CCL2 is associated with microglia and macrophage recruitment in chronic traumatic encephalopathy

Post-mortem human brain tissue was obtained from 261 subjects from different study groups using previously described procedures [18‐20]. Different sets of cases were used for histology and immunoassay experiments based on the availability of frozen or formalin-fixed paraffin embedded (FFPE) tissue. Overall, a total of 224 cases with frozen samples were used for the immunoassay experiments and 53 cases with FFPE tissue were used for histology experiments. There were 16 cases that overlapped for both histology and immunoassay analysis and were used for both. Cases for both histology and immunoassay experiments were drawn from multiple brain bank sources. First, 124 individuals that had a history of exposure to American football at either the professional or amateur level were selected from the Understanding Neurological Injury and Traumatic Encephalopathy (UNITE) group consisted of 18 control cases were obtained from the national PTSD brain bank. Control cases lacked a diagnosis of a neurodegenerative disease and did not carry a diagnosis of PTSD. The next group consisted of 119 subjects from clinic and community aging-based brain banks: Boston University Alzheimer’s Disease Center (BU ADC) and Framingham Heart Study (FHS). In the FHS, an athletic history assessment identical to UNITE was performed with the donor’s next of kin [21]. Athletic history was not available for BU ADC participants. In all groups, cases were excluded from the study if they carried a neuropathologic diagnosis of frontotemporal lobe degeneration, neocortical Lewy bodies, or motor neuron disease. A complete description of the neuropathologic analysis is found in the “Neuropathologic examination” section of the methods. Next-of-kin provided written consent for participation and donation. Institutional review board approval for brain donation was obtained through the Boston University Alzheimer’s Disease and CTE center, Human Subjects Institutional Review Board of the Boston University School of Medicine, and Edith Nourse Rogers Memorial Veterans Hospital (Bedford, MA). Demographics, athletic history (type of sports played, level, position, age of first exposure to sports and years playing contact sports), military history (branch, location of service and duration of combat exposure), and traumatic brain injury (TBI) history (including number of concussions) were queried during a telephone interview as detailed previously [22]. Sample data including mean age, gender, and years of playing American football is present in Table 1. A histogram showing the distribution of age of death between all the groups is present in Supplemental figure 1.

Pathological processing and evaluation were conducted using previously published methodology [5, 6]. All brain tissue was processed identically by fixation in periodate-lysine-paraformaldehyde and stored at 4 °C. Brain volume and macroscopic features were recorded during initial processing. Twenty-two sections of paraffin-embedded tissue were stained for Luxol fast blue, hematoxylin and eosin, Bielschowsky’s silver, phosphorylated tau (pTau) (AT8), alpha-synuclein, amyloid-β (Aβ), and phosphorylated TDP-43 using methods described previously [23]. A neuropathological diagnosis of CTE was made using the NINDS criteria [6]. Neuropathological criteria for CTE require at least one perivascular pTau lesion consisting of aggregates in neurons, astrocytes, and cell processes around a small vessel; these pathognomonic lesions (referred to as the CTE lesion) are most often distributed at the depths of the sulci in the cerebral cortex and are distinct from the lesions of aging-related tau astrogliopathy by the presence of neuronal tau pathology [24]. Neuropathological evaluation occurred blinded to the clinical evaluation and was reviewed by four neuropathologists (VA, BH, TS, AM); discrepancies in the neuropathological diagnosis were resolved by consensus conference. Cases were only included in the RHI group if they received a negative neuropathologic diagnosis for AD, neocortical Lewy body disease, frontotemporal lobar degeneration, or motor neuron disease. Cases that received a neuropathologic diagnosis of CTE stage 1 or 2 were grouped together as “Low CTE” (n = 27) while cases that were CTE stage 3 or 4 were grouped as “High CTE” (n = 47). Cases that had a history of playing American football but were not found to have CTE were labeled “RHI without CTE” (n = 20). Individuals in the AD cohorts were grouped for AD using the NIA-Reagan criteria [25]. The criteria was as followed: for “High” there needed to be neuritic plaques and neurofibrillary tangles in the neocortex (CERAD frequent, Braak stage of V/VI) (n = 24), “Intermediate” required moderate neocortical neuritic plaques and neurofibrillary tangles in limbic regions (CERAD moderate, Braak Stage III/IV) (n = 28), “Low” was denoted when there were sparse neuritic plaques and/or neurofibrillary tangles in a more limited distribution and/or severity (CERAD infrequent and/or Braak I/II) (n = 60). Cases in the AD group were excluded from the study if they carried a neuropathologic diagnosis of CTE, neocortical Lewy bodies, frontotemporal lobar degermation, or motor neuron disease.

