Comprehensive Analysis of Individual Variation in the Urinary Proteome Revealed Significant Gender Differences

Urinary proteomes from 49 healthy adult donors were comprehensively analyzed. Besides age, gender was found to be a crucial factor contributing to individual variation. Proteins that were increased in the male urine samples include major prostate-secreted proteins, and those that were increased in the female urine samples are enriched in immunological pathways. Reference intervals of each urinary protein were estimated, providing the baseline for the discovery of abnormalities.

Graphical Abstract

Highlights

• Comprehensive analysis of inter-individual variation of normal urinary proteome.

• Significant gender differences were observed.

• Proteins increased in female urine are enriched in immunological pathways.

• Estimated reference intervals of proteins as the baseline for biomarker discovery.

Abstract

Disease biomarkers are the measurable changes associated with a pathophysiological process. Without homeostatic control, urine accumulates systematic changes in the body. Thus, urine is an attractive biological material for the discovery of disease biomarkers. One of the major bottlenecks in urinary biomarker discovery is that the concentration and composition of urinary proteins are influenced by many physiological factors. To elucidate the individual variation and related factors influencing the urinary proteome, we comprehensively analyzed the urine samples from healthy adult donors (aged 20–69 years). Co-expression network analysis revealed protein clusters representing the metabolic status, gender-related differences and age-related differences in urinary proteins. In particular, we demonstrated that gender is a crucial factor contributing to individual variation. Proteins that were increased in the male urine samples include prostate-secreted proteins and TIMP1, a protein whose abundance alters under various cancers and renal diseases; however, the proteins that were increased in the female urine samples have known functions in the immune system. Nine gender-related proteins were validated on 85 independent samples by multiple reaction monitoring. Five of these proteins were further used to build a model that could accurately distinguish male and female urine samples with an area under curve value of 0.94. Based on the above results, we strongly suggest that future biomarker investigations should consider gender as a crucial factor in experimental design and data analysis. Finally, reference intervals of each urinary protein were estimated, providing a baseline for the discovery of abnormalities.

Disease biomarkers are the measurable changes associated with a pathophysiological process (1). Without homeostatic control, urine accumulates systematic changes in the body (1, 2). Thus, urine is an attractive biological material for disease biomarker discovery. With the development of techniques, urinary proteomics has become one of the most popular fields in biomarker discovery. Although the total amount of proteins in normal urine is much lower than that in plasma, urine contains a large variety of protein species. To date, over six thousand proteins have been identified by the deep profiling of the normal human urinary proteome (3). The urinary proteome is mainly composed of plasma proteins that pass the glomerular barrier and proteins shed by cells within the urogenital system; therefore, it could reflect both systemic and local conditions of the body (4, 5). Applications of urinary proteomics have been focused on urogenital diseases (6, 7), and there is also a large amount of evidence supporting that changes in urinary proteins could reflect disorders at distant locations such as pancreatitis (8, 9), cardiovascular system diseases (10, 11), brain diseases (12), and breast cancers (13–15).

One of the major bottlenecks in urinary protein biomarker discovery is that both the concentration and composition of urinary proteins are influenced by physiological factors such as aging, sex hormones, diet and exercise (16–24). Previous observations showed a considerable level of intra-individual variation (i.e. variation in the proteins from urine samples collected from the same individual at different time points) (25–29) and an even higher level of inter-individual variation (26, 27) in the normal urinary proteome. In 2011, Mann and colleagues quantified over 600 proteins in the urine samples from seven individuals collected over three consecutive days and evaluated the technical, intra-individual, inter-individual and overall variations in the urinary proteome. In one of our early studies (30), the variation in the urinary proteome was measured in samples from ten male and ten female healthy volunteers by a semi-quantitative method. Inter-gender variation was observed to be greater than intra-gender variation.

Determining the levels and sources of variation among normal urinary proteomes is the foundation to distinguishing the real disease-mediated alterations from those caused by physiological conditions. This goal could not be achieved using a small sample size, as most of the previous studies have used. Recently, Leng et al. established a highly efficient workflow to analyze the sediment of urine samples acquired by ultracentrifugation at high speed (31). They measured the variation among 497 urine samples collected from 167 healthy donors and established reference intervals for ∼2000 proteins. However, to the best of our knowledge, studies of the urinary supernatant, which is the most popular urine sample in biomarker studies with comparable sample sizes, have not been conducted.

In this study, the urine samples from a total of 134 healthy donors were analyzed with proteomic approaches. A total of 49 samples were comprehensively analyzed by 2D LC-MS/MS in the discovery phase to evaluate individual variation in the urinary proteome. A network was established to explore the co-expression patterns of urinary proteins. Markedly different patterns between the male and female urinary proteomes were observed. The gender-related differences were then validated with a set of samples from the other 85 donors, demonstrating that gender is one of the main factors that contributes to individual variation in the normal urinary proteome. Finally, the reference intervals for each gender were estimated, providing a baseline to discover changes under disease conditions. The workflow of this study is illustrated in Fig. 1.

