ORIGINAL ARTICLE Stromal Cell–Derived Factor-1/CXCL12 Stimulates Chemorepulsion of NOD/LtJ T-Cell Adhesion to Islet Microvascular Endothelium Christopher D. Sharp,1 Meng Huang,1 John Glawe,1 D. Ross Patrick,1 Sible Pardue,1 Shayne C. Barlow,2 and Christopher G. Kevil1

OBJECTIVE—Diabetogenic T-cell recruitment into pancreatic islets faciltates ␤-cell destruction during autoimmune diabetes, yet specific mechanisms governing this process are poorly un- iabetogenic T-cell infiltration into pancreatic derstood. The stromal cell–derived factor-1 (SDF-1) islets is a key pathophysiological feature of controls T-cell recruitment, and genetic polymorphisms of SDF-1 autoimmune diabetes. Recruitment of autoreac- are associated with early development of type 1 diabetes. tive T-cells into islets, known as insulitis, ini- D ␤ tiates the process of -cell damage, which may occur over RESEARCH DESIGN AND METHODS—Here, we examined a protracted period of time, eventually leading to frank the role of SDF-1 regulation of diabetogenic T-cell adhesion to ␤ islet microvascular endothelium. Islet microvascular endothelial destruction of -cells and loss of insulin production (1,2). cell monolayers were activated with tumor necrosis factor-␣ Cellular and molecular mechanisms necessary for recruit- (TNF-␣), subsequently coated with varying concentrations of ment and homing of diabetogenic T-cells have been pur- SDF-1 (1–100 ng/ml), and assayed for T-cell/endothelial cell posed, with presentation and changes in adhesion interactions under physiological flow conditions. molecule expression playing important roles in this pro- cess (3–6). However, T-cell recruitment is regulated by RESULTS—TNF-␣ significantly increased NOD/LtJ T-cell adhe- other factors besides antigen presentation and adhesion sion, which was completely blocked by SDF-1 in a dose-depen- molecule expression. Recent studies have revealed that dent manner, revealing a novel chemorepulsive effect. serve critically important roles in properly Conversely, SDF-1 enhanced C57BL/6J T-cell adhesion to TNF- ␣ directing T-cell adhesion and migration, which are neces- –activated islet endothelium, demonstrating that SDF-1 aug- sary for immune cell surveillance and recruitment to ments normal T-cell adhesion. SDF-1 chemorepulsion of NOD/ ␣ discreet tissue compartments (7,8). Given the immunolog- LtJ T-cell adhesion was completely reversed by blocking Gi - protein–coupled activity with pertussis toxin. CXCR4 ical importance of these molecules, no information exists protein expression was significantly decreased in NOD/LtJ T- regarding the manner in which chemokine activity con- cells, and inhibition of CXCR4 activity significantly reversed trols diabetogenic T-cell adhesion and recruitment to islet SDF-1 chemorepulsive effects. Interestingly, SDF-1 treatment microvascular endothelium. significantly abolished T-cell resistance to shear-mediated de- The process of T-cell recruitment involves a dynamic tachment without altering adhesion molecule expression, thus series of events involving cell capture, rolling, firm adhe- demonstrating decreased integrin affinity and avidity. sion, and emigration ultimately resulting in T-cell move- ment into the extravascular tissue. Multiple leukocyte and CONCLUSIONS—In this study, we have identified a previously endothelial cell adhesion molecules orchestrate this event unknown novel function of SDF-1 in negatively regulating NOD/ with selectins regulating cell capture and rolling and LtJ diabetogenic T-cell adhesion, which may be important in integrins regulating T-cell firm adhesion and transmigra- regulating diabetogenic T-cell recruitment into islets. Diabetes 57:102–112, 2008 tion (9–11). Distinct cellular responses accompany the transition of T-cell recruitment from one phase to the next with chemokine receptor interactions serving to activate signaling pathways necessary for firm adhesion. Chemo- From the 1Department of Pathology, Louisiana State University Health Sci- kines are small heparin-binding proteins that have been ences Center, Shreveport, Louisiana; and the 2Department of Pharmacology, defined based on amino acid composition of conserved Physiology and Neuroscience, University of South Carolina, Columbia, South tetra-cysteine motifs, resulting in two major subclasses, Carolina. Address correspondence and reprint requests to Christopher Kevil, PhD, CXC (separation by a nonconserved amino acid) or CC Department of Pathology, Louisiana State University Health Sciences Center- (adjacent cysteine location), along with three other homol- Shreveport, 1501 Kings Hwy., Shreveport, LA 71130-3932. E-mail: ckevil@ ogous molecules of differing motifs (12). Numerous che- lsuhsc.edu. Received for publication 11 May 2007 and accepted in revised form 26 mokines have been identified to date that bind to various September 2007. receptors in redundant fashion. Chemokine receptors are Published ahead of print at http://diabetes.diabetesjournals.org on 1 Octo- surface G-protein–coupled receptors that contain seven ber 2007. DOI: 10.2337/db07-0494. C.D.S. and M.H. contributed equally to this work. membrane-spanning domains that activate downstream FACS, fluorescence-activated cell sorting; GST, glutathione S-transferase; G-protein signal cascades. Chemokines avidly bind glycos- HBSS, Hanks’ balanced salt solution; ICAM-1, intracellular adhesion mole- aminoglycans associated with cells or matrix proteins and cule-1; NIH, National Institutes of Health; PMSF, phenylmethylsulfonyl fluo- are readily diffusible because of their small size (7–15 ride; SDF-1, stromal cell–derived factor-1; TBS, Tris-buffered saline; TNF-␣, tumor necrosis factor-␣; VCAM-1, vascular cell adhesion molecule-1. kDa). Moreover, chemokines may also be transported © 2008 by the American Diabetes Association. through or around microvascular endothelium, thus fur- The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance ther identifying discreet regions for leukocyte recruitment with 18 U.S.C. Section 1734 solely to indicate this fact. (13). Together, these structural and biochemical features