Flash frozen brain tissue was obtained from the dorsolateral frontal cortex (DLFC) and the calcarine cortex, weighed, and placed on dry ice. Cases were used based on the presence of tissue within the Brain Bank. Not all cases had available calcarine cortex. Freshly prepared, ice cold 5 M guanidine hydrochloride in Tris-buffered saline (20 mM Tris-HCl, 150 mM NaCl, pH 7.4 TBS) containing 1:100 Halt protease inhibitor cocktail (Thermo Scientific) and 1:100 phosphatase inhibitor cocktail 2 & 3 (Sigma) was added to the brain tissue at 5:1 (5 M Guanidine hydrochloride volume (ml): brain wet weight (g)) dilution and homogenized with Qiagen Tissue Lyser LT, at 50 Hz for 5 min. The homogenate was then incubated while rocking overnight at room temperature. Lysate was diluted according to manufacture protocol and spun down at 17,000 g, 4 °C, for 15 min. The supernatant was then applied to Meso Scale Discovery (MSD) Chemokine Panel 1 (human) Kit V-PLEX Plus (Thermo Scientific) and Aβ₄₂ ELISA (CAT K15200E-2), following manufactures protocols. Guanidine hydrochloride extraction methods will result in the extraction of both soluble and insoluble forms of proteins for analysis of total levels. Final concentrations were expressed as pg/g.

Histological staining and analysis of total AT8 pTau density, Iba1+, and CD68+ cell count in the DLFC at the depth of the cortical sulcus was performed using the Aperio ScanScope (Leica) as previously described [26]. For immunofluorescence staining, tissue was extracted from the DLFC, embedded in paraffin, and cut at 20 μm. Immunofluorescence staining was performed using Akoya Bioscience Opal Polaris 7 color manual IHC detection kit as per the manufactures instructions as previously described (tau isoform paper when published). Sections were incubated with antibodies to anti-Iba1 (1:500, Wako), anti-PHF-tau (AT8) (1:1000, Pierce Endogen), anti-TMEM119 (1:100, Abcam), and DAPI. Stained sections were digitized using an Axio Scan.Z1 slide scanner (Zeiss) and visualized using Zen Blue (Zeiss).

Quantification of Iba1+ cell density around CTE lesions as well as microglia vs macrophages density was carried out using Indica Laboratory HALO through manual counts by a blinded observed. To determine the total Iba1+ cell density around CTE lesion blood vessels, three blood vessels that were surrounded by tau and identified as a CTE neuropathologic lesion and three blood vessels not surrounded by tau were selected in each case and used for immunofluorescent analysis. All blood vessels selected were present at the depth of the cortical sulcus in the DLFC. Blood vessels were identified by DAPI stain marking a distinct circular structure and verified with neuropathologists. The number of Iba1+ cells directly contacting the vessel were counted and averaged together.

To determine the abundance of brain derived microglia vs peripherally derived macrophages, dual Iba1 and TMEM119 staining was utilized. As both microglia and macrophages label with Iba1, the presence or absence of TME119 was used to identify cell types [27]. Iba1+ TMEM119+ cells were identified as possible microglia, while Iba1+ TMEM119− cells were identified as possible macrophages. Similar to the total Iba1+ density analysis, blood vessels at the depth of the sulcus in the DLFC were examined to determine if there were changes specific to tau accumulation. Three blood vessels surrounded by tau consistent with a CTE lesion and three blood vessels that lacked tau were selected in each case. The number of Iba1+ TMEM119+ and Iba1+ TMEM119− cells that directly contacted the vessel were counted and averaged together. The number of TMEM119+ and TMEM119− cells were divided by the total Iba1+ cell count to establish percentages of each cell type. Percentages of microglia and macrophages were compared around non-tau and tau containing vessels (control vs lesion vessels). Control cases with no tau pathology were also included. As no vessels contain tau in control cases, only non-tau associated vessels were examined.

Statistical analysis was performed using SPSS (v24, IBM) and Prism (v8, Graphpad Software). Separate two-way ANOVAs were used to examine Iba1+ cell density and the microglia/macrophage ratio around CTE lesioned blood vessel compared to blood vessels that lacked perivascular tau in low and high CTE. Shapiro-Wilk testing revealed CCL2, Aβ42, and AT8 tau density did not have a normal distribution. Since linear regression analysis requires normally distributed data, CCL2, Aβ₄₂, and AT8 tau density were transformed using a rank-based method as previously described [28‐30]. Briefly, the technique transforms the non-normal variable into a percentile rank for each value, then applies the inverse-normal transformation to the ranks to form a variable which consists of normally distributed Z scores. Further Shapiro-Wilk testing demonstrated transformed values had a normal distribution and were sufficient for linear regression analysis. Separate linear regression analyses were run to compare CCL2 in the DLFC to the total years of playing American football, Iba1+ cell density, and CD68+ cell density. One-way ANOVAs with a Kruskal-Wallis post-test was performed to test differences between CCL2 in the DLFC, calcarine cortex, and the DLFC/calcarine cortex ratio. Ordinal regression was used to examine if the DLFC/calcarine cortex ratio increased according to pathologic disease stage. Multiple linear regressions were used to determine which pathologic protein best correlated with CCL2 levels in the DLFC with age at death, Aβ₄₂, and AT8 density as independent predictor variables and CCL2 as the dependent variable across the different analysis groups. Sensitivity analyses run separately within male AD cases and female AD cases were also performed.

