Hypoglycemia-induced changes in complement pathways in type 2 diabetes

Background and aims An association between hypoglycaemia and adverse cardiovascular events has been suggested from longitudinal and retrospective cohort studies. The complement pathway proteins in hypoglycemia are not well studied. Here, we hypothesized that these circulating proteins would be elevated in response to hypoglycemia in type 2 diabetes (T2D) through the inflammatory response. Methods A prospective, parallel study in T2D (n = 23) and controls (n = 23). Subjects underwent insulin-induced hypoglycemia with blood sampling at baseline, hypoglycemia and post-hypoglycemia; SOMAscan proteomic analysis of complement pathway-related proteins, cytokines and inflammatory proteins was undertaken. Results At baseline: Complement C2 (p < 0.05) and Factor B (p < 0.05) were elevated in T2D. At hypoglycemia: Complement C2 (p < 0.05) and Factor B (p < 0.01) remained elevated, whilst Factor I became elevated (p < 0.05) in T2D; Complement C4b became elevated in controls (p < 0.05). In the post-hypoglycemia follow up period, Complement C2, Factor B and Factor I remained elevated in T2D; in addition, Factor D, Factor H and mannose-binding protein C showed elevations in T2D, whilst properdin, complement C3b, Factor H-related protein 5, complement C1q and decay-accelerating factor (DAF) showed elevations in controls. Granger causality analysis showed that inflammatory proteins appeared to drive complement protein changes in T2D; conversely, in controls, complement proteins drove inflammatory protein changes. Conclusions Baseline elevations in C2 and Factor B indicate upregulation of the complement pathway in T2D. Changes in complement pathway-related protein levels in response to hypoglycemia suggest both intrinsic and alternative pathway activation at 2-h that appears driven by the underlying inflammation in T2D and could contribute to a cardiovascular event. ClinicalTrials.gov NCT03102801. Date of registration April 6, 2017, retrospectively registered. https://clinicaltrials.gov/ct2/show/NCT03102801?term=NCT03102801&draw=2&rank=1.


Introduction
Results emanating from a number of large randomized controlled studies designed to investigate the effect of improving glycemic control on vascular complications in type 2 diabetes (T2D) are in accord with finding an association between severe hypoglycaemia and adverse events, notably cardiovascular-related events, and increased mortality [1e5]. The action to control cardiovascular risk in diabetes (ACCORD) study was terminated prior to completion due to an increase in mortality in the intensive glucose control arm in T2D patients [6], suggesting that the increased frequency of hypoglycemic events likely underlay the poor outcomes. Consistent with these findings of association between hypoglycaemia and adverse cardiovascular events were reports from a number of longitudinal and retrospective cohort studies [7e9] as well as a post hoc analysis of the Veteran Affairs Diabetes Trial (VADT) [10]; conversely, other studies, such as the Bypass Angioplasty Revascularization Investigation in Type 2 Diabetes study, failed to find such association [11]. Nevertheless, the weight of evidence has fostered significant concerns regarding the risk of hypoglycaemia and adverse cardiovascular outcomes in patients with T2D.
Patients with T2D have an increased risk of thrombotic disease and a number of mechanisms working in concert underlie this risk: coagulation factor activation [12], reduced fibrinolysis [13] and activated dysfunctional platelets [14]. Hypoglycemia, an adverse outcome of glucose-lowering medication in T2D that is escalated by tight glucose control regimens, provokes these mechanisms, there being an established relationship of hypoglycaemia with thrombogenicity [15] and platelet hyper-reactivity [16,17]. In addition to inhibiting fibrinolysis and activating platelets, hypoglycemia also induces pro-inflammatory proteins, such as the interleukins [18e20] and it is likely that recurrent hypoglycemic events further provoke thrombotic processes, such as atherosclerosis, in a cumulative manner through induction of the inflammatory response [21].
The inflammatory response is rapidly induced in response to hypoglycaemia and likely drives other acute response pathways, such as the complement cascade, with significant crosstalk between the two [22,23]. Complement system activation has been reported in obesity, insulin resistance and T2D [24] and has additionally been implicated in diabetic micro-and macrovascular complications [25,26]. Activation of the complement system has been reported in T2D patients diagnosed with myocardial infarction (MI), hyperactivation being associated with a poor prognosis [27].
While recognised as involved in the response to the hyperglycemia of diabetes, complement cascade pathways have not been well studied in the context of hypoglycaemia in T2D. Here, we hypothesized that hypoglycaemia would trigger an inflammatory response that would, in turn, activate complement pathways. Therefore, in this study, we determined a panel of circulatory complement pathway-related proteins over a 24-h timecourse in T2D and control subjects.