102 DIABETES, VOL. 57, JANUARY 2008 C.D. SHARP AND ASSOCIATES contribute to the strong ability of chemokines to control velocity. Firmly adherent cells were defined as those that did not move one directional leukocyte recruitment and migration. cell diameter over a 5-s period as determined by automated tracking and manual review of individual cells in each experimental field of view. Chemokines can facilitate leukocyte recruitment and ϩ Western blot analysis. CD3 cells were isolated as described above. Cells migration through alterations of adhesion molecule func- were rinsed in Tris-buffered saline (TBS) and spun for 5 min at 1,500 rpm. The tion or cellular location. Chemokine stimulation facilitates resulting pellet was lysed in radioimmunoprecipitation assay buffer (50 mmol/l integrin-mediated adhesion by altering the state of integrin Tris-HCL, pH 8.0, 150 mmol/l NaCl, 1% Nonidet-40, 0.5% deoxycholate, and activation by converting it from a “closed” nonbinding 0.1% SDS) supplemented with 0.1 ␮mol/l leupeptin, 0.3 ␮mol/l aprotinin, and state to an “open” high binding state, thereby increasing 1 ␮mol/l phenylmethylsulfonyl fluoride (PMSF). Samples were sonicated for affinity for ligand or altering integrin cell surface location three 5-s intervals on ice. Protein determination was preformed using a into discreet clusters and, thus, enhancing avidity for Bio-Rad DC Protein kit (Bio-Rad, Hercules, CA) according to the manufactur- er’s instructions. Whole-cell protein homogenates (25 ␮g total protein) were ligand (14). Both of these molecular events may occur loaded on 12% polyacrylamide SDS gels, and electrophoresis was performed simultaneously, providing a very rapid and effective re- as we have previously reported (23). Gels were transferred overnight to sponse to enhance leukocyte adhesion. The chemokine Immobilon-P7 (Bio-Rad), and subsequent membranes were blocked with 5% stromal cell–derived factor-1 (SDF-1)/CXCL12 has been BSA in TBS for 2 h. Anti-CXCR4 antibody (E-Bioscience, San Diego, CA) was reported to rapidly stimulate integrin-dependent T-cell incubated overnight at 1:1,000 dilution at 4°C in blocking buffer supplemented firm adhesion under hydrodynamic flow conditions, which with 0.1% polyoxyethylenesorbitan monolaurate (Tween-20). The remaining involves changes in integrin affinity or avidity in a G- washes and incubations were performed at room temperature in TBS con- protein–coupled receptor-dependent manner (15–17). In- taining 0.1% milk and 0.1% Tween-20. Membranes were washed three times for 5 min and allowed to incubate with the peroxidase secondary anti-rabbit terestingly, recent reports suggest that polymorphisms of antibody for 2 h. After three 10-min washes, membranes were rinsed for 10 SDF-1/CXCL12 may be associated with the early develop- min in TBS alone. Chemiluminesence was preformed using ECL detection ment of autoimmune diabetes, yet the effects of SDF-1 on reagents (Amersham, Piscataway, NJ) according to the manufacturer’s direc- regulating diabetogenic T-cell adhesion are completely tions. Various exposures to Hyblot film (Denville Scientific, Metuchen, NJ) unknown (18,19). In this study, we examined the effect of were performed to insure exposure linearity. Films were scanned and SDF-1 on NOD/LtJ diabetogenic T-cell adhesion to acti- quantified using Image J (National Institutes of Health [NIH], Bethesda, MD). vated islet microvascular endothelial cells to obtain a Densitometric values were reported as means Ϯ SE. Three isolations from each mouse strain were performed, and samples were run in triplicate. better understanding of chemokine-dependent regulation RAP1 activity assay. Rap1 activation was measured using a kit from of diabetogenic T-cell recruitment. Stressgen (Ann Arbor, MI) according to the manufacturer’s directions. SDF-1 was added to C57BL/6J or NOD/LtJ T-cells at a concentration of 100 ng/ml and allowed to react for 5 or 15 min. Controls were performed according to the RESEARCH DESIGN AND METHODS manufacturer’s directions. The Rap1 activation kit used to detect active Mice used in this study were bred and housed at the Association for GTP-Rap1 uses a glutathione S-transferase (GST)-fusion protein containing Assessment and Accreditation of Laboratory Animal Care, international- the Rap1 binding domain to affinity purify active Rap1 from cell lysates. The accredited Louisiana State University Health Sciences Center-Shreveport fusion protein is incubated with the cell lysate and captured using a glutathi- animal resource facility and maintained according to the National Research one-conjugated filter disc. The affinity-purified product is detected by Western Council Guide for Care and Use of Laboratory Animals. Twelve-week-old blot analysis using anti-Rap1 antibody. Affinity purification of active Rap1 was female NOD and C57BL/6 mice were used for cell isolations. performed using 500 ␮g total protein from each sample. Approximately 15 ϫ Cell culture materials. All tissue culture media and reagents were pur- 106 CD3ϩ cells from either C57BL/6J or NOD/LtJ mice for each sample were chased from Sigma (St. Louis, MO). Purified recombinant murine SDF-1␣ and rinsed in TBS and lysed at 4°C using the Tris-based kit buffer supplemented tumor necrosis factor-␣ (TNF-␣) proteins were purchased from R&D Systems with 0.1 ␮mol/l leupeptin, 0.3 ␮mol/l aprotinin, and 1 ␮mol/l PMSF. Samples (Minneapolis, MN). The MS1 mouse pancreatic islet microvascular endothelial were passed over resin discs containing the GST-fusion capture peptide; cell line was cultured and maintained as we have previously reported (20). bound protein was eluted using sample buffer (0.060 mol/l Tris-HCL, pH 6.8, Endothelial cell cultures were routinely cultivated in T-75 flasks. For parallel 0.1% SDS, 2% glycerol, 5% ␤-mercaptoethanol, and 0.05% bromphenol blue) by plate flow chamber studies, endothelial cells were seeded into 35-mm culture heating to 95°C for 3 min and loaded on 12% polyacrylamide SDS gels; and dishes and grown to confluency. Confluent endothelial cell monolayers were Western analysis was performed. Gels were transferred to nitrocellulose stimulated with 10 ng/ml TNF-␣ 4 h before use in flow cytometric analysis or membranes and blocked for2hin5%BSAinTBS. Rap1 antibody (1:1,000) was parallel plate flow chamber adhesion assays. Anti-CXCR4 antibody (10 ␮g/ml; incubated overnight at 4°C in 5% BSA and 0.1% Tween-20 in TBS and then R&D Systems) was used in some experiments to block T-cell CXCR4 binding washed three times for 5 min in TBS containing 0.1% milk and 0.1% Tween. A to SDF-1. peroxidase-conjugated secondary antibody, anti-rabbit IgG (1:2000), was T-cell isolation procedure. Whole splenocytes from female NOD/LtJ or allowed to react for1hatroom temperature in wash buffer. Membranes were C57BL/6 mice were obtained from spleens that were ground between two again washed three times for 5 min in the same buffer followed by a 10-min frosted slides into a Petri dish containing RPMI as we have previously wash in TBS alone. Chemiluminesence was preformed using Amersham ECL reported (21). The cell suspension was filtered twice through 70-␮m cell detection reagents according to manufacturer’s directions. Various exposures strainers before lysis of erythrocytes. Target T-cell subpopulations were to Hyblot film (Denville Scientific) were performed to insure exposure obtained by purification with CD3, CD4, or CD8 SpinSep Mouse linearity. Films were scanned and quantified using Image J (NIH). Densito- Enrichment kits from StemCell Technologies (Vancouver, BC, Canada). metric values were reported as means Ϯ SE. In vitro hydrodynamic flow chamber adhesion assay. Hydrodynamic Flow cytometry analysis. Measurement of T-cell and endothelial cell surface parallel plate flow chamber studies were performed as we have previously adhesion molecule expression was performed by flow cytometry as we have reported (20,22). Briefly, mouse leukocytes were labeled with a fluorescent previously reported (20,22,24). Antibodies and the dilutions used against these dye by 30-min incubation at 37°C with 200 nmol/l Cell Tracker Green molecules are as follows: CD18 (1:320), CD29 (1:320), CD11a (1:80), CD49d purchased from Molecular Probes. The labeled cells were resuspended in (1:80), intracellular adhesion molecule-1 (ICAM-1) (1:80), vascular cell adhe- Hanks’ balanced salt solution (HBSS) at 2 ϫ 105 cells/ml in a 200-ml beaker sion molecule (VCAM-1) (1:80), E-selectin (1:320), and P-selectin (1:320) and kept at 37°C and stirred at 60 rpm. A Glycotech flow chamber insert and isotype controls at the respective dilutions. Cultured MS1 cell lines were gasket were used to form a laminar plate flow chamber that could be viewed harvested and washed in 10 ml fluorescence-activated cell sorting (FACS) on a microscope. The labeled cells were drawn from the beaker into the flow buffer (PBS plus 1% fetal bovine serum). An aliquot of 5 ϫ 105 cells was used chamber across endothelial cell monolayers at a physiological shear rate of 1.5 for adhesion molecule analysis. All cells were preincubated on ice for 20 min dynes/cm2 using a programmable digital syringe pump. All endothelial cell with 50 ␮l 1:100 anti–FC-receptor antibody to block nonspecific binding. The monolayers were washed three times with HBSS to remove remaining above-mentioned diluted antibodies were then added to the cells and incu- /chemokines before flow chamber assembly. Fluorescently labeled bated on ice for 20 min. The cells were washed twice with 1 ml FACS buffer cells were viewed using a Nikon Eclipse TE-2000 epifluorescent microscope and resuspended in 300 ␮l FACS buffer. Immunofluorescence-stained samples equipped with a Hamamatsu digital camera, and real-time digital video was were analyzed on a FACS Calibur flow cytometer (Becton Dickinson) made captured using SIMPLE PCI software from Compix. The motion-tracking available through the Research Core Facility at Louisiana State University analysis feature of the software enabled calculation of individual cell-rolling Health Sciences Center. Data analysis was performed using CELL Quest

