Increased Levels of Antigen-Bound β-Amyloid Autoantibodies in Serum and Cerebrospinal Fluid of Alzheimer’s Disease Patients

Recent studies have suggested a protective role of physiological β-amyloid autoantibodies (Aβ-autoantibodies) in Alzheimer’s disease (AD). However, the determination of both free and dissociated Aβ-autoantibodies in serum hitherto has yielded inconsistent results regarding their function and possible biomarker value. Here we report the application of a new sandwich enzyme-linked immunosorbent assay (ELISA) for the determination of antigen-bound Aβ-autoantibodies (intact Aβ-IgG immune complexes) in serum and cerebrospinal fluid (CSF) of a total number of 112 AD patients and age- and gender-matched control subjects. Both serum and CSF levels of Aβ-IgG immune complexes were found to be significantly higher in AD patients compared to control subjects. Moreover, the levels of Aβ-IgG complexes were negatively correlated with the cognitive status across the groups, increasing with declining cognitive test performance of the subjects. Our results suggest a contribution of IgG-type autoantibodies to Aβ clearance in vivo and an increased immune response in AD, which may be associated with deficient Aβ-IgG removal. These findings may contribute to elucidating the role of Aβ-autoantibodies in AD pathophysiology and their potential application in AD diagnosis.


Introduction
Alzheimer's disease (AD) is the most prevalent form of dementia among the aging population. Its long preclinical phase and the lack of biomarkers that would allow an early diagnosis pose great challenges for the development of effective therapeutic approaches. The neuropathology of AD is characterized by the accumulation of intracellular neurofibrillary tangles and extracellular betaamyloid (Ab) plaques, associated with axonal, dendritic and synaptic degeneration [1][2][3][4]. Several species of aggregated Ab, such as small oligomers, annular oligomers and fibrils, precede the formation of amyloid plaques in the AD brain. The small Ab oligomers, consisting of 3-50 monomer units, appear to be the most neurotoxic species [5].
In transgenic mouse models of AD, both active immunization with full-length Ab peptides or Ab fragments [6][7][8][9] and passive immunization with monoclonal anti-Ab-antibodies [10][11][12] were effective in preventing Ab-aggregation, clearing amyloid plaques and improving cognitive performance. Based on the promising preclinical results, immunotherapy has been proposed as a possible therapeutic approach for AD [13,14]. A phase II multicenter clinical trial of active immunization with preaggregated Ab42 (AN1792(QS-21) vaccine) showed a reduction of amyloid plaque burden and slower cognitive decline in AD patients. However, the trial was interrupted due to the occurrence of meningoencephalitis in some of the immunized participants [14,15]. A follow-up study of the AN1792 clinical trial with yearly assessments and postmortem neuropathological examinations indicated progression of AD-related neurodegeneration and cognitive decline, despite vaccination [16]. Another study reporting the clinical effects of a phase IIa immunotherapeutic trial of AN1792 showed similar results, but also revealed a significantly higher score in one of the neuropsychological test batteries in antibody responders compared to the placebo group, suggesting that Ab-immunotherapy may be useful for the treatment of AD [17]. Several clinical trials are carried out to further evaluate the therapeutic potential of Abbased active immunization and to assess the effect of passive immunization with anti-Ab-antibodies in AD patients [18]. Two phase 3 clinical trials designed to evaluate the efficacy and safety of a humanized N-terminal anti-Ab monoclonal antibody, Bapineuzumab, in patients with mild to moderate AD have recently been completed (http://clinicaltrials.gov). Results presented at the 16 th EFNS congress in Stockholm showed that the treatment with Bapineuzumab did not reach clinical endpoints (no significant benefit on cognitive or functional performance); however, reduced CSF levels of phospho-tau were observed in the Bapineuzumabtreated group (http://www.stevenderoover.be/EFNS/ Presentations/EFNS2012/WC220/; http://www.stevenderoover. be/EFNS/Presentations/EFNS2012/WC219/). Considering the difficulty to find an efficient treatment that would improve the cognitive functions of AD patients, a promising approach would be the administration of potential drugs (e.g., antibodies) at the earliest possible stage, before or just after the onset of AD symptoms, in order to prevent the disease progression [19].
Recently, physiological antibodies binding Ab (Ab-autoantibodies) have been detected in serum and CSF of AD patients and healthy individuals [20][21][22][23][24], as well as in intravenous immunoglobulin preparations (IVIg), which are fractionated blood products used for the treatment of immune deficiencies and other disorders [25]. Dodel et al. [26] reported that administration of Ab-autoantibodies led to reduced plaque formation and improvement of behavior in a mouse model of AD. Moreover, in AD patients, promising effects on cognition were observed in small pilot trials involving passive immunization with IVIg [20,23,27]. These findings suggest that Ab-autoantibodies might exert a protective function against AD and could play an important role in AD treatment.
In addition to their potential therapeutic applications for AD, the biomarker value of Ab-autoantibodies was also investigated. Currently available data on the serum/plasma levels of Abautoantibodies in AD patients compared to healthy individuals are controversial. Several groups found that the serum levels of free, non-antigen-bound Ab-autoantibodies were lower in AD patients than in controls [28][29][30], while others reported either higher values [31] or no difference [32,33]. So far, there is only one reported study on the CSF levels of free Ab-autoantibodies, showing decreased values in AD patients compared to control subjects [34]. Gustaw et al. [35] suggested that the presence of Abautoantibodies not only in free, non-antigen-bound state, but also as preformed immune complexes with Ab-peptides, may be a possible reason for these controversial results. Subsequent serum determinations of Ab-autoantibodies after acidic dissociation of the Ab-immune complexes indicated higher levels of Ab-autoantibodies in AD patients compared to controls [35,36]. However, using a similar procedure, Klaver et al. [37] found no difference between the groups.
Based on the finding that Ab-autoantibodies recognize the Ab (21-37) epitope [38,39], unlike the antibodies produced by active immunization that bind the Ab (4-10) epitope [40], we have recently developed a sandwich ELISA for the determination of intact Ab-IgG immune complexes and applied it for the analysis of serum samples from healthy individuals aged 18 to 89 years. The serum levels of Ab-IgG immune complexes were not correlated with age or cognitive performance of healthy adults [41]. To date, there are no other reports on the determination of Ab-IgG immune complexes in serum or CSF by ELISA.
In the present study, we have employed the sandwich ELISA to determine the levels of Ab-IgG immune complexes in both serum and CSF of AD patients and age-and gender-matched control subjects and evaluated their correlations with the neuropsychological performance and age of the study participants, as well as their diagnostic power.

