Preliminary observational study of metabonomics in patients with early and late-onset type 2 diabetes mellitus based on UPLC-Q-TOF/MS

Non-targeted metabonomic techniques were used to explore changes in metabolic profiles of patients with early onset and late onset T2DM. Newly diagnosed early onset T2DM (EarT2DM) and late onset T2DM (LatT2DM) patients were recruited, and the matched age, sex, and low-risk population of diabetes mellitus were selected as the control group. 117 adults were recruited in the study, including 21 in EarT2DM group with 25 in corresponding control group (heaCG1), and 48 in LatT2DM group with 23 in corresponding control group (heaCG2). There were 15 relatively distinctive metabolic variants in EarT2DM group and 10 distinctive metabolic variants in LatT2DM group. The same changing pathways mainly involved protein, aminoacyl-tRNA biosynthesis, fatty acid biosynthesis, taurine metabolism, arginine biosynthesis, lysosome and mTOR signaling pathway. The independent disturbed pathways in EarT2DM included branched chain amino acids, alanine, aspartate and glutamate metabolism. The independent disturbed pathways in LatT2DM involved linoleic acid metabolism, biosynthesis of unsaturated fatty acids, arginine, proline metabolism and FoxO signaling pathway. T2DM patients at different diagnosed ages may have different metabolite profiles. These metabolic differences need to be further verified.


Comparison of metabonomics results between EarT2DM group and heaCG1.
The results of the PCA and OPLS-DA in the EarT2DM and control groups were shown in Fig. 2. The PCA results showed R2X > 0.5.The two OPLS-DA models were evaluated using the R2Y and Q2 parameters.The values of R2Y were 0.981 and Q2 0.969 in the positive mode, and R2Y 0.980 and Q2 0.924 in the negative mode.The samples of the EarT2DM group were separated intelligibly from the control samples in both the positive and negative modes, and the obvious separation suggested that there was pronounced metabolic differences at an overview level.
The differential metabolites between the groups were filtered using the multivariate and univariate statistical significance criteria (VIP > 1 and p < 0.05).Figure 3 showed the hierarchical clustering results of metabolites with significant differences in the EarT2DM group.In total, 66 identified metabolites showed significant differences between the two groups, including 23 species in positive mode and 43 species in negative mode.

Comparison of metabonomics results between LatT2DM group and heaCG2.
The results of the PCA and OPLS-DA in the LatT2DM and control groups were shown in Fig. 4. The PCA results showed Table 1.Comparison of baseline data between the recruited groups.EarT2DM Early-onset T2DM, HeaCG1 Healthy Control Group 1, LatT2DM Late-onset T2DM, HeaCG2 Healthy Control Group 2, Tg triglyceride, Tc total cholesterol, HDL-c high-density cholesterol, LDL-c low-density cholesterol, BMI body mass index.

P-value LatT2DM vs HeaCG2
Age (years) R2X > 0.5.The two OPLS-DA models were evaluated using the R2Y and Q2 parameters.The values of R2Y were 0.979 and Q2 0.958 in the positive mode and R2Y 0.981 and Q2 0.937 in the negative mode.The samples of the LatT2DM group were separated intelligibly from the control samples in both positive and negative modes, and the obvious separation suggested that there were pronounced metabolic differences at an overview level.Figure 5 showed the hierarchical clustering results of the metabolites with significant differences in the LatT2DM group.In total, 60 identified metabolites showed significant differences, including 20 species in the positive mode and 40 species in negative mode.

