ANALYSIS of EARLY STROKE-INDUCED CHANGES in CIRCULATING LEUKOCYTE COUNTS USING TRANSCRIPTOMIC Grant C

Research Article • DOI: 10.1515/tnsci-2018-0024 • Translational Neuroscience • 9 • 2018 • 161-166

Translational Neuroscience

ANALYSIS OF EARLY STROKE-INDUCED CHANGES IN CIRCULATING LEUKOCYTE COUNTS USING TRANSCRIPTOMIC Grant C. O’Connell*, DECONVOLUTION Julia H.C. Chang

Abstract School of Nursing, Growing evidence suggests that stroke alters the phenotype of the peripheral immune system; better Case Western Reserve University, Cleveland, OH characterization of this response could provide new insights into stroke pathophysiology. In this investigation, we employed a deconvolution approach to informatically infer the cellular composition of the circulating leukocyte pool at multiple timepoints following stroke onset based on whole blood mRNA expression. Microarray data generated from the peripheral blood of 23 cardiovascular disease controls and 23 ischemic stroke patients at 3, 5, and 24 hours post-symptom onset were obtained from a public repository. Transcriptomic deconvolution was used to estimate the relative counts of nine leukocyte populations based on the expression of cell-specific transcripts, and cell counts were compared between groups across timepoints. Inferred counts of lymphoid cell populations including B-cells, CD4+ T-cells, CD8+ T-cells, γδ T-cells, and NK-cells were significantly lower in stroke samples relative to control samples. With respect to myeloid cell populations, inferred counts of neutrophils and monocytes were significantly higher in stroke samples compared to control samples, however inferred counts of eosinophils and dendritic cells were significantly lower. These collective differences were most dramatic in samples collected at 5 and 24 hours post-symptom onset. Findings were subsequently confirmed in a second dataset generated from an independent population of 24 controls and 39 ischemic stroke patients. Collectively, these results offer a comprehensive picture of the early stroke-induced changes to the complexion of the circulating leukocyte pool, and provide some of the first evidence that stroke triggers an acute decrease in eosinophil counts. Keywords Received 09 August 2018 Complete blood count • CBC • NLR • Neutrophil lymphocyte ratio • Immune suppression accepted 14 October 2018 • WBC • White blood cell • WBC Differential • Eosinophil • Infection

Introduction routine clinical evaluation; unfortunately, the [2–6]. Recent work by our group suggests clinical white blood cell differential provides that similar to other conditions [7], several It is becoming increasingly evident that the quantification of a limited number of cell of the gene expression changes observed peripheral immune system responds robustly populations, often only total neutrophils, between stroke patients and controls in to stroke, and that this response influences monocytes, and lymphocytes, and thus these investigations were likely artifacts of clinical outcome. For example, peripheral provides a relatively low-detail picture underlying changes in leukocyte counts, and immune changes triggered by stroke are regarding peripheral immune status. Multi- not true changes in transcription at the cellular believed to contribute to the pathogenesis color flow cytometry experiments have been level [8,9]. Transcriptomic deconvolution is a of adverse complications such as secondary used to examine more discrete subpopulations process which leverages such phenomena to tissue damage, hemorrhagic transformation, of leukocytes, however they have often only informatically infer the cellular composition and post-stroke infection [1]. Thus, better focused on small numbers of cell types in a single of complex biological samples based on characterization of the stroke-induced analysis. Thus, more detailed characterization aggregate gene expression through the peripheral immune response could provide of the stroke-induced changes to the cellular analysis of cell-specific transcripts [10]. In this novel insights into stroke pathophysiology and complexion of the peripheral immune system study, in an attempt to better characterize the open new avenues for immunotherapeutic could reveal nuanced alterations which are stroke-induced peripheral immune response, intervention. pathologically relevant. we employed a deconvolution approach to Many studies which have investigated Several prior studies have performed infer the counts of nine major circulating the peripheral immune response to stroke genome-wide transcriptomic profiling of leukocyte populations at multiple timepoints in humans have done so using the standard peripheral whole blood with the goal of following stroke onset using publicly available white blood cell differential collected as part of identifying clinically-useful stroke biomarkers human whole blood gene expression data.

