Pre-Existing Hypertension Dominates γδT Cell Reduction in Human Ischemic Stroke

T lymphocytes may play an important role in the evolution of ischemic stroke. Depletion of γδT cells has been found to abrogate ischemia reperfusion injury in murine stroke. However, the role of γδT cells in human ischemic stroke is unknown. We aimed to determine γδT cell counts and γδT cell interleukin 17A (IL-17A) production in the clinical setting of ischemic stroke. We also aimed to determine the associations of γδT cell counts with ischemic lesion volume, measures of clinical severity and with major stroke risk factors. Peripheral blood samples from 43 acute ischemic stroke patients and 26 control subjects matched on race and gender were used for flow cytometry and complete blood count analyses. Subsequently, cytokine levels and gene expression were measured in γδT cells. The number of circulating γδT cells was decreased by almost 50% (p = 0.005) in the stroke patients. γδT cell counts did not correlate with lesion volume on magnetic resonance diffusion-weighted imaging or with clinical severity in the stroke patients, but γδT cells showed elevated levels of IL-17A (p = 0.048). Decreased γδT cell counts were also associated with older age (p = 0.004), pre-existing hypertension (p = 0.0005) and prevalent coronary artery disease (p = 0.03), with pre-existing hypertension being the most significant predictor of γδT cell counts in a multivariable analysis. γδT cells in human ischemic stroke are reduced in number and show elevated levels of IL-17A. A major reduction in γδT lymphocytes also occurs in hypertension and may contribute to the development of hypertension-mediated stroke and vascular disease.


Introduction
Stroke is a leading cause of death and disability in the United States [1]. The immune system is recognized as playing a major role in the evolution and pathophysiology of stroke and microvascular dysfunction, with a growing number of reports on the role of T lymphocytes [2][3][4][5][6][7].
cdT cells are a population of T lymphocytes that comprise between 0 and 7% of circulating CD3 + cells [8]. These cells express cd T cell receptors (TCR), whereas the majority of T lymphocytes express abTCR. cdT cells secrete interleukin-17 (IL-17) when stimulated by interleukin-23 (IL-23) and interferon gamma (INF-c) when stimulated by interleukin-12 (IL-12) [9,10]. As a bridge cell population cdT cells break the classical immune system paradigm by having features of both the adaptive (e.g., the antigen experienced response) and innate (i.e., the rapid response) immune responses. cdT cells that secrete IL-17 have been characterized as ligand-naive and cdT cells that secrete INF-c have been characterized as ligand-experienced [9].
In a prior study depletion of cdT cells in mice ameliorated ischemia-reperfusion injury in the brain [11]. When analyzing the mechanistic basis of cdT cells in this model, Shichita et al. [11] identified important roles and interactions of IL-23 and IL-17 but not INF-c. After ischemia reperfusion injury, increased macrophage IL-23 secretion commenced after day 1 and increased cdT cell IL-17 secretion commenced after day 3. In this same model, infarct size was reduced on day 1 in IL-23 knockout (KO) mice and on day 4 in IL-17 KO mice but was not altered in INF-c KO mice. Intracellular cytokine staining confirmed that cdT cells were the main source of IL-17, CD4 + and cdTCR-T cells were the main sources of INF-c.
To date there is little data on the role of cdT cells in human stroke [12]. The mice in Shichita et al's study [11] were young and apparently free of vascular risk factors, while older age, a major risk factor for stroke, has been associated with reduced cdT cell counts [13,14]. Furthermore, hypertension and coronary artery disease, other major stroke risk factors, have been associated with alterations in IL-17 and IFN-c [15][16][17][18]. Therefore, our objectives were to determine cdT cell counts and function in the clinical setting of ischemic stroke and the associations of cdT cell counts with lesion volume, measures of clinical severity and with major stroke risk factors. We have investigated whether circulating cdT cell counts are 1) altered in patients presenting with acute ischemic human stroke, and 2) correlate with: lesion volume on diffusionweighted magnetic resonance imaging (DW MRI), and with clinical severity and demographic and risk factors. Next we examined cdT cell levels of IL-17A and INF-c and the cellular gene expression of interleukin-23 receptor (IL-23R) and interleukin-23 subunit alpha (IL-23A).

Methods
Approval was obtained from the SUNY Downstate Medical Center Institutional Review Board for this prospective observational study. All patients and/or authorized representatives gave full and signed informed consent.