DIABETES, VOL. 57, JANUARY 2008 103 CHEMOREPULSION OF DIABETOGENIC T-CELL ADHESION software (Becton Dickinson). The cells were gated based on forward versus 1B shows that pretreatment of NOD/LtJ CD3 T-cells with 1 side scatter, and 10,000 events were collected. ␮g/ml pertussis toxin for 30 min completely restores T-cell detachment assay. Shear-mediated detachment assays were per- ␣ formed to determine changes in T-cell integrin affinity and avidity necessary adhesion to TNF- –activated endothelium coated with 100 for T-cell firm adhesion (25). Briefly, NOD/LtJ T-cells were introduced into the ng/ml SDF-1. flow chamber and allowed to adhere to endothelial monolayers for 15 min By way of comparison, we then performed C57BL/6J under static conditions. After the incubation period, flow was resumed at 0.15 CD3 T-cell firm adhesion studies to various islet endothe- dyne/cm2 using cell-free perfusate, and the shear stress was doubled every lial monolayers. Figure 1C demonstrates that TNF-␣ in- 2 30 s using a programmable syringe pump (0.15–38.4 dynes/cm ). The percent- creases C57BL/6J CD3 T-cell firm adhesion and that 50 age of cells remaining adherent was determined by counting the number of adherent cells remaining on completion of the respective shear stress interval ng/ml SDF-1 further enhances firm adhesion consistent and dividing that number by the total number of initially adherent cells. The with previous reports (15,17,28). However, it has been number of cells remaining adherent during a given shear stress interval shown that low doses (5–10 nmol/l) of SDF-1 stimulate reflects the degree of adhesion molecule affinity and/or avidity. T-cell , whereas a high dose of SDF-1 (100 Statistical analysis. Data were statistically compared using Prism 4.0 nmol/l) stimulates chemorepulsion of T-cell migration software (GraphPad). The number of firmly adherent cells was compared (26). Therefore, we examined whether 100 nmol/l (800 using a one-way ANOVA with Bonferroni’s post hoc test to determine statistical differences between experimental groups. Firm adhesion data are ng/ml) SDF-1 might also stimulate chemorepulsion of reported as the mean and SE. Rolling velocity data from 1,200 cells per C57BL/6J CD3 T-cell firm adhesion to TNF-␣–activated treatment group were compared using a Kruskal-Wallis nonparametric monolayers. Figure 1C illustrates that 800 ng/ml SDF-1 ANOVA with a Dunn’s post hoc test to determine statistical differences significantly prevented C57BL/6J CD3 T-cell firm adhesion between experimental groups. Rolling velocity data are reported as a bar to TNF-␣–activated islet endothelium, which was reversed graph illustrating the mean rolling velocity and SE and a relative frequency by pretreating C57BL/6J T-cells with pertussis toxin. histogram distribution identifying cell populations rolling at various velocity intervals. The percentage of remaining firmly adherent cells during the We further examined whether SDF-1 stimulated che- detachment assay was compared by one-way ANOVA with Bonferroni’s post morepulsion of different NOD/LtJ T-cell populations. Fig- hoc test between experimental groups at each shear interval. Densitometric ure 1D and E illustrates that SDF-1 treatment resulted in values for CXCR4 protein expression between C57BL/6J and NOD/LtJ mice chemorepulsion of firm adhesion of both CD4 and CD8 were compared using an unpaired Student’s t test. Rap1 activity assays were T-cells from NOD/LtJ mice. Importantly, pretreatment of compared using a one-way ANOVA with Bonferroni’s post hoc test to Ͻ these T-cell populations with pertussis toxin prevented determine differences between time points. A P value of 0.05 was required SDF-1 chemorepulsion of firm adhesion, reiterating a role to achieve statistical significance for all experimental procedures. ␣ for Gi -protein–coupled signaling. SDF-1 does not alter T-cell–rolling parameters. The RESULTS process of leukocyte recruitment is a multistep event in SDF-1 stimulates chemorepulsion of diabetogenic T- which leukocyte capture and rolling are necessary to cell adhesion. Previous reports demonstrate that SDF-1 initiate firm adhesion. Moreover, SDF-1 has been reported facilitates firm adhesion of normal T-cells under hydrody- to destabilize L-selectin–dependent T-cell rolling indepen- namic flow conditions (16,17). However, the effect of dent of shedding, suggesting that SDF-1 could alter rolling SDF-1 on diabetogenic or autoimmune T-cell firm adhe- behavior (29). Therefore, we examined whether SDF-1 sion is not known. Figure 1A illustrates the experimental chemorepulsion of firm adhesion involved defective T-cell design used to investigate the effect of SDF-1 on NOD/LtJ rolling. Figure 2 reports biophysical rolling data compar- T-cell firm adhesion to TNF-␣–activated mouse islet mi- ing C57BL/6J and NOD/LtJ CD3 T-cell–rolling interactions crovascular endothelial cells under hydrodynamic flow. under hydrodynamic flow conditions with unstimulated, Mouse T-cells were drawn across unstimulated, 10 ng/ml TNF-␣–stimulated, or TNF-␣–stimulated plus 100 ng/ml TNF-␣–stimulated, or 10 ng/ml TNF-␣–stimulated and SDF-1–coated mouse pancreatic islet endothelium. Figure SDF-1–coated islet microvascular endothelial cell mono- 2A illustrates the number of T-cells rolling on the various layers using a hydrodynamic parallel plate flow chamber islet microvascular endothelial cell cultures. TNF-␣ stim- to measure T-cell–endothelial cell biophysical interactions ulation significantly increased the number of rolling (22). Figure 1B shows biophysical data of diabetogenic C57BL/6J and NOD/LtJ CD3 T-cells. Interestingly, SDF-1 NOD/LtJ CD3 T-cell firm adhesion to control, TNF-␣– treatment further augmented the number of rolling NOD/ stimulated, or TNF-␣–stimulated plus SDF-1–coated (1, LtJ CD3 T-cells but not C57BL/6J T-cells. Figure 2B 10, 50, and 100 ng/ml) islet endothelial cells in a dose- illustrates the average rolling velocity of CD3 T-cells. Both response fashion. NOD/LtJ CD3 T-cell firm adhesion to C57BL/6J and NOD/LtJ CD3 T-cells rolled significantly TNF-␣–stimulated islet endothelium was significantly in- slower on TNF-␣–stimulated or TNF-␣–stimulated plus creased under hydrodynamic flow conditions compared SDF-1–treated monolayers compared with unstimulated with unstimulated islet endothelial monolayers. Surpris- monolayers. These data clearly demonstrate that SDF-1 ingly, NOD/LtJ T-cell firm adhesion to TNF-␣–activated does not antagonize NOD/LtJ CD3 T-cell rolling. endothelium was significantly inhibited by surface coating The process of leukocyte slow rolling is important in of SDF-1 in a dose-dependent manner, demonstrating facilitating cell signaling, which results in cellular activa- chemorepulsion of T-cell firm adhesion. Complete inhibi- tion and conversion to firm adhesion (22,30). Thus we tion of TNF-␣–dependent firm adhesion was accomplished determined the population frequency of CD3 T-cells rolling using a dose of 100 ng/ml SDF-1. Importantly, 100 ng/ml at various velocity intervals. Figure 2C and D report rolling SDF-1 had no effect on NOD/LtJ T-cell firm adhesion to velocity histograms for C57BL/6J and NOD/LtJ CD3 T- unstimulated endothelium. cells, respectively. A similar fraction of CD3 T-cells from SDF-1 binds to its primary receptor CXCR4 and signals both C57BL/6J and NOD/LtJ mice rolled at slow velocities ␣ Ͻ ␮ through a pertussis toxin–sensitive Gi -protein–coupled ( 50 m/s), which are necessary for conversion to firm pathway (26,27). Therefore, we examined whether pertus- adhesion. In all of the rolling parameters analyzed, NOD/ sis toxin pretreatment of NOD/LtJ CD3 T-cells could LtJ CD3 T-cells showed an equivalent response to prevent SDF-1–mediated chemorepulsion of T-cell adhe- C57BL/6J CD3 T-cells, indicating that rolling responses are sion to TNF-␣–activated endothelial monolayers. Figure not defective in response to SDF-1.