161 Translational Neuroscience

Methods null hypothesis was rejected when p<0.05. immune response to stroke could provide novel insights into stroke pathophysiology and Microarray data processing Validation of results identify new targets for immunotherapeutic Raw microarray data generated from the Findings were subsequently confirmed via intervention. In this study, we employed a peripheral whole blood of 23 controls, as well as deconvolution of a second publically available transcriptomic deconvolution approach to infer 23 ischemic stroke patients at 3, 5, and 24 hours microarray dataset generated from an the relative counts of nine major circulating post-symptom onset, were downloaded as .CEL independent population of 24 controls and leukocyte populations in blood samples files from the National Center for Biotechnology 39 ischemic stroke patients (GEO accession collected at multiple timepoints over the first Information (NCBI) Gene Expression Omnibus number GSE16561) [14]. 24 hours following stroke onset. With respect (GEO) via accession number GSE58294. to several cell types, our results confirmed the Probe annotations were updated via the Results findings of prior cytometric studies, however, ‘annotate’ package for R (R project for our analysis also revealed changes in other Statistical Computing). Raw perfect match Clinical and demographic leukocyte populations that have yet to be probe intensities were background corrected, characteristics widely reported on in human stroke. quantile normalized, and summarized at the set Stroke samples originated from patients Our observations suggest that the level via robust multi-array averaging using the which were signiﬁcantly older than control circulating counts of lymphoid cell populations rma() function of the ‘affy’ package. Data were counterparts, but were well matched in are ubiquitously suppressed in response to further summarized at the gene level via max terms of sex and ethnicity. In terms of risk stroke. This is consistent with prior cytometric intensity using the collapserows() function of factors for cardiovascular disease, groups investigations which have reported stroke- the ‘WGCNA’ package. were well matched with regards to rates of induced decreases in total lymphocyte counts hypertension and diabetes, however control [15–18], as well as reductions in counts of more Deconvolution subjects displayed a signiﬁcantly higher discrete lymphoid subpopulations such as Estimated counts of B-cells, CD4+ T-cells, CD8+ prevalence of dyslipidemia relative to stroke B-cells, CD4+ T-cells, CD8+ T-cells, and NK-cells T-cells, gamma delta (γδ) T-cells, natural killer patients. All stroke patients had received [15,16,18]. To our knowledge, only three other (NK) cells, monocytes, neutrophils, eosinophils, thrombolytic intervention via recombinant human studies have reported on circulating γδ and dendritic cells were generated from tissue plasminogen activator (rtPA) following 3 T-cell counts in stroke; one reporting that γδ normalized expression data using a list of 226 hour blood collection (Table 1). T-cell counts are elevated [19], one reporting cell-specific genes (Figure 1) aggregated from that they are unaffected [20], and one reporting a compendium of immune cell microarray Inferred leukocyte counts that they are decreased [21]. Our results provide data compiled by Newman et al. [11]. Inferred counts of lymphoid populations additional evidence suggesting that stroke Weighted correlation network analysis was including B-cells, CD4+ T-cells, CD8+ T-cells, triggers an acute decrease in circulating γδ used to produce a relative count for each cell γδ T-cell and NK-cells were all significantly T-cell numbers similar to that which is observed population based on the expression levels of reduced in stroke samples relative to controls in other lymphoid populations. its associated genes using the collapserows() (Figure 2A-E); this effect was most pronounced Our results are also consistent with those of function of the ‘WGCNA’ package according in samples collected at 5 and 24 hours following several prior cytometry-based investigations to the method described by Miller et al. [12]. onset of symptoms. With regards to myeloid reporting that counts of the two most abundant Relative counts of each cell population were populations, inferred counts of neutrophils and peripheral blood myeloid cell populations, arbitrarily scaled from zero to one using unity- monocytes were significantly higher in stroke neutrophils and monocytes, become robustly based normalization. samples in comparison to controls (Figure elevated in response to stroke [15,17,22,23]. 2F-G), however, inferred counts of dendritic However, our findings also suggest that two Demographic information cells and eosinophils were significantly lower lesser-studied myeloid populations, dendritic Clinical and demographic information (Figure 2H-1). Once again, these differences cells and eosinophils, are significantly reduced. associated with samples was aggregated from were most pronounced at 3 and 5 hours post- With respect to dendritic cells, our observations the descriptors reported in Stamnova et al. [13]. onset. An identical overall pattern of changes are consistent with those reported in the was observed in a second microarray dataset limited number of prior cytometric studies Statistics generated from an independent patient performed in humans [24]. With respect to All statistics were performed using R 3.3. population (Supplemental Figure 1). eosinophils, this current study is one of the Fisher’s exact test was used for comparison first case-control analyses to report evidence of dichotomous variables. T-test or one- Discussion of an acute reduction in eosinophil counts in way ANOVA was used for comparisons of response to stroke; our observations align well continuous variables where appropriate. The Better characterization of the peripheral with two recent associative studies reporting