Study Subjects
The peripheral blood samples of 69 subjects were used: 43 acute stroke patients admitted to the University Hospital of Brooklyn at SUNY Downstate Medical Center and at Long Island College Hospital, and 26 gender and race matched healthy control subjects recruited from the local community. The clinical and laboratory characteristics of patients and controls are shown in Table 1. The age range of the subjects was 30-95 years. The median time of blood draw was 36629 hours post stroke. Stroke was diagnosed according to World Health Organization stroke criteria.
The study inclusion criteria were: over 18 years of age and acute ischemic stroke. The exclusion criteria were: current immunological diseases, taking steroid or immunosuppressive therapies, severe allergies, acute infection and severe anemia. The following clinical data were recorded: age, gender, race, self-reported risk factors, medications at the time of stroke onset (based on 40 subjects) or, in the control subjects, medications at the time of blood draw, National Institutes of Health Stroke Scale (NIHSS) score in the stroke subjects and complete blood counts (CBC), including total white blood cell count and white cell differential counts. Hypertension was defined as a prior (at any time in the  past) diagnosis of hypertension by the subject's physician or currently receiving treatment for hypertension. Diabetes was defined as a past medical history of known diabetes mellitus. Coronary artery disease was defined as a physician-diagnosed past history of ischemic heart disease or angina. Hyperlipidemia was defined as a past history of documented elevation in total cholesterol (.200 mg/dl). Smoking was defined as current or prior smoking. Atrial fibrillation was defined as a past or current history of physician-diagnosed atrial fibrillation.
In the stroke patients lesion volumes were independently measured by two observers with neuroimaging expertise who were blinded to the clinical and laboratory data. Ischemic lesion volumes were measured on magnetic resonance imaging diffusion weighted images using Medical Image Processing Analysis and Visualization software. Neurological severity was based on the admission National Institutes of Health Stroke Scale (NIHSS) score. Follow-up assessment used the Barthel scores at 3 to 6 months (based on 33 patients) that provides a measure of the amount of stroke recovery and disability.

Flow Cytometry
Peripheral blood from the study participants was drawn to BD Vacutainer tubes containing ethylenediaminetetraacetate (EDTA). Within an hour from blood collection 100 ml of blood was stained in the dark for 30 minutes in room temperature with four monoclonal antibodies (Ab) to CD3, CD4, CD8 and cdTCR ( Table 2). Following staining, erythrocytes were lysed and leukocytes were fixed according to manufacturer No-Wash Lysis Procedure (Cal-Lyse solution Invitrogen). Counting beads (Ac-cuCheck Counting Beads Invitrogen) were added, following manufacturer protocol, to obtain absolute cell counts (gating strategy presented in the Figure 1). Data were collected on the Epics XL flow cytometer (Beckman Coulter). Three populations of T lymphocytes were analyzed: CD8+, CD4+ and cdTCR+. T lymphocytes were defined as CD3+ cells within the lymphocyte gate. The lymphocyte gate was set based on side scatter characteristics. Gating strategy was set to include CD3+ cells regardless of size, not to exclude activated (larger) and apoptotic (smaller) lymphocytes ( Figure 1). Based on forward scatter (FSC) values, size of CD3+ lymphocytes was not different from the size of cdT cells (mean 6standard deviation, respectively: 182614 and 187616; p.0.05) as presented on FCS vs. cdTCR+ plot ( Figure 1). cdT cells size did not differ between stroke and control (mean 6standard deviation, respectively: 188617 and 184614; p.0.05). 200,000 events were acquired for each sample. Raw data were compensated, analyzed and presented using FlowJo software version 9.2 for Mac (Tree Star, San Carlos, California). Ab were titrated to achieve optimal resolution and gates for CD8+, CD4+ and cdTCR+ T lymphocytes were set using fluorescence minus one (FMO) method [19].

