Protein signatures linking history of miscarriages and metabolic syndrome: a proteomic study among North Indian women

Ethics statement

The study was approved by Ethics Committee, Department of Anthropology, University of Delhi, New Delhi, India (Ref. No. Anth/2010/455/1). All participants provided pre-informed written consent to participate in this study.

Participants and plasma sample preparation

Fieldwork was carried out in 15 different villages of Palwal District of Haryana State, India (28.1487°N, 77.3320°E). A total of 1,014 married women of 30–65 years were selected from a Mendelian population of same gene pool to participate in the study. Each of the recruited women was subject to structured interview schedule for collection of data on household composition and reproductive history after obtaining pre-informed written consent. Since menopause itself is a risk factor for adverse cardiovascular outcomes, women with menopause (both natural and surgical) (N = 447) were excluded for the present study. Five women in perimenopausal state were also excluded. Of the 1,009 recruited females, 562 women were in the premenopausal category with an age of 30–46 years. These women were categorized into four groups: (i) Group A—absolute cases with history of miscarriage and MeS, (ii) Group B—absolute controls without any history of miscarriage and MeS, (iii) Group C—cases with MeS but lack any history of miscarriage, and (iv) Group D—cases with history of miscarriage but lack MeS. Age-matched women were selected from each group resulting into a total sample size of N = 80 for the present study (20 each group). Any woman who had experienced one or more spontaneous pregnancy loss or abortion till the second trimester of the pregnancy was grouped in the case of history of miscarriage (Group D). Woman was classified with MeS (Group C) following guidelines of NCEP-ATP (III) 2005 Revised Guidelines (Huang, 2009), that is, a case of MeS shows the presence of a minimum of three abnormalities out of waist circumference, fasting glucose, triglyceride levels, HDL levels, and blood pressure. Exclusion criteria included menopause, established CVDs (like myocardial infarction, angina, etc.), and medication for any disease/infection in the past 1 month (Table S1).

The blood samples from twenty women (subjects) from Group A (absolute cases) were collected. Similarly, age-matched subjects were selected from the other three groups (groups B, C, and D) and 60 blood samples were collected. A total of 80 blood samples analyzed for the present study. Intravenous blood samples (5 ml) were collected in evacuated tubes with and without EDTA (2.5 ml each) from every participant after overnight fasting, and the samples were transported on ice to the Molecular Anthropology Laboratory, Department of Anthropology, University of Delhi within 2 h of collection and processed for further analysis. Plasma was separated from evacuated tubes with EDTA after centrifugation at 3000×g for 10 min. Serum was separated from evacuated tubes without EDTA after centrifugation at 3000×g for 10 min. Plasma and serum samples were aliquoted for single use (to avoid repeated freezing-thawing of the sample) and stored at −80 °C till further use. Serum sample was characterized for biochemical parameters (fasting glucose, total cholesterol or TC, triglycerides or TG, high-density lipoprotein-cholesterol or HDL-C) using enzymatic assay by spectrophotometry within 24 h of sample collection. The protein content of plasma samples was determined by Bradford’s dye-binding method using bovine serum albumin (BSA) as standard (Bradford, 1976).