Further analysis of differential metabolites between EarT2DM group and LatT2DM group.
The differentiated between the groups were filtered using multivariate and univariate statistical significance criteria (VIP > 1 and p < 0.05).Compared with the corresponding healthy control groups, 12 metabolites increased and The significantly elevated metabolites included l-leucine, betaine, thioetheramide-PC, l-glutamine, dihydroxyacetone, vitamin E, cholesterol 3-sulfate, 2'-deoxy-d-ribose, l-isoleucine, methylmalonic acid and deoxycholic acid.Metabolites with significantly decreased expression included 1,2,3-benzenetriol, l-carnitine, pristanic acid and embelin.There were 10 relatively distinctive metabolic variants in the LatT2DM group.The elevated metabolites included d-proline, and l-arginine.The decreased metabolites included 3-phenylpropanoic acid, alphalinolenic acid, linoleic acid, capric acid, pentadecanoic acid, 2E-Eicosenoic acid, 11(Z), 14(Z)-Eicosadienoic acid, and heptadecanoic acid.The detailed information on these metabolites was shown in Table 2. KEGG was used to gain further understanding of the metabolic disturbances in the EarT2DM and LatT2DM groups.The significantly perturbed pathways were shown in Fig. 6 [21][22][23] .Compared with HeaCG1, EarT2DM had 22 significantly altered pathways, mainly including protein, fatty acid, amino acid biosynthesis and degradation.LatT2DM had 20 significantly altered pathways, mainly including fatty acid, linoleic acid and galactose metabolism.The same changing pathways mainly involved protein, aminoacyl-tRNA biosynthesis, fatty acid biosynthesis, taurine metabolism, arginine biosynthesis, lysosome and mTOR signaling pathway.
Excluding the same metabolic pathways, the independent disturbed pathways in EarT2DM included branched chain amino acids, alanine, aspartate and glutamate metabolism.The independent disturbed pathways in LatT2DM involved linoleic acid metabolism, biosynthesis of unsaturated fatty acids, arginine, proline metabolism and FoxO signaling pathway.

