Abstract

   We used a gas chromatography-mass spectrometry (GC-MS) based
   metabolomics approach to obtain the metabolic profiling of unexplained
   male infertility (UMI), and identified seminal plasma biomarkers
   associated with UMI by a two-stage population study. A robust OPLS-DA
   model based on these identified metabolites was able to distinguish 82%
   of the UMI patients from health controls with a specificity of 92%. In
   this model, 44 metabolites were found differentially expressed in UMI
   subjects compared with health controls. By pathway enrichment analysis,
   we identified several major changed metabolic pathways related to UMI.
   Our findings provide new perspective for the diagnosis of UMI.

Introduction

   Declines in fertility have become a serious concern in recent years,
   and have been accompanied by increasing requirements for infertility
   treatment [[36]1]. Infertility is clinically defined as failure of a
   couple to conceive after one year of regular sexual intercourse. An
   estimated 4–17% of couples seek medical treatment in order to rectify
   their infertility. It is reasonable to assume that there are many more
   cases of infertility that are unreported [[37]2]. About 50% infertile
   cases are attributed to male factor. Despite considerable efforts to
   identify the pathophysiology of the disease, the underlying mechanisms
   of approximately half of infertility cases are still unclear.

   Unexplained male infertility (UMI) is a diagnosis reserved for men in
   whom routine semen analysis results are within normal values, and
   physical as well as endocrine abnormalities were ruled out. Currently,
   routine semen analysis remains the backbone of the evaluation of the
   UMI [[38]3]. Technological developments in the field of metabolomics
   may bring robust and cost-effective clinical tools to fix the
   shortcomings of the routine semen analysis and provide new insights
   into UMI. Seminal plasma has long been used as a key source to
   investigate male infertility. It has been reported that infertile
   patients represent significant change at metabolome level compared to
   fertile men, based on metabolomics fingerprinting studying infertile
   patient seminal plasma [[39]4]. Previous studies have shown that
   metabolic fingerprinting can be applied as a screening tool for male
   infertility problems [[40]5].

   Further tests are certainly required beyond routine semen analysis for
   evaluating subfertile men. Genomic [[41]6–[42]9] and proteomic
   [[43]10–[44]12] efforts were paid in diagnostic and mechanism study for
   male infertility. As metabolomics is a newly emerging technology which
   holds promise for diagnosis of disease and discovery of mechanisms
   linked to disease processes [[45]13, [46]14], metabolomics approaches
   based on serum or urine samples have shown potential of establishing
   useful research tools for male infertility [[47]15, [48]16]. Even
   though the metabolomics changes in seminal plasma, which is the most
   valuable sample type that mediates the chemical surrounding of sperm
   and the chemical function of the ejaculate, are still unclear. Seminal
   plasma is also a favorable noninvasive material which can provide the
   direct understanding of the dysregulated metabolism in reproductive
   organs, and may have great potential in the diagnosis of UMI.

   In this study, we used a gas chromatography-mass spectrometry (GC-MS)
   based metabolomics approach to obtain the metabolic profiling,
   identified 153 seminal plasmatic metabolites from UMI patients. A
   robust OPLS-DA model based on these identified metabolites was able to
   distinguish 82% of the UMI patients from healthy controls with a
   specificity of 92%. In this model, 44 metabolites were found
   differentially expressed in UMI subjects. Our finding provides new
   perspective for diagnosis of UMI.

Materials and methods

Study population

   The study was conducted in full compliance with government policies and
   the Helsinki Declaration, and the procedures were approved by
   institutional ethics committee of Nanjing Medical University. The
   subjects were the volunteers sequentially recruited from affiliated
   hospitals of Nanjing Medical University (NJMU). Every participant was
   fully informed about the purpose of this research, and written consent
   was obtained from each participant. All participants were ethnically
   Han Chinese. A questionnaire was used to collect information including
   personal background, lifestyle factors, occupational and environmental
   exposures, genetic risk factors, sexual and reproduction status,
   medical history, and physical activity. Cases included in this study
   were UMI with normal semen quality in accordance with WHO guidelines
   and normal physical as well as endocrine status. They were husbands of
   couples who attended the affiliated hospitals of NJMU because of
   conception failure for at least 12 months. Their wives had undergone a
   series of clinical tests and their reproductive function was evaluated
   as normal. The men had undergone complete historical and physical
   examinations and measurement of hormones as well as routine semen
   analysis; their partners had undergone complete historical and physical
   examinations, including conventional gynecological examination, serum
   hormone level measurements, tubal assessment, ovarian function testing,
   hysteroscopy, β-ultrasonic examination, immunologic test, microbial
   inspection, and karyotype analysis. Fertile controls (FC) in the
   present study were fertile men recruited and sampled at the same period
   as the UMI cases. They were from the early pregnancy registry of the
   same hospitals who were in the third month following a successful
   pregnancy. They were healthy men with normal reproductive function,
   prospectively confirmed by them fathering healthy babies 6~8 months
   later. Eighty UMI cases and 80 matched FCs were randomly included in
   training set; 50 UMI cases and 50 matched FC were randomly included in
   the validation set.

