Neutrophil Transcriptional Profile Changes During Transit from Bone Marrow to Sites of Inflammation

RESEARCH ARTICLE

Neutrophil transcriptional profile changes during transit from bone marrow to sites of inflammation

Flavia S Lakschevitz1,2, Michelle B Visser2,3, Chunxiang Sun2 and Michael Glogauer1,2

It has recently been established that neutrophils, the most abundant leukocytes, are capable of changes in gene expression during inflammatory responses. However, changes in the transcriptome as the neutrophil leaves the bone marrow have yet to be described. We hypothesized that neutrophils are transcriptionally active cells that alter their gene expression profiles as they migrate into the vasculature and then into inflamed tissues. Our goal was to provide an overview of how the neutrophil’s transcriptome changes as they migrate through different compartments using microarray and bio-informatic approaches. Our study demonstrates that neutrophils are highly plastic cells where normal environmental cues result in a site-specific neutrophil transcriptome. We demonstrate that neutrophil genes undergo one of four distinct expression change patterns as they move from bone marrow through the circulation to sites of inflammation: (i) continuously increasing; (ii) continuously decreasing; (iii) a down-up-down; and (iv) an up-down-up pattern. Additionally, we demonstrate that the neutrophil migration signaling network and the balance between anti-apoptotic and pro-apoptotic signaling are two of the main regulatory mechanisms that change as the neutrophil transits through compartments. Cellular & Molecular Immunology (2015) 12, 53–65; doi:10.1038/cmi.2014.37; published online 9 June 2014

Keywords: fMLP signaling pathway; microarray; neutrophil; transcriptome

INTRODUCTION Neutrophils undergo spontaneous apoptosis in order to As the main cellular component of the innate immune system, maintain homeostasis, but also to regulate efficient resolution neutrophils are responsible for responding to bacterial infec- of inflammation.2,10,11 Impaired clearance as well as delays in tions. They are rapidly recruited to sites of infection where they apoptotic events is critical to the progression of chronic inflam- eliminate bacteria through both reactive oxygen dependent and mation.3,12,13 A complex signaling interaction regulates neu- independent mechanisms.1 Their removal from the inflammat- trophil apoptosis and is initiated by two distinct pathways: the ory site alongside elimination of deleterious stimulus is indis- extrinsic pathway (death receptor pathway) and the intrinsic pensable for maintaining host health.1,2 It has been established pathway (mitochondria pathway), where both pathways lead that neutrophils are capable of changes in gene expression dur- to activation of the characteristic caspase cascade4,10 It is very ing inflammatory responses.3 Subrahmanyam demonstrated likely that tight regulation of neutrophil apoptosis at different that while neutrophils synthesize lower levels of protein and locations is a key feature of normal neutrophil function. RNA, they are able to markedly change their gene expression To investigate gene expression profiles in the mouse neutro- following bacterial exposure.4 The concept that neutrophils are phil, as they leave the bone marrow (BM), enter the circulation, static quiescent cells that merely kill bacteria has been replaced and finally arrive in the site of inflammation, we used three by the idea that neutrophils play a more active role in the host approaches: first using microarrays we identified differences response. For example, activated neutrophils are able to secrete in the gene expression profile of neutrophils as they transit a variety of cytokines,5 act as antigen-presenting cells expressing through different compartments. Secondly, using a bioinfor- MHC Class II,6 allow activation of T cells7 and play an active matics approach, we examined whether the differences in gene regulatory role in angiogenesis8 and tumoural fate.9 expression are associated with specific pathways. Finally, we

1Department of Periodontology, Faculty of Dentistry, University of Toronto, Toronto, Ont., Canada; 2Matrix Dynamics Group, Faculty of Dentistry, University of Toronto, Toronto, Ont., Canada and 3Current address: Department of Oral Biology, School of Dental Medicine, University at Buffalo, The State University of New York, Buffalo, NY, USA Correspondence: Dr. M Glogauer, Faculty of Dentistry, University of Toronto, Room 221 Fitzgerald Building, 150 College Street, Toronto, Ont. M5S 3E2, Canada. E-mail: [email protected] Received: 17 October 2013; Revised: 2 May 2014; Accepted: 2 May 2014 Neutrophil gene expression profiles changes FS Lakschevitz et al 54

sought to investigate if these different gene expression profiles gating was performed based on fluorescence-minus one con- result in differences in the neutrophil phenotype. Using a trols (Supplementary Figures 1 and 2). Viability was confirmed combination of transcriptomic, bio5informatic and func- using eFluor 506 (eBioscience, San Diego, CA, USA), which tional approaches, we highlight the importance of altered gene stains necrotic cells, during flow cytometric analysis and/or expression in the apoptosis pathway and N-formyl-Met-Leu- trypan blue exclusion test. Following gradient separation alone, Phe (fMLP) receptor pathways as the neutrophil migrates from a yield of 93106 cells/mouse with a purity of 85%63% and storage compartments into inflamed sites. 95%65% viability. Negative selection from BM cell resulted in purity similar to the Percoll gradient alone (80%65%) while MATERIAL AND METHODS negative selection following Percoll resulted in the highest pur- Mice ity (98%62%) (Table 1). However, it produced a very low cell 6 Experiments were carried out in accordance with the Guide for the yield (162310 cells/mouse), which is in accordance with the 15 Humane Use and Care of Laboratory Animals and were approved literature. For microarray analysis and functional assays, we by the University of Toronto Animal Care Committee (protocol # used the Percoll method alone for neutrophil isolation. For 20009597). C57BL/6 male mice (Charles River, Sherbrooke, QC, Quantitative real-time PCR (qRT-PCR) analysis and western Canada), 8 - 12 weeks of age, were housed in the animal facility at blot analysis, Percoll gradient followed by additional purifica- the University of Toronto, and fed mouse chow and water ad tion using immunomagnetic negative selection was performed. libitum. Preparation of blood PMN Neutrophil isolation Methods similar to those for PMN-BM were performed for We sought to identify the best neutrophil isolation method for isolation of blood neutrophils (PMN-B). Blood from twelve each compartment in mice that would result in adequate yield, mice were pooled to obtain appropriate numbers of cells. viability and purity. With that in mind, we tested different meth- Mice were anesthetized by intraperitoneal injection of keta- ods for each of the three compartments: For isolation of PMN-BM mine (Ketaset 125 mg/kg) and xylazine (10 mg/kg). Blood and PMN-B we used Percoll gradient separation14 or immuno- (1 ml) was collected by cardiac puncture using a sodium citrate magnetic separation.15Additionally, for PMN-B we used the coated syringe and transferred to a 15 ml falcon tube with 100 ml FACS-sorted samples, while peritoneal exudate (PE) PMN were of sodium citrate. Erythrocyte lysis was immediately performed collected with or without Percoll purification as detailed below. by addition of 5 ml of Pharm Lysis buffer (BD Biosciences) as described above. The cell pellet was resuspended in 2 ml of a- Preparation of BM PMN (PMN-BM) MEM and laid on the top of a three-layer Percoll gradient, as Percoll gradient separation was performed as previously described for PMN-BM isolation. Following gradient separa- 6 described. In brief, BM cells were resuspended14 in 4 ml of a- tion, a yield of 0.5310 cells/mouse with purity of 85% and MEM and overlaid on Percoll gradients (82%/65%/55%). After viability of 95% was obtained. Purity and viability was assessed centrifugation (2500 r.p.m., 4 uC, 30 min) the lower band was as described for PMN-BM. Additionally, we used flow cyto- collected and washed with PBS. Erythrocytes were eliminated metric analysis to sort cells based on expression of Ly6G (1A8). by addition of 5 ml of Pharm Lysis buffer (BD Biosciences, Gating was performed based on fluorescence-minus one con- Mississauga, ON, Canada) followed by centrifugation trols. Neutrophils were identified based on the following gating 1 2 (2500 r.p.m., 4 uC, 5 min) as per manufacturer’s instructions. strategy: FSC vs. SSC, then Singlet , followed by eFluor 506 Immunomagnetic negative selection using a Mouse Neutrophil and CD11b1Ly6G1. Cells were kept on ice during all proce- Enrichment Kit (Stemcell Technology, Vancouver, BC, dures. For microarray analysis and functional assays, we used Canada) was also performed following the manufacturer’s Percoll separation alone for neutrophil isolation. For qRT-PCR instructions, on either the whole BM cell or cell populations analysis and western blot analysis, FACS-sorted and Percoll following Percoll gradient separation. Sample purity was con- gradient separation with and without an additional purification firmed by flow cytometry. Neutrophils were identified based on by negative selection, was performed as described for PMN- the following gating strategy: FSC vs. SSC, then Singlet1, fol- BM. After immunomagnetic separation, a yield of 0.33106 lowed by eFluor 506- and CD11b1Ly6G1. Flow cytometry cells/mouse with purity of 98% and viability of .95% was

Table 1 Neutrophil isolation and purity (%6s.d.) Compartment (%)

BM Blood PE Isolation method Percoll only 85.263.6 84.763.6 95.961.7 Percoll1negative selection 97.662.1 98.264.4 n/a Negative selection only 80.165.1 73.968.6 93.764.1 FACS 93.965.2 94.462.2 92.063.7

Abbreviations: BM, bone marrow; PE, peritoneal exudate.