Fig. 1.

The flowchart of the urinary proteome analysis of individuals.

EXPERIMENTAL PROCEDURES

Experimental Design and Statistical Rationale

All donors of the urine samples were recruited from a cohort at the PLA General Hospital in Beijing, China. Donors were provided written informed consent. All the protocols for urine collection were approved by the Ethics Board. All donors were free of acute or chronic illnesses and were not taking any prescription or over-the-counter medications at the time of urine collection. Female donors were not pregnant or menstruating at the time of urine collection. All donors completed a series of physical examinations and laboratory tests. Measurements such as blood pressure, body mass index (BMI)¹, plasma glucose level, total cholesterol (TC), triglyceride (TG), urinary white blood cell, red blood cell and total protein levels, and estimated glomerular filtration rate (eGFR) were employed to select healthy donors. A total of 134 healthy donors aged 20 to 69 years were included in this study. The detailed characterization of donors is provided in supplemental Table S1.

One second voided urine specimen was collected from each donor. Samples were divided into two sets. Set I, including 49 urine samples, was used as the discovery group and analyzed by 2D LC-MS/MS. Set II, including the remaining 85 samples, was used as the validation group for targeted multiple reaction monitoring (MRM) analysis (tier 3). In both sets I and II, each gender and age group were represented by a similar number of donors. The donors were categorized by their age and gender as summarized in Table I. To evaluate the technical reproducibility of the profiling and targeted MS experiments, quality control (QC) samples were generated by pooling all samples in sets I and II in equal amounts and were repeatedly analyzed throughout the entire MS process.

Table I. Characteristics of urine donors. F, female; M, male.

Age group	Set I (F/M)	Set II (F/M)
A (20–30 years)	7/3	9/9
B (30–40 years)	6/4	9/5
C (40–50 years)	4/6	10/10
D (50–60 years)	7/3	6/10
E (>60 years)	4/5	9/8

Urine Sample Collection and Storage

Urine samples were collected in 250 ml conical tubes, immediately acidified to pH 2.7 with hydrochloric acid and then stored at −80 °C to prevent bacterial growth and proteolysis.

Sample Preparation

Samples were centrifuged at 5000 × g for 30 min, and the precipitates were removed. The supernatants were precipitated overnight at 4 °C using 3 times the volume of ethanol for 2 h. After 30 min of centrifugation at 10000 × g, the pellets were resuspended in lysis buffer (7 m urea, 2 m thiourea, 0.1 m DTT, and 50 mm Tris). The protein concentration of each sample was quantified by the Bradford method.

Five hundred micrograms of protein from each sample were digested with the filter-aided sample preparation (FASP) method (32). The protein samples were reduced with 20 mm DTT at 37 °C for 1 h and then carboxyamidomethylated with 50 mm IAA at room temperature in the dark for 45 min. Then, the sample was loaded onto a 10 KD ultracentrifugation filter, washed twice with UA buffer (containing 7 m urea and 50 mm Tris), and washed twice with 25 mm NH₄HCO₃. The treated samples were digested with trypsin (2 μg per 100 μg protein) in 25 mm NH4HCO3. The digested peptides were eluted from the 10 KD filter, and the samples desalted on C18 columns (3 cc, 60 mg, Oasis, Waters Corporation, Milford, MA). The desalted peptides were lyophilized by vacuum centrifugation and stored at −80 °C.

2DLC-MS/MS Analysis

Offline HPLC separation. The digested peptides were fractionated using a high pH RPLC column from Waters (4.6 mm×250 mm, Xbridge C18, 3 μm, Waters Corporation, Milford, MA). The samples were loaded onto the column in buffer A1 (H₂O, pH = 10). The elution gradient was 5–35% buffer B1 (90% ACN, pH = 10; flow rate = 1 ml/min) over 30 min. The eluted peptides were collected at a frequency of one fraction per minute. The dried 30 fractions were resuspended in 0.1% formic acid and pooled into 8 fractions by combining fractions 6–7, 8–9, 10–11, 12–13, 14–16, 17–19, 20–22 and 23–26.