Cytokine Levels and Gene Expression in cdT cells
cdT cells were extracted from 35 study subjects (17 stroke, 18 control). A detailed protocol for cdT cell sorting has been published previously [20]. Briefly, cdT cells were extracted from whole blood using magnetic bead separation (Miltenyi Biotec) -cdT cell purity was over 80%. RNA and proteins were extracted using column separation (Norgen). Quality and concentration of RNA and protein was measured using Qubit 1.0 Fluorometer (Invitrogen). Protein concentration was measured in 28 study subjects (10 stroke, 18 controls). The concentration of IL-17A and INF-c in proteins extracted from cdT cell was determined by flow cytometry using BD cytometric bead array. Samples processing was performed according to the manufacturers instructions (BD Bioscience) using Human IL-17A Flex Set and Human IFN-c Flex Set. Results were analyzed using FCAP Array version 3 software (BD Bioscience). Gene expression was measured in 29 study subjects (17 stroke, 12 control). RNA reverse transcription PCR was based on random hexamers (Life Technologies). The primers were designed using NCBI/Primer-Blast (Primer3 and BLAST), and wet tested. High throughput real time PCR run on BioMark HD system (Fluidigm) was used to measure gene expression of IL-23R and IL-23A genes. To compare gene expression between samples relative analytic approach was used. The relative analysis was based on the cell count input for each sample and was normalized to commercial cDNA (Universal cDNA Reverse Transcribed by Random Hexamer: Human Normal Tissues; Biochain, Newark, CA). Subjects selected for cytokine expression and gene expression were representative for the entire group used for cell phenotyping.
on cdTCR expression). (b) Gating strategy to identify counting beads. First, two populations of counting beads were selected based on the forward and side scatter characteristics. Second, FITC positive beads were counted for the cell count analyses. Third, the frequency of each bead population was assessed to compare with bead lot characteristics. doi:10.1371/journal.pone.0097755.g001 Table 3. Leukocyte subset cellular counts in stroke and control subjects.

Statistical Analyses
The data were analyzed using R version 2.15.2 [21]. Nonparametric and parametric tests, as appropriate, were used to compare demographic and laboratory values between the stroke and control subjects (Wilcoxon rank sum and Student's t, tests).Spearman correlation coefficients were used for correlational analyses. Chi-square and Fishers' exact tests were used to compare grouped data. The hierarchical cluster analysis used Ward's method and log transformed and normalized cell count data.
A factorial analysis of variance (section 4 of the Results) was used to evaluate for potential interactions in the associations of stroke and hypertension on cdT cell subset counts. Levene's tests confirmed homogeneity of the variances of the independent factors. A linear regression analysis was performed to adjust for baseline differences between patients and controls and to control for major potential confounders in the associations of hypertension and stroke with cdT cell counts. Measurements of the global validation statistic, skewness, kurtosis, link function and heteroscedasticity confirmed that all model assumptions were met. The cdT cell count data were log transformed for this analysis. P values ,0.05 were considered statistically significant.

Reduction of cdT cell numbers in human ischemic stroke
The absolute number of circulating cdT cells was reduced in the stroke subjects relative to the controls, by almost 50% (p = 0.005, Table 3). In the stroke patients the median cdT cell count was 79.4 cells/ml while in the control subjects the median number of cdT cells was 137.0 cells/ml. cdT cells accounted for a median of 1.92% (range 0.28-22.9%) of CD3 + lymphocytes. Figure 2. cdT cells were differentially altered in this clinical cohort relative to 5 other leukocyte subsets. This heatmap illustrates the normalized cell counts for 6 leukocyte cell subsets (columns) in the study subjects (rows). Cell numbers for neutrophils, monocytes and total lymphocytes were determined from the complete blood count, and cell numbers for CD4 + , CD8 + and cd + T lymphocytes were measured using flow cytometry. The column dendrogram reveals clustering by leukocyte subset, demonstrating that the cdT cell subset formed a separate and dominant cluster from the other five leukocyte subsets. The cdT cell cluster separates into decreased and increased cellular counts, with decreased cellular counts predominating in the stroke subjects. This hierarchical cluster analysis used Ward's method and log-transformed and normalized cell count data. doi:10.1371/journal.pone.0097755.g002 cdT cells were disproportionately altered relative to the five other leukocyte subsets studied ( Figure 2). Mild increases in neutrophils (p = 0.045) and decreases in the total number of lymphocytes (p = 0.03) were present in the ischemic stroke patients, while CD4 + and CD8 + T cells were non-significantly reduced ( Table 3) and monocyte counts were unaltered (p = 0.99, Table 3).

cdT cell numbers did not correlate with lesion volume or clinical severity
There was no association of cdT cell count with ischemic lesion volume on DW MRI (coefficient = 20.03, p = 0.9). There was also no association of cdT cell number with stroke severity as measured by the NIHSS (coefficient = 2012, p = 0.4) or with outcome (coefficient = 0.02, p = 0.9).  Table 5. T cell numbers in subjects with and without pre-existing hypertension.

