Accuracy of human sperm DNA oxidation quantification and threshold determination using an 8-OHdG immuno-detection assay

Abstract

STUDY QUESTION

Can a discriminant threshold be determined for human sperm DNA oxidation?

SUMMARY ANSWER

A discriminant threshold was found with 65.8% of 8-hydroxy-2’-deoxyguanosine (8-OHdG)-positive sperm cells and a mean intensity of fluorescence (MIF) of 552 arbitrary units.

WHAT IS KNOWN ALREADY

Oxidative stress is known to interfere with sperm quality and fertilizing capacity. However, current practice does not include the routine determination of oxidative DNA damage in spermatozoa; optimized consensus protocols are lacking and no thresholds of normality have been established.

STUDY DESIGN, SIZE, DURATION

Intra- and inter-method comparisons between four protocols (I–IV) were conducted to determine the most relevant and efficient means of assessing human sperm 8-OHdG content. Tests of assay repeatability, specificity, sensitivity and stability were performed to validate an optimized methodology for routine diagnostic use.

PARTICIPANTS/MATERIALS, SETTING, METHODS

This prospective study compared three immuno-detection methods including immunocytochemistry, fluorescence microscopy and flow cytometry. Sperm DNA oxidation for 80 patients was determined relative to semen parameters and clinical conditions, using the selected immuno-detection protocol in comparison with a commercial kit. These patients (age 35 ± 1 years: mean ± SEM) presented with normozoospermic (n = 40) or altered parameters (necro- or/and astheno- or/and teratozoospermia or/and leukocytospermia).

MAIN RESULTS AND THE ROLE OF CHANCE

Significant positive Pearson and Spearman correlations were determined for 8-OHdG values and sperm parameters using protocol III. A notable high and positive correlation was revealed for MIF with BMI and leukocyte concentration. Protocol III was the most discriminating method regarding assay repeatability, specificity, sensitivity, stability and reliability for sperm parameter alterations, in particular leukocytospermia according to parametric or non-parametric tests, effect-size determinations and factorial analysis such as principal component analysis and factor discriminant analysis. Of interest is that 39% of the subjects with ‘pathological’ sperm DNA oxidation values were normozoospermic.

LIMITATIONS, REASONS FOR CAUTION

The oligozoospermic population was not evaluated in this study because insufficient material was available to carry out the comparisons. However, spermatozoa concentration was taken into account in the statistical analysis.

WIDER IMPLICATIONS OF THE FINDINGS

Our study is the first validation of a protocol to determine a discriminant threshold for human sperm DNA oxidation. The protocol’s detection accuracy for 8-OHdG human sperm DNA residues, stability over time, and relationship to human sperm quality were demonstrated. The assay should find application in the diagnosis of male factor infertility associated with oxidative stress.

STUDY FUNDING/COMPETING INTEREST(S)

This work was funded by institutional grants from the CNRS, INSERM and Université Clermont Auvergne (to J.R.D.) and by Clermont-Ferrand Hospital-CECOS research funds (to L.J. and F.B.). P.G., A.M., R.J.A. and J.D. are, respectively, CEO, scientific director and scientific advisors of a US-based biotech company (Celloxess, Princeton, NJ, USA) involved in preventative medicine with a focus on the generation of antioxidant oral supplements.

8-OHdG (8-hydroxy-2’-deoxyguanosine), human spermatozoa, DNA oxidation, pathological threshold, percentage of 8-OHdG-positive spermatozoa, fluorescence, leukocyte concentration, stability, repeatability

Introduction

Oxidative stress (OS) is the major contributing cause of damage to sperm DNA (Aitken, 1994; Lewis et al., 1995; Fisher and Aitken, 1997; Irvine et al., 2000; Aitken et al., 2012, 2014). An early marker of DNA oxidation is the formation of base adducts such as 8-hydroxy-2’-deoxyguanosine (8-OHdG) (De Iuliis et al., 2009). DNA oxidation triggers activation of the base excision repair pathway (BER), which eliminates oxidized residues, such as 8-OHdG, via the 8-oxoguanine glycosylase 1 (OGG1) enzyme, and inserts a new nucleotide on free 3’-OH ends using the APE1 protein (Smith et al., 2013). APE1 is not present in human spermatozoa, making it impossible to complete the BER pathway, which remains halted at the generation of an abasic site (Hoeijmakers, 2001; Smith et al., 2013). However, the oocyte does possess APE1 and XRCC1, permitting continuation of the BER pathway and effective DNA repair following fertilization (Lord and Aitken, 2015). Since the oocyte is deficient in OGG1 (Wood et al., 1992; Lord and Aitken, 2015), non-resolution of sperm DNA 8-OHdG adducts means they will persist in the zygote and create the opportunity for mutations to occur (typically transversion mutations) prior to the initiation of embryonic development (Wood et al., 1992; Lord and Aitken, 2015). Consistent with this situation, it has been reported that high sperm DNA oxidation can impair the oocyte’s ability to activate a normal developmental program (Morado et al., 2013; Park et al., 2015). In this context, assessment of sperm DNA oxidation should be of interest to manage male factor infertility associated with OS.