Online LC/MS/MS Analysis

The peptide fractions from both the individual urine samples and the QC samples were analyzed by the same LC-MS/MS configuration. Each fraction was analyzed with a reverse-phase C18 self-packed capillary LC column (75 μm×100 mm). The eluted gradient was 5–30% buffer B2 (0.1% formic acid, 99.9% ACN; flow rate = 0.3 μl/min) over 40 min. A TripleTOF 5600 mass spectrometer (AB Sciex, Framingham, MA) was used to analyze the eluted peptides from LC. The MS data were acquired using high-sensitivity mode with the following parameters: 30 data-dependent MS/MS scans per full scan; full scans acquired at a resolution of 40,000 and MS/MS scans at a resolution of 20,000; rolling collision energy; charge state screening (including precursors with a +2 to +4 charge state); dynamic exclusion (exclusion duration 15 s); MS/MS scan range of 100–1800 m/z; and scan time of 100 ms. A total of 432 LC-MS/MS runs were conducted, including 392 runs for the 49 individual samples and 40 runs for the five technical replicates of the QC sample. To avoid time-dependent systematic bias, the samples from different HPLC fractions were arranged in random order for LC-MS/MS analysis.

2DLC-MS/MS Data Analysis

All MS/MS data were searched against the human section of the SwissProt database (www.uniprot.org, 20121 entries, downloaded on October 2016) using Mascot (Matrix Science, London, UK; version 2.5.01) with the following parameters: fully tryptic digestion and up to two miss cleavage sites; 0.05 Da for both precursor and fragment ion mass tolerance; carbamidomethyl of cysteine as a fixed modification; and methionine oxidation and +43 on Kn (carbamyl) as variable modifications. The database search results obtained with Mascot were further processed by Scaffold (version 4.0.7, Proteome Software Inc., Portland, OR) for protein inference and false discovery rate (FDR) calculation. The probabilities of protein identification were estimated by the ProteinProphet algorithm (33). Only protein groups with at least 2 unique peptides were accepted as positively identified. The false discovery rates (FDR) were set at <1% at both the peptide and protein levels.

The acquired wiff files were imported into the Progenesis LC-MS software (Nonlinear Dynamics, Newcastle, UK, Version 4.0) for label-free quantification analysis. LC alignments were automatically conducted between the runs of the same offline HPLC fractions based on nonlinear mapping of the extracted features. Peptide identifications were then transferred between the aligned features. The abundances of peptide ions based on the robust estimation of the mean log peak area ratio of all peptide ions in the target run were normalized against an automatically selected reference run. Only features of peptide ions with charge states from +2 to +5 were selected for quantification. Protein abundances were calculated by the sum of the normalized abundances of their unique peptides.

MRM Experiments and Data Analysis

Data derived from a spectral library of the normal urine proteome generated by conventional 1D LC-MS/MS and 2D LC-MS/MS using HCD collision were imported into Skyline version 3.5 (33). Skyline was employed to manually select the most intense peptide transitions. Up to five transitions per peptide were traced on a QTRAP 6500 mass spectrometer (AB Sciex, Framingham, MA). Peptides with potential modification sites (cysteine and methionine) and those with missed cleavage sites were excluded. A total of 1–3 peptides from one protein were selected for quantification. All samples were loaded onto a self-packed C18 RP capillary column (100 mm×0.075 mm, 3 μm) with buffer A (0.1% formic acid). The peptides were eluted with 5–30% buffer B (0.1% formic acid, 99.9% ACN; flow rate = 300 nL/min) for 60 min. The samples were analyzed in a random order, and each sample was analyzed three times. All the MRM data were imported into Skyline, which was used for further visualization, transition detection, and abundance calculation. The peptide abundance in a sample was calculated as the mean summed peak area of all selected transitions.

To observe the stability of the MS signal, the mixed peptide sample was used as a QC sample and analyzed 3 times every 2 days during the process of MS analysis.

Estimation of Absolute Protein Concentrations and Reference Intervals

First, a protein MS abundance measurement like iBAQ (34) was calculated for each protein. The abundance was defined as (protein molecular weight) * (summed peptide peak areas)/(number of theoretical peptides that could be detected by mass spectrometry). As reported in an epidemiological study, the mean concentration of urinary albumin among healthy 40–59-year-old Chinese individuals was ∼2.675 mg/L (35). A conversion factor between the protein MS abundance measurement and its absolute concentration was then calculated by dividing 2.675 by the mean albumin abundance among samples within the same age range. Consequently, the concentration of each protein in each sample could be estimated by multiplying its MS abundance measurement by the conversion factor. For each protein, the 2.5% and 97.5% quantiles of concentration values were calculated to estimate the 95% reference interval of protein concentrations in the normal population.

Evaluation of the Effects of Potential Influencing Factors on Urinary Proteome

Hierarchical clustering was employed to evaluate and visualize the overall similarities between samples. This analysis was performed on log₂ transformed protein abundance data using the Pearson correlation coefficient for similarity calculation and the Ward agglomeration method. Only proteins with the biological coefficients of variation (CVs) above the median were used for clustering. PERMANOVA (permutational multivariate analysis of variance) was employed to test whether there was a significant effect of each potential influencing factor on the urinary proteome. All tests were performed with the adonis2 function in the vegan package of R using the same distance matrix as in the hierarchical clustering, and the number of permutations was set to 999.

