Molecular form and concentration of serum α2-macroglobulin in diabetes

α2-Macroglobulin is a highly abundant serum protein involved in the development of atherosclerosis and cardiac hypertrophy. However, its circulating molecular form and exact concentrations in human health/diseases are not known. Blue native-polyacrylamide gel electrophoresis of human serum was used to confirm the native conformation of α2-macroglobulin. We created an enzyme-linked immunosorbent assay suitable for quantifying its circulating molecular form and undertook a cross-sectional study to measure its serum levels in 248 patients with diabetes mellitus and 59 healthy volunteers. The predominant circulating molecular form of α2-macroglobulin was the tetramer, whereas its dimer was detectable in patients with high serum levels of α2-macroglobulin. The serum α2-macroglobulin concentration was not associated with glycated hemoglobin or any other glycemic variable as evaluated from 48-h continuous glucose monitoring, but showed close correlation with left ventricular posterior wall thickness, carotid artery intima-media thickness, urinary albumin:creatinine ratio (ACR) and brachial–ankle pulse wave velocity (baPWV). Multivariate analysis revealed only the ACR and baPWV to be independent variables influencing serum levels of α2-macroglobulin. Thus, an increased ACR and baPWV are associated with higher serum concentrations of α2-macroglobulin, and the latter may contribute to the mechanism by which albuminuria increases the risk of developing cardiovascular diseases.

In humans, α 2 -macroglobulin is the largest non-immunoglobulin molecule among the highly abundant proteins in the peripheral blood circulation. α 2 -macroglobulin is synthesized mainly in the liver as a result of coordination between endothelial cells and hepatocytes 1 . α 2 -macroglobulin can inhibit a broad spectrum of serine, threonine, and metalloproteases as well as pro-inflammatory cytokines 2 . It can also induce transcriptional activation of various genes essential for the proliferation/hypertrophy of cells, oncogenesis and atherosclerosis 3 .
Studies using stored pooled human plasma have suggested that two identical α 2 -macroglobulin subunits of size 182 kDa are disulfide-bonded to form dimers, which interact non-covalently to yield a tetrameric structure 4,5 . In biologic fluids, α 2 -macroglobulin tetramers appear to be predominant, but may undergo structural changes during manipulation and preservation 6 . Thus, the exact circulating molecular forms of α 2 -macroglobulin associated with health and diseases have yet to be elucidated.
A 182-kDa protein termed "cardiac isoform of α 2 -macroglobulin" was shown to induce expression of muscle-specific genes associated with the pressure-overloaded heart and to cause cardiac hypertrophy directly [7][8][9][10] . This putative "isoform" was claimed to be a key molecule inducing myocardial infarction and cardiac hypertrophy, especially in people with diabetes mellitus (DM) on the basis of serum level measurements [11][12][13] . Recently, we and others demonstrated using mass spectrometry that the corresponding proteins in rats and humans were indistinguishable from α 2 -macroglobulin molecule 9,14 .
Microalbuminuria has been measured to assess renal risk in early diabetic nephropathy, but is also a strong predictor of cardiovascular diseases [28][29][30] . People with type-1 diabetes mellitus (T1DM) and nephropathy carry a tenfold greater risk of adverse cardiovascular outcome compared with those without these disorders 31 . A slightly increased urinary albumin:creatinine ratio (ACR) carries an increased atherosclerotic risk 32 , which is reduced by the regression of albuminuria 28,33 . The latter is also associated with premature death [34][35][36][37] which, in those with T1DM, is due mainly to cardiovascular factors 37 . However, the mechanisms relating albuminuria with an increased cardiovascular risk are not known.
We established a new ELISA to quantify circulating levels of the molecular form of human α 2 -macroglobulin using an antibody that can recognize its monomers, dimers and tetramers in human serum. We designed a cross-sectional study to ascertain if this important molecule is related to diabetic microvascular complications and the risk factors of cardiac/atherosclerotic diseases.