162 Translational Neuroscience

Figure 1.

Gene Entrez ID B-cell CD4+ T-cell CD8+ T-cell γδ T-cell NK-cell Neutropphil Monocyte Dendritic cell Eosinophil Gene Entrez ID B-cell CD4+ T-cell CD8+ T-cell γδ T-cell NK-cell Neutropphil Monocyte Dendritic cell Eosinophil Gene Entrez ID B-cell CD4+ T-cell CD8+ T-cell γδ T-cell NK-cell Neutropphil Monocyte Dendritic cell Eosinophil ABCB4 5244 DSC1 1823 MMP25 64386 ACP5 54 EAF2 55840 MMP9 4318 ADAM28 10863 EBI3 10148 MNDA 4332 AIF1 199 ELANE 1991 MS4A1 931 ALOX15 246 EMR1 2015 MS4A6A 64231 APOL6 80830 EPHA1 2041 MSC 9242 AQP9 366 ETV3 2117 MXD1 4084 ARHGAP22 58504 FAM198B 51313 NAALADL1 10004 ARRB1 408 FASLG 356 NFE2 4778 ASGR1 432 FCGR2B 2213 NKG7 4818 ASGR2 433 FCGR3B 2215 NLRP3 114548 BACH2 60468 FCN1 2219 NOD2 64127 BANK1 55024 FCRL2 79368 NOX3 50508 BCL7A 605 FFAR2 2867 OSM 5008 BLK 640 FLT3LG 2323 P2RX5 5026 BTNL8 79908 FOXP3 50943 P2RY13 53829 C1orf54 79630 FPR1 2357 P2RY14 9934 C5AR1 728 FPR2 2358 P2RY2 5029 CASP5 838 FPR3 2359 PASK 23178 CCL1 6346 GNG7 2788 PDCD1LG2 80380 CCL13 6357 GNLY 10578 PGLYRP1 8993 CCL17 6361 GPR171 29909 PIK3IP1 113791 CCL18 6362 GSTT1 2952 PLA1A 51365 CCL19 6363 GUSBP11 91316 PLA2G7 7941 CCL20 6364 GZMA 3001 PMCH 5367 CCL22 6367 GZMB 3002 PNOC 5368 CCL4 6351 GZMK 3003 PRF1 5551 CCND2 894 GZMM 3004 PTGDR 5729 CCR2 729230 HAL 3034 QPCT 25797 CCR3 1232 HESX1 8820 RALGPS2 55103 CD160 11126 HK3 3101 RASGRP3 25780 CD180 4064 HLA-DOB 3112 RASSF4 83937 CD19 930 HLA-DQA1 3117 RCAN3 11123 CD1A 909 HNMT 3176 REPS2 9185 CD1B 910 HPSE 10855 RGS13 6003 CD1C 911 HSPA6 3310 S100A12 6283 CD1D 912 ICOS 29851 S1PR5 53637 CD1E 913 IDO1 3620 SIK1 150094 CD209 30835 IFNA10 3446 SKA1 220134 CD22 933 IGHD 3495 SLAMF8 56833 CD27 939 IGHM 3507 SMPDL3B 27293 CD28 940 IGKC 3514 SPIB 6689 CD33 945 IGLL3P 91353 ST8SIA1 6489 CD40LG 959 IL12B 3593 STAP1 26228 CD72 971 IL12RB2 3595 STEAP4 79689 CD79A 973 IL26 55801 TCF7 6932 CD79B 974 IL2RA 3559 TCL1A 8115 CD80 941 IL3 3562 TLR2 7097 CD8A 925 IL7 3574 TLR7 51284 CD8B 926 IL7R 3575 TMEM255A 55026 CDA 978 IL9 3578 TNFRSF10C 8794 CDH12 1010 IRF8 3394 TNFRSF17 608 CEACAM3 1084 KIAA0226L 80183 TNFRSF4 7293 CEACAM8 1088 KIR2DL1 3802 TNFSF14 8740 CFP 5199 KIR2DL4 3805 TNIP3 79931 CHI3L1 1116 KIR2DS4 3809 TRAC 28755 CHST15 51363 KLRC3 3823 TRAT1 50852 CLC 1178 KLRD1 3824 TRAV12-2 28673 CLEC10A 10462 KLRF1 51348 TRDC 28526 CLIC2 1193 KYNU 8942 TREM1 54210 CR2 1380 LAMP3 27074 TREM2 54209 CREB5 9586 LCK 3932 TREML2 79865 CRTAM 56253 LEF1 51176 TTC38 55020 CSF2 1437 LILRA2 11027 TXK 7294 CSF3R 1441 LILRA3 11026 TYR 7299 CTLA4 1493 LILRB2 10288 UBASH3A 53347 CTSW 1521 LIME1 54923 UPK3A 7380 CXCL10 3627 LTA 4049 VNN1 8876 CXCL11 6373 MAK 4117 VNN2 8875 CXCL13 10563 MAP3K13 9175 VNN3 55350 CXCR1 3577 MAP9 79884 VPREB3 29802 CXCR2 3579 MARCO 8685 ZFP36L2 678 CXCR5 643 MEFV 4210 ZNF165 7718 DCSTAMP 81501 MGAM 8972 ZNF442 79973 DHRS11 79154 MICAL3 57553 DHX58 79132 MMP12 4321 Max expression

