Complement dysregulation and Alzheimer's disease in Down syndrome

Abstract Introduction Down syndrome (DS) is associated with immune dysregulation and a high risk of early onset Alzheimer's disease (AD). Complement is a key part of innate immunity and driver of pathological inflammation, including neuroinflammation in AD. Complement dysregulation has been reported in DS; however, the pattern of dysregulation and its relationship to AD risk is unclear. Methods Plasma levels of 14 complement biomarkers were measured in 71 adults with DS and 46 controls to identify DS‐associated dysregulation; impact of apolipoprotein E (APOE) ε4 genotype, single nucleotide polymorphisms (SNPs) in CLU and CR1, and dementia on complement biomarkers was assessed. Results Plasma levels of complement activation products (TCC, iC3b), proteins (C1q, C3, C9), and regulators (C1 inhibitor, factor H, FHR4, clusterin) were significantly elevated in DS versus controls while FI and sCR1 were significantly lower. In DS with AD (n = 13), C3 and FI were significantly decreased compared to non‐AD DS (n = 58). Neither APOE genotype nor CLU SNPs impacted complement levels, while rs6656401 in CR1 significantly impacted plasma sCR1 levels. Conclusions Complement is dysregulated in DS, likely reflecting the generalized immune dysregulation state; measurement may help identify inflammatory events in individuals with DS. Complement biomarkers differed in DS with and without AD and may aid diagnosis and/or prediction. Highlights Complement is significantly dysregulated in plasma of people with DS who show changes in levels of multiple complement proteins compared to controls. People with DS and dementia show evidence of additional complement dysregulation with significantly lower levels of C3 and factor I compared to those without dementia. rs6656401 in CR1 was associated with significantly elevated sCR1 plasma levels in DS.

• People with DS and dementia show evidence of additional complement dysregulation with significantly lower levels of C3 and factor I compared to those without dementia.
• rs6656401 in CR1 was associated with significantly elevated sCR1 plasma levels in DS.

BACKGROUND
Down syndrome (DS), caused by triplication or partial triplication of chromosome 21, is the most common chromosomal abnormality in humans, affecting 6 million people worldwide. 1 In addition to intellectual disability, DS predisposes individuals to autoimmune diseases and several chronic pathologies, including congenital heart defects, and ophthalmic diseases. 2 Although advances in medicine have substantially improved outcome in recent years, these multiple pathologies impact both life expectancy and quality of life for people with DS. 3 Life expectancy remains low, now averaging 60 to 65 years, almost 20 years below that of the general population. With increasing longevity, numerous age-related comorbidities have been identified in DS. 4,5 By the age of 40, almost all individuals affected by DS display typical Alzheimer's disease (AD) neuropathology, notably the presence of amyloid beta (Aβ) plaques, a pathological hallmark of AD, 6,7 associated with early onset dementia. 8 Accelerated amyloid accumulation is likely a consequence of triplication of the amyloid precursor protein (APP) gene located on chromosome 21, leading to increased APP expression and rapid and sustained deposition of amyloid protein in the brain. 8 Additionally, neurofibrillary tau has been shown to accompany Aβ deposits in DS brains as early as Braak stage I. 9 DS is now recognized as an immune dysregulation disorder, with anomalies affecting both the innate and adaptive immune systems. 10 Children with DS frequently present with autoimmune conditions, are more likely to get severe infections, and spend more time in hospital as a consequence. 11 Aberrant neuroinflammatory processes in DS likely contribute to the development of AD-like pathology, neurodegeneration, and dementia. 12 Complement, a core part of the innate immune system, comprises >30 factors in plasma and on cells; these cooperate in a complex network to recognize and eliminate a variety of pathogens. Activation on pathogens, debris, or damaged cells triggers an enzymatic cascade that generates opsonic and pro-inflammatory products; these facilitate elimination of pathogens and debris but can, if dysregulated, also drive pathology. 13 There is a large body of evidence implicating complement dysregulation as a driver of neuroinflammation in AD. Aggregated Aβ directly activates the complement cascade by interacting with C1q, 14 and several studies have shown C1q immunoreactivity colocalizing with Aβ in amyloid plaques in AD brain tissue. 15,16,17 Similar patterns of C1q, C3 fragments, clusterin, and activated microglia localizing to amyloid plaques were reported in brains from DS individuals as young as 15 years old. 15,18 Complement biomarker studies in AD cerebrospinal fluid (CSF) and plasma have provided additional evidence of complement dysregula-tion and shown that complement activation products and proteins can aid diagnosis and predict progression of AD. 19,20 Additionally, single nucleotide polymorphisms (SNPs) in CLU and CR1 were significantly associated with AD risk in genome-wide association studies (GWAS). 21 The AD risk SNP rs6656401 in CR1 was strongly associated with CR1 length polymorphism, minor allele carriers expressing the longer form of CR1. 22

Ethics
Ethical approval for the London Down Syndrome Consortium (Lon-

Subjects
Plasma samples for complement analysis from individuals with DS (n = 71) were obtained from the LonDownS, a prospective longitudinal study following adults with DS. 26 Demographic and health data including body mass index (BMI) were collected contemporaneously.