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 55 obtained. For FACS-sorted cells, we obtain a yield of 0.13106 (Illumina, San Diego, CA, USA). The quantile normalization cells/mouse with purity of 94% and viability of .95%. method implemented in lumi R package was used to normalize the data. We use LIMMA (linear models for microarray data) Preparation of PE PMN to identify differentially expressed genes. To compare gene Inflammation was induced by injecting 1 ml of 5 mM sodium expression in multiple groups, we used an empirical Bayes periodate (Sigma, Oakville, ON, Canada) solution into the approach to assess differentially expressed genes, allowing for peritoneal cavity of the mouse. Similar to thioglycollate, multiple testing adjustments. This method does not require the sodium periodate induces cell death in the peritoneum which data to be normally distributed.19 Briefly, it starts by fitting a is the triggering event for neutrophil recruitment. This is highly linear model for each gene in the data, and then empirical Bayes relevant for neutrophil recruitment in diseases with underlying method is used to moderate the standard errors for estimating endogenously triggered inflammation including myocardial the moderated t-statistics for each gene, which shrinks the infarction, atherosclerosis, type 2 diabetes mellitus and some standard errors towards a common value. The corresponding autoimmune diseases.16,17 PE cells were harvested 3 h after P-values for the modified t-statistics were adjusted using the injection by two lavages of the peritoneal cavity with 5 ml of multiple testing procedures developed by Benjamini and cold PBS (Sigma). Collected cells were then washed by cent- Hochberg.20 Genes of interest were selected by setting the false rifugation (2500 r.p.m., 5 min, 4 uC), followed by erythrocyte discovery rate at the level of 0.01 and by fold change of 2. lysis with Pharm Lysis buffer (BD Biosciences). Additional Further analyses were carried out using Ingenuity Pathway purification was performed by resuspending collected cells in Analysis (IPA—http://www.ingenuity.com) software (Ingenuity 100% Percoll solution and ultracentrifugation (23 000 r.p.m., Systems, Redwood City, CA, USA). Additional analysis were 20 min, 4 uC). PMN were collected from the lower opaque layer performed using DAVID 6.7 (http://david.abcc.ncifcrf.gov) and washed twice by adding 10 ml of PBS then centrifuged at (the database for annotation, visualization and integrated dis- 2500 r.p.m. for 5 min at 4 uC. Purity and viability was assessed covery) bioinformatics resources as described elsewhere21 and as described above. Sodium periodate injection into peritoneal using GOEAST (http://omicslab.genetics.ac.cn/GOEAST/) cavity gave a yield of 13106 cells/mouse with PMN purity of (Gene Ontology Enrichment Analysis Software Toolkit) bioin- 95% and viability .95%. For neutrophil chemotaxis and apop- formatics tools.22 For cluster analysis and to generate heat- tosis assays, we collected peritoneal cells without further puri- maps, we used MultiExperiment Viewer (http://www.tm4. fication. Microarray and qRT-PCR analysis was performed org/mev/).23,24 with cells isolated with Percoll purification. All the other experiments were carried out as described. For some experi- Accession codes ments, purification was performing using negative selection as The microarray data complies with MIAME guidelines, and the described for PMN-BM. data set was deposited at Gene Expression Omnibus (National Center for Biotechnology Information), accession number RNA isolation from purified PMN GSE43513. Neutrophils (1.03107 cells) from BM, PE and blood from sepa- rate pools of up to 12 C57BL/6 male mice were immediately qRT-PCR prepared for RNA isolation, to avoid degradation, with MirVana Selected genes were analyzed using quantitative RT-PCR on kit (Ambion, Austin, TX, USA) as described by the manufac- the same RNA samples used for microarray analysis. qRT- turer. RNAse free techniques were utilized during all procedures PCR was performed in triplicates using a CFX96TM Real- with surface and equipment decontamination by a surface Time System (Bio-Rad, Hercules, CA, USA). The protocol decontamination solution (RNAse Zap solution; Ambion). was followed according to Wang et al.,25 with a slight modi- Genomic DNA was eliminated by RNase-free DNase I digestion fication. Total RNA (0.12 mg) was reverse transcribed into (Qiagen, Mississauga, ON, Canada) during the isolation proced- cDNA using Superscript II (Invitrogen Life Technologies, ure. Isolated total RNA was stored at 280 uC and later analyzed Burlington, ON, Canada) and Oligo dT18VN primer (ACGT on an Agilent 2100 bioanalyzer using a RNA 6000 picolabchip Corp., Toronto, ON, Canada) in a 20 ml reaction volume. Two kit (Agilent Technologies, Santa Clara, CA, USA) to certify that negative controls were used to ensure there was no contam- all samples are with minimal degradation as possible. Isolated inating DNA: one without template RNA and another lacking RNA presenting RIN above 8 was selected for further analysis.18 the reverse transcriptase. A reaction mixture was also prepared containing: 5 ml of template cDNA, 15 ml of master mix (1 mlof Microarray analysis forward and 1 ml of reverse primer (both 10 mM stock), 10 mlof Total RNA was measured for gene expression by microarray BioRad Ssofast EvaGreen Super Mix (BioRad) and 3 mlof using the Illumina Mouse-6 v2 Expression BeadChip (48 000 RNase-free distilled water was used. Primers were designed gene transcripts) at the Centre for Applied Genomics (Sick from the PrimerBank ID number (Supplementary Table 2) Kids Hospital, Toronto, ON, Canada), and all experiments and GAPDH was used to normalize the expression data. For were performed in triplicate. data analysis, CFX manager software (Bio-Rad) was used. To Raw mouse microarray data were analyzed using lumi R investigate the effect of neutrophil isolation method on gene package. Background correction was done in Beadstudio expression profiles, we compared gene expression of selected

