Diagnosing Alzheimer’s Disease Specifically and Sensitively With pLG72 and Cystine/Glutamate Antiporter SLC7A11 AS Blood Biomarkers

Abstract Background Reliable blood biomarkers for Alzheimer’s disease (AD) have been lacking. The D-amino acids oxidase modulator (named pLG72) modulates glutamate N-methyl-D-aspartate receptor activity. The cystine/glutamate antiporter contains a SLC7A11 subunit, which mediates glutamate release. This study aimed to determine the accuracy of pLG72 protein and SLC7A11 mRNA in diagnosing AD. Methods This study enrolled 130 healthy controls and 109 unmatched AD patients; among them, 40 controls and 70 patients were selected to match by age. We measured their pLG72 protein in plasma and SLC7A11 mRNA in white blood cells. Results AD patients had markedly higher pLG72 levels and SLC7A11 mRNA ΔCT values than healthy controls (in both unmatched and matched cohorts; all 4 P values <.001). The receiver operating characteristics analysis in the unmatched cohorts demonstrated that the pLG72 level had a high specificity (0.900) at the optimal cutoff value of 2.3285, the ΔCT of SLC7A11 mRNA displayed an excellent sensitivity (0.954) at the cutoff of 12.185, and the combined value of pLG72 and SLC7A11 ΔCT determined a favorable area under the curve (AUC) (0.882) at the cutoff of 21.721. The AUC of the combined value surpassed that of either biomarker. The specificity, sensitivity, and AUC of the matched cohort were like those of the unmatched cohort. Conclusions The findings suggest that pLG72 protein and SLC7A11 mRNA can distinguish AD patients from healthy controls with excellent specificity and sensitivity, respectively. The combination of pLG72 and SLC7A11 yields better AUC than either, suggesting the superiority of simultaneously measuring both biomarkers in identifying AD patients.


Introduction
The prevalence of dementia in the elderly is increasing rapidly in the aging society, and the deteriorating clinical course is a heavy burden to both the patients and their family (Livingston et al., 2020). Early detection and intervention of Alzheimer's disease (AD) is pivotal for the outcome (Budd et al., 2011), while, to date, the diagnosis of AD relies on medical history and behavioral observations (McKhann et al., 1984). The presence of characteristic neurological and neuropsychological features and the absence of other physical or mental conditions are supportive in diagnosis (Erkkinen et al., 2018). Advanced medical imaging with computed tomography (CT) or magnetic resonance imaging and with single-photon emission computed tomography or positron emission tomography can be used to help exclude other cerebral pathology or subtypes of dementia (Weiner et al., 2015).
Medical organizations have created diagnostic criteria to standardize the diagnostic procedure. For example, the National Institute of Neurological and Communicative Disorders and Stroke (NINCDS) and the Alzheimer's Disease and Related Disorders Association (ADRDA, now known as the Alzheimer's Association) established the most commonly used NINCDS-ADRDA criteria for diagnosis of AD (McKhann et al., 1984). However, the aforementioned diagnosis methods are timeconsuming and dependent on the physicians' experience. Therefore, it is necessary to search for blood biomarkers that can give rise to a quick, reliable, and accurate diagnosis of AD.
Currently, most studies of AD biomarkers in blood have mainly focused on known pathological substrates for the disease, such as amyloid plaques and neurofibrillary tangles, which are respectively composed of the abnormally aggregated amyloid-β peptide (Aβ) and hyperphosphorylated tau (Lewczuk et al., 2018(Lewczuk et al., , 2020. Recently, measuring the levels of t-tau/Aβ 42 , hyperphosphorylated tau 181 /Aβ 42 , and Aβ 42 /Aβ 40 in blood samples successfully distinguished the AD patients from healthy controls (Kim et al., 2020).
However, the pathogenesis of AD remains still unresolved, and heterogeneous etiologies may be implicated (Lewczuk et al., 2020;Ahmed et al., 2021;Cheng et al., 2021;Tatulian, 2022). Optimal glutamate N-methyl-D-aspartate receptor (NMDAR) activation is required for synaptic plasticity, memory, and cognitive function (Tilleux and Hermans, 2007;Mattson, 2008). Attenuation of NMDAR-mediated neurotransmission can result in loss of neuronal plasticity and cognitive deficits in the aging brain, which may account for clinical deterioration and brain atrophy; on the other hand, overactivation of NMDAR leads to neurotoxicity (Madeira et al., 2018;Pinheiro and Faustino, 2019;Segovia et al., 2001).
Cystine/glutamate antiporter system x c − has been also implicated in the pathogenesis of AD (Qin et al., 2006). System x c − , a sodium-independent acidic amino acid transporter, regulates the uptake of cystine into cells in exchange for glutamate in a 1:1 ratio (Bannai, 1986). System x c − is composed of a light chain subunit (xCT, SLC7A11, which mediates the antiporter activity) and a heavy chain subunit (4F2hc, SLC3A2, which anchors the light chain subunit to the plasma membrane) (Bridges et al., 2012). Cystine is reduced to cysteine intracellularly after being taken up by system x c − . Cysteine is the rate-limiting substrate for the biosynthesis of glutathione, which is one of the most important antioxidants in the brain (Dringen and Hirrlinger, 2003). System x c − also modulates glutamate release, while glutamate, the most abundant amino acid neurotransmitter in the mammalian brain (Tilleux and Hermans, 2007;Mattson, 2008), plays an important role in regulating cognitive aging (Yao and Zhou, 2017;Chang et al., 2020). Patients with AD were found to have altered glutamate terminals in the hippocampus (Cowburn et al., 1988).
To our knowledge, the potential roles of pLG72 protein or cystine/glutamate antiporter SLC7A11 in diagnosing AD have yet to be studied. To explore new blood biomarkers, this study aimed to determine the diagnostic accuracy of pLG72 protein and SLC7A11 mRNA in detection of AD.

