A New Strategy for Early Diagnosis of Type 2 Diabetes by Standard-Free, Label-Free LC-MS/MS Quantification of Glycated Peptides

Figure 1 shows a complete workflow of the experiment. First, ¹⁸O-labeling, a convenient proteomics technique (36), was used to screen glucose-sensitive and glucose-insensitive peptides within HSA characteristic peptides. Next, a glucose-insensitive peptide was selected as the internal standard, and glucose-sensitive peptides underwent further screening as biomarker candidates. With the glucose-insensitive peptide as the internal standard, a standard-free quantitative method was then applied in the discovery of candidate peptide biomarkers for T2DM in a clinical plasma sample set. This sample set included three groups that were classified according to the fasting plasma glucose (FPG) levels and 2-h postprandial glucose (2hPG) of the individuals: T2DM group: FPG ≥7.0 mmol/L or 2hPG ≥11.1 mmol/L; impaired glucose tolerance (IGT) group: FPG 11.1 mmol/L and >2hPG >7.8 mmol/L; and normal glucose tolerance (NGT) group: FPG <7.0 mmol/L and 2hPG ≤7.8 mmol/L. Potential biomarker peptides were verified to show significant differences among groups, and the receiver operating characteristic (ROC) analysis showed excellent sensitivity and specificity as well.

Sequencing-grade modified trypsin was purchased from Promega (Madison, WI), H₂¹⁸O (97%) was purchased from the China Isotope Company (Beijing, China), high-performance LC (HPLC)-grade acetonitrile and formic acid (FA) were purchased from Fisher Scientific (Fair Lawn, NJ, and Edmonton, AB, Canada), and inorganic salts for making buffers (NH₄HCO₃, KH₂PO₄, and K₂HPO₄) were purchased from Beijing Chemical Company (Beijing, China). All other chemicals (dithiothreitol [DTT], iodoacetamide, d-glucose, urea, and xylene) were purchased from Sigma-Aldrich (St. Louis, MO). Water was prepared by a Milli-Q system (Millipore, Billerica, MA). A reverse-phase VYDAC LC column (C¹⁸, 2.1 × 150 mm, 300 Å) was purchased from Grace (Deerfield, IL).

An Agilent 1100 series HPLC system was coupled with an Agilent electrospray ionization–time of flight (ESI-TOF) MS (6210) with a VYDAC C¹⁸ column (300 Å, 2.1 × 150 mm) at a flow rate of 0.2 mL/min for quantitative analysis. The sample injection amount was 10 μg. The mobile phase consisted of buffer A (0.1% FA in H₂O) and buffer B (0.1% FA in acetonitrile). The gradient elution for buffer B was as follows: 3% from 0 to 5 min, 40% from 5 to 35 min, 40–95% from 35 to 40 min, 95% from 40 to 43 min, and 95–3% from 43 to 45 min, with a final flow from 45 to 55 min. Nitrogen was used as the drying and nebulizing gas (flow rate, 10 L/min; pressure, 35 ψ; temperature, 350°C). The MS ionization mode was positive, and the capillary voltage was set at −3.5 kV. A mass window of 300–1,800 amu was used for the TOF MS instrument.

The same Agilent 1100 series HPLC system was coupled with an Agilent MSD trap for qualitative analysis. The same chromatographic and ESI conditions were used. MS and MS/MS analyses were performed on favorable fragmentations to double- and triple-charged ions. Ion trap MS also had a mass window of 300–1,800 amu. The Mascot database (Matrix Science) was used to identify peptide sequences associated with each MS/MS spectrum, and the human protein database SwissProt was used for protein identification.

In vitro glycation of pure HSA was done with glucose according to a published procedure (37). Briefly, unmodified HSA (40 mg/mL in 50 mmol/L KH₂PO₄-K₂HPO₄ buffer [pH 7.4], containing 0.1 mmol/L toluene as a bacteriostat) was incubated with d-glucose at four different HSA-to-glucose molar ratios (1:10, 1:41.5, 1:83, and 1:415) at 37°C for 10, 20, or 30 days, and then lyophilized. These four concentrations were calculated by the HSA-to-glucose molar ratio to give a physiological glucose condition (∼12.05% site-occupancy), theoretical 50% site-occupancy, theoretical 100% site-occupancy, and fivefold glucose excess, respectively. An equal volume of HSA incubated under the same conditions without glucose was used as a control.

Two hundred micrograms of HSA (both glycated and control HSA) was denatured and reduced by a 20 μL solution containing 8 mol/L urea and 10 mmol/L DTT in 50 mmol/L NH₄HCO₃ buffer (pH 8.3) at 37°C for 4 h. Alkylation was performed in a 50 mmol/L iodoacetamide solution at room temperature for 1 h in the dark. After alkylation, the sample was diluted using a 50 mmol/L NH₄HCO₃ buffer to give a final urea concentration of 1 mol/L. Tryptic digestion was then performed at a protein-to-trypsin concentration ratio of 50:1 (w/w) for 20 h at 37°C. All digested peptide samples were lyophilized to complete dryness and then used for the labeling reaction.