Two-Dimensional gel electrophoresis and analysis of gel images

To analyze differential protein expression in different case groups and to discern the proteomic link between the history of miscarriage and MeS, plasma proteins of different case groups were subjected to 2-dimensional gel electrophoresis. The protein profiles of case groups A, C, and D were compared with group B. Each protein sample (50 μg) was mixed with rehydration buffer (7M urea, 2M thiourea, 2% CHAPS containing 40 mM dithiothreitol or DTT and 0.2% ampholyte, Sigma-Aldrich, St. Louis, MO, USA). Isoelectric focusing (IEF) was carried out using Protean IEF cell (Bio-Rad, Hercules, CA, USA) as per manufacturer’s protocol. Proteins were separated in the first dimension using IPG strips (immobilized pH gradient strips, pH 4–7, seven cm, Bio-Rad, Hercules, CA, USA). The IEF conditions used were: 200 V for 1 h followed by 500 V for 1 h and 1,000 V for next 1 h and rest of the focusing at 8,000 V for 5 h for a total of 13,000 VH (Bio-Rad, Hercules, CA, USA). After completing the IEF, strips were equilibrated for 20 min each in equilibration buffer (50 mM Tris-HCl, pH 8.8, 6M urea, 4% (w/v) SDS, and 20% (w/v) glycerol) with 10 mg/ml DTT followed by that containing 40 mg/ml iodoacetamide (Sigma-Aldrich, St. Louis, MO, USA). Proteins were further separated in the second dimension using SDS-PAGE (12–5% gel system). The gels were silver stained and analyzed using Alpha Digidoc 1201 software. The densities of protein bands of groups A, C, and D were compared with group B. Any protein band from groups A, C, and D showing a ≥1.2-fold change in density than that of group B was considered differentially expressed. The relative intensities of protein spots were determined, and densitometric results were subjected to cluster analyses (Saroha et al., 2012).

In-gel digestion of proteins

To extract the protein from silver-stained gels, the gel blocks containing protein spots were carefully excised and trypsinized using trypsin gold following manufacturer’s protocol (Promega, USA). The gel blocks were washed with deionized water and kept in destaining solution (0.2% K₃FeCN₆ and 0.04% Na₂S₂O₃; Sigma-Aldrich, St. Louis, MO, USA) till the stain disappeared. The washing was continued with 100 mM NH₄HCO₃ and 100% CH₃CN in 1:1 ratio. The gel blocks were dehydrated with 40 μl of 1:1 acetonitrile (ACN) and deionized water for 15 min and then washed with 40 μl of ACN. The gel blocks were treated with 50 μl of 100 mM ammonium bicarbonate followed by 50 μl ACN (Sigma-Aldrich, St. Louis, MO, USA). The gel blocks were digested with 10 μl digestion buffer containing 0.1 μg trypsin/μl (Promega, USA) for 45 min on ice. The 60 μl digestion buffer without trypsin (1M CaCl₂, 1M NH₄CO₃) was added again and incubated at 37 °C for overnight. Supernatants were collected and dried in a vacuum centrifuge. The peptides were extracted by varying the concentration of ACN and trifluoroacetic acid (TFA). Gel pieces were sonicated in a water bath for 30 min, dried in speed-vac to remove remaining TFA and ACN and finally stored at −20 °C till further use (Wang et al., 2012).

Matrix-assisted laser desorption ionization-MS/MS

To identify the differentially expressed proteins, the trypsinized peptides were analyzed by MALDI-TOF MS/MS (Applied Biosystems, Foster City, CA, USA; Life Technologies, Carlsbad, CA, USA) system. One μl of peptide solution was mixed with an equal amount of α-cyano hydroxy cinnamic acid (CHCA) matrix solution prepared in 70% CH₃CN and 30% of 0.1% TFA. Minimum S/N (signal/ noise) filter was set at 25 μm with an exclusion list for α-cyano-4-hydroxycinnamic acid (CHCA) matrix peaks for the MS/MS precursor selection. A statistically significant probability of the proteins acceptance was based on Mowse score (p ≤ 0.05).

Western blotting

The altered expression of TTR in the plasma of all the three case groups, that is, Group A, C, and D was evaluated using Western blot analysis. Plasma samples of all four groups were separated on 12–5% SDS–PAGE and transferred onto nitrocellulose membrane (Millipore, Burlington, MA, USA). To reduce the chance of non-specific binding, the membrane was blocked using 5% BSA (Sigma-Aldrich, St. Louis, MO, USA) in 1× PBS (phosphate buffered saline, Merck Chemicals, Germany) for 1 h at RT with shaking. Anti-TTR (Abcam, Cambridge, MA, USA) (1:3,000) and GAPDH (1:1,000) were used as a primary antibody for 2 h at RT. The membrane was then incubated with horseradish peroxidase (HRP) conjugated anti-mouse antibody for 1 h with mild shaking. After each step, the membrane was washed with PBST to remove unbound proteins and antibody. Enhanced chemiluminescence assay was used to develop the membrane as per manufacturer’s protocol (G Biosciences, St. Louis, MO, USA) (Mishra et al., 2018). Variation in expression of TTR was analyzed by comparing intensities of the signals observed in the absolute case group (Group A) with that of the absolute control (Group B), only MeS (Group C), and only history of miscarriage (Group D).