Ethics Statement
This study was approved by the ethics committee of the University of Ulm, Germany, and conducted according to the guidelines outlined in the Declaration of Helsinki. Prior to participation, written informed consent was obtained.

Participants
Demographic data are depicted in Table 1. Altogether, 58 AD patients were recruited at the Memory Clinic of the Hospital for Neurology of the University of Ulm, Germany. Patients underwent a comprehensive clinical neurological examination, a routine blood analysis, structural imaging (MRI or CT), apolipoprotein E (APOE) genotyping and a detailed neuropsychological assessment, including the Mini Mental State Examination test (MMSE, range 0-30 points; [42]) and the Alzheimer's Disease Assessment Scale -Cognitive subscale (ADAS-Cog, range 0-70 errors; e.g., [43]). Probable AD was diagnosed according to NINCDS-ADRDA [44] and DSM-IV-TR criteria [45]. Furthermore, 54 unrelated ageand gender-matched control subjects were recruited at the same site and did not display any cognitive or neurological deficits following thorough clinical and neuropsychological examination.

Determination of Ab42 and Total tau (T-tau) Levels in CSF
The collection of CSF samples by lumbar puncture and the preanalytical processing were performed using a standardized protocol [46]. In brief, CSF samples were collected into polypropylene tubes, centrifuged immediately and stored at 280uC. The CSF levels of total tau (T-tau) were determined using a sandwich ELISA (INNOTESTH hTau Ag, Innogenetics, Belgium), by which both normally phosphorylated and nonphosphorylated tau were detected. The assay was performed according to the manufacturers' instructions and the laboratory reference ranges were as follows: ,200 ng/L and ,300 ng/L for control individuals below 65 and older than 65 years, respectively. The concentrations of total tau in the analyzed CSF samples were estimated from standard curves obtained for each assay. Tau levels .350 ng/L were regarded as indicative of a neurodegenerative process. The analytical sensitivity of the assay was 75 pg/mL, and the intra-assay and inter-assay variations were ,8%. The CSF levels of Ab (1-42) (Ab42) were determined using a commercially available sandwich ELISA kit (INNOTESTH b-amyloid(1-42), Innogenetics, Belgium), according to the protocol supplied with the kit. CSF Ab42 concentrations of the samples were estimated from standard curves obtained for each assay. Ab42 levels below 550 ng/L were regarded as abnormal.
For both serum and CSF determinations, triplicate 3-fold dilutions from a stock solution (7 mg/mL in blocking buffer) of human serum c-globulin (immunoglobulin preparation, Calbiochem, Merck, Darmstadt, Germany) were used as reference, to allow data to be normalized between plates and different experiments. The non-specific binding (NSB) of the IgG preparation and analyte samples was assessed from triplicate wells containing all components except the mAb 6E10. The average OD values, NSB subtraction, standard deviation (SD) and intra2/ inter-assay coefficients of variation (CV) were calculated with the WorkOut2.0 software (Perkin Elmer, Rodgau, Germany). Both serum and CSF determinations of Ab-IgG immune complexes showed intra-assay CVs,10% and inter-assay CVs,15%. For the newly developed CSF ELISA, the cut-off values of the assay, defined as the linearity limits of the reference curve (R 2 .0.97), were as follows: 0.065 min. and 1.129 max. Since there is no unique method for expressing ELISA responses and arbitrary units are derived from absorbance readings, we considered it adequate to present the results of Ab-IgG determinations in serum and CSF as OD values.