Figure 1. Cell-specific genes used for deconvolution. 226 cell-specific genes used for the deconvolution of whole blood microarray data along with their predominant leukocyte population of expression.

163 Translational Neuroscience

Table 1. Clinical and demographic characteristics. Cardiovascular Disease (n=23) Ischemic Stroke (n=23) p

aAge mean±SD 57.9 ± 3.3 71.7 ± 7.9 <0.001*

bFemale n(%) 11 (47.8) 11 (47.8) 1.000

bNon-caucasian n(%) 4 (17.4) 8 (34.8) 0.314

bDyslipidemia n(%) 16 (69.6) 6 (26.1) 0.007*

bHypertension n(%) 16 (69.6) 16 (69.6) 1.000

bDiabetes n(%) 5 (21.7) 4 (17.4) 1.000

aBaseline NIHSS mean±SD 0.0 ± 0.0 15.4 ± 7.4 <0.001*

brtPA n(%) 0 (0.0) 23 (100.0) <0.001* aCompared via two-sample two-tailed t-test; bCompared via Fisher’s exact test; *Statistically significant

Figure 2.

A B C D E

1.0 1.0 1.0 1.0 1.0

0.9 0.9 0.9 0.9 0.9

0.8 0.8 0.8 0.8 0.8

0.7 0.7 0.7 0.7 0.7

0.6 0.6 0.6 0.6 0.6

0.5 0.5 0.5 0.5 0.5

0.4 0.4 0.4 0.4 0.4

0.3 0.3 0.3 0.3 0.3

0.2 0.2 0.2 0.2 0.2

0.1 0.1 0.1 0.1 0.1 Relative NK-cell count (AU) Relative B-cell count (AU) 0.0 count (AU) Relative CD4+ T-cell 0.0 count (AU) Relative CD8+ T-cell 0.0 count (AU) Relative γδ T-cell 0.0 0.0 Control Control Control Control Control Stroke (3h) Stroke (5h) Stroke (3h) Stroke (5h) Stroke (3h) Stroke (5h) Stroke (3h) Stroke (5h) Stroke (3h) Stroke (5h) LYMPHOID Stroke (24h) Stroke (24h) Stroke (24h) Stroke (24h) Stroke (24h)

