×
![]()
Introduction
With growing concerns about the increasing global incidence of type 1 diabetes, there is a need for biochemical markers that can help monitor progression, treatment and remission [1]. As part of the INNODIA study (Innovative approaches to understanding and arresting type 1 diabetes), we previously demonstrated serum protein differences between newly diagnosed (ND) individuals in the first year after diagnosis and unaffected family members (UFMs), together with statistically significant associations between serum proteins and C-peptide levels [2]. As a follow-up to the initial investigation, we analysed these proteins in samples from a separate and larger group of ND youth (n=146; 560 samples) and UFMs (n=272) who were subsequently recruited for the study.
Anzeige
Methods
Samples
Sera were collected from autoantibody-positive ND individuals with type 1 diabetes who were consecutively recruited to the INNODIA study. In keeping with our previous study [2], samples from individuals diagnosed under the age of 18 years were selected for analysis (n=146; 91 male and 55 female participants). Sex was based on reporting by parents or adult study participants. No selection restrictions were applied concerning regional or socioeconomic factors. Accurate data on ethnicity were not available and no specific ethnicity criteria were applied. Additional samples were collected from autoantibody-negative UFMs (n= 272; 169 male and 103 female participants) with a similar age range (see electronic supplementary material [ESM] Table 1, ESM Fig. 1). The ND samples were collected within the first 6 weeks of diagnosis (n=146) and then 3 months (n=132), 6 months (n=138) and 12 months (n=144) after diagnosis. Only one sample was collected from each UFM.
The study followed the guidelines of the Declaration of Helsinki for research on human participants, and the study protocols were approved by the ethics committees of the participating hospitals. Either the parents or participants gave their written informed consent.
Fasting C-peptide and fasting glucose measurements
Sample preparation and targeted LC-MS/MS
Serum samples were prepared and analysed as previously described with slight modifications, as detailed in ESM Methods. In brief, sera were digested with trypsin, spiked with isotope-labelled analogues of the targeted peptides and analysed by selected reaction monitoring (SRM) using a TSQ Vantage Triple Quadrupole Mass Spectrometer (Thermo Scientific, USA), coupled with an Evosep One liquid chromatograph (Evosep, Denmark).
Anzeige
Data analysis
Pre-processing, normalisation and false discovery rate calculations
Data analysis was conducted as previously described [2]. Briefly, linear mixed models (LMMs) were used to normalise the log2-transformed peptide abundances, adjusting for acquisition batch and run order. Periodic analysis of quality control samples indicated similar data quality metrics to those in our earlier study.
Changing peptide levels and beta cell function
Regression analysis on the natural logarithm-transformed fasting C-peptide/glucose ratios during the first year from diagnosis was carried out using LMMs adjusted for sex, height, BMI score (age-based BMI expressed as SD score), study centre and individual variation. Sex, height and BMI score were included as fixed effects, while individual and study centre were included as random effects, with individual nested under the study centre. To combine the peptide data protein-wise, meta p values were calculated using the sum of z (Stouffer’s) method with the R package metap 1.10 and adjusted for multiple correction using Benjamini–Hochberg procedure.
Differences in the levels of tryptic peptides measured from sera
To determine whether there were significant differences in the levels of tryptic peptides between individuals with type 1 diabetes and UFMs, peptide-wise LMMs were used. Age at baseline and sex were included as fixed effects in the LMMs, and individual and study centre were included as random effects, with individual nested under the study centre. Meta p values for the proteins were calculated as above.
Results
Apolipoprotein B-100, apolipoprotein M and glutathione peroxidase 3 are inversely associated with fasting C-peptide/glucose ratios in newly diagnosed individuals
To strengthen the verification of the previously observed associations between the target proteins and fasting C-peptide/glucose ratios [2], additional peptides were included in the analysis. Significant inverse associations (p<0.05) with fasting C-peptide/glucose ratios were demonstrated for two peptides from each of glutathione peroxidase 3 (GPX3) and apolipoprotein M (ApoM) and three peptides from apolipoprotein B-100 (ApoB; Table 1, ESM Table 2, ESM Fig. 2). The combined peptide data for these proteins confirmed that the effects were significant after false discovery rate (FDR) correction.