A limited number of reports have shown that 8-OHdG levels are negatively correlated with male fecundity (Fraga et al., 1991; Loft et al., 2003; Lewis and Aitken, 2005) and sperm quality (Cambi et al., 2013). However, others have found no correlation between oxidative DNA alterations and semen parameters (Thomson et al., 2009; Montjean et al., 2010; Zribi et al., 2011). These discrepancies underpin the lack of consensus regarding the predictive utility of such evaluations and stem from differences in several parameters including: the type of assays used to evaluate DNA oxidation, operator bias, variable patient selection criteria and the sperm isolation protocols used. The literature reveals several methods to detect sperm DNA oxidation, particularly the detection of 8-OHdG residues. Recently, the specificity and reliability of commercially available kits for the detection of human sperm DNA 8-OHdG residues have been questioned (Cambi et al., 2013) revealing the lack of a proper and reliable tool to evaluate sperm DNA oxidation. The aims of this study were to validate a standardized protocol using specific 8-OHdG immuno-detection, compare its accuracy with the commercial OxyDNA Test^® and examine its output in relation to various parameters of sperm quality. The assay we have developed has allowed us to propose a discriminant threshold of human sperm DNA oxidation that should find application in the diagnosis of male factor infertility associated with OS.

Material and Methods

Patient samples and ethics statement

Samples were obtained from men undergoing routine semen analysis at the Center for Reproductive Medicine, Clermont-Ferrand, France between May 2014 and November 2017. Patients were recruited in accordance with the 1975 Declaration of Helsinki principles on human experimentation. Only surplus semen was used in this study. Written informed consent was obtained from each man prior to inclusion of any sperm sample in the ‘Germetheque’ biobank. This study was approved by the ‘Germetheque’ Scientific Committee and the Ethical Committee under the French Institute Review Board reference CPP Sud-Est 6, DC2008-DC58. Patient clinical data, notably fecundity (having had a child naturally, n = 22), were compiled during medical counseling. Semen samples were collected and conventional semen analysis was performed in accordance with the World Health Organization 2010 guidelines (WHO, 2010). Leukocytes were detected using the LeucoScreen Kit (FertiPro N.V., Belgium).

Chemicals

All chemicals were purchased from Sigma-Aldrich (St-Quentin-Fallavier, France) with the exception of monoclonal anti-8-OHdG antibody (mouse anti-8-OHdG monoclonal antibody DNA/RNA Damage Antibody-15A3, NB110-96878, Novus Biological^®, France), polyclonal anti-mouse IgG-coupled horse-radish peroxidase [HRP] (P.A.R.I.S Anticorps, France), Alexa Fluor 488 Goat anti-mouse IgG secondary antibody (Molecular Probes^®, Eugene, OR, USA), OxyDNA Test^® (EKF Diagnostics, Biotrin International, Dublin, Ireland), Vector Nova Red substrate kit (Vector Laboratories, AbCys^®, Paris, France), Super Frost^® slides (Thermo Fisher Scientific, Villebon sur Yvette, France), exogeneous 8-OHdG and 3-deoxyguanosine (3-dG, Biolog Life Science Institute^®, Bremen, Germany) and H₂O₂ (Gilbert, Herouville-Saint-Clair, France).

Experimental design

The experimental design is described in Fig. 1 and Supplementary Fig. S1. In the first set of experiments, we carried out an intra-methods comparison between three methods of detection (pairwise) using the same anti-8-OHdG primary antibody and light microscopy (LM, protocol I), fluorescence microscopy (FM, protocol II) and flow cytometry (FCM, protocol III). In the second set of experiments, we assessed the reliability of values obtained with protocol III and those obtained with the OxyDNA Test^® (protocol IV). Using FCM for both protocols III and IV, the percentage of 8-OHdG-positive spermatozoa and the mean intensity level of fluorescence (MIF) were evaluated.