Semen collection, analysis and storage

   Semen samples were obtained in private by masturbation into a sterile
   wide-mouth and metal-free glass container after an at least 3-day
   abstinence. Routine semen analyses were performed using a computer
   aided semen analyzer (CASA) (WLJY 9000, Weili New Century Science and
   Tech Dev., China), while referring to guidelines in the fourth edition
   World Health Organization (WHO) Laboratory Manual for the Examination
   of Human Semen. After collection, the semen was liquefied at 37°C for
   30 min and then analyzed in accordance with WHO guidelines; variables
   examined included semen volume, concentration, number per ejaculum,
   motility, vitality, progression, and motion parameters. Strict quality
   control measures were enforced throughout the entire study. Semen
   samples were assessed twice, and the counting procedure was automatic
   and performed by two technicians according to WHO recommendations. CASA
   results were compared directly with manual assessments using an
   improved Neubauer hemocytometer over a wide range of sperm
   concentrations. For sperm motility analysis, we used prepared
   videotapes to ensure that all motile spermatozoa were being identified
   using the current and consistent setup parameters.

   After CASA analysis, each semen sample was prepared by Percoll gradient
   methods for separating viable spermatozoa and seminal plasma in the
   laboratory. Then all samples were stored at -80°C prior to GC-MS
   metabolic analysis.

Sample preparation and metabolic profiling

   The metabolic analysis and data processing were performed basically
   according to previous report [[49]17]. Briefly, a 25 μl mixture of
   chloroform, methanol, and water (2:5:2) was added into 25 μl of seminal
   plasma and the samples were vortexed for 1 min. The samples were placed
   at -20°C for 20 min to extract metabolites, followed by centrifugation
   at 12,000 rpm for 10 min, and the liquid layer was transferred into a
   new tube. The residue was extracted with 50 μl of methanol using the
   homogenizer for 10 min followed by centrifugation at 12,000 rpm for 10
   min. The supernatant was then combined with the previous extraction.
   After vortexing, a volume of 100 μl mixture was transferred to a glass
   vial spiked with internal standards (10 μl heptadecanoic acid at 1
   mg/ml and 4-chlorophenylalanine at 0.3 mg/ml). Then the samples were
   vacuum dried at room temperature. The residue was chemically
   derivatized with a two-step procedure and then analyzed with Pegasus HT
   system (Leco Corporation, St Joseph, USA) coupled with an Agilent 6890N
   gas chromatography. Briefly, a 1 μl derivate was injected in splitless
   mode at 270°C. Helium was used as carrier gas; the flow rate was 1.0
   ml/min. The transfer interface and ion source were set to 270°C and
   220°C respectively. The oven temperature was set as a program which
   started at 80°C for 2 min, and then ramped to 180°C at a rate of
   10°C/min, to 230°C with a rate of 6°C/min, finally to 295°C with a rate
   of 40°C/min and held for 8 min. MS data were acquired with m/z range of
   30–600 at an acquisition rate of 20 spectrum/second. Compound
   identification for GC-TOFMS was performed by comparing the mass
   fragments with NIST Standard mass spectral databases using ChromaTOF
   software with a similarity of more than 70% and verified by available
   reference compounds.

Quality control

   An aliquot of 5 μl of each seminal plasma sample was pooled as the
   quality control sample. The QC samples were stored at -80°C and
   prepared once for each batch, and were injected every 10 samples in the
   GC-MS sequence. Responses of internal standards were used for creating
   a QC chart, those samples where deviation of internal standards were
   over control limit of 3 times of relative standard deviation (RSD) were
   re-analyzed. RSD of compounds in each QC sample were calculated;
   compounds that possessed an RSD over 30% were obsoleted.