Discussion
EarT2DM and LatT2DM patients had metabolic spectrum changes.In this study, it was found that compared with the corresponding control group, they had common metabolic spectrum changes, but also had their own relatively specific changes.The same changing pathways mainly involved protein, aminoacyl-tRNA biosynthesis, fatty acid biosynthesis, taurine metabolism, lysosome and mTOR signaling pathway.In this study, they also had relatively specific metabolic changes.Combined with KEGG analysis, the difference of amino acid metabolism was obvious in the metabolic spectrum of the two groups.These pathways mainly Leucine and isoleucine levels in EarT2DM group were higher.Branched chain amino acids, including valine, leucine and isoleucine, are essential amino acids.The study showed that l-leucine and l-isoleucine levels increased in patients with early onset T2DM.A cohort of 2422 patients without diabetes who underwent physical examinations between 1991 and 1995 was studied.Fasting serum samples were taken at baseline for metabonomics.Metabolic changes were found to occur in the early stages of diabetes mellitus.During the 12-year follow-up period, 201 patients developed type 2 diabetes.Isoleucine, leucine, valine, tyrosine and phenylalanine were significantly higher, and the analysis of their metabolites could have important predictive value for new-onset diabetes mellitus 24 .Varieties of amino acids, especially leucine, are important regulators of the mTORC1 signal 25 .The increased level of branched chain amino acids in plasma over a long period of time may lead to the over-activation of mTOR signalling and may lead to early β cell dysfunction and destruction 26 .
Branched chain amino acids can also promote glucose uptake in the liver and skeletal muscle and promote glycogen synthesis through the phosphatidylinositol 3-kinase or protein kinase C pathway 27 .Branched-chain amino acids may also affect glucose and lipid metabolism through the intestinal flora 28 .In a study of the American population, the predictive value of phenylalanine and valine was better than the other three amino acids 24 .Studies have shown that branched chain amino acids are the strongest predictors of the progression and prognosis of metabolic diseases such as diabetes and obesity, which are stronger than those related to lipid metabolism 29 .In the study we found that leucine and isoleucine levels in early onset T2DM patients were significantly higher compared to those in the control group, and this phenomenon was not observed in late-onset T2DM.
l-glutamine (Gln) levels were increased in EarT2DM group.l-glutamine is one of the most abundant amino acids in the human body.It is a conditional essential amino acid that plays a vital role in nitrogen exchange, intermediate metabolism, immunity, and pH homeostasis 30 .Studies have found that plasma Gln is negatively correlated with body mass index, blood pressure, blood triglyceride and insulin levels, and positively correlated with high-density lipoprotein 31 .Gln circulation pathway may stimulate the pancreas by promoting the release of glucagon like peptide 1 and the activation of glucose transporter β cellular insulin release and insulin gene transcription 32 .Metabonomic analysis of human adipocytes showed that Gln attenuated glycolysis and reduced the level of uridine diphosphate N-acetylglucosamine (UDP GlcNAc).UDP GlcNAc is an o-junction mediated by O-GlcNAc transferase β-N-acetylglucosamine post-translational modified substrate is closely related to the existence of proinflammatory transcriptional response of human adipocytes 33 .Glutamine reduces macrophage infiltration in white adipose tissue and reduces inflammatory responses 34 .Glutamine metabolism in macrophage polarisation regulation function or by manipulating macrophage polarisation to prevent or improve obesity or type 2 diabetes 35 .Animal studies have shown that L-Gln supplementation can improve insulin resistance in liver and muscle of obese mice 36 .www.nature.com/scientificreports/l-arginine was increased significantly in LatT2DM group.l-arginine is a functional amino acid and a precursor of nitric oxide.It plays an important role in animal maintenance, reproduction, growth, anti-aging and immunity.More and more clinical evidence shows that dietary supplementation of l-arginine can reduce obesity, lower arterial blood pressure, antioxidant and endothelial dysfunction, thereby relieving T2DM 37 .The signal pathway of l-arginine beneficial effect may include l-arginine-nitric oxide pathway, through which cellular signal proteins can be activated.More and more studies have shown that l-arginine may have the potential to prevent and/or alleviate T2DM by restoring insulin sensitivity in vivo 37 .The observed safe level for oral administration of Arg is ∼20 g/d 38 .Arginine and glycine increased the risk of T2DM in the western countries' subgroup 39 .
In addition, the slight improvements in T2DM prediction beyond traditional risk factors were observed when these metabolites were added in the predictive analysis 39 .
Alpha-linolenic acid and linoleic acid were decreased in LatT2DM group.Linoleic acid can be metabolized to n-6 via desaturase, resulting via the biosynthesis of gamma-linolenic acid, dihomo-gammalinolenic acid, and finally arachidonic acid 40 .The largest amount of arachidonic acid is found in phospholipid membranes, competing with n-3 acids for metabolism and with their products for receptors 40 .Linoleic acid as an unsaturated fatty acid linoleic acid has been proved to be beneficial to human body, including lipid-lowering, anti atherosclerosis, anti diabetes and anti-inflammatory effects 41,42 .
The decrease of pristanic acid level was in EarT2DM group.Pristanic acid originates mainly from the endogenous transformation of phytanoic acid in the liver and acts as a peroxisome proliferator activated receptor (PPAR)-α, PPAR-γ, and the natural ligand of retinoid X receptor, which is involved in peroxisome and mitochondrial function, oxidative stress, inflammatory pathway, cell signal transduction, glucose/energy metabolism, and microbial effects 43 , and can promote the expression of glucose carriers on the cell membrane and improve insulin sensitivity 44 .In this study, the decrease of pristanic acid level in patients with early onset T2DM needs further research to confirm whether it is related to insulin resistance in patients with early onset T2DM.There were also other characteristics in the metabolic spectrum between patients with early onset T2DM and late-onset T2DM compared to their healthy control groups.The findings of this study had several limitations.Firstly, this was a single-centre investigation.Secondly, the research object in this study was mainly the urban population of Tianjin, China.Thirdly, the study was only a preliminary study with a small sample size, and the metabolites and pathways involved need to be further verified.
In summary, patients with type 2 diabetes at different diagnosed ages may have different metabolite profiles.In the metabolism of amino acids and their derivatives, patients with early onset T2DM may have more level of l-glutamine and branched chain amino acids (especially leucine and isoleucine).For newly diagnosed T2DM patients over 50 years of age, there may be metabolic changes in arginine level and the decrease of unsaturated fatty acids.Therefore, it is necessary to consider the diagnosed age factor when using metabolic indicators as predictor of T2DM.Further studies are needed to determine whether the higher risk of complications in earlyonset type 2 diabetes is related to its specific metabolic profile.