104 DIABETES, VOL. 57, JANUARY 2008 C.D. SHARP AND ASSOCIATES

FIG. 1. SDF-1 stimulates chemorepulsion of NOD/LtJ T-cell firm adhesion. A: Various treatments of islet microvascular endothelial cell monolayers used in the parallel plate flow chamber to measure T-cell adhesion. B: Effect of a dose-response curve of SDF-1 treatment on NOD/LtJ CD3 T-cell adhesion to TNF-␣–activated islet endothelial monolayers and the effect of pertussis toxin pretreatment of NOD/LtJ T-cells. C: Effect of a dose-response curve of SDF-1 treatment on C57BL/6J CD3 T-cell adhesion to TNF-␣–activated islet endothelial monolayers and the effect of pertussis toxin treatment of C57BL/6J T-cells. D and E: Effect of 100 ng/ml SDF-1 on NOD/LtJ CD4 and CD8 T-cell adhesion to TNF-␣–stimulated islet endothelium, respectively. ***P < 0.001 vs. control; #P < 0.05 vs. TNF-␣ alone; ##P < 0.01 vs. TNF-␣ alone; ∧P < 0.05 vs. TNF-␣ alone.

Role of CXCR4 in SDF-1 chemorepulsion. We next showed a significant 60% reduction of CXCR4 protein examined the role of CXCR4 in modulating SDF-1 che- expression compared with C57BL/6J T-cells. Figure 3B morepulsive effects on NOD/LtJ CD3 T-cells. Figure 3A reports the effect of anti-CXCR4 antibody (10 ␮g/ml) shows CXCR4 protein expression between C57BL/6J and blockade on SDF-1 chemorepulsion. Inhibition of SDF-1/ NOD/LtJ T-cells. Interestingly, NOD/LtJ CD3 T-cells CXCR4 binding conferred a significant yet partial 43 Ϯ 2%

DIABETES, VOL. 57, JANUARY 2008 105 CHEMOREPULSION OF DIABETOGENIC T-CELL ADHESION

FIG. 2. SDF-1 does not alter NOD/LtJ or C57BL/6J T-cell–rolling parameters. A: Number of rolling CD3 T-cells per 500 ␮m2 between NOD/LtJ and C57BL/6J on control, TNF-␣–stimulated, or TNF-␣–stimulated and 100 ng/ml SDF-1–treated islet endothelial monolayers. B: Average rolling velocity of CD3 T-cells from either NOD/LtJ or C57BL/6J on control, TNF-␣–stimulated, or TNF-␣–stimulated and 100 ng/ml SDF-1–treated islet endothelial monolayers. C: Population frequency of C57BL/6J CD3 T-cell rolling at various velocity intervals. D: Population frequency of NOD/LtJ CD3 T-cell rolling at various velocity intervals. **P < 0.01 vs. respective control.