Data Analysis
Statistical analysis was performed using the R statistical software package of The R Foundation of Statistical Computing (www.rproject.org; version 2.11.1 for Mac OS X, GUI 1.34 Leopard).
Welch's two-sample t-tests (two-tailed with modified degrees of freedom) were applied to examine differences in demographic and cognitive data between AD patients and controls. Analysis of variance with group as factor and age as covariate were computed in order to investigate differences in the levels of Ab-IgG immune complexes between both groups. Models' residuals were tested for normality using the Shapiro-Wilk normality test. For categorical variables, Pearsons's Chi-squared (x 2 ) test was computed. Pearson's r product moment correlation coefficient was calculated in order to investigate possible associations of serum and CSF levels of Ab-IgG immune complexes with age and neuropsychological performance (MMSE, ADAS-Cog). The diagnostic power of the Ab-IgG immune complexes in serum and CSF was calculated using receiver-operating characteristic (ROC) curve analysis (package Daim and pROC for R; [47,48]). All tests for statistical significance referred to a significance level with a #0.05.

Demographic and Clinical Characteristics of Alzheimer's Disease Patients and Controls
Demographic and clinical characteristics of Alzheimer's disease patients and controls are shown in Table 1 and Table S1 in the Supporting Information. The statistical evaluation indicated a similar distribution of age (t (85) = 0.79, p = 0.43 for serum donors; t (64) = 21.12, p = 0.27 for CSF donors) and gender (x 2 (1) = 0.05, p = 0.82 for serum donors; x 2 (1) = 0.43, p = 0.51 for CSF donors) in the AD and the control group. As expected, AD patients scored lower in the MMSE (t (47) = 214.00, p,0.0001 for serum donors; t (44) = 210.16, p,0.0001 for CSF donors) and committed more errors in the ADAS-Cog neuropsychological test battery (t (37) = 11.30, p,0.0001 for serum donors; t (26) = 6.12, p,0.0001 for CSF donors) than the control subjects. They also presented significantly lower levels of Ab42 (t (69) = 29.39, p,0.0001) and higher levels of T-tau (t (69) = 8.88, p,0.0001) in CSF. Furthermore, an increased incidence of APOE e4 allele was observed in the AD cases (x 2 (1) = 15.26, p,0.0001). In the following paragraphs we compare the levels of Ab-IgG immune complexes in serum and CSF samples from AD patients and age-and gender-matched control subjects. Since old age and APOE e4 status are considered to be associated with an increased risk of AD pathology [49], we also included age as covariate into the group comparison and further investigated potential differences between the levels of Ab-IgG immune complexes in serum and CSF with respect to APOE genotype.

Ab-IgG Immune Complexes in Serum of AD Patients and Control Subjects
Two samples (from one AD patient and one control subject) were excluded from the statistical analysis, since the Ab-IgG levels exceeded the ELISA cut-off values.
Higher levels of Ab-IgG immune complexes were determined in serum of AD patients compared to the controls (F (1,84) = 4.94, p = 0.03; Table 1, Figure 2A). According to ROC curve analyses, the serum Ab-IgG levels discriminate the AD patients from the control subjects with 81% specificity and 44% sensitivity (AUC = 0.63, 95% CI: 0.75-0.51; Figure 2B). When the assay sensitivity was set to 80%, specificity reached a maximum of 33%.
The serum levels of Ab-IgG immune complexes increased with advancing age in the AD patients (r = 0.37, p = 0.01) but not in the controls (r = 20.14, p = 0.38; Figure 3A). Furthermore, they were negatively correlated with the MMSE scores (r = 20.23, p = 0.04; Figure 3B) and positively with the ADAS-Cog scores across groups (r = 0.32, p = 0.02; Figure 3C), i.e., reaching higher values with decreasing cognitive test performance. There was no difference between the serum levels of Ab-IgG immune complexes in the case of APOE e4 (homo-and heterozygotes) and non-APOE e4 carriers, either in the AD or the control group.