Results
To study the native conformation of circulating α 2 -macroglobulin molecules, we incubated purified α 2 -macroglobulin protein and fresh human serum treated with and without dithiothreitol (DTT) and subjected them to sodium dodecyl sulfate (SDS)-PAGE and western blotting. A single band representing α 2 -macroglobulin-like immunoreactivity corresponding to the size of standard α 2 -macroglobulin protein was detected on western blots from healthy human serum. Treatment of serum and an α 2 -macroglobulin standard with increasing concentrations of DTT (1, 10 and 25 mM) reduced the standard α 2 -macroglobulin protein to form monomers of size 182 kDa (Fig. 1a). Blue native (BN)-PAGE of serum and standard α 2 -macroglobulin protein and subsequent immunoblotting revealed a single band corresponding to an α 2 -macroglobulin tetramer of size 725 kDa in untreated samples, whereas treatment of serum and standard α 2 -macroglobulin with DTT reduced the protein into monomers of size 182 kDa (Fig. 1b). A band corresponding to an α 2 -macroglobulin dimer was very faint or barely visible in the lane for a sample of healthy serum or in the lane for standard tetrameric α 2 -macroglobulin. However, in some untreated sera of patients with DM who showed increased serum levels of α 2 -macroglobulin, dimers of α 2 -macroglobulin were visible (Fig. 1c). These results suggested that circulating α 2 -macroglobulin molecules in healthy humans consisted of tetramers that could be monomerized completely by reducing intermolecular disulfide bonds, and that dimers could be present in some patients with DM.
We established an ELISA using anti-human monoclonal mouse α 2 -macroglobulin IgG, which has been confirmed to recognize monomeric, dimeric and tetrameric α 2 -macroglobulin molecules. A curve prepared with tetrameric α 2 -macroglobulin standards (0.0-0.2 µg/mL) showed parallelism with those of serially diluted serum samples in the ELISA of α 2 -macroglobulin (Fig. 2a). When serial dilutions of serum were applied to the standard curve, a very good correlation was obtained (R 2 = 0.9646). The concentration of unknown samples was determined by interpolation, which is reliant on a standard curve being generated appropriately. The intra-assay coefficient of variation (CV) of serum samples was 2.3% (n = 8) and inter-assay CV was 13.8% (n = 14). These results suggested that our ELISA was suitable for quantifying serum concentrations of α 2 -macroglobulin in humans. Using serum samples from 40 DM patients, we compared the results obtained with the current ELISA and those determined using quantitative western blotting 14 (in which α 2 -macroglobulin molecules in serum should undergo reduction/monomerization during SDS-PAGE). Data for the ELISA of α 2 -macroglobulin showed close correlation with the total concentration of monomerized α 2 -macroglobulin (R 2 = 0.575, p < 0.0001) (Fig. 2b).
We measured serum concentrations of α 2 -macroglobulin in 59 healthy volunteers and 248 patients with DM (Table 1). Patients with DM showed significantly higher serum levels of α 2 -macroglobulin (310.4 ± 114.2 mg/dL) than healthy volunteers (272.9 ± 76.5 mg/dL) (p = 0.0169) (Fig. 3). When the data of all 307 subjects were analyzed together, serum levels of α 2 -macroglobulin did not correlate with HbA 1c level. To confirm no association between serum α 2 -macroglobulin with glycaemic control or glycaemic fluctuation in DM, we analyzed the results of 57 individuals who underwent 48-h CGM at the time of measuring serum levels of α 2 -macroglobulin. Neither the mean glucose level nor data on glycemic fluctuation (standard deviation, CV, mean amplitude of glycemic excursions) showed a correlation with serum levels of α 2 -macroglobulin. These results negate the current belief that serum concentrations of α 2 -macroglobulin reflect glycemic control.
Pearson's univariate correlation analysis revealed a positive correlation of serum levels of α 2 -macroglobulin with the ACR, baPWV, maximum intima media thickness (maxIMT), left ventricular posterior wall thickness  www.nature.com/scientificreports www.nature.com/scientificreports/ (LVPWth) and interventricular septum thickness (IVSth), but not with HbA 1c levels ( Table 2). Serum levels of α 2 -macroglobulin in DM patients with retinopathy were higher than those without ( Table 2). Of 248 patients with diabetes, 51 had cardiovascular disease as evidenced by coronary angiography, CT of coronary arteries  www.nature.com/scientificreports www.nature.com/scientificreports/ and/or cardiac ultrasonography. Least square multivariate analyses revealed that only the ACR and baPWV were associated independently with higher serum levels of α 2 -macroglobulin, and that age, sex, eGFR or the presence of retinopathy or cardiovascular disease were not independent variables influencing serum levels of α 2 -macroglobulin (Table 3). Also, serum levels of α 2 -macroglobulin in DM patients with normoalbuminuria (281.0 ± 95.1 mg/dL, n = 131) were not significantly higher than those of healthy volunteers (272.9 ± 76.5 mg/ dL, n = 59). Taken together, these results suggested that the ACR and baPWV had an impact on higher serum concentrations of α 2 -macroglobulin.