Participants
The study was approved by the Institutional Review Board of Chang Gung Memorial Hospital, Taiwan (104-9692B) and conducted in accordance with the current revision of the Declaration of Helsinki.
Both patients and healthy controls were evaluated by research psychiatrists after a thorough medical workup.
Patients were enrolled into this study if they (1) satisfied NINCDS- ADRD (McKhann et al., 1984) criteria for probable AD and had a clinical dementia rating (CDR) (Morris, 1993) score of ≥1, (2) were physically healthy and had all laboratory assessments (including blood routine and biochemical tests) within normal limits, and (3) agreed to participate in the study and provided informed consent.
Exclusion criteria included history of significant cerebrovascular disease with Hachinski Ischemic Score >4; major neurological, psychiatric, or medical conditions other than AD; substance (including alcohol) abuse or dependence; delusion, hallucination, or delirium symptoms; severe visual or hearing loss; and inability to follow protocol.

Significance Statement
While blood biomarkers for Alzheimer's disease (AD) have been lacking, this study enrolling 130 healthy controls and 109 AD patients found that AD patients had markedly higher pLG72 levels and SLC7A11 mRNA ΔCT values than healthy controls. While SLC7A11 mRNA levels showed better sensitivity than the pLG72 levels, the pLG72 levels revealed better specificity. Importantly, the combination of pLG72 and SLC7A11 mRNA generated better AUC than either of the 2 biomarkers. Overall, the findings suggest that pLG72 protein joint with SLC7A11 mRNA can constructively distinguish AD patients from healthy controls.
All healthy volunteers, aged ≥18 years, were free of any Axis I or II psychiatric disorder. To exclude potential confounding effects, all participants were non-smokers and had no DSM-IV diagnosis of substance (including alcohol) abuse. All healthy volunteers were physically and neurologically healthy and had normal laboratory assessments (including blood routine and biochemical tests).

