Drebrin is a novel biomarker of cognitive deterioration in Alzheimer's disease

Abstract

Background

The cerebrospinal fluid (CSF) and plasma amyloid-β (Aβ)_40/42 ratio, p-217tau and p-181tau, and neurofilament light chain are biomarkers of Aβ proteinopathy, tau proteinopathy, and neuronal injury, respectively, in Alzheimer's disease (AD). However, direct biomarkers of cognitive function have yet to be identified.

Objective

Therefore, the present study investigated the potential of CSF and plasma levels of drebrin, a postsynaptic protein, as biomarkers of synaptic activity and cognitive function in the human brain in clinical settings.

Methods

We developed a novel ELISA to measure CSF and plasma levels of drebrin and analyzed 68 CSF and 128 plasma samples from patients with AD and other neurological diseases.

Results

CSF drebrin levels were significantly reduced in samples of mild cognitive impairment due to AD, the dementia stages of AD, and idiopathic normal pressure hydrocephalus. Plasma drebrin levels were also significantly reduced in samples of MCI due to AD.

Conclusions

CSF and plasma drebrin levels are specific biomarkers of cognitive decline in the MCI stage of AD.

Introduction

In the Alzheimer's disease (AD) continuum, amyloid-β (Aβ) deposition starts two decades before the onset of symptomatic cognitive decline, and is followed by the accumulation of phosphorylated tau. After a long incubation period, neurofibrillary tangles and neurodegeneration emerge. Cognitive decline progresses to clinical dementia, which lasts further for more than 10 years. The signatures of these pathological and clinical processes are now traced using biomarkers that are present in cerebrospinal fluid (CSF)^1,2 or plasma as well as via positron emission tomography (PET) of Aβ and tau amyloid deposition.^3,4 In addition to Aβ₄₂ and Aβ_42/40 ratios as biomarkers of Aβ proteinopathy (A), phosphorylated tau at threonine 181 (p-181tau), 217 (p-217tau), or 205 (p-205tau) and total tau (t-tau) have recently been proposed as biomarkers of phosphorylated and secreted AD tau (T1) and AD tau proteinopathy (T2) in the National Institute on Aging and Alzheimer's Association's (NIA-AA) revised criteria for the diagnosis and staging of AD by the Alzheimer's Association workgroup.^5,6 Additionally, neurofilament light chain (NfL) has been proposed as a biomarker of non-specific processes involved in the AD pathology “N (injury, dysfunction, or degeneration of neuropils)”.^6,7 However, this marker reflects non-specific axonal damage. ⁸

Surrogate biomarkers that directly monitor cognitive decline as an alternative to clinical cognitive function tests are still expected. Drebrin was the first postsynaptic protein to be identified that markedly decreased in spite of the preservation of the presynaptic protein synaptophysin in the AD brains.^9,10 These findings were subsequently confirmed at the protein and mRNA levels in the AD brains.^11,12 Aβ oligomer-induced aberrations in drebrin-associated synaptic composition, shape, and density resulted in a loss of connectivity in AD. ¹³ However, the development of ELISA to measure drebrin levels in body fluids and its confirmation as a surrogate biomarker of cognitive decline have yet to be attempted. Drebrin, an actin-binding postsynaptic protein, is distributed at spine heads and stabilizes the spine structure.^14–16 Dendritic spines change their shape in response to synaptic input ¹⁷ and this change correlates with synaptic plasticity, such as long-term potentiation or depression, which are mechanisms underlying learning and memory. Drebrin levels decreased in the hippocampus with mild cognitive impairment (MCI),^10,18 and almost disappeared in the hippocampus ⁹ and cerebral cortex^11,12 with AD. Anti-drebrin autoantibodies define a chronic syndrome of recurrent seizure and neuropsychiatric impairment. ¹⁹ The drebrin expression pattern in the dentate gyrus of refractory temporal lobe epilepsy has been negatively associated with seizure frequency and positively associated with verbal memory. ²⁰ These findings indicate the potential of decreases in drebrin, which lead to synapse loss, as a surrogate biomarker for synaptic function as measure of cognitive function.^13,21,22

Here, we developed ELISA and tried to assay drebrin amount and evaluated association with disease specificity and clinical cognitive function in fluid CSF and plasma in neurological disease including AD.