Protein Coexpression Network Analysis

Network analysis was performed to identify co-expressed urinary protein clusters in different individuals. First, the Spearman correlation coefficients between proteins were calculated, and tests of significance were conducted. p values were then corrected for multiple hypothesis testing using the method described by Hochberg (36). A protein co-expression network was constructed based on the correlation analysis. In this network, each vertex denotes a protein; two vertices were linked by an edge if and only if the correlation between the two proteins was significant (adjusted p value < 0.05), and the weight of this edge was determined by the absolute value of the correlation coefficient of the two proteins. Densely connected subnetworks were discovered by the fast greedy community detection algorithm in the iGraph package of R (37, 38). Each subnetwork was required to contain at least 7 vertices and contain more edges than vertices. To identify hub proteins, Kleinberg's hub centrality scores considering edge weights were calculated for all vertices in a subnetwork (39).

Identification of Differentially Expressed Proteins Between Male and Female Urine Samples

To determine the gender-related proteins in urine, a t test with the assumption of unequal variances was performed on the log transformed abundance value for each protein. Adjusted p values for multiple hypothesis tests were then calculated using the Benjamini-Hochberg method (40). Proteins with an adjusted p value less than 0.05 and a fold change over 1.5 were significantly differentially expressed between the two groups.

Discriminate Model

In the training set (Set I), the peak area of each candidate protein was log₂ transformed and then standardized (mean = 0 and standard deviation = 1). The LASSO logistic regression model was performed with the glmnet package in R (41). Ten-fold cross-validation was used to determine the lambda value for the LASSO model. The lambda value was the largest value with an error within 1 standard error of lambda that gives the minimum mean cross-validated error. The fitted “urinary gender model” included a panel of proteins that perfectly separated male and female samples. In the test set (Set II), the abundance of each protein was calculated as the weighted sum of peptide peak areas. The weight of a peptide was defined as its median peak area among all samples divided by the sum of median peak areas of the peptides belonging to the same protein. The protein abundances were log2 transformed and standardized in the same way as the training set. The performance of the “urinary gender model” was evaluated by the ROC curve with the pROC package in R (42).

Functional Analysis

The subcellular localization of proteins was annotated by the IPA software (Ingenuity Systems, Mountain View, CA). The significance of enrichment for each subcellular localization term was then calculated by one-sided Fisher's exact test, using all urinary proteins identified in this study as the background. Overrepresentation analysis of protein functions and pathways was performed by IPA and Reactome (43), respectively.

RESULTS

Protein Identification and Label-free Quantification

Urine samples from 49 healthy donors aged 20–69 years in sample Set I were comprehensively analyzed by 2D LC-MS/MS. Donors were assigned to five categories based on age, and each category spanned ten-year interval. Thus, the whole sample set was classified into ten groups based on donor age and gender. A total of 3008 proteins were identified with a protein FDR of <1% and with at least two unique peptides. For each sample, a median of 1162 protein groups was identified by peptide-spectrum-matching, whereas 1884 protein groups could be identified by either peptide-spectrum-matching or the match-between-runs approach (Fig. 2A and supplemental Table S2A).

Fig. 2.

Protein identification and CV distributions. A, Number of protein groups identified in each sample. Darker color: number of proteins identified by MS/MS spectral match; Lighter color: number of proteins identified by match between runs. The first and second letters of each label on the x axis denote the age and gender group of the sample, respectively. B, Technical and biological CVs. The x axis denotes the median log10 protein abundance among the QC runs.

A total of 1872 proteins were quantifiable (supplemental Table S2B). A total of 1738 of these with quantitative data in more than half of the samples were selected for further analysis. Technical variation of this analysis was evaluated by calculating the CV of the protein abundances among the five QC replicates. The median and 75% quantile of technical CVs were 0.23 and 0.39, respectively. Considering that the whole mass spectrometry analysis process included 432 LC-MS/MS runs and lasted for 60 days, the technical reproducibly with the label-free quantification used in this analysis is acceptable.

Individual Variation in the Normal Urinary Proteome

For each protein, the biological variation was evaluated as the CV of the measured protein abundances in the 49 urine samples. The median biological CV was 0.60, remarkably greater than the median technical CV (0.23). The distributions of the technical and biological variations are shown in Fig. 2B. The dynamic range spanned seven orders of magnitude. As more abundant proteins are generally easier to precisely quantify, protein abundance was negatively correlated with the technical CV.

