Serum Anti-AP3D1 Antibodies Are Broad-Spectrum Biomarkers for Atherosclerosis, Acute Ischemic Stroke, Cardiovascular Disease, Diabetes Mellitus, Chronic Kidney Disease, and Digestive Organ Cancer

Background: Atherosclerosis has been considered as the main cause of morbidity, mortality, premature incapacity, and disability worldwide. For early and sensitive diagnosis, development of novel biomarkers is expected and of signicant practical importance. Methods: The rst screening for antigen markers was conducted using the serological identication of antigens by recombinant cDNA expression cloning (SEREX). Serum antibody levels were examined using the amplied luminescent proximity homogeneous assay-linked immunosorbent assay (AlphaLISA) using a recombinant protein as an antigen. Results: SEREX screening has identied adaptor-related protein complex 3 subunit delta 1 (AP3D1) as an antigen recognized by serum IgG antibodies of patients with atherosclerosis. As per the results of AlphaLISA, it was determined that the serum antibody levels against AP3D1 were higher in patients with acute ischemic stroke (AIS), transient ischemic attack (TIA), diabetes mellitus (DM), cardiovascular disease (CVD), chronic kidney disease (CKD), esophageal squamous cell carcinoma (ESCC), and colorectal carcinoma than those in the healthy donors. The area under the curve values of DM, nephrosclerosis type of CKD, and ESCC calculated using receiver operating characteristic curve analysis were noted to be higher than that of other diseases. Correlation analysis showed that the anti-AP3D1 antibody levels were highly associated with maximum intima-media thickness, which indicates that this marker reected the development of atherosclerosis. The Japan Public Health Center-based Prospective Study results have veried that the serum antibody levels against AP3D1 showed signicantly higher odds ratios with the risk of AIS for persons with the third quartiles and the highest quartiles versus the lowest quartile, indicating that this antibody marker is deemed useful as risk factors for AIS. Conclusions: Serum anti-AP3D1 antibodies, which are broad-spectrum biomarkers of atherosclerotic diseases and digestive organ cancers, could be useful in predicting the onset of AIS.

In this study, using the SEREX method screening, adaptor-related protein complex 3 subunit delta 1 (AP3D1) was identi ed as a target antigen recognized by serum IgG antibodies in the sera of patients with atherosclerosis. The levels of serum anti-AP3D1 antibodies in patients with atherosclerosis-related diseases were then examined.

Patients and HDs' sera
This study was approved by the Local Ethical Review Board of Chiba University Graduate School of Medicine (Chiba, Japan) and the review boards of the participating hospitals.
Blood samples were collected from patients who had provided their written informed consent. Each serum sample was centrifuged at 2000 × g for 10 min at 4 °C, and the supernatant was stored at − 80 °C until use. Repeated thawing and the freezing of samples were avoided.
Sera of patients with DM and CVD were obtained from Chiba University Hospital, and patients with AIS and transient ischemic attack (TIA) were provided by the Chiba Prefectural Sawara Hospital, Chiba Rosai Hospital, and Chiba Aoba Municipal Hospital. Serum samples associated with AIS, TIA, and CVD were obtained within 2 weeks after disease onset. Sera of chronic kidney disease (CKD) patients were obtained from the Kumamoto cohort [22,23], and Chiba University Hospital provided the serum samples of patients with ESCC and colorectal carcinoma (CRC). Sera of HDs were obtained from Chiba University Hospital, Port Square Kashiwado Clinic, and Chiba Prefectural Sawara Hospital. Sample of HDs from Port Square Kashiwado Clinic and Chiba Prefectural Sawara Hospital were selected from individuals who exhibited no abnormalities in cranial magnetic resonance imaging.

Sequence analysis of identi ed clones
Monoclonal phage cDNA clones were converted to pBluescript phagemids by in vivo excision using the ExAssist helper phage (Stratagene; Agilent Technologies, La Jolla, CA). Plasmid DNA was obtained from the E. coli SOLR strains transformed by the phagemids. Homology search of the inserted and sequenced cDNAs using a public database provided by the the National Center for Biotechnology Information (https://blast.ncbi.nlm.nih.gov/Blast.cgi) identi ed the genes.