Statistical analysis

   Fisher's exact test was used to analyze the differences of categorical
   variables such as drinking and smoking status between cases and
   controls. T-test was used to analyze for differences in continuous
   variables such as age and body mass index (BMI). Student-T tests were
   engaged to identify the differential seminal plasma metabolites between
   cases and controls, the data were log-transformed when appropriate. To
   obtain statistically robust results, the Bonferroni method was employed
   for correcting for multiple comparisons.

   Those statistical methods above mentioned were all carried out within R
   programming language. Principal component analysis (PCA) and orthogonal
   partial least squares discriminant analysis (OPLS-DA) models were
   performed with SIMCA-P+ 13.5 (Umetrics, Umea, Sweden).

Results

Characteristics of the study population

   As shown in [50]Table 1, cases and controls were well matched for age,
   BMI, drinking status, and smoking status. No significant difference was
   observed between the two groups ([51]Table 1).

Table 1. Summary description of the subjects used in this study.

   Variables Training Set (n = 160) Validation Set (n = 100) Total (n =
   260)
   Case (n = 80) Control (n = 80) Case (n = 50) Control (n = 50) Case (n =
   130) Control (n = 130)
   Age (years, mean ± s.d.) 29.47 ± 3.78 29.32 ± 3.87 29.77 ± 4.79 30.80 ±
   4.46 29.59 ± 4.18 29.89 ± 4.15
   BMI (mean ± s.d.) 23.64 ± 2.66 23.07 ± 2.83 23.25 ± 2.87 23.09 ± 3.05
   23.49 ± 2.74 23.08 ± 2.91
   Smoking [n (%)]
       Ever 33 (41.25%) 36 (45.00%) 26 (52.00%) 30 (60.00%) 59 (45.40%) 66
   (50.77%)
       Never 47 (58.75%) 44 (55.00%) 24 (48.00%) 20 (40.00%) 81 (62.31%)
   64 (49.23%)
   Drinking [n (%)]
       Yes 39 (48.75%) 34 (42.50%) 22 (44.00%) 29 (58.00%) 61 (46.92%) 63
   (48.8%)
       No 41 (51.25%) 46 (57.50%) 28 (56.00%) 21 (42.00%) 69 (53.08%) 67
   (51.3%)
   [52]Open in a new tab

Seminal plastic metabolomics profile

   A total of 333 metabolites were obtained from metabolic profiling of
   seminal plasma. 153 metabolites were identified by library matching,
   and 96 of those were validated by local standards. After application of
   QC criteria, 136 metabolites were found with RSD < 0.3. Ninety
   compounds of those were identified, and 70 were confirmed by local
   standards. A PCA model was created with a learning set using the 136
   metabolites. PCA score plots ([53]Fig 1) showed the separation trend
   between UMI and FS groups.

Fig 1. PCA score plot for training set.

   [54]Fig 1
   [55]Open in a new tab

Comparison of OPLS-DA models created with metabolomics data and seminal
quality parameter data

   Subsequently, OPLS-DA models were established and demonstrated
   satisfactory modeling and predictive abilities with 1 predictive
   component and 2 orthogonal components (R^2Xcum = 0.333, R^2Ycum =
   0.833, Q2cum = 0.634) with metabolic profiling data ([56]Fig 2A). As a
   comparison, quality parameters of seminal plasma, including semen
   volume, concentration, number per ejaculum, motility, vitality,
   progression, and motion parameters, were used for constructing another
   OPLS-DA model (R^2Xcum = 0.360, R^2Ycum = 0.423, Q2cum = 0.331). The
   score plot showed a poorer discriminate efficiency (a sensitivity of
   82% and specificity of 92%) than that created with metabolomics data (a
   sensitivity of 98% and specificity of 95%) ([57]Fig 2B).

Fig 2. OPLS-DA score plot for training set.

   [58]Fig 2
   [59]Open in a new tab

   A. Generated with metabolomic profiling data; B. Generated with quality
   parameter of seminal plasma data.

OPLS-DA model was validated by an independent sample set

   Furthermore, an independent set of samples were profiled as the
   validation set which consisted of 50 UMI and 50 FC. Profiling data from
   the validation set were used to test the prediction ability of the
   established OPLS-DA model. In the Y prediction scores plot, 41 out of
   50 UMI samples and 46 out of 50 control samples in validation groups
   were correctly assigned to proper class, using a cutoff value of 0.5
   ([60]Fig 3, the Y value was set at 1 for UMI, and 0 for controls in the
   learning group). This result showed the ability of the OPLA model to
   predict the unknown samples to the right groups with a sensitivity of
   82% and specificity of 92%. As all these UMI were diagnosed as fertile
   male under current WHO seminal plasmatic parameter standards, the
   OPLS-DA model is of great significance in the diagnosis of UMI.