Figure 1

Experimental design of the study to assess the accuracy of human sperm DNA oxidation quantification and threshold determination using an 8-OHdG immune-detection assay. 8-OHdG: 8-hydroxy-2’-deoxyguanosine

8-OHdG detection

For all four conditions, PBS-washed samples were incubated for 10 min in the dark in a decondensation buffer comprising 2 mM dithiothreitol (DTT), 0.5% Triton X-100 and 1× PBS, as previously described (Noblanc et al., 2012), followed by fixation with 4% paraformaldehyde. For protocols I, II and III, spermatozoa were incubated in 1.5% normal goat serum saturation solution. Incubations with anti-8-OHdG monoclonal antibody (1:1000) and secondary (1:500 and 1:1000 for IgG-coupled [HRP] and Alexa Fluor 488, respectively) antibodies were conducted overnight and for 90 min, respectively. For protocol I, peroxidase signal detection was achieved using the Vector Nova Red substrate kit. For both protocols I and II each sample was spread on two Super Frost^® slides and the slides were read ‘blind’ by two different observers. Spermatozoa were considered as 8-OHdG-positive when a spot of peroxidase (pink–brown color) or green fluorescence was observed (Supplementary Fig. S1). In protocol IV, the OxyDNA Test^® was used per the manufacturer’s indications. For each assay, control incubations were conducted comprising a negative control untreated with anti-8-OHdG primary antibody or the binding conjugate, and a positive control treated with an 8 M H₂O₂ solution for 1 h before fixation. The specificity and sensitivity of protocols III and IV were controlled (n = 3) by pre-incubating anti-8-OHdG antibody or FITC-conjugate with increasing concentrations of exogenous 8-OHdG or 3-dG (negative control) for 3 h. To measure signal stability as a function of time, spermatozoa underwent fixation and were then stored in a 0.01% sodium azide/PBS solution for 0, 15 and 30 days before final labeling. An analysis was then performed to compare any changes in oxidized sperm percentage and MIF. Tests of intra- and inter-assay repeatability were also performed to determine coefficient of variation (CV) using five sperm samples assayed in triplicate (intra-assay CV) on three different days (inter-assay CV).

Analyses of Protocols III and IV were performed on a BD FACS Aria SORP (BD Biosciences, Franklin Lakes, NJ, USA) cell sorter using an 85 μm nozzle. The forward-scattered light and side-scattered light were detected on a linear scale and used to gate sperm cells. Sperm cell gating was confirmed by microscopy after cell sorting. Cytometry detection using protocol IV differentiated two subpopulations (Supplementary Fig. S1). Throughout this work, we have considered both the total fluorescent sperm population termed ‘IV population’ and the subpopulation with the highest MIF termed ‘HF IV’ (HF for highly fluorescent).

Statistical analysis

All analyses were performed using GraphPad PRISM^® 5 (La Jolla, CA, USA) and STATA^® (version 13, StataCorp, College Station, TX, USA) software. Data were presented as the mean ± SEM or median [interquartile range]. Concordance and correlations tests were studied using Bland–Altman test and Lin’s concordance coefficient [kLIN] (Bland and Altman, 1986) and, Pearson or Spearman (rho coefficient) correlations, respectively. For each pairwise comparison, the statistical difference was evaluated using parametric (Student’s t-test) or non-parametric (Wilcoxon test) paired tests and effect-size (ES) values (Cohen, 1998). Receiver operating characteristic (ROC) curve analyses were performed using AUC and 95% CI. Then, the optimal threshold to predict a pathological state was determined according to biological and clinical relevance and according to usual indexes recommended in the literature These usual inferential analyses were followed by factorial analysis: principal component analysis (PCA) used to study relationships between quantitative parameters, and factor discriminant analysis (FDA). Statistical analysis details are described in Supplementary Data.

Results

Intra-methods and inter-methods comparisons

Protocols I (LM) and II (FM) were moderately, but significantly concordant with a Bland–Altman bias of measure of 4.7 [−36.51 to 45.84] (Table I). Protocols II and III (FCM) were both significantly and substantially positively correlated (P = 0.004) and concordant (κ_LIN = 0.66, P < 0.001) with a Bland–Altman bias of 2.4 [−26.66–31.34] which was highly reliable giving comparable values for the proportion of 8-OHdG-positive spermatozoa, for both subjects and positive controls.

