Abstract
Background:
Alzheimer’s disease (AD) and Lewy body disease (LBD) are complex neurodegenerative disorders that have been associated with brain inflammation and impaired neurotransmission.
Objective:
We aimed to determine concentrations of multiple cytokines, chemokines, and neurotransmitters previously associated with brain inflammation and synapse function in cerebrospinal fluid (CSF) from AD and LBD patients.
Methods:
We examined a panel of 50 analytes comprising neurotransmitters, cytokines, chemokines, and hormones in CSF in a cohort of patients diagnosed with mild cognitive impairment (MCI), AD, LBD, or non-demented controls (NDC).
Results:
Among neurotransmitters, noradrenaline (NA) was increased in AD CSF, while homovanillic acid (HVA), a dopamine metabolite, was reduced in both AD and LBD CSF relative to NDC. Six cytokines/chemokines out of 30 investigated were reliably detected in CSF. CSF vascular endothelial growth factor (VEGF) was significantly reduced in LBD patients relative to NDC.
Conclusions:
CSF alterations in NA, HVA, and VEGF in AD and LBD may reflect pathogenic features of these disorders and provide tools for improved diagnosis. Future studies are warranted to replicate current findings in larger, multicenter cohorts.
Keywords
INTRODUCTION
Alzheimer’s disease (AD) and Lewy body disease (LBD) are complex neurodegenerative disorders whose pathophysiological mechanisms have not been fully elucidated. AD diagnosis can be currently reached by a combination of brain PET imaging, assessment of cerebrospinal fluid (CSF) amyloid-β1–42 (Aβ42) and tau protein, and neuropsychological testing [1]. On the other hand, diagnosis of LBD, a disease characterized by brain deposition of α-synuclein aggregates, remains challenging and mostly relies on clinical evaluation [2, 3].
In addition to brain deposition of Aβ and tau aggregates, neuroinflammation and synapse failure have been centrally implicated in AD pathogenesis [4–10]. Increased blood-brain barrier permeability, infiltration of peripheral immune cells, and aberrant function of microglia and astrocytes appear to contribute to proinflammatory brain cytokine/chemokine signaling in AD [11–14]. Disease progression further leads to impaired synapse function and neurotransmission, as evidenced by decreased synapse density, network dysfunction, and altered levels of neurotransmitters in the AD brain [13, 16]. Peripheral and brain inflammation, as well as impaired neurotransmission, have further been reported in LBD [17, 18]. Moreover, markers of glial reactivity have been found to be elevated in CSF from patients suffering from a range of neurodegenerative disorders [19].
We have assessed CSF concentrations of a panel of neurotransmitters, cytokines, chemokines, and hormones in a Brazilian memory clinic cohort comprising patients diagnosed with mild cognitive impairment (MCI), AD, LBD, and non-demented controls (NDCs). We found altered CSF levels of noradrenaline (NA), homovanilic acid (HVA), and vascular endothelial growth factor (VEGF), which may reflect pathogenesis and be potentially useful for clinical diagnosis of AD and LBD.
MATERIALS AND METHODS
Study approval and ethics
Experimental procedures involving human CSF were approved by the Committee for Research Ethics of Copa D’Or Hospital, Rio de Janeiro, Brazil (protocol no. 47163715.0.0000.5249). Donors gave written informed consent for use of CSF. Samples were anonymized prior to analyses and all measurements were performed in blinded fashion by trained investigators. All studies were performed according to international ethical regulations and standards.
Study population
From a total of 225 individuals (men and women) initially recruited at a memory clinic at D’Or Institute of Research and Education (IDOR) in Rio de Janeiro, Brazil, 62 met inclusion criteria for the current study (age≥60; native Brazilian Portuguese speakers; formal education ≥8 years; absence of other neurodegenerative, neurodevelopmental, or genetic diseases; no restriction for MRI studies). The cohort studied included non-demented controls (NDC; N = 25), AD (N = 14), amnestic mild cognitive impairment (aMCI; N = 14) and Lewy body dementia (LBD; N = 9). All groups were evaluated with the same extensive clinical, neuropsychological and neuroimaging protocols as described previously [20]. For demographics and biomarker information, see Table 1.