Table 1
Significant associations between SRM data and fasting C-peptide/glucose ratios among the proteins targeted (n=146 ND individuals; FDR <0.05)
Gene | Peptide sequence | Peptide effect size | Peptide p value | Peptide FDR | Protein meta p value | Protein FDR |
|---|---|---|---|---|---|---|
APOB | EVGTVLSQVYSK | −0.13 | 0.025 | 0.096 | 1.2 × 10−6 | 0.025 |
NIQEYLSILTDPDGKa | −0.23 | 4.0 × 10−4 | 0.005 | |||
ITENDIQIALDDAKa | −0.26 | 0.002 | 0.019 | |||
APOM | AFLLTPR | −0.45 | 3.2 × 10−4 | 0.005 | 1.3 × 10−5 | 7.0 × 10−5 |
SLTSCLDSKa | −0.25 | 0.006 | 0.038 | |||
GPX3 | FLVGPDGIPIMR | −0.20 | 0.025 | 0.096 | 0.0018 | 0.0068 |
QEPGENSEILPTLKa | −0.27 | 0.024 | 0.096 | |||
NSCPPTSELLGTSDRa | –0.16 | 0.14 | 0.45 |
Comparison of ND individuals and UFMs demonstrates differences in peptide levels during the first year after diagnosis
The comparisons of peptide levels between ND individuals and UFMs verified most of our earlier findings [2]. The results included comparable significant differences in the levels of 19 peptides, representing ten proteins (p<0.05; Table 2, ESM Table 3, ESM Figs 3, 4). Furthermore, the significant peptides included four of the additional peptides that were included in the analysis for hepatocyte growth factor activator (HGFAC), haemoglobin subunit beta (HBB) and GPX3. The combined peptide data further demonstrated that the differences were significant for these proteins after FDR correction.
Table 2
Significant differences in levels of targeted proteins between ND individuals (n=146) and UFMs (n=272) (FDR<0.05)
Gene | Peptide sequence | Peptide effect size | Peptide p value | Peptide FDR | Protein meta p value | Protein FDR |
|---|---|---|---|---|---|---|
AFM | DADPDTFFAK | −0.14 | 5.8 × 10−7 | 3.1 × 10−6 | 1.9 × 10−7 | 3.7 × 10−7 |
GQCIINSNK | −0.19 | 0.016 | 0.025 | |||
AESPEVCFNEESPK | −0.08 | 0.037 | 0.049 | |||
APOC1 | EWFSETFQK | −0.26 | 4.7 × 10−4 | 0.001 | 6.4 × 10−6 | 9.6 × 10−6 |
EFGNTLEDK | −0.16 | 0.002 | 0.004 | |||
C2 | HAFILQDTK | 0.14 | 1.1 × 10−5 | 4.1 × 10−5 | 8.3 × 10−10 | 2.5 × 10−9 |
AVISPGFDVFAK | 0.14 | 9. 6 × 10−6 | 4.1 × 10−5 | |||
F2 | TATSEYQTFFNPR | 0.42 | 6.4 × 10−26 | 1.7 × 10−24 | 3.2 × 10−17 | 1.9 × 10−16 |
GPX3 | NSCPPTSELLGTSDRa | 0.14 | 4.0 × 10−5 | 1.4 × 10−4 | 2.5 × 10−9 | 6.1 × 10−9 |
QEPGENSEILPTLKa | 0.09 | 2.9 × 10−4 | 8.6 × 10−4 | |||
FLVGPDGIPIMR | 0.09 | 0.003 | 0.006 | |||
HBB | VNVDEVGGEALGRa | 0.36 | 5.2 × 10−4 | 0.001 | 2.6 × 10−6 | 4.4 × 10−6 |
SAVTALWGK | 0.35 | 7.7 × 10−4 | 0.002 | |||
HGFAC | LEACESLTR | 0.3 | 7.6 × 10−11 | 6.9 × 10−10 | 2.3 × 10−10 | 9.0 × 10−10 |
VANYVDWINDRa | 0.14 | 0.008 | 0.014 | |||
HRG | DGYLFQLLR | 0.1 | 0.012 | 0.021 | 0.0038 | 0.0051 |
ADLFYDVEALDLESPKa | 0.08 | 0.064 | 0.083 | |||
TGFBI | LTLLAPLNSVFK | −0.08 | 0.016 | 0.025 | 0.016 | 0.018 |
TTR | AADDTWEPFASGK | −0.28 | 1.4 × 10−11 | 1.9 × 10−10 | 3.1 × 10−18 | 3.7 × 10−17 |
TSESGELHGLTTEEEFVEGIYK | −0.3 | 1.4 × 10−8 | 9.6 × 10−8 |
Discussion
Following on from our earlier study of serum protein markers of type 1 diabetes in youth among the first 100 ND participants in the INNODIA study [2], we have now analysed sera from the next 150 ND individuals recruited to the study. The previously reported inverse associations with fasting C-peptide/glucose ratios were confirmed for targeted peptides from the proteins ApoB, ApoM and GPX3. Furthermore, we verified the majority of the peptide-level differences between ND individuals and UFMs reported previously [2].