F G H I

1.0 1.0 1.0 1.0

0.9 0.9 0.9 0.9

0.8 0.8 0.8 0.8

0.7 0.7 0.7 0.7

0.6 0.6 0.6 0.6

0.5 0.5 0.5 0.5

0.4 0.4 0.4 0.4

0.3 0.3 0.3 0.3

0.2 0.2 0.2 0.2

0.1 0.1 0.1 0.1 Relative eosinophil count (AU) Relative neutrophil count (AU) 0.0 Relative monocyte count (AU) 0.0 Relative dendritic cell count (AU) 0.0 0.0

Mean Control Control Control 95% confidence interval Control Stroke (3h) Stroke (5h) Stroke (3h) Stroke (5h) Stroke (3h) Stroke (5h) Stroke (3h) Stroke (5h) MYELOID Stroke (24h) Stroke (24h) Stroke (24h) Significant vs. control Stroke (24h)

Figure 2. Inferred counts of circulating leukocyte populations. (A-I) Estimated relative counts of B-cells, CD4+ T-cells, CD8+ T-cells, γδ T-cells, NK-cells, monocytes, neutrophils, eosinophils, and dendritic cells in blood sampled from controls and stroke patients at 3, 5, and 24 hours post symptom onset. Counts were statistically compared between stroke and control samples across time points using one-way ANOVA with subsequent planned comparisons via Bonferoni-corrected two-sample two-tailed t-test.

164 Translational Neuroscience

Table 1. Clinical and demographic characteristics. that post-stroke circulating eosinophil counts and the magnitude of transcriptional changes Acknowledgements Cardiovascular Disease (n=23) Ischemic Stroke (n=23) p are negatively correlated with stroke severity observed in stroke at the level of whole blood a Age mean±SD 57.9 ± 3.3 71.7 ± 7.9 <0.001* and risk of mortality [25,26], suggesting are of a relatively small [2,4,6]. Furthermore, The authors would like to thank the members bFemale n(%) 11 (47.8) 11 (47.8) 1.000 that the pathophysiological relevance of the fact that our findings align well with those of FBP School of Nursing at Case Western bNon-caucasian n(%) 4 (17.4) 8 (34.8) 0.314 a stroke-induced reduction in peripheral of previous cytometry-based investigations Reserve University for providing general blood eosinophil counts warrants further suggests that our deconvolution analysis was research support and critical review of the bDyslipidemia n(%) 16 (69.6) 6 (26.1) 0.007* investigation. well implemented. Nonetheless, our findings, manuscript. bHypertension n(%) 16 (69.6) 16 (69.6) 1.000 It should be stated that this study is not in particular those regarding stroke-induced bDiabetes n(%) 5 (21.7) 4 (17.4) 1.000 without limitations; most notable in this regard changes in eosinophil counts, should be Funding is the fact that we did not directly measure confirmed in future work using direct cytometric aBaseline NIHSS mean±SD 0.0 ± 0.0 15.4 ± 7.4 <0.001* leukocyte counts, and instead informatically analysis. Work was funded via Case Western Reserve brtPA n(%) 0 (0.0) 23 (100.0) <0.001* inferred them using gene expression data. While It is also important to note that because University FPB School of nursing start-up aCompared via two-sample two-tailed t-test; bCompared via Fisher’s exact test; *Statistically significant transcriptional deconvolution approaches have samples originated from patients who received funds issued to GCO and a National Institute of been shown to accurately enumerate cell counts thrombolytic treatment following the initial Nursing Research T32 predoctoral fellowship in numerous benchmarking studies [10–12], 3 hour blood collection, it is possible that awarded to JHCC (5T32NR15433-3). the fact that they rely on reference expression some of the differences in inferred leukocyte signatures generated from isolated healthy cells counts observed between stroke patients Conflict of interest has the potential to introduce confounds. The and controls at the 5 and 24 hour timepoints handling and manipulation of reference cells were driven by effects of rtPA. However, we GCO has a patent pending re: genomic patterns during isolation could potentially alter gene find this scenario unlikely due to the fact we of expression for stroke diagnosis. GCO has expression and introduce artifacts. Furthermore, observed a similar pattern of changes in a received consulting fees from Valtari Bio because reference signatures are generated second dataset originating from blood samples incorporated. The remaining authors report no from the cells of healthy donors, disease-specific which were collected prior to administration of potential conflicts of interest. differential regulation of reference signature thrombolytics. genes could reduce the accuracy of analyses. Collectively, our results offer a comprehensive Compliance with ethical standards However, the aforementioned limitations are picture of the early stroke-induced changes to most likely to introduce confounds when trying the complexion of the circulating leukocyte All procedures performed in studies involving to discriminate between highly similar cell pool. Our findings confirm the results of prior human participants were in accordance with populations or in disease states which introduce cytometric investigations, and additionally, the ethical standards of the institutional and/or dramatic alterations in transcription; the cell provide some of the first human evidence that national research committee and with the 1964 types which were enumerated in our analysis stroke triggers an acute decrease in circulating Helsinki declaration and its later amendments are relatively distinct at the molecular level, eosinophil counts. or comparable ethical standards.