Ab-IgG Immune Complexes in CSF from AD Patients and Control Subjects
The levels of Ab-IgG immune complexes were higher in AD patients compared to the controls (F (1,63) = 4.98, p = 0.03; Table 1, Figure 4A). ROC curve analyses indicated 59% specificity and 70% sensitivity (AUC = 0.65, 95% CI: 0.79-0.52; Figure 4B) for the diagnostic discrimination of the assay between AD cases and controls. When specificity of the Ab-IgG determinations was set to 80%, sensitivity reached a maximum of 33%. When sensitivity was set to 80%, specificity reached a maximum of 31%. The ratio of the CSF to serum levels of the Ab-IgG immune complexes showed 82% specificity and 50% sensitivity in ROC curve analysis (AUC = 0.67, 95% CI: 0.83-0.50). When sensitivity was set to 80%, specificity was only 35%. The ROC curve analysis of the CSF Ttau/Ab42 concentration ratio showed 91% specificity and 93% sensitivity (AUC = 0.97, 95% CI: 0.10-0.94).
The CSF levels of Ab-IgG immune complexes across all subjects were negatively correlated with the MMSE scores (r = 20.30, p = 0.02) and positively correlated with the ADAS-Cog test scores (r = 0.48, p = 0.006), increasing with the decline of cognitive performance ( Figure 5A, B). Furthermore, they were positively correlated with the Ab42 concentration in CSF of AD patients (r = 0.35, p = 0.04; Figure S1 in the Supporting Information), but not of control subjects. A highly significant positive correlation was observed across groups between the levels of Ab-IgG immune complexes in CSF and serum (r = 0.54, p = 0.0002; Figure S2 in the Supporting Information). There was no effect of age or APOE e4 (homo-and heterozygote) genotype on the Ab-IgG levels in the CSF of either AD patients or control subjects.