One of the challenges in serum proteomics is the pre-analytical variability, which can impact protein quantification [7, 8]. However, despite these challenges, the validations of the protein differences between ND individuals and UFMs were highly consistent. In this respect, our study benefited from a tightly controlled longitudinal sample collection protocol, along with the inclusion of matched UFMs. With recruitment based on diagnosis of type 1 diabetes, there were, however, more male than female participants in the cohort, following the tendency for a higher frequency of type 1 diabetes in male than female populations [9]. Nevertheless, for both the comparisons of peptide measurements between ND individuals and UFMs and the analysis of associations between fasting C-peptide/glucose ratios and target proteins, sex was included in the LMMs. This inclusion supports the generalisability of the findings to both male and female populations.
Although the results from the ND and UFM comparison were mostly validated, the peptide and C-peptide/glucose associations were consistent for only three of the 11 previously reported target proteins [2]. Notably, the parabolic course of the fasting C-peptide/glucose ratios in the first year was less pronounced in the follow-up data than in our previous study [2] (ESM Fig. 5).
Insulin plays a vital role in the regulation of lipid metabolism and lipid disorders are commonly diagnosed in individuals with type 1 diabetes [10]. In this follow-up study, we confirmed the inverse associations between fasting C-peptide/glucose ratios and peptides from ApoB and ApoM. Both of these proteins are expressed extensively in the major sites of insulin action, including liver and adipose tissue [10, 11]. Insulin indirectly inhibits ApoB-containing triacylglycerol-rich lipoprotein production and promotes clearance of the particles [10]. Interestingly, although ApoM is primarily produced in hepatocytes, it is also formed in human adipose tissue, and adipose ApoM levels have been reported to correlate positively with plasma ApoM levels [11]. Certain SNPs of ApoM have been associated with type 1 diabetes [12]. Additionally, ApoM has been shown to increase insulin secretion by binding to bioactive sphingolipid sphingosine 1-phospate (S1P) [13] and, in turn, insulin has been shown to inhibit ApoM expression [14]. Therefore, the inverse relationship between fasting C-peptide/glucose ratios and peptides from ApoB and ApoM may reflect how changes in endogenous insulin affect the secretion of both proteins and the clearance of ApoB-containing particles. Additionally, our data confirm the lower levels of apolipoprotein C1 (ApoC1) in sera from ND individuals compared with UFMs. Interestingly, recent reports have demonstrated that plasma/serum ApoC1 levels are already reduced after seroconversion [7, 15].
In keeping with our earlier study [2], and putatively indicative of oxidative stress, three peptides from GPX3 were detected at higher levels in ND individuals than in UFMs (Table 2). The previously noted inverse association with fasting C-peptide/glucose ratios observed for GPX3 [2] was significant for two of the measured peptides (p<0.05; Table 1). As before, and noted in relation to its role in the transport of the antioxidant vitamin E, the three peptides measured for afamin (AFM) were less abundant in ND individuals than in UFMs.