Demographic, clinical and biomarker characteristics of CSF donor subjects
Values are presented as means±SD (range). Statistical significances presented as F (p-value) based on two-tailed one-way ANOVA followed by Holm-Sidak adjustment for multiple comparisons, except for sex and APOE4 (Chi-Square Test, χ2 (p-value), Aβ 42 and tau (Kruskal-Wallis test followed by Dunn’s adjustment for multiple comparisons). Asterisks indicate statistically significant differences from NDC (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001). Aβ40, amyloid-β1–40; Aβ 42, amyloid-β1–42; AD, Alzheimer’s disease; APOE, apolipoprotein E; APOE4, apolipoprotein E4; CSF, cerebrospinal fluid; LBD, Lewy body dementia; MCI, mild cognitive impairment; MMSE, Mini-Mental State Exam; NDC, Non-demented controls; t-tau, total tau.
CSF samples
CSF was collected by lumbar puncture performed around 11 a.m. in all cases to minimize circadian fluctuations in biomarkers. CSF was centrifuged, aliquoted, immediately frozen at –80°C and stored at IDOR, as described [21, 22]. Samples were thawed prior to assays, and further processing was conducted on ice until use. Samples and calibrators were run in duplicates in all analyses.
ELISA
Aβ40, Aβ42, and t-tau were measured using Euroimmun (Lübeck, Germany) ELISA kits. Total APOE levels and APOE4 phenotypes were determined by ELISA (MBL Biosciences; #7635). FABP3 and FABP4 were detected by ELISA (R&D Systems kits DY1678 and DY3150, respectively). Cytokines and chemokines were detected using a commercially available multiplex kit (LHC6003M; Thermo) on a Luminex 100 system with xPONENT software (Luminex Corp.). From a total of thirty cytokines/chemokines investigated (EGF, eotaxin, FGF-basic, G-CSF, GM-CSF, HGF, IFN-α, IFN-γ, IL-1ra, IL-1β, IL-2, IL-2r, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12(p40/p70), IL-13, IL-15, IL-17, IP-10, MCP-1, MIG, MIP-1α, MIP-1β, RANTES, TNF-α, and VEGF), six (IL-8, IP10, MCP-1, MIP-1α, RANTES, and VEGF) had CSF levels consistently above the detection limit in our analyses.
High performance liquid chromatography
Reverse-phase HPLC coupled to electrochemical detection (HPLC-ED; Shimadzu, Japan) was used to determine CSF levels of monoamines and related metabolites (dopamine, DOPAC, HVA, L-DOPA, noradrenaline, serotonin, 5-HIAA), as well as amino acids. Samples were deproteinized by addition of perchloric acid (0.1 M) for 30 min followed by centrifugation (10,000 g for 10 min) to remove protein pellets. Supernatants were used to determine monoamines and amino acids, and standard curves were obtained using commercial reagents of the highest grade. For monoamine determination, fast isocratic separation was performed using a reverse phase LC-18 column (Supelco; 5μm particle size, 250 mm×4.6 mm) with the following mobile phase: 20 mM sodium phosphate dibasic, 20 mM citric acid, 10%methanol, 0.12 mM Na2EDTA, and 566 mg/liter of heptanesulfonic acid, pH 2.64, as described [23]. For amino acid determination, a separate but similar reverse phase C-18 column (Kromasil; 5μm particle size, 250 mm×4.6 mm) was used. OPA-sulfite derivatives of amino acids were generated before each run with the aid of an autosampler, as described [24].
Statistics
Statistical analyses were performed using GraphPad Prism 6 (GraphPad Software Inc., La Jolla, CA). Data were checked for normal distribution using D’Agostino & Pearson Omnibus normality test. When the data passed the normality test, they were analyzed using two-tailed one-way ANOVA with Holm-Sidak post hoc test to assess significance. Data not conforming to normal distributions were analyzed using non-parametric Kruskal-Wallis test followed by Dunn’s post hoc test to assess significance, as indicated in Figure Legends.
RESULTS
Population characteristics
Demographics and AD biomarker results are presented in Table 1. There were no differences in sex or APOE4 status among groups. Age was higher in the AD group compared to NDCs. AD and LBD patients presented evident cognitive impairment, as revealed by Mini-Mental State Exam (MMSE) scores compared to the NDC group. CSF Aβ42 was specifically reduced in the AD group, and no differences were detected in Aβ40 among groups. Total tau (t-tau, a marker of neurodegeneration) was consistently increased in AD (Table 1).
Neurotransmitters and metabolites in CSF
Because AD and LBD present with impairments in several neurotransmitter systems [2, 25–27], we initially determined CSF concentrations of amino acids and monoamine neurotransmitters and their metabolites in both diseases. Results showed that noradrenaline was increased in the AD CSF compared to NDCs (Table 2).