Proteins with CV > 0.77 (the 95% quantile of the technical CV) were considered outliers and removed from the subsequent analysis. A total of 1640 proteins remained in the final dataset for quantitative analysis. Hierarchical clustering analysis was performed to classify 48 out of the 49 samples (a female sample in age group C was not included) into ten small clusters (Fig. 3). Each cluster is dominated by samples from an age and gender group. Only five samples were “mistakenly” classified into four clusters in which the dominant samples were from a different group. This observation suggests that gender and age are the core contributing factors to the individual variability of the normal urinary proteome. The significance of their effects was tested by PERMANOVA (p < 0.001 for both age and gender). The effects of five other physiological factors (TC, TG, blood glucose level, BMI and eGFR) were also examined. Only eGFR showed a significant (p = 0.003) effect based on the PERMANOVA tests. It is important to note that eGFR is well known to decline with the normal aging process (44), and it showed a strong negative correlation with age in our dataset, as expected (Pearson correlation coefficient = -0.78). For the other four physiological factors, no significant correlations were observed with either age or gender, and none of them showed significant effects on the urinary proteomes (p = 0.17, 0.11, 0.78, 0.41 for TC, TG, blood glucose level, and BMI, respectively).

Fig. 3.

Hierarchical clustering of the 49 individual urinary proteomes. Samples were categorized into ten groups based on donor age and gender, as illustrated by the colored bars beside both rows and columns of the heatmap. Based on the clustering tree, the 48 samples were classified into 10 clusters, and 1 was an outlier. The clusters are labeled by their dominant sample groups on the corresponding branches of the dendrogram with the same color code.

Urinary Protein Co-expression Network

Protein co-expression analysis was then performed to identify proteins with similar urinary expression profiles across different individuals. A protein co-expression network containing 332 vertices and 480 edges was built as described in the Experimental Procedures section. In this network, each vertex represents a protein, whereas each edge linking two vertices represents a significant abundance correlation between two proteins. Densely connected subnetworks representing protein clusters with similar expression profiles were detected and characterized in Fig. 4A and 4B.

Fig. 4.

Protein co-expression networks. A, Characteristics of the 9 protein co-expression subnetworks. #Vertices. Number of vertices (i.e. proteins) in the subnetwork; #Edges, the median edge number of all vertices; Correlation, the median absolute Spearman correlation coefficient of all vertex pairs, regardless of whether they are linked by edges; CV, the median biological CV of all proteins in that subnetwork; Hub proteins, proteins with a hub score > 0.9. B, Expression profiles of the protein co-expression subnetworks. For the convenience of visualization, the high dimension abundance matrix in each subnetwork was summarized into a one-dimensional array (an eigengene in co-expression network analysis[45]) by principal component analysis. In detail, the abundance values for each protein were log2 transformed and scaled to a mean of 0 and standard deviation of 1. Then, the eigengene for each subnetwork was calculated as its first principle component, which explained 49.65% to 75.59% of the total protein variance. Thus, the major information of protein abundance distribution in a subnetwork could be represented by the distribution of its eigengene. C, The metabolism subnetwork. Proteins involved in different metabolic pathways are displayed in different colors. All of the proteins were positively correlated with each other.

Subnetwork No.1, the largest one, is composed of 55 proteins (Fig. 4C). It is a relatively stable component of the individual urinary proteomes, with a median biological CV of 0.51. Proteins in this subnetwork did not exhibit different abundance patterns between the two gender groups. According to the REACTOME database, a total of 36 proteins are involved in metabolic pathways. Four metabolic pathways were significantly enriched: metabolism of carbohydrates (12 proteins, FDR adjusted p value = 1.39E-7), metabolism of amino acids and derivatives (11 proteins, FDR adjusted p value = 1.18E-04), biological oxidation (9 proteins, FDR adjusted p value = 1.18E-04) and metabolism of vitamins and cofactors (5 proteins, FDR adjusted p value = 0.05). The tissue expression patterns of the proteins in these pathways were then queried in The Human Protein Atlas (www.proteinatlas.org) (45). Twelve proteins were observed to be highly expressed in the kidney, liver or both. Proteins highly expressed in kidney and liver are particularly enriched in the pathway of amino acid and derivative metabolism, in which 4 proteins are increased in both kidney and liver, whereas another one is increased in kidney and some other tissues. Because kidney and liver play important roles in the metabolism of the body, this subnetwork might act as a urinary indicator of metabolic state. Significant alterations observed in this subnetwork might indicate abnormalities in the metabolic system.

The remaining 8 subnetworks are shown in supplemental Fig. S1. Two gender-related subnetworks exhibit the top two largest individual variations: Subnetwork No. 2 contains 46 proteins that were increased in the female urine samples as well as only one increased in the male urine samples, whereas all the proteins in Subnetwork No. 7 were more abundant in the male urine samples than in the female urine samples. Subnetwork No. 7 represents the prostate secretion states of individuals, as it contains all the three predominate prostate-secreted proteins: prostate-specific antigen (abbreviated as PSA or KLK3), prostatic acid phosphatase (ACPP) and beta-microseminoprotein (MSMB) (47). In-depth analysis of the gender-related urinary proteome differences are provided in the next section.

