Biased S1PR1 Signaling in B Cells Subverts Responses to Homeostatic Chemokines, Severely Disorganizing Lymphoid Organ Architecture This information is current as of October 3, 2021. Il-Young Hwang, Chung Park, Kathleen Harrison and John H. Kehrl J Immunol published online 23 September 2019 http://www.jimmunol.org/content/early/2019/09/20/jimmun

ol.1900678 Downloaded from

Supplementary http://www.jimmunol.org/content/suppl/2019/09/20/jimmunol.190067 Material 8.DCSupplemental http://www.jimmunol.org/

Why The JI? Submit online.

• Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication by guest on October 3, 2021

*average

Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Published September 23, 2019, doi:10.4049/jimmunol.1900678 The Journal of Immunology

Biased S1PR1 Signaling in B Cells Subverts Responses to Homeostatic Chemokines, Severely Disorganizing Lymphoid Organ Architecture

Il-Young Hwang, Chung Park, Kathleen Harrison, and John H. Kehrl

Ligand-engaged chemoattractant receptors trigger Gai subunit nucleotide exchange, stimulating the activation of downstream effector molecules. Activated chemoattractant receptors also dock G –coupled receptor kinases (GRKs) that help mediate receptor desensitization. In this study, we show that the B cell–specific loss of GRK2 severely disrupts B cell trafficking and immune cell homeostasis. The GRK2 deficiency in developing murine B cells leads to a severe immune phenotype, including a major reduction of bone marrow IgD+ cells, splenomegaly with a loss of white pulp and grossly expanded red pulp, a deficit of Peyer patches, and small lymph nodes with marked reductions in B cell numbers. The major phenotypes in these mice arise from

excessive S1PR1 signaling combined with inadequate homeostatic signaling. CXCL13 signaling is the most Downloaded from severely compromised. In B cells, our data also indicate that S1PR1 signals constitutively, as blocking S1PR1 signaling with an S1PR1 antagonist enhanced CXCL13-triggered wild-type B cell migration. Furthermore, blocking S1PR1 signaling in the GRK2-deficient B cells partially corrected their poor response to chemokines. Treating mice lacking GRK2 expression in their B cells with an S1PR1 antagonist partially normalized B cell trafficking into and splenic follicles. These findings

reveal the critical interdependence of Gai-linked signaling pathways in controlling B trafficking. The Journal of

