Increased prefrontal cortex interleukin-2 protein levels and shift in the peripheral T cell population in progressive supranuclear palsy patients

Accumulating evidence suggests neuroinflammation to be an integrated feature of neurodegeneration. Profiling inflammatory mediators across diseases may reveal common and disease-specific signatures. Here, we focused on progressive supranuclear palsy (PSP), a tauopathy presenting motor and cognitive dysfunction. We screened for 21 cytokines and growth factors in the dorsomedial prefrontal cortex of 16 PSP and 16 control brains using different quantitative techniques. We found and validated increased interleukin (IL)-2 protein levels in the PSP group expressed locally by neurons and glia cells. We further investigated central players in neuroinflammatory pathways and found increased mRNA expression of glycogen synthase kinase 3 beta (GSK3B). IL-2 and GSK3B proteins are T and natural killer (NK) cell regulators and have previously been associated with other neurodegenerative diseases such as Alzheimer’s disease, Parkinson’s disease and multiple system atrophy. In addition, we identified a shift in peripheral CD4+ and CD8+ T cell populations toward increased numbers of memory and reduced numbers of naive T cells. We also observed increased numbers of CD56+ NK cells, but not of CD56+CD57+ or CD57+ NK cells. Our findings suggest a role for IL-2 in PSP disease processes and point toward active and possibly dysfunctional peripheral immune responses in these patients.

Accumulating evidence suggests neuroinflammation to be an integrated feature of neurodegeneration. Profiling inflammatory mediators across diseases may reveal common and disease-specific signatures. Here, we focused on progressive supranuclear palsy (PSP), a tauopathy presenting motor and cognitive dysfunction. We screened for 21 cytokines and growth factors in the dorsomedial prefrontal cortex of 16 PSP and 16 control brains using different quantitative techniques. We found and validated increased interleukin (IL)-2 protein levels in the PSP group expressed locally by neurons and glia cells. We further investigated central players in neuroinflammatory pathways and found increased mRNA expression of glycogen synthase kinase 3 beta (GSK3B). IL-2 and GSK3B proteins are T and natural killer (NK) cell regulators and have previously been associated with other neurodegenerative diseases such as Alzheimer's disease, Parkinson's disease and multiple system atrophy. In addition, we identified a shift in peripheral CD4 + and CD8 + t cell populations toward increased numbers of memory and reduced numbers of naive T cells. We also observed increased numbers of CD56 + NK cells, but not of CD56 + CD57 + or CD57 + NK cells. Our findings suggest a role for IL-2 in PSP disease processes and point toward active and possibly dysfunctional peripheral immune responses in these patients.
Progressive supranuclear palsy (PSP) is a progressive, neurodegenerative disease that shares clinical features with other parkinsonian disorders and Alzheimer's disease (AD) 1 . Neuropathologically, PSP is characterized by accumulation of tau protein in neurofibrillary tangles (NFT) in tufted astrocytes. The highest load of NFTs are localized in different areas of the basal ganglia and the brainstem, but cortical areas including the frontal lobe are also affected 2,3 . Brain imaging studies have shown that brain atrophy in PSP patients is accompanied by microglial activation in the frontal cortex 4,5 . This suggests that neuroinflammatory processes are part of the course of the disease. Still, there is a knowledge gap regarding how neuroimmunomodulating factors are affected in PSP brains. The few existing studies approaching this topic have focused solely on gene expression without reporting protein levels 6,7 . This information is important, not only for understanding the pathophysiology behind this disease, but also for illustrating to what extend neuroinflammatory processes are a common manifestation in neurodegenerative diseases. To our knowledge, no study has investigated the immune cell profile in PSP patients to shed light on the possible involvement of the peripheral immune system in PSP.
The primary aim of the present study was to investigate the neuroinflammatory signaling profile in the dorsomedial prefrontal cortex (dmPFC) of PSP patients. Using multiplex assays, we screened for protein levels of 21 cytokines and growth factors in post-mortem brain samples from 16 PSP patients and 16 normal controls (NCs). Significant differences were validated using sensitive electrochemiluminescence technology. To depict the underlying processes associated with cytokine aberrancies, we performed quantitative gene expression analyses of up-and down-stream targets as well as of signaling effector molecules. Lastly, using flow cytometry we screened for peripheral changes in CD4 + and CD8 + T cell, and NK cell numbers in PSP patients.