Table I

Intra-method validation using the same primary anti- 8-OHdG antibody (Protocols I vs. II, II vs. III), and inter-methods validation using anti-8-OHdG immunodetection (Protocol III) vs. the OxyDNA Test® (Protocol IV)

	Protocol I vs. Protocol II		Protocol II vs. Protocol III		Protocol III vs. Protocol IV
n	10		15		80
	8-OHdG spz (%)		8-OHdG spz (%)		8-OHdG spz (%)		MIF (a.u.)
Positive controls Mean ± SEM	97.0 ± 3.0	93.8 ± 2.1	72.9 ± 5.0	79.1 ± 4.6	83.5 ± 1.7	90.9 ± 1.1***	1986 ± 277	2867 ± 121**
[CI]	[90.2–103.8]	[89.1–98.5]	[62.1–83.7]	[69.2–89]	[80.1–87.0]	[88.7–93.2]	[1434–2538]	[2625–3109]
	8-OHdG spz (%)		8-OHdG spz (%)		8-OHdG spz (%)		MIF (a.u.)
Subjects Mean ± SEM	70.7 ± 8.2	66.0 ± 6.3	52.8 ± 4	50.4 ± 5	66.6 ± 2.6	84.9 ± 1.6***	937 ± 126	830 ± 53
[CI]	[52.2–89.2]	[51.7–80.3]	[44.3–61.3]	[38.9–61.8]	[61.4–71.8]	[81.7–88.1]	[685–1208]	[726–936]
Concordance
Kappa LIN	0.57**		0.66***		0.28***		0.06
Bland and Altman Bias	4.7		2.4		−17.6		115.6
Pearson or Spearman correlation
r	0.61		0.69**		0.41***		0.09

Intra-methods and inter-method validations were determined by comparing pairwise protocols I vs. II (n = 10), protocols II vs. III (n = 15), and protocols III vs. IV (n = 80) by considering for subjects and controls: statistical difference using parametric (Student’s t-test) or non-parametric (Wilcoxon test) paired tests according to statistical distribution and number of subjects; Statistical concordance using Lin’s concordance coefficient (Kappa LIN) and Bland–Altman test (Bland and Altman Bias). Kappa LIN values are usually framed in a scheme where <0.20 is negligible, 0.21–0.40 weak, 0.41–0.6 moderate, 0.61–0.8 substantial and >0.8 excellent concordance. The Bland–Altman test measured the difference between the two methods, also called ‘bias of measure’; Pearson or Spearman correlations (r) according to statistical distribution and number of subjects. Values are mean ± SEM for percentage (%) of 8-hydroxy-2’-deoxyguanosine-positive spermatozoa (8-OHdG spz) and the mean intensity of fluorescence (MIF) using arbitrary units (a.u.) with agreement limits corresponding to 95% CI.

*, **, *** indicate significant difference.

The mean percentage of 8-OHdG-positive spermatozoa labeled with protocol IV was significantly higher than with protocol III (84.9% vs. 66.6% P < 0.001, Table I) and very weakly concordant (κ_LIN = 0.28, P < 0.001, Bland–Altman bias of −17.6 [−61.4–26.1]) and correlated (r = 0.41, P = 0.0002). MIF values were not significantly different, but the two methods were neither significantly correlated nor concordant between themselves: 830 vs. 937 arbitrary units (a.u.) for protocols IV and III, respectively (Table I). Flow cytometry detection using protocol IV distinguished two populations with different fluorescent intensity profiles (Supplementary Fig. S1). Comparison of only the highly fluorescent (HF) subpopulation revealed the protocol IV values were also neither significantly concordant (κ_{LIN =} 0.051, P = 0.515) nor correlated (r = 0.074, P = 0.516) with protocol III, with a very high bias of 23.6% [−44 to 91.3]. These results were confirmed by the individual distribution profiles (Supplementary Fig. S2) showing no similarity in the staining profiles of protocols III and IV. This was notable for the 8-OHdG-positive sperm percentage.

The discordance between the three profiles suggests that protocols III and IV did not detect the same target in the human sperm nucleus. Competition experiments with exogenous 8-OHdG and 3-dG confirmed this hypothesis (Supplementary Fig. S3). Indeed, we observed a significant and dose-dependent shift in the percentage of 8-OHdG-positive cells detected using protocol III when pre-incubation with exogenous 8-OHdG was performed (from 83.6% for the basal condition to 39.7% when 2.5 μmol of competitor was employed P < 0.001, Supplementary Fig. S3). On the contrary, there was no change to the 8-OHdG-positive percentages when an identical competition was carried out with protocol IV. Although MIF values were decreased for both protocols, the measured effects were not statistically significant due to inter-individual variability (P = 0.14 and P = 0.74 for protocols III and IV, respectively). In contrast, when pre-incubation was carried out with an unrelated 3-dG residue we observed no impact on either the percentage or on the MIF in either protocol III or IV. These results highlighted the higher sensitivity and specificity of protocol III compared with protocol IV for the detection of human sperm 8-OHdG residues.