Immunology, 2019, 203: 000–000. http://www.jimmunol.org/

n array of chemoattractant receptors modulates the Activated G protein–coupled receptors (GPCRs) also recruit positioning and trafficking of B (1–3). GPCR kinases (GRKs). They are proposed to dock the 20-aa A Each of these receptors use Gai to activate N-terminal a helix of GRKs much like they bind the C-terminal downstream effector molecules (4, 5). Mice whose lymphocytes a5 helix of the Ga subunit of GDP-bound heterotrimeric G pro- lack Gai2 and Gai3 proteins or that possess Gai proteins un- teins (8). Consequently, GRKs and heterotrimeric G proteins able to interact with regulator of G protein–signaling proteins compete for the same site on the activated receptor. Typically, G exhibit severe defects in chemoattractant receptor signaling protein binding precludes GRK binding and vice versa. However, by guest on October 3, 2021 and in the organization of B cell compartments (5, 6). By specific receptor conformations may favor the binding of one adopting an active conformation , engaged receptors protein versus the other. Structural studies indicate that activated recruit heterotrimeric G proteins (7). The activated receptor GPCRs adopt a dynamic conformational landscape rather a sin- opens a cleft between the helical and ras domains in the Gai gle “active” conformation (7). Based on the structural analysis subunit, which facilitates GDP release, allowing GTP to bind. of GRK5/b2–adrenergic receptors interactions, receptor docking The GTP-bound Gai and -freed Gbg subunits can then stimulate causes a GRK conformation change reorienting the catalytic do- the activity of downstream effector molecules. Precise regulation main increasing catalytic activity (9). This contrasts with most of Gbg– and GTP–Gai signaling coordinate an assortment of other protein kinases, whose catalytic activity depends upon a post- signaling pathways that enable B cell integrin activation and di- translational modification such as phosphorylation. Once recruited rected migration. and activated, GRKs phosphorylate GPCRs, typically on their C-terminal serine/threonine residues. An activated, phosphory- lated receptor becomes a target for b-arrestins, whose recruitment B Cell Molecular Immunology Section, Laboratory of Immunoregulation, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, leads to receptor desensitization and receptor endocytosis. Fol- MD 20892 lowing endocytosis, the receptor is degraded or recycles to the ORCIDs: 0000-0002-7819-5333 (C.P.); 0000-0002-6526-159X (J.H.K.). plasma membrane (10). Received for publication June 18, 2019. Accepted for publication August 29, 2019. The GRKs are the serine/threonine protein kinases most related This work was supported by an intramural program of the National Institute of to the AGC kinase subfamily (8). They have a central catalytic Allergy and Infectious Disease. domain located within a regulator of G protein–signaling homology I.-Y.H. designed and performed the experiments, helped analyze the data; C.P. did the (RH) domain, which is flanked by a short N-terminal a-helical intravital imaging and confocal microscopy; K.H. managed the mice and helped with domain and a variable C-terminal lipid-binding region. Among the the experiments; J.H.K. conceived the experiments, helped in data analysis, and wrote the manuscript. seven mammalian GRKs, Grk2 and Grk6 are most prominently Address correspondence and reprint requests to Dr. John H. Kehrl, Laboratory of expressed in lymphocytes (http://www.immgen.org/databrowser/ Immunoregulation, National Institute of Allergy and Infectious Diseases, National index.html). Linking heterotrimeric G protein signaling to GRK2 Institutes of Health, Building 10, Room 6A01, 10 Center Drive, MSC 1876, regulation, the C-terminal lipid-binding domain in GRK2 (PH Bethesda, MD 20892. E-mail address: [email protected] domain) allows Gbg subunits to recruit GRK2 to the plasma mem- The online version of this article contains supplemental material. brane. In contrast, GRK6 undergoes C-terminal palmitoylation to Abbreviations used in this article: FO, follicular; GPCR, G protein–coupled receptor; GRK, GPCR kinase; HEV, high endothelial venule; KO, knockout; LN, lymph node; mediate membrane localization (11, 12). A limited immune pheno- MZ, marginal zone; S1P, sphingosine-1-phosphate; T, transitional-type; WT, wild-type. typing of Grk6-deficient mice revealed normal B cell chemotaxis to

www.jimmunol.org/cgi/doi/10.4049/jimmunol.1900678 2 GRK2-DEFICIENT B CELLS

CXCL12 but reduced transendothelial migration (13). Whereas Beads (Thermo Fisher Scientific) individually stained with each fluorochrome. GRK2 deficiency causes embryonic lethality, an analysis of mice Compensation matrices were calculated with FACSDiva software. Data with conditional deletion of Grk2 in B or T lymphocytes has been acquisition was done on FACSCanto II (BD Biosciences) flow cytometer and analyzed with FlowJo software version 9.9.6 (Treestar). reported (14). Follicular (FO) B cells from these mice resisted sphingosine-1-phospate (S1P) R1 desensitization, migrated better S1PR1 immunostaining, internalization, and recovery to S1P in standard chemotaxis assays, but entered lymph nodes For the S1PR1 immunostaining, cells were fixed in 2% paraformaldehyde (LNs) poorly. The S1PR1 receptors on marginal zone (MZ) B cells for 10 min, washed, and stained with S1PR1-PE, which is directed against also resisted desensitization, which impaired MZ B cell shuttling. an external S1PR1 epitope (R & D Systems). To assess S1PR1 internali- In contrast, CXCR4 and CXCR5 signaling was reported as not zation, the splenocytes were rested in DMEM/10 mM HEPES for 30 min at 37˚C/5% CO2 before stimulation with 1 mM S1P or solvent (DMSO) for significantly altered. This study implicated GRK2 as a central 10 or 30 min. Cells were stained with S1PR1-PE prior to stimulation and at regulator of S1PR1 desensitization. the indicated time points. To examine S1PR1 recovery, cells were incu- Building on these results, we have examined in greater detail bated for 10, 30, or 60 min in DMEM/10 mM HEPES at 37˚C/5% CO2. the origins of the phenotypes when B cells lack GRK2. We Cells were washed and stained with S1PR1-PE prior to incubation and at confirmed the finding of impaired S1PR1 desensitization but the indicated times. have also found severe defects in B cell responses to homeostatic Intracellular flow cytometry chemokines. At least some of these defects result from the Purified B cells were rested in DMEM containing 1% charcoal-stripped dysregulated S1PR1 signaling. Together, they led to defective FCS, antibiotics (100 IU/ml penicillin and 100 mg/ml streptomycin), B cell physiology and, surprisingly, abnormalities in immune 1 mM sodium pyruvate, and 50 mM 2-ME for 30 min at 37˚C/5% CO2 homeostasis. These include impaired bone marrow and splenic before stimulation for varying durations with 1 mg/ml CXCL13 (R&D B cell development, pronounced splenomegaly with a marked Systems or PeproTech) or 1 mM S1P (Sigma-Aldrich or Cayman Chem- Downloaded from ical) for 2, 5, 10, 30, or 60 min. The labeling of dead cells, fixation, and disruption of the splenic architecture, a loss of Peyer patches, permeabilization were performed as described in the manufacturer’s pro- reduced LN homing because of transmigration defects, small lym- tocol. Following permeabilization, the cells were immunostained with anti- phoid organ B cell follicles, and accelerated B cell LN and bone B220, anti-IgD, anti-IgM, CD21, CD23, and CD93 for 30 min at 4˚C. To marrow egress. detect phosphorylated signaling molecules Abs against phospho-Akt (Ser473) Alexa Fluor 647 (D9E), phospho-Erk (Thr202/Tyr204) (197G) 567 564