References

[1] Kamel H, Iadecola C. Brain-immune interactions and ischemic stroke: [6] Tang Y, Xu H, Du X, Lit L, Walker W, Lu A, et al. Gene expression in Clinical implications. JAMA Neurol. 2012;69:576–81. blood changes rapidly in neutrophils and monocytes after ischemic [2] O’Connell GC, Petrone AB, Treadway MB, Tennant CS, Lucke-Wold N, stroke in humans: a microarray study. J Cereb Blood Flow Metab Off J Chantler PD, et al. Machine-learning approach identifies a pattern Int Soc Cereb Blood Flow Metab. 2006;26:1089–102. of gene expression in peripheral blood that can accurately detect [7] Whitney AR, Diehn M, Popper SJ, Alizadeh AA, Boldrick JC, Relman ischaemic stroke. Npj Genomic Med. 2016;1:16038–16038. DA, et al. Individuality and variation in gene expression patterns in [3] O’Connell GC, Treadway MB, Petrone AB, Tennant CS, Lucke-Wold human blood. Proc Natl Acad Sci U A. 2003;100:1896–901. N, Chantler PD, et al. Peripheral blood AKAP7 expression as an early [8] O’Connell GC, Treadway MB, Tennant CS, Lucke-Wold N, Chantler PD, marker for lymphocyte-mediated post-stroke blood brain barrier Barr TL. Shifts in Leukocyte Counts Drive the Differential Expression disruption. Sci Rep. 2017;7:1172–1172. of Transcriptional Stroke Biomarkers in Whole Blood. Transl Stroke [4] Stamova B, Xu H, Jickling G, Bushnell C, Tian Y, Ander BP, et al. Gene Res [Internet]. 2018 [cited 2018 Jun 14]; Available from: http://link. expression profiling of blood for the prediction of ischemic stroke. springer.com/10.1007/s12975-018-0623-1 Stroke J Cereb Circ. 2010;41:2171–7. [9] O’Connell GC, Treadway MB, Petrone AB, Tennant CS, Lucke-Wold N, [5] Jickling GC, Ander BP, Stamova B, Zhan X, Liu D, Rothstein L, et al. RNA Chantler PD, et al. Leukocyte Dynamics Influence Reference Gene in blood is altered prior to hemorrhagic transformation in ischemic Stability in Whole Blood: Data-Driven qRT-PCR Normalization Is a stroke. Ann Neurol. 2013;74:232–40. Robust Alternative for Measurement of Transcriptional Biomarkers.

