Hypoglycaemia in type 2 diabetes exacerbates amyloid‐related proteins associated with dementia

Hypoglycaemia in diabetes (T2D) may increase the risk of Alzheimer's disease (AD). We hypothesized that hypoglycaemia‐induced amyloid‐related protein changes would be exacerbated in T2D.

Alzheimer's disease (AD) comprises 60-80% of all dementia cases. 7 Epidemiological studies indicate an increased risk for patients with T2D developing AD. [8][9][10] Elevation in plasma amyloid precursor protein (APP) has been reported in association with AD, levels tending to increase with increasing cognitive impairment. 11,12 While no difference in plasma levels of alpha-synuclein (SNCA) had previously been reported, 13 a recent study showed lower concentrations of SNCA in red blood cells in subjects with AD. 14 Improved management of T2D involves stricter glucose control, increasing the risk and frequency of hypoglycaemic episodes.
Hypoglycaemia itself is directly linked to cognitive dysfunction, complicating diabetic management and perhaps increasing the risk of dementia. 15 We hypothesized that changes in hypoglycaemia-induced amyloid-related protein levels would be exacerbated during/following hypoglycaemia in T2D, providing a potential mechanistic link between T2D-related hypoglycaemia and AD. This study was specifically designed to mimic the physiological responses to hypoglycaemia as would be seen in patients with diabetes in clinical practice. 16 To this end, we analysed amyloid-related proteins levels following acute iatrogenic-induced hypoglycaemia in subjects with and without T2D.

| Study design
This was a prospective parallel study performed in 46 subjects (23 adults with T2D and 23 controls) at the Diabetes Centre at Hull Royal Infirmary; all were white and aged 40-70 years. The duration of diabetes was <10 years and all subjects with T2D were on a stable dose of medication (metformin, statin and/or angiotensin-converting enzyme inhibitor/angiotensin receptor blocker) over the previous 3 months. For those with T2D no medication for glycaemic control except metformin was allowed, haemoglobin A1c levels were <10% (86 mmol/mol) and none had either hypoglycaemic unawareness or hypoglycaemia within a 3-month period. In the control group, diabetes was excluded with an oral glucose tolerance test. All subjects had a body mass index between 18 and 49 kg/m 2 , and all had normal renal and hepatic biochemical indices and no previous history of cancer or any contraindication to insulin infusion to achieve hypoglycaemia (ischaemic heart disease, epilepsy, seizure history, drop attacks, history of adrenal insufficiency and treated hypothyroidism).

| Study participants
Medical history, clinical examination, routine blood tests and an electrocardiogram were performed for all participants. A continuous insulin infusion was performed to induce hypoglycaemia as previously detailed 16 with blood samples taken at 0, 30, 60, 120 and 240 min post-hypoglycaemia. After 240 min, participants were provided lunch and were given their (morning) diabetes medications. Patients later took their evening medication as prescribed. Subjects reattended 24 h following the induction of hypoglycaemia; patients withheld their medications until they completed the blood tests in the fasted state, after which breakfast was provided. Before discharge, blood glucose was checked using a glucose analyser (HemoCue® glucose 201+) to ensure normal levels, together with other vital signs.

| Insulin infusion
The insulin infusion was performed as previously detailed. 16 Following an overnight fast, bilateral antecubital fossa indwelling cannulas were inserted 30-60 min before the commencement of the test (08:30 h).
To induce hypoglycaemia, soluble intravenous insulin (Humulin S; Lilly, Basingstoke, Hampshire, UK) was given in a pump starting at a dose of 2.5 mU/kg body weight/min with an increment of 2.5 mU/kg body weight/min every 15 min by hypoglycaemic clamp, 17 until two readings of capillary blood glucose measured by a glucose analyser (Hemo-Cue® glucose 201+) ≤2.2 mmol/L (<40 mg/dL) or reading of ≤2.0 mmol/L (36 mg/dL). 17 The blood sample schedule was timed subsequently in respect to the timepoint that hypoglycaemia occurred. Following the identification of hypoglycaemia, 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. The comparison of blood glucose levels at baseline, at hypoglycaemia and post-hypoglycaemia up to 24 h is shown in Figure 1A.