Cytokines, chemokines, hormones and neurotransmitters in CSF
Values are presented as means±SD (range). Statistical significances presented as F (p-value) based on Kruskal-Wallis test followed by Dunn’s adjustment for multiple comparisons, except for RANTES, leptin, arginine, glutamine, 5-HT, and noradrenaline, which passed normality test and were analyzed by one-way ANOVA followed by Holm-Sidak adjustment for multiple comparisons. Asterisks indicate statistically significant differences from NDC (*p < 0.05; **p < 0.01). FABP3, fatty acid binding protein 3; FABP4, fatty acid binding protein 4; IL-8, interleukin-8; IP10, interferon-γ-induced protein 10; MCP-1, monocyte chemoattractant protein 1; MIP, macrophage inflammatory protein; RANTES, regulated on activation, normal T cell expressed and secreted; VEGF, vascular endothelial growth factor; GABA, γ-aminobutyric acid; 5-HIAA, 5-hydroxyindoleacetic acid; 5-HT, 5-hydroxytryptamine; DOPAC, 3,4-dihydroxyphenylacetic acid; HVA, homovanillic acid; L-DOPA, l-3,4-dihydroxyphenylalanine.
Interestingly, a recent study demonstrated that noradrenaline sensitizes Aβ-induced tau phosphorylation in AD mice and in human brain tissue [28]. This prompted us to examine potential correlations between CSF noradrenaline, classical AD biomarkers and cognitive performance in AD and controls. CSF noradrenaline positively correlated with CSF tau (Fig. 1A), but not with CSF Aβ42 or with MMSE scores (Fig. 1B,C).
CSF noradrenaline correlates with total tau. Correlations between CSF noradrenaline and CSF tau (A), CSF Aβ42 (B), or Mini-Mental State Exam (MMSE) scores (C) in non-demented controls (NDC; blue symbols; N = 25) and AD patients (AD; red symbols; N = 14). Lines represent Pearson linear regressions (R2 and p-values as indicated).
We further found that HVA was reduced in AD and LBD CSF compared to NDCs (Table 2). Of note, CSF HVA positively correlated with MMSE scores (Fig. 2A), but not with CSF Aβ42 or tau (not shown). Concentrations of arginine, GABA, glutamate, glutamine, taurine, serotonin, 5-HIAA, L-DOPA, DOPAC, and taurine in CSF were unchanged among patient groups (Table 2).
CSF HVA and VEGF correlate with MMSE scores. Correlations between CSF HVA (A) or VEGF (B) with Mini-Mental State Exam (MMSE) scores in non-demented controls (NDC; blue symbols; N = 25) and LBD patients (LBD; black symbols; N = 9). Lines represent Pearson linear regressions (R2 and p-values as indicated).
Cytokines, chemokines, and hormones in CSF
Since neuroinflammation is a prominent feature of both AD and LBD, we next employed a multiplex assay for simultaneous detection of a panel of cytokines and chemokines in CSF. Of the 30 analytes investigated (see “Methods”), 6 (IL-8, IP10, MCP-1, MIP-1α, RANTES, and VEGF) were detected at concentrations consistently above the detection limit in our analyses. In addition, we determined CSF concentrations of leptin and fatty acid binding proteins 3 (FABP3) and 4 (FABP4), known to play key roles in peripheral inflammation and metabolic disease [29]. CSF VEGF correlated positively with MMSE scores (Fig. 2B), but not with CSF Aβ42 or tau (not shown), and was substantially reduced in LBD compared to NDCs (Table 2). Concentrations of other cytokines, chemokines, and hormones in CSF were unchanged among patient groups (Table 2).
DISCUSSION
The goal of the current study was to determine CSF concentrations of inflammatory mediators and neurotransmitters thought to play relevant roles in AD and LBD. Brain inflammation and synapse defects have been linked to the pathogenesis of both AD and LBD [5, 31], prompting us to interrogate levels of amino acid and monoamine neurotransmitters, cytokines, chemokines and hormones in the CSF of AD and LBD patients. Parallel investigation of a large number of analytes in CSF from the same patients represents an advantage of the current work, as previous studies have largely attempted to measure a single or a limited number of analytes in CSF.
