All experimental animals were obtained from an intensive dairy farm in accordance with the requirements of the Veterinary Medical Ethical Committee of the Local Agricultural Department of Mishan, Heilongjiang province, China. All experimental animals were maintained in accordance with the International Guiding Principles for Biomedical Research Involving Animals. All cows were fed a total mixed ration (TMR) during the prepartum period, which consisted of 9 kg of concentrated feed, 17 kg of silage maize, 4 kg of hay, and 400 g of fat. The nutritional analysis of the TMR was 55.60% dry matter, 16% crude protein, 5.60% fat, 39.10% neutral detergent fiber, 20.30% acid detergent fiber, and 1.75 mcal/net energy for lactation, 180 g of Ca, 116 g of phosphorus.

Blood samples were collected from randomly selected Holstein cows at three time points: 3rd day before calving (day −3), day of calving (day 0), and 3rd day after calving (day +3). None of the cows showed any clinical symptoms on day −3 and diseases other than hypocalcemia on day 0 were excluded. The cows were assigned to one of three groups based on serum Ca concentration on day 0: the CH group (n = 9), obvious clinical symptoms of hypocalcemia and Ca concentration of <1.80 mmol/L; the SH group (n = 9), no clinical symptoms of hypocalcemia and Ca concentrations of 1.80 to 2.00 mmol/L; and a control (C) group (n = 9), no clinical symptoms of hypocalcemia and Ca concentrations of 2.00 to 3.00 mmol/L.

Blood samples (10 mL) from all tested cows were collected from the caudal vein at three time points in accordance with the International Guiding Principles for Biomedical Research Involving Animals. The samples were centrifuged immediately at 4,000 g for 10 min. Then, the serum fraction was frozen in liquid nitrogen and stored at −80°C until subsequent analyses. The serum Ca concentration was detected using a commercial kit (651564-01; Roche Diagnostics, Mannheim, Germany) with an automatic biochemical analyzer (Modular DPP; Roche Diagnostics, Germany). The serum Ca concentrations at the three time points are shown Figure 1 and group values are shown in Supplementary Table S1.

To reduce individual differences, every sample for 2D-DIGE was obtained from serum samples of three animals. The test design is depicted in Table 1 and the study strategy is shown in Supplementary Table S2.

Serum samples were thawed at room temperature and pooled in accordance with the study design. Protein concentrations were measured using the Bradford protein assay. Lysis buffer and 3.5 mg/mL of bovine serum albumin (BSA) at a ratio of 1:3 with ddH₂O at a ratio of 1:9 with the sample were added to a test tube. A 5-μL aliquot of the sample, which was diluted according to the above method, was homogenized in BSA (standard protein buffer) and 875 μL of protein assay buffer (Bio-Rad Laboratories, Inc., Hercules, CA, USA). Protein concentrations were quantified spectrophotometrically at 595 nm with BSA as a standard.

The abundance of albumin and immunoglobulins are varied in serum, and can account for up to 95% of proteins in serum. To increase resolution of the gels, highly abundant proteins were extracted and purified using a protein purification kit (GE Healthcare, Buckinghamshire, UK), while impurities were removed using a 3-kD hyperfiltration tube (UFC800396; Merck Millipore, Billerica, MA, USA), and then centrifuged at 10,000 g for 30 min. All protein concentrations were quantified twice according to the above method.

NaOH (50 mmol/L) was added to the samples to adjust the pH to 8.5 and lysis buffer was added to adjust the concentration to 5 mg/mL. The interior label was a mixed liquor of all nine samples and 400 pmol Cy2 fluorescent label was added to every 50 μg of protein in the interior label. The samples were randomly labeled with Cy3 and Cy5. The CH1 sample was added twice to the different gels because of the odd number of total samples. This design was chosen to reduce the occurrence of technical errors and increase replication.

Cy3- and Cy5-labeled samples (50 μL) were mixed with the interior label and then two-fold buffer (7 M urea, 2 M thiourea, 4% CHAPS (3-[{3-Cholamidopropyl} dimethylammonio] propanesulfonic acid), 1% Bio-Lyte, pH 3 to 10, 20 mg/mL dithiothreitol (DTT) and hydration buffer (7 M urea, 2 M thiourea, 4% CHAPS, 0.5% Bio-Lyte, 10 mg/mL DTT) were added for a final volume of 410 μL.