To determine if the pathology found in CTE was directly related to increased glial cell recruitment, the Iba1+ cell density directly around blood vessels that presented with pTau pathology (i.e., the CTE pathognomonic lesion) was compared to neighboring blood vessels in the same tissue without pTau pathology (i.e., control vessels) (Fig. 1). Overall, an increase in Iba1+ cells was observed around lesion vessels compared to control vessels in the same individual (Fig. 1a, b). Increases were observed in both low and high stage CTE. To determine if the increase in Iba1+ cells was the result of CNS derived microglia or infiltration of peripheral macrophages, dual Iba1/TMEM119 staining and quantitation was performed. TMEM119 has been previously found to only label CNS derived microglia and not peripheral macrophages [27]. When counting the number of Iba1+/TMEM119+ and Iba1+/TMEM119−, it was observed that the majority of cells around all the blood vessels in control, low, and high stage CTE cases were TMEM119+ (Fig. 1c, d). However, a significant increase in TMEM119− cells was found around the lesion vessel in both low and high stage CTE, where they accounted for 15.8% and 19.0% respectively of the total Iba1+ cells when compared to control vessels that had an average of 5.3% and 6.8% respectively (Fig. 1c).

The mechanism behind the Iba1+ recruitment was investigated next. CCL2 is a potent glial chemokine that that is upregulated after impacts, potentially linking RHI to glial recruitment. Analysis of the number of years playing American football (a correlate to the amount of total RHI received) demonstrated a significant correlation between playing longer and increased protein levels of CCL2 within the DLFC (Fig. 2a). Furthermore, CCL2 levels significantly correlated with the overall Iba1+ cell density (Fig. 2b) and CD68+ inflammatory cell density (Fig. 2c). When examining the slops of each linear regression present in Fig. 2a-c, Iba1 density was found to have a larger slope (Y = 11.41(X)−17.48) when compared to the number of years playing American football (Y = 1.594(X)−89.13) and CD68 density (Y = 1.001(X)−19.85). To rule out age as confounding factor, a multiple linear regression model demonstrated that CCL2 significantly correlated with the Iba1+ cell density (β = 1.466, p = 0.014), but not CD68 cell density (β = 0.035, p = 0.951), independent of age at death (β = 2.955, p = 0.061).

To further examine if CCL2 levels were related to neuropathology, CCL2 protein levels were investigated across CTE stages and through multiple brain regions. In the DLFC, a brain region where CTE pathology can be observed early in disease, CCL2 was found to be increased in both low and high stage CTE when compared to control cases (Fig. 3a). When looking at the calcarine cortex, a brain region that is relatively spared from CTE pathology in all but the most severe cases, only differences between control and high stage CTE were observed (Fig. 3b). In order to determine if the observed CCL2 changes were representative of a brain wide increase of CCL2 or a region-specific increase, CCL2 in the DLFC was standardized to CCL2 levels in the calcarine cortex. When comparing across disease groups using an ANOVA, only high stage CTE had a significantly elevated DLFC/calcarine ratio when compared to control cases (Fig. 3c). However, adjusting for age using an ordinal retrogression analysis, the CCL2 ratio was observed to correlate with the step-by-step increase across control, RHI without CTE, low CTE, and high CTE (estimate = 1.622, p < 0.001), independent of age at death (estimate = 0.055, p < 0.001).

We next wanted to compare CCL2 levels and the differential brain region response, to a similar neurodegenerative disease, Alzheimer’s disease (AD). Using the NIA-Reagan criteria for AD, no significant difference was observed in the DLFC (Fig. 3d) or the calcarine cortex (Fig. 3e). When standardizing DLFC values to calcarine cortex values, no overall difference was observed (Fig. 3f).

Although AD and CTE are related neurodegenerative diseases that can present with both pTau and Aβ pathology, the pathologic protein most commonly believed to drive disease (pTau for CTE and Aβ for AD) differs. This presents an opportunity to examine differential effects of various pathologic proteins on CCL2 levels and explore possible novel interaction pathways. Multiple linear regression analysis was performed to determine if pTau or Aβ was related to CCL2 levels in the DLFC. Overall, the amount of pTau, as measured by AT8 staining, was significantly correlated with CCL2 independent of age at death in both CTE and AD (Table 2). Aβ₄₂ levels, measured by immunoassay, were found to not correlate with CCL2 when grouping all cases together. However, when segregating male and female AD cases, a trend toward a negative correlation between Aβ₄₂ and CCL2 in men was observed (Table 3).