Alexa Fluor 647, and phospho–ezrin (Thr )/radixin (Thr )/moesin http://www.jimmunol.org/ Materials and Methods 558 Mice and bone marrow reconstitutions (Thr ) (pERM), all from Cell Signaling Technology, were used. Isotype control staining was performed using rabbit IgG isotype mAb Alexa Fluor fl/fl a b C57BL/6, C57BL/6 Grk2 , and B6.SJL-Ptprc Pepc /BoyJ (CD45.1) mice 647 (DA1E; Cell Signaling Technology). Secondary F(ab9)2 fragment of were obtained from The Jackson Laboratory. The C57/BL6 mb1-cre mice goat anti-rabbit IgG (H+L) Alexa Fluor 647 (Thermo Fisher Scientific) was were kindly provided by Dr. M. Reth and used to breed with the Grk2fl/fl used to detect the pERM Abs. After washing, the cells were resuspended in mice to generate the Grk2fl/flmb1-cre mice. For the bone marrow recon- 250 ml of 1% BSA/PBS and filtered prior to acquisition on a FACSCanto II stitutions, 6-wk-old CD45.1 mice were twice irradiated with 550 rad for flow cytometer (BD Biosciences). total of 1100 rad. Mixed chimeric mice were made by reconstituting the irradiated CD45.1 mice with a 1:1 mixture of bone marrow prepared from Chemotaxis assays fl/fl C57BL/6 CD45.1 wild-type (WT) mice and either CD45.2 Grk2 mice or Chemotaxis assays were performed using a transwell chamber (Costar). fl/fl by guest on October 3, 2021 CD45.2 Grk2 mb1-cre mice. The engraftment was monitored by sam- Splenic B cells were immunostained for B cell subsets with fluorochrome- pling the blood 28 d later. The mice were used 6–8 wk after reconstitution. conjugated Abs against B220, CD21, CD23, CD93, anti-IgM, and anti-IgD; All mice used in this study were 6–14 wk of age. Mice were housed under they were washed twice, resuspended in complete RPMI 1640 media using specific pathogen–free conditions. All the animal experiments and proto- charcoal-stripped FCS, incubated for 1 h at 37˚C, and added to the upper cols used in the study were approved by the National Institute of Allergy wells of a 24-well transwell plate containing a 5-mm insert. In some ex- and Infectious Diseases’ Animal Care and Use Committee at the National periments, the indicated concentrations of S1P (Sigma-Aldrich or Cayman Institutes of Health. Chemical) or chemokine (R & D Systems or PeproTech) were added to the Cells upper well. The lower wells contained various concentrations of CCL19, CXCL12, CXCL13, or S1P in 600 ml of media. The numbers of cells that and LNs and were removed and gently dissociated into single-cell migrated to the lower well after 3-h incubation were counted using a suspensions. Bone marrow cells were collected by flushing isolated femurs MACSQuant flow cytometer (Miltenyi Biotec). The percentage of cells with PBS. Peripheral blood samples were collected by retro-orbital eye that migrated in the absence of chemoattractant was designated as non- bleeding. After removing RBCs with Tris–NH4Cl, the cells were resus- specific migration. The specific migration was calculated by subtracting pended in PBS containing 1% fatty acid–free BSA at 4˚C. B cells were the nonspecific migration from the migration in the presence of chemo- isolated by negative depletion using biotinylated Abs to CD4, CD8, Gr-1 attractant. The percentage migration was calculated as the numbers of cells (Ly-6C and Ly-6G), and CD11c and Dynabeads M-280 Streptavidin that migrated into the lower part of the chamber divided by the total (Thermo Fisher Scientific). The B cell purities were .95%. When needed, number of cells in the starting cell suspension and multiplying the result by splenic, lymph, bone marrow, or B cells were cultured in RPMI 1640 100. For those assays using Ex 26 (Tocris Bioscience), the compound containing 10% charcoal-stripped FCS (Life Technologies), 2 mM L-glutamine, (1 mM) was added during the 1-h preincubation phase and present in antibiotics (100 IU/ml penicillin and 100 mg/ml streptomycin), 1 mM the both chambers of the chemotaxis chamber. DMSO alone served as sodium pyruvate, and 50 mM 2-ME. the control. Standard flow cytometry Intracellular calcium measurements Single cells were resuspended in PBS, 2% FCS, and strained with Splenic B220+ cells were seeded at 3 3 105 cells per 100-ml loading fluorochrome-conjugated or biotinylated Abs against B220 (RA3-6B2), medium (RPMI 1640, 10% charcoal-stripped FCS) into poly-D-lysine– CD19 (1D3), CD23 (B3B4), CD21/35 (4E3), CD93 (AA4.1), CD43 (S7), coated 96-well black wall, clear-bottom microtiter plates (Nalgene IgD (11-26c-2a), IgM (R6-60.2), CD24 (M1/69), BP-1 (6C3), CD3 (145-2C11), Nunc). An equal volume of assay loading buffer (FLIPR Calcium 4 CD4 (GK1.5 or RM4-5), CD8 (53-6.7), CD11c (HL3), CD11b (M1/70), CD184 Assay Kit; Molecular Devices) in HBSS supplemented with 20 mM (CXCR4, 2B11), CCR7 (4B12), CXCR5 (2G8), CD11a (M17/4), CD49d HEPES and 2 mM probenecid was added. Cells were incubated for 1 h (9C10, MFR4.B), CD54 (3E2), CD62L (MEL-16), NK1.1 (PK136), TCRgd at 37˚C before adding chemokine or S1P, and then the calcium flux (GL3), Ly-6G (1A8), Ly-6C (AL-21), CD45.1 (A20), or CD45.2 (104) (all peak was measured using a FlexStation 3 (Molecular Devices). For from eBioscience, BioLegend, or BD Pharmingen). Biotin-labeled Abs those assays using Ex 26, the compound was added during the 1-h were visualized with fluorochrome-conjugated streptavidin (eBioscience). preincubation and added to the upper and lower chamber of the che- A LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Thermo Fisher Sci- motaxis chamber. The data were analyzed with SoftMax Pro 5.2 entific) was used in all experiments to exclude dead cells. Compensation (Molecular Devices). Data are shown as fluorescent counts, and the was performed using CompBeads and ArC Amine Reactive Compensation y-axisislabeledasLm1. The Journal of Immunology 3