IL-2 protein is expressed by NeuN + and GFAP + brain cells.
To investigate if IL-2 is produced locally by brain cells, we performed double immunofluorescence labelling using specific antibodies against IL-2, neuronal nuclei (NeuN), and glial fibrillary acidic protein (GFAP) on brain sections from both PSP patients and NCs. We observed co-localization of IL-2 with both NeuN and GFAP in both groups (Fig. 1C,D) confirming that IL-2 is produced locally in the brain.
The number of peripheral T and NK cells are altered in blood of PSP patients. Lastly, using flow cytometry on a new cohort (18 NCs, nine PSP patients) we investigated whether the elevated levels of IL-2 in PSP brains could be reflected in the composition of peripheral T lymphocytes. We observed a significant increase in numbers of CD4 + T cells in PSP patients (t = 3.812, p < 0.001; Fig. 2B) accompanied by a shift in the ratio between memory and naive CD4 + T cells (t = 3.915, p < 0.001, and t = 7.940, p < 0.001, respectively; Fig. 2C,D). CD8 + T cells numbers did not differ between the groups (t = 1.602, p = 0.122; Fig. 2E), however, we also observed a shift in the ratio between memory and naive CD8 + T cells (t = 7.667, p < 0.001, and t = 8.595, p < 0.001, respectively; Fig. 2F,G) in PSP patients. We did not see any differences in the levels of total T cells (t = 0.318, p = 0.754; Fig. 2A). Since IL-2 protein not only affects classical T cells but also interacts with different types of natural killer (NK) cells, we included NK cell markers in our setup. We saw no differences in the numbers of total NK cells (t = 1.071, p = 0.295; Fig. 2H) between the groups, however, CD56 + NK cell numbers were increased in PSP patients (t = 7.007, p < 0.001; Fig. 2I). We found a tendency towards increased CD57 + NK cells (t = 1.956, p = 0.062; Fig. 2K) whereas there was no difference in CD56 + CD57 + NK cells (t = 0.699, p = 0.491; Fig. 2J).

Discussion
To our knowledge, this is the first study extensively evaluating cytokine protein levels in brains affected by PSP. We screened for 18 cytokines and three growth factors and identified increased IL-2 protein levels in the prefrontal cortex of the brain in PSP patients. Generally, the effects of IL-2 are maturation and survival of T cells of the peripheral immune system 8 . However, the expression of IL-2 and IL-2 receptors is found in brain cells, and we have previously shown how NeuN + cells express IL-2 protein in both healthy brains, and brains from patients diagnosed with Parkinson's disease or multiple system atrophy 9 . Concordantly, in this study we observed co-localization of IL-2 inside both NeuN + and GFAP + cells. IL-2 protein has been suggested to be a neurokine 10 . Hence, the increase in IL-2 protein levels could be of neuronal origin and caused by disease processes. In the current study, for IL-2 protein levels, we observed both sex-specific differences and correlations to clinical scores. To our knowledge, no study has identified sex-specific differences in PSP, nor does the incidence rate differ between sexes 11 . The correlations to the clinical scores were significant for the Luminex measurements, however, this was not replicated for the MSD measurements. This is probably due to the nature of both methods and needs further validation. These observations might be relevant for evaluation of the clinical aspects of changes in brain IL-2 levels. Nevertheless, the current study was not designed to address such issues as the sample size is a limiting factor. Due to our modest sample size included in this experimental setup, additional studies using larger cohorts are needed to validate these findings.
One regulator of IL-2 and T cell proliferation is GSK3B protein 12 . We identified increased mRNA levels of GSK3B in PSP brains. In the brain, GSK3B protein participates in the production of pro-inflammatory mediators secreted by microglia cells 13 . GSK3B has previously received considerable attention in relation to PSP as GSK3B protein activity mediates aggregation of tau 14 . However, interventional strategies targeting GSK3B have been unsuccessful 15 . Although mRNA and protein levels are not directly correlated, our findings are in addition to the previous observations of a central role of GSK3B in the brains of PSP patients.
For the first time, we report on lymphocyte populations in blood of PSP patients. IL-2 protein is produced by CD4 + and CD8 + T cells as well as CD56 + and CD57 + NK cells either as developing or mature cells 8,16,17 . Overall, the observed shift in the pool of T cells toward an increase in CD45RA − CD45RO + memory T cells may be indicative of an active, adaptive immune response. Further, the concurrent decrease in CD45RA + CD45RO − naive T cells, accompanied by increased numbers of CD4 + , but not CD8 + , T cells, could be indicative of a possible reduced capacity of the adaptive immune system toward novel stimuli 18 . Lastly, the observed increase in CD56 + NK cells, but not of activated CD56 + CD57 + NK cells, points toward an increased capacity to respond to innate immune challenges. These results are supported by the results from Santiago and Potashkin 19 reporting on affected gene clusters in PSP that are overrepresented in biological pathways of both leukocyte and lymphocyte activation. Nevertheless, limitations to our study are the low number of individuals as well as the lack of demographic and clinical information for factors affecting peripheral lymphocyte levels. Therefore, we encourage that our observations should be validated in a separate study using a larger and well-defined cohort.
Lately, IL-2 has received much attention as an immune effector in AD since animal studies have shown a beneficial effect of IL-2 on amyloid pathology 20,21 . We could therefore speculate that the increased levels observed in the PSP brains are reflecting compensatory effects to the disease progression rather than necessarily being the cause of the detrimental effects. Therefore, IL-2 treatment aiming at specifically expanding and activating regulatory T cells recently proposed in AD 22,23 could also be an interesting venue to explore in relation to PSP, which at present is a disease without curative treatment.
Other studies have reported on cytokine gene expression in several brain areas of PSP patients with differing results. One study observed increased expression of IL-1beta(β) in the substantia nigra and no change in the expression of transforming growth factor β in the frontal cortex of PSP patients 7 , whereas another study identified increased expression of the latter in the frontal cortex of PSP patients compared to controls 6 . In the current study, we did not assess protein levels of transforming growth factor β, nor did we investigate cytokine expression in the Scientific RepoRts | (2019) 9:7781 | https://doi.org/10.1038/s41598-019-44234-y www.nature.com/scientificreports www.nature.com/scientificreports/ substantia nigra. Here, we focused on the prefrontal cortex which is an area that is only mildly affected in PSP 4 . Our aim was to identify disease pathology distant to the epicenter of the disease that may reflect earlier stages of degenerative processes. This is a limiting factor to our study, and we can therefore not exclude that cytokine expression may be different in other more affected brain areas in PSP patients.
To conclude, in the present study we have identified and validated an increase in IL-2 protein levels in the prefrontal cortex of PSP patients compared with NCs. PSP thereby presents a similar cytokine profile as the www.nature.com/scientificreports www.nature.com/scientificreports/ one identified for both Parkinson's disease and multiple system atrophy in the same brain area. Further, we have observed changes in the peripheral immune system. Our results are indicative of an active and dysregulated adaptive and innate immune response. These data warrant future investigations into the possible role of the immune systems in PSP pathophysiology.