Study design
As has previously been described [28], "This prospective parallel study was performed in 46 subjects, adult T2D (n ¼ 23) and control (n ¼ 23) at the Diabetes Centre at Hull Royal Infirmary; all subjects were Caucasian and between the ages of 40e70 years. In the T2D cohort, duration of diabetes was <10 years and all subjects were on a stable dose of medication (metformin, statin and/or angiotensin converting enzyme inhibitor/angiotensin receptor blocker) for the preceding 3 months. For inclusion in the T2D cohort, no glucoselowering agents other than metformin were allowed, HbA1c levels were required to be <10% (86 mmol/mol)], and patients must not have had either hypoglycemic unawareness or hypoglycemia within the preceding 3-month period. In the control group, an oral glucose tolerance test was used to exclude diabetes. For study inclusion, subjects in both T2D and control cohorts were required to have a body mass index (BMI) between 18 and 49 kg/m 2 , normal renal and hepatic biochemical indices and no prior history of cancer nor any contraindication to insulin infusion to achieve hypoglycemia (ischemic heart disease, epilepsy, seizure history, drop attacks, history of adrenal insufficiency and treated hypothyroidism).
The trial was approved by the North West-Greater Manchester East Research Ethics Committee (REC number:16/NW/0518), registered at www.clinicaltrials.gov (NCT03102801) and conducted according to the Declaration of Helsinki."

Insulin infusion
The insulin infusion was performed as previously detailed [29]. "Following an overnight fast, bilateral ante-cubital fossa indwelling cannulas were inserted 30e60 min prior to the commencement of the test (0830h). To induce hypoglycemia, soluble intravenous insulin (Humulin S, Lilly, UK) was given in a pump starting at a dose of 2.5mU/kg body weight/min with an increment of 2.5mU/kg body weight/min every 15min by hypoglycemic clamp [30], until two readings of capillary blood glucose measured by a glucose analyser (HemoCue® glucose 201þ) of 2.2 mmol/L (<40 mg/dl) or a single reading of 2.0 mmol/L (36 mg/dl) [30], after which the hypoglycemia was immediately reversed with an infusion of 10% dextrose. The blood sample schedule was timed subsequently in respect to the time point that hypoglycemia occurred. Following the identification of hypoglycemia, intravenous glucose was given in the form of 150 ml of 10% dextrose and a repeat blood glucose check was performed after 5 min if blood glucose was still <4.0 mmol/L" SOMA-scan assay "The SOMAscan assay was used to quantify the complement pathway-related proteins, cytokines and inflammatory proteins reported in this study. The protein assay was performed on an inhouse Tecan Freedom EVO liquid handling system (Tecan Group, Maennedorf, Switzerland) utilizing buffers and SOMAmers from the SOMAscan HTS Assay 1.3K plasma kit (SomaLogic, Boulder, CO) according to manufacturer's instructions and as described previously [31,32]. The assay was performed in 96-well plates containing up to 85 plasma samples, 3 quality control and 5 calibrator plasma samples. Briefly, EDTA plasma samples were diluted into bins of 40%, 1% and 0.05% and incubated with streptavidin-coated beads immobilized with dilution-specific SOMAmers via a photocleavable linker and biotin. After washing bound proteins were first biotinylated and then released from beads by photocleaving the SOMAmer-bead linker. The released SOMAmer-protein complex was treated with a polyanionic competitor to disrupt unspecific interactions and recaptured on the second set of streptavidincoated beads. Thorough washing was performed before 5' Cy3 fluorophore labelled SOMAmers were released under denaturing conditions, hybridized on microarray chips with SOMAmercomplementary sequences, and scanned using the SureScan G2565 Microarray Scanner (Agilent, Santa Clara, CA).