Methods

Patients

Sixty-eight CSF and 128 plasma samples were examined from 30 patients with AD, 17 with idiopathic normal pressure hydrocephalus (iNPH), ²³ 10 with inflammatory demyelinating diseases (IMD), 32 with amyotrophic lateral sclerosis (ALS), ²⁴ 13 with immune-mediated polyneuropathy (IMN), 6 with other neurological diseases (OND), and 20 healthy and cognitively unimpaired control subjects (HCU) were examined (Table 1). Thirty AD cases fulfilled the NIA-AA criteria. ⁵ Cognitive function in AD subjects was evaluated by the Mini-Mental State Examination (MMSE) and Clinical Dementia Rating (CDR) and was classified into the following clinical stages: cognitively unimpaired (ADCU), mild cognitive impairment due to AD (ADMCI), and dementia stages of AD (ADD; Table 1a, b). CSF biomarker criteria to exclude AD consisted of decreased levels of t-tau (<297 pg/mL) and p-181tau (<50 pg/mL). In the analysis of 68 CSF samples, 24 AD cases comprising 7 cases of ADD, 8 of ADMCI, and 9 of ADCU were examined. IMD included 7 cases of multiple sclerosis (MS), 2 of neuromyelitis optica spectrum disease, and 1 of myelin oligodendrocyte glycoprotein antibody-associated disease. OND consisted of cervical spondylotic myelopathy, non-AD dementia, semantic dementia, multiple system atrophy, polymyositis, and progressive supranuclear palsy (Table 1a). In 60 plasma samples, it was not possible to measure CSF drebrin levels.

Table 1.

Cerebrospinal fluid (CSF) and plasma drebrin levels. CSF and plasma drebrin levels were both measured in 68 subjects (a). Plasma drebrin levels in samples from 68 subjects (a) and those from other 60 subjects are summarized in b.

a. CSF
Disease	N	Age (y)	F/M	MMSE	CDR	tau pg/mL	ptau181 pg/mL	Aβ1−42 pg/mL	Drebrin pg/mL
AD	24	71	11/13	22	0.8	530 ± 274	73 ± 23	895 ± 618	53 ± 7
ADD	7	68	3/4	13	1.2	708 ± 318	78 ± 33	502 ± 253	7 ± 12
ADMCI	8	73	6/2	22	0.5	618 ± 318	81 ± 25	434 ± 167	17 ± 39
ADCU	9	72	2/7	29	0	334 ± 54	62 ± 8	1610 ± 378	121 ± 73
iNPH	17	78	6/11	20	1.1	166 ± 57	30 ± 5	472 ± 222	6 ± 9
IMD	10	36	7/3	27	ー	ー	ー	ー	42 ± 58
OND	6	65	2/4	25	ー	223 ± 8	40 ± 3	373 ± 245	33 ± 30
HCU	11	70	2/9	29	0	155 ± 31	34 ± 7	1069 ± 239	176 ± 225
Total	68	67	28/40	24	0.5	325 ± 255	51 ± 26	793 ± 513	59 ± 113

AD: Alzheimer's disease; ADD: Alzheimer's disease dementia; ADMCI: MCI due to AD; ADCU: cognitively unimpaired stage in AD; iNPH: idiopathic normal pressure hydrocephalus (Hakim Disease); IMD: immune-mediated demyelinating disease; ALS: amyotrophic lateral sclerosis; IMN: immune-mediated neuropathy; OND: other neurological diseases; HCU: healthy and cognitively unimpaired controls; MMSE: Mini-Mental State Examination; CDR: Clinical Dementia Rating; All measured values were described as the mean ± standard deviation.

The present study was approved by the Ethics Committees of the Geriatric Research Hospital (2021–78) and Hirosaki University (2017–112). All participants provided their written informed consent.