AP3D1 protein expression, extraction, and puri cation
The region of 2490-4347 of the AP3D1 gene was isolated and was recombined into the EcoRI/NotI site of pGEX-4T-1 (GE Healthcare Life Sciences, Pittsburgh, PA), followed by con rmation by DNA sequencing.
The E. coli BL-21 transfected with pGEX-4T-1-AP3D1 was then treated with 0.1 mM IPTG at 37 °C for 3 h to induce the expression of cDNA products. Then, the cells were lysed in BugBuster Master Mix (Merck KGaA, Darmstadt, Germany). With the Glutathione Sepharose (GE Healthcare Life Sciences, Pittsburgh, PA) column chromatography according to the manufacturer's instructions, glutathione S-transferase (GST)-fused-AP3D1 protein was puri ed, as has been described previously [14,15,18,27].

JPHC cohort analysis
The Japan Public Health Center-based Prospective Study (JPHC) cohort analysis was performed using the above AlphaLISA detection antibody levels. The study nested within JPHC [29,30], involving approximately 30,000 Japanese individuals aged 40-69 years at the baseline period of 1990-1994 whose plasma were stored. The antibody levels of AP3D1 protein were then measured in 202 cases of AIS in the cohort developed between the baseline and 2008 and in 202 controls whose age, sex, and area were matched with the cases. A conditional logistic regression model was used to estimate the odds ratios (ORs) and 95% con dence intervals (CIs) for cerebral infarction with respect to the antibody levels of AP3D1 protein.

Statistical analysis
All statistical analyses were conducted using GraphPad Prism 5 (GraphPad Software, La Jolla, CA). The Kruskal-Wallis test (Mann-Whitney U test with Bonferroni's correction applied) was used to evaluate differences among > 3 groups, and the Mann-Whitney U test was employed to determine signi cant differences between the two groups. Correlations were calculated using Spearman's correlation analysis and logistic regression analysis. The predictive values of the putative disease markers were assessed using a receiver operating characteristic (ROC) curve analysis, and the cutoff values were set to maximize the sums of sensitivity and speci city. All tests were two-tailed, and P-values lower than 0.05 were considered to be statistically signi cant.

Results
Initial screening of AP3D1 antigens using SEREX As an initial SEREX screening, sera of patients with atherosclerosis were used to search for antigens that could be recognized by serum IgG antibodies, one of which was AP3D1 (accession no. NM_003938.8).
The region of 2490-4347 of the AP3D1 gene was then isolated and recombined into the EcoRI/NotI site of pGEX-4T-1, followed by con rmation by DNA sequencing. The cDNA was then expressed in E. coli, puri ed by a nity chromatography, and employed as an antigen in order to examine the serum antibody levels.
The levels of anti-AP3D1 antibodies were elevated in patients with AIS and TIA The serum anti-AP3D1-antibody (s-AP3D1-Ab) levels in patients with AIS and TIA were examined using AlphaLISA. AIS and TIA sera were provided by Chiba Prefectural Sawara Hospital, Chiba Rosai Hospital, and Chiba Aoba Municipal Hospital, whereas samples of HDs were obtained from Chiba University, Port Square Kashiwado Clinic, and Chiba Prefectural Sawara Hospital. The average ages [± standard deviations (SDs)] of the HDs and patients with AIS and TIA were 52.33 ± 8.92, 57.99 ± 7.97, and 69.45 ± 11.64 years, respectively (Table 1a). The levels of s-AP3D1-Abs were determined to be signi cantly higher in patients with AIS and TIA than those in HDs (Fig. 1a). At a cutoff value equivalent to the average plus two SDs of the HD values, the s-AP3D1-Ab-positive rates in HDs and patients with AIS and TIA were 2.4%, 10.1%, and 10.4%, respectively (Table 1b). ROC curve analysis revealed that the area under the curve (AUC) values for s-AP3D1-Abs vs. AIS and vs. TIA were 0.616 and 0.662, respectively (Figs. 1b and c). No signi cant difference was found in the positive rates and the AUC values between AIS and TIA.
Elevation of s-AP3D1-Abs levels in patients with DM The levels of s-AP3D1-Abs were also examined for DM. Sera of HD were obtained from Chiba University, whereas the sera of patients with DM were provided by the Chiba University Hospital. The average ages (± SDs) of the HDs and patients with DM were 45.20 ± 10.95 and 63.12 ± 12.04 years, respectively. The AlphaLISA results revealed that s-AP3D1-Ab levels were signi cantly higher in patients with DM than in the HDs (Fig. 2a). When the positive samples for which the AlphaLISA counts exceeded the cutoff value were scored, the positive rates of s-AP3D1-Abs in the HDs and the patients with DM were 3.7% and 41.8%, respectively (Table 2b). The AUC value of s-AP3D1-Abs vs. DM was as high as 0.791 (Fig. 2b). Therefore, it can be concluded that the s-AP3D1-Ab levels were closely associated with DM.
The s-AP3D1-Abs levels were associated with CVD For the next step, the antibody levels in samples from CVD patients were examined. The samples of CVD patients were obtained from Chiba University Hospital, and those in HDs were from Chiba University, Port Square Kashiwado Clinic, and Chiba Prefectural Sawara Hospital. The average ages (± SDs) of the HDs and CVD patients were 45.27 ± 11.20 and 66.07 ± 11.32 years, respectively. Compared with HDs, s-AP3D1-Abs levels were signi cantly higher in patients with CVD ( Fig. 3a), and the s-AP3D1-Ab positivity rates in HDs and patients with CVD were 5.1% and 24.0%, respectively (Table 3). ROC curve analysis revealed that AUC of s-AP3D1-Abs for CVD was 0.758 (Fig. 3b).
The s-AP3D1-Ab levels were closely related to CKD The antibody levels in the sera of CKD patients were examined, which were assumed to be closely related to atherosclerosis. The sera of the CKD patients were obtained from the Kumamoto cohort [22,23], including 145 from patients with diabetic kidney disease (type 1 CKD), 32 from patients with nephrosclerosis (type 2 CKD), and 123 from patients with glomerulonephritis (type 3 CKD). The sera of HDs (82 specimens) were obtained from Chiba University, Chiba Prefectural Sawara Hospital, and the National Hospital Organization of Shimoshizu Hospital (Table 4a). Patients from all three groups of CKD were found to have signi cantly higher levels of s-AP3D1-Abs compared to that in HDs (Fig. 4a). The s-AP3D1-Ab-positive rates in HDs and patients with type 1, type 2, and type 3 CKD were 4.9%, 27.6%, 37.5%, and 22.8%, respectively (Table 4). ROC curve analysis revealed AUC of s-AP3D1-Abs of type 1, type 2, and type 3 CKD to be 0.791, 0.874, and 0.735, respectively (Figs. 4b, c, and d). Type 2 CKD showed the highest AUC value among all diseases examined.
Association of s-AP3D1-Ab levels with ESCC or CRC The s-AP3D1-Abs levels were also measured in serum samples from the HDs and patients with ESCC or CRC. AlphaLISA results revealed that s-AP3D1Ab levels were signi cantly higher in patients with ESCC and CRC than in HDs (Fig. 5a). The positivity rates of s-AP3D1-Abs in HDs and patients with ESCC and CRC were 3.1%, 42.2%, and 15.6%, respectively ( Table 5). The AUC values were 0.872 and 0.743 for ESCC and CRC, respectively (Figs. 5b and c).

