Bioinformatic analysis
To uncover plausible molecular mechanisms responsible for the observed impact of GLP-1R agonists treatment on urinary peptides in T2DM patients, the proteases potentially responsible for cleavage of the 70 statistically significant peptides were investigated using Proteasix. In total, 10 endopeptidases were retrieved as a result of the default search with the “Observed Prediction tool” of Proteasix, putatively responsible for cleaving 38 urinary peptides (36 downregulated and 2 upregulated) out of the 70 urinary peptides. The results are provided in Table S2. Most of the predicted endopeptidases belonged to the matrix metalloproteinase (MMP) family of proteases (7 out of 10 proteases), responsible for cleaving peptides at both the N’ and C’ terminals. Further proteases predicted as potentially responsible for cleaving the N’ terminal belonged to the cathepsin family (CTSL and CTSD) while that cleaving at the C’ terminus was A disintegrin and metalloproteinase with thrombospondin motifs 5 (ADAMTS5). Notably, the proteases MMP2, MMP9 and MMP13 were mapped to at least 6 cleavage sites each.
The protein-protein interactome was constructed using 26 parental proteins identified from 70 urinary GLP-1R agonist-associated peptides, using the STRING database. The network consisted of 27 nodes and 113 edges, as depicted in Figure 2I. The protein-protein interaction enrichment yielded a significant p-value < 1.0e–16. While, most of the collagen proteins can be observed to interact with all the other collagen proteins, interestingly none of these interacted with the non-collagen proteins. Furthermore, a protein-protein interaction network within the non-collagen proteins could also be observed, indicating their involvement in the pathophysiology of T2DM. Within this network, a total of 9 KEGG pathways were predicted to be significantly enriched, including pathways related to protein digestion and absorption, ECM-receptor interaction, AGE-RAGE signaling pathway in diabetic complications, amoebiasis, relaxin signaling, focal adhesion, human papillomavirus infection, PI3K-Akt signaling pathway, small cell lung cancer and platelet activation, detailed results are provided in Table S3.
Discussion
In the last decade, GLP-1R agonists have been the recommended and preferred second line treatment for T2DM patients. Despite the various advantages of GLP-1R agonists over other anti-hyperglycemic drugs, the underlying molecular mechanisms of treatment with GLP-1R agonists have not been studied in-depth. Aiming to understand the effect of GLP-1R agonist treatment on T2DM patients, the urinary peptidome of thirty-two T2DM patients were analyzed for the first time in this study with CE-MS. The untargeted peptidomic analysis coupled with statistical tools identified 70 statistically significant (adjusted for multiple testing) urinary peptide fragments in abundance between the pre- and post-treatment samples. These urinary peptides generated from 26 parental proteins. Uniform distribution of intensity of the 70 peptides (red spots in Figure 2A), emphasized that their observed significant change with GLP-1R agonist treatment was not a function of abundance in the urine samples. For most of these 70 peptides, a comparative analysis further revealed a combined downregulation with GLP-1R agonist treatment (66/70 peptides).
In total, 59 out of the 70 statistically significant urinary peptides, majorly generated from three prominent collagen proteins COL3A1 (n =16), COL1A1 (n =15) and COL1A2 (n =10). Recently, He et al. ,[41] reported about the high abundance of collagen peptides observed in urine samples, as a result of proline hydroxylation which plausibly inhibits its reabsorption in the kidney. 55 out of the 59 collagen peptides were observed to be significantly downregulated with GLP-1R agonists treatment in the T2DM urinary proteome, while 4 peptides (COL3A1;n =3 and COL1A2; n =1) showed a significant upregulation on treatment which could plausibly be attributed to the variation in the post-translational modification by hydroxylation of proline residues in the peptides and varied proteolytic cleavage. In-line with the extensive recent report by Mavrogeorgis et al. ,[42] all the urinary collagen peptides identified in this study were devoid of the signal peptide, N-terminal pro-peptide and C-terminal pro-peptide; corresponding only to the mature protein region (Figure 2G and 2H). The observed downregulation of the collagen peptides in our study could potentially represent rather an attenuated degradation of the mature collagen protein, instead of resulting from protein synthesis or protein assembling processes. Rossing et al.[43] and Genovese et al. ,[44] have earlier speculated that the decrease of urinary collagen proteins could attribute to decreased proteolysis of collagen molecules, resulting from an increased resistance to proteolytic cleavage or an increased expression of protease inhibition.
To further corroborate the above speculation, endopeptidases responsible for putatively cleaving 38 statistically significant collagen peptides were majorly accounted to the MMP family of proteases (89% of the cleavage sites, i.e., 34 out of the 38 urinary peptides) by MMP2 (21.1%), MMP9 (21.1%) and MMP13 (15.5%). The suggested downregulation in the activity of MMP peptidases as observed by decrease in intensity of collagen peptides in this study, is in-line with literature. Down-regulation in expression of MMP9 was observed on treatment with Liraglutide in a study that included induced-DM rabbit models[45]. In an another study, 45% and 60% reduction in activity of MMP2 and MMP9, respectively, in addition to 60% reduced COL1A1 levels, was observed in a male C57BL/6 mice on treatment with Semaglutide (GLP-1R agonist type)[46]. Research groups exploring the effect of Exenatide (GLP-1R agonist type) treatment on tumor necrosis factor-α human coronary artery smooth muscle cells[47] and human retinal pigment epithelium cells[48], reported the downregulated expression of MMP2 and MMP9, respectively on treatment. Another study analyzing atherosclerosis associated biomarkers in T2DM female subjects, reported decrease in MMP2 and MMP9 levels, with an increase in GLP-1 and GLP-1R levels[49]. Interestingly, treatment of Human SW1353 with Dulaglutide (GLP-1R agonist type)[50] and Fibroblast-like synoviocytes cell lines with Exenatide[51], also resulted in the downregulation of MMP13 and ADAMTS5 proteases.