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 56

genes across multiple methods (Supplementary Tables 2 and 3). Quantification of neutrophil apoptosis Comparisons performed included: BM isolation by Percoll vs. Neutrophil apoptosis was quantified in the three different com- immunoselection, FACS-sorted blood-PMN isolation with min- partments by flow cytometry using FlowJo software for analysis. imal manipulation vs. Percoll and FACS-sorted blood-PMN from Neutrophils were identified based on the following gating strat- mice with and without induction of peritonitis (Supplementary egy: FSC vs. SSC, then Singlet1, followed Ly6G1. Apoptotic cells Table 4). Results are reported using the DCt method, which cal- were gated based on exclusion of eFluor506 staining and culates the difference between the reference and target Ct values Annexin V staining. Cells, which stained positive for Annexin for each sample. First, we determined the relative expression of V only, were considered early apoptotic and cells that stained for each gene for each compartment (BM, blood and PE) using the eFluor506 and Annexin V were considered late apoptotic events. following equation: 2(Ct(GAPDH)2Ct(Target gene))5relative expression. Then, the ratio of expression was determined by dividing Caspase 9 cleavage Assay therelativeexpressionofeachtarget gene from each site (blood Neutrophils were isolated from each compartment (BM, blood or PE) to the relative expression of each target gene from BM and PE) and cultured overnight in complete medium to induce samples. The graphic illustrates the relationships of obtained from apoptosis, followed by a 30 minutes incubation using Fluo- Blood to BM and PE to BM. rescein Active Caspase-9 Staining kit (eBioscience). Cells were then harvested, washed and stained with Annexin V. Finally, Western blots the cells were analyzed in the presence of DAPI Staining Protein lysates from neutrophils were prepared with SDS Solution. Viable cells were gated based on exclusion of DAPI buffer. The lysates were clarified by centrifugation (5 min, staining and then these gated events were analyzed for staining 13 000 r.p.m. at 4 uC). Total protein concentration was mea- of Annexin V and Fluorescein Active Caspase-9. sured with a BCA protein assay kit (Pierce, Rockford, IL USA). Fifteen micrograms of heat-denatured protein was loaded on a Statistical analysis 12% polyacrylamide gel followed by transfer to nitrocellulose For experiments in which there were multiple observations per membrane (Amersham, Baie d’Urfe,QC, Canada). After trans- sample, numerical results were expressed as mean6s.e.m. Each fer,the membrane was incubated for 1 h in blocking buffer (5% experiment was performed at least three times, and within nonfat milk powder in Tris-buffered saline and 0.1% Tween 20 each experiment, each data point had a sample size of no3. (TBST)). The membrane was incubated overnight in primary Statistical analysis was performed using Student’s two-tailed t- antibody in TBST with 5% bovine serum albumin, washed test, unless specified otherwise. Pf0.05 was considered statist- with TBST (3310 min) and incubated for 60 min in horse- ically significant, by GraphPad (http://www.graphpad.com) La radish peroxidase-conjugated secondary antibody/5% nonfat Jolla, CA, USA. milk powder/TBST at room temperature. After washing with TBST (3310 min), the membrane was developed with Western RESULTS Lightning solution (Perkin-Elmer, Guelph, ON, Canada) and Neutrophil isolation and purity exposure to film (Kodak, Rochester, NY, USA). Images Neutrophils from BM, blood and inflamed tissues are part of obtained were quantitated using ImageJ software as described mixed cell populations. Since transcriptome profiling and phe- elsewhere.26 The following antibodies were purchased from notyping assays could be influenced by contaminating cell Cell Signaling (Beverly, MA, USA): ARP2, Wiskott–Aldrich populations we optimized our neutrophil isolation protocols syndrome protein (WASp), CDC42, caspase-9 and Bcl-xL. (Table 1) and verified that the isolation protocols and contaminating cells themselves did not significantly alter the gene Neutrophil chemotaxis assays expression profiles of the isolated neutrophils. qRT-PCR ana- Neutrophils from the three compartments (BM, blood and PE) lyses were performed to evaluate how the method of PMN were suspended in HBSS with 1% gelatin. Neutrophils (13106/ isolation may affect the neutrophil transcriptome. Com- ml) were allowed to attach to bovine serum albumin-coated parison of PMN-BM isolated by Percoll gradient separation glass coverslips (22340 mm) at 37 uC for 15 min. The coverslip with immunomagnetic separation, PMN-B FACS sorted (to was inverted onto a Zigmond chamber, and 100 ml HBSS media minimize manipulation) with Percoll gradient separation yielded was added to the left chamber with 100 ml HBSS media contain- no significant differences in gene expression (Supplementary ing fMLP (1026) was added to the right chamber. Time-lapse Tables 2 and 3). video microscopy was used to examine neutrophil movement in Zigmond chambers. The microscope (Nikon Eclipse E400) The neutrophil is a transcriptionally active cell that alters its was equipped with differential interference contrast optics and gene expression profile as it migrates from storage a 320 objective. Images were captured at 15-s intervals for compartments into sites of inflammation 10 min with a Nikon Coolpix 995 camera. Cell-tracking soft- Following purification of neutrophils from the three compart- ware (Retrac version 2.1.01 Freeware) was used to characterize ments in mice (BM, blood and PE) (Supplementary Figures 1 cellular chemotaxis and the percentage of cells migrating and 2) we performed gene expression analysis using an Illu- toward chemoattractant source was calculated from the cap- mina gene array. Of the 45 282 probe sets, we identified 1216 tured images. genes that were differentially expressed when we compared

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 57 neutrophils from BM (PMN-BM) with neutrophils isolated from 1. Continuously increasing: Overall, 69 genes were consis- blood (PMN-B), 1468 genes when we compared PMN-B and tently increased in their expression from BM through neutrophils from PE (PMN-PE) and 1545 genes when comparing blood through to the peritoneum. Gene ontology analysis PMN-BM vs. PMN-PE, using a cutoff of two-fold change (FC) indicates over representation of processes involved in the and a P-value of 0.001 (Table 2 and Supplementary Table 1). We regulation of apoptosis and cell death, cytokine–cytokine confirmed these results using a second statistical approach, the receptor interaction, and cell adhesion and activation in significance analysis of microarray technique, and similar gene this category (Figure 2a and Supplementary Table 9). numbers and significance levels were found (data not shown). 2. Continuously decreasing: We found 170 genes in which their expression decreased as they migrated from the BM Confirmation of microarray data by qRT-PCR to the blood and into the inflamed tissue. Cell cycle func- To validate the microarray data, changes in expression of 10 tions were among the pathways enriched within this cat- selected genes were assessed by qRT-PCR, in samples taken egory (Figure 2b and Supplementary Table 10). from the three different compartments (Figure 1a). From our 3. Down-up-down pattern: This group consisted of 617 initial microarray analysis using IPA software, we identified genes being low in the BM, then upregulated in PMN-B several pathways that were differentially regulated as neutro- with subsequent downregulation in PMN-PE; cell adhe- phils transit through compartments. For validation of our sion molecules, antigen processing and presentation, and microarray data, we selected cell migration due to its high inflammation mediated by chemokine and cytokine sig- 26 significance value (P51.85310 ) and the high frequency of naling pathway were upregulated in this group (Figure 2c these genes appearing in the array results. For the neutrophil and Supplementary Table 11). chemotaxis pathway 12 genes were differentially regulated: 4. Up-down-up pattern: This group consisted of 360 genes C5AR1, CCL3, CXCL1, CXCL2, CXCL3, fMLP receptor 1, that are upregulated in PMN-BM, downregulated in IL-1B, Lilrb3, MAPKAPK2, NOD2, PDE4B and TLR4. From PMN-B and later upregulated in PMN-PE. Regulation those 12 genes, we selected the 10 genes with the highest fold of actin cytoskeleton and the PI3/kinase pathway were change for further analysis. These initial data were confirmed among those downregulated in PMN-B (Figure 2d and by correlating the signal intensities of the microarray with qRT- Supplementary Table 12). We confirmed the expression PCR experiments using the Pearson linear correlation test to pattern groupings using Cluster analysis with a K-mean measure the relationship between microarray and qRT-PCR clustering algorithm of the high differentially expressed results. The results were highly concordant for the majority 2 transcripts identified in the three neutrophil subsets. The of genes selected (r 50.7180), although some minor differ- up-down-up pattern with decreased regulation in PMN-B ences were found in the magnitude of the change (Figure 1b). is represented by the heat map found in Figure 2e; addi- Additional qRT-PCR analyses were also performed to evaluate tional clusters, demonstrating an up- or downregulatory if the PMN-B transcriptome would be altered by peritonitis. pattern, can be found in Supplementary Table 5—pat- We found that the transcriptome of PMN-B was not altered by terns of expression. the presence of peritonitis (Supplementary Table 4).

Microarray transcriptome analysis through the neutrophil Location of neutrophils significantly influences the gene migration cycle reveals four different gene regulation patterns regulation pattern During our analysis, we could clearly divide genes based on Using IPA software to identify the main biological processes their expression patterns through the three compartments: with changing gene expression as neutrophils leave circulation

Table 2 The neutrophil is a transcriptionally active cell that alters its gene expression profile as it migrates from sterile compartments into sites of inflammation. Bone marrow neutrophils (PMN-BM) were compared with blood neutrophils (PMN- B), PMN-B were compared with peritoneal exudate neutrophils (PMN-PE) and PMN-BM were compared with PMN-PE, using a cutoff of fivefold change (FC) and a P-value of 0.001. For a complete list of genes for each comparison, see Supplementary Table 1 BM vs. blood BM vs. PE Blood vs. PE

Fold change # of genes Fold change # of genes Fold change # of genes

BM f15 39 BM f15 52 Blood f15 19 10–15 75 10–15 46 10–15 33 5–10 249 5–10 168 5–10 145 Blood 5–10 638 PE 5–10 311 PE 5–10 346 10–15 81 10–15 70 10–15 106 f15 82 f15 126 f15 144 Total 1164 Total 773 Total 793

Abbreviations: BM, bone marrow; PE, peritoneal exudate.