Correlation analysis
Correlation analysis of s-AP3D1-Ab levels and subject data was performed using 635 specimens from Chiba Prefectural Sawara Hospital, including 139 samples from HDs, 121 from patients with deep and subcortical white matter hyperintensity, 17 from patients with asymptomatic cerebral infarction, 43 from patients with TIA, 228 from patients with AIS, 57 from patients with chronic-phase cerebral infarction (cCI), and 30 from other diseases. Using the Mann-Whitney U test, the s-AP3D1-Ab levels were compared between participants with body mass index (BMI) < 25 and BMI ≥ 25; participants with or without diseases of DM, hypertension (HT), CVD, and dyslipidemia; and between those patients who were smokers or nonsmokers and those who consumed alcohol or not. The analysis showed that the s-AP3D1-Ab levels were signi cantly higher in the subjects with HT than in those without HT and those with DM than without DM (Table 6). Conversely, no signi cant differences in s-AP3D1-Ab levels were observed in the other categories.
Performing a logistic regression analysis of the predictors for AIS using the results of the Sawara Hospital was also considered, which included 139 samples from HDs and 228 from patients with AIS. An elevated s-AP3D1-Ab level was associated with an increased risk of AIS as shown by the univariate logistic regression analysis (P < 0.0001). A multivariate logistic regression analysis has identi ed age, HT, and DM, but not s-AP3D1-Ab, as independent predictors of AIS (Table 7).
Next, correlation analysis was performed using a Spearman application in order to determine the correlation between s-AP3D1-Ab levels and subject parameters, including general information such as age, body height, weight, BMI, and the degree of artery stenosis [the maximum intima-media thickness (max IMT)]. The serum s-AP3D1-Ab levels were determined to be closely correlated with age (P < 0.001), max IMT (P < 0.001), blood pressure (BP) (P < 0.001), and smoking period (P < 0.001) ( Table 8).
Conversely, inverse correlation was observed between s-AP3D1-Ab levels and height, weight, Ca, and lowdensity lipoprotein cholesterol. Blood glucose and glycated hemoglobin, which are identi ed as DM markers, were not signi cantly correlated with the s-AP3D1-Abs levels. These results suggest that s-AP3D1-Ab re ected atherosclerosis and its causal HT and smoking.