106 DIABETES, VOL. 57, JANUARY 2008 C.D. SHARP AND ASSOCIATES

FIG. 3. CXCR4 function and SDF-1 chemorepulsion. Western blot analysis was used to determine CXCR4 protein expression between C57BL/6J and NOD/LtJ T-cells. A: NOD/LtJ cells have a significantly reduced amount of CXCR4 expression compared with C57BL/6J T-cells. B: Anti-CXCR4 blockade significantly reverses SDF-1 chemorepulsion of NOD/LtJ T-cell firm adhesion to TNF-␣. C: Effect of SDF-1 treatment on Rap1 activity between C57BL/6J vs. NOD/LtJ T-cells at 0, 5, and 15 min. *P < 0.01 C57 vs. NOD (in A). *P < 0.001 control vs. TNF-␣;#P < 0.001 TNF-␣ vs. TNF-␣ plus SDF-1; ␨P < 0.05 TNF-␣ plus SDF-1 plus anti-CXCR4 (10 ng/ml) vs. TNF-␣ plus SDF-1 (in B). reversal of SDF-1 chemorepulsion, suggesting involvement examined whether 100 ng/ml SDF-1 treatments altered endo- of additional pathways. Given the significantly reduced thelial cell surface adhesion molecule expression. Flow cy- CXCR4 expression and its partial involvement in mediat- tometric analysis was performed on nonstimulated, TNF-␣– ing SDF-1 chemorepulsion, we next examined whether stimulated, and TNF-␣–stimulated plus SDF-1–coated mouse SDF-1–dependent Rap1 activation was also altered be- pancreatic islet endothelial cells to measure the surface tween C57BL/6J versus NOD/LtJ T-cells. SDF-1/CXCR4 expression of endothelial cell adhesion molecules P-selectin, interactions result in activation of Rap1 activity, E-selectin, ICAM-1, and VCAM-1 (Fig. 4). Figure 4 shows that which alters adhesion molecule functions (31–33). Figure low basal levels of constitutive P-selectin, ICAM-1, and 3C reports Rap1 kinase activity from C57BL/6J or NOD/LtJ VCAM-1 expression were observed on unstimulated endothe- T-cells stimulated with 100 ng/ml SDF-1 at 0, 5, and 15 min. lium. Expression of P-selectin, E-selectin, ICAM-1, and SDF-1 treatment stimulated a rapid, transient increase in VCAM-1 were all significantly upregulated on stimulation C57BL/6J Rap1 activity that is consistent with previous with 10 ng/ml TNF-␣. Importantly, treatment with SDF-1 did studies (32,33). Conversely, Rap1 activation by SDF-1 was not significantly alter TNF-␣ induction of increased adhesion significantly delayed in NOD/LtJ T-cells, and basal Rap1 molecule expression, and SDF-1 treatment alone did not activity was distinctly absent in NOD/LtJ T-cells compared affect resting adhesion molecule expression. with C57BL/6J. Together, these data highlight a significant SDF-1 does not alter surface expression of T-cell difference in NOD/LtJ CXCR4 expression and function in adhesion molecules. VLA-4 (CD49d/CD29) and LFA-1 response to SDF-1, which suggests that other signaling (CD11a/CD18) are the primary T-cell integrin adhesion mol- pathways likely contribute to SDF-1 chemorepulsion of ecules that mediate firm adhesion to endothelial cells (35). NOD/LtJ T-cells. Therefore, we next examined whether 100 ng/ml SDF-1 SDF-1 does not alter surface expression of endothe- altered the surface expression of these adhesion molecules. lial cell adhesion molecules. Proper surface expression Figure 5 reports surface expression staining by flow cytom- of endothelial cell adhesion molecules is an obligatory re- etry analysis for CD29 and CD49d on either C57BL/6J or quirement for T-cell firm adhesion (34). Therefore, we next NOD/LtJ CD3 T-cells. Stimulation with 100 ng/ml SDF-1 for

DIABETES, VOL. 57, JANUARY 2008 107 CHEMOREPULSION OF DIABETOGENIC T-CELL ADHESION

FIG. 4. SDF-1 does not alter islet endothelial cell surface adhesion molecule expression. Flow cytometry analysis was used to measure surface expression of endothelial cell adhesion molecules. Surface staining for P-selectin, E-selectin, ICAM-1, or VCAM-1 was done on unstimulated, TNF-␣–stimulated, TNF-␣–stimulated and SDF-1–treated, or SDF-1–treated endothelial monolayers. The thick line indicates flow cytometry intensity for the specified antigen, whereas the thin line illustrates isotype control staining of the same endothelial cell population. either 5 or 15 min did not significantly change C57BL/6J or CD3 T-cells were perfused onto various endothelial cell NOD/LtJ CD49d/CD29 surface expression. Likewise, Fig. 6 monolayers and allowed to adhere under static conditions demonstrates that SDF-1 treatment also did not alter for 15 min. Hydrodynamic flow was then reestablished and C57BL/6J or NOD/LtJ CD3 T-cell surface expression of doubled every 30 s to determine the resistance of NOD/LtJ CD11a/CD18 over the 15-min experimental period. CD3 T-cell adhesion to increasing shear, which is directly SDF-1 decreases diabetogenic T-cell integrin activa- proportional to degree of integrin affinity and avidity. tion. Data presented thus far demonstrate that SDF-1 Figure 7A illustrates that NOD/LtJ CD3 T-cells readily treatment is chemorepulsive for NOD/LtJ CD3 T-cell firm detach from unstimulated islet endothelial cells, with adhesion to activated islet endothelium that does not Ͻ10% remaining adherent at a shear stress of 1.5 dynes/ involve alterations in cell capture, rolling, or surface cm2. Conversely, NOD/LtJ CD3 T-cells adherent to TNF- adhesion molecule expression. An additional layer of ␣–stimulated islet endothelium were significantly resistant regulation of firm adhesion is accomplished by altering to shear-mediated detachment, with ϳ40% of the popula- integrin affinity and/or avidity for its counter-ligand. SDF-1 tion still adherent at a shear stress of 38.4 dynes/cm2. has been reported to stimulate changes in VLA-4 and SDF-1 treatment (100 ng/ml) completely inhibited NOD/ LFA-1 affinity and avidity, which are important for firm LtJ CD3 T-cell resistance to shear-mediated detachment adhesion by mediating resistance against increasing shear- on TNF-␣–activated islet endothelium, with Ͻ10% re- mediated detachment (17,28,36). Therefore, we performed maining adherent at 0.6 dyne/cm2. Importantly, pretreat- shear-mediated detachment assays to examine whether ment of NOD/LtJ CD3 T-cells with pertussis toxin SDF-1 stimulates chemorepulsion of firm adhesion by completely reversed SDF-1–dependent loss of shear altering changes in integrin affinity or avidity. NOD/LtJ resistance.

108 DIABETES, VOL. 57, JANUARY 2008 C.D. SHARP AND ASSOCIATES

FIG. 5. SDF-1 does not alter CD3 T-cell CD29 or CD49d surface expression. CD3 T-cells from either C57BL/6J or NOD/LtJ were stimulated with 100 ng/ml SDF-1 at various times and examined for CD49d/CD29 surface expression by flow cytometry analysis. Time points are indicated in the top right corner. The thick line indicates flow cytometry intensity for the specified antigen, whereas the thin line illustrates isotype control staining of the same cell population.