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 58

ab18 2.5 16 R2 = 0.718 14 2 12

10 1.5 8 PE 1 6 BLOOD qRT-PCR 4 0.5 2 Fold Change (relative to BM) 0 00.51 1.5 2 2.5 3 Microarray IL1B CCL3 FPR1 C5AR1 CXCL1 CXCL2 CXCL3 NOD2 PDE4B

MAPKAPK2

Figure 1 Confirmation of microarray identified genes by qRT-PCR. (a) Genes with increased expression levels in neutrophils identified by microarray analysis: 10 regulated genes were assessed in samples from BM, blood and PE by qRT-PCR. The graphic illustrates fold change expression of PE (%) and blood (&)relativetoBMexpression.*Pf0.001; **Pf0.003; ***Pf0.02; n53. (b) Relationship between mean expression of the selected genes in a obtained by microarray (X-axis) and qRT-PCR (Y-axis) by Pearson linear correlation analysis of BM vs. blood expression. Each dot represents log fold change of one selected gene. The diagonal line represents the ideal correspondence trend (R250.718, P-,0.01). Similar trends were observed for PE vs. BM comparison. BM, bone marrow; PE, peritoneal exudates; qRT-PCR, quantitative real-time PCR.

and enter a site of inflammation, we grouped differentially pathway. As a first step toward better understanding neutrophil expressed genes into known molecular and cellular functions. chemotaxis following exposure to fMLP, we analyzed the fMLP Molecular functions such as antigen presentation, cell-to-cell receptor signaling pathway and specific key components signaling and interaction and cell death were differentially involved in chemotaxis. It is known that activation of cell sur- regulated in both comparisons (PMN-BM vs. PMN-B and face receptors initiates intracellular signal transduction path- PMN-B vs. PMN-PE), while cellular movement and cell signal- ways, leading to responses including superoxide production, ing were upregulated only when comparing PMN-B vs. PMN- actin reorganization and cell migration.27 PE. Cellular morphology and cellular compromise were found Through our microarray analysis and subsequent additional to be upregulated only in the PMN-BM vs. PMN-B comparison protein expression analysis, we observed upregulation of the (Supplementary Table 6—IPA Summary). Rho family small GTPase CDC42 and actin binding protein Additionally, pathway analysis was performed with DAVID, ARP2 in PMN-PE compared with PMN-B, yet no expression a web-based software package that organizes genes based on difference was found for WASp. CDC42 regulates actin poly- GO Terms, and interrelationship between genes of inquiry. merization through its direct binding to WASp, which subse- Among the altered pathways in both PMN-BM vs. PMN-B quently activates the Arp2/3 complex leading to de novo actin and PMN-B vs. PMN-PE are inflammation mediated by che- assembly28 (Figure 3a–c). mokine and cytokine inflammation signaling pathway, the To verify if these signalling regulation differences translate to MAPK signaling pathway, antigen processing and presentation functional differences, we also analysed neutrophil chemotaxis and T cell activation (Supplementary Table 7A). Pathways that in a gradient of fMLP using Zigmond chambers. Neutrophils are altered in PMN-BM vs. PMN-B analysis only are: regulation from BM migrated with increased speed compared with PMN- of actin cytoskeleton, PI3/kinase pathway and integrin signal- PE (8.8660.271 mm/min vs. 6.1260.103 mm/min, Pf0.0001), ing pathway (Supplementary Table 7B). Finally, we found Toll- while no difference was found between PMN-BM when com- like receptor signaling pathway, NOD-like receptor signaling pared with PMN-B (8.8660.271 mm/min vs. 8.5160.322 mm/ pathway, oxidative stress response, caspase cascade in apopto- min, P50.4542) (Figure 3d1). However, when we analyzed the sis and regulation of BAD phosphorylation among the path- direction of migration (chemotaxis compass), after 10 minutes ways that are enriched when PMN-B were compared with of fMLP stimulation, we found that 87.7%63.2% of PMN-PE PMN-PE (Supplementary Table 7C). Since multiple leukocyte were migrating towards fMLP vs. 69%65.9% of PMN-B migration and apoptosis containing pathways were upregu- (P50.0495) and 59.5%64.0% of PMN-BM (P50.0029) lated in both comparisons we chose to further explore cell (Figure 3d2 and Supplementary Figure 4). movement and cell death categories in the neutrophils from the three compartments. Death receptor pathway Our microarray analysis demonstrated enrichment of apoptosis Evaluation of leukocyte migration—fMLP receptor associated pathways as neutrophils migrates through different signaling and actin sites. It is widely recognized that neutrophils undergo apopto- As cell migration was one function differentially regulated siseven in the absence of any extracellular stimuli.4,10,29,30 The between sites we chose to further analyze components of this Bcl-2 family, which we observed in our microarray analysis to

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 59

a be 16 18 Sox5 Mpo 3.0 8.0 15.0 14 16 Bcl2a1d Ctsg 12 14 Cma1 Ela2 10 12 10 BM - 1 BM - 2 BM - 3 Blood - 10 Blood - 11 Blood - 12 PE - 7 PE - 8 PE - 9 8 Ccr6 lgh-VJ558 8 6 Traf1 Rasl11b 6 4 Cd83 4 Camp Expression level (a.u.) Expression level (a.u.) 2 ll4i1 2 Ngp 0 Ccl24 0 Cdca3 BM Blood PE BM Blood PE

c 16 d 16 Prg4 14 14 Fcnb Retnla 12 12 ltgb2l Ptgis 10 10 Cd177 8 Ecm1 8 Gpr97 6 Fcrls 6 Mapk13 4 C4b 4 Expression level (a.u.) Expression level (a.u.) Serpinf1 2 Ctsa 2 Vcam1 0 Cxcl4 0 BM Blood PE BM Blood PE

Figure 2 Microarray transcriptome analysis of neutrophilsreveals four specific expression patterns. (a) Genes, which are continuously increasing, (b) genes which are Continuously decreasing in their pattern of gene expression. (c) with a down-up-down pattern and (d) up-down-up pattern. Relative fold change of gene expression is shown expressed in a.u. (e) Cluster analysis performed by applying a K-mean clustering algorithm of the most highly differentially expressed transcripts identified in the three neutrophils subsets, represented in this heatmap, demonstrates a up-down- up pattern with decreased regulation in PMN-B (full list can be found in Supplementary Table 5 and DAVID analysis of each cluster can be found in Supplementary Table 7). Green represents downregulation and red upregulation. a.u., arbitrary units; BM, bone marrow; PE, peritoneal exudate. be differentially expressed between compartments, is an impor- confirm results from western blot analysis, we also used a flow tant regulator of neutrophil apoptosis.5,31 Here we found a cytometry-based assay to detect active Caspase 9 in neutrophils mixed pattern with both pro-apoptosis and pro-survival genes from the three compartments (Figure 4d2). Apoptosis among with altered regulation, as neutrophils leave the circulation and sites was also measured by Annexin V expression with arrive at the site of inflammation with downregulation of FasL, 2.2%60.91% in the PMN-BM, 8.7%62.27% in the PMN-B CR3 (Mac-1), caspase 2 and 9 and upregulation of Fas, Daxx, and 3.8%60.72% in PMN-PE (vs. PMN-BM, P50.0333; vs. cIAP and NfkB. Additionally, members of the TLR family PMN-B, P50.05) (Figure 4d1). PMN-PE presented with the including TLR2 and TLR4 were decreased in PMN-PE lowest active caspase 9 activity with median fluorescence intens- (Figure 4a and b and Supplementary Table 13). Interestingly, ity (MFI)58.41310661.113106 (vs. PMN-BM, P50.01; vs. MCL-1 (Bcl2L11), a known regulator of neutrophil apopto- PMN-B, P50.01), PMN-BM with MFI51.43310761.003106 sis,6,32 is highly expressed in all three compartments, but with and PMN-B with MFI51.34310766.453105 (Figure 4d2). no difference in its regulation. Another major regulator of neutrophil apoptosis is Annexin-A1.7,33 Expression of Annexin-A1 DISCUSSION in neutrophils, promotes pro-apoptotic mechanisms8,34 and It has been demonstrated that neutrophils produced in the BM, promote the removal of cells that have undergone apoptosis.9,35 enter the circulation and finally arrive at the site of inflam- From our data, we observed a downregulation of Annexin-A1 mation/infection.1,2,10,11,36–38 Although previous studies have suggesting the balance between anti-apoptotic and pro-apopto- shown that neutrophils can undergo rapid changes in gene tic signaling is active as these cells migrate through compart- expression upon bacterial challenge3,4,12,13,39 or exposure to ments and is an essential element of cellular innate immunity. endotoxin,3,4,10,40 it was previously unknown if and to what In order to extend our microarray results, selected apoptosis extent the neutrophil alters its transcriptome as it transits from proteins were also analyzed by western blot or flow cytometry the BM through the circulation and into sites of inflammation. (Figure 4c and d). We quantified protein expression of caspase Microarrays have significantly enhanced our knowledge of 9 and Bcl-xL (Bcl2L1) and found that consistent with the neutrophils by providing a general indication of the number of microarray data, PMN-PE presented with decreased expression transcripts involved in their biology.14,41In analyzing the tran- of full-length caspase 9 and increased expression of Bcl-xL. To scriptome of murine neutrophils from BM, blood and PE, we