Gender-related Urinary Proteins

A total of 35 and 76 proteins that were significantly increased in the male and female urine samples, respectively, with changes larger than 1.5-fold and p values <0.05 were identified (supplemental Table S3 and Fig. 5A). This gender-related protein list includes 6 proteins in Subnetwork No. 7 and 39 proteins in Subnetwork No. 2.

Fig. 5.

Gender-related differential protein analysis. A, Volcano plot for protein abundance changes between the male and female urine samples. B, Comparison of the subcellular localization distributions of the proteins that were increased in the different gender groups. C, Functional analysis by IPA. The significance of the enrichment was tested by Fisher's exact test. The x axis represents the -log₁₀ p value of this test. The color of the bar represents the Z-score, which was calculated to predict the effect of protein changes. A positive/negative Z-score indicates that proteins with observed changes in abundance have a potential activating/inhibitory effect on a function. D, Pathway analysis with REACTOME. The significance of the enrichment for each pathway is shown as the -log₁₀ p value adjusted for multiple hypothesis tests. E, Enrichment of related diseases analyzed by IPA.

First, proteins that might originate from the reproductive system were analyzed. Three predominant proteins secreted by the human prostate gland (PSA, ACPP, and MSMB) were observed in all 49 samples. Despite not being uniquely identified in the male urine samples, these proteins were 3 to 7 times more abundant in the male urine samples than in the female urine samples. In addition, several proteins that were observed to be highly expressed in the vagina and cervix were increased in the female urine samples. Among them, fatty acid binding protein 5 (FABP5) and small proline-rich protein 3 (SPRR3) were reported to be characteristic vaginal fluid proteins (48).

Next, differences in the subcellular localization distribution and protein functions between the proteins that were increased in males and females were investigated (Fig. 5B). The proteins that were increased in the male urine samples are enriched with extracellular proteins (odds ratio = 3.20 and p < 0.01), whereas those increased in the female urine samples had a significantly greater proportion of cytoplasmic proteins (odds ratio = 1.82 and p < 0.01) and a lower proportion of plasma membrane proteins (odds ratio = 0.41 and p = 0.02).

The molecular functions of the gender-related proteins were analyzed by IPA (Fig. 5C). An interesting phenomenon is that the proteins that were increased in the different gender groups showed an opposite effect on cell invasion (supplemental Fig. S2). In the group of proteins that were increased in males, 4 of the 5 proteins related to cell invasion were predicted to have an inhibitory effect, whereas in the group of proteins that were increased in females, 10 of the 13 proteins related to this function showed a potential activating effect. The two protein groups also display opposite effects on apoptosis, which was predicted to be inhibited by the proteins that were increased in males and slightly activated by the proteins that were increased in females. Additionally, proteins with several molecular functions, such as the migration of cells, catabolism of protein, and cross-linking of peptides, were significantly enriched only in the group of proteins that were increased in females.

Pathway analysis was further performed by the online analysis tool of the REACTOME database (Fig. 5D). The immune system is the most significantly enriched category, containing nearly half (36/76) of the proteins that were increased in females. The markedly enriched immunological pathways included neutrophil degranulation, which contained 24 and 6 proteins that were increased in females and males, respectively, and antimicrobial peptides, containing 9 proteins enriched in females. The differential proteins in urine might reflect the gender-related immunological differences across the whole body. The second enriched pathway category is keratinization, which contains 15 proteins that were increased in females that are specifically involved in the formation of the cornified envelope. This observation might suggest more shed keratinocytes in women's urine than in men's urine.

Nine proteins with interesting functions were chosen for further validation (Table II). An independent set of urine samples from 85 donors (Set II) was employed for validation by the MRM approach. The result of the validation analysis is shown in Fig. 6A. Detailed information on the monitored transitions and their quantitative information in each sample is listed in supplemental Table S4. The same direction of significant abundance changes (p < 0.05 by two-sided t test on log₂ transformed peptide peak areas) was observed as in the Set I analysis for all peptides of the nine proteins quantified in the MRM experiments.

Table II. Nine gender-related proteins that were validated by MRM approach.