Peptide ¹⁸O-labeling was done under optimized conditions (38). Peptide samples were dissolved in 50 mmol/L KH₂PO₄-K₂HPO₄ buffer, pH ∼5. This step was followed by a second lyophilization to completely dry the sample. Finally, the control HSA peptides and the glycated HSA peptides were labeled in H₂¹⁸O and H₂¹⁶O, respectively, with 1.6 μg trypsin. After labeling at 37°C for 20 h, residual trypsin activity was quenched by boiling the sample for 10 min and adding 5% FA (v/v). All samples were centrifuged at 17,000g for 15 min before HPLC-MS analysis.

Clinical samples for the T2DM, IGT, and NGT groups were provided by the Beijing Institute of Technology Hospital, Beijing, China. During 2009–2010, upon request, 389 volunteers provided blood samples with questionnaires about their characteristics, including basic characteristics (age, sex, and BMI), family history of diabetes, and symptoms of early diabetes. Biochemical examinations of their blood samples were also performed and recorded. This clinical study was conducted in accordance with the requirements of medical ethics.

The participants who provided blood samples were classified into 3 groups: T2DM, IGT, and NGT, as described in the research design and methods. Table 1 lists the basic characteristics of the study participants. As expected, patients classified into the T2DM group had a higher BMI and a higher probability of having a family history of diabetes than patients classified into the NGT group. Furthermore, patients in the T2DM group were more inclined to feel that they had symptoms of diabetes and its complications than patients in the NGT group. Biochemical examinations showed that patients in the T2DM group had significantly higher glucose, cholesterol, alanine aminotransferase, triglyceride, and LDL levels than patients in the NGT group.

In an environment with high glucose concentrations, glycation will occur on certain sites of HSA (i.e., lysines, arginines, and the N-terminal free amine), which are called glycation-sensitive sites. Peptides with glucose-sensitive sites were potential glucose-sensitive peptides, whereas peptides without glucose-sensitive sites were potential glucose-insensitive peptides. With increasing degrees of glycation, three types of peptides were expected after HSA was digested to peptides: 1) glucose-insensitive peptides, 2) glucose-modified glucose-sensitive peptides, and 3) unmodified glucose-sensitive peptides. In this study, type 1 and type 3 peptides were quantitatively analyzed and used for diagnosis of T2DM.

The ¹⁸O-labeling proteomics technique was first used in in vitro samples to search for type 1 and type 3 peptides. Within glucose-insensitive peptides (summarized in Supplementary Table 1), the peptide AAFTECCQAADKAACLLPK (m/z = 977.4 and retention time of 28.7 min) was chosen as the internal standard peptide. As shown in Fig. 2, the peak area ratio (¹⁶O-to-¹⁸O) of peptide AAFTECCQAADKAACLLPK was close to 1 (0.961 ± 0.077; Fig. 2A) in 72 MS runs of 36 samples, where 4 different glucose concentrations were used for incubations of 10, 20, or 30 days. After glycated HSA samples were digested with trypsin, 57 unmodified peptide sequences with usable sequence coverage of 83% in total were matched in the database search query by exporting their MS/MS data to Mascot (Supplementary Table 2). Similar results have been reported (36). As an example, Fig. 2B shows the concentration variation of one of these glucose-sensitive peptides, FKDLGEENFK (m/z = 614.8), with a retention time of 25.5 min.

Glycation is dependent on the concentration of glucose and has an accumulation effect. As a result, the concentration of a glucose-sensitive peptide in glycated HSA samples will be a function of the glucose concentration and the incubation duration. Supplementary Table 3 reports the relative intensity (peak area ratio ¹⁶O-to-¹⁸O of the LC-TOF MS data) variations of selected peptide ions (usually detected in each MS run) in glycated HSA samples. By calculating their statistical correlations with respect to glucose concentration and incubation duration, eight peptides were provisionally recognized as unmodified glucose-sensitive peptides (or type 3 peptides) for HSA (summarized in Table 2) because they exhibited a steady decrease in peak area ratio (significant difference) against increasing glucose concentration and incubation duration. The decreasing ratio among peak area ratios (¹⁶O-to-¹⁸O) of these glucose-sensitive peptides with different incubation durations is presented in Supplementary Table 4, and the peak area ratios of three peptides with respect to glucose concentration and incubation times is shown in Fig. 3. Detailed information describing statistical correlation values (P values) with respect to glucose concentrations is given in Supplementary Table 5.

