Non-Phosphorylated Tau in Cerebrospinal Fluid is a Marker of Alzheimer’s Disease Continuum in Young Urbanites Exposed to Air Pollution

INTRODUCTION

The notion that in the scenario of high exposures to air pollution, Alzheimer’s disease (AD) pathology starts very early, progresses relentlessly in the first two decades of life, and is indicative of an active individual complex response to environmental and genetic factors is at the core of our research [1–17]. We have identified subcortical pretangle stage b [18] and olfactory bulb hyperphosphorylated tau and Lewy neurites in <12-month-old residents in the highly polluted Metropolitan Mexico City (MMC) [15, 16]. Cortical tau pre-tangles, neurofibrillary tangles (NFT) Stages I-II, amyloid phases 1-2, tau phosphorylated at threonine 181 (P-Tau) in substantia nigra and auditory, oculomotor, and trigeminal nuclei are identified by the 2nd decade and progression to NFT stages III-V is documented in 24.8 % of 30–40-year-old subjects [18, 19]. Apolipoprotein E allele 4 (APOE4) carriers have 23.6 times higher odds of NFT V (p < 0.0001) versus APOE4 non-carriers having similar cumulative fine particulate matter (CPM_{2 ·5}) exposure, and age [15]. The issue of combustion-derived nanoparticles (CDNPs) associated with early and progressive damage to the neurovascular unit and pathology in mitochondria, endoplasmic reticulum (ER), mitochondria-ER contacts (MERCs), axons, and dendrites in MMC residents and the extensive literature of nanoparticles (NPs), their transportation to the brain and their direct association with aggregation and propagation of abnormal proteins are key for our current work [20–28]. Ambient air NPs can be very high and even the smallest NPs, called nanocluster aerosol particles, emitted by road transportation (1.3–3.0 nm) can contribute more than 50% to the ambient air particle number concentration [22].

We have shown that MMC children cerebrospinal fluid (CSF) samples have significant reductions in amyloid-β (Aβ)_{1 - 42} compared to clean air controls [13], and significant increases in CSF Macrophage Inhibitory Factor (MIF), IL6, IL1ra, and IL-2, versus controls, strongly suggest there is an important neuro-immune and inflammatory intrathecal component at play in MMC children [7]. Eighty-seven percent of MMC children also exhibit severe deficiency of serum 25-hydroxyvitamin D (<30 ng/mL) and given the role of vitamin D in immune regulation, cognitive impairment, dementia, and the association between maternal 25(OH) D status with childhood attention and executive function, we also pursued the measurements of vitamin D in CSF samples in this work [29–31]. Given that tau alterations [total tau (T-Tau), P-Tau] are at the core of CSF diagnostic biomarkers in AD [32], and since CSF tau occurs as a subset of distinct tau species [33], we explored the importance of non-P-Tau epitopes, described as potential biomarkers for axonal damage and AD [34, 35] in the CSF of MMC versus lower air pollution controls, knowing beforehand that both T-Tau and P-Tau are not significantly different in MMC children versus controls [10].

Non-P-Tau CSF has been described in high concentrations in Creutzfeldt-Jakob disease and is elevated in AD [34]. Non-P-Tau has been suggested as a reliable biomarker for AD in stage of mild cognitive impairment (MCI) or dementia [35]. The work of Lewczuk and Ermann describe non-P-Tau with high sensitivity and specificity for discrimination of the targeted groups [34, 35].

Thus, our selection of the monoclonal tau antibody with strong affinity for two KTTP motifs without phosphorylation: T175 and T181 [35]. Since the epitope resides in the N-terminal region of the tau protein, the selected monoclonal is favorable for CSF detection, because C-terminal tau fragments are almost absent in CSF [33].

Since young residents in Mexico City [15] have the biomarker profiles corresponding to the AD continuum [36], to further examine the damage to the supratentorial white matter (WM) [37–39], we did transmission electron microscopy in frontal samples from previously staged AD autopsy cases [15] (Supplementary Tables 1 and 2) and measured axonal diameters in anterior cingulate cortex (ACC) WM from 4 MMC and 2 controls (age 16.16±2.92 y).

Our results strongly suggest that non-P-Tau discriminates highly exposed young urbanites from low pollution controls and more importantly puts tau research into focus in the early AD stages, 5 to 7 decades away from the late NFT stages and terminal AD pathology. The increases in non-P-Tau support that tau changes are early, potentially related to axonal damage and the accumulation of cortical hyperphosphorylated tau [37–39] and seen across tau species in highly exposed populations.