Measurement of complement proteins by enzyme-linked immunosorbent assay
Fourteen complement components, regulators, and activation products were measured by sandwich enzyme-linked immunosorbent assay (ELISA) using commercial or in-house produced antibodies. Antibodies and proteins used in the ELISA are listed in Table S1 in supporting information. Plasma samples stored at -80 • C were thawed immediately prior to use in ELISA, vortexed briefly, diluted in 0.2% bovine serum albumin (BSA) in phosphate-buffered saline containing 0.05% Tween-20 (PBST), and kept on ice until use. Activation products (TCC, iC3b, C5a) were measured immediately after thawing to avoid in vitro complement activation, while other proteins were measured in either undiluted or prediluted samples kept on ice and measured within 24 h of thawing. Standard proteins were purified by affinity chromatography from healthy donor serum/plasma and purity confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis. Protein concentrations were determined by bicinchoninic acid assay. In-house monoclonal antibodies (mAbs) were produced using hybridoma technology; mAb were tested for specificity in ELISA and western blotting. Sandwich ELISAs were developed in house for quantification of complement activation products, regulators, and proteins in plasma samples; they were evaluated for sensitivity, reproducibility, and intra- Drug Administration Q2(R1) Validation Guidelines. 27 Purified protein was spiked into the protein-depleted plasma at three known concentrations (i.e., 50, 100, and 150 μg/ml) and concentration measured in the ELISA to determine accuracy, repeatability, linearity, and range.
Assay reproducibility was calculated using a published method and template. 27 Assay performance was assessed by taking multiple measures from independently diluted aliquots of the same plasma sample set; the intra-assay CV, calculated as described, 28 was less than 10% and inter-assay CV (n = 3) was between 1% and 14% in the different assays (Table S1).  When a sample qualified as an outlier (1.5 times higher than the third quartile or 1.5 times lower than the first quartile in the interquartile range of the whole cohort), this was remeasured at a dilution that permitted interpolation on the standard curve. Control and DS samples were analyzed separately on four ELISA plates. The same two standard plasma samples were included on each plate in each assay to control for between-plate variation; all inter-assay CVs were < 15% (Table S1).

Data analysis and statistics
Data were analyzed by constructing a 7-to 10-point standard curve

Plasma levels of complement proteins are elevated in DS
Complement proteins were quantified by ELISA in DS and control subjects. Among the measured components, C1q, C3, and C9 were significantly elevated in DS compared to control (p < 0.05, Figure 1A). Two of the measured activation products TCC and iC3b, were also significantly increased in DS plasma (p < 0.001, Figure 1B). Plasma levels of the complement regulators clusterin, FH, C1 inhibitor, and FHR4 were significantly elevated in DS (p < 0.001), while FI and sCR1 were significantly decreased (p < 0.01, Figure 1C). There were no significant changes in levels of C4, C5a, or FHR125 in DS compared to controls.
Within the DS group, several significant positive correlations were identified between complement proteins and are shown in Figure 2A (C3 and clusterin, FH; C1q and C1; TCC and C5a). Within the control group, a few significant positive correlations were identified between complement proteins and are shown in Figure 2B (C9 and FI, FH; clusterin and C4, C1 inhibitor). All other correlations are shown in Figure   S1 in supporting information. Regarding demographic factors, C3 levels significantly correlated with BMI (Pearson r = 0.26, p < 0.05, Table   S2), and FHR4 was significantly higher in women with DS, while TCC was significantly lower (Table S2).

Complement protein levels are altered in DS with AD
Out of 71 individuals with DS, 13 were classified as having dementia at the time of sampling; as expected, the DS with AD group were significantly older. We tested whether any changes in complement protein levels occurred in DS with AD compared to DS without AD. Levels of C3 (p < 0.01) and FI (p < 0.05) were significantly decreased in DS with AD ( Figure 3). No other significant differences in levels of complement biomarkers between the groups were identified. Clusterin levels were reduced and C1q levels increased in DS with AD but these did not reach significance ( Figure 3).

The rs6656401 SNP is associated with increased sCR1 levels in DS
Plasma levels of sCR1 were compared between DS homozygous for the major allele at CR1 SNP rs6656401 (G) and carriers of the minor allele (A); GG donors had significantly lower sCR1 levels in (A) (B)

F I G U R E 2 Correlation scores and P-values between complement markers in the Down syndrome (DS) group (A) and control group (B)
plasma compared to GA/AA donors (mean = 11.8 vs. 14.6 ng/ml, p < 0.05; Figure 4A). The CR1 SNP rs669117 had no significant effect on plasma sCR1 levels ( Figure 4B), and the CLU SNP rs1113600 was not associated with changes in plasma clusterin levels in the DS cohort ( Figure 4C).
APOE genotyping was available on 63 of the DS subjects; from these we identified 16 heterozygotes for the ε4 allele, 2 of whom had AD. No significant changes in complement proteins were found in carriers of the ε4 allele compared to non-carriers ( Figure 4D).