Confocal microscopy and immunohistochemistry parameters: autoregressive motion algorithm, estimated diameter 10 mm, background subtraction true, maximum distance 20 mm, and maximum Immunohistochemistry was performed as described previously (15). gap size 3. Tracks acquired that could be tracked for at least 20% of total Briefly, freshly isolated LNs or spleens were fixed in newly prepared 4% imaging duration were used for analysis. Some tracks were manually ex- paraformaldehyde (Electron Microscopy Sciences) overnight at 4˚C on an amined and verified. Calculations of the cell motility parameters (speed, agitation stage. Spleens or LNs were embedded in 4% low-melting agarose track length, displacement, straightness, and speed variability) were (Invitrogen) in PBS and sectioned with a vibratome (Leica VT1000 S) at a performed using the Imaris software v.9.0.1 x64 (Bitplane AG). Statis- 30-mm thickness. Thick sections were blocked in PBS containing 10% tical analysis was performed using Prism software (GraphPad Software). FCS, 1 mg/ml anti-Fcg receptor (BD Biosciences), and 0.1% Triton X-100 Annotations on videos and video editing were performed using Adobe (Sigma-Aldrich) for 30 min at room temperature. Sections were stained Premiere Pro CS3 (Adobe Systems). Video files were converted to overnight at 4˚C on an agitation stage with Abs directed at B220 MPEG4 format with ImTOO Video Converter Ultimate (v. 7.8.19) (ImTOO (RA3-6B2; BD Biosciences), CD4 (RM4-5; BD Biosciences), CD169 Software Studio). (3D6.112), CD21/35 (4E3), CD45.1 (A20), CD45.2 (104) (all from Bio- Legend), Ki67 (SolA15; Thermo Fisher Scientific), and LYVE-1 (223322; Statistical analysis R&D Systems). Stained thick sections were microscopically analyzed using a Leica SP8 confocal microscope equipped with an HC PL APO CS2 In vivo results represent samples from 2 to 10 mice per experimental 403 (numerical aperture, 1.30) oil objective (Leica Microsystems), and group. Results represent mean values of at least triplicate samples. SEM images were processed with Leica LAS AF software (Leica Microsystems) and p values were calculated with a t test or ANOVA using GraphPad and Imaris software v.9.0.1 x64 (Bitplane AG). The intensities of fluorescent Prism (GraphPad Software). signals in regions of interests were measured by LSA AF Lite software (Leica Microsystems). Results Homing, egress, and competition assays Loss of Fr. F cells from the bone marrow B cell compartment in

