brain: negatively correlating gene products reveal neuronal polarity as a potential cause. J      All cut-off value was decided by the predicted probability that the individual semen