CSF and Brain Indices of Insulin Resistance, Oxidative Stress and Neuro-Inflammation in Early versus Late Alzheimer’s Disease

Alzheimer’s disease (AD) is characterized by progressive impairments in cognitive and behavioral functions with deficits in learning, memory and executive reasoning. Growing evidence points toward brain insulin and insulin-like growth factor (IGF) resistance-mediated metabolic derangements as critical etiologic factors in AD. This suggests that indices of insulin/IGF resistance and their consequences, i.e. oxidative stress, neuro-inflammation, and reduced neuronal plasticity, should be included in biomarker panels for AD. Herein, we examine a range of metabolic, inflammatory, stress, and neuronal plasticity related proteins in early AD, late AD, and aged control postmortem brain, postmortem ventricular fluid (VF), and clinical cerebrospinal fluid (CSF) samples. In AD brain, VF, and CSF samples the trends with respect to alterations in metabolic, neurotrophin, and stress indices were similar, but for pro-inflammatory cytokines, the patterns were discordant. With the greater severities of dementia and neurodegeneration, the differences from control were more pronounced for late AD (VF and brain) than early or moderate AD (brain, VF and CSF). The findings suggest that the inclusion of metabolic, neurotrophin, stress biomarkers in AβPP-Aβ+pTau CSF-based panels could provide more information about the status and progression of neurodegeneration, as well as aid in predicting progression from early- to late-stage AD. Furthermore, standardized multi-targeted molecular assays of neurodegeneration could help streamline postmortem diagnoses, including assessments of AD severity and pathology.


Multiplex ELISA
We used Luminex bead-based multiplex ELISA's to measure immunoreactivity to proinflammatory cytokines, chemokines, growth factors, and insulin-related gut hormones (Millipore or Bio-Rad; Table 1). After re-centrifuging the brain, VF, and CSF samples to remove particulate debris (12000×g/10 minutes, 4°C), the supernatants were filtered (0.45 μm pores). For these assays, the samples were diluted in PBS (brain) or used undiluted (VF and CSF). Samples (200 μl) were incubated with the beads, and captured antigens were detected with biotinylated secondary antibody and phycoerythrin-conjugated Streptavidin according to the manufacturer's protocol. Immunoreactivity was measured in a Bio-Plex 200 system (Bio-Rad, Hercules, CA). Data are expressed as fluorescence light units corrected for protein concentration.

Statistical analysis
Inter-group comparisons of brain tissue results were made by analysis of variance (ANOVA) with linear trend and Fisher LSD post-hoc tests. VF and CSF AD versus control comparisons were made using Student t-tests. Box plots reflect group medians (horizontal bar), 95% confidence interval limits (upper and lower box limits) and range (whiskers). Data were analyzed using GraphPad Prism 6 software (GraphPad Software, Inc., San Diego, CA).  Figure 1F) in AD Braak Stage 6 relative to the other groups. In contrast, 8-OHdG immunoreactivity was significantly elevated in Braak 3-4 AD brains, and only modestly increased at Braak 6 ( Figure 1E) relative to control. This suggests that elevated levels of 8-OHdG may mark early to intermediate stages of neurodegeneration, and that a more sensitive assay may be required to assess DNA damage in late stages of disease. In addition, we observed a significant linear trend for increasing AGE levels and AD severity (F=28.39; P<0.0001), consistent with previous observations [47,48]. AGE marks oxidation related to insulin resistance.
In postmortem VF samples, we detected significantly lower levels of PDGF (P=0.0002), VEGF (P=0.0003), and β-FGF (P=0.0005), and higher levels of HGF (P=0.0001) in AD (Braak Stage 5-6) relative to control ( Table 2). In CSF from patients diagnosed with probable AD (confirmed by postmortem examination), VEGF (P<0.0001) levels were increased while β-FGF (P=0.025) levels were reduced relative to normal aged controls (Table 3). In addition, a trend for reduced NGF (P=0.10) in AD CSF was observed. The slightly higher CSF levels of HGF in the AD group were not statistically significant. Note that the VF and CSF results were concordant with respect to HGF and β-FGF, but discordant with respect to β-NGF, PDGF, and VEGF, indicating that the trends in trophic factor expression/secretion may shift with AD progression.
In contrast to the findings in brain and VF, cytokine and chemokine levels were mainly increased in AD relative to control CSF (Table 3). Significant differences or trends corresponding to increased inflammation in AD were observed for IL-6, IL-16, LIF, and MCP-1. In addition, CSF levels of IL-8, SCF, MIP-1b, and IP-10 were also increased in AD, although the differences from control did not reach statistical significance. However, as observed in brain and VF, evidence for AD-associated CNS suppression of inflammatory mediators was marked by the significant reductions or trends in reduced levels of IL-1β, IL-18, and TNF-α. The findings with respect to neuro-inflammation in CSF at the early and intermediate stages of AD were largely discordant with postmortem brain and VF results. On the other hand, the reduced levels of IL-1β and IL-18, and increased levels of IL-16 and MCP-1 in AD CSF did correspond with the postmortem findings in Braak 6 AD brains ( Figures 5 and 6), and to some extent (IL-1β and IL-16) postmortem VF. Overall, the most informative and consistent correlate of AD was IL-1β suppression in brain, VF, and CSF. In contrast, neuro-inflammatory indices in CSF (early or moderate AD) seem not to inform about the levels of neuroinflammation in brains with Braak Stage 3-4 AD.