CSF and plasma sampling

Lumbar puncture to obtain a CSF sample was performed in the morning after fasting in accordance with the standard protocol. CSF samples were immediately centrifuged at 1400xg at 4°C for 10 min and stored in polypropylene vials at −80°C. Ten milliliters of morning fasting blood was collected into an EDTA-2Na tube and immediately centrifuged at 1400xg for 10 min, separated into plasma, and stored in a polypropylene tube at −80°C.

ELISA

We generated two novel independent monoclonal antibodies targeting human drebrin, AA001 and AA003. AA001 binds to amino acid (aa) residues 216–232 and AA003 to aa residues 420–475 of the human drebrin sequence (Figure 1). In ELISA, 100 μL of the capture antibody AA001 (10 μg/mL) in 100 mM carbonate buffer (pH 9.5) was added to each of the 96 wells of an ELISA plate. The plate was incubated at 4°C for 30–36 h for antibody coating, washed twice with phosphate-buffered saline (PBS; pH 7.4), and 200 μL of PBS containing 1% bovine serum albumin (BSA) was then added for 1 h for blocking. After discarding the blocking solution, 100 μL of a recombinant human drebrin protein as the standard, 100 μL of CSF or 25 μL of plasma mixed with 75 μL of a diluent buffer (1% BSA in PBS, 0.05% Tween20, and 0.05% Proclin300 [Sigma-Aldrich, St Louis, MO], and 50 μg/mL normal mouse IgG) were added to each well. The plate was incubated at 4°C overnight. After washing 4 times with PBS, 100 μL of peroxidase-labeled AA003 Fab’ detection antibodies (0.1 μg/mL) was added to each well. The plate was incubated at 4°C for 30 min. After washing the plate 5 times with PBS, 100 μL of peroxidase substrate solution (TMB PLUS2^® Cat. No. 4395, Kementec, Denmark) was added. After an incubation for 30 min, 100 μL of 1.5 N sulfuric acid was added to stop the reaction. Optical density values were measured at a wavelength of 450 nm (SpectraMax^® ABS, SoftMax Pro 7.1.2, Molecular Devices, CA). All measured values were described as the mean ± standard deviation. Recovery rates ranged between 97.8 and 129.3%. The intra-assay coefficient of variation (CV) was 11.8% and the inter-assay CV was 4.8%. Basic data of ELISA for CSF t-tau, p-181tau, and Aβ₁₋₄₂ were previously reported.^25,26

Figure 1.

Schematic representation of the structure of human drebrin. (A) The green and purple lines indicate the approximate positions of the areas recognized by the antibodies AA001 and AA003, respectively. (B) Epitope mapping of antibodies using GFP-tagged drebrin fragments. Antibody AA001 recognized drebrin fragments encompassing amino acids (aa) 212–232 and 216–232 (left), indicating that its epitope lies within aa 216–232. Antibody AA003 bound to fragments spanning aa 420–695 and 420–475 (right), localizing its epitope to the intrinsically disordered region (IDR) between aa 420–475. (C) Specificity of antibodies for drebrin. Both antibodies detected drebrin in wild-type (WT) brain lysates, but showed no signal in drebrin knockout (DXKO) samples, confirming their specificity for drebrin.

Statistical analysis

Since normality tests (the D’Agostino & Pearson, Anderson-Darling, Shapiro-Wilk, and Kolmogorov-Smirnov tests) and a normal QQ plot analysis showed that data were not normally distributed, the Kruskal-Wallis test for non-parametric data was used. Dunn's multiple comparisons test was adopted as a post-hoc test for comparisons with HCU. In comparisons with HCU, no significant differences were observed in the sex distribution of any disease group, whereas the age distribution of the MS group significantly differed.

An area under the receiver operator characteristic curve (ROC) analysis was performed to examine the diagnostic utility of drebrin for discriminating between the respective disease groups and HCU. The optimal cut-off value for each CSF marker was selected based on the likelihood ratio [sensitivity/(1-specificity)]. A correlation analysis was performed using Spearman's correlation coefficient by rank test. All tests were two-tailed and performed with GraphPad Prism, version 10 (GraphPad Software, San Diego, CA), or R, version 4.4.2, and the significance level was set at 5%.