| Biochemical markers
Blood samples were separated immediately by centrifugation at 2000 g for 15 min at 4 C, and the aliquots were stored at −80 C, within 30 min of blood collection, until batch analysis. Fasting plasma glucose, total cholesterol, triglycerides and high-density lipoprotein cholesterol levels were measured enzymatically using a Beckman AU 5800 analyser (Beckman-Coulter, High Wycombe, UK).

| SOMA scan assay
The SOMAscan assay used to quantify proteins was performed on an in-house 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, USA) according to the manufacturer's instructions and as described previously. 18,19 The assay was performed in 96-well plates containing up to 85 plasma samples, three quality control and five 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 reca-  Statistical analyses were performed on log 2 RFU values using R version 3.5.2 (R Foundation for Statistical Computing, Vienna, Austria) including base R package. Data handling and differential protein expression were analysed using the autonomics and limma 20 packages. For differential protein analysis, we applied limma models containing contrasts between timepoints, as well as contrasts between healthy subjects and patients with diabetes at single timepoints. In both models, blocking by patient ID was performed to account for random effects. Batch effect correction was performed by adding batch as a covariate to the model. Limma obtained P values were corrected using the Benjamini-Hochberg method. 21

| Statistical analysis
There are no studies detailing the changes in Alzheimer proteins in response to hypoglycaemia on which to base a power calculation.
Sample size for pilot studies has been reviewed by Birkett and Day. 22 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. Data trends were visually evaluated for each parameter and non-parametric tests were applied on data that violated the assumptions of normality when tested using the Kolmogorov-Smirnov test. Comparison between groups was performed at each timepoint using Student's t-test. P < .05 was considered statistically significant. Within-group comparisons are as follows: changes from baseline, and from hypoglycaemia, to each subsequent timepoint were compared using Student's t-test. The sample size was too small to adjust for baseline covariates. Statistical analysis was performed using Graphpad Prism (San Diego, CA, USA).
For the proteomic analysis we fitted an intercept-free general linear model as a function of a subgroup (i.e. condition:timepoint), while taking the patient ID as a random effect using the R package limma.

| RESULTS
In total, 46 subjects were recruited (23 people with T2D, 23 controls). 16 Demographic and clinical characteristics of the subjects are presented in Table 1

| Differences between type 2 diabetes and controls
In the T2D cohort, baseline levels of APP were elevated (P < .01) ( Figure 1B

| Within cohort changes at the point of hypoglycaemia
Significant changes occurred in the control group in response to hypoglycaemia with APP and noggin being elevated [APP P < .01 ( Figure 1A); noggin P < .05 ( Figure 2C)] and SNCA levels decreasing (P < .01; Figure 1B).
In the T2D cohort, APP levels were already elevated as compared with controls, so while there was a further increase in response to hypoglycaemia, this did not reach significance. While SNCA levels were already relatively decreased in T2D, as in controls, SNCA levels were further depressed in T2D in response to hypoglycaemia (P < .05; Figure 1B).  Figure 1B). Canonical pathways that were affected by key AD-related proteins showed a direct relevance to the induced hypoglycaemia reported here.