Gene	Increase group	Subcellular localizationa	Tissue specificity	Functions and pathways	Related diseasesa,e
ANXA1	Female	Plasma Membrane	Enriched in esophagusd	Activation of cell invasiona; Innate and adaptive immune respondsb,c; chemotaxisa,c; keratinocyte differentiationc	Breast, gastrointestinal cancer; glomerulopathy; inflammatory arthritis
S100A9	Female	Cytoplasm	Highly expressed in squamous epithelial cells of vagina and cervixd	Activation of cell invasiona; neutrophil chemotaxisc; defense response to bacteriumb,c; chemokine and cytokine productionc; apoptoisisa,c	Various inflammatory and autoimmune diseases[49]; various cancers[50]; cardiovascular diseases[51]
TXN	Female	Cytoplasm		cell redox homeostasisb; Innate Immune Systemb	Cardiovascular disease; heart failure, stroke, inflammation, metabolic syndrome, neurodegenerative diseases, arthritis, and cancer[52]
AZU1	Female	Cytoplasm	Enriched in bone marrowd	Induction of positive chemotaxisa,c; Neutrophil degranulationb; antimicrobial humoral responsec	Periodontitis; cornea cancer
SPRR3	Female	Cytoplasm	Characteristic vaginal fluid protein; highly expressed in keratinocytesc	Formation of the cornified envelopeb	Esophageal and skin tumors[53]
FABP5	Female	Cytoplasm	Highly expressed in squamous epithelial cells of vagina and cervixd	Activation of cell invasiona; lipid metabolismc; neutrophil degranulationb; Keratinocyte Differentiationb;[54]	Psoriasis; basal and squamous cell carcinomas[55]; tongue carcinoma[56]
PSA	Male	Extracellular Space	Prostate secreted protein	Inhibition of cell invasiona	Prostate cancer and other prostate diseases[57]
TIMP1	Male	Extracellular Space		Metalloproteinase inhibitor; regulation of cell differentiation, migration and cell deathc	Various cancer and renal diseases
AZGP1	Male	Extracellular Space		Inhibition of cell invasiona; stimulation of lipolysisc	Diabetes; cachexia; obesity[58]; ovarian, bladder and prostate cancers[59]

^a IPA;

^b REACTOME;

^c UniProtKB;

^d The Human Protein Atlas, data available from v18.proteinatlas.org;

^e Urinary Protein Biomarker Database, http://upbd.bmicc.cn.

Fig. 6.

Urinary gender model. A, MRM validation results for nine gender-related proteins. B, Distribution of the “gender score” among the training (2D LC-MS/MS) set. C, Performance of the urinary gender model on the test (MRM) set. ROC curves represent the urinary gender model as well as five predictors using individual proteins in this model.

The nine gender-related proteins were then used to develop a “urinary gender model” to distinguish male and female urine samples. The 2D LC-MS/MS results (Set I) were used to train a LASSO logistic regression model. As shown in Fig. 6B, male and female samples were perfectly distinguished by their gender scores. The same model was then applied to predict donors' genders for sample Set II. The ROC curve illustrating the discriminant power is displayed in Fig. 6C. Most of the samples were correctly classified except for three outliers.

Reference Intervals of Urinary Proteins

In addition to gender, many other factors might also influence the urinary proteome. The reference interval for each protein provides a baseline to discover abnormalities in urinary biomarker analysis. Therefore, absolute concentrations of urinary proteins were estimated based on the mean value of urinary albumin concentration in the Chinese population. The 95% reference interval for each protein was calculated (supplemental Table S5).

Table III lists the concentrations of the top 10 most abundant proteins in the normal urinary proteome. None of these proteins had significant differences in abundance in female and male urine. Most of these proteins are extracellular glycoproteins existing in plasma. Among them, uromodulin (UMOD) is the only one that is exclusively secreted by the kidney.

Table III. The top 10 most abundant proteins in urinary proteome. The unit of protein concentration is mg/l. Proteins were ranked based on the median concentration among all of the 49 samples.

Protein ID	Protein Name	CV	Concen. all	Concen. male	Concen. female
P02768	ALB	0.27	2.74 (1.63–4.57)	2.63 (1.7–4.54)	2.93 (1.77–4.4)
P07911	UMOD	0.92	1.14 (0.24–5.72)	1.04 (0.2–4.73)	1.23 (0.31–4.51)
P02760	AMBP	0.38	1.07 (0.53–1.95)	1.34 (0.82–1.71)	0.93 (0.49–2.09)
P05090	APOD	0.47	0.56 (0.11–1.03)	0.71 (0.28–1)	0.45 (0.08–1.03)
P01834	IGKC	0.45	0.51 (0.28–1.09)	0.53 (0.33–1.05)	0.47 (0.28–1.2)
P41222	PTGDS	0.41	0.49 (0.23–1)	0.61 (0.33–1.09)	0.44 (0.19–0.91)
P01042	KNG1	0.4	0.35 (0.11–0.66)	0.35 (0.18–0.67)	0.34 (0.1–0.65)
P10451	SPP1	0.67	0.3 (0.05–0.84)	0.32 (0.14–0.77)	0.28 (0.04–0.95)
P13987	CD59	0.28	0.23 (0.11–0.34)	0.23 (0.1–0.35)	0.21 (0.13–0.34)
Q7Z5L0	VMO1	0.46	0.22 (0.07–0.42)	0.26 (0.11–0.48)	0.18 (0.07–0.36)

DISCUSSION

The characterization of the normal urine proteome provides fundamental information for the diagnosis and monitoring of diseases. In this study, we found that in addition to age, gender is a crucial factor that contributes to individual variation. The female and male urinary proteomes exhibited distinct expression profiles and could be accurately distinguished by a linear model consisting of merely five proteins. The effect of gender on the urinary proteome has not yet been well studied except for our previous studies using small sample sizes (16, 30).