Materials and Methods
Patient material. The tissue for the present study originated from donated human brains. All donors provided written informed consent prior to death. PSP samples were analyzed alongside NC samples 9 . Brain samples were provided by the Bispebjerg Brain Bank (University Hospital of Copenhagen, Bispebjerg Hospital, DNK), the Netherlands Brain Bank (Netherlands Institute for Neuroscience, NLD), and the Harvard Brain Tissue Resource Center (Harvard Medical School Teaching Hospital, USA). A summary of demographic data for the 16 NC and 16 PSP brain samples are shown in Table 3. All brains were neuropathologically examined to verify the clinical diagnosis 1, 24 . Medical records were available for 13 of the 16 patients. The clinical characteristics are shown in Table 1. Clinical scorings were based on the clinicians' notes as well as appropriate, clinical tests, e.g., the Mini-Mental State Examination, the Montreal Cognitive Assessment, or the Addenbrooke's Cognitive Examination tests for dementia and cognitive dysfunction. All brain samples have been collected and handled in accordance with Danish ethical standards of the Brain Bank and the Danish Health and Medicine Authorities. Samples were stored at −80 °C prior to handling. Flow cytometry was performed on a new cohort of samples, that included 18 NC and nine PSP samples, a summary of the demographic data are shown in Table 3. These samples were generously donated by Bispebjerg Movement Disorder Biobank. This project was approved by the Regional Ethics Protein analyses. Protein extraction and quantification of total levels were performed as earlier described 9 .
Validation of IL-2 and G-CSF protein levels were investigated on MSD V-PLEX Human IL-2 Assays (Meso Scale Discovery (MSD), USA; #K151QQD; lower limit of detection (LLOD): 0.09 pg/ml) and MSD U-PLEX G-CSF Assays (MSD; #K151VGK; LLOD: 1.6 pg/ml), respectively, following the manufacturer's instructions. Plates were analyzed on a SECTOR S 600 instrument (MSD). mRNA analyses. Extraction, quality assessment of total RNA, DNA contamination, and analysis of mRNA levels was performed as previously described 9 . All RNA experiments were performed in accordance with the MIQE guidelines 27 . Extraction of RNA was performed using the miRNeasy Mini Kit (Qiagen, NLD; #217004). In short, app. 30 mg brain tissue was homogenized by intensive pipetting in a lysis reagent. Samples were then phase separated in chloroform, and subjected to on-column DNase treatment. Samples were eluted in 30 µl  www.nature.com/scientificreports www.nature.com/scientificreports/ analyses of clinical characteristics and IL-2 protein levels. Student's t-test or Welch's t-test was used for flow cytometry analyses. The remaining statistical analyses were performed in R v. 3.4.1 35 . Normality was assessed using the Shapiro-Wilk Normality test. Scedasticity and collinearity were assessed using ncvTest and vif, respectively, from the car package 36 . If necessary, data were log10-transformed. Data were analyzed using multiple linear regression with models that included relevant variables that have previously been shown to possibly impact results 9 . These variables included group (NC vs. PSP), age, sex (female vs. male), and PMI as well as RIN for the mRNA targets. Familywise error rate was adjusted using Bonferroni correction for the multiplex data (adj. p = 4.17*10 −3 , n = 12). Graphs were made in GraphPad Prism v. 7.02.

Data Availability
The datasets generated and analysed during the current study are available from the corresponding author on reasonable request.