Correlative analyses and reliability of 8-OHdG levels with sperm and clinical parameters

Using protocol III, a significant positive correlation was observed between MIF and BMI (r = 0.6, P < 0.001, Table II) as overweight patients showed a significantly higher MIF compared with the normal-weight group (P < 0.01, Table II). The MIF showed a significant positive correlation with round cell concentration (r = 0.26, P < 0.05 Table II and PCA, Supplementary Fig. S3B, sperm concentration (r = 0.25, P < 0.05) and polymorphonuclear neutrophil cell (PMN) concentration (r = 0.5, P < 0.001). Furthermore, we revealed significant differences in 8-OHdG⁺ spermatozoa percentage and/or MIF for all sperm parameters investigated, i.e. necrozoospermia (P < 0.05), asthenozoospermia (P < 0.01), teratozoospermia (P < 0.05) and leukocytospermia (P < 0.001) subjects in comparison with respective control groups (Table II). The highest ES was observed for leukocytospermia subjects with a value of 1.82 [1.12; 2.52] in comparison with subjects with a normal seminal PMN concentration (Table II). When subjects with altered sperm parameters (irrespective of the alteration) were compared to normozoospermic subjects, a significant difference in MIF was also seen with an ES = 0.6 ([0.15; 1.05], P < 0.01). We did not determine any significant difference between fecund or infecund subject groups regarding percentage (P = 0.75) or MIF (P = 0.26). Using protocol IV, neither the 8-OHdG spermatozoa percentage nor the MIF showed any significant correlation with sperm parameters, fecundity or clinical data (Table II and PCA analysis: Supplementary Fig. S3B).

Table II

Relationship between sperm DNA oxidation and clinical and sperm parameters.

		Protocol III Anti 8-OHdG antibody			Protocol IV Kit
	Mean ± SEM	% 8-OHdG spz	MIF	EFFECT-SIZE	% 8-OHdG spz	MIF
Clinical parameters		Spearman r correlation
Age (Y)		r = −0.19	r = 0.06		r = −0.15	r = 0.09
<35 (n = 41)	30.8 ± 0.5	71 ± 3	850 ± 148		87.6 ± 1.4	836 ± 86
35–44 (n = 29)	37.7 ± 0.5	60 ± 5	940 ± 224		78.9 ± 3.6	831 ± 67
≥45 (n = 10)	49.0 ± 1.3	65 ± 9	1286 ± 502		78.5 ± 7.9	850 ± 142
BMI		r = 0.21	r =0.6***		r = 0.07	r = −0.05
Normal (n = 14)	22.8 ± 0.4	63.4 ± 6	379 ± 65		86.4 ± 1.7	794 ± 105
≥25 kg/m2 (n = 20)	28.7 ± 1.1	66.6 ± 6	1478 ± 373**		77.3 ± 5.7	871 ± 128
Semen parameters		Spearman r correlation
Vitality (%)		r = 0.04	r = −0.08		r = −0.01	r = −0.28
Control n = 39	47.6 ± 1.2	71.9 ± 3.3	1068 ± 201		86.2 ± 2.0	805 ± 71
Necrozoospermia n = 8	49 ± 2.7	50.5 ± 8.5*	768 ± 301	−0.25 [1.01; 0.52]	79.1 ± 4.5	1093 ± 131
Progressive motility (%)		r = 0.05	r = −0.13		r = 0.07	r = 0.06
Control n = 68	49.3 ± 1.1	66.7 ± 2.7	877 ± 142		84.3 ± 1.8	835 ± 57
Asthenozoospermia n = 12	22.5 ± 2.2	65.9 ± 8.4	1277 ± 236**	0.35 [0.26; 0.97]	77.4 ± 6.3	808 ± 140
Typical morphology (%)		r = 0.12	r = −0.04		r = 0.25	r = 0.17
Control n = 54	11.8 ± 0.9	70.2 ± 2.9	985 ± 168		84.6 ± 2.5	883 ± 71
Teratozoospermia n = 26	1.6 ± 0.2	59.4 ± 5.1*	843 ± 176	−0.13 [−0.59; 0.34]	80.6 ± 2.2**	727 ± 66
Sperm concentration (×106/ml)		r = 0.21	r = 0.25*		r = 0.19	r = 0.03
Round cell concentration (×106/ml)		r = 0.07	r = 0.26*		r = -0.05	r = −0.19
PMN concentration (×106/ml)		r = 0.19	r = 0.5***		r = −0.11	r = −0.07
Control n = 38	0.2 ± 0.3	66.9 ± 3.5	609 ± 75		84.4 ± 2.0	819 ± 57
Leukocytospermia n = 14	9.0 ± 4.0	72.4 ± 7.0	2435 ± 488***	1.82 [1.12; 2.52]	82.4 ± 6.2	819 ± 165

PMN: polymorphonuclear neutrophil cells.