Immobiline DryStrips (length, 18 cm; GE Healthcare, UK), containing immobilized ampholytes forming a linear pH gradient of 4 to 7, were used for isoelectric focusing (IEF) in the first dimension. The strips were rehydrated in 100 μL of a buffer containing 7 mmol/L urea, 2 mmol/L thiourea, 4% CHAPS, 30 mmol/L Tris buffer, and deionized water. Immobilized pH gradient (IPG) rehydration buffer contained 4.5 mg of 1% DDT, 4.5 μL of 40% ampholyte, and trace amounts of bromophenol blue. The samples were applied to the strips following a cup loading protocol and IEF was performed at room temperature using the IPGPhor IEF system (Amersham Biosciences, Amersham, UK) at 50 V per strip for 13 h, according to the following program: fast gradient at 200 V for 30 min (to remove impurities), linear gradient at 1,000 V for 6 h, focusing at 10,000 V for 60,000 vh, and fast gradient at 500 V to the end of the run.

After IEF, the IPG strips were incubated in equilibration buffer I (6 mol/L urea, 20% glycerol, 2% sodium dodecyl sulfate (SDS), 0.375 mol/L Tris-HCl, 0.1 g DDT) while shaking for 15 min, and then in equilibration buffer II (buffer I in which DTT was replaced by 0.4 g iodoacetamide) for a further 15 min. Thereafter, the strips were individually applied to the surface of a vertical 12% polyacrylamide gel and sealed with 0.5% agarose dissolved in electrophoresis buffer (3 g Tris-alkali, 14.4 g of glycerol, 0.1% SDS) with traces of bromophenol blue. SDS-polyacrylamide gel electrophoresis (PAGE) Broad Range or All Blue Precision Plus Protein (Bio-Rad Laboratories, Inc., USA) molecular weight standards were loaded at each end of the strip. Second-dimension electrophoresis was performed using an EttanDALT II gel at a constant gain of 30 V for 60 min, then 80 V for 16 h.

Following electrophoresis, signals from the three fluorophores in each gel were sequentially scanned using a Typhoon 9410 fluorescence imager (GE Healthcare, UK). The corresponding excitation and emission wavelengths for Cy3 and Cy 5 were 532 and 580 nm, and 633 and 670 nm, respectively. Cy2 images were obtained at wavelengths of 488 and 520 nm. The DeCyder (V6.5) was used for analysis of the DIGE gel to determine differences in protein content. The t-test was used to identify significant differences in protein expression. A probability (p) value of ≤0.05 and an increase in expression of >1.5-fold were considered statistically significant.

According to the results of the second protein quantification, the preparation gel, which was used for MS, was used for all samples according to the above-mentioned method. The analysis gel, which was used to determine differences in protein expression, was used to compare the preparations to ensure the places of proteins exhibiting differences in expression profiles. Then, spots containing proteins were carefully excised using a scalpel. After freeze-drying, the spots were mixed with trypsin solution and then subjected to enzymolysis at 37°C in a water bath for 16 h. Before MS analysis, peptides were extracted by the addition of 5% trifluoroacetic acid (TFA) solution. Then, 0.5 μL of each peptide sample was placed on target of the mass spectrograph. After addition of 5 mg/mL of matrix solution (α-cyan-4-hydroxy-cinnamic acid, 0.1% TFA, 50% acetonitrile), the peptide samples were analyzed by MALDI–TOF MS using an ABI-4800 mass spectrometer (Applied Biosystems, Foster City, CA, USA) under the following parameters: scanning mode, reflection-type; range, 700 to 4,000 Da; laser energy of full scan, 5,000; second scan, 5,500. All proteins were identified by a search of the National Center for Biotechnology Information (NCBI) database.

Gene ontology (GO) analysis is an important analytical method in bioinformatics software for high-throughput data. Cytoscape (http://www.cytoscape.org/) is an open source software platform for analysis of high-throughput data. The many plugins with Cytoscape are used for GO analysis and prediction of protein structure, and so on. In this study, the Bin/GO tool, a plugin of Cytoscape (version 2.8.3), was used for GO analyses.

All identified proteins were used for pathway analysis with the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (http://www.genome.jp/kegg/).