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 60

A(a) C(b) BM vs. Blood 1.8 fMLP Signaling in Neutrophils fMLP 1.6 * ND Extracellular space fMLPR 1.4 Cytoplasm α GTP Cdc42 Neutrophil PIP2 G i Ras PIP2 Gαi 1.2 Gγ Gβγ PLCβ Pertussis Gβ PI3K 1 toxin WASP 0.8 DAG c-Raf ARP 0.6 IP3 PIP3 Rac 2/3 0.4 Ca2+ PKC CDC42 to B-actin ratio 0.2 MEK p47 1/2 Phox IP3R Actin 0 CALM organization IκB BM Blood PE Ca2+ NFκB ERK C(c) Calcineurin ER 1/2 NADPH 1.4 P Oxidase IκB ND ND P 1.2 NFAT NFAT Proteasomal NFκB Respiratory 1.0 degradation burst

Nucleus 0.8 ERK 1/2 0.6

κ NFAT NF B EIk1 0.4

WASP to B-actin ratio WASP 0.2 Chemokine gene expression © 2000-2011 Ingenuity Systems, Inc. All rights reserved. 0.0 BM Blood PE A(b) Blood vs. PE C(d) 1.4 fMLP Signaling in Neutrophils fMLP ** 1.2 Extacellular space fMLPR ND GTP Cytoplasm PIP2 Gαi Ras PIP2 Cdc42 Neutrophil 1.0 Gαi Gγ βγ β G PLC β Pertussis G PI3K 0.8 toxin WASP 0.6 DAG c-Raf ARP IP3 PIP3 Rac 2/3 0.4 2+ PKC Ca ARP2 to B-actin ratio

MEK p47 0.2 1/2 Phox Actin CALM IP3R organization IκB 0.0 Ca2+ NFκB ERK BM Blood PE Calcineurin ER 1/2 NADPH D(a) P Oxidase IκB P 100 ND * NFAT NFAT Proteasomal 90 NFκB Respiratory degradation burst 80

Nucleus 70 ERK 60 1/2 50 κ NFAT NF B EIk1 40 30

8 m/min) μ β-actin 4 6 4 Expression level (a.u.) 0 Arp 2 BM Blood PE β-actin 2 CDC42 WASP ARP2 Migration speed ( 0 BM Blood PE

Figure 3 Analysis of the fMLP receptor signaling pathway. IPA canonical pathway analysis of fMLP receptor signaling pathway, (a1) in PMN-BM vs. PMN-B and (a2) PMN-B vs. PMN-PE. Red represents upregulated genes, green are downregulated genes and white symbols depict neighbouring genes in this analysis. Box highlight genes that were further evaluated by western blot analysis. (b) Graphical representation of gene regulatory

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 61 patterns of CDC42, WASp and ARP2 by microarray analysis. (c1) Representative western blot of CDC42, WASp and ARP2 protein levels. PMN cell lysates from BM, blood and PE (15 mg of total protein per lane) were subjected to SDS–PAGE, transferred to nitrocellulose and probed with a monoclonal antibody against CDC42, WASp or ARP2. Expression was normalized to b-actin used as internal control. Graphs of densitometry analysis of western blots are shown in c2, c3 and c4. Neutrophil migration toward fMLP was assessed using Zigmond chambers. (d1) The proportion of neutrophils oriented towards the fMLP source (leading edge within 180u of the source) and (d2) and absolute speed of migration after 10 min exposure are shown. Graphs represent mean6s.e.m. of 3–5 independent experiments (ND, no difference; *P,0.001; **P,0.05). BM, bone marrow; fMLP, N-formyl-Met-Leu-Phe; PE, peritoneal exudate; WASp, Wiskott–Aldrich syndrome protein. A(a) and A(b) have been granted permission by Ingenuity Systems, Inc. to use copyrighted figures generated from Ingenuity Pathways Analysis in this publication. noted a significant alteration in the neutrophil transcriptome endogenously triggered inflammation including myocardial as these cells migrate through compartments, characterized by infarction, atherosclerosis, type 2 diabetes mellitus and some four distinct gene expression patterns: a continuously increas- autoimmune diseases.16,17 As a result, we believe that the gene ing, a continuously decreasing, a down-up-down and an up- changes here in the sterile inflammation model may also be down-up pattern. Our demonstration that neutrophils consistent with changes in these other inflammatory responses. undergo transcriptome changes as they transition through To perform gene expression analysis, we used a bioinfor- compartments suggest that (i) neutrophils are able to alter their matics approach that allowed us to organize the neutrophil transcriptome as required after leaving the BM; (ii) unique transcriptome into known molecular functions, cellular neutrophil phenotypes may be found within different environ- components and canonical pathways. Using these tools, we ments; and (iii) careful consideration is required when deter- demonstrated that the three populations of neutrophils are mining the experimental source of cells when carrying out significantly different in their gene expression profiles. To bet- neutrophil studies. ter understand these differences, we carried out bio-informatic The need to obtain a highly purified neutrophil population is investigations on those traditionally associated neutrophil important for many studies, especially when dealing with more functions, such as chemotaxis and apoptosis, as well as identify sensitive molecular biology techniques.15,42–44 The neutrophil less well-established functions such as cell-mediated immune literature contains numerous methodologies for neutrophil response and cell-to-cell interaction (Supplementary Table 7). purification; ranging from density gradient centrifugation to When analyzing functions involved in neutrophil chemotaxis, immunomagnetic selection, cell sorting and culture mod- such as cell movement and leukocyte migration, our results els.14,41,45–47 In this study, obtaining high-quality mRNA from were the opposite of what has been previously reported in neutrophils for all three compartments was the greatest obs- neutrophils found in tumors.16,17,49 We found enrichment of tacle and required the most amount of refining. While the leukocyte migration and cell movement in tissue neutrophils current optimal method for obtaining pure neutrophils is (PMN-PE), while Fridlender et al.18,49 reported downregula- immunomagnetic negative selection,15,43,48 this approach tion of the same pathways in neutrophils in tumoral tissue. This resulted in very low cell yields (loss of up to 90% of neutro- difference likely reflects differences in tumor-associated neu- phils—see the section on ‘Materials and methods’) and low trophils and tissue neutrophils, where at a tumor site, once mRNA quality. Both of these drawbacks made microarray neutrophils arrive, they lose their ability to leave the area and impossible with this method. Since neutrophils synthesize therefore no longer require the cell migration apparatus.19,49 low levels of mRNA, we opted to use a Percoll gradient, which Not surprisingly, highlighted by DAVID analysis, we also we manipulated to allow us to obtain a high yield and viability, found upregulation of neutrophil responses, such as Toll-like with minimal levels of contaminating cell lineages. Interes- receptor signaling, NOD-like receptor signaling and the oxid- tingly, western blot analysis of neutrophils isolated by either ative stress response in PMN-PE when compared with PMN-B. Percoll gradients or negative immunomagnetic selection yielded These are all key pathways involved in the neutrophil response identical results, supporting the use of the gradient isolation at inflammatory sites, allowing for detection of invading patho- approach for these experiments (data not shown). We also gens and induction of host antimicrobial defense. obtained similar results by qRT-PCR analysis for FACS-sorted Neutrophil chemotaxis is tightly regulated by a group of samples and samples isolated using the gradient method, also receptors divided into two main groups: non-classical and clas- confirming the validity of our isolation method and data sical chemotactic receptors.20,50 The classical chemotactic obtained. Additionally, to further confirm the low frequency receptors belong to the G-protein coupled receptor family of non-granulocytic cells in the isolated populations, microarray and are also capable of inducing additional immune functions analysis failed to detect transcripts encoding lineage-specific of neutrophils. These receptors include the N-formylated pep- genes highly expressed in T and B cells (CD4, CD19), monocytes tide (N-formyl-L-methionyl-L-leucyl-phenylalanine) fMLP (M-CSFR, CD14) and eosinophils (IL-5R) in any of the com- receptor.21,51 Low concentration of fMLP induces change of partment isolates. Of particular relevance is the triggering event shape (polarization), formation of pseudopod extension and for neutrophil recruitment to the peritoneum. We used sodium ultimately chemotaxis of neutrophils22,52 through activation of periodate, a model of sterile inflammation that induces cell downstream signaling pathways including small GTPase- death without bacterial byproducts. This system is highly rel- associated pathways and subsequent cytoskeletal rearrange- evant for neutrophil recruitment in diseases with underlying ments, while higher concentrations also induce degranulation