Data processing and analysis
Initial Relative Fluorescent Units (RFUs) were obtained from microarray intensity images using the Agilent Feature Extraction Software (Agilent, Santa Clara, CA). Raw RFUs were normalized and calibrated using the software pipeline provided by SomaLogic. Samples with a high degree of hemolysis (Haptoglobin log RFU < 10) were excluded from the analysis."

Statistical analysis
There are no studies detailing the changes in complement pathway-related proteins in response to hypoglycaemia on which to base a power calculation. Sample size for pilot studies has been reviewed by Birkett and Day [33] They concluded that a minimum of 20 degrees-of-freedom was required to estimate effect size and variability. Hence, we needed to analyse the samples from a minimum of 20 patients per group. Comparison between groups was performed at each timepoint using Student's t-test. A p-value of <0.05 was considered statistically significant. Within-group comparisons are as follows: changes from baseline, and from hypoglycemia, to each subsequent time point were compared using a mixed linear model performed using the R package 'lmerTest' (R version 4.0.2). Statistical analysis was performed using Graphpad Prism (San Diego, CA, USA). The resulting p-values were further subjected to multiple testing correction using the FDR procedure.

Partial correlation and Granger causality analysis
Partial correlations between inflammatory and complement proteins were calculated using the Graphical Marcov Model implemented in the GGM package in R. To assess causal relationships between inflammatory and complement proteins, we used the Granger test implemented in R package 'lmtest': Briefly, the test evaluates whether previous values of an inflammatory protein are predictive of future values of a complement system protein. For each pair of inflammatory and complement proteins, the test was also performed in the opposite direction. Pairs where the test was significant in both ways were ignored since causation could not possibly be inferred.
Protein-protein interaction tools STRING 11.0 (Search Tool for the Retrieval of Interacting Genes) was used to visualize the known and predicted Protein-Protein Interactions for proteins identified by SOMAscan assay in plasma of T2D versus control subjects (https://string-db.org/). Interactions between proteins are evidence-based and collated from databases, experiments, neighbourhood, gene fusion, co-occurrence, text mining, co-expression, and homology. Here, we determined the relationships between the complement pathway-related proteins presented in this study."
Supplementary Table 1A details the p-value, % difference, fold difference and standard deviation (SD) for the proteins that differed between T2D and controls.

Within cohort changes at hypoglycaemia and post-hypoglycemia
There were no significant differences for any of the complement pathway-related proteins at the point of hypoglycaemia versus baseline in either the T2D or control group.

Post-hypoglycemia changes within group for controls only
Factor I increased at 2-h post-hypoglycemia versus hypoglycemia (p < 0.05) (Fig. 2C).
Supplementary Table 1B details the p-value, % change, fold change and SD for the proteins that showed 'within cohort' changes over the study time course.
There were no differences either between T2D and control groups or within T2D and control groups throughout the timecourse for the following complement pathway-related proteins: Complement C3, iC3b, C3a, C8a, MASP1 and CCL2 ( Supplementary  Fig. 1, A-F).
STRING analysis revealed the close linkage between complement pathway-related proteins (Fig. 5).
Correlation between age, BMI and inflammatory markers with complement pathway-related proteins that differed at baseline between T2D and controls No correlation was found between age or BMI and basal levels of Complement C2 and Factor B, the two complement pathwayrelated proteins that differed at baseline between T2D and control subjects.
Interleukin-1 correlated with baseline C2 in T2D only (r ¼ 0.42, p ¼ 0.046) and correlated with Factor B in controls only (r ¼ 0.43, p ¼ 0.039). Interferon-g did not correlate with baseline levels of C2 and Factor B in either T2D or controls.