Laboratory Assessments
A total 10 mL of peripheral venous blood was collected into an ethylenediamin tetra-acetic acid tube by well-trained personnel. The specimens were immediately centrifuged at 1500×g for 10 minutes at 4°C for further assaying both pLG72 levels in plasma and SLC7A11 mRNA levels in white blood cells (WBC).
Laboratory Methods for pLG72-The laboratory methods for measuring pLG72 levels in plasma were previously detailed . In brief, the blood specimens were centrifuged at 4°C and plasma was quickly dissected and immediately stored at −80°C until analysis. For western blotting, 100 µL plasma was depleted using ProteoPrep Blue Albumin and IgG Depletion Kit (Sigma, St. Louis, Missouri). The low-abundant protein fractions were collected to 100 μL, and 10 μL of them were mixed with 4× sample buffer (500 mM Tris-HCl [pH 6.8], 16% SDS, 80% glycerol, 400 mM DTT, and 0.08% bromophenol blue) and separated on 12% SDS-PAGE). Thereafter, proteins in the gels were transferred to 0.45 μm polyvinylidene difluoride membrane (Millipore), which was placed in 5% nonfat dry milk in TBST (20 mM Tris-HCl, pH 7.6, 500 mM sodium chloride, 0.1% Tween 20) for 1 hour at room temperature and then incubated with goat anti-pLG72 antibody (sc-46118, Santa Cruz Biotechnology) (Sacchi et al., 2008) diluted by 1:1000 in TBST overnight at 4°C. The membrane was washed thrice in TBST and incubated for 2 hours with an HRP-linked anti-goat IgG secondary antibody (sc-2030, Santa Cruz) diluted by 1:5000 in TBST. After 3 washes in TBST, the western blots were visualized with an ECL Advance Western Blotting Detection Kit (RPN2135, GE Healthcare). The stained membranes were photographed on ImageQuant LAS 4000 mini (GE Healthcare) and quantified using ImageQuant TL 7.0 software (GE Healthcare) by measuring the relative intensity from each band and normalized to the pLG72 recombinant protein (20 ng) signals. The commercial pLG72 antibodies were able to specifically recognize LG72 recombinant proteins. A standard curve was generated by serial dilutions of the pLG72 protein (50, 20, 10, 5, 2.5, 1.25, and 0.625 ng), and its detection limit was as low as 0.625 ng. The western blotting was repeated by 2 experienced technicians separately for quality control. The results of the blotting were very similar between the 2 technicians. The R-squared of the linearity between the blotting signals and the amounts of the pLG72 proteins was 0.988. In the western blotting, the molecular weight of the pLG72 protein band was approximately 18 kDa. The molecular weight of the standard recombinant pLG72 protein (as the control), which had a tagged protein on it, was marginally higher than that of the plasma pLG72 protein. The noise-signal ratios around the points of the western blotting were between 0.04 and 0.13.
Laboratory Methods for SLC7A11-The laboratory procedure for determining the SLC7A11 mRNA levels in WBC have been described elsewhere (Lin et al., 2016). In short, red blood cells were removed by 1× RBC lysis buffer (Genepure Technology Co., http:// www.genepure.com.tw/index.asp), and WBC was obtained after blood centrifugation at 4°C. RNA was isolated from the WBC using the Tri-reagent method (MRC) according to the protocol provided by the manufacturer. Four housekeeping genes were used as endogenous controls: glyceraldehyde-3-phosphate dehydrogenase NM_002046 and NM_002046.3, beta-2-microglobulin NM_004048.2, and hypoxanthine-guanine phosphoribosyltransferase NM_000194.2. The real-time quantitative PCR reaction was carried out twice for each sample. The relative mRNA level of SLC7A11 in the samples was calculated by the ΔCt value (Ct , target − Ct , housekeeping ) (Livak and Schmittgen, 2001).

Statistical Analysis
Baseline characteristics were calculated for the AD and control groups. Numeric data are presented as means ± SD. The P value between the AD and control groups was calculated based on t test or Fisher's exact test, and P < .05 was considered to indicate statistical significance.
A receiver operating characteristics (ROC) analysis for pLG72 Protein, SLC7A11 mRNA in ΔCT value, or the combination thereof was applied by plotting the proportion of true-positive results (sensitivity) vs the proportion of false-positive results (1 − specificity).
Logistic regression was applied to generate the predictive model for AD. In the model, AD (vs. control) was the dependent variable, and pLG72 protein level and SLC7A11 mRNA ΔCT value were the covariates. In this way, an equation (I) for the weighted value of pLG72 protein level and SLC7A11 mRNA ΔCT value in combination as shown below was formulated using the logistic regression model: (1) A = Weighted value B = pLG72 protein level (ng/μL) C = ΔCT value of SLC7A11 mRNA