Discussion
In AD, significant impairments in brain insulin signaling begin early in the clinical course and progress with disease severity [7,11,14]. A probable role for brain metabolic dysfunction in the pathogenesis of AD is further supported by the: 1) findings of cognitive impairment and neurodegeneration in experimental models of brain insulin/IGF resistance [6,[49][50][51][52][53]; 2) halting or reversal of cognitive deficits by insulin, GLP-1 analog, or insulin sensitizer treatments in humans and experimental animals [18][19][20][21][22][23][24][25][26][27]; and 3) effectiveness of lifestyle changes for reducing insulin resistance and preserving cognition [54][55][56][57]. Despite these conceptual gains, AD diagnostic panels have not been revised to accommodate the metabolic/insulin resistance hypothesis, and instead remain largely focused on detecting altered levels of AβPP-Aβ and pTau in CSF. Data from neuroimaging and human brain studies strongly suggest that CNS metabolic indices, particularly those related to brain insulin signaling, could help with early detection of AD, monitoring the clinical course, and evaluating responses to treatment. This concept is reinforced by evidence of brain mitochondrial dysfunction which reflects significant perturbations in brain energy metabolism and begins early in the course of AD [43,44,58]. Finally, independent evidence suggests that consequences of, or co-factors in brain metabolic dysfunction, e.g. neuroinflammation and oxidative stress, should be considered as they likely exacerbate or perpetuate the cascade of neurodegeneration [15,16,43]. The present study attempts to address these concepts by identifying clusters of additional potential biomarker indices that might be incorporated into AD diagnostic panels. The long-range objectives are to enhance sensitivity and specificity of AD detection, particularly in the early and most treatable phases of disease.
Consistent with the well-characterized neuropathology of AD and studies utilizing pTau and AβPP-Aβ CSF-based biomarker assays to diagnose or monitor AD [59][60][61], we detected increased levels of pTau and AβPP-Aβ in postmortem brains that had intermediate (Braak [3][4] or advanced (Braak 6) stages of AD. In addition, we observed increased levels of lipid peroxidation (4-HNE), advanced glycation end-products, and a trend for increased DNA damage (8-OHdG) with severity of AD. In sporadic AD, which was present in all AD cases in this study, Tau pathology is caused by aberrant activation of kinases that cause hyperphosphorylation of the protein, leading to the formation of insoluble aggregates that undergo ubiquitination. Fibrillar aggregates of pTau promote oxidative injury and stress, which activate or exacerbate AβPP-Aβ pathology, neuro-inflammation, and cell death cascades [62]. Hyper-phosphorylated tau-associated lesions in AD are recognized as neurofibrillary tangles, dystrophic neurites, and neuropil threads, and their accumulations correlate with clinical severity of dementia. AβPP-Aβ pathology results from aberrant cleavage of AβPP, resulting in AβPP-Aβ fibril accumulation. Fibrillar aggregates of AβPP-Aβ undergo ubiquitination and promote oxidative stress, which can trigger or worsen Tau pathology, as well as promote cellular stress-related injury and inflammation [63]. In addition, soluble, diffusible and toxic AβPP-Aβ oligomers, which accumulate late in the course of AD, may have a role in AD progression due to neurotoxic injury [64] and inhibitory effects on insulin signaling [13].
Activation of glycogen synthase kinase 3β (GSK-3β) which has a pivotal role in promoting Tau hyper-phosphorylation [65], is a major consequence of impaired insulin signaling and insulin resistance [66][67][68][69][70]. Increased GSK-3β activation promotes oxidative stress and DNA damage [71], and oxidative stress is sufficient to increase AβPP-Aβ accumulation and Tau phosphorylation [72]. Likewise, insulin resistance promotes brain accumulations of pTau and AβPP-Aβ [49,52]; AβPP-Aβ toxic fibrils impair insulin signaling by down-regulating insulin receptors [13]. Together, these responses promote oxidative stress, neuroinflammation, neurotoxicity, and synaptic dysfunction through a positive feedback loop that exacerbates insulin/IGF resistance [13,15]. Given this scenario, it is likely that multipronged biomarkers that detect different components of the neurodegeneration cascade will provide a more informative and sensitive diagnostic aid for detecting and monitoring AD at different stages of disease. Moreover, this strategy holds promise for early detection of AD, when the disease is most likely to respond to treatment. Lastly, simple, cost-effective biochemical and molecular tools are needed to objectively monitor therapeutic responses, as well as help to streamline postmortem diagnoses of AD.