Partial correlation and Granger causality analysis
To determine the degree of association between proteins from the complement and inflammatory pathways, a partial correlation analysis was performed. Whilst many proteins within either the complement or the inflammatory pathways showed close association, that was to be anticipated. Close association of proteins between pathways, defined as a p-value <0.05, was limited to 12 pairs of proteins (Table 1).
In order to test whether changes in complement pathway proteins were prompted by changes in inflammatory pathway proteins or vice versa, the Granger causality analysis was performed on the inflammatory and complement proteins. This analysis tests whether a variable of interest can be predicted based on the measurement of another variable from a previous timepoint.
We have previously shown that a number of inflammatory proteins are affected by hypoglycaemia [34] and these include C-X-C motif chemokine ligand 10 (CXCL10), interleukin 1 alpha (IL-1A), interleukin 10 (IL-10), interleukin 12 alpha (IL-12A), CeC motif chemokine ligand 19 (CCL19), tumour necrosis factor alpha (TNFalpha) and TANK-binding kinase 1 (TBK1) and therefore these specific proteins were correlated with the complement proteins. Protein pairs involving one complement and one inflammatory protein revealed by this lagged analysis are shown in Table 2. It is apparent that inflammatory proteins drive complement proteins more frequently in T2D than in controls, likely because the baseline levels of inflammatory proteins are elevated in T2D. By contrast, in controls, it is the complement proteins that more often drive the inflammatory proteins, whereas this lagged association is very rare in T2D.