| Correlation between age and body mass index and Alzheimer's disease-related proteins
Basal levels of APP increased with age in subjects with T2D, but not in controls ( Figure S1A; see Supporting Information); however, basal levels of SNCA did not differ with age in either group ( Figure S1B branes. 28 Thus, these diverse conditions are all considered diseases of protein misfolding. [29][30][31] Much evidence points to the prefibrillar oligomers as being the primary cytotoxic form of these proteins. [32][33][34][35] AD is characterized by an accumulation of β-amyloid (Aβ) and tau proteins in the brain. Evidence points to the generation of Aβ from APP as the critical step in the development of AD 36 and elevated serum levels have been reported, 11,12 as are reported here. In this study, it can be seen that the APP levels continue to fall after the hypoglycaemic event in those with T2D. The comparison of blood glucose in post-hypoglycaemic timepoints between control and T2D revealed a significant increase in blood glucose level occurring 1 h after the hypoglycaemic event and continuing up to 24 h in T2D compared with controls. This elevation of blood glucose in T2D occurred in response to a meal that the patients received 10 min posthypoglycaemia ( Figure 1A). This elevation of blood glucose posthypoglycaemia is concurrent with the decrease of APP in T2D. A previous report demonstrated the same, where an increase in blood insulin concentration, followed by an increase in blood glucose concentration, was associated with a decrease in plasma APP concentration. 37 Moreover, previous reports demonstrated that an increase in insulin stimulates the synthesis of the lipoprotein receptor-related protein (LRP), within minutes, in vitro in hepatic cells. 38 As APP binds directly to LRP, which in turn mediates its uptake into cells, 39 and we observed a rapid rise of insulin just after starting the insulin clamp F I G U R E 5 Ingenuity pathway analysis. Ingenuity pathway analysis (IPA) of four amyloid-related protein genes, amyloid P component (APCS), amyloid precursor protein (APP), pappalysin (PAPPA) and serum amyloid A1 (SAA1). IPA demonstrates the most significantly affected pathways relating to the proteins in question. Of particular relevance to the induction of hypoglycaemia reported here, canonical pathways (CPs) affected by APP include the neuroprotective role of Thimet oligopeptidase (THOP1, an enzyme able to degrade the amyloid-beta precursor protein and generate amyloidogenic fragments) in Alzheimer's disease and the neuroinflammation signaling pathway; CPs affected by APCS and SAA1 include acute phase response signaling (data not shown), it is possible that an insulin-induced decrease in plasma APP occurred in our study. Taken together, hyperinsulinaemia and postprandial hyperglycaemia probably drive the depression of plasma APP levels in subjects with T2D at post-hypoglycaemic timepoints (up to 24 h), and this action is probably mediated by rapid uptake of APP by the lipoprotein-binding protein LRP.
It is also possible that the decreasing APP levels in the T2D cohort post-hypoglycaemia reflect a general stress response. A study assessing proteomic changes in military personnel following moderate blast exposure resulting in traumatic brain injury found an acute decrease in APP lasting for up to 3 days post-injury and suggested that this was a protective mechanism. 40 IPA in that study revealed that the APP network was central in the response to traumatic brain injury. 40 Of relevance here, acute hypoglycaemia has been suggested to result in a form of traumatic brain injury and patients with T2D may be more severely affected than their nondiabetic counterparts, 41 which may help to explain the differences in protein levels in the peripheral circulation between the two groups.
Given that diabetes may exacerbate changes in traumatic brain injury, 42 this may also account for the differences seen between controls and subjects with T2D for MAPT and for the changes in noggin.
Plasma MAPT levels had also fallen post-hypoglycaemia in PAPPAs are known to cleave insulin-like growth factor binding proteins, and overexpression of PAPPA-2 has been shown to play a role in Aβ peptide accumulation in AD. 49 Increased serum levels of PAPPA-1 have been reported in subjects with T2D 50 and this is in accord with their suggested shared pathophysiology. 51 Elevated levels of PAPPA were only seen in patients with T2D after the hypoglycaemic event and were increased for up to 4 h, as levels returned to baseline at 24 h. Studies are inconclusive as regards circulating APOE levels in AD, 52 with some reporting an increase, 53 some a decrease 54 and some no change. 55 In this study, we found an increase in APOE3 levels in T2D following hypoglycaemia and lasting until the 4-h timepoint, but with no change in controls. APOE and APOE4 levels were unchanged throughout the study in either cohort.
Higher circulating APOA1 levels are associated with a lower risk of AD 56 and dementia. 57 In this study, APOA1 showed an overall decrease, still present at 24 h, following hypoglycaemia in T2D but not in controls. SAA1 is an acute-phase protein that may play a housekeeping role in healthy tissue, but increased expression has been shown in the brain in AD. 58 There were no changes in SAA in either the controls or T2D in response to hypoglycaemia. That so many of the proteins changed in response to induced hypoglycaemia is not surprising given the interrelated protein interactions shown clearly in the STRING analysis that were in accord with other reports 11,12,44,48,50,56 ; however, the differential effect of the protein response in T2D compared with controls was marked.
It can be seen in this study that basal APP levels increased with age only in the T2D cohort, and a decrease in the basal APOE proteins, that have been reported to be protective, 59,60 was also seen only in the T2D cohort. Taken together, these results suggest that an AD pattern of proteins may be seen at baseline for patients with T2D and the changes in APP and APOE proteins are more pronounced with increasing age; hypoglycaemia causes their enhanced expression that may lead to the accelerated development of AD in susceptible patients or perhaps those particularly prone to severe and recurrent hypoglycaemia, as hypoglycaemia has been associated with AD. 15 What is of potential concern is that the patients with T2D in this study had a relatively short duration of disease (<5 years) and they were all on a single therapy (metformin) that was not associated with hypoglycaemia and, yet, they showed exaggerated changes in the AD-related proteins. Repetition of this study in patients of longer diabetes duration or on hypoglycaemic agents, including insulin, may show an even greater differential in ADrelated protein expression. Further, it would be of interest to determine whether a similar profile of protein changes is observed when study subjects are subjected to a less severe but more prolonged hypoglycaemic episode, and for blood samples to be taken beyond 24 h and perhaps up to 1 week. Given the changes in APP, further work on the protective effects of APP metabolites needs to be undertaken. 61 Finally, that hypoglycaemia is having these effects is being assumed, rather than a response to counter-regulatory parameters.
The strengths of this study were inclusion of a group of T2D subjects of short disease duration who were relatively treatment naïve and not on polypharmacy. The main study limitation is the small study numbers and, with a larger population, even greater differences in plasma levels of amyloid-related proteins may have been discerned.
However, it is important to note that these patients were subjected to a severe hypoglycaemic episode, during which it would be anticipated that changes in protein levels would have become apparent. While subjects with T2D were older and more obese, this should not have altered the expression of these proteins. A further limitation is that peripheral blood samples were only collected up to the 24-h timepoint, and it would be important in future studies to extend this period, perhaps for up to 1 week, to profile the time course of posthypoglycaemia protein changes better.