Fig 3. Classification result by OPLS-DA model.

   [61]Fig 3
   [62]Open in a new tab

   A. training set; B. validation set.

Seminal plastic biomarkers of UMI

   Significantly altered metabolites in seminal plasma with a VIP
   threshold (VIP > 1) in above mentioned OPLS-DA model, as well as the
   Mann-Whitney U test between UMI and FC, were selected and summarized in
   [63]Table 2. Among these 44 differential metabolites, 37 were confirmed
   by reference standards as shown in [64]Table 2. Variations of these
   differential metabolites were described by fold changes where UMI
   subjected to FC. P-values and fold changes in validation group were
   also listed, and coincided well with those in the training set.

Table 2. Differential metabolites between UMI and FC.

   Metabolites VIP value Training set Validation set
   P Fold change [65]^c P Fold change [66]^c
   Itaconic acid [67]^a 2.14 7.1E-13 0.68 9.1E-06 0.73
   Urea [68]^a 1.97 6.0E-11 14.95 1.6E-07 4.97
   4-Hydroxyphenylacetic acid [69]^a 1.87 6.6E-10 [70]^b 0.19 8.9E-10
   [71]^b 0.2
   Glutamic acid [72]^a 1.8 3.8E-09 0.54 6.8E-07 0.62
   Threonic acid [73]^a 1.79 9.1E-10 [74]^b 0.74 2.2E-06 [75]^b 0.73
   Glucose [76]^a 1.77 6.5E-09 12.46 2.3E-07 13.97
   Pyroglutamic acid [77]^a 1.77 6.9E-09 0.68 2.0E-06 0.73
   Valine [78]^a 1.77 7.1E-09 0.73 2.8E-07 0.76
   Isoleucine [79]^a 1.74 1.3E-08 0.73 9.6E-07 0.76
   Mannose [80]^a 1.71 2.6E-08 2.42 1.5E-07 4.25
   Glyceric acid [81]^a 1.71 3.9E-11 [82]^b 0.44 1.1E-06 [83]^b 0.59
   Hydroxyacetic acid [84]^a 1.7 3.1E-08 0.61 5.4E-05 0.62
   Threonine [85]^a 1.67 5.6E-08 0.74 2.2E-06 0.77
   Serine [86]^a 1.63 1.3E-07 0.77 1.5E-06 0.8
   Gluconic acid [87]^a 1.62 1.6E-07 2.07 3.3E-06 7.85
   Galactose [88]^a 1.61 2.0E-07 0.71 2.5E-03 0.76
   Alanine [89]^a 1.6 2.1E-07 0.63 9.6E-05 0.75
   Spermidine [90]^a 1.59 2.9E-07 1.51 2.8E-03 1.64
   2,3-Dihydroxy-Propanoic acid 1.59 2.9E-07 0.79 3.4E-03 0.84
   Uridine [91]^a 1.52 3.2E-08 [92]^b 0.61 4.2E-04 [93]^b 0.65
   Ethanolamine [94]^a 1.48 1.8E-06 0.78 5.4E-03 0.81
   Glutamine [95]^a 1.48 1.9E-06 5.18 1.6E-06 2.55
   Aspartic acid [96]^a 1.48 2.0E-06 0.33 7.6E-05 0.51
   Phenylalanine [97]^a 1.47 2.2E-06 0.76 2.8E-03 0.8
   Proline [98]^a 1.46 2.7E-06 0.46 8.1E-05 0.6
   Inositol [99]^a 1.42 5.2E-06 0.78 9.8E-04 0.81
   Cadaverine [100]^a 1.42 5.6E-06 0.37 5.0E-05 0.43
   Aminomalonic acid 1.41 5.8E-06 0.77 1.1E-02 0.76
   Tryptophan [101]^a 1.39 8.6E-06 0.68 6.1E-03 0.69
   Rhamnose [102]^a 1.38 1.1E-05 0.76 6.5E-04 0.75
   Xanthine [103]^a 1.37 1.2E-05 0.88 1.6E-01 0.88
   N-Acetyl glucosamine [104]^a 1.35 1.6E-05 0.81 6.1E-03 0.79
   Asparagine [105]^a 1.35 3.4E-06 [106]^b 0.77 7.5E-04 [107]^b 0.76
   Dopamine 1.27 5.2E-05 0.92 1.6E-04 0.87
   Glycine [108]^a 1.27 5.2E-05 0.8 5.2E-04 0.76
   Hypoxanthine [109]^a 1.24 5.3E-06 [110]^b 0.54 1.9E-03 [111]^b 0.63
   Spermine [112]^a 1.21 2.1E-05 [113]^b 4.48 8.5E-03 [114]^b 2.51
   Citric acid [115]^a 1.21 4.8E-06 [116]^b 0.8 6.4E-05 [117]^b 0.78
   Xylitol [118]^a 1.2 1.4E-05 [119]^b 0.79 2.8E-03 [120]^b 0.75
   Dehydroascorbic acid 1.16 2.3E-04 4.49 1.5E-02 3.59
   3-Hydroxyisobutyric acid 1.14 3.1E-04 0.95 9.0E-01 1.02
   Glycerol 2-phosphate 1.09 5.4E-04 0.72 1.2E-04 0.76
   Threitol [121]^a 1.08 1.6E-07 [122]^b 0.59 5.1E-07 [123]^b 0.68
   Dithiothreitol [124]^a 1.06 7.7E-04 0.9 1.8E-01 0.67
   [125]Open in a new tab