METHODS

This prospective pilot study was approved by the review boards and ethics committees at the hospitals involved. Normal CSF samples were obtained from two cohorts: 1) children admitted to the Mexico City hospital from a clean air city or Mexico City, with a work up diagnosis of acute lymphoblastic leukemia (ALL) entering a clinical protocol, which included a spinal tap and 2) young adults admitted to the hospital for a work-up that required a spinal tap, with varied diagnoses including potential neoplastic and infectious involvement of the CNS, leukemia, and migraine headache. The cohort of young adults also included permanent residents in clean environments and lifetime residents in highly polluted Mexico City. The normal CSF samples used for this study were destined to be destroyed after the diagnosis of normal CSF was completed.

RESULTS

Electron microscopic axonal findings

The major finding in Mexico City residents was the extensive WM axonal changes, the widened Virchow-Robin spaces and the deposition of NPs in axonal organelles as well in glial cells (Figs. 1 and 2). NPs in the size range of 13–73 nm, average size 43.9±23 nm, exhibited the iron-rich strongly magnetic CDNP morphology: rounded particles, reflecting crystallization upon cooling from an initially heated, iron-bearing source material [49]. CDNPs were seen in mitochondria, ER, MERCs, axons, and dendrites (Fig. 2).

OPEN IN VIEWER

Fig. 1.

Representative one micron toluidine blue anterior cingulate white matter sections from Metropolitan Mexico City and low pollution control subjects in the second decade of life. A) Fourteen-year-old control male (CTL1) with unremarkable white matter and unremarkable myelinated large, medium and small axons. 1μm Toluidine blue. Scale bar 10μm. B) Fourteen-year-old female (MMC2) with areas where large and medium myelinated axons are few at the same power as the clean air child (A). 1μm Toluidine blue. Scale bar 10μm. C) Thirteen-year-old female (MMC1) in an area where myelinated axons are few (arrows) and mostly small. Several areas of the neuropil show no structures (^*). Three glial cells, presumably oligodendrocytes (OL), show round nuclei with heterochromatin beneath the nuclear envelope and patches of heterochromatin throughout the nucleoplasm. The cytoplasm in any of them is not clearly defined. 1μm Toluidine blue. Scale bar 10μm. D) Twenty-year-old male (MMC4) with widened Virchow-Robin spaces and neuropil spaces with no visible structures (^*). Blood vessels (BV) are seen along oligodendroglia-like nuclei (OL), and few small and medium size axons (arrows). 1μm Toluidine blue. Scale bar 10μm. E) Seventeen-year-old male, APOE 3/4, showing extensive enlargement of the Virchow-Robin spaces and empty neuropil spaces (^*). Few small and medium size axons (arrows) and larger axons (arrowheads) are seen. 1μm Toluidine blue. Scale bar 10μm.

OPEN IN VIEWER

Fig. 2.

A) Fourteen-year-old female (MMC2), a small number of medium (arrowheads) and small myelinated axons (arrows) in a loose neuropil. Empty neuropil spaces (^*). Scale bar 2μm. B) Twenty-year-old male (MMC4) with abundant lipofuscin (Lf) in pericytes. The red blood cell in the lumen of the vessel is marked RBC and empty neuropil spaces (^*) are numerous. Smaller myelinated axons are few (arrows). Scale bar 2μm. C) Close-up of the lipofuscin (Lf) in pericytes in a 20-year-old male. Scale bar 2μm. Nanoparticles are seen associated to Lf and throughout the Lf (close-up of the white rectangle). Mag 30, 000x. Scale bar 500 nm. D) Seventeen-year-old APOE4 male (MMC3), myelinated axons with round nanoparticles (arrowhead). Scale bar 500 nm. E) Fourteen-year-old girl (MMC2) several nanoparticles (arrows) are free in the axoplasma. Focal myelin sheet disintegration is present in several places (arrowheads). Scale bar 500 nm. F) Seventeen-year-old male (MMC3), oligodendrocyte with numerous nanoparticles inside the nucleus (arrowheads), in the cytoplasm (arrows) and inside mitochondria (M). Scale bar 500 nm. G) Same 17-year-old (MMC3) with numerous nanoparticles inside mitochondria (M) and in ill-defined structures in the axoplasm (arrowheads). The larger NP (upper left) measures 73 nm. Scale bar 100 nm.