By looking at gender-related proteins, physiological differences between men and women could be revealed. Typical proteins secreted by the prostate (PSA, ACPP, and MSMB) were enriched in male urine, but did not specifically exist in the male urine. It was reported that Skene's gland, the homolog of the prostate in females, is a source of PSA secretion into female urine (60, 61). These proteins were also found to be secreted or expressed by other tissues, such as lung, kidney, stomach, ovary, and breast, suggesting that their existence in female urine might come from multiple sources (62–65).

The most interesting finding in our study is that many proteins increased in the female urine have functions in the immune system, such as neutrophil degranulation and antimicrobial peptides. Sex differences cause complicated differences in immune response throughout lifetime. These differences occur in both innate and adaptive immune responses, which are thought to be contributed by both the genes located on the X chromosome and the sex hormones (66, 67). For example, females are more susceptible to autoimmune diseases (68), whereas infectious diseases are more prevalent in males. Our study indicates that certain gender differences in the immune system could be reflected in the urinary proteome. A proteomic study on normal saliva also observed elevated abundances of immune-related proteins in females (69). These findings together suggest that the systematic status of the human body might be reflected by various biofluids.

The overrepresented disease terms that are associated with the gender-related proteins identified by IPA are highly enriched in a diverse dermatological, immunological and inflammatory-related diseases and disorders, including dermatitis, hypersensitive reaction, psoriasis, and lupus erythematosus (Fig. 5E). Some of these diseases are known to be more prevalent in females than in males (70–73). In addition, many clues suggest that the levels of these gender-related proteins change in the urine under various other disease conditions. Detailed data can be found in the Urinary Protein Biomarker Database (6).

Individual variation is thought to be an important factor for the inability to validate the biomarker candidates discovered in pilot proteomic studies (17). The bias of factors causing individual variation in case-control studies may lead to an inaccurate list of differentially expressed proteins. Considering the wide range of diseases associated with gender-related proteins, we strongly suggest that future biomarker studies should consider gender as a crucial factor in both experimental design and data analysis. Gender-stratified analysis is preferable when the proper sample size is available; otherwise, the number of female and male subjects must be balanced between groups.

Aging is another important factor that influences the urinary proteome, because aging kidneys show a progressive decline of renal function, such as loss of GFR and decline of the capacity to concentrate urine (74). In a previous study, Bakun et al. reported altered extracellular matrix turnover and declining immune function from the changes in proteins in an elderly group (aged from 70 to 90 years) compared with the young and middle-age groups (23). The correlation with age and the urinary proteome was also observed in our datasets from relatively younger donors (aged 20 to 69 years). Trends of aging-related differences were also observed in Subnetworks No. 1 and 5 in both gender groups. To accurately measure the age-related differences, our further investigations will include donors with an extended age range and use a larger sample size.

In our data set, plasma glucose levels, TC, TG and BMI showed insignificant effects on the normal urinary proteome. Other known (e.g. diet and exercise) and unknown factors may also cause changes in urinary proteins. The results of the protein co-expression network analysis might offer clues for further studies. For example, Subnetwork No. 9, the third most variable subnetwork, includes a hub vertex UMOD. All of the other six proteins showed a strong correlation with this hub protein. UMOD is a tubular specifically secreted protein and a famous marker of renal function (75). However, none of the other proteins have been observed to be highly expressed in the kidney. The functional associations of proteins in this subnetwork remain to be discovered.

Finally, we estimated reference intervals for 1640 urinary proteins quantified in Set I among normal males and females providing a baseline for detecting abnormalities in urine. We previously used the same approach to determine the absolute concentration of 2571 urinary proteins in a pooled sample and showed a medium correlation with immunoassay results (3). It is an interesting phenomenon that UMOD exhibited much broader intervals than the high abundant proteins that originated from the plasma. This suggests that the levels of renal secretion are variable among healthy donors in our study. We hope that the omic-scale reference interval would be helpful for urinary proteome clinical applications and related research.

DATA AVAILABILITY

All raw MS data are stored at the iProx (www.iprox.org) data repository (Project ID: IPX0001256000).