   ^a Metabolites confirmed by local standards.

   ^b Data was log transformed before U test.

   ^c Folder changes of metabolites were calculated with mean value in FC
   and UMI group, while those with log transformed were calculated with
   the median. A fold change more than 1 means an increase in UMI group,
   whereas less than 1 represents a decrease.

Correlations between seminal plasmatic metabolites and quality parameters of
seminal plasma

   Pearson correlation coefficient was calculated with metabolic profiling
   data and quality parameters of seminal plasmatic, including semen
   volume, concentration, number per ejaculum, motility, vitality,
   progression, and motion parameters. Significant correlated metabolites
   and quality parameters were listed in [126]Table 3.

Table 3. Correlations between seminal plastic metabolites and semen
parameters.

   AT pH SPMS VCL VSL VAP BCF SC
   P r P r P r P r P r P r P r P r
   4-Hydroxyphenylacetic acid 6E-01 0.05 2E-06 -0.42 8E-03 0.25 6E-03
   -0.25 1E-02 -0.23 8E-03 -0.24 1E-01 0.15 3E-20 0.72
   Fumaric acid 4E-06 0.41 1E-03 -0.29 2E-04 0.34 4E-05 -0.37 1E-05 -0.39
   1E-05 -0.40 3E-05 0.38 4E-14 0.63
   Mannose 3E-04 -0.33 5E-01 0.06 4E-04 -0.32 3E-04 0.33 8E-05 0.36 2E-04
   0.33 8E-08 -0.47 1E-10 -0.55
   Glycerol phosphate 8E-01 -0.03 5E-01 -0.06 5E-01 0.06 1E-01 -0.14 2E-01
   -0.12 2E-01 -0.13 5E-01 0.06 2E-08 0.49
   Itaconic acid 3E-02 0.20 6E-04 -0.31 7E-04 0.31 5E-05 -0.36 2E-04 -0.33
   8E-05 -0.36 2E-02 0.22 3E-08 0.49
   Glucose 6E-05 -0.36 6E-01 0.05 7E-06 -0.40 7E-05 0.36 4E-06 0.41 1E-05
   0.39 9E-09 -0.50 5E-08 -0.48
   Glutamic acid 1E-06 0.43 1E-05 -0.39 3E-08 0.49 3E-08 -0.48 4E-07 -0.45
   1E-07 -0.46 9E-08 0.47 8E-08 0.47
   3-Phenyllactic acid 9E-06 0.40 3E-03 -0.28 7E-08 0.47 8E-08 -0.47 1E-07
   -0.47 5E-08 -0.48 5E-07 0.44 1E-07 0.47
   Lactic acid 4E-02 0.19 6E-01 -0.05 8E-03 0.24 3E-01 -0.10 4E-02 -0.19
   1E-01 -0.15 1E-02 0.22 1E-07 0.47
   Aspartic acid 2E-06 0.42 5E-05 -0.37 4E-09 0.51 3E-08 -0.49 9E-08 -0.47
   3E-08 -0.49 1E-08 0.50 4E-07 0.45
   Glycerol 2-phosphate 1E-01 0.15 2E-01 -0.11 2E-04 0.34 2E-05 -0.39
   4E-06 -0.41 1E-06 -0.43 2E-02 0.21 2E-06 0.42
   Ethanolamine 1E-01 0.14 4E-01 -0.08 2E-02 0.21 7E-04 -0.31 4E-03 -0.26