On the other hand, collectively, all the non-collagen peptides (11 out of 70) showed lower abundance after GLP-1R agonist treatment and each generated from a different protein namely, SERPINA1, APOC3, CD99, CPSF6, CRNN, SERPINA6, HBA2, MB, VGF, PIGR and TTR. In studies analyzing the effect of Liraglutide (GLP-1R agonist type) on T2DM patients, Rafiullahet al. ,[52] reported the downregulation of the urinary protein SERPINA1 and Adiels et al. ,[53] reported decreased secretion of APOC3, respectively with treatment. No literature was found reporting regulation of the other non-collagen peptides by GLP-1R agonist treatment. However, impact of diabetes and/or obesity on these molecules has been reported in the literature. CD99 transcripts have been stated to up-regulate in T2DM profiles[54] and Pasello et al. ,[55] reported the proteins involvement in biological processes such as cell death and inflammation. CPSF6 indirectly modulates glucose homeostasis and insulin secretion[56]; while, HBA2, a commonly known marker for anemia and β-thalassaemia, interferes with glycemic markers of T2DM patients[57].
Increased levels of TTR have been associated with glucose intolerance, obesity and decreased pancreatic β-cells percentage in T2DM[58, 59]. Along the same lines, increased levels of SERPINA6 have been identified in obese patients and is reported to play a crucial role in glucose homeostasis, along with reducing insulin resistance and inflammation[60, 61]. Benchoula et al. ,[62] in their extensive review reported that VGF is expected to induce obesity, while also playing a role in lipolysis and insulin secretion, hence, acting as a potential target in T2DM therapy. CRNN is reportedly associated with the immune system and acts as an inflammation marker in chronic diseases[63, 64]. In addition, elevated levels of MB, a known cardiac marker, were reported in T2DM patients[65] and has been associated with insulin resistance, dyslipidemia and abnormal glucose metabolism with elevated levels acting as a biomarker for diabetic kidney disease[66]. Similarly, inflammatory mediators have been reported to increase PIGR protein levels in the renal tubular cells, linking its role in renal injury[67]. As a result of this exploratory study and literature search, we report the plausible molecular mechanisms affected by the treatment of GLP-1R agonists on the pathophysiology of T2DM, as hypothesized from the functions of the down-regulated non-collagen proteins in Figure 3. The results in this study may therefore indicate towards the beneficial effect of GLP-1R agonists in the context of management of T2DM and prevention or delaying the progression of its associated diseases.
Regardless of the novel findings, this study comes with its own limitations. Firstly, the large difference in time points between the pre-treatment and administration of GLP-1R agonists of 4.4 ± 4.11 may have resulted in unidentified variations of clinical parameters as well as the composition of urinary peptides, which were not accounted in this study. Secondly, the administering of multiple anti-hypertensives and GLP-1R agonist drugs at varied dosages and different combinations to the T2DM patients may have produced different effects of the treatment, which were also not analyzed in this study. Thirdly and surprisingly, within the follow-up we did not observe significant changes in Hb1Ac and BMI. The study was not powered to detect such changes, which would require about 10 times the number of subjects to be included, however, this does help in eliminating the argument that the reported changes in this study could be a result of weight loss, and instead support the proteins’ role in T2DM pathophysiological mechanisms. Fourthly, since only one urinary peptide was identified per non-collagen protein, the reported effects of GLP-1R agonist treatment on these proteins cannot be definitive and require further experimental studies with increased power. However, to overcome this short-coming, we additionally, performed a paired Wilcoxon test on a cohort of thirty-two T2DM patients administering only the anti-hypertensive drugs and no GLP-1R agonists drugs from the same PROVALID study, that were matched to the GLP-1R agonist treated cohort by age, sex, BMI, SBP, DBP and eGFR. Interestingly, we did not identify any statistically significant urinary peptides between the paired urine samples collected at a similar time difference as in this study. Finally, we could identify 15 urinary peptides downregulated by GLP-1R agonist treatment in our study, whose elevated levels have been previously reported as markers of heart failure[68]. This observation may further indicate the positive effect of GLP-1R agonist treatment.
To conclude, this untargeted peptidomic analysis to identify the effect of GLP-1R agonists treatment on the urinary peptidome of T2DM patients, indicated as a prominent finding the downregulation of MMP proteases, as identified by the downregulation of urinary collagen peptides on GLP-1R agonists treatment. Treatment with GLP-1R agonists also resulted in the decrease of SERPINA1, APOC3, CD99, CPSF6, CRNN, SERPINA6, HBA2, MB, VGF, PIGR and TTR peptides; indicating a potential benefit as many of these proteins express increased levels in T2DM patients. The results also merit the possibility of larger cohort studies to further understand the impact on underlying molecular mechanisms such as insulin resistance and inflammation, behind the findings.
Associated data
Data will be made available upon request directed to the corresponding author. Proposals will be reviewed and approved by the investigators and collaborators based on scientific merit. After approval of a proposal, data will be shared through a secure online platform.