| CONCLUSION
In conclusion, these data are in accord with the hypothesis that hypoglycaemia has a detrimental effect on AD-associated proteins. At baseline, subjects with T2D had elevated APP and reduced SNCA levels in plasma, reflecting changes previously reported in AD. These baseline changes were exacerbated in response to induced hypoglycaemia with increased APP and decreased SNCA in both T2D and controls. The other measured proteins did not differ at baseline between the cohorts. In the post-hypoglycaemia period, APCS, MAPT and APOA1 decreased, and APOE3 and PAPPA increased in T2D only, while noggin levels decreased in both cohorts, and SAA levels remained unchanged. Taken together, the circulating protein levels reported here suggest that subjects with T2D have an increased risk of the development of AD, and this risk may be exacerbated by hypoglycaemia.

ACKNOWLEDGMENTS
We thank the research nurses at the Diabetes Research Centre, Hull Royal Infirmary, for helping with blood sample collection. SLA 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. No funding was received to undertake this work. 61. Plummer S, Van den Heuvel C, Thornton E, Corrigan F, Cappai R. The neuroprotective properties of the amyloid precursor protein following traumatic brain injury. Aging Dis. 2016;7:163-179.

SUPPORTING INFORMATION
Additional supporting information may be found online in the Supporting Information section at the end of this article.