Participants
This study enrolled 239 participants, including 130 healthy controls and 109 unmatched AD patients. The mean age (74.6 ± 7.9 years [SD] years) of unmatched patients with AD was older than that of healthy controls (43.1 ± 18.2 years, P < .001). Among them, we selected 70 patients from the AD group and 40 individuals from the healthy controls group to match by age.
The demographic and clinical data of age-matched and unmatched AD patients and healthy controls are summarized in Table 1.
Among the 109 unmatched patients with AD, 34 were medicated and the other 75 were drug free. Among the 70 matched patients with AD, 24 were medicated and the other 46 were drug free (Table 1).

pLG72 Protein
Typical western-blot results for G72 protein in plasma samples from randomly selected AD patients (n = 6) and healthy controls (n = 6) are shown in Figure 1. Table 1, the pLG72 protein levels in the plasma of unmatched and matched AD patients were markedly higher than those of healthy controls.

SLC7A11 mRNA
Also shown in Table 1, the ΔCT values of SLC7A11 mRNA in WBC of healthy controls were markedly lower than those of AD patients.

ROC Analysis for Specificity and Sensitivity
Shown in Table 2 and Figure 2, ROC analysis was applied to determine the cutoff values of pLG72 protein and SLC7A11 mRNA as potential AD predictors by plotting the proportion of truepositive results (sensitivity) vs the proportion of false-positive results (1 − specificity).
The ROC analysis of the pLG72 levels for all AD patients vs healthy controls determined an optimal cutoff value (2.3285) with a modest sensitivity (0.541) and a high specificity (0.900) (AUC = 0.791) ( Table 2). The ROC analysis of the ΔCT values of SLC7A11 mRNA for all AD patients vs healthy controls determined an optimal cutoff value (12.185) with an excellent sensitivity (0.954) and a modest specificity (0.523) (area under the curve [AUC] = 0.803) ( Table 2).  pLG72 and SLC7A11 as Biomarkers for AD | 5

Patients with Alzheimer's disease
The ROC analysis of the combined values of pLG72 levels and the ΔCT values of SLC7A11 mRNA for all AD patients vs healthy controls determined an optimal cutoff value (21.721) with a favorable sensitivity (0.881) and a modest specificity (0.738) (AUC = 0.882) (Table 2; Figure 2). The AUC of the combined values of the 2 biomarkers was better than the AUC of either of the 2 biomarkers (Table 2; Figure 2).
The specificity, sensitivity, and AUC of the matched AD patients and healthy controls were similar to those of the unmatched cohort (Table 2). ROC curves of pLG72 protein and mRNA of SLC7A11 of matched healthy controls vs AD patients are not shown because they are similar to Figure 2.
Among the drug-free patients, the ROC analysis of the combined values of pLG72 levels and the ΔCT values of SLC7A11 mRNA for AD patients vs healthy controls determined an optimal cutoff value (24.985) with a favorable sensitivity (0.907) and a modest specificity (0.777) (AUC = 0.915). The AUC of the combined values of the 2 biomarkers were better than the AUC of either of the 2 biomarkers (0.829 and 0.824, respectively) ( Table 2).
From the above results, the differentiating power of the combined values of pLG72 levels and the ΔCT values of SLC7A11 mRNA was better in the drug-free AD patients than the medicated.

Correlations Between Levels of pLG72 and SLC7A11 mRNA and Scores of CDR, MMSE, and ADAS-cog
Among the AD patients, we checked the correlations between levels of pLG72 and SLC7A11 mRNA and scores of CDR, MMSE,  and ADAS-cog. There was no correlation between levels of pLG72 and scores of CDR (r = −0.037, P = .703), MMSE (r = −0.116, P = .247), and ADAS (r = −0.024, P = .808). There was also no correlation between levels of SLC7A11 and scores of CDR (r = 0.175, P = .069), MMSE (r = −0.164, P = .100), and ADAS (r = 0.091, P = .361). Among the healthy controls, we checked the correlations between levels of pLG72 and SLC7A11 mRNA and scores of MMSE and ADAS-cog (but not CDR), because all CDR scores of healthy controls were zero. There was no correlation between levels of pLG72 and scores of MMSE (r = −0.126, P = .514) and ADAS (r = 0.030, P = .876). There was also no correlation between levels of SLC7A11 and scores of MMSE (r = −0.214, P = .265) and ADAS (r = 0.265, P = .164).