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 62

A(a) BM vs. Blood C(b) Death Receptor Signaling 2.0 FasL TNF-a* TNF-a* APO3L APO2L TL1 ND **

Extracellular space TNF- TNF- Fas DR3 DR4/5 DR6 1.5 R1 R2 Cytoplasm Daxx FADD FADD TRADD RAIDD TBK1 FADD TRADD TRAF2 RIP TRADD FADD HSP27 FLIP RIP TANK TRADD Caspase TRADD TRAF2 Caspase TRAF2 8/10 ASK1 Caspase 1.0 8/10 2 NIK TRAF2 ASK1 FLIP clAP k Bib JNK1 NF- B Signaling IKK 0.5 tBib Mitochondria-mediated Mitochondria IkB apoptosis JNK1 Caspase 9 to B-actin ratio tBib NF-kB 0.0 Diablo P Diablo IkB BMBlood PE Bcl-2* HtrA2 HtrA2 NF-kB Cytochr Cytochrome Apaf1 clAP C C Caspase C(c) 9 Caspase Caspase Caspase 7 3 6 1.6

Cell shrinkage 1.4 ** DNA repair DNA fragmentation Chromatin condensation Membrane blebbing 1.2

Apoptosis 1.0 ND

Extracellular space BMBlood PE TNF- TNF- Fas DR3 DR4/5 DR6 R1* R2 Cytoplasm Daxx FADD FADD TRADD RAIDD TBK1 FADD TRADD TRAF2RIP TRADD FADD D(a) HSP27 FLIP RIP TANK* TRADD Caspase TRADD TRAF2 Caspase TRAF2 8/10 ASK1 Caspase NIK 8 8/10 2 TRAF2 BM ASK1 FLIP ** clAP NF-kB Signaling 7 MKK Bib JNK1 Blood 4/7 IKK tBib 6 PE Mitochondria-mediated Mitochondria IkB apoptosis JNK1 5 tBib NF-kB Diablo P Diablo IkB 4 * ** Bcl-2* HtrA2 HtrA2 NF-kB 3 % of cells Cytochrome Cytochr clAP C C Apaf1* Caspase 2 9 Caspase Caspase Caspase 1 7 3 6 0 Cell shrinkage DNA repair DNA fragmentation Chromatin condensation eFluor506+, Annexin V+ eFluor506–, Annexin V+ Membrane blebbing

Apoptosis D(b)

C(a) Millions 14.0 10 12.0 8 10.0 bm blood pe 8.0 6 6.0

BCL-XL MFI - Caspase 9 4.0 Caspase 9 4 CASP9 2.0

Expression level (a.u.) 0.0 MCL-1 Bcl-xL 2 BM Blood PE ANXA1 β-actin 0 BMBlood PE

Figure 4 Analysis of death receptor pathway. (a) Graphic representation of death receptor pathway using IPA canonical pathway analysis. Green represents downregulation and red upregulation (a1) in PMN-BM vs. PMN-B and (a2) PMN-B vs. PMN-PE. Box highlights genes that were further evaluated by western blot analysis. (b) Graphical representation of gene regulatory patterns of Bcl-xL, caspase 9, MCL-1 and ANXA1 by microarray analysis. (c1) Representative western blot of caspase 9 and Bcl-xL protein levels. PMN cell lysates from BM, blood and PE (15 mg of total protein per lane) were subjected to SDS–PAGE, transferred to nitrocellulose and probed with a monoclonal antibody against caspase 9 and Bcl-xL. Protein expression was normalized to b-actin as an internal control. Graphs of densitometry analysis of western blots are shown in c2 and c3. Apoptosis was assessed using flow cytometry analysis. The percentage of early apoptotic (Annexin V1, eFluor 5061) and late apoptotic cells (Annexin V1, eFluor 5062)(d1) and caspase 9 activation (d2) were measured by FACS analysis. Graphs represent mean6s.e.m. of 3–5 independent experiments (ND, no difference; *P,0.01; **P,0.05). a.u., arbitrary units; BM, bone marrow; IPA, Ingenuity Pathway Analysis; MFI, median fluorescence intensity; PE, peritoneal exudate. A(a) and A(b) have been granted permission by Ingenuity Systems, Inc. to use copyrighted figures generated from Ingenuity Pathways Analysis in this publication.

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 63 and generation of oxygen-derived free radicals.23,24,53 In this inhibits neutrophils extravasation and downregulates the study, we selected the fMLP pathway to further investigate its magnitude of the inflammatory response. Without Annexin-A1 regulation during neutrophil migration through different com- in mediating this response, neutrophil extravasation is highly partments. Our bio-informatic approaches identified fMLP active.33 Our data demonstrate increased early apoptotic events receptor 1 to be upregulated in PMN-PE. fMLP receptor 1 in the blood, characterized by increased Annexin-V expression, mediates fMLP-induced actin polymerization and chemo- which likely represents a safety mechanism to promote a smooth taxis.25,54 It has recently been reported that the CDC42/WASp turnover of cells, since apoptotic cells suppress the production of 56 axis downstream of the fMLP receptor mediates neutrophil pro-inflammatory cytokines. Further, a shift occurs in peritoneal polarity,26,48 where CDC42 controls WASp activation, which neutrophils with reduced apoptosis rate, reflecting that preser- in turn induces CD11b clustering, related cytoskeletal stabiliza- vation of neutrophil functions is required at inflammatory sites. tion, cell polarity and efficient migration. Here we also found the Consistent with our previous report, where neutrophils actin regulatory protein, ARP2 to be upregulated likely allowing altered their transcriptome and phenotype as they transited 58,59 for increased directionality in PMN-PE, where neutrophil polar- from circulation into the oral cavity in humans, we also ization is critical. Our microarray analysis suggested that the observed pathways not traditionally associated with neutrophil regulation of cell migration may be altered as the neutrophils biology such as cell-to-cell signaling and interaction, IL-17 transit through compartments. We therefore looked at the signaling and regulation of T-cell activation to be enriched in protein levels for some of the key proteins in this pathway murine blood when compared with BM and the peritoneum. A and assayed chemotaxis. Since ARP2 is key element of actin large number of genes commonly expressed in T cells can also mediated cell migration and was upregulated as the neutrophil be expressed by neutrophils. For example, ZAP70, a compon- migrates from blood into the peritoneum, we assayed expression ent of the T-cell receptor complex, as well as CXCL9 and of this protein as well as the small GTPase CDC42 and WASp. In CXCL10, which are known chemoattractant to T cells, is highly agreement with the microarray data, the expression of ARP2 was regulated in neutrophils, and to finalize some activators of T increased as the neutrophils enter the peritoneum. This result is cell, pathways such as CCL2 and CCL20 can be expressed and produced by activated neutrophils. More importantly, this consistent with data from other cells where ARP2 is required for 7,60 normal directional movement.55 In agreement with the micro- supports the idea that innate and adaptive immunity do not function independently of each other, and neutrophils play array data, WASp showed significantly altered expression in the an important role in these interactions.7 peritoneum compared to the blood (Figure 3). Our data presented here is supported by de Kleijn et al.,61 Apoptosis plays an important role in the neutrophil cycle with where they looked at neutrophil subsets in humans, based on involvementincontinuationorresolutionoftheinflammatory expression of CD16/CD62L. After inducing endotoxemia, they response. Our microarray data highlighted a shift in the balance observed upregulation of genes involved in regulation of between pro-apoptosis and anti-apoptosis molecules between immune responses and apoptosis. This upregulation of apoptosis compartments. Although we recognize that our microarray associated pathways in both our murine model and their human analysis only shows that the regulation of the death receptor model allow us to speculate that similar mechanisms drive pathway is significantly altered as the neutrophils enter the changes in the neutrophil transcriptome in a mice model of tissue compartment, our flow cytometry data further confirm inflammation and an induced endotoxemia model in humans.61 increased apoptosis in blood and tissue as demonstrated by Interestingly, analysis of blood neutrophils from trauma patients increased membrane alterations measured by Annexin V also identified the death receptor pathway, antigen presentation expression. Previously, a microarray-based approach analyzed pathway and oxidative phosphorylation to be upregulated,41 differences in neutrophil gene expression during spontaneous similar to our results presented in this study. These are significant apoptosis after extended culture (12–24 h) with and without 32 observations, considering the differing species and study designs, GM-CSF. Similar to our study, they observed changes in cas- method of PMN isolation and microarray platforms.41 pase 9, as well as Fas and Bcl2 members. Interestingly, we show Our study demonstrates that neutrophils are highly plastic that the expression of these pro-apoptotic genes and pro- cells where compartment cues determine a site-specific neutro- survival genes is also altered as the cell arrives at the site of phil transcriptome. The differences in the cell functions inflammation. Importantly, our data highlight that regulatory between the three compartments allow us to suggest that the mechanisms of neutrophil apoptosis in vivo are differently regu- neutrophil has a site-specific phenotype in each compartment. 32 lated than in vitro. It is well recognized that delay in neutrophil This study focused on gene expression changes in response to a 11,56 apoptosis coincides with preservation of their functions. sterile injury, which suggests that these responses may also be Another interesting finding that is worth to mention, is the vari- extrapolated to other inflammatory models, including contact ability of Annexin-A1 expression found in our microarray ana- hypersensitivity and autoimmune diseases, although further lysis. During an inflammatory insult, Annexin-A1 is known to studies will be required to confirm this. reduce inflammation of tissues by interacting with Annexin-A1 receptors on neutrophils. In turn, upon the activation of these AUTHOR CONTRIBUTIONS receptors, leukocytes are directed to the site of infection and target FL designed, conducted experiments, analyzed data, inter- thesourceofinflammationdirectly.34,57 Consequently, this preted experiments and wrote the paper; MV conducted