Relationship between sperm DNA oxidation and clinical and sperm parameters were measured using Spearman correlations (r), parametric or non-parametric tests according to statistical distribution and number of subjects (n) and the effect-size determination. Data are presented as the mean ± SEM with agreement limits corresponding to 95% CI.*, **, *** indicate significant difference with P < 0.05, P < 0.01 and P < 0.001 in comparison with respective normal groups. Numbers in bold characters are statistically significant.

ROC curve analysis was used to determine the empirical optimal cut point in relation to sperm parameters. With regard to protocol III, optimal cut-off points of 65.8% for 8-OHdG-positive sperm with a MIF of 552 a.u. were determined, with an AUC of 0.62 ([0.49; 0.74]; sensitivity = 0.54; specificity = 0.72, Fig. 2A) and 0.57 ([0.44; 0.70]; sensitivity = 0.72; specificity = 0.46) for MIF and percentage, respectively. Using this threshold, we clearly distinguished two major subpopulations within the value distributions presented (Fig. 2B): in the first, 23 subjects had DNA sperm oxidation values higher than the cut-off (light gray area surrounded by a red line box), 61% possessed altered sperm parameters, notably leukocytospermia (71%); in the second, 47 subjects had a MIF below or equal to 552 a.u., regardless of the positive spermatozoa proportion (dark gray area surrounded by a blue line), 40% presented with altered sperm parameters and of these 89% had teratozoospermia. Only two patients exhibited leukocytospermia. It follows that 39% of the subjects with ‘pathological’ sperm DNA oxidation were otherwise normozoospermic.

Figure 2

Human sperm DNA oxidation threshold related to sperm parameters. A: Receiver operating characteristic (ROC) curve illustrating the relationship between mean intensity level of fluorescence (MIF) (determined using protocol III) and sperm alterations. B and C: respective (B & C) distributions of the percentage of 8-OHdG-positive spermatozoa in the 80 subjects related to their MIF determined by protocols III and IV. In B, two mean populations were clearly distinguished: subjects with higher DNA sperm oxidation values than the mean for both percentage (66%) and MIF (552 a.u.) (light gray area surrounded by a red line box); subjects with a MIF under or equal to 552 a.u. regardless of the positive spermatozoa proportion (dark gray area surrounded by a blue line). In C, the distribution of marked sperm cell values did not distinguish any particular profile. au: arbitrary units. spz: spermatozoa.

With regard to protocol IV, ROC analysis allowed us to determine an empirical threshold for MIF and percentage of 758 a.u. and 84.4%, respectively, but their AUC were not correct. Moreover, the value distribution analysis (Fig. 2C) did not reveal any significant discrimination regarding global sperm alteration.

The protocol III thresholds determined for 8-OHdG and MIF were particularly discriminating with regard to PMN concentration and leukocytospermia as shown by the high sensitivity (0.86) and specificity (0.79) of the ROC curve (AUC = 0.87 [0.76; 0.99], Fig. 3A) and the factorial discriminant analysis (FDA, Fig. 3B and C). Indeed, we observed that each subject represented by a point in FDA was discriminated satisfactorily according to PMN concentration, clearly distinguishing two populations. This was not the case for protocol IV, with respect to either the total or the HF population (FDA representations show all the points gathered at the center of Fig. 3C).

Figure 3

Discrimination related to polymorphonuclear neutrophil cell concentration. A: ROC curve illustrating the relationship between MIF determined using protocol III and polymorphonuclear neutrophil cell (PMN) concentration. B and C: factor discriminant analysis representing MIF values obtained using protocol III (B) or IV (C, IV for total population and HF IV for HF subpopulation) according to whether the subject has a normal PMN concentration (group 1) or is leukocytospermic (group 2). IV population: total fluorescent sperm population. HF IV: subpopulation with the highest fluorescence intensity mean termed (HF: highly fluorescent).

Stability and repeatability assays

To determine signal stability over the time, fixed spermatozoa were stored in 0.01% sodium azide solution in PBS for 0, 15 and 30 days before labeling. With protocol III, the first significant alteration to MIF and oxidized sperm proportions appeared after 30 days of storage for both subject samples and positive controls (Supplementary Fig. S4A and B). With protocol IV, a significant increase was observed for the oxidized sperm proportion from Day 15 for the positive controls (P < 0.05 Supplementary Fig. S4C). Finally, intra- and inter-assay CVs (Supplementary Table SI) were determined using protocol III. The CVs were less than or equal to 7.5% for both percentage and MIF values for patients and H₂O₂ control values, indicating an acceptable repeatability for diagnostic use.