DISCUSSION

Non-P-Tau CSF concentrations are increasing significantly with age in a cohort of 507 normal CSF samples from MMC versus clean air controls, age 12.82±6.73 y. This finding is in accordance with our previous reports of early axonal damage associated with CDNPs and the evolving AD neuropathology in the first two decades of life [15–17]. The early development of AD pathology (the AD continuum [36]) and the ubiquitous brain presence of iron-rich CDNPs in Mexico City residents [49] associated with axonal mitochondria, ER, and MERCs damage is a plausible explanation for the increases in axonal damage biomarkers in children, teens, and young adults. In the context of CSF biomarkers and PM_2.5 research we have shown Aβ_{1 - 42}, BDNF, MIF, Cellular prion protein (PrP^C), IL6, IL1ra, and IL-2, are key in differentiating young MMC versus clean air controls [7, 13]. These results point into at least three directions: 1) classically accepted AD biomarkers [32] including the low concentrations of Aβ_{1 - 42} in young MMC residents: a prime example of a marker associated with the deposit of extracellular amyloid previously shown in MMC children and teens autopsy cases [15, 16]; 2) increased concentrations of inflammatory and cytokine markers such as MIF, IL6, IFN γ, and IL2 suggesting a dysregulated neural immune response in keeping with the brain early and significant inflammatory changes [3, 6]; and last but not least, 3) the strong evidence that a number of neuroprotectors are no longer accomplishing their functions in highly exposed children (cellular prion protein and BDNF fall in this category) [50–54].

The issue of WM pathology, through a number of pathways, and its contribution to functional decline and AD pathogenesis is at the core of this work. Small blood vessel disease, amyloid angiopathy, neurovascular instability resulting in WM ischemic damage, neurofilament light alterations, and hypoperfusion could be critical in highly exposed individuals [55–58]. Endothelial cell dysfunction is also potentially a key factor altering myelination and blocking oligodendroglial differentiation [59].

There is clear evidence that patients with MCI and AD [60] develop microstructural WM changes occurring during the AD course, APOE strongly influences regional WM axonal density loss in AD [61], and there is a correlation between CSF Aβ levels, WM lesion loads, and diffusion-weighted images suggesting the pattern characterization of WM changes might be useful for assessing AD progression [62]. The work of McAleese et al. and Mito et al. [37–39, 63] ought to be mentioned in the context of WM damage in AD and its association with Wallerian degeneration as a consequence of P-Tau cortical burden. Mito and co-workers [63] emphasized the involvement of long association fibers, while McAlesse et al. [37] reported P-Tau cortical pathology is significantly associated or is a predictor of white matter hyperintensities (WMH) in parietal and temporal regions.

Jackson et al. [64] discussed WM tauopathy and damage of axonal myelin lamella and myelin remodeling, while Ferrer [65] discusses oligodendrocytes and their precursors, development and functions, and the importance of dysfunction of oligodendrocytes in selected neurodegenerative diseases with abnormal protein aggregation. Jackson and colleagues [64] support that WM dysfunction early in tauopathy will give rise to altered neural circuits and potentially contribute to early clinical deficits in dementia, a situation we cannot dismiss in MMC children with significant cognitive deficits as early as age 10.3±2.4 y [4], and young adult cohorts age 20.14±2.39 y with mean Montreal Cognitive Assessment scores of 22.76±2.09 (personal communication Dr. Lilian Calderón-Garcidueñas, September 25, 2018). The work of Dr. Roxana Carare [66–68] ought to be discussed in the context of significant neurovascular unit and basement membrane damage as we have described in MMC children and young adults [3, 69]. Specifically, the integrity of periarterial basement membrane pathways altering Intramural Peri-Arterial Drainage pathways for the elimination of interstitial fluid and solutes from the brain [66, 67]; the dilated perivascular spaces in amyloid angiopathy [68] and certainly the presence of cerebral small vessel disease [70] in our young exposed urbanites are all potential pathways to explain WM damage. Poggi et al. [71] discussed progressive hypomyelination in Mbp heterozygous mice, the unaffected parental strain of shiverer, a classical neurological mutant. In the work of Poggi et al., the key is that subtle abnormalities of CNS myelin can cause persistent cortical network dysfunction, and, as discussed by Gao et al. [72] in a rat model, WM atrophy and disruption of myelinated fibers may contribute to the pathophysiology underlying depression. Also, in the context of vascular pathology, we have described in each MMC case, extensive microvascular leakage related to severe damage to the neurovascular unit, a pathology described in diabetes mellitus patients’ brains [73]. The issue of breakdown of the blood-brain barrier is indeed relevant to highly exposed urbanites, because is one of the earliest findings in our exposed infants [15].