The MS results showed that expression of the alpha-2-macroglobulin (A2M) protein was up-regulated in the CH group and down-regulated in the SH group, as compared with the C group. To verify the results of 2D-DIGE/MALDI-TOF-MS, differentially expressed proteins were chosen for western blot (WB) analysis using rabbit anti-A2M antibody (bs-9505R; Bioss, Inc., Woburn, MA, USA).

Crude serum proteins from three groups (CH group, n = 3; SH group, n = 2; C group, n = 3) at day −3 before calving were used for WB analysis. Equal amounts of proteins were separated by 5% to 20% SDS-PAGE and transferred to polyvinylidene fluoride membranes. After blocking with 5% skim milk powder for 2 h at 37°C, target proteins were incubated with primary antibodies for 2 h while shaking overnight at 4°C. Then, the membranes were incubated with secondary antibodies overnight at 4°C. Afterward, the membranes were dried to form a gel for scanning. The A2M protein content of the three groups was analyzed to verify relative expression levels.

The Ca concentrations of the samples are shown in Figure 1. All data are provided in Supplementary Table S1. The Ca concentrations of all three groups were decreased on day 0, and there were no significant difference on day −3. Although the Ca concentrations of three groups were decreased on day 0, that of the C group was still in the normal range. The Ca concentration in the C group was restored to a normal range on day +3, but not the other two groups, especially the CH group.

A representative fluorescent image is shown in Figure 2. All images are provided in Supplementary Table S3. Fifty-seven protein spots were separated. Each of these protein spots was significantly different between at least two groups (1.5-fold difference in abundance; p<0.05). Twenty-seven spots were selected for MALDI-TOF-MS, each spot on the gel was required not to be connected to any other spot, and the blot was clear and complete.

Thirteen peptide mass fingerprints (PMFs) of 27 spots were obtained after MALDI–TOF-MS identification. The NCBI database was searched to identify the 13 PMFs to confirm the identification of the proteins. Of the 13 PMFs, eight different proteins were identified.

The results of MS analysis are shown in Table 2. The eight differentially expressed proteins included third party data (TPA): vitamin D-binding protein precursor (GC), complement C4 precursor (C4), paraoxonase 1 (PON1), A2M, endopin 1 (EN1), TPA: apolipoprotein A-IV precursor (APOA4), alpha-1-antitrypsin, and haptoglobin (HP).

The expression levels shown in Figure 5 were calculated as the ratio among the three groups. All data are shown in Supplementary Table S5. A2M expression was increased (CH/C: 6.10%) in the CH group and reduced (SH/C: 15.53%) in the SH group. The WB analysis strongly supported the results of 2D-DIGE combined with MALDI–TOF-MS.

In this study, the serum Ca concentrations of all cattle decreased on day 0, which was consistent with the results of previous studies [11,12]. This decrease likely resulted from providing nutrition to the fetus and formation of the beestings. Mechanisms are in place to efficiently maintain normal serum Ca concentration most of the time, but occasionally these homeostatic mechanisms fail, which results in hypocalcemia [3]. Physiologically, hormones, including calcitonin (decreases Ca concentration), parathyroid hormone (PTH, increases Ca concentration), and 1,25-dihydroxyvitamin D₃ (1,25(OH)₂D₃, increases Ca concentration), act to control Ca metabolism [1]. 1,25(OH)₂D₃ is required to stimulate the intestine to absorb dietary Ca. This hormone is produced in the kidney from vitamin D [13]. About 85% of serum 1,25(OH)₂D₃ binds with high affinity to GC [14]. GC is a 55-kDa protein that is a member of the albumin superfamily and is secreted by the liver. One of the physiological functions of the GC protein is vitamin D transport and storage [15]. According to the results of this study, GC was down-regulated in the CH group, as compared with the C and SH groups, while there was no significant difference between the C and SH groups. As a potential pathogenesis of CH, GC was reduced before calving. When the serum Ca concentration was decreased on calving day, the amount of GC is insufficient to combine with vitamin D. Thus, the role of vitamin D was inhibited leading to a severe decrease in serum Ca concentration. Ca is continually consumed by providing nutrition to the fetus and forming the beestings, which prevents restoration of serum Ca to normal concentrations.