Discussion
The results of this study show an elevation of Complement C2 and Factor B at baseline in T2D versus control subjects. Upon induction of hypoglycaemia, C2 and Factor B remained elevated, whilst Factor I became elevated in T2D and C4b became elevated in controls. Perturbations of complement pathway-related proteins occurred largely around the 2-h post-hypoglycemia timepoint and had, in almost all cases, returned to baseline at 24-h, demonstrating an acute but transient response.
The elevated complement C2 and Factor B at baseline in T2D subjects without any diabetes-related complications, and that remained consistently higher at all time points of the hypoglycemic challenge, are novel observations. Complement C2 is important in the lectin and classical pathways of complement activation and is important in the formation of C3-convertase and C5-convertase activation [24]; however, its upregulation in T2D has not been described before and theoretically may indicate that the complement system is primed for activation in diabetes. This is reflected in the observation that the complement pathway shows increased activation in T2D [35]. Interestingly, Factor B, a complement factor in the alternative pathway, is also important in the formation of C3convertase and C5-convertase activation [24]. Upregulation of both C2 and Factor B seen at baseline in T2D may be responsible for the well described increased levels of C3 and C3 activator in diabetes that are related to glucose intolerance and elevated fasting plasma glucose [35]. From the Granger causality analysis, it can be seen that inflammation was driving complement, and all the more so in the T2D cohort, where underlying chronic inflammation is well recognised and where these factors may be primed to respond. However, it is evident that complement changes were also driving inflammation, suggesting that a vicious circle of complement and inflammatory marker activation may result. That there is upregulation of complement C2 and Factor B may indicate that these are driven by metabolic factors [35] rather than inflammation. C2 and Factor B are closely linked, as they are encoded by single loci in the Class III region of the major histocompatibility complex on the short arm of chromosome 6 [36]. C2 in the classical pathway is the functional homologue of Factor B in the alternative pathway.
In the classical pathway, C2 associates with activated C4 (C4b) and is cleaved by activated C1 into 2 fragments, C2a and C2b. C2a combines with factor C4b to generate C3-and C5-convertases In the alternative pathway, C3-convertase is formed from C3 and Factor B: Factor B associates with C3b and is cleaved by Factor D into Ba and Bb, and the association of C3b and Bb is the alternative pathway C3-convertase [36] (Fig. 1).
C2 and Factor B are synthesized in the liver, the source of the majority of circulating protein, and in monocyte/macrophage lineage cells. Their expression can be induced by inflammatory proteins, such as interferon-g and interleukin-1 [37], the latter correlating with C2 and Factor B here, and likely accounting for the increased basal levels in T2D versus controls through the increased basal levels of the inflammatory proteins seen in T2D [34].
In accord with our findings, baseline Factor B has previously been reported to be elevated in serum of Asian patients with T2D [38] and its expression in adipose tissue correlates with fasting glucose and circulatory lipids [39,40]. Elevated levels of circulatory Factor B are reported to increase the risk of endothelial dysfunction [41] and coronary heart disease [42]. Despite its homology to Factor B, little has been reported about the role of C2 in disease.
Acquired causes of low complement levels include autoimmune disorders, such as systemic lupus erythematosis, several forms of glomerulonephritis, vasculitides and autoimmune pancreatitis; the common underlying mechanism is accelerated consumption of proteins by immune complexes [43,44]. This mechanism is the likely cause of the transient fall in complement pathway-related proteins seen in this study at 2-h following the hypoglycemic insult. This study benefited from the inclusion of normal controls to define what a normal physiological response may be to hypoglycemia, with the caveat that non iatrogenic hypoglycemia is a rare and pathological event in non-diabetic individuals. However, it highlights the impact of the pathophysiological changes that T2D may have even when of relatively short duration. The complement protein changes seen in response to hypoglycemia were mainly seen at 2-h in both controls and T2D, and often showed a decrease in complement protein levels that was transient, reverting to basal levels within 2-h. In the controls, complement factors C3, C4, D, C5, C5a, C5b-C6 complex and C1r all significantly decreased at 2 h compared to baseline. In T2D, complement factors C3, C3b, C3a desArg, C3d, C5, C5a, C5b-C6 complex C1r and CD55. This suggests that in controls and in T2D that both the classical and alternative complement pathways were being activated as shown in Fig. 1. Complement levels can fall to very low levels within a few hours due to the development of immune complexes so that the fall in complement here may be a reflection of activation; consumption of complement in response to the physiological stress response and reduction of complement levels have been reported in acute pancreatitis with the levels of C3 and C4 being sensitive and accurate to the severity of the pancreatitis [45], whilst a fall in C3 and its reversibility was predictive in renal allograft rejection and recovery [46].
The drivers of the change in complement response to hypoglycemia differed between controls and T2D. Beyond its role in pathogen elimination, the complement system is a recognised mediator of the inflammatory response [22,23]. While partial correlation analysis infers considerable interaction between the complement and inflammatory pathways, as evidenced by the string pathway analysis, the Granger causality analysis revealed that in controls hypoglycemia induced a response in complement that then affected the inflammatory proteins, whilst, conversely, in T2D it was largely the inflammatory proteins that drove the complement cascade. That inflammatory proteins are the drivers in T2D may be explained by their upregulation even in the basal state in T2D [34]. Hypothetically, chronic low-grade inflammation may prime the complement pathway proteins; when a hypoglycemic episode occurs, this may initiate an inflammatory response/cascade that may be prolonged beyond the transient complement changes and could thereby contribute to a cardiovascular event. Others have shown that the impact of a hypoglycemic effect may still be evident after seven days and, whilst complement C3 was not altered at the time of the hypoglycemic event, it became elevated at seven days suggesting that the inflammatory cascade had continued for this period of time [47]. This accords with the well described vicious cycle of the inflammatory process leading to complement activation leading to further inflammatory stimulation [48].
This study has a number of limitations. Firstly, the relatively small subject numbers in this intensive study protocol may not have identified additional alterations in complement proteins though a lack of power; as this was a pilot study, larger appropriately powered confirmatory studies should be undertaken. Secondly, we report the complement protein levels and not their activation, and functional studies are needed to confirm these findings particularly with regards to the link between inflammation and complement levels. Thirdly, this was a homogenous Caucasian population and its applicability to other ethnicities needs confirmation.
In conclusion, baseline elevations in C2 and Factor B indicate upregulation of the complement pathway in T2D and changes in complement pathway-related protein levels in response to hypoglycemia suggest both intrinsic and alternative pathway activation at 2-h that appear driven by the underlying inflammation in T2D and that could contribute to a cardiovascular event. It is, however, important to note that this was a pilot study and further larger, appropriately powered studies are needed to confirm and extend these findings.  Author contributions ASMM, MN and AEB analyzed the data and wrote the manuscript. ID performed the statistical analysis. AA-Q performed the clinical studies. TS supervised clinical studies and edited the manuscript. SLA contributed to study design, data interpretation and the writing of the manuscript. All authors reviewed and approved the final version of the manuscript. Alexandra E Butler is the guarantor of this work.

Ethics approval and consent to participate
The Newcastle & North Tyneside Ethics committee approved this study that was conducted according to the Declaration of Helsinki. All study participants signed an informed consent form prior to participation.

Consent for publication
All authors gave their consent for publication.

Availability of data and materials
All the data for this study will be made available upon reasonable request to the corresponding author.