Methods and materials
Patients.Newly diagnosed T2DM patients treated at the MMC of the Tianjin 4th Central Hospital from April to August 2018 were recruited as the research subjects, and healthy people were recruited as the control group at the same time.In this study, healthy people were referred to as volunteers who had no diabetes and had a low risk of diabetes when they entered the group (China Diabetes Risk Score < 25) 45 .We collected information on the patient's name, sex, smoking and drinking history, family history of diabetes, waist circumference, age, diabetes course, and blood pressure from the MMC system.The body mass index (BMI) was calculated and Chromatographic conditions and process: the samples were separated by Agilent 1290 infinity LC ultra high performance liquid chromatography system (UHPLC) HILIC column; Column temperature 25 ℃; Flow rate: 0.3 ml / min; Injection volume 2 μL; Mobile phase composition a: water + 25 mM ammonium acetate + 25 mm ammonia, B: acetonitrile; The gradient elution procedure is as follows: 0-1 min, 95% B; 1-14 min, B changes linearly from 95 to 65%; 14-16 min, B changes linearly from 65 to 40%; 16-18 min, B maintained at 40%; 18-18.1 min, B changes linearly from 40 to 95%; 18.1-23 min, B maintained at 95%; During the whole analysis process, the samples were placed in a 4 ℃ automatic sampler.In order to avoid the influence caused by the fluctuation of instrument detection signal, the random sequence is used for continuous analysis of samples.
Mass spectrum condition setting and process: The ESI source conditions were set as follows: Ion Source Gas1 (Gas1) as 60, Ion Source Gas2 (Gas2) as 60, curtain gas (CUR) as 30, source temperature, 600 °C; and IonSpray voltage floating (ISVF) ± 5500 V. On MS acquisition, the instrument was set to acquire over the m/z range 60-1000 Da, and the accumulation time for TOF MS scan was set at 0.2 s/spectra.In the auto MS/MS acquisition, the instrument was set to acquire over the m/z range of 25-1000 Da, and the accumulation time for the product ion scan was set at 0.05 s/spectra.The product ion scan was acquired using information-dependent acquisition (IDA) with a high sensitivity mode.The parameters were set as follows: the collision energy (CE) was fixed at 35 V with ± 15 eV; declustering potential (DP),60 V ( +), and − 60 V ( −), excluding isotopes within 4 Da, candidate ions to monitor per cycle: 6.
In the HILIC separation, samples were analysed using a 2.1 mm × 100 mm ACQUIY UPLC BEH 1.7 µm column (Waters, Ireland).In both ESI positive and negative modes, the mobile phase contained 25 mM ammonium acetate and 25 mM ammonium hydroxide in water and B = acetonitrile.The gradient was 95% eluent B for 1 min and was linearly reduced to 65% in 14 min, and then reduced to 40% in 2 min, maintained for t2 min, and then increased to 95% in 0.1 min, with a 5 min re-equilibration period.The gradients were at a flow rate of 0.3 mL/ min, and the column temperatures were kept constant at 25 °C.A 2 µL aliquot of each sample was injected.During the analysis, quality control samples (QC) were inserted into the sample queue to monitor and evaluate the stability of the system and the reliability of the experimental data.

Metabolite data analysis.
The raw MS data (wiff.scan files) were converted to mzXML files using the Proteo Wizard MS Convert before importing into freely available XCMS software.For peak picking, the following parameters were used: centre-wave m/z = 25 ppm, peak width = c (10, 60), and prefilter = c (10, 100).For peak grouping, bw = 5, mzwid = 0.025, and minfrac = 0.5.In the extracted ion features, only the variables with more than 50% of the nonzero measurement values in at least one group were retained.The identification of the metabolites using MS/MS with an in-house database was established using available authentic standards.After the normalisation to total peak intensity, the processed data were uploaded before importing into SIMCA-P (version 14.1, Umetrics, Umea, Sweden), where the data were subjected to multivariate data analyses, including Pareto-scaled principal component analyses (PCAs) and orthogonal partial least-squares discriminant analyses (OPLS-DAs).The variable importance in the projection (VIP) value of each variable in the PLS-DA model was calculated to indicate its contribution to the classification.Metabolites with a VIP value > 1 were further subjected to Student's t-tests at the univariate level to measure the significance of each metabolite, and P values less than 0.05, were considered to be statistically significant.The hierarchical cluster algorithm was used to cluster the differentially expressed proteins among groups, and the data were displayed in the form of a heat map.
The hierarchical cluster algorithm was used to cluster the differentially expressed proteins among groups, and the data were displayed in the form of a heat map.Taking the KEGG pathway as a unit and taking the metabolic pathway participated by this species or species with close genetic relationships as the background, the significance level of metabolite enrichment of each pathway was analysed and calculated using the Fisher's exact test to make a preliminary determination of the significantly affected metabolic and signal transduction pathways.

Statistical analysis. Statistical Program for Social
Sciences software (version 20.0; SPSS, Inc., Chicago, IL, USA) was used for data collation and analysis.The Kolmogorov-Smirnov normal test was performed on the measurement data.Means ± standard deviations were used for the statistical description of the variables that conformed to a normal distribution, and percentages (%) was used for counting data.An independent sample t-test was used to compare the measurement data.The chi-squared test was used for the comparison of the counting data.All the statistical tests were performed using bilateral tests with alpha = 0.05.