Discussion
We developed a highly sensitive ELISA to recognize human α 2 -macroglobulin molecules in serum specifically. The limit of detection was 0.5 ng/mL, and the standard curves obtained by diluting the serum from 1:40,000 to 1:320,000 showed parallelism with the standard purified tetramer protein of α 2 -macroglobulin. Inter-assay and intra-assay CVs were small, and the results were stably reproducible. α 2 -macroglobulin levels in human serum have ranged by six orders of magnitude, and reported to be 0.5-2.5 µg/dL when measured using commercial ELISA kits 22,23,25 . Albumin and Ig molecules make up the greatest proportion of plasma proteins, but α 2 -macroglobulin is regarded to be one of top-12 abundant proteins. In the present study, the serum concentration of α 2 -macroglobulin in healthy volunteers was 272.9 ± 76.5 mg/dL, which was quite reasonable.
Considering the well-described potent biologic activities of α 2 -macroglobulin 10,38-41 and its predominance over other bioactive substances in plasma, significant fluctuations of this important molecule in the peripheral blood circulation could affect the onset and/or progression of human diseases. Our system is the first ELISA suitable for quantifying α 2 -macroglobulin levels in human serum accurately.
BN-PAGE revealed that the α 2 -macroglobulin-like immunoreactivity in human serum migrated to a position corresponding to that of the purified tetramer protein of α 2 -macroglobulin of size 725 kDa. Reduction of the disulfide bonds of purified α 2 -macroglobulin protein and α 2 -macroglobulin molecules in human serum by DTT treatment followed by BN-PAGE and western blotting revealed that α 2 -macroglobulin molecules in serum were separated completely into monomers of size 182 kDa even in the absence of SDS. Furthermore, neither serum nor purified α 2 -macroglobulin protein were monomerized following SDS-PAGE due to non-covalent bonding, whereas both were monomerized completely with reduction of disulfide bonds after DTT treatment. These data argue against the current belief based on studies using stored human plasma that two disulfide-bonded α 2 -macroglobulin dimers are non-covalently bonded to form tetramers in the peripheral blood circulation 2,4,5,42 .
In some patients with DM whose serum α 2 -macroglobulin levels were increased markedly, dimeric components were observed in addition to tetramers according to BN-PAGE. Thus, dimers of α 2 -macroglobulin may be formed in human blood if α 2 -macroglobulin synthesis is enhanced. Our results demonstrate that the α 2 -macroglobulin antibody that we used had cross-reactivity with monomers, dimers and tetramers, which could be present in human serum. We compared the results of semiquantitative measurement of serum α 2 -macroglobulin monomerized after DTT treatment and SDS-PAGE with measurements by ELISA using 40 samples of human serum, and found that both were highly correlated (Fig. 2b). We concluded that our ELISA reflected the circulating components of human α 2 -macroglobulin molecules.
We also found significantly high serum levels of α 2 -macroglobulin in DM patients compared with healthy volunteers. Several studies have reported a positive correlation between the HbA 1c level and α 2 -macroglobulin level in serum 20-22 , but we did not. To negate the possibility that blood glucose levels in DM affect serum concentrations of α 2 -macroglobulin, we obtained the exact glycemic profiles of 57 individuals through 48-h CGM. Neither the mean blood glucose levels of the entire 48-h recordings nor variables of glycemic fluctuation (standard deviation, CV) correlated with serum levels of α 2 -macroglobulin. Univariate regression analysis of 248 DM patients revealed that the α 2 -macroglobulin level correlated with the ACR, baPWV, maxIMT, LVPWth, IVSth, but not with the HbA 1c level. Serum levels of α 2 -macroglobulin were significantly higher in DM patients with retinopathy than those without retinopathy. Among these variables, however, multiple regression analysis with the α 2 -macroglobulin level as the objective variable revealed only the ACR and baPWV to be significant independent variables affecting the serum level of α 2 -macroglobulin. The association of the urinary albumin concentration with an increased level of this important molecule may be linked to the mechanism by which albuminuria increases the risk of developing cardiac and atherosclerotic diseases.