Discussion

Protocols I, II and III showed consistent reliability. However, several practical characteristics concerning protocols I and II render them difficult for continued diagnostic application. Peroxidase detection coupled with light microscopy is time-consuming, has a long learning curve, and observers must be experienced in order to read 8-OHdG sperm labeling in a consistent manner without introducing inter-operator variability. Detection by FCM in protocol III is impartial and enables evaluation of a large number of cells over a short period of time.

Our study shows a relatively high percentage of positive sperm cells identified by all the protocols when compared with previously reported observations (Meseguer et al., 2008; Zribi et al., 2010, 2011; Cambi et al., 2013). The DNA decondensation pretreatment could partly explain this situation, by increasing antibody access to DNA 8-OHdG sites. This step was optimized in order to ensure access of the anti-8-OHdG antibody and OxyDNA conjugate to most sites of DNA oxidation in sperm chromatin with the goal of obtaining the best signal with the lowest concentration of antibody or conjugate (data not shown). We also reduced the time of decondensation to 10 min with 2 mM DTT to preserve head morphology without seeing a significant difference in labeling in comparison with the 45-min incubation currently used (Twigg et al., 1998; De Iuliis et al., 2009; Muratori et al., 2015).

Moreover, we defined 8-OHdG positivity as any spermatozoa with a spot of fluorescence rather than previous methods that defined positivity when over 50% of the head was stained (Park et al., 2015). This was adopted to give the most sensitive indication of oxidative damage and to derive an objective threshold value for determining correlations with sperm quality. In a similar manner, the 8 M H₂O₂ pretreatment was selected for the positive control after comparison with different published oxidative conditions including various concentrations of H₂O₂ (25 μM (Cambi et al., 2013), 5 mM (Santiso et al., 2010), 4 M (Zribi et al., 2011); Fenton chemistry (De Iuliis et al., 2009; Micillo et al., 2016); or, pesticide-mediated endocrine disruption (Bagchi et al., 1995; Grizard et al., 2007)). To the best of our knowledge, no previous studies have tested 8 M H₂O₂ pretreatment, however we found that 8 M H₂O₂ (1 h, at room temperature, in the dark) provided the most consistent intra- and inter-CV on both the 8-OHdG-positive sperm percentage and MIF (2.2% and 6.9%, respectively for protocol III) regardless of the initial sperm sample quality and/or protocol. As shown (Supplementary Fig. S1) on visible or epifluorescence photographs, or on FCM histograms, sperm morphology was conserved in these conditions. Moreover, in order to verify nuclei structure integrity, in few experimentation sperm cells were countermarked using Hoechst and observed using confocal microscope. The fact that the treatment is performed on raw semen samples may also partly explain why such a high concentration of H₂O₂ had to be used for the positive controls.

The competition assay using the OxyDNA Test® (protocol IV) and high level of exogenous 8-OhdG revealed the test’s poor sensitivity and specificity, whereas a significant and dose-dependent shift was observed with protocol III. Similar results were reported earlier by Cambi et al. (2013), raising the question of the pertinence of the OxyDNA assay’s conjugate for DNA 8-OHdG. Moreover, neither the 8-OHdG spermatozoa percentage nor the MIF determined using protocol IV showed any significant correlation with clinical endpoints, patient fertility or sperm quality. Therefore, data from the competition assay, similar to the stability and repeatability assays, demonstrated the superiority and the accuracy of protocol III to detect human sperm 8-OHdG.