   2E-03 -0.29 1E-01 0.14 3E-06 0.42
   Cadaverine 9E-04 0.30 3E-02 -0.20 3E-06 0.42 1E-05 -0.39 2E-05 -0.39
   6E-06 -0.41 9E-04 0.30 4E-06 0.41
   Pyroglutamic acid 7E-04 0.31 2E-04 -0.34 3E-07 0.45 2E-08 -0.49 5E-07
   -0.45 6E-08 -0.48 4E-05 0.37 2E-05 0.38
   Valine 3E-04 0.33 4E-04 -0.32 5E-08 0.48 8E-09 -0.50 9E-08 -0.47 2E-08
   -0.49 2E-05 0.38 2E-05 0.38
   Isoleucine 2E-04 0.34 1E-03 -0.30 7E-08 0.47 2E-08 -0.49 2E-07 -0.46
   3E-08 -0.48 3E-05 0.38 6E-05 0.36
   Alanine 4E-05 0.37 2E-04 -0.34 4E-09 0.51 2E-08 -0.49 2E-08 -0.49 5E-09
   -0.51 9E-07 0.44 7E-05 0.36
   Threonine 3E-04 0.33 4E-04 -0.32 9E-08 0.47 9E-09 -0.50 8E-08 -0.47
   1E-08 -0.49 1E-05 0.39 1E-04 0.35
   Proline 6E-07 0.44 1E-04 -0.35 3E-09 0.52 3E-08 -0.48 7E-08 -0.47 2E-08
   -0.49 1E-07 0.47 1E-04 0.35
   Urea 1E-05 -0.39 6E-02 0.18 3E-05 -0.38 4E-04 0.32 1E-03 0.30 9E-04
   0.30 1E-03 -0.29 1E-04 -0.35
   Tryptophan 2E-06 0.43 2E-03 -0.29 9E-09 0.50 3E-08 -0.48 4E-08 -0.48
   2E-08 -0.49 5E-08 0.48 3E-04 0.33
   Threitol 2E-03 0.29 2E-03 -0.28 4E-06 0.41 4E-05 -0.37 2E-04 -0.34
   2E-04 -0.34 9E-04 0.30 4E-04 0.32
   Glycine 1E-05 0.40 1E-02 -0.23 4E-07 0.45 2E-08 -0.49 3E-08 -0.49 8E-09
   -0.50 1E-06 0.43 5E-04 0.32
   Serine 1E-03 0.30 4E-03 -0.27 1E-06 0.43 3E-08 -0.49 3E-07 -0.45 5E-08
   -0.48 2E-04 0.34 9E-04 0.30
   Aminomalonic acid 7E-07 0.44 1E-01 -0.15 5E-06 0.41 2E-04 -0.33 4E-06
   -0.41 6E-06 -0.40 1E-05 0.39 3E-03 0.27
   Phenylalanine 2E-04 0.34 1E-02 -0.24 8E-07 0.44 1E-06 -0.43 3E-06 -0.42
   8E-07 -0.44 9E-05 0.35 4E-03 0.27
   Homocysteine 8E-06 0.40 2E-02 -0.22 2E-09 0.52 3E-07 -0.45 7E-08 -0.47
   4E-08 -0.48 1E-08 0.50 6E-03 0.25
   Ribonic acid 2E-04 0.34 5E-02 -0.18 8E-08 0.47 3E-04 -0.33 2E-04 -0.34
   2E-04 -0.34 2E-05 0.39 7E-03 0.25
   Arabionse 2E-04 -0.34 3E-01 0.09 3E-07 -0.45 3E-04 0.33 4E-07 0.45
   5E-07 0.44 7E-06 -0.40 9E-03 -0.24
   Maltose 3E-04 -0.33 8E-02 0.16 2E-05 -0.38 3E-05 0.37 3E-05 0.38 1E-05
   0.39 2E-05 -0.39 1E-02 -0.23
   Lysine 2E-04 0.33 2E-02 -0.21 4E-06 0.41 4E-07 -0.45 6E-07 -0.44 2E-07
   -0.46 7E-06 0.40 6E-02 0.18
   3-Hydroxyisobutyric acid 8E-05 0.36 1E-01 -0.14 4E-07 0.45 3E-02 -0.20
   1E-04 -0.34 2E-04 -0.34 2E-03 0.28 7E-02 0.17
   [127]Open in a new tab