Albuminuria was originally identified as an incipient marker of diabetic nephropathy 43,44 , but is now well documented as an independent predictor of cardiovascular morbidity and mortality not only in T1DM and www.nature.com/scientificreports www.nature.com/scientificreports/ T2DM 28,37,45,46 , but across varied non-DM populations 29,36,47,48 . Accumulating evidence suggests that proteinuria is a stronger predictor of cardiovascular disease than traditional risk factors such as cholesterol and blood pressure 30,49 , whereas albuminuria is associated with an increased risk starting well below the cutoff for microalbuminuria 29,48,49 . However, the exact pathophysiologic conditions responsible for the overwhelming cardiovascular consequences induced specifically by trace albuminuria, but not in patients without albuminuria, have remained unknown to date.
Several mechanisms have been proposed to contribute to the cardiovascular consequences of albuminuria, including inflammation, endothelial dysfunction, thrombogenic factors, and insulin resistance [50][51][52][53] . In nephrotic syndrome, hepatic synthesis of the high-molecular-weight protein α 2 -macroglobulin has been shown to be enhanced significantly to replace lost liver-derived proteins in experimental animals 54 and humans 55 , resulting in a net increase in its serum levels 54 . The present study demonstrated a close correlation between serum levels of α 2 -macroglobulin and the ACR, and that the former start to rise when a trace amount of albumin is excreted. Considering the magnitude of increased post-transcriptional α 2 -macroglobulin synthesis in the liver in response to urinary protein excretion [54][55][56] , it is reasonable to assume that circulating α 2 -macroglobulin levels may be increased far more profoundly by urinary albumin than by other factors such as sex, age, glycemic control or diabetic retinopathy. A proposed mechanistic link relating albuminuria to increased serum level of α 2 -macroglobulin and cardiovascular disease in diabetes is depicted in Fig. 4.
Considering the limited number of individuals employed in this cross-sectional study, whether an increased serum level of α 2 -macroglobulin in diabetes mellitus could serve as a mediator of albuminuria that, ultimately, results in cardiovascular diseases, is not known. Also, whether preventive therapy can reverse albuminuria and reduce the serum level of α 2 -macroglobulin is not known. Further studies to ascertain if measurement of the serum level of α 2 -macroglobulin could help to identify patients at a high risk of cardiovascular disease should be carried out.
In conclusion, using a sensitive and specific ELISA, we identified, among other laboratory parameters, a close association between serum levels of α 2 -macroglobulin with ACR and baPWV. Measurement of serum levels of α 2 -macroglobulin could help to identify patients at high risk of cardiovascular diseases.

Methods
Study participants. We enrolled 248 patients with DM (39 with T1DM, 204 with T2DM, and 5 with gestational DM) and 59 healthy volunteers at Kitasato University Medical Center or Kitasato University Hospital. Excluded from the analyses were patients with infectious diseases, malignancy, end-stage renal failure, autoimmune thyroid diseases and those receiving corticosteroids. We obtained information on the medical and life histories of all participants.
Patients underwent routine evaluation of systemic diseases covered by the universal health coverage system in Japan 57,58 . These included electrocardiography (ECG), radiography (chest and abdomen), ultrasonography (neck and/or abdomen), urinalysis, complete blood count, biochemical analyses of serum, as well as tests for levels of free thyroxine, free triiodothyronine, thyroid-stimulating hormone (TSH) and thyroid antibodies. DM patients underwent ophthalmologic and neurologic tests, the anti-glutamic acid decarboxylase antibody test as well as measurements of glycated albumin and/or glycated hemoglobin (HbA 1c ), fasting serum insulin, and urinary ACR. Cardiovascular diseases were diagnosed by ECG, exercise electrocardiography, ankle-brachial pressure index, brachial-ankle pulse wave velocity (baPWV), cardiac catheterization/angiography, and/or coronary computed tomography. For healthy volunteers, the blood-test data at the most recent medical checkup were used to confirm the absence of other systemic diseases and as analytical data. collection of serum samples. Blood samples from study participants were collected into vacutainers.