JPHC cohort analysis
To determine whether s-AP3D1-Ab marker can be applied to predict the onset of AIS, JPHC cohort samples were examined. JPHC [29,30] conducted a baseline survey for 30,000 registered residents aged 40-69 years between 1990 and 1994 whose plasma were stored. The antibody levels of AP3D1 protein were detected by the above AlphaLISA in 202 cases of incident cerebral infarction in the cohort developed between the baseline and 2008 and in 202 controls whose age, sex, and area were matched with the cases. The antibody level of AP3D1 protein was positively and strongly associated with the risk of AIS: the ORs (95% CI) were 1.40 (0.75-2.63), 1.97 (1.07-3.65), and 2.28 (1.26-4.13) for the samples with the second, third, and highest quartiles of antibody level, respectively, compared with the lowest quartile (Table 9). These results indicate that the antibody markers against AP3D1 are useful in predicting the onset of AIS.

Discussion
Through the initial SEREX screening, AP3D1 was identi ed as an antigen recognized by serum IgG in patients with atherosclerosis. The s-AP3D1-Ab levels were higher in patients with AIS, TIA, DM, CVD, CKD, ESCC, and CRC than in the HDs (Figs. 1-5 and Tables 1-5). Among these diseases, the highest positive rates were observed for ESCC, DM, and type 2 CKD (Tables 1-5). The AUC values for nephrosclerosis type 2 CKD and ESCC, diabetic type 1 CKD, and DM were 0.874 and 0.872, 0.791, and 0.791, respectively, which were higher than other diseases. The comparison using the Mann-Whitney U test revealed that the s-AP3D1-Ab levels were signi cantly higher in the subjects with DM than in those without DM (Table 6). In contrast, no signi cant correlation was found between the s-AP3D1-Ab levels and DM markers, including blood glucose and glycated hemoglobin (Table 8). Consequently, the s-AP3D1-Ab levels do not directly re ect DM, but are associated with DM-induced atherosclerotic disorders, which are also related to CKD and cancer. Consistently, Spearman correlation analysis revealed a signi cant association between s-AP3D1-Ab levels and max IMT (P < 0.001), which re ects arterial stenosis, namely, atherosclerosis ( Table 8). The antibody levels signi cantly correlated with HT (P < 0.0001) ( Table 6), which are wellknown risk factors for atherosclerosis [31]. A univariate logistic regression analysis revealed that an elevated AP3D1-Ab level was associated with an increased risk of AIS (P < 0.0001). A multivariate logistic regression analysis has also identi ed age, HT, and DM, but not AP3D1-Ab as independent predictors of AIS (Table 7). Therefore, s-AP3D1-Ab marker may discriminate a certain type, if not all, of atherosclerosis caused by HT or DM, leading to the development of AIS and CVD.
AP3D1 is a subunit of the AP3 adaptor-like complex [32], which is expressed in the ubiquitous AP3 complex and also in the neuronal form [33]. AP3 is a heterotetrameric adaptor protein involved in the biogenesis of lysosome-related organelles, such as platelet-dense bodies. Mouse mutants of the null AP3D1 allele were reported to show abnormal bleeding due to the absence of a storage pool of dense platelet particles, raising the possibility that AP3D1 functions as a thrombogenic regulator through platelet function [34]. Platelets have been identi ed to play a signi cant role in hemostatic and thrombotic processes, where abnormal platelet adhesion/activation can lead to the formation of clots (thrombosis) [35]. Thrombosis is known to be closely associated with atherosclerosis [36]. Hirokawa et al.
[37] reported that AP3D1-DOT1L-SF3A2 was identi ed as a new susceptibility locus for myocardial infarction (MI) by European genome-wide association studies (GWAS), which is consistent with our observation that s-AP3D1-Abs levels were signi cantly higher in patients with CVD including MI (Fig. 3a, Table 3). On the other hand, Xiao et al. reported that angiotensin II facilitated the binding of AP3D1 with beta-arrestin, which resulted in the activation of AP3D1 as a scaffold protein [38]. Angiotensin II plays a key role in the pathogenesis of HT [39-43], leading to endothelial dysfunction and atherosclerosis [44,45]. AP3D1 can possibly mediate angiotensin II-induced HT and atherosclerosis. This was compatible with the results that the s-AP3D1-Ab levels were signi cantly associated with HT (P < 0.0001) ( Table 6) and BP (P < 0.001) ( Table 8).
HT is also known as a risk factor for various types of cancers [46][47][48][49][50]. Previous reports have proved that angiotensin is associated with cancer development [41,52]. Angiotensin is a major upstream regulator of cancer cachexia [53] and can further stimulate angiogenesis and tumor growth of breast cancer [54,55]. Thus, AP3D1 could also mediate angiotensin II-induced carcinogenesis. Based on the results, s-AP3D1-Abs levels were elevated in cancer patients (Fig. 5).
The progression of atherosclerosis and cancer often takes several years or more and, in the early stages, is sometimes accompanied by low levels of tissue destruction, which can lead to leakage of proteins from the cells. During this repeated leakage of proteins, even low levels of antigens can induce ampli ed expression of the antibodies. Thus, antibody markers are deemed more sensitive than antigen markers and may be useful for the early diagnosis of solid cancers including ESCC and CRC. s-AP3D1-Ab marker was closely associated with max IMT, an index of atherosclerosis, which then leads to the onset of AIS and AMI. Thus, predicting these onsets using s-AP3D1-Ab marker is possible. In fact, the results of JPHC cohort analysis indicated that the antibody marker against AP3D1 is useful in predicting the onset of AIS (Table 9). Atherosclerotic AIS and AMI and cancer have been identi ed as the leading causes of death; thus, the s-AP3D1-Ab marker would be highly useful to reduce its mortality.
As antihypertensive agents, statins, and antiplatelet agents are generally known to prevent the pathogenesis of atherosclerosis [56][57][58][59], the potential modulatory effects of these drugs on s-AP3D1-Ab levels must be considered. Second, as the controls were healthy volunteer donors, potential confounding factors between patients with atherosclerosis, cancers, and controls (age, BMI, HT, DM, and hyperlipidemia) were not adjusted in the analysis of this study. Lastly, physiological testing, such as baPWV, or coronary artery calci cation, was not performed to evaluate atherosclerosis in subjects subjected to the s-AP3D1-Ab analysis. Nevertheless, these tests might be expected to con rm the results of this study. The study population included only Japanese patients; thus, further studies are required in patients who are not taking drugs that can affect atherosclerosis and in other ethnic groups. Development of more biomarkers for the early diagnosis of atherosclerotic diseases and early detection of the development of tumors may improve the quality of life.