The above results indicate that NOD/LtJ integrin affinity clearly implicate SDF-1 in regulating the development of or avidity is diminished by SDF-1, which suggests that autoimmune diabetes; however, the manner in which this integrin activity could be constitutively greater in NOD/LtJ could occur is not known. Our data corroborate the CD3 T-cells. Therefore, we measured the absolute num- importance of SDF-1 in controlling diabetogenic T-cell bers of C57BL/6J or NOD/LtJ firmly adherent CD3 T-cells function and provide a potential mechanistic explanation to control or TNF-␣–stimulated islet endothelial monolay- for how SDF-1 may be critical in regulating diabetogenic ers. Figure 7B illustrates that NOD/LtJ CD3 T-cells show a T-cell entry into pancreatic islets. significantly greater number of firmly adherent cells under The process of leukocyte recruitment is essential for both control and TNF-␣–stimulated conditions compared initiating and sustaining inflammatory states within tis- with C57BL/6J CD3 T-cells. These results demonstrate that sues. As such, the ability to recruit a leukocyte into a integrin affinity and avidity is significantly greater in NOD/ particular tissue niche is regulated at several levels, with LtJ CD3 T-cells, consistent with the results from shear- chemokines serving important roles in directing cell-cell mediated detachment experiments. adhesion and chemotaxis responses (12). It has been reported that genetic mutations or increased expression of DISCUSSION various chemokines facilitate the development of autoim- Recent reports have shown that SDF-1 genetic polymor- mune diabetes in NOD mice and patients (41–45). How- phisms are associated with early onset of autoimmune ever, the specific cellular responses and mechanisms of diabetes (18,19). Interestingly, some studies suggest that chemokine regulation of diabetogenic T-cell recruitment this may not be true for all ethnic backgrounds or in are completely unknown. Moreover, nothing is known disease conditions involving other autoimmune disorders regarding the manner in which SDF-1 affects diabetogenic (37,38). Nonetheless, a recent study reported that genetic versus normal T-cell homing and adhesion, highlighting a polymorphisms of SDF-1 result in decreased protein ex- key deficiency in our understanding of autoimmune T-cell pression in lymphoblasts on activation using allele-specific recruitment. In this study, we have revealed that SDF-1 transcript quantification methods (39). Consistent with mediates a novel chemorepulsive effect on diabetogenic these findings, a recent study suggested that SDF-1/CXCR4 T-cell firm adhesion compared with enhancing normal signaling protects against autoimmune diabetes in NOD T-cell firm adhesion. This finding is in stark contrast to all mice as inhibition of CXCR4 activity exacerbates adoptive previous studies investigating the effect of SDF-1 on T-cell transfer of diabetes (40). Together, these observations recruitment, which show that SDF-1 augments T-cell ad-

DIABETES, VOL. 57, JANUARY 2008 109 CHEMOREPULSION OF DIABETOGENIC T-CELL ADHESION

FIG. 6. SDF-1 does not alter CD3 T-cell CD18 or CD11a surface expression. CD3 T-cells from either C57BL/6J or NOD/LtJ were stimulated with 100 ng/ml SDF-1 at various times and examined for CD11a/CD18 surface expression by flow cytometry analysis. Time points are indicated in the top right corner. The thick line indicates flow cytometry intensity for the specified antigen, whereas the thin line illustrates isotype control staining of the same cell population. hesion to recombinant adhesion molecules and endothe- ments examining the effect of NOD T-cell detachment on lial monolayers (16,17,46). recombinant VCAM-1 Ϯ SDF-1 coating showed that SDF-1 How then might SDF-1 stimulate chemorepulsion of NOD/ augments NOD T-cell resistance to shear-mediated detach- LtJ T-cell firm adhesion? One explanation could be related to ment compared with islet microvascular endothelial cell the doses of SDF-1 previously used because we report here monolayers (M. Hueng, C.G. Kevil, unpublished observa- for the first time that SDF-1 can prevent normal T-cell firm tions). Such a co-repressor function of the endothelium is adhesion at a concentration known to stimulate chemorepul- possible because the Slit-2/Robo system has been shown to sion of T-cell migration (800 ng/ml) (26). However, several counteract SDF-1/CXCR4 function and to regulate leukocyte lines of data support the hypothesis that SDF-1 may be chemotaxis (49,50). Future experiments are necessary to working through alternative receptor-signaling pathways determine whether RDC1/CXCR7 signaling or Slit/Robo in- besides CXCR4. First, CXCR4 protein expression is teractions participate in SDF-1–mediated chemorepulsion of significantly decreased in NOD/LtJ T-cells. Second, im- NOD/LtJ T-cell firm adhesion. muno-neutralization of CXCR4 partially attenuates The remaining question is how does SDF-1 mediate che- ␣ SDF-1 chemorepulsion, whereas Gi -protein inhibition morepulsion of diabetogenic versus normal T-cell firm adhe- by pertussis toxin completely reverses chemorepulsion. sion? We have presented data demonstrating that SDF-1 As such, SDF-1 signaling responses between normal and treatment renders diabetogenic T-cells vulnerable to shear- diabetogenic T-cell populations appear to have a com- mediated detachment that is due to changes in integrin ␣ mon point of origin, Gi -protein, but show divergent affinity and avidity. This result suggests that integrin affinity activation of downstream Rap1 pathways. This alterna- or avidity is likely to be constitutively greater on diabeto- tive receptor could be RDC1/CXCR7, which avidly binds genic versus normal T-cells. Support for this idea is strong SDF-1 and signals through a G-protein–coupled recep- because the absolute numbers of adherent diabetogenic tor pathway (47,48). Thus, it is possible that down- T-cells was greater on either unstimulated or TNF-␣–stimu- stream SDF-1 signaling pathways mediated by CXCR4 lated islet endothelial cells compared with normal T-cells. versus RDC1/CXCR7 may be different between diabeto- These findings are striking for two reasons. First, these data genic versus normal T-cells and that the activity be- demonstrate that NOD/LtJ T-cell surface integrins are acti- tween the two may differentially control for vated to a greater degree, thereby increasing the likelihood chemoattraction versus chemorepulsion. It is also pos- that these cells will readily adhere and migrate into islets. sible that a co-repressor signal from the endothelium Second, our data identify a previously unknown response may also be involved. Interestingly, preliminary experi- that integrin affinity and avidity is reversible by SDF-1. Figure

110 DIABETES, VOL. 57, JANUARY 2008 C.D. SHARP AND ASSOCIATES

FIG. 7. SDF-1 treatment alters NOD/LtJ CD3 T-cell integrin activity. A: Resistance of NOD/LtJ CD3 T-cells to shear-mediated detachment. The black, red, blue, and green lines indicate control, TNF-␣–stimulated, TNF-␣–stimulated plus 100 ng/ml SDF-1–treated, and TNF-␣–stimulated plus 100 ng/ml SDF-1–treated with pertussis toxin–pretreated CD3 T-cells, respectively. B: Absolute numbers of firmly adherent C57BL/6J and NOD/LtJ CD3 T-cells to unstimulated or TNF-␣–stimulated islet endothelial monolayers. C: Our working hypothesis in which integrin activation is constitutively greater in NOD/LtJ T-cells, which can be reversed by SDF-1 treatment in a pertussis toxin–dependent manner. It is possible that other factors could influence integrin activation in NOD T-cells, such as differences in binding pocket structure or subunit assembly, which are not portrayed in this diagram. *P < 0.05 NOD/LtJ vs. C57BL/6J.