Serum was separated immediately in a refrigerated centrifuge and stored in aliquots at −30 °C until processing. eLiSA of α 2 -macroglobulin. All measurements were carried out in triplicate. One-hundred microliters of human α 2 -macroglobulin protein diluted to 0.5-80 ng/mL and serum samples diluted to 1:160,000 with phosphate-buffered saline (PBS) were incubated overnight at 4 °C in a 96-well microplate. Plates were washed thrice and blocked with 200 µL/well of PBS containing 0.05% Tween ® 20 (PBS-T) and 5% (w/v) skimmed milk/ Tris-buffered saline (TBS) for 2 h at room temperature. After washing thrice with PBS-T (200 µL/well), plates were incubated overnight with 100 µL of anti-human monoclonal mouse α 2 -macroglobulin at 1:5,000 dilution. Then, plates were washed thrice with PBS-T (200 µL/well) and reacted with 1:10,000 goat anti-mouse IgG (H + L)-HRP conjugate (100 µL/well) and allowed to stand at room temperature for 1 h under light-shielding. After washing four times with PBS-T (200 µL/well), o-phenylenediamine/0.05 M phosphate-citrate buffer (100 µL/well) was added for 30 min at room temperature under light-shielding, then overlaid with 3 N HCl (100 µL/well) to terminate the reaction. Absorbance of the plate was measured at 490 nm using a microplate reader (iMark ™ ; Bio-Rad Laboratories). polyacrylamide gel electrophoresis (pAGe) followed by western blotting. SDS-PAGE was performed using purified human α 2 -macroglobulin protein and human serum essentially as described 61,62 except for the following modifications. Ten-microliters of fresh human serum diluted to 1:80 and 0.5 µg of purified α 2 -macroglobulin protein pretreated with or without DTT were subjected to 4-20% gradient gel electrophoresis using Mini-PROTEAN ® TGX ™ and subsequent immunoblotting using human α 2 -macroglobulin antibody.

Materials.
Blue native PAGE was performed as described 63 except for the following modifications. Native PAGE ™ Sample Buffer was used to dissolve 0.1 µg of standard α 2 -macroglobulin and to dilute 0.025 µL of serum. After pretreatment with or without 25 mM of DTT, 10 µL of diluted samples was loaded on a 5-20% gradient Perfect NT Gel using Native PAGE Running Buffer as a positive electrode and Dark Blue Cathode Buffer as a negative electrode according to manufacturer instructions. Native Mark ™ Unstained Protein Standard was used as a marker of molecular size. After migrating samples at 150 V to one-third of the entire gel, Dark Blue Cathode Buffer in the negative electrode was replaced with Light Blue Cathode Buffer to complete the remaining electrophoresis.
After electrophoresis, the gel was photographed, transferred to a PVDF membrane (Immune-Blot ® ) at 75 V for 120 min, incubated with 8% acetic acid for 15 min, then with sterilized water for 5 min, and air-dried to fix the protein. The PVDF membrane was then immersed in MeOH to remove Coomassie brilliant blue G-250 dye. Then, it was washed with sterilized water, blocked at 4 °C overnight with Blocking One and, after washing thrice with TBS-T, incubated overnight with human α 2 -macroglobulin antibody (1:5000 dilution) at 4 °C. After incubation with goat anti-mouse IgG (H + L)-HRP Conjugate (1:10000 dilution) for 1 h, protein bands were detected using ECL and photographed and analyzed with ImageQuant LAS 4000 (GE Healthcare).
Quantitative western blotting to measure levels of monomerized α 2 -macroglobulin. Serum samples from 40 consecutive DM patients were diluted 40-fold with binding buffer (50 mM DTT, 0.5% SDS, 10% glycerol/1 M Tris-HCl) and subjected to SDS-PAGE with a protein marker (Precision Plus Protein ™ Dual Color Standards) using a 5% Perfect NT Gel and a Perfect NT Gel System SDS-PAGE Running Buffer at 120 V for 60-90 min.
Proteins in the gels were then transferred to a PVDF membrane (Sequi-Blot ™ ) at 60 mA for 60 min and incubated overnight at 4 °C with Blocking One. After washing thrice with TBS-T, the PVDF membrane was incubated with human α 2 -macroglobulin antibody (1:3000 dilution) for 90 min, and then with goat anti-mouse IgG(H + L)-HRP Conjugate (1:5000 dilution) for 1 h at room temperature. The PVDF membrane was visualized using Brown Stain. The intensity of each band was quantified using the ImageQuant TL program (GE Healthcare). All densitometric data were converted to relative monomer concentrations using twofold serially diluted samples of control protein.
Statistical analyses. Data are the mean ± SD unless indicated otherwise. Statistical analyses were carried out using JMP v5 (SAS institute, Cary, NC, USA) and Prism 5 (GraphPad Software, La Jolla, CA, USA). Differences between groups were examined for significance using the unpaired t-test. Correlation analysis was