Using protocol III, we observed significant correlations between sperm DNA oxidation and alterations in sperm parameters. Most notable was the positive correlation between MIF, PMN concentration and BMI, consistent with inflammatory and oxidative states. Even though this was expected since PMN are known to be a major source of reactive oxygen species in semen, our data clearly confirm that sperm DNA oxidation is related to a high PMN concentration. Previous studies have already reported relationships between sperm DNA oxidation and isolated sperm parameters including sperm concentration (Kodama et al., 1997; Ni et al., 1997; Park et al., 2015; Micillo et al., 2016), sperm motility (Shen et al., 1997; Kao et al., 2008; Meseguer et al., 2008; Chen et al., 2012; Cambi et al., 2013) or sperm morphology (Cambi et al., 2013). In Meseguer et al. (2008) a correlation between asthenozoospermia and oxidative stress was found when using both the average 8-OHdG signal intensity and the percentage of positive cells, but the authors did not propose a threshold. We are the first to propose thresholds for the percentage of 8-OHdG-positive cells and the average intensity of the signal generated per cell. Optimal cut-off points of 65.8% for 8-OHdG-positive sperm with a MIF of 552 a.u. were determined, with AUC of 0.62 and 0.57 for MIF and percentage, respectively. Even if AUC were moderate, MIF and percentage values were complementary to each other, MIF bringing a good specificity (0.72) and the percentage a good sensitivity (0.72). Of interest was the observation that 39% of the subjects with what we could define as ‘pathological’ sperm DNA oxidation were otherwise normozoospermic. This suggests that motile, fertile spermatozoa could still possess high levels of oxidative DNA damage. It is also worth noting that some leukocytospermic patients did show a normal DNA oxidation value and conversely others with oxidized sperm DNA showed a normal PMN concentration revealing inter-individual variation. Consequently, human sperm DNA 8-OHdG detection should not be seen as a unique definitive marker of sperm DNA integrity. It should be used in combination with conventional sperm quality indicators such as those included in the WHO standardized checklist together with a sperm DNA fragmentation assay. Indeed, base oxidation is a precocious indicator of DNA oxidative damage. Depending on the level of oxidative stress it may or may not be associated with sperm nuclear decondensation and fragmentation conditions that have in turn been associated with male infertility. Regardless, an oxidized sperm nucleus will challenge the oocyte BER pathway, which may result in an increase in the mutational load subsequently carried by the offspring (Aitken, 1999; Crow, 2000; Baker and Aitken, 2005). In our technical validation study we did not observe significant correlation between sperm DNA oxidation and male fecundity. However, having a child is not a presage of current sperm quality or the absence of transgenerational impact.

Other parameters come into play in accomplishing a pregnancy including the oocyte’s repair power. In order to investigate these aspects, complementary investigations are now necessary and should focus on deciphering the relationship existing between sperm DNA oxidation and it’s impact on embryo development and the offspring’s future health.

Conclusion

Our study is the first validation of a protocol to assess human sperm DNA oxidation levels using the anti-8-OHdG antibody and FCM to determine an associated discriminant threshold in relation to sperm quality. We have demonstrated the protocol accuracy with regard to 8-OHdG detection in human sperm. The stability over time (up to 30 days) and good intra- and inter-assay CV provide additional, supporting arguments for potential practical routine diagnostic use. In addition, this test could be a helpful decision-making tool regarding whether or not to pursue therapeutic intervention such as the use of oral antioxidant supplementation. Since over-supplementation of antioxidants could exacerbate sperm DNA decondensation (Agarwal and Allamaneni, 2004; Greco et al., 2005; Ménézo et al., 2007; Lanzafame et al., 2009; Ross et al., 2010; Gharagozloo and Aitken, 2011; Showell et al., 2014) it is critical to know when to apply such therapeutic intervention. However, when antioxidant use is relevant, it may improve live birth rates and increase clinical pregnancy rates, as suggested by the Cochrane review (Showell et al., 2014). Therefore, a reliable, accurate and sensitive human sperm DNA 8-OHdG detection assay, such as the one described here, could be a pertinent addition to the andrology clinician’s toolbox.

Supplementary data

Supplementary data are available at Human Reproduction online.

Acknowledgements

The authors would like to thank Christelle Damon-Soubeyrand, Caroline Vachias, Alexandre Champroux for expert technical assistance during the course of the study.

Authors’ roles

S.V. designed the study, performed the research, performed the statistical analyses and wrote the article in collaboration with H.P. F.B. contributed to design the study, interpret data and write the manuscript in collaboration with H.P. and S.V. A.K. contributed to interpretations and manuscript revision. S.D. performed the research and contributed to interpretations. C.B. and M.B. contributed to realize FCM analysis and interpretations. L.J. contributed to interpretations and to manuscript revision. B.P. performed the statistical analyses in collaboration with S.V. and H.P. and contributed to write manuscript. R.J.A., A.M., and P.G. contributed to the development of the assay and to manuscript revision. J.R.D. contributed to design the study, interpret and write the manuscript in collaboration with H.P. and S.V. H.P. designed the study, performed the research, performed the statistical analyses and wrote the article in collaboration with S.V.

Funding

This work was funded by institutional grants from the Centre National de la Recherche Scientifique, Institut National de la Santé et de la Recherche Médicale and Université Clermont Auvergne (to J.R.D.) and by Clermont-Ferrand Hospital-Centre d'Etude et de Conservation des Oeufs et du Sperme humains research funds (to L.J. and F.B.).

Conflict of interest

P.G. is the Managing Director of CellOxess LLC (New-Jersey, USA), a US-based Biotech which has a commercial interest in the resolution of oxidative stress. A.M. is an employee of CellOxess LLC, while J.R.D. and R.J.A. are honorary members of the CellOxess scientific advisory board.