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 64

experiments and wrote the paper; CS conducted experiments; neutrophil granulocytes from murine bone marrow for functional and MG designed, interpreted experiments and wrote the studies in vitro and in vivo. PLoS ONE 2011; 6: e17314. 16 Jiang N. The effect of inflammation on the generation of plasma DNA paper. from dead and dying cells in the peritoneum. J Leuk Biol 2004; 77: 296–302. COMPETING CONFLICTS OF INTEREST 17 Shen H, Kreisel D, Goldstein DR. Processes of sterile inflammation. The authors declare no competing financial interests. J Immunol 2013; 191: 2857–2863. 18 Schroeder A, Mueller O, Stocker S, Salowsky R, Leiber M, Gassmann M et al. The RIN: an RNA integrity number for assigning integrity ACKNOWLEDGEMENTS values to RNA measurements. BMC Mol Biol 2006; 7:3. This work was funded by an operating grant from The Canadian 19 Smyth GK. Linear models and empirical Bayes methods for assessing Institutes of Health Research (CIHR, Ottawa, ON, Canada) to MG. FL differential expression in microarray experiments. Stat Appl Genet is supported by CIHR Training Fellowship, TGF-53877 and the Mol Biol 2004; 3: 1–26. Harron scholarship (Faculty of Dentistry, University of Toronto). 20 Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc The authors give special thanks to Dionne White from The Flow Ser B (Methodol) 1995; 57: 289–300. Cytometric Facility of the Immunology Department, Faculty of Medicine 21 Huang DW, Sherman BT, Lempicki RA. Systematic and integrative of University of Toronto for assistance with flow cytometric analysis, also analysis of large gene lists using DAVID bioinformatics resources. Nat to Guillermo Casallo and Pingzhao Hu of the Centre for Applied Protoc 2008; 4: 44–57. Genomics, The Hospital for Sick Children, Toronto, Canada, for 22 Zheng Q, Wang XJ. GOEAST: a web-based software toolkit for Gene assistance with microarray analysis and statistical analysis respectively. Ontology enrichment analysis. Nucleic Acids Res 2008; 36: W358– W363. 23 Chu VT, Gottardo R, Raftery AE, Bumgarner RE, Yeung K. MeV1R: Supplementary Information accompanies the paper on Cellular & using MeV as a graphical user interface for Bioconductor applications Molecular Immunology’s website. (http://www.nature.com/cmi). in microarray analysis. Genome Biol 2008; 9: R118. 24 Yeung KY, Haynor DR, Ruzzo WL. Validating clustering for gene expression data. Bioinformatics 2001; 17: 309–318. 25 Wang Y, Lebowitz D, Sun C, Thang H, Grynpas MD, Glogauer M. 1 Nathan C. Neutrophils and immunity: challenges and opportunities. Identifying the relative contributions of Rac1 and Rac2 to Nat Rev Immunol 2006; 6: 173–182. osteoclastogenesis. J Bone Miner Res 2007; 23: 260–270. 2 Kobayashi SD, Deleo FR. Role of neutrophils in innate immunity: a 26 Luhtala N. Parker R. LSM1 over-expression in Saccharomyces systems biology-level approach. Wiley Interdiscip Rev Syst Biol Med cerevisiae depletes U6 snRNA levels. Nucleic Acids Res 2009; 37: 2009; 1: 309–333. 5529–5536. 3 Zhang X, Ding L, and Sandford A.J. Gene expression in mature 27 Wang F. The signaling mechanisms underlying cell polarity and neutrophils: early responses to inflammatory stimuli. Jo Leuk Biol chemotaxis. Cold Spring Harbor Perspect Biol 2009; 1: a002980– 2004; 75: 358–372. a002980. 4 Subrahmanyam YV, Newburger PE, Weissman SM. RNA expression 28 Kumar S, Xu J, Perkins C, Guo F, Snapper S, Finkelman FD, Zheng Y patterns change dramatically in human neutrophils exposed to et al. Cdc42 regulates neutrophil migration via crosstalk between bacteria. Blood 2001; 97: 2457–2468. WASp, CD11b, and microtubules. Blood 2012; 120: 3563–3574. 5 Cassatella MA. The production of cytokines by polymorphonuclear 29 Moulding DA, Akgul C, Derouet M, White MR, Edwards SW. BCL-2 neutrophils. Immunol Today 1995; 16: 21–26. family expression in human neutrophils during delayed and 6 Gosselin E, Wardwell K, Rigby W, Guyre P. Induction of MHC class II accelerated apoptosis. J Leuk Biol 2001; 70: 783–792. on human polymorphonuclear neutrophils by granulocyte/ 30 Kirschnek S, Vier J, Gautam S, Frankenberg T, Rangelova S, Eitz- macrophage colony-stimulating factor, IFN-gamma, and IL-3. Ferrer P et al. Molecular analysis of neutrophil spontaneous J Immunol 1993; 151: 1482–1490. apoptosis reveals a strong role for the pro-apoptotic BH3-only 7 Pelletier M, Maggi L, Micheletti A, Lazzeri E, Tamassia N, Costantini C protein Noxa. Cell Death Differ 2011; 18: 1805–1814. et al. Evidence for a cross-talk between human neutrophils and Th17 31 Atallah M, Krispin A, Trahtemberg U, Ben-Hamron S, Grau A, cells. Blood 2010; 115: 335–343. Verbovetski I et al. Constitutive neutrophil apoptosis: regulation by 8 Shaw JP, Chuang N, Yee H, Shamamian P. Polymorphonuclear cell concentration via S100 A8/9 and the MEK–ERK pathway. PLoS neutrophils promote rFGF-2-induced angiogenesis in vivo. JSurg ONE 2012; 7: e29333. Res 2003; 109: 37–42. 32 Kobayashi SD. Spontaneous neutrophil apoptosis and regulation of 9 Fridlender ZG, Sun J, Kim S, Kapoor V, Cheng G, Ling L et al. cell survival by granulocyte macrophage-colony stimulating factor. Polarization of tumor-associated neutrophil phenotype by TGF-beta: J Leuk Biol 2005; 78: 1408–1418. ‘‘N1’’ versus ‘‘N2’’ TAN. Cancer Cell 2009; 16: 183–194. 33 Perretti, M. Editorial: To resolve or not to resolve: Annexin A1 pushes 10 Luo HR, Loison F. Constitutive neutrophil apoptosis: mechanisms and resolution on track. J Leuk Biol 2012; 92: 245–247. regulation. Am J Hematol 2008; 83: 288–295. 34 Vong L, D’Acquisto F, Pederzoli-Ribeil M, Lavagno L, Flower RJ, 11 Witko-Sarsat VR, Pederzoli-Ribeil M, Hirsh E, Sozzani S, Cassatella Witko-Sarsat V et al. Annexin 1 cleavage in activated neutrophils: a MA. Regulating neutrophil apoptosis: new players enter the game. pivotal role for proteinase 3. J Biol Chem 2007; 282: 29998–30004. Trends Immunol 2011; 32: 117–124. 35 Arur S, Uche UE, Rezaul K, Fong M, Scranton V, Cowan AE et al. 12 Hudock KM, Liu Y, Mei J, Marino RC, Hale JE, Dai N et al. Delayed Annexin I is an endogenous ligand that mediates apoptotic cell resolution of lung inflammation in IL- 1RN2/2 mice reflects elevated engulfment. Dev Cell 2003; 4: 587–598. IL-17A/G-CSF expression. Am J Respir Cell Mol Biol 2012; 47: 436– 36 Theilgaard-Monch K. The transcriptional program of terminal 444. granulocytic differentiation. Blood 2005; 105: 1785–1796. 13 J. Gamonal J, Sanz M, O’Connor A, Acevedo A, Suarez I, Sanz A et al. 37 Cheretakis C, Leung R, Sun CX, Dror Y, Glogauer M. Timing of Delayed neutrophil apoptosis in chronic periodontitis patients. J Clin neutrophil tissue repopulation predicts restoration of innate Periodontol 2003; 30: 616–623. immune protection in a murine bone marrow transplantation model. 14 Sun CX. Rac1 is the small GTPase responsible for regulating the Blood 2006; 108: 2821–2826. neutrophil chemotaxis compass. Blood 2004; 104: 3758–3765. 38 Pillay J, Braber I, Vrisekoop N, Kwast LM, de Boer RJ, Borghans JAM 15 Hasenberg M, Ko¨hler A, Bonifatius S, Borucki K, Riek-Burchardt M, et al. In vivo labeling with 2H2O reveals a human neutrophil lifespan Achilles J et al. Rapid immunomagnetic negative enrichment of of 5.4 days. Blood 2010; 116: 625–627.