Discussion
The main findings of this study suggest that while pLG72 (DAOA) protein or SLC7A11 mRNA alone may be useful as a potential biomarker for the detection of AD, the combination of pLG72 protein and SLC7A11 mRNA can lead to an even more reliable diagnosis of AD.
pLG72, existing in mitochondria of humans and the other 3 primate species (Chumakov et al., 2002), has been reported to correlate with DAAO (Mechelli et al., 2012;Birolo et al., 2016;Keller et al., 2018;Lin et al., 2020b). DAAO levels in serum have been found to be increased with the severity of the cognitive aging; however, DAAO blood concentration has been inadequate to be a good enough biomarker for AD (Lin et al., 2017). The current study indicates that pLG72 levels in plasma display a high specificity (0.900) in differentiating AD patients from healthy controls.
Cystine/glutamate antiporter system x c − also has been implicated in the pathogenesis of not only AD (Qin et al., 2006) but also schizophrenia (Fournier et al., 2017). In fact, there are overlapping pathophysiology and clinical manifestations between AD and schizophrenia: both reveal cognitive decline (McKhann et al., 1984;Lin et al., 2014c), psychotic and behavioral symptoms Kendler, 2020), implication with NMDAR pathogenesis (Yao and Zhou, 2017;, and clinical improvement by DAAO inhibition (Harrison, 2018;Lin et al., 2020a). Previous study (Lin et al., 2016) has suggested that the mRNA expression of cystine/glutamate antiporter system x c − SLC3A2 may have potential to be a biomarker of schizophrenia in the future. The present study demonstrated that cystine/glutamate antiporter SLC7A11 mRNA had supersensitivity for identification AD.
Of note, the combination of pLG72 and SLC7A11 generated higher AUC than either biomarker, therefore suggesting the superiority of the combination in diagnosing AD. Since both pLG72 and cystine/glutamate antiporter system x c − mediate glutamaterelated neurotransmission and redox reaction (Filiou et al., 2012;Fournier et al., 2017;, it is reasonable for simultaneously measuring both pLG72 and SLC7A11.
We compared the 2 biomarkers in age-unbalanced and age-balanced cohorts, mainly to examine whether age may affect the biomarker-based differentiation between the AD patients and healthy controls. We found that the results in the age-unbalanced cohorts and age-balanced cohorts were quite similar, implying that age may not confound the differentiation.
The present study has several limitations. First, the findings of this study were obtained from a cross-sectional design. Future longitudinal studies are warranted. Second, the blood-brain (or CSF) relationship of pLG72 and SLC7A11 remain uncertain in the participants. Third, we did not apply mass spectrometry to confirm the recognized band in western blotting corresponding to pLG72. Fourth, only Han Chinese populations were recruited in this study. Whether the findings can be extrapolated to other populations deserves investigation. Fifth, the sample size in the matched cohort was modest. Sixth, western-blot analyses were not done in triplicate. Seventh, we did not measure biomarkers of AD pathophysiology (Aβ or tau) for confirmation. Eighth, APOE4 was not assayed. And lastly, there was an almost significant P value for the age-matched cohort in terms of age; in addition, the control group and the AD group differed in gender distribution and education levels. Since women are more likely to have AD (Prince et al., 2016) and low education is an important risk factor for dementia (Maccora et al., 2020), we regard the difference in gender (female predominance) and education in the study as nature. Further study on participants with better matched age, gender distribution, and education levels is necessary.
In conclusion, the above findings suggest that the pLG72 protein levels and SLC7A11 mRNA levels (as expressed in ΔCT values) can differentiate AD patients and healthy controls with good AUC. SLC7A11 mRNA levels have better sensitivity than pLG72 levels, while pLG72 levels have better specificity. The more interesting finding is that the combination of pLG72 and SLC7A11 yields better AUC than either, suggesting the superiority of simultaneously measuring both biomarkers in identifying AD patients.

Author Contributions
H.Y. Lane and C.H. Lin were responsible for the conception and design, literature review, data collection and analysis, and manuscript writing. Both authors reviewed and approved the manuscript.