DISCUSSION
Despite DS may also contribute to increased C1q plasma levels. 33,34 We also found that C3, the most abundant complement protein, and C9, essential for terminal pathway activity, were significantly increased in DS plasma; C3 is an acute phase reactant so elevated C3 may also be a consequence of inflammation. C3 levels are also increased in obesity, likely a consequence of obesity-driven inflammation; 35 notably, a recent study reported elevated C3 and C4 levels in adolescents with DS that strongly correlated with obesity. 23 We also identified a significant positive correlation with BMI for plasma levels of C3, but not C4, in the DS group ( Figure S1).
Regulatory mechanisms have evolved to prevent complement damage to self-cells during responses to foreign pathogens or other activators. 36 We quantified plasma levels of C1 inhibitor, FI, sCR1, clusterin, FH, and FH-related proteins and found striking differences between the DS and control groups. C1 inhibitor, a critical regulator of the classical pathway that inactivates the C1 complex, was significantly elevated in DS compared to controls (mean 150 vs. 118 μg/ml) and correlated positively with C1q levels. Interestingly, higher C1 inhibitor levels were also reported in plasma from mothers carrying fetuses with DS compared to carriers of normal fetuses. 37  terin as a biomarker for late-onset AD. 38,20 The CLU gene has also been implicated in AD through GWAS 21,39 with the minor T allele at rs11136000 associated with better cognitive function. 24 In the DS cohort, this CLU SNP was not associated with a significant difference in clusterin plasma levels. FH is a plasma regulator of the alternative pathway C3 convertase while FHR4 is a FH-related plasma protein that controls the activity of FH; levels of both were significantly elevated in DS plasma, with female subjects showing significantly higher levels compared to males in the DS group. In contrast, another study found increased FH levels in brain, liver, and spleen tissue in subjects with DS; however, this involved post mortem tissue from five subjects considerably older than our cohort. 40 Levels of FI, an enzyme that provides powerful regulation of complement by inactivating complement convertases, were significantly decreased in DS plasma. Taken together, these findings support a substantial dysregulation of complement, primarily impacting the alternative pathway convertases. CR1 is a cell surface receptor for the activation fragments C3b and C4b; CR1 is cleaved from expressing cells under some circumstances and the released sCR1 can be measured in plasma. Elevated plasma levels of sCR1 have been reported in AD and suggested to have functional relevance. 41 In our study sCR1 levels were significantly reduced in DS plasma compared to controls. Genetic variation in CR1 has been shown to significantly contribute to AD risk and influence plasma sCR1 levels. 21 The minor allele at rs6656401 was associated with expression of the long form of CR1, 22 more rapid decline of cognitive function, 42 and increased sCR1 plasma levels. 43   and FI were significantly lower in the dementia group, though whether this is cause or consequence of the pathology is unclear. A recent report described significantly decreased CSF C3 levels in AD patients positive for Aβ, tau, and neurodegeneration markers. 45 Others have reported association of low plasma C3 levels with increased risk for AD, particularly in APOE ε4/ε4 carriers. 46 These findings suggest that AD pathology consumes C3 in the AD brain resulting in lower plasma levels; the demonstration that C3-deficient APP transgenic mice displayed increased Aβ burden at 17 months compared to C3-sufficient APP mice supports this mechanism. 47 Additionally, C3 processing in the periphery may be accelerated by AD-related comorbidities. Low levels of FI, linked to specific SNPs in the FI gene, are strongly associated with risk of age-related macular degeneration (AMD). 48 The association of low plasma FI with AMD has been ascribed to overactivation of complement in the retina; although we did not test the FI SNP as a cause of low FI in the DS cohort, the same pro-inflammatory consequences are likely. Clusterin has been reported as a biomarker for late-onset AD in several studies; 49,50 however, there was no significant difference in clusterin levels in demented compared to non-demented subjects with DS. Notably, while biomarker studies have implicated clusterin as a late-onset sporadic AD biomarker, dementia in DS is an early onset genetically determined type of AD; no published studies address clusterin levels in familial AD.

CONCLUSIONS
Our study demonstrates that complement is dysregulated in DS.
Plasma levels of activation products and key components were significantly elevated compared to controls; regulators showed significant changes compatible with a failure of regulation in the activation and terminal pathways. Subjects with DS and AD showed a distinct complement profile compared to non-AD individuals, prompting further studies into complement system involvement in DS and the development of AD. While complement biomarkers provide clear evidence of immune dysregulation, they may not be sufficient to differentiate individuals with DS who will rapidly progress to dementia; they may, however, complement AD-specific biomarkers to enhance risk prediction or early diagnosis of dementia in people with DS.

ACKNOWLEDGMENTS
Thank you to all the participants of the LonDownS study and their families and caregivers. We also thank our NHS network of sites that helped to identify participants. Recruitment support and data collec-