Correlation of cdT cell numbers with demographic and risk factors
cdT cell numbers were decreased in individuals over age 60 years (p = 0.004) and in individuals with a prior history of coronary artery disease (p = 0.03). There was no association of cdT cell count with race or gender ( Table 4). There was a substantial decrease in cdT cell counts in individuals with pre-existing hypertension. cdT cell counts were reduced by over 60% in hypertensive subjects (80.6 cells/ml) relative to normotensive subjects (217.0 cells/ml, p = 0.0005, Table 4). cdT cell counts were reduced in patients taking antithrombotics (p = 0.03) or bblockers (p = 0.02) but not altered in patients taking angiotensin converting enzyme (ACE) inhibitors, angiotensin II receptor blockers (ARB), diuretics or calcium channel blockers.
To further explore this finding, the cellular counts of the other T cell subsets were compared between subjects with and without hypertension. The CD8 + subset (p = 0.01) was moderately reduced in individuals with hypertension while there was no significant difference in CD4 + subset numbers in the subjects with hypertension ( Table 5). cdT cell numbers were compared between the hypertensive and normotensive controls where substantial reductions were found-81.8 cells/ml in hypertensive control subjects compared to 238.0 cells/ml in the normotensive control subjects (p = 0.0006, Table 5). This was not explained by age differences, as the median age was not significantly different between the control subjects with and without hypertension (58.0 vs. 52.0 years, respectively). CD8 + T cells were non-significantly decreased in hypertensive control subjects relative to normotensive controls ( Table 5).

Relative contributions of hypertension and stroke on cdT cell counts
In a factorial analysis of variance hypertension was the major determinant of cdT cellular counts (F value = 12.8; p = 0.0006). Stroke and hypertension:stroke interaction were not significant (respectively, F value = 1.54; p = 0.22 and F value = 2.1; p = 0.15). In linear regression analysis, adjusting for age, race and gender, pre-existing hypertension was also the most significant predictor of cdT cell counts (p = 0.008, Table 6). The estimated effect size of hypertension from the model was 2.77, indicating that hypertension is associated with a 64% reduction in cdT cell counts ( Table 6). Coronary artery disease had no effect when entered into the model, nor did smoking history or beta-blocker or antithrombotic therapy.

Cytokine levels and gene expression in cdT cells
In cdT cells there were increased levels of IL-17A in the stroke patients relative to the controls (p = 0.048) with a similar trend in IL-17A/IFN-c ratio (p = 0.1) and with no alterations in IFN-c levels ( Table 7). There was a 3.3 fold non significant increase in IL-23R gene expression in cdT cells in the stroke patients (p.0.05) with a 1.1 fold non significant decrease in IL-23A gene expression (p.0.05). cdT cellular cytokine levels and gene expression were not significantly different between individuals with and without hypertension, with and without CAD and those over age 60 years relative to those less than age 60 years. Linear regression modeling was used to determine the relative impacts of hypertension and stroke on cdT cell counts and to adjust for clinical and demographic factors and potential interactions (for age, race and gender). This revealed that pre-existing hypertension was the significant predictor in the model accounting for cdT cell counts. Coronary artery disease had no effect when entered into the model, nor did smoking history, beta-blocker treatment, or antithrombotic treatment. CIconfidence intervals. doi:10.1371/journal.pone.0097755.t006 Table 7. cdT cellular cytokine levels (n = 28).