Results

CSF drebrin levels were 53 ± 7 pg/mL in AD, 6 ± 9 pg/mL in iNPH, 42 ± 58 pg/mL in IMD, 33 ± 30 pg/mL in OND, and 176 ± 225 pg/mL in HCU. CSF drebrin levels in the 3 clinical stages of AD were as follows: 7 ± 12 pg/mL in ADD, 17 ± 39 pg/mL in ADMCI, and 121 ± 73 pg/mL in ADCU (Table 1a). CSF drebrin levels were significantly lower in ADD (p = 0.0008), ADMCI (p = 0.0011), and iNPH (p < 0.0001) than in HCU (Figure 2). No significant difference was noted between ADCU and HCU. Therefore, CSF drebrin levels were reduced in cognitively deteriorated patients with ADD, ADMCI, and iNPH.

Figure 2.

CSF drebrin. ADD: Alzheimer's disease dementia; ADMCI: MCI due to AD; ADCU: cognitively unimpaired stage in AD; iNPH: idiopathic normal pressure hydrocephalus; IMD: Inflammatory demyelinating disease; OND: other neurological diseases; HCU: healthy and cognitively unimpaired controls. Bars and numbers indicate p-values in the respective diseases relative to HCU. Scatter plot of individual values and mean with standard deviation. Numbers show p values.

The ratios of plasma to CSF drebrin were 66 in AD and HCU, 70 in ADCU, 95 in iNPH. Ratios increased in ADD (125), IMD (538), and OND (274) and decreased in ADMCI (10). A characteristic decrease in the plasma/CSF ratio was only observed in ADMCI.

Plasma drebrin levels in 128 samples were as follows: 5877 ± 5728 pg/mL in AD (n = 30), 3754 ± 6795 pg/mL in iNPH (n = 17), 18,485 ± 11,031 pg/mL in IMD (n = 10), 13,283 ± 8295 pg/mL in ALS (n = 32), 16,553 ± 9437 pg/mL in IMN (n = 13), 9717 ± 11,777 pg/mL in OND (n = 6), and 6701 ± 6553 pg/mL in HCU (n = 20) (Table 1b). In the clinical subgroups of AD, drebrin levels were 9670 ± 10,640 pg/mL in ADD (n = 12), 1539 ± 545 pg/mL in ADMCI (n = 9), and 5492 ± 2563 pg/mL in ADCU (n = 9). A significant decrease in the level of drebrin was only noted in ADMCI (p = 0.045; Figure 3).

Figure 3.

Plasma drebrin. ADD: Alzheimer's disease dementia; ADMCI: MCI due to AD; ADCU: cognitively unimpaired stage in AD; iNPH: idiopathic normal pressure hydrocephalus; IMD: immune-mediated demyelinating disease; ALS: amyotrophic lateral sclerosis; IMN: immune-mediated neuropathy; OND: other neurological diseases; HCU: healthy and cognitively unimpaired controls. A significant difference was observed between the ADMCI and HCU groups (p = 0.045). Bars and numbers show p-values relative to the HCU group. Scatter plot of individual values and mean with standard deviation. Numbers show p values.

The ROC analysis of CSF drebrin identified a cut-off value of 40 pg/mL to discriminate between ADD + ADMCI and ADCU + HCU with 100% sensitivity (95% CI: 83–100%), 93% specificity (95% CI: 70–100%), and a likelihood ratio of 15 (Figure 4(a)). Between iNPH and HCU, a cut-off value of 24 pg/mL demonstrated 100% sensitivity (95% CI: 72–100%), 94% specificity (95%CI: 73–100%), and a likelihood ratio of 17 (Figure 4(b)). The ROC analysis of plasma drebrin identified a cut-off value of 2217 pg/mL to discriminate between ADMCI and ADCU + HCU with 95% sensitivity (95% CI: 83–99%), 89% specificity (95% CI: 57–99%), and a likelihood ratio of 8.5 (Figure 4(c)). Between iNPH and HCU, the cut-off value of 2598 pg/mL showed 90% sensitivity (95% CI: 70–98%), 82% specificity (95%CI: 59–94%), and a likelihood ratio of 5.1 (Figure 4(d)).