The Strain et al. paper [74] is also highly relevant to our work. They evaluated 59 cognitively normal subjects and 10 cognitively impaired individuals with diffusion tensor imaging, PET Aβ ([¹⁸F] AV-45 florbetapir), and PET tau ([¹⁸F] AV-1451 flortaucipir) imaging. Their results are indeed very exciting and not really a surprise: Tau, not Aβ, was associated with changes in anterior temporal WM integrity. Highly relevant to our work: WMH, a proxy for vascular damage, was strongly associated with axonal damage, and tau independently contributed to the model, suggesting an additional degenerative mechanism within tracts projecting from regions vulnerable to AD pathology. Strain et al. [74] concluded WM decline was associated with early tau accumulation, and certainly we support the same mechanism could be at play in Mexico City, where we documented WMH in 55% of children at age 10.73±2.7 y [4].

Microstructural WM changes are also significant in APOE4 carriers [61, 75] and we have shown significant interactive and additive influences of gender, body mass index, and APOE4 on cognition in MMC children, teens, and young adults [12, 69]. Operto et al. [75] examined by diffusion-weighted imaging 532 cognitively healthy middle-aged participants including 68 APOE ɛ4 homozygotes, 207 heterozygotes, and 257 non-carriers. APOE ɛ4 homozygotes display increased WM diffusivity in regions known to be affected by AD. The effects in radial water diffusivity suggested a disruption of the myelin sheath rather than pure axonal damage and it was interpreted by the authors as resulting from a reduced capacity of the ɛ4 isoform of the APOE protein to keep cholesterol homeostasis in the brain. The work of Slattery et al. [61] is equally relevant. Diffusion tensor imaging and neurite orientation dispersion and density imaging with tract-based spatial statistics was used to investigate APOE4 modulation of WM damage in 37 patients with young onset AD (22, 59% APOE ɛ4 carriers) and 23 age-matched controls. WM disruption was more widespread in APOE4 individuals and fractional anisotropy changes were related to combinations of axonal loss and morphological change. So, indeed, APOE4 status is associated with different patterns of WM neurodegeneration [61]. What is clear so far in our cohorts: WM decline starts early, is associated with tau accumulation, and progresses with age [15–17].

CSF alterations are critical in the early development of AD and tau biomarkers are at the core of early AD diagnosis for intervention trials and clinical practice [35]. Non-P-Tau measurements in young highly exposed air pollution cohorts puts forward a candidate biomarker that distinguishes low from highly exposed individuals along with classical AD markers (Aβ_{1 - 42}). More importantly, non-P-Tau is likely reflecting axonal damage in association with early AD development in a very young population where CDNPs are key players and hyperphosphorylated tau in infra and supratentorial locations is a key likely irreversible finding [18].

Low concentrations of Aβ_{1 - 42} are associated to diffuse cortical Aβ_{1 - 42} plaques (Aβ Phase 2 [19]) present in about 80% of children and young adults in the first two decades of life (versus 0% in clean air controls) and supported by the literature associating long-term exposure to PM_2.5 above current USEPA standards and increased AD risk [76, 77]. We know that concentrations of Aβ_{1 - 42} reach maximum abnormality level in the asymptomatic stage of AD [78], and there is a consensus in AD familial cases of a trend for the decrease in CSF Aβ_{1 - 42} at least four decades before expected onset of clinical symptoms [79]. There is also a current agreement that CSF changes in Aβ_{1 - 42}, T-tau, and P-Tau are diagnostic of AD in its prodromal stage and have proved diagnostic accuracy for MCI and AD [32, 80–83]. On the other side of the coin, de Leon et al. [84] have shown the prediction of cognitive decline was improved by considering both high and low levels of Aβ_{1 - 42} and more importantly, their data suggest an earlier preclinical stage marked by CSF elevations in tau and accompanied by either elevations or reductions in Aβ_{1 - 42}. On the same line of discussion is the work of Merluzzi et al. [85] showing that cognitively unimpaired individuals with AD neuropathology differ from individuals with AD dementia on biomarkers of neurodegeneration, synaptic dysfunction, and glial activation. In their work, a group of individuals with p-tau/Aβ₄₂ and Aβ₄₂/Aβ₄₀ levels comparable to those of the AD-Dementia group had significantly lower levels of neurofilament light and total tau. Merluzzi et al. [85] concluded that monitoring multiple biomarkers provides a much better insight into the biological processes phenotypic of dementia. We fully agree.