Discussion
In this study we found that cdT cell numbers were reduced in human ischemic stroke and that cdT cells showed elevated IL-17A secretion. Reductions in cdT lymphocytes were also associated with hypertension, older age and, to a lesser degree, with prevalent coronary artery disease. Hypertension was associated with cdT cell count reductions of almost 65%. To the best of our knowledge this is the first clinical study showing the potential roles of cdT lymphocytes in stroke, hypertension and hypertension-mediated stroke.
In this study, the increase in IL-17A but not in INF-c levels in the cdT cells of the stroke patients suggests activation of the IL-23 and IL-17 cytokine expression pathway. This is in line with previous reports by Shichita et al. from a murine stroke model and by Li et al. from human ischemic brain tissues [11,22]. This increase in IL-17 expression from antigen-naive cdT cells may reflect a non-specific/antigen naive immune response to stroke [9,10,23]. Contrary to this murine study of Shichita et al. where cdT cell depletion ameliorated ischemia-reperfusion injury [11] we did not find associations of cdT cell counts with lesion volume, stroke severity and outcome. However, associations with lesion size and outcome and severity might be found in a larger patient sample.
Changes in leukocyte numbers, as reflected by elevated neutrophil counts and reductions in lymphocyte counts are common in acute stroke and reflect the phenomenon of adrenergic mediated stroke-induced immunodepression [24]. In stroke, activation of the sympathetic nervous system causes shrinkage of the spleen due to the release of residual cells [25][26][27] and is essential to the release of hepatic invariant NKT cells [6]. Blockade of the sympathetic nervous system has been shown to modulate circulating regulatory T cell numbers [28] and to change the activity of CD8 + and CD4 + T cells [29]. Stimulation of the sympathetic nervous system by acute stress has also been associated with increased numbers of circulating T cells expressing receptors for chemokines secreted by activated endothelial cells [30]. The reductions in cdT cells seen in this study may indicate that this is one of the populations of T lymphocytes that are regulated by the stress induced sympathetic response [6,30]. Although, we did not specifically study the role of sympathetic activation in this study, we note that patients treated with bblockers had significantly decreased numbers of circulating cdT cells.
The reductions of cdT cellular numbers in the stroke patients were disproportionately larger than reductions in the other lymphocyte subsets. Our results show that older age, hypertension and coronary artery disease were also associated with cdT cellular counts, with hypertension dominating cdT cell reductions. Older age has previously been reported to be associated with altered cdT cell counts [13,14]. In CAD, no alterations in cdT cell counts have been reported to date, but decreases in CD3 + lymphocytes and increases of IFN-c secreting CD8 + CD56 + T lymphocytes and of CD28 -CD4 + T lymphocytes -also a significant source of IFN-chave been reported [15][16][17]. Coronary artery infiltrating T lymphocytes producing IL-17 and IFN-c [18] have been described. Given that cdT cells are a major source of both IL-17 and IFN-c, our finding of decreased numbers of cdT cells in CAD is deserving of further study.
Hence, while the stroke patients and control subjects were matched on race and gender the subjects were not matched on age and hypertension status, which were associated with cdT cell counts. Given this finding, and given that most stroke patients are hypertensive, in order to fully and further explore the decrease in the gamma delta T cell counts in stroke patients stroke patients should be matched with controls based on their hypertension score.
Hypertension was overall the strongest determinant of cdT cell counts. Hypertension was the factor that was most strongly associated with cdT cell counts and was the most significant predictor in factorial and linear regression analyses. Markedly reduced cellular counts were also seen in the hypertensive control subjects, relative to the normotensive control subjects.
Hypertension is by far the most significant, modifiable risk factor for both ischemic and hemorrhagic stroke [31][32][33]. The high prevalence of hypertension in our non-Hispanic Black Caribbean stroke population is typical of many stroke cohorts. Hypertension has classically been considered a disorder of the renin angiotensin system or of the brain. Recently in rodent studies it has been demonstrated that T lymphocytes are required for the full development of angiotensin II induced hypertension [34]. T lymphocytes were found to infiltrate the adventitia and periadventitial fat of vessels [34] and T cells secreting IL-17 have been found in the vessel walls of hypertensive animals [34][35][36], potentially due to induction of hypertension-induced expression of homing receptors [35]. Furthermore, plasma and tissue levels of IL-17 and IFN-c have been independently related with the development of hypertension, but the cellular sources were not identified [36]. As cdT cells secrete and are key sources of IL-17 and IFN-c in the blood and IL-17 in the brain [4,8,37], the decrease in the number of circulating cdT cells in our study may be due to their active role in the evolution of hypertension and possible mobilization into vascular walls or their removal from the circulation. As the number of cdT cells distinguished the already diagnosed hypertensive subjects from non-hypertensive subjects, regardless of the medications being taken, as yet unrecognized pathophysiological mechanisms may be involved-other than targeted by current antihypertensive agents [38].
The implications of our results are that we have recognized two T lymphocyte subsets -the cdT and the CD8 subsets -that may be involved in human hypertension [39], with cdT cells possibly linking hypertension to stroke. cdT cells were identified for the first time as a potential marker of stroke risk in hypertensive individuals or of vascular disease burden. Alterations in CD8 + T lymphocytes support a recent finding where subsets of these cells were found to contribute to human hypertension [40]. The importance of these findings are highlighted by the fact that hypertension is the most common vascular disease worldwide with 5-30% of individuals remaining resistant to available treatments [31]. Hypertensionrelated morbidity comes largely from its downstream vascular consequences that include atherosclerotic vascular disease, heart disease and stroke, with stroke being the preeminent complication.
Elucidating as yet unrecognized mechanisms of both hypertension and stroke could give new perspectives to the understanding of these diseases and possibly lead to novel treatments. Furthermore, this study demonstrates the importance of incorporating vascular risk factors in stroke translational research studies.