165 Translational Neuroscience

Lab Med [Internet]. 2017;48. Available from: http://fdslive.oup.com/ Suppresses Immunity in the Periphery and Brain via Different www.oup.com/pdf/production_in_progress.pdf Neurogenic Innervations. Immunity. 2017;46:474–87. [10] Abbas AR, Wolslegel K, Seshasayee D, Modrusan Z, Clark HF. [19] He Y, Ma N, Xing C, Wang X, Xiao H, Zheng M, et al. Novel IL-6- Deconvolution of Blood Microarray Data Identifies Cellular Activation secreting γδT cells increased in patients with atherosclerotic cerebral Patterns in Systemic Lupus Erythematosus. Tan P, editor. PLoS ONE. infarction. Mol Med Rep. 2015;11:1497–503. 2009;4:e6098. [20] Peterfalvi A, Molnar T, Banati M, Pusch G, Miko E, Bogar L, et al. [11] Newman AM, Liu CL, Green MR, Gentles AJ, Feng W, Xu Y, et al. Robust Impaired function of innate T lymphocytes and NK cells in the acute enumeration of cell subsets from tissue expression profiles. Nat phase of ischemic stroke. Cerebrovasc Dis Basel Switz. 2009;28:490– Methods. 2015;12:453–7. 8. [12] Miller JA, Cai C, Langfelder P, Geschwind DH, Kurian SM, Salomon [21] Adamski MG, Li Y, Wagner E, Yu H, Seales-Bailey C, Durkin H, et al. DR, et al. Strategies for aggregating gene expression data: The Pre-Existing Hypertension Dominates γδT Cell Reduction in Human collapseRows R function. BMC Bioinformatics. 2011;12:322. Ischemic Stroke. Lees JR, editor. PLoS ONE. 2014;9:e97755. [13] Stamova B, Jickling GC, Ander BP, Zhan X, Liu DZ, Turner R, et al. [22] Hug A, Dalpke A, Wieczorek N, Giese T, Lorenz A, Auffarth G, et al. Gene expression in peripheral immune cells following cardioembolic Infarct Volume is a Major Determiner of Post-Stroke Immune Cell stroke is sexually dimorphic. PLoS ONE. 2014;9:1–9. Function and Susceptibility to Infection. Stroke. 2009;40:3226–32. [14] Barr TL, Conley Y, Ding J, Dillman a, Warach S, Singleton a, et al. [23] Urra X, Cervera Á, Obach V, Climent N, Planas AM, Chamorro A. Genomic biomarkers and cellular pathways of ischemic stroke by Monocytes are major players in the prognosis and risk of infection RNA gene expression profiling. Neurology. 2010;75:1009–14. after acute stroke. Stroke. 2009;40:1262–8. [15] Vogelgesang A, Grunwald U, Langner S, Jack R, Bröker BM, Kessler C, [24] Yilmaz A, Fuchs T, Dietel B, Altendorf R, Cicha I, Stumpf C, et al. et al. Analysis of lymphocyte subsets in patients with stroke and their Transient decrease in circulating dendritic cell precursors after acute influence on infection after stroke. Stroke. 2008;39:237–41. stroke: potential recruitment into the brain. Clin Sci. 2009;118:147– [16] Urra X, Cervera Á, Villamor N, Planas a. M, Chamorro Á. Harms and 57. benefits of lymphocyte subpopulations in patients with acute stroke. [25] Wang J, Ma L, Lin T, Li S-J, Chen L-L, Wang D-Z. The significance of Neuroscience. 2009;158:1174–83. eosinophils in predicting the severity of acute ischemic stroke. [17] Haeusler KG, Schmidt WUH, Föhring F, Meisel C, Helms T, Oncotarget [Internet]. 2017 [cited 2018 Jun 14];8. Available from: Jungehulsing GJ, et al. Cellular Immunodepression Preceding http://www.oncotarget.com/fulltext/22199 Infectious Complications after Acute Ischemic Stroke in Humans. [26] Guo L-B, Liu S, Zhang F, Mao G-S, Sun L-Z, Liu Y. The role of eosinophils Cerebrovasc Dis. 2008;25:50–8. in stroke: a pilot study. Eur Rev Med Pharmacol Sci. 2015;19:3643–8. [18] Liu Q, Jin W-N, Liu Y, Shi K, Sun H, Zhang F, et al. Brain Ischemia

166