Figure 4.

ROC analysis of sensitivity and specificity of CSF and plasma drebrin. (a) The ROC analysis of CSF drebrin showed that a cut-off value of 40 pg/mL discriminated between ADD + ADMCI and ADCU + HCU with 100% sensitivity (95% CI: 83–100%), 93% specificity (95% CI: 70–100%) and a likelihood ratio of 15. (b) Between iNPH and HCU, the cut-off value of 24 pg/mL showed 100% sensitivity (95% CI: 72–100%), 94% specificity (95%CI: 73–100%), and a likelihood ratio of 17. (c) The ROC analysis of plasma drebrin showed that a cut-off value of 2217 pg/mL discriminated between ADMCI and ADCU + HCU with 95% sensitivity (95% CI: 83–99%), 89% specificity (95% CI: 57–99%), and a likelihood ratio of 8.5. (d) Between iNPH and HCU, the cut-off value of 2598 pg/mL showed 90% sensitivity (95% CI: 70–98%), 82% specificity (95%CI: 59–94%), and a likelihood ratio of 5.1.

Spearman's correlation coefficient by rank test showed a moderate correlation between MMSE scores and drebrin levels in CSF samples (p < 0.0001; r = 0.6) and plasma samples (p < 0.0003, r = 0.4).

Discussion

The present results showed that CSF drebrin levels were markedly lower in ADD, ADMCI and iNPH than in IMD, OND, and HCU. Other central and nervous system diseases did not show a decrease in drebrin. Furthermore, drebrin levels were reduced in ADMCI and ADD, but not in ADCU. These results imply a disturbance in the stability and plasticity of the dendritic spine structure, leading to learning and memory impairments in AD and iNPH. The positive relationship observed between MMSE scores and drebrin levels supports this hypothesis. These results also suggest that a decrease in drebrin was closely associated with cognitive impairment, but not with a disease-specific pathology, such as Aβ and tau amyloidosis in AD or chronic communicating hydrocephalus with incomplete obstruction of the normal pathway of CSF flow in iNPH. The results showing that drebrin levels were unchanged in ADCU but decreased in ADMCI and ADD suggests that it is a more accurate marker of established cognitive impairment than of preclinical disease. This contrasts with evidence showing that non-cognitive symptoms (e.g., motivational and affective changes linked to limbic circuits such as the nucleus accumbens) may precede overt cognitive decline. ²⁷

Drebrin levels were reduced by 35–81% in the hippocampus and cortex of autopsy brains from MCI, mild to moderate AD dementia,^10,11,18 and Down syndrome ¹² despite the preserved levels of other presynaptic proteins, such as synaptophysin, synaptotagmin, ¹⁰ and synaptosomal-associated protein-25 (SNAP-25). ²⁸ Consistent with these findings, our assay showed an 89% reduction in CSF drebrin levels in ADMCI and ADD.

This is the first study to report decreases in CSF drebrin levels in iNPH as well as AD. To exclude the possibility of the complication of AD in iNPH cases, we carefully selected typical iNPH cases without elevated CSF t-tau or p-181tau. Therefore, a different pathological mechanism other than tau pathology may reduce drebrin levels in iNPH. A previous study suggested that higher levels of ventricular CSF neurogranin, a postsynaptic marker, correlated with a favorable postoperative outcome, indicating the presence of postsynaptic dysfunction in iNPH. ²⁹ In contrast, low CSF phosphorylated tau and Aβ₄₂ levels have been reported in iNPH. ³⁰ Our iNPH cases also showed low p-181tau and Aβ₄₂ levels. However, these findings and the present results do not rule out synaptic dysfunction caused solely by comorbid Aβ deposition. A follow-up study of improvements in cognitive function and drebrin levels after shunt surgery is needed to elucidate whether obstructed CSF flow or mild Aβ amyloidosis is more strongly associated with cognitive impairment in iNPH.