   AT: abstinence time; SPMS: slowly progressive motile sperm; VCL:
   eurvilinear velocity; VSL: straight-line velocity; VAP: average path
   velocity; BCF: beat-cross frequency; SC: sperm concentration.

Metabolic pathway changes in UMI

   Metabolic pathway analysis were carried out with MetaboAnalyst
   [[128]18]. The seminal plastic biomarkers of UMI were input as data
   source. Seven from 80 homo-sapiens (human) pathways were significantly
   changed as shown in [129]Table 4.

Table 4. Result from pathway analysis.

   Pathway Total [130]^a Hits [131]^b Raw p [132]^c Holm p [133]^d Impact
   [134]^e
   Aminoacyl-tRNA biosynthesis 75 13 1.57E-10 1.26E-08 0.17
   Nitrogen metabolism 39 7 3.52E-06 2.78E-04 0.01
   Glutathione metabolism 38 6 4.10E-05 3.19E-03 0.08
   Arginine and proline metabolism 77 8 4.46E-05 3.43E-03 0.21
   Alanine, aspartate and glutamate metabolism 24 5 4.74E-05 3.60E-03 0.75
   Cyanoamino acid metabolism 16 4 1.38E-04 1.03E-02 0.00
   Glycine, serine and threonine metabolism 48 6 1.59E-04 1.18E-02 0.42
   [135]Open in a new tab

   ^a Total is the total number of compounds in the pathway.

   ^b Hits is the actually matched number from the user uploaded data.

   ^c Raw p is the original p value calculated from the enrichment
   analysis.

   ^d Holm p is the p value adjusted by Holm-Bonferroni method.

   ^e Impact is the pathway impact value calculated from pathway topology
   analysis.

Discussion

   UMI is considered a complicated disease with a genetic, proteomic, and
   metabolic basis. Several causes of UMI are still unknown. Previous
   studies have shown that metabolic fingerprinting can be applied as a
   screening tool for male infertility problems. In our present study,
   using a GC-MS-based metabolite profiling platform, a robust OPLS-DA
   model was able to distinguish 82% of the UMI patients from FC, from
   which 44 metabolites were identified differentially expressed in UMI
   subjects. This finding was validated with the validation set,
   confirming the result was robust. Pearson correlation coefficient
   analysis found that the relationships between the concentrations of
   seminal plasmatic metabolites and seminal quality parameters,
   indicating the result was biologically relevant.

   Metabolomics has been proven to be a promising tool in disease
   diagnosis. The applications of metabolic-based approaches in
   discriminate asthenozoospermia or oligoasthenozoospermia using seminal
   plasma and urine have been emerging recently, confirming metabolic
   approach has great potential in aiding in the routine diagnosis of male
   infertility. However, asthenozoospermia or oligoasthenozoospermia can
   be identified by rountine semen analysis, there is a great need to
   apply metabolic approach in the identification of biomarkers related to
   UMI and further diagnosis of this disease. A robust OPLS-DA model based
   on these identified metabolites was built which was able to distinguish
   82% of the UMI patients from healthy controls, from which the
   concentration of 44 metabolites were found aberrantly expressed in UMI
   subjects. Our findings, which enable the diagnostic of UMI using merely
   seminal plasmatic specimen, indicating our finding has great clinical
   significance, and will be a possible tool for identify UMI for couples
   that decide to conceive a baby.