Consistent with our previous results [2], peptides from TGF-beta-induced protein ig-h3 (TGFBI) and transthyretin (TTR) were less abundant in ND individuals. These two proteins have been associated with islet cell survival and beta cell integrity, respectively [16], and their lower levels could signify how this milieu is compromised in type 1 diabetes.
Anzeige
Lastly and consistent with our previous study [2], coagulation factors and complement-related proteins were found at higher levels in ND individuals (Table 2). The increased levels of prothrombin (F2) and HGFAC may reflect thrombotic differences manifested in ND individuals [17].
Summary
These analyses confirm the previously reported differences in relative levels of proteins associated with type 1 diabetes between individuals with type 1 diabetes and unaffected individuals. These data, together with relationships between the target proteins and C-peptide/glucose ratios, further highlight the importance of these proteins, together with their associated pathways, in the pathogenesis of type 1 diabetes. Further research is needed to explore the clinical value of these targets and how these findings reflect and contribute to underlying disease pathogenesis.
Acknowledgements
We are grateful to staff at the University of Cambridge Department of Paediatrics laboratory, particularly A. Qureshi, for their contributions to the management of the samples. M. Hakkarainen and S. Heinonen at Turku Bioscience are thanked for their excellent technical assistance. We thank the personnel of the Turku Proteomics Facility at Turku Bioscience, which is supported by the University of Turku, Åbo Akademi University and Biocenter Finland.
Data availability
Access to these person-sensitive data is only through secure environment by application to the INNODIA Data Access Committee (see https://www.innodia.eu/).
Anzeige
Funding
Open Access funding provided by University of Turku (including Turku University Central Hospital). This project has received funding from the Innovative Medicines Initiative 2 Joint Undertaking (IMI2-JU) under grant agreement no. 115797 (INNODIA) and no. 945268 (INNODIA HARVEST). This IMI2-JU receives support from the European Union’s Horizon 2020 research and innovation programme, EFPIA, Breakthrough T1D (formerly known as JDRF) and The Leona M. and Harry B. Helmsley Charitable Trust. The views and opinions expressed are, however, those of the authors only and do not necessarily reflect those of the IMI2-JU, and the IMI2-JU cannot be held responsible for them. RL received funding from the Academy of Finland (grants 31444, 329277, 331793) and Business Finland and grants from the JDRF, Sigrid Jusélius Foundation, Jane and Aatos Erkko Foundation, Finnish Diabetes Foundation and Finnish Cancer Foundation. LLE reports grants from the European Research Council ERC (677943), Academy of Finland (310561, 329278, 335434, 335611 and 341342) and Sigrid Jusélius Foundation during the conduct of the study. MK has also received grants supported by the Sigrid Jusélius Foundation, Helsinki University Hospital Research Funds and Liv and Hälsa Fund. Research at the Turku Bioscience Centre (LLE and RL) was supported by the University of Turku Graduate School (UTUGS), Biocenter Finland, ELIXIR Finland and the InFLAMES Flagship Programme of the Academy of Finland (decision no. 337530). TV is supported by the Doctoral Programme in Mathematics and Computer Sciences (MATTI) of the University of Turku. MKH was supported by the Turku Doctoral Programme of Molecular Medicine (TuDMM), Päivikki and Sakari Sohlberg Foundation and Yrjö Jahnsson Foundation.
Authors’ relationships and activities
The authors declare that there are no relationships or activities that might bias, or be perceived to bias, their work.
Contribution statement
RM prepared the samples, conducted the analyses, prepared the tables and figures, evaluated and interpreted the data and co-wrote the manuscript. MKH automated and performed the sample preparation, evaluated and interpreted the data, prepared the figures and co-wrote the manuscript. TV analysed the data, prepared the figures and co-wrote the manuscript. TS supervised the analysis of the data. CM, LO, MP and SB initiated, designed and supervised the study. MK, LLE and RL designed and supervised the study. All authors edited, reviewed and approved the final version of the manuscript. RL 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
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