Another surmise was that APOA4, a 46-kDa high-density lipoprotein (HDL), is secreted from the small intestine in response to lipid absorption and chylomicron formation [16]. It is regulated by leptin, a major component of energy homeostasis [17]. The integrated pathway (Figure 5) shows that ApoA4 enters the circulation on lymph chylomicrons and is subsequently released as a lipid-free protein during lipolysis of chylomicrons [18]. Haas and Mooradian [19] reported that vitamin D levels were positively correlated with plasma HDL levels, but inversely with plasma low-density lipoprotein cholesterol levels. The result of this study showed that APOA4 was down-regulated in the CH group, as compared with the C and SH groups, while there was no significant different between the C and SH groups. An addition physiological function of the GC protein is VD₃ storage. In summary, serum APOA4 levels may be affected by Ca regulation and may be positively correlated with GC expression.

In this study, the A2M protein was detected by 2D-DIGE/MALDI-TOF-MS and verified by WB analysis. A2M expression was up-regulated in the CH group and down-regulated in the SH group, as compared with controls. Wang et al [20] reported that A2M plays a role in osteoarthritis and suggested that A2M is a powerful inhibitor of many cartilage catabolic factors through protease activities. A2M inhibits cartilage catabolic factor through decreased gene expression to protect the cartilage. In addition, tartrate-resistant acid phosphatase (TRACP) 5b is derived from osteoclasts, which circulate in large complexes that contain A2M and Ca [21]. Many reports have suggested that serum TRACP 5b is a specific and sensitive marker of bone restoration that can be used for the diagnosis and follow-up of skeletal pathologies [22]. Bone-resorbing osteoclasts contain high amounts of TRACP 5b and secrete it into the bloodstream.

This evidence suggests that A2M expression is related to Ca concentration. Although the relationship between A2M and Ca remains unclear, A2M was useful as an early detection index to distinguish CH from SH from healthy dairy cows before calving. However, further studies are needed to elucidate the underlying mechanisms.

Other proteins identified in this study included PON1, C4, HP, EN1, and AAT. PON1 is a 44-kDa glycoprotein that hydrolyzes organophosphates found in pesticides, neurotoxins, and arylesters [22,23]. PON1 is synthesized mainly in the liver, with lower expression found in the lungs, heart, brain, kidneys, and small intestine [24]. The results of 2D-DIGE/MALDI-TOF-MS showed that PON1 expression was down-regulated in the CH group, as compared with the SH and C groups, respectively. A possible mechanism for the effects of PON1 is the up-regulation of PTH receptor 1, as described by Lu et al [25]. PTH is an important hormone for the balance of Ca and P. Since the PTH receptor 1 specifically binds to PTH, the level of PTH receptor 1 is positively correlated with the level of PTH, suggesting that PON1 affects serum Ca concentration via the PON1 receptor 1 and available PTH.

C4 activation acts as immune surveillance system in healthy and altered host cells to detect the presence of foreign matter. C4 is a component of C3 convertase in the classical and lectin complement pathways, which is expressed in two polymorphic isotypes, C4A and C4B [26].

HP is primarily produced in the liver and is functionally important for binding free hemoglobin from lysed red cells in, preventing its toxic effects [27]. Decreased HP is a marker of hemolysis, as HP levels become depleted in the presence of large amounts of free hemoglobin [28].

EN1 is present in endorine tissues, as specified by “endo” and its serpin-like characteristic indicated by “pin”, thus designating this new serpin as endopin 1. EN1 is present within isolated chromaffin granules, secreted from chromaffin cells in culture, and inhibits trypsin and plasmin via cleavage at basic residues [29].

AAT is a 52-kD glycoprotein produced mainly by hepatocytes and secreted into the blood. AAT deficiency is a rare genetic disorder associated with the development of liver and lung diseases [30].

In summary, the results of this study identified several proteins associated with hypocalcemia, although this evidence could also explain other diseases that often occur during and after calving caused by hypocalcemia, including fatty liver (AAT, C4) and endometritis (HP, C4).

In this study, 2D-DIGE/MALDI-TOF-MS, bioinformatics, and western blotting were performed to distinguish healthy cows from those with CH and SH. A total of eight differentially expressed proteins were identified among the three groups. Due to its specific expression, A2M was selected for validation, and results show that it was as an early detection index. The possible pathogenesis of CH caused by GC and APOA4 was speculated. Further evaluation of these findings is required for clinical translation.