CSF and blood NfL are highly sensitive nerve injury biomarkers for neuroaxonal disturbances^7,8; however, NfL increases in a wide range of neurological diseases in non-disease-specific and damaged location-dependent manners. ⁸ The presynaptic synaptic markers SNAP-25 ^{4 31–33} and Visinin-like protein 1 (VILIP-1) ³⁴ and the postsynaptic marker neurogranin^4,33,35 are available as biomarkers of synaptic dysfunction in AD and other diseases. These biomarkers were significantly increased in CSF from AD patients with the development of clinical stages; however, the increases were modest, and the overlap in individual values reduced their sensitivity and specificity as biomarkers.^4,32,33,35 These markers increased in traumatic brain injuries, ³⁶ MS, ³⁷ Creutzfeldt-Jakob disease,^38–41 and non-AD dementia,^40,42 but decreased in some neurodegenerative diseases. ³³ Based on the increased pattern of the synaptic markers, SNAP-25, VILIP-1, and neurogranin, not only pre- and postsynaptic dysfunctions, but also extensive neuronal damage may occur in these diseases, similar to the increase in t-tau, which suggests neuronal injuries. In comparisons with increased pre- and postsynaptic markers, CSF drebrin levels markedly decreased in symptomatic AD and iNPH. Since mRNA ⁴³ and protein levels^9,10,18 of drebrin markedly decreased in AD brains, low CSF drebrin levels appear to correspond to these changes. Presynaptic VILIP-1 ⁴⁴ and postsynaptic neurogranin^45,46 are distributed in nerve cell bodies and axons; however, drebrin mainly localizes to dendritic spines. Differences in the degree of reduction and histological distribution of VILIP-1 and neurogranin may explain why drebrin shows a different pattern of change.

Plasma biomarkers of synaptic dysfunction are expected for non-invasive and easy-to-use clinical laboratory tests. However, a reliable plasma biomarker has yet to be identified. ⁶ Plasma ß-synuclein and serum VILIP-1 levels were previously shown to be increased in AD; however, overlap among controls and various neurological diseases reduced their sensitivity and specificity.^41,47 Plasma SNAP-25 and neurogranin levels were not evaluated because of the difficulties associated with their measurement.^48,49

Plasma drebrin levels significantly decreased in ADMCI, but not in iNPH. Plasma drebrin levels were 66-fold higher than those of CSF levels. The majority of biomarkers were higher in CSF than in blood, except for α-synuclein, which is present extracerebrally in red blood cells. ²⁵ Drebrin is also expressed in immune system cells, such as T lymphocytes and mast cells, ⁵⁰ and in stomach and kidney epithelial cells, ⁵¹ and this peripheral expression may result in higher drebrin levels in plasma than CSF, similar to α-synuclein (Table 1a, b). This hypothesis may also explain why inflammation increased plasma drebrin levels in cases of ADD, MS, IMN, and OND.

The present study has a number of limitations that need to be addressed. The main limitations were the small number of subjects and narrow range of neurological diseases examined for a clinical evaluation of biomarkers of synaptic dysfunction. Most cases were diagnosed clinically and using CSF biomarkers, but not by definite autopsy findings. Therefore, since the evidence obtained in the present study alone is insufficient to use CSF or plasma drebrin as a surrogate biomarker, large-scale prospective longitudinal studies are needed to address these issues and establish more precise cut-off values for social implementation. Plasma drebrin may be useful for the screening of cognitive decline, particularly in MCI due to AD, and the application of newly developed disease-modifying therapy for Aβ amyloidosis if cut-off values are established for various neurological diseases to improve accuracy. Another limitation is the lack of patients with non-AD dementia or other tauopathies for comparison in the stage of MCI. We need to confirm and standardize our novel ELISA for drebrin using CSF and plasma in large cohort studies with a long-term follow-up.