   The biomarkers identified in metabolomics may be related to the
   pathogenesis of disease, and may provide novel therapeutic targets of
   disease [[136]19, [137]20]. Autoimmune infertility and deficient sperm
   function such as reactive oxygen species and fertilization defects were
   considered to be the causes of UMI, but the metabolic causes of UMI are
   still unknown [[138]3]. [139]Fig 4 was built based on the major
   findings of pathway enrichment revealed in [140]Table 3. A notable
   metabolic feature of UMI patients was the remarkably disturbed amino
   acid metabolism. Various amino acids, including tryptophan,
   phenylalanine, glycine, serine, threonine, isoleucine, proline, and
   valine significantly decreased in seminal plasma of UMI, indicating a
   possible increased consumption of these amino acids. Notably, they all
   can be converted to glutamic acid by transamination, which can further
   produce ammonia. In the UMI seminal plasma, we even found decreased
   glutamic acid, indicating the potential ammonia burden in UMI. The
   human body controls the ammonia concentration in an accepted level by
   consumption of ammonia to generate glutamine and urea. Urea is
   eliminated from body by urination, and glutamine is not toxic.
   Interestingly, among the numerous decreased metabolites in the enriched
   pathways related to amino acid metabolism in UMI, we found urea and
   glutamine increased significantly in UMI seminal plasma, indicating the
   major metabolic signature of UMI seminal plasma is the increased
   catabolism of various amino acids. Furthermore, these changed amino
   acids might play roles in male reproduction. Aminoacyl-tRNAs are
   substrates for translation and are pivotal in determining how the
   genetic code is interpreted as amino acids. Some of these amino acids
   such as tryptophan, serine, and glycine can form aminoacyl-tRNAs. We
   also found the concentrations of tryptophan, phenylalanine, glycine,
   serine, threonine, isoleucine, proline, and valine all showed positive
   relationships with sperm counts, while the concentration of their
   deamination product urea was negatively related to sperm counts.

Fig 4. Sketch map of metabolic changes in UMI.

   [141]Fig 4
   [142]Open in a new tab

   In this study, 4-hydroxyphenylacetic acid was identified as a key
   metabolite in differentiating the UMI and controls, and its
   concentration in seminal plasma was positively related to sperm counts.
   Imbalance between the oxidant burden and antioxidant defense system has
   aroused great attention on UMI, because it exerts a critical
   pathological effect on sperm function through sperm DNA damage in the
   nucleus and mitochondria as well as by inducing lipid peroxidation in
   the sperm plasma membrane [[143]21]. A natural antioxidant defense
   system consists of radical scavengers, chain breaking antioxidants, and
   ROS-metabolizing enzymes offering protection against ROS. Vitamin C,
   uric acid, tryptophan, and taurine are accepted radical scavengers
   [[144]22]. The lack of radical scavengers in seminal plasma contributes
   to UMI. Notably, it is widely reported that 4-hydroxyphenylacetic acid
   and related phenolic compounds exert an anti-oxidant effect
   [[145]23–[146]25]. Therefore, one reasonable mechanism underlying the
   involvement of 4-hydroxyphenylacetic acid in UMI is the lack of this
   anti-oxidant chemical for radical clearage. 4-hydroxyphenylacetic acid
   is an oxidative deaminated metabolite of p-tyramine and also a
   metabolite of tyrosine via enteric bacteria. However, this metabolite
   is not only endogenous, it can also be supplemented directly from foods
   such as olive oil [[147]22, [148]23]. Therefore, there may be a
   convenient therapeutic intervention for UMI by nutrition, which needs
   further investigation.

   Strengths of the present study include its sample size and the
   two-stage population study design, as well as the use of well-validated
   metabolic approach which can provide reliable and accurate metabolic
   data. Future studies need to verify the metabolic changes in this study
   and their effects on male fertility both in in vitro and in vivo and to
   elucidate the underlying mechanism. The application of metabolomics in
   discriminating more clinical types of male reproductive abnormalities
   is also needed in the future.

   In conclusion, using a GC-MS-based metabolite profiling platform, a
   robust OPLS-DA model was established for distinguishing UMI patients
   from FC. From the differentially expressed metabolites in UMI subjects,
   we identified the metabolic signature of UMI. Our findings provide new
   potential diagnosis biomarker and treatment target of UMI and may
   improve the current treatment and future clinical practice of assisted
   reproductive technology. With further study of linkage between
   metabolic pattern and other environmental risk factors, such as diet
   pattern, oxidative stress, etc., it may prompt new couples how to
   increase chances to have a baby.

Data Availability

   All relevant data are within the paper.

Funding Statement

   This work was supported by the National Natural Science Foundation of
   China and Priority Academic Program for the Development of Jiangsu
   Higher Education Institutions. The funders had no role in study design,
   data collection and analysis, decision to publish, or preparation of
   the manuscript.

References