Cellular & Molecular Immunology Neutrophil gene expression profiles changes FS Lakschevitz et al 65

39 Kobayashi SD, Braughton KR, Whitney AR, Voyich JM, Schwan TG, 50 Servant G, Weiner OD, Neptune ER, Sedat JW, Bourne HR. Dynamics Musser JM et al. Bacterial pathogens modulate an apoptosis of a chemoattractant receptor in living neutrophils during chemotaxis. differentiation program in human neutrophils. Proc Natl Acad Sci Mol Biol Cell 1999; 10: 1163–1178. USA 2003; 100: 10948–10953. 51 Zhelev DV, Alteraifi AM, Chodniewicz D. Controlled pseudopod 40 Coldren CD, Nick JA, Poch KR, Wollum MD, Fouty BW, O’Brien JM extension of human neutrophils stimulated with different et al. Functional and genomic changes induced by alveolar chemoattractants. Biophys J 2004; 87: 688–695. transmigration in human neutrophils. Am J Physiol Lung Cell Mol 52 Wong CH, Heit B, Kubes P. Molecular regulators of leucocyte Physiol 2006; 291: L1267–L1276. chemotaxis during inflammation. Cardiovasc Res 2010; 86: 183–191. 41 Kotz KT, Xiao W, Miller-Graziano C, Qian WJ, Russom A, Warner EA 53 Sun R, Iribarren P, Zhang N, Zhou Y, Gong W, Cho EH et al. et al. Clinical microfluidics for neutrophil genomics and proteomics. Identification of neutrophil granule protein cathepsin G as a novel Nat Med 2010; 16: 1038–1043. chemotactic agonist for the G protein-coupled formyl peptide 42 Pelletier M, Micheletti A, Cassatella MA. Modulation of human receptor. J Immunol 2004; 173: 428–436. neutrophil survival and antigen expression by activated CD41 and 1 54 Ye RD, Boulay F, Wang JM, Dahlgren C, Gerard C, Parmentier M et al. CD8 T cells. J Leuk Biol 2010; 88: 1163–1170. International union of basic and clinical pharmacology. LXXIII. 43 Davey MS, Tomassia N, Rossato M, Bazzoni F, Calzetti F, Bruderek K Nomenclature for the formyl peptide receptor (FPR) family. et al. Failure to detect production of IL-10 by activated human Pharmacol Rev 2009; 61: 119–161. neutrophils. Nat Immunol 2011; 12: 1017–1018. 55 Suraneni P, Rubinstein B, Unruh JR, Durnin M, Hanein D, Li R. The 44 Zhang X, Ding L, Sandford AJ. Selection of reference genes for gene Arp2/3 complex is required for lamellipodia extension and directional expression studies in human neutrophils by real-time PCR. BMC Mol fibroblast cell migration. JCellBiol2012; 197: 239–251. Biol 2005; 6:4. 56 Maianski NA, Maianski AN, Kuijpers TW, Roos D. Apoptosis of 45 Hu Y. Isolation of human and mouse neutrophils ex vivo and in vitro. neutrophils. Acta Haematol 2004; 111: 56–66. Leuc Methods Protoc 2011; 844: 101–113. 57 Hannon R. Aberrant inflammation and resistance to glucocorticoids in 46 Siemsen DW, Quinn M. Neutrophil isolation from nonhuman species. 2/2 Methods Mol Biol 2007; 412: 21–34. Annexin 1 mouse. FASEB J 2003; 17: 253–255. 47 de Santo C, Arscott R, Booth S, Karydis I, Jones M, Asher R et al. 58 Lakschevitz FS, Aboodi GM, Glogauer M. Oral neutrophil Invariant NKT cells modulate the suppressive activity of IL-10- transcriptome changes result in a pro-survival phenotype in secreting neutrophils differentiatedwithserumamyloidA.Nat periodontal diseases. PLoS ONE 2013; 8: e68983. Immunol 2010; 11: 1039–1046. 59 Lakschevitz FS, Aboodi GM, Glogauer M. Oral neutrophils display a 48 Kumar S, Xu J, Perkins C, Guo F, Snapper S, Finkelman FD, Zheng site-specific phenotype characterized by expression of T-cell Y et al. Cdc42 regulates neutrophil migration via crosstalk receptors. J Periodontol 2012; 84: 1493–1503. between WASp, CD11b, and microtubules. Blood 2012; 120: 60 Puellmann K, Kaminski WE, Vogel M, Nebe CT, Schroeder J, Wolf H et al. 3563–3574. A variable immunoreceptor in a subpopulation of human neutrophils. Proc 49 Fridlender ZG, Sun J, Mishalian I, Singhal S, Cheng G, Kapoor V et al. Natl Acad Sci USA 2006; 103: 14441–14446. Transcriptomic Analysis Comparing Tumor-Associated Neutrophils with 61 de Kleijn S, Langereis JD, Leentjens J, Kox M, Netea MG, Granulocytic Myeloid-Derived Suppressor Cells and Normal Koenderman L et al.IFN-c-stimulated neutrophils suppress Neutrophils. PLoS ONE 2012; 7(2): e31524. doi:10.1371/journal. lymphocyte proliferation through expression of PD-L1. PLoS ONE pone.0031524 2013; 8: e72249.

Cellular & Molecular Immunology