Herein, we examined three clusters of potential biomarker for detecting AD neurodegeneration: insulin resistance, trophic factors, and inflammatory indices. The goals were to assess: 1) trends in AD-associated abnormalities in insulin resistance-related proteins; 2) AD-associated abnormalities in trophic factors that support different functions in the brain, including neuronal plasticity; and 3) patterns of neuro-inflammation in brain versus VF and CSF.
The results obtained with the multiplex gut hormone panel were reassuring with regard to the roles of insulin resistance and metabolic dysfunction in AD because they reported AD Braak stage declines in insulin and GLP-1, and increases in leptin. The significant reductions in GLP-1 correspond with the reduced insulin levels in brain and CSF. The trends with regard to GIP-1 and PYY were novel and suggest further studies should be done to characterize the full spectrum of AD-associated abnormalities in gut-pancreatic type polypeptides that occur over the course of disease. It is particularly noteworthy that the reductions in GLP-1 and GIP-1 could exacerbate the deficiencies in brain insulin levels and worsen impairments in brain insulin signaling, since both GLP-1 and GIP-1 are incretins with insulinotropic functions, and they are important regulators of glucose metabolism [73] that could be used therapeutically to treat cognitive impairment and neurodegeneration in AD [18,74]. Reduced levels of these polypeptides in AD correlate with decreased levels of insulin, thereby supporting their use in diagnostic panels as well as targeted therapy for AD. In addition, since many of the insulin resistance-related abnormalites in AD also occur in metabolic syndrome which contributes to cognitive decline [75], both CNS and systemic factors mediating brain metabolic dysfunction and insulin resistance could serve as therapeutic targets in AD [75][76][77].
Increased levels of leptin [78] and reduced levels of PYY [79] are features of obesity with peripheral insulin resistance [80]. The presence of similar abnormalities in AD brains suggests that leptin and PYY levels could also serve as indices of brain insulin resistance. Finally, ghrelin, a ligand for growth hormone secretagogue receptor, is downregulated in aging [81] and morbid obesity, which are insulin resistance states [82]. Therefore, reduced levels of ghrelin in AD correspond with brain aging and insulin resistance. The constellation of insulin, GLP-1, and GIP-1 deficiencies, together with alterations in other polypeptides that report brain insulin resistance is consistent with the hypothesis that AD represents Type 3 diabetes with combined features of insulin deficiency and resistance in the brain [12,14]. The consequences of these metabolic derangements were reflected by the increased levels of AGE, HNE, and 8-OHdG in postmortem AD brains.
The findings with respect to trophic factors were interesting because most of the trends showed increased expression levels in relation to AD severity. Exceptions were PDGF and VEGF, which declined, and GDNF, which was unchanged. Increased levels of NGF and BDNF in AD could reflect effects of receptor resistance, particularly given the impairments in neuronal plasticity and the roles these neurotrophins play in synaptic remodeling [83,84]. Despite its name, HGF is expressed in the brain, particularly the hippocampus and may have neurotrophic properties [85]. Its prominent localization in the CA3-CA4 regions of the hippocampus [85] where neurogenesis occurs [86], further suggests that HGF plays a key role in maintaining neuronal populations as well as mediating synaptic plasticity. The increased levels of HGF in AD brains corresponds with previous observations [87], and the somewhat higher levels in AD VF and CSF could also reflect HGF receptor resistance given the progressive impairments in neuronal plasticity that occur with AD progression. The findings with regard to β-FGF are entirely consistent with earlier observations in human postmortem brains [88]. Previous studies correlated increased β-FGF expression in AD with increased gliosis [88], which characteristically marks several aspects of neurodegeneration, including loss of neurons and fibers.