ELISA

To evaluate the expression pattern of TTR, ELISA was carried out using plasma samples in all the groups A, B, C, and D (N = 20 each). Each well of a 96-well microtiter plate (Thermo Scientific, Waltham, MA, USA) was coated with anti-TTR antibody (ProSci Antibody, Poway, CA, USA) in 1:2,000 dilution with coating buffer (0.01M Na₂CO₃ and 0.035M NaHCO₃, pH 9.6) and incubated at 4 °C for overnight. Next day, the plate was washed three times with 100 μl PBST (phosphate buffered saline with tween 20) and incubated with 200 μl blocking buffer (2% BSA; Sigma-Aldrich, St. Louis, MO, USA) for 1 h at RT. The blocking buffer was removed, washed and incubated with plasma samples (1:500) for 3 h at RT. After treating with plasma, plates were washed and incubated with primary antibody (anti-TTR). Again, plate was washed and incubated with 100 μl of anti-mouse HRP conjugate secondary antibody for 1 h and developed with ortho phenylene diamine (1 mg/ml in 0.05M citrate buffer and five μl/ml H₂O₂; Sigma-Aldrich, St. Louis, MO, USA) for 30 min. The reaction was stopped by adding 2N H₂SO₄ to each well and OD_492nm was measured using ELISA plate reader (Biswas et al., 2013).

Protein networking analysis

Differentially expressed proteins were functionally annotated, and their network of interactions was predicted using bioinformatics web tools. Blast2Go bioinformatics platform (Conesa & Götz, 2008) was used for functional annotation of 18 proteins differentially expressed in different case groups. To understand the possible biological processes linking the history of miscarriage with MeS, protein-protein interactions were examined: (i) involving all the 18 differentially expressed proteins, and (ii) considering only the seven proteins (HP proteins, ApoA1, transferrin (TF), Serpin38, and SYNE 1) commonly dysregulated in all the three case groups and TTR (involved only in absolute cases exclusively). Functional association among differentially expressed proteins in all case groups was revealed by networking analysis using STRING webtool (Search Tool for the Retrieval of Interacting Genes/Proteins ver. 10.0). The analyses were based on the known and predicted protein-protein interactions (physical and functional) available in the database. The database houses the information derived from sources, such as interactions known from high-throughput experiments, co-expression analysis, and previously available literature, and interactions predicted from computational analyses of genomic characters and gene/protein functions based on orthologs and interologs. The protein–protein interaction with confidence score >0.700 was considered. The confidence score was calculated by weighing and integrating the data for all protein interactions. Colored nodes represent direct interaction, white nodes represent the absence of interaction, whereas the thickness of lines signifies the strength of associations (Bag et al., 2014).

Statistical analyses

One-way ANOVA and principal component analysis was performed using SPSS ver. 16.0 for Windows (SPSS Inc., Chicago, IL, USA) and GraphPad Prism 7.01 for Windows (GraphPad Software, La Jolla, CA, USA). Statistical significance was tested at p < 0.05. Cluster analysis was done using Multiple Experiment Viewer software (MeV, Boston, MA, USA).

Identification of differentially expressed proteins in only miscarriage, only metabolic syndrome, and absolute cases

Densitometric analyses of 2-D gel electrophoretic profiles of plasma proteins of different case groups revealed 18 differentially expressed proteins in groups A, C, and D (Fig. 1). MALDI-TOF/MS-MS analysis revealed that eighteen differentially expressed proteins (Table 1) play significant roles in lipid metabolism, ion transport, and binding processes in the cell. Most of the proteins (15/18) were upregulated in case groups (A, C, and D), while a few (3/18) showed down-regulation, that is, apolipoprotein A, chain B TF binding protein and chain A TF.