Conclusions
Serum anti-AP3D1 antibody levels have been determined to be elevated in atherosclerotic diseases and digestive organ cancers; thus, these can be applied to diagnose atherosclerosis and early-stage cancer.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
The present study was approved by the Local Ethical Review Board of Chiba University Graduate School of Medicine (Chiba, Japan) as well as the review boards of co-operating hospitals. Serum was collected from patients who had provided written informed consent.

Consent for publication
Not applicable.

Competing interests
The present study was performed in collaboration with Fujikura Kasei Co., Ltd. RN, NS and HK are employees of Fujikura Kasei Co., Ltd.  Part a indicates the number of total samples, samples from male and female participants, and ages (average ± SD). Part b summarizes the s-AP3D1-Ab levels examined using AlphaLISA. Numbers are as shown in Table 1; P-values of <0.05 and positive rates of >10% are marked in bold font. The plots for these data are shown in Fig. 2b. Part a indicates the number of total samples, samples from male and female participants, and ages (average ± SD). Part b summarizes the s-AP3D1-Ab levels examined using AlphaLISA. Numbers are as shown in Table 1; P-values of <0.05 and positive rates of >10% are marked in bold font. The plots for these data are shown in Fig. 2. Table 4 Comparing the s-AP3D1-Ab levels between HDs and patients with chronic kidney disease (CKD) a, Numbers for the total samples, samples from male and female participants and ages (average ± SD). The numbers shown are as described in Table 1. CKD was divided into three groups as follows: type 1, diabetic kidney disease; type 2, nephrosclerosis; and type 3, glomerulonephritis. P-values of <0.05 and positive rates of >10% are marked in bold font. The plots for these data are shown in Fig. 4a. The s-AP3D1-Ab levels examined using AlphaLISA in HDs and patients with ESCC and CRC are shown.

Sample information HD
Puri ed AP3D1-GST proteins were used as antigens. The numbers shown are as described in Table 1. Pvalues of <0.05 and positive rates of >10% are marked in bold font. The plots for these data are shown in     Table 9 Results of JPHC cohort subjects