We found that most cytokines and chemokines tested were present at very low levels and could not be reliably detected in human CSF using a commercial multiplex ELISA assay. This is in line with a study indicating that cytokine profiling in the CSF is challenging and may not necessarily reflect clinical diagnosis [32]. Furthermore, the source of these cytokines in the CSF remains to be clarified. Whereas microglia become dysfunctional in AD [11] and neutrophils are known to infiltrate the AD brain [12], recent evidence indicates that immune CD8 T cells patrol the CSF in AD patients [33]. Additional studies may help identify specific neuroinflammatory cues in the AD CSF.
Our finding that NA is elevated in the CSF of AD patients is in line with very recent evidence implicating NA signaling in the pathogenic cascade initiated by Aβ that culminates in aberrant tau phosphorylation and neurodegeneration in AD [28]. Furthermore, plasma NA was recently reported to be elevated and to correlate with memory impairment in AD [34]. Current results thus confer clinical relevance to these recent pathophysiological findings and encourage further studies aimed at dissecting the specific roles of increased central NA in AD.
CSF NA levels further correlated with tau in our investigation. Tau pathology in the locus coeruleus, which harbors NA-producing neurons in the brainstem, is an early marker of AD [35], and is facilitated by the NA metabolite 3,4-dihydroxyphenylglycolaldehyde [36]. Additional evidence indicates reduced levels of NA in postmortem AD brains [37, 38].
Accurate diagnosis of LBD remains challenging due to the lack of effective biomarkers and limited insight into pathogenesis. Our results from a Brazilian cohort showed that HVA is reduced in LBD CSF, consistent with previous findings in Asian and Dutch cohorts [39–41], and add support to the notion that CSF HVA could integrate a potential biomarker panel for LBD. Results are in further agreement with previous studies showing that CSF HVA is reduced in AD [42] and frontotemporal lobar degeneration [43], raising the possibility that CSF HVA could be a marker for multiple forms of dementia. We further report that VEGF is selectively reduced in LBD CSF, consistent with evidence of reduced brain VEGF in LBD [44]. Our results also report divergent trends on CSF GABA in AD and LBD. Because differential diagnosis between LBD and AD comprises an additional clinical challenge, combined measures of CSF VEGF, HVA, and GABA may provide a useful tool for diagnosis of LBD if supported by replication studies in additional cohorts.
The reduced size of the study cohort is a limitation of the current study that precluded a more detailed investigation of the influence of known dementia-linked factors (e.g., sex, APOE status). In addition, future studies investigating the association of biomarker and cognitive trajectories are warranted to shed light on pathogenic mechanisms in AD or other forms of dementia.
In conclusion, our findings of altered CSF concentrations of NA, HVA, and VEGF may reflect underlying pathogenic mechanisms in AD and LBD, and may contribute to biomarker development for these disorders. Future studies encompassing larger and multicentric cohorts are warranted to establish the potential of these biomarkers in clinical practice.
Footnotes
ACKNOWLEDGMENTS
This work was supported by grants from the Brazilian funding agencies Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) (434093/2018-1 and 311487/2019-0 to M.V.L., 467546/2014-2 to F.G.D.F) and Fundação Carlos Chagas Filho de Amparo á Pesquisa do Estado do Rio de Janeiro (FAPERJ) (202.817/2016 and 202.744/2019 to M.V.L., 201.432/2014 to S.T.F., and 202.944/2015 to F.G.D.F.), the National Institute for Translational Neuroscience (INNT/Brazil) (465346/2014-6 to S.T.F. and F.G.F.), the International Society for Neurochemistry (CAEN 1B to M.V.L.), Alzheimer’s Association (AARG-D-615741 to M.V.L.) and intramural grants from the D’Or Institute for Research and Education (IDOR) and Rede D’Or São Luiz Hospital Network. M.V.L., F.C.R, L.E.S, D.B., and H.M.M. were supported by fellowships granted by FAPERJ, CNPq, or Comissão de Aperfeiçoamento de Pessoal de Nível Superior (CAPES/Brazil; financial code 001). The funding bodies had no role in study design, data collection or interpretation, or manuscript writing. We thank Ana Claudia Rangel, Bruno G. Caroli, Bruna Viana, Thainá Moreira, Maíra S. Oliveira, Mariângela V. Melo, and Dr. Pedro M. P. Coelho, from the Federal University of Rio de Janeiro, for experimental and/or administrative assistance. We also thank the IDOR clinical and imaging staff for patient recruitment and diagnostic investigation.