Finally, when the standard-free and label-free method was applied to the clinical samples, the amount of three peptide ions (m/z = 537.7, 614.8, and 820.5) of the eight potential biomarkers showed significant differences among the T2DM, IGT, and NGT groups. Figure 4 shows the box plot of these three peptides for the T2DM, IGT, and NGT groups, and Supplementary Table 6 provides the corresponding P values. These three peptides (LDELRDEGK, FKDLGEENFK, and KVPQVSTPTLVEVSR) showed significant decreases in their concentrations in the T2DM group compared with the IGT group. Furthermore, there were significant differences (P < 0.01) between both NGT/IGT and IGT/T2DM groups for peptides FKDLGEENFK and KVPQVSTPTLVEVSR, indicating that these three peptides could be used as potential biomarkers for the early diagnosis of T2DM.

In addition, Supplementary Fig. 1 shows the ROC analysis for early diagnosis of T2DM. Peptides FKDLGEENFK and KVPQVSTPTLVEVSR showed excellent sensitivities and specificities between the NGT/IGT groups (Supplementary Fig. 1A) and the NGT/IGT and T2DM groups (Supplementary Fig. 1B). The parameters of the ROC analysis (sensitivity, specificity, area under the curve, and cutoff point) are given in Supplementary Table 7.

In this study, we developed a novel method (outlined in Fig. 1) for the detection of glycated peptides as a potential technique for the early diagnosis of T2DM. During the process of ¹⁸O-labeling, control HSA was prepared in the presence of an ¹⁸O-enriched PBS buffer, whereas glycated HSA was prepared in an ¹⁶O-enriched buffer. After incubation and digestion, the corresponding glycated HSA peptide sample was mixed with the control sample in a 1:1 ratio (v/v), and the mixtures were injected into HPLC/ESI-TOF MS for quantitative analysis. This allowed for the identification of an internal standard peptide and three promising glucose-sensitive peptides, which may have potential application in the early diagnosis of T2DM. Then, following the identification of an internal standard peptide and putative biomarker peptides, a standard-free and label-free method was developed and applied to clinical samples. Thus, instead of using the ¹⁸O-labeling technique for final quantification and validation of the data, we performed quantification of the putative biomarker peptides by calculating the peak area ratios of the putative biomarker peptide over the internal standard peptide (m/z = 977.4).

Peptide AAFTECCQAADKAACLLPK (m/z = 977.4 and retention time of 28.7 min) was chosen as the internal standard peptide because its concentration was independent of the concentration of glucose solution and the incubation duration. In addition, this peptide had a stable and strong enough signal (×10⁵ in vitro and ×10⁴ in vivo) in the TOF mass spectrum, which would ensure analysis accuracy. Internal standards in quantitative analysis usually have to be exogenously added. However, there are many limits to the selection of an internal standard; for example, the internal standard substance must be completely dissolved in the sample, no interactions must occur between the internal standard and the component to be measured, and the retention time of the two peaks should be close in the chromatogram. We solved this problem in our study by using the above-mentioned peptide, which was cleaved from HSA itself.

In addition, as shown in Supplementary Fig. 2, the glucose-insensitive peptide used as the internal standard (AAFTECCQAADKAACLLPK) was located at the center of the HSA protein. Thus, the location of this peptide within the HSA protein may contribute to its glucose-insensitivity, because it would be difficult for small molecules (such as glucose) to attack the modification site.

As reported in Supplementary Table 1, three glucose-insensitive peptides were obtained from the in vitro study. That the three peptides are insensitive to glucose in the clinical samples (the T2DM group, 73 cases) was also confirmed. Therefore, the peptide (AAFTECCQAADKAACLLPK) was used as an internal standard for the clinical study.

HSA has no unique peptide sequence, and today we know that there are hundreds of different proteins in the human plasma. HSA characteristic peptides are the peptides that are present in HSA but are not the highest abundance proteins in human plasma. However, HSA itself is the highest abundance protein in human plasma, and proteins that have similar amino acid sequences to HSA are all albumins. Importantly, studies have demonstrated that no other albumin protein is among the 137 highest abundance proteins in human plasma (22).

In conclusion, this report describes the development of a standard-free and label-free LC-MS/MS method for the discovery of putative biomarkers for T2DM. In vitro glycation of the most abundant protein, HSA, in human blood plasma was monitored through quantitative analysis of HSA characteristic peptides using ¹⁸O-labeling techniques. Verified by clinical samples, three peptides (FKDLGEENFK, LDELRDEGK, and KVPQVSTPTLVEVSR) exhibited significant differences between T2DM and NGT groups, and two of these peptides (FKDLGEENFK and KVPQVSTPTLVEVSR) also showed significant differences between both NGT/IGT and IGT/T2DM groups. Our results indicate that these peptides could be putative biomarkers for the early diagnosis of T2DM.

This research was financially supported by the Ministry of Science and Technology of the People’s Republic of China (grant numbers 2012YQ040140 and 2012CB910603) and the National Natural Science Foundation of China (grant number 21205005).

No potential conflicts of interest relevant to this article were reported.

M.Z. collected data and prepared the manuscript. W.X. was involved in data integration and analysis and also in preparing the manuscript. Y.D. conceived the idea and designed the study. Y.D. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank David R. Raisor and Editage for providing the English language editing.