VEGF is expressed in microglia and endothelial cells. Alterations in VEGF expression occur in cerebral microvascular disease and in AD. In addition to its role in angiogenesis, VEGF has neuroprotective actions that may have relevance for treatment of AD and other neurodegenerative diseases [89]. In this regard, low VEGF levels have been shown to mediate neurodegeneration, which could be due to hypoperfusion or reduced neuronal protection from oxidative stress [89]. Therefore, reduced levels of VEGF in AD brains and VF could mark the presence and/or severity of neurodegeneration mediated by brain hypoperfusion and neuronal death. This concept opens the door to additional treatment modalities for AD, as well as investigating whether the VEGF responses in AD are primary or secondary. For example, insulin and IGF-1 regulate expression of VEGF [90], and impairments in brain insulin and IGF-1 levels begin early in the course of AD [11].
Platelet-derived growth factor (PDGF) mediates β-γ secretase mediated cleavage of AβPP [91]. In addition, PDGF-BB, which is only expressed in neurons, is abundant in neurofibrillary tangles and associated with synaptic loss and dystrophic sprouting, whereas PDGF-AA is vascular associated [92] and mediates oligodendrocyte development. PDGF-AA, which was measured in the gut hormone panel, has an important role in myelin maintenance [93]. PDGF-A receptor is regulated by β-FGF [93], and PDGF regulates oligodendrocyte progenitor cells functions, including myelination [94,95]. Therefore, the reduced levels of PDGF in AD correspond with the previously demonstrated early loss of white matter and hypomyelination in this disease [96,97]. The fact that PDGF expression was reduced in AD brain and VF samples but not in the clinical CSF samples suggests that these abnormalities may be detectable in CSF only in the later stages of disease.
Neuro-inflammation remains a focus of research in AD because it occurs early in the course of disease [98], and already has been addressed in several clinical trials [99,100]. The failure to obtain conclusive evidence that anti-inflammatory measures are neuroprotective and can halt neurodegeneration most likely reflects the complexity and non-static nature of the problem. For example, inflammation may mediate disease at selected stages rather than throughout its clinical course. Multiplex ELISAs are an efficient way to simultaneously assess arrays of pro-inflammatory mediators in human subject material. A major unexpected finding was the broad-based suppression rather than activation of pro-inflammatory mediators in AD brains and postmortem ventricular fluid. In brain tissue, only IL-16, TNFα, MCP-1, and Interferon-γ were elevated at either Braak Stage 3-4 or 6. For the other 12 cytokines/chemokines measured, 8 were expressed at significantly lower levels in brains with Braak Stage 6 or both Braak 3-4 and 6 AD relative to control. Similarly, in VF samples, only 5 of the 16 cytokines measured were elevated in AD but none of those differences were statistically significant. In contrast, in CSF, 4 cytokines were significantly elevated in AD, and 11 were moderately although not significantly elevated relative to control. However, since TNF-α and Interferon-γ expression were elevated in brains with Braak Stage 3-4 but not Braak 6 AD, and higher percentages of the inflammatory mediators were up-regulated in CSF as compared with brain or VF, conceivably the activation of neuro-inflammation occurs early in the course of AD, but as disease progresses, neuroinflammation subsides or is suppressed. The mechanisms and consequences of these responses are not known. However, the findings suggest that as a tool for evaluating AD diagnosis, severity, and responses to treatment, pro-inflammatory cytokines do not represent viable targets. On the other hand, the higher levels and profiles of inflammatory mediators in the clinical CSF samples from patients with probable AD suggest that anti-inflammatory therapeutic approaches may have value in the early stages of disease.
In conclusion, this study demonstrates the utility of evaluating indices of insulin resistance, neuronal plasticity, glial function, and oxidative stress in conjunction with pTau and AβPP-Aβ in CSF-based multiplex assays. This multi-pronged approach to assess different aspects of the neurodegeneration cascade will likely be more informative with respect to using streamlined biochemical and molecular assays for clinical as well as postmortem diagnoses, monitoring the clinical course of AD, and evaluating responses to treatment. This study suggests that the use of neuro-inflammatory markers will likely not be beneficial due to the transient nature of their activation in relation to disease severity. Future studies should assess the time course of shifts in biomarker indices in relation to cognitive decline and structural and functional neuroimaging abnormalities.      Trophic factors and cytokines probed in alzheimer brain, ventricular fluid and cerebrospinal fluid samples.