Discussion
AD-related pathological processes start well before the onset of clinical manifestations [2,4,50,51]. Hence, the identification and evaluation of biochemical markers that enable an early diagnosis should be of substantial value. CSF levels of Ab42 and tau protein are currently the only reliable biomarkers for the diagnosis of AD, with sufficient sensitivity and specificity [52], while efforts to establish less invasive blood-derived biomarkers have been hitherto unsuccessful. Reports on potential serum-biomarkers for AD diagnosis have provided contradictory results and it is unclear whether changes in the periphery reflect pathologies within the brain [53][54][55]. Thus, studies of Ab42 levels in serum of AD patients showed both reduced [56,57] and increased values [58,59] compared to control subjects, while other reports indicated no differences between groups [60,61]. Also, no correlation was found between the serum Ab42 levels and the Ab42 levels in CSF of AD patients and healthy individuals [62,63], the accumulation of Ab peptides in AD brain [64] or the progression of cognitive deterioration in AD [65,66]. Furthermore, research focused on blood protein signatures recently revealed that epidermal growth factor (EGF), platelet-derived growth factor (PDG-BB) and macrophage inflammatory protein 1d (MIP-1d) differentiated Ab-IgG Complexes in Serum and CSF of AD Patients PLOS ONE | www.plosone.org AD from control subjects, but not from patients with other types of dementia [67].
Physiological Ab-autoantibodies have been detected in serum and CSF [20][21][22][23][24]27] and generated a high interest as a potential biomarker for AD, however with hitherto inconsistent results. In AD patients compared to controls, the serum levels of free, nonantigen-bound Ab-autoantibodies were found to be reduced [28][29][30]34], enhanced [31] or unchanged [32,33]. Other studies reported increased levels of Ab-autoantibodies after acidic dissociation of preformed Ab-immune complexes in serum of AD patients [35,36]. Nevertheless, these results could not be reproduced by Klaver et al. [37], who found no difference in the levels of dissociated Ab-autoantibodies between AD and control subjects.
Due to the postulated imbalance between Ab production and removal in AD [11], we have evaluated the contribution of Abautoantibodies to Ab-clearance and the diagnostic potential of Ab-IgG immune complexes. Our method could be applied as an alternative or complementary approach to the previously reported direct ELISAs for the determination of total Ab-autoantibodies levels. It does not require additional sample preparation steps such as acidic dissociation and may provide valuable information on possible problems related to antibody avidity and clearance of the immune complexes. Another important aspect is the subtraction of the NSB from the OD response of each sample, a procedure previously reported only in a few ELISA studies of Abautoantibodies [37]. We initially optimized and employed the sandwich ELISA protocol for the analysis of serum samples from healthy adults aged 18-89 years [41]. In the present study, the experimental procedure was also optimized for CSF analysis and applied to determine the levels of Ab-IgG immune complexes in serum and CSF samples from a total number of 112 AD patients and age-and gender-matched control subjects. Ab-IgG immune complexes were detected in all serum and CSF samples, suggesting a contribution of IgG-type Ab-autoantibodies to Ab clearance in vivo. Higher Ab-IgG levels were found in both serum and CSF of AD patients compared to controls, in agreement with two previous studies [35,36] that revealed increased total levels of Ab-autoantibodies in AD patients. An elevated antibody production would be expected in response to Ab accumulation, which is either due to deficient clearance mechanisms [68,69] or to an increased formation of Ab peptides. The latter is mainly the case in familial AD, owing to genetic mutations of amyloid precursor protein (APP) and presenilin 1 and 2, but it can also occur in sporadic AD, where it was suggested to be partially caused by the enhanced  expression and activity of APP cleaving enzyme 1 (BACE 1) [70]. The progression of the disease, despite increased Ab-IgG levels in serum and CSF of AD patients, could indicate defective clearance mechanisms, leading to the accumulation of Ab-IgG immune complexes in AD. In healthy individuals, antigen-bound antibodies are captured by macrophages through Fc receptor-mediated recognition and transferred to mastocytes in liver or spleen for degradation during the process of ''immune adhesion'', which is regulated by antibody avidity [10,71]. A possible explanation for the apparent clearance deficiency of Ab-IgG immune complexes is provided by the observations of Jianping et al. [72] who found the avidity of Ab-autoantibodies to be lower in AD patients than in healthy controls and suggested that this could impair the removal of Ab-IgG immune complexes by macrophages.
Our results further revealed that serum and CSF levels of Ab-IgG immune complexes were negatively correlated with the cognitive performance of the study participants. Thus, subjects with higher Ab-IgG levels had weaker performances during MMSE screening and ADAS-Cog neuropsychological testing. The increased levels of Ab-IgG immune complexes and their inverse correlation with the cognitive status would point to a pathological process, potentially associated with defective clearance mechanisms, as discussed above. Thereby, the reported cognitive improvements of AD patients treated with IVIg [20,23,27] might be partially attributed to the replacement of deficient Abautoantibodies by passive immunization.
In agreement with our previous work [41], serum Ab-IgG levels were not correlated with the age of control subjects. In the AD group, however, increased age was associated with higher levels of Ab-IgG immune complexes in serum and might therefore represent a factor for reduced Ab clearance in AD. A positive correlation with age was also reported by Gustaw-Rothenberg et al. [36] for the difference values between the Ab-autoantibody levels before and after acidic dissociation of the Ab-IgG immune complexes, which might be comparable with the levels of intact Ab-IgG immune complexes.
As shown in Figure S1 in the Supporting Information, our results also indicated a positive correlation between the CSF levels of Ab-IgG immune complexes and Ab42 peptide. Furthermore, we found a strong correlation across groups between the serum and CSF levels of Ab-IgG immune complexes. Considering the dilution factors applied in ELISA, the Ab-IgG levels were approximately 100 fold lower in CSF than in serum, suggesting that the Ab-autoantibodies are produced and bind to Ab mainly in the periphery.
In summary, we report here for the first time the determination of intact Ab-IgG immune complexes in serum and CSF of AD patients and age-and gender-matched control subjects, employing a sandwich ELISA approach. Our results showed higher serum and CSF levels of Ab-IgG immune complexes in AD patients relative to controls; however, due to the variability within groups leading to overlapping values, the Ab-IgG levels displayed only moderate discrimination powers in ROC analyses. A possible application of serum Ab-IgG immune complexes for AD diagnosis in a panel of blood-derived biomarkers remains to be further tested. Nevertheless, the correlation of both serum and CSF Ab-IgG levels with the cognitive status across groups represents a valuable characteristic and it would be interesting to assess their potential use for predicting conversion to AD or evaluating the efficacy of therapeutic interventions in AD (e.g., passive immunization with intravenous immunoglobulin preparations containing Ab-autoantibodies).
Our findings additionally suggest an increased immune response in AD, presumably associated with deficiencies in the clearance of Ab-IgG immune complexes. A better understanding of the mechanisms causing the apparent accumulation of Ab-IgG immune complexes could reveal in future studies new approaches for diagnosis or targeted treatment of AD.