7C illustrates our working hypothesis in which integrin normal and inflammatory states is constantly evolving activation is constitutively greater in NOD/LtJ T-cells, which with the identification of novel functions and new mole- can be reversed by SDF-1 treatment. It is possible that other cules (47,48,51). Future studies are clearly needed to factors could influence integrin activation, such as differ- better understand the unique findings above and how ences in binding pocket structure or subunit assembly, which chemokines may differentially regulate immune cell traf- cannot be dismissed. However, specific questions that need ficking during autoimmune diabetes. to be answered in the future are 1) whether SDF-1 treatment specifically affects integrin affinity or avidity or both and 2) ACKNOWLEDGMENTS which of the NOD/LtJ T-cell integrin proteins are primarily C.G.K. has received American Diabetes Association Award affected by SDF-1 treatment. 1-05-JF-26. In conclusion, we provide novel evidence that SDF-1 exerts a diverse response in regulating diabetogenic im- REFERENCES mune cell recruitment versus nonautoimmune cells. Here, 1. Kawasaki E, Abiru N, Eguchi K: Prevention of type 1 diabetes: from the we show that diabetogenic T-cell integrin activity is con- view point of beta cell damage. Diabetes Res Clin Pract 66 (Suppl. stitutively elevated in NOD/LtJ T-cells, which can be 1):S27–S32, 2004 inactivated by SDF-1 that could serve to diminish the 2. Roep BO: T-cell responses to autoantigens in IDDM: the search for the autoimmune cell recruitment. This may appear counterin- Holy Grail. Diabetes 45:1147–1156, 1996 tuitive because several reports suggest that several che- 3. Roep BO, Heidenthal E, de Vries RR, Kolb H, Martin S: Soluble forms of mokines and their receptors (e.g., IP-10, RANTES, CXCR3, intercellular adhesion molecule-1 in insulin-dependent diabetes mellitus. Lancet 343:1590–1593, 1994 CCR7, CCR5, etc.) may contribute to the disease process 4. Barlow SC, Langston W, Matthews KM, Chidlow JH Jr, Kevil CG: CD18 of type 1 diabetes (41,42). However, our understanding of deficiency protects against multiple low-dose streptozotocin-induced dia- chemokine regulation of immune cell trafficking during betes. Am J Pathol 165:1849–1852, 2004