Clustering of differentially expressed proteins based on their expression

In cluster analysis, the 18 proteins were segregated based on the expression pattern (Fig. 2A) and formed five clusters (Fig. 2B). Different isoforms of ApoA1 showed different expression patterns, and therefore formed different clusters. In cases with only MeS (Group C), one of the isoforms (S15) was upregulated (cluster 2), while the other isoform (S14) was downregulated (cluster 4). Similarly, the expression pattern of two isoforms of chain A, and TF (S2 and S3) also formed distinct clusters (cluster 2 and 3). It may be noted that ApoA1 and chain A, and TF are the major components of the vital cellular processes, that is, lipid metabolism and iron transport, respectively. The distinct functions of these isoforms in different case groups need further analyses. Mostly the proteins were involved in biological regulation and cellular processes followed by the proteins involved in metabolic processes (Fig. 3).

The Venn diagram shows the distribution of differentially expressed proteins among three case groups (Figs. 4A and 4B). Of the 18 differentially expressed proteins, 11 proteins were shared by the groups C (MeS) and group D (history of miscarriage) that include, ApoA1, TF (two isoforms), chain B TF binding protein B, HP proteins (three isoforms), SYNE protein 1, serpin38, CRA_a and coiled-coil domain protein (regions II and V). Of these, seven proteins were also shared by Group A case (region V). Cases from group C and group D shared only four dysregulated proteins (region II), whereas group A and group C cases shared only two proteins (region VI). Interestingly, group A and group D cases lacked any shared protein (region IV). The cases from group A (region VII) and group D (region I) showed one group-specific protein, whereas cases of group C (region III) showed three group-specific proteins. Transthyretin (TTR, S1) showed upregulation in group A (absolute cases) exclusively and occupied distinct quadrant I in the PCA plot (Fig. 4C), so TTR was chosen for further verification in all the samples.

Analyses of interactions among commonly dysregulated proteins in the case groups

The differentially expressed proteins showed varying levels of interactions. For example, TTR, TF, ApoA1, and HP showed strong interactions with each other (Fig. 5A), whereas, FGF1 showed a weak interaction with TF. The interaction among TTR, TF, ApoA1, and HP remained strong whether we analyzed all eighteen differentially expressed proteins, or only the seven proteins (HP proteins, ApoA1, TF, Serpin38, and SYNE 1) commonly dysregulated in the three case groups and TTR (involved only in absolute cases exclusively) (Fig. 5).

Validation of higher expression of transthyretin in absolute cases

ELISA and Western blot analyses confirmed upregulated expression of TTR in the plasma of absolute cases with both history of miscarriage and MeS. As compared with absolute control or group B (n = 20), TTR showed significant upregulation in absolute cases or group A, (2.5-fold, n = 20) only MeS cases or group C (2.23-fold, n = 20) and only miscarriage cases or group D (4.1-fold, n = 20) (Fig. 6C). Similarly, the densitometric analysis of Western blots validated a significantly higher expression of TTR in absolute cases (1.56-fold), only MeS cases (1.33-fold) and only history of miscarriage (2.21-fold) as compared with absolute control (Figs. 6A and 6B).

Strengths

The major strength of this study is a collection of samples from a single Mendelian population sharing a common gene pool with a similar lifestyle, physical activity and dietary habits. Such a retrospective study in a single population thereby reduces the effect of genomic variations, dietary habits etc., in different ethnic groups on proteins to the maximum. In a hospital based case-control study, such matching may not be possible. Since plasma proteome is largely affected by intervention through drugs, insights into the pathways may be obtained through the recruitment of participants who are not under the influence of medication. Therefore, another major strength of the study is the recruitment of naïve participants who were not on any drug/medical prescription. The study also involved precise phenotyping by cross-validation of reproductive history through repeated multiple-questions involving “miscarriage.”