Low concentrations of BDNF in young urbanites deserves comment. BDNF has key roles in cell survival, axonal and dendritic growth, and synaptic plasticity and is the most abundant neurotrophin in the brain [86–88]. A common single-nucleotide polymorphism in the pro-region of the human BDNF gene, resulting in a valine to methionine substitution (Val66Met), has been associated with the susceptibility, incidence, and clinical features of neurodegenerative disorders [89]. The involvement of BDNF in the regulation of psychomotor speed, working memory, and executive function has been shown in healthy subjects [90]. Of key relevance for this work, CSF BDNF levels are significantly reduced in AD compared to MCI and healthy controls (p = 0.009) [88]. And equally important, logistic regression models show that lower CSF BDNF levels, lower Aβ_{1 - 42} and lower Mini-Mental State Examination scores are significantly associated with progression from MCI to AD [88]. Thus, the finding of low CSF BDNF in Mexico City young is an expected finding. Goltz et al. [91] explored the putative associations and interactions between serum BDNF and vitamin D levels with depressive symptoms and abdominal obesity in a large population-based cohort. Vitamin D was inversely associated with depression and obesity, while BDNF was associated with abdominal obesity, but not with depression. Interestingly, in the MMC cohort, vitamin D level was negatively correlated with BDNF (p = 0.0294), a finding that obligates us to run both analytes in future CSF studies in young urban residents along with APOE4 genotyping [91–93].

Cellular prion protein (PrP^C), a conserved GPI-anchored membrane protein with neuroprotective activity, located on the surface of neurons, at both pre and post-synaptic sites showed significant differences in its correlations with T-Tau and P-Tau between MMC and control cases, raising questions about physiological and pathological interactions between PrP^C and tau under high pollution exposures. In 2009, Laurén et al. [94] identified PrP^C as a high-affinity receptor for Aβ oligomers. The role of PrP^C in protecting neurons from threatening Ca^{2 +} overloads and excitotoxicity [95], the interaction of the N-terminal domain (N-PrP) with neurotoxic ligands, such as Aβ and Scrapie prion protein [96], the modulation of glucose homeostasis and the promotion of iron uptake through divalent-metal-transporters [97], and the unexpected interaction between inflammatory mediators and reduced PrP^C on endothelium [98] certainly provides additional mechanisms by which PrP^C alterations may lead to CNS damage. The importance of PrP^C cannot be ignored in subjects exposed to air pollutants for one major reason: our previous report of significant 15-fold downregulation of PrP^C in the frontal cortex of 18±8.7 y Mexico City residents [6]. The downregulation of PrP^C in the prefrontal cortex is likely detrimental given the protective effects of PrP^C against oxidative stress and its role in cell signaling, metal interactions, memory, T cell differentiation, myelin maintenance, axonal growth and neuronal development [94–99].

Finally, the significant differences in a T helper Th1 cytokine between MMC and controls: Interferon γ (IFN γ) deserve a comment. IFN γ is a prototypical pro-inflammatory Th1 cytokine that drives cell-mediated immunity. Ott et al. [100] described the results of altered blood-cerebrospinal fluid barrier (BCSFB) function in MCI and AD patients versus cognitively healthy controls. Paired serum and CSF samples were assayed for a panel of cytokines, chemokines, and related trophic factors and dominance analysis models were conducted to determine the relative importance of the inflammatory factors in relationship to BCSFB permeability, as measured by CSF/serum ratios for urea, creatinine, and albumin. Their results showed severe perturbations in BCSFB function in MCI and AD and IFN γ was one of the inflammatory proteins with a shifting distribution from serum to CSF [100]. Otto and coworkers’ report is key for our findings, since highly exposed children had IFN γ concentrations lower than the controls and thus dysregulation of immune responses could be an early finding in the AD continuum associated with early neurovascular unit pathology [3, 12].