DIABETES, VOL. 57, JANUARY 2008 111 CHEMOREPULSION OF DIABETOGENIC T-CELL ADHESION

5. Hanninen A, Salmi M, Simell O, Jalkanen S: Endothelial cell-binding 29. Grabovsky V, Dwir O, Alon R: Endothelial chemokines destabilize L- properties of infiltrated into human diabetic pancreas: impli- selectin-mediated rolling without inducing selectin shedding. cations for pathogenesis of IDDM. Diabetes 42:1656–1662, 1993 J Biol Chem 277:20640–20650, 2002 6. Yang XD, Michie SA, Mebius RE, Tisch R, Weissman I, McDevitt HO: The 30. Ni N, Kevil CG, Bullard DC, Kucik DF: Avidity modulation activates role of cell adhesion molecules in the development of IDDM: implications adhesion under flow and requires cooperativity among adhesion receptors. for pathogenesis and therapy. Diabetes 45:705–710, 1996 Biophys J 85:4122–4133, 2003 7. Campbell DJ, Kim CH, Butcher EC: Chemokines in the systemic organiza- 31. McLeod SJ, Shum AJ, Lee RL, Takei F, Gold MR: The Rap regulate tion of immunity. Immunol Rev 195:58–71, 2003 integrin-mediated adhesion, cell spreading, actin polymerization, and Pyk2 8. Godessart N, Kunkel SL: Chemokines in autoimmune disease. Curr Opin tyrosine phosphorylation in B lymphocytes. J Biol Chem 279:12009–12019, Immunol 13:670–675, 2001 2004 9. Simon SI, Green CE: Molecular mechanics and dynamics of leukocyte 32. Tohyama Y, Katagiri K, Pardi R, Lu C, Springer TA, Kinashi T: The critical recruitment during inflammation. Annu Rev Biomed Eng 7:151–185, 2005 cytoplasmic regions of the alphaL/beta2 integrin in Rap1-induced adhesion 10. Kubes P: The complexities of leukocyte recruitment. Semin Immunol and migration. Mol Biol Cell 14:2570–2582, 2003 14:65–72, 2002 33. Katagiri K, Shimonaka M, Kinashi T: Rap1-mediated lymphocyte function- 11. Kevil CG, Bullard DC: Roles of leukocyte/endothelial cell adhesion mole- associated antigen-1 activation by the T cell antigen receptor is dependent Am J Med cules in the pathogenesis of vasculitis. 106:677–687, 1999 on -gamma 1. J Biol Chem 279:11875–11881, 2004 12. Rot A, von Andrian UH: Chemokines in innate and adaptive host defense: 34. Kevil CG: Endothelial cell activation in inflammation: lessons from mutant basic chemokinese grammar for immune cells. Annu Rev Immunol mouse models. Pathophysiology 9:63–74, 2003 22:891–928, 2004 35. Springer TA: Traffic signals for lymphocyte recirculation and leukocyte 13. Middleton J, Neil S, Wintle J, Clark-Lewis I, Moore H, Lam C, Auer M, Hub emigration: the multistep paradigm. Cell 76:301–314, 1994 E, Rot A: Transcytosis and surface presentation of IL-8 by venular 36. Vicente-Manzanares M, Montoya MC, Mellado M, Frade JM, del Pozo MA, endothelial cells. Cell 91:385–395, 1997 Nieto M, de Landazuri MO, Martinez AC, Sanchez-Madrid F: The chemo- 14. Laudanna C, Kim JY, Constantin G, Butcher E: Rapid leukocyte integrin activation by chemokines. Immunol Rev 186:37–46, 2002 kine SDF-1alpha triggers a chemotactic response and induces cell polar- 15. Peled A, Grabovsky V, Habler L, Sandbank J, Arenzana-Seisdedos F, Petit ization in human B lymphocytes. Eur J Immunol 28:2197–2207, 1998 I, Ben-Hur H, Lapidot T, Alon R: The chemokine SDF-1 stimulates 37. Shigihara T, Shimada A, Yamada S, Maruyama T, Hirose H, Saruta T: integrin-mediated arrest of CD34(ϩ) cells on vascular endothelium under Stromal cell-derived factor-1 chemokine gene polymorphism is not asso- shear flow. J Clin Invest 104:1199–1211, 1999 ciated with onset age of Japanese type 1 diabetes. Ann N Y Acad Sci 16. Campbell JJ, Hedrick J, Zlotnik A, Siani MA, Thompson DA, Butcher EC: 1005:328–331, 2003 Chemokines and the arrest of lymphocytes rolling under flow conditions. 38. Kawasaki E, Ide A, Abiru N, Kobayashi M, Fukushima T, Kuwahara H, Kita Science 279:381–384, 1998 A, Uotani S, Yamasaki H, Eguchi K: Stromal cell-derived factor-1 chemo- 17. Constantin G, Majeed M, Giagulli C, Piccio L, Kim JY, Butcher EC, kine gene variant in patients with type 1 diabetes and autoimmune thyroid Laudanna C: Chemokines trigger immediate beta2 integrin affinity and disease. Ann N Y Acad Sci 1037:79–83, 2004 mobility changes: differential regulation and roles in lymphocyte arrest 39. Kimura R, Nishioka T, Soemantri A, Ishida T: Allele-specific transcript under flow. Immunity 13:759–769, 2000 quantification detects haplotypic variation in the levels of the SDF-1 18. Ide A, Kawasaki E, Abiru N, Sun F, Fukushima T, Takahashi R, Kuwahara transcripts. Hum Mol Genet 14:1579–1585, 2005 H, Fujita N, Kita A, Oshima K, Uotani S, Yamasaki H, Yamaguchi Y, 40. Aboumrad E, Madec AM, Thivolet C: The CXCR4/CXCL12 (SDF-1) signal- Kawabata Y, Fujisawa T, Ikegami H, Eguchi K: Stromal-cell derived ling pathway protects non-obese diabetic mouse from autoimmune diabe- factor-1 chemokine gene variant is associated with type 1 diabetes age at tes. Clin Exp Immunol 148:432–439, 2007 onset in Japanese population. Hum Immunol 64:973–978, 2003 41. Hanifi-Moghaddam P, Kappler S, Seissler J, Muller-Scholze S, Martin S, 19. Dubois-Laforgue D, Hendel H, Caillat-Zucman S, Zagury JF, Winkler C, Roep B, Strassburger K, Kolb H, Schloot N: Altered chemokine levels in Boitard C, Timsit J: A common stromal cell–derived factor-1 chemokine individuals at risk of type 1 diabetes mellitus. Diabet Med 23:156–163, 2005 gene variant is associated with the early onset of type 1 diabetes. Diabetes 42. Meagher C, Sharif S, Hussain S, Cameron M, Arreaza G, Delovitch T: 50:1211–1213, 2001 Cytokines and chemokines in the pathogenesis of murine type 1 diabetes. 20. Huang M, Matthews K, Siahaan TJ, Kevil CG: Alpha L-integrin I domain In Cytokines and Chemokines in Autoimmune Disease. Santamaria P, Ed. cyclic peptide antagonist selectively inhibits T cell adhesion to pancreatic New York, Plenum Publishers, 2003, p. 133–158 islet microvascular endothelium. Am J Physiol Gastrointest Liver Physiol 43. Christen U, Von Herrath MG: IP-10 and type 1 diabetes: a question of time 288:G67–G73, 2005 and location. Autoimmunity 37:273–282, 2004 21. Pavlick KP, Ostanin DV, Furr KL, Laroux FS, Brown CM, Gray L, Kevil CG, 44. Shimada A, Morimoto J, Kodama K, Suzuki R, Oikawa Y, Funae O, Kasuga Grisham MB: Role of T-cell-associated lymphocyte function-associated A, Saruta T, Narumi S: Elevated serum IP-10 levels observed in type 1 antigen-1 in the pathogenesis of experimental colitis. Int Immunol 18:389– diabetes. Diabetes Care 24:510–515, 2001 398, 2006 45. Zhernakova A, Alizadeh BZ, Eerligh P, Hanifi-Moghaddam P, Schloot NC, 22. Kevil CG, Chidlow JH, Bullard DC, Kucik DF: High-temporal-resolution Diosdado B, Wijmenga C, Roep BO, Koeleman BP: Genetic variants of analysis demonstrates that ICAM-1 stabilizes WEHI 274.1 monocytic cell RANTES are associated with serum RANTES level and protection for type rolling on endothelium. Am J Physiol Cell Physiol 285:C112–C118, 2003 1 diabetes. Genes Immun 7:544–549, 2006 23. Langston W, Chidlow JH Jr, Booth BA, Barlow SC, Lefer DJ, Patel RP, Kevil 46. Grabovsky V, Feigelson S, Chen C, Bleijs DA, Peled A, Cinamon G, Baleux CG: Regulation of endothelial glutathione by ICAM-1 governs VEGF-A- F, Arenzana-Seisdedos F, Lapidot T, van Kooyk Y, Lobb RR, Alon R: mediated eNOS activity and angiogenesis. Free Radic Biol Med 42:720– Subsecond induction of alpha4 integrin clustering by immobilized chemo- 729, 2007 kines stimulates leukocyte tethering and rolling on endothelial vascular cell 24. Goebel S, Huang M, Davis WC, Jennings M, Siahaan TJ, Alexander JS, Kevil adhesion molecule 1 under flow conditions. J Exp Med 192:495–506, 2000 CG: VEGF-A stimulation of leukocyte adhesion to colonic microvascular 47. Infantino S, Moepps B, Thelen M: Expression and regulation of the orphan endothelium: implications for inflammatory bowel disease. Am J Physiol receptor RDC1 and its putative ligand in human dendritic and B cells. Gastrointest Liver Physiol 290:G648–G654, 2006 J Immunol 176:2197–2207, 2006 25. Chan JR, Hyduk SJ, Cybulsky MI: Chemoattractants induce a rapid and 48. Balabanian K, Lagane B, Infantino S, Chow KY, Harriague J, Moepps B, transient upregulation of alpha4 integrin affinity for vascular cell Arenzana-Seisdedos F, Thelen M, Bachelerie F: The chemokine SDF-1/ adhesion molecule 1 which mediates arrest: an early step in the process of CXCL12 binds to and signals through the orphan receptor RDC1 in T emigration. J Exp Med 193:1149–1158, 2001 lymphocytes. J Biol Chem 280:35760–35766, 2005 26. Poznansky MC, Olszak IT, Foxall R, Evans RH, Luster AD, Scadden DT: 49. Wu JY, Feng L, Park HT, Havlioglu N, Wen L, Tang H, Bacon KB, Jiang Z, Active movement of T cells away from a chemokine. Nat Med 6:543–548, Zhang X, Rao Y: The neuronal repellent Slit inhibits leukocyte chemotaxis 2000 induced by chemotactic factors. Nature 410:948–952, 2001 27. Murdoch C: CXCR4: chemokine receptor extraordinaire. Immunol Rev 50. Guan H, Zu G, Xie Y, Tang H, Johnson M, Xu X, Kevil C, Xiong WC, Elmets 177:175–184, 2000 C, Rao Y, Wu JY, Xu H: Neuronal repellent Slit2 inhibits 28. Peled A, Kollet O, Ponomaryov T, Petit I, Franitza S, Grabovsky V, Slav migration and the development of immune responses. J Immunol 171: MM, Nagler A, Lider O, Alon R, Zipori D, Lapidot T: The chemokine SDF-1 6519–6526, 2003 activates the integrins LFA-1, VLA-4, and VLA-5 on immature human 51. Heinzel K, Benz C, Bleul CC: A silent chemokine receptor regulates CD34(ϩ) cells: role in transendothelial/stromal migration and engraftment steady-state leukocyte homing in vivo. Proc Natl Acad SciUSA104:8421– of NOD/SCID mice. Blood 95:3289–3296, 2000 8426, 2007

112 DIABETES, VOL. 57, JANUARY 2008