Trophic Factor Abbreviation Function
Basic Fibroblast Growth Factor b-FGF Present in basement membranes and sub-endothelial extracellular matrix.
Regulates angiogenesis and cell survival, division, differentiation, and migration.
Modulates nervous system development and wound healing.
Beta-Nerve Growth Factor β-NGF Neurotrophin family member that regulates survival and maintenance of sensory and sympathetic neurons. Implicated in neuronal growth, proliferation, differentiation and plasticity, as well as cognition. Functions through receptor tyrosine kinase. Incretin whose secretion is regulated by nutrients, e.g. carbohydrate, protein, and lipid. Promotes glucagon-dependent stimulation of insulin secretion, and survival and proliferation of pancreatic beta cells. Enhances insulin sensitivity and satiety.

Hepatocyte Growth Factor HGF
Typically secreted by mesenchymal cells with actions on epithelial and endothelial cells. Mediates embryogenesis. Stimulates mitogenesis, cell motility, matrix invasion and angiogenesis via c-MET receptor tyrosine kinase. Regulates VEGF. Neuroprotective for cortical and hippocampal neurons during aging and ischemic injury.

Insulin INS
Reduces blood glucose. Increases cellular permeability to monosaccharides, amino acids and fatty acids. Increases rates of glycolysis, pentose phosphate cycle, and glycogen synthesis in liver.
Leptin LEP Produced in adipocytes and regulates brain energy intake and expenditure, metabolism, and behavior.

Pancreatic Polypeptide (Human) PP
Polypeptide secreted by PP endocrine cells in pancreas in response to hypoglycemia, fasting, or protein meal and decreased by glucose infusion or somatostatin. Closely related to neuropeptide Y and PP.
Peptide YY (tyrosine-tyrosine) PYY Secreted by intestinal L cells in response to feeding. Reduces appetite. Also produced in brainstem neurons, pancreatic islets. Improves nutrient absorption by slowing gastric motility and emptying.
Platelet-derived Growth Factor-AA PDGF-AA Regulates cell growth and angiogenesis, and mitogenic for glial and mesenchymal cells. Signals through PI3 Kinase to regulate cell growth and motility, tissue remodeling, differentiation, and migration. Maintains proliferation of oligodendrocyte progenitor cells.

Vascular Endothelial Growth Factor VEGF
Stimulates angiogenesis and vasculogenesis and endothelial cell growth. Inhibits apoptosis and induces vascular permeability, revascularization of injured tissue, endothelial cell migration and proliferation.

Cytokine/Chemokine Abbreviation Function
Granulocyte Macrophage Colony-Stimulating Factor GM-CSF Stimulates the growth and differentiation of hematopoietic precursor cells from various lineages, including granulocytes, macrophages, eosinophils and erythrocytes.
Interferon-gamma IFN-γ Produced by innate NK cells, acquired antigen-specific cytotoxic CD4+ and effector CD8+ T cells. Activates macrophages and critical for innate and adaptive