fl/fl Downloaded from Purified splenic B cells from Grk2fl/fl and Grk2fl/flmb1-cre B cells mice Grk2 mb1-cre mice were labeled with 1 mM eFluor 450 or 2.5 mM CMTMR (Thermo Fisher Based on Grk2 mRNA expression profiling, bone marrow B cells Scientific) for 15 min at 37˚C, and equal numbers of viable cells (8–10 prominently express Grk2 (http://www.immgen.org/databrowser/ million) were injected i.v. into recipient mice. After 2 h, , inguinal LNs, and mesenteric LNs were removed and gently dissociated into single- index.html). To determine the role of GRK2 in the mouse bone fl/fl cell suspensions. Data acquisition was done on a FACSCanto II flow marrow B cell compartment, we crossed Grk2 mice to mb1-cre cytometer and analyzed with FlowJo software (Tree Star). The egress assay mice. Based on the known expression pattern of mb1 and the used a similar initial approach, but 2 h after the cell transfer, mice were previous functional assessment of deleted with mb1-cre, the http://www.jimmunol.org/ injected i.v. with CD62L Ab (100 mg/mouse). After 18 h, spleen, inguinal loss of GRK2 should become evident in Fr. B cells in the Hardy LNs, and popliteal LNs were removed and gently dissociated into single- cell suspensions. Peripheral blood was collected by retro-orbital eye scheme of B cell development, in which Fr. A-C cells correspond bleeding. After removing RBCs with Tris–NH4Cl, the cells were resus- to pro-B cell stages, Fr. D cells correspond to pre-B cells, Fr. E pended in PBS containing 1% BSA at 4˚C. The ratios between the numbers cells correspond to immature B cells, and Fr. F cells correspond to of cells in the various compartments 2 h after transfer versus after CD62L mature recirculating cells (5, 16). Assessing the B cell compart- Ab injection were calculated. The competition assays were performed by fl/fl adoptively transferring 20 million CMTMR-labeled Grk2fl/fl or Grk2fl/flmb1-cre ment in bone marrow from control and Grk2 mb1-cre mice B cells to separate WT mice. One hour later each mouse was injected with revealed similar overall numbers of B cells; however, we noted a 10 million eFluor 450–labeled Grk2fl/fl B cells. Two hours later, multiple 50% increase in Fr. D cells and, strikingly, a 4–5-fold reduction in by guest on October 3, 2021 LNs were collected, and cell suspensions were made from each of them. Fr. F cells in bone marrow prepared from the Grk2fl/flmb1-cre The number of labeled cells in each LN cell suspension was assessed by mice (Fig. 1A, 1B). A representative flow pattern of IgM versus flow cytometry. IgD on bone marrow cells gated on cells expressing B220 and Intravital two-photon laser-scanning microscopy CD19 but not CD43 is shown (Fig. 1C). Mature IgD+ B cells (Fr. F Inguinal LNs were prepared for intravital microscopy as described previ- cells) occupy an anatomically distinct bone marrow niche and can ously (15). Cell populations were labeled for 10 min at 37˚C with functionally generate specific IgM Abs to blood-borne pathogens 2.5–5 mM CellTracker Red CMTMR (Thermo Fisher Scientific) or 2 mM (17, 18). Their loss from the Grk2fl/flmb1-cre mouse bone marrow of eFluor 450 (Thermo Fisher Scientific). Five to ten million labeled cells likely results from a recruitment failure, a retention defect, or a of each population in 200 ml of PBS were adoptively transferred by tail major loss of recirculating IgD+ B cells. Among the chemoattractant vein injection into recipient mice. After anesthesia, the skin and fatty tissue over inguinal LN were removed. The mouse was placed in a prewarmed receptors, CXCR4 has the dominant role in the recruitment of coverglass chamber slide (Nalgene Nunc). The chamber slide was then blood-borne B cells and in the retention of B cells within spe- placed into the temperature control chamber on the microscope. The cialized bone marrow niches (19). To assess the impact on the loss 6 temperature of air was monitored and maintained at 37.0 0.5˚C. Inguinal of GRK2 on CXCR4-mediated migration, we first checked LN was intravitally imaged from the capsule over a range of depths CXCR4 expression and that of several cell-trafficking molecules on (10–220 mm). All two-photon imaging was performed with a Leica SP8 fl/fl inverted five-channel confocal microscope (Leica Microsystems) equipped bone marrow B cell fractions of control and Grk2 mb1-cre mice with 253 water-dipping objective, 0.95 numerical aperture (immersion (Fig. 1D). Although we noted some variation in CD11a, CD54, and medium used distilled water). Two-photon excitation was provided by a CD62L (L-selectin), none likely explained the loss of Fr. F cells or Mai Tai Ti:Sapphire laser (Spectra-Physics) with a 10-W pump, tuned the gain of Fr. D cells. The Grk2fl/flmb1-cre Fr. D-F cells expressed wavelength ranges from 820 to 920 nm. Emitted fluorescence was col- lected using a four-channel nondescanned detector. Wavelength separation 2-fold more CXCR4 cells, predicting an enhanced responsiveness to was through a dichroic mirror at 560 nm and then separated again through CXCL12. We also assessed S1PR1 expression because of its known a dichroic mirror at 495 nm followed by 525/50-nm emission filter for role in B cell bone marrow egress (20). Among the bone marrow Alexa Fluor 488 (Thermo Fisher Scientific); the eFluor 450 (Thermo fractions, Fr. F cells expressed the most S1PR1, but we did not Fisher Scientific), or second harmonic signal, was collected by 460/50-nm detect much difference between the cells from the two mouse emission filter, a dichroic mirror at 650 nm, followed by 610/60-nm emission filter for CMTMR or Alexa Fluor 594 (Thermo Fisher Scien- strains (Fig. 1D). Next, we tested the responsiveness of the different tific), and the Evans blue (Sigma-Aldrich). For four-dimensional analysis B cell bone marrow fractions to CXCL12 and S1P in a standard of cell behavior, stacks of various numbers of section (z step = 2, 3, 4, 5, or chemotaxis assay (Fig. 1E). The GRK2-deficient Fr. F cells had an 6 mm) were acquired every 2.5, 5, or 30 s to provide an imaging volume of increased percentage of spontaneously migrating cells, an enhanced 20–120 mm in depth. Sequences of image stacks were transformed into volume-rendered four-dimensional videos using Imaris software v.9.0.1 S1P migratory response to low concentrations of S1P. Based on x64 (Bitplane AG), and the tracks analysis was used for semiautomated their elevated CXCR4 expression, they had a suboptimal response tracking of cell motility in three dimensions by using the following to CXCL12. In addition, the Grk2-deficient Fr. B/C cells responded 4 GRK2-DEFICIENT B CELLS Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021

FIGURE 1. Increased Fr. D cells and loss of Fr. F cells in bone marrow of Grk22/2mb1-cre mice. (A) Flow cytometry results from analysis of bone marrow cells from Grk2fl/fl and Grk2fl/flmb1-cre mice. (B) Flow cytometry results from bone marrow showing CD43 expression and the percentage of Fr. A–Fr. F cells. (C) Flow cytometry of bone marrow cells gated on B220+CD19+CD432 examining IgD versus IgM. (D) Flow cytometry results showing receptor ex- pression and representative S1PR1 histograms for each fraction. (E) Spontaneous and directed migration of Grk2fl/fl and Grk2fl/flmb1-cre bone marrow B cell fractions to CXCL12 and S1P. All results are from B cells purified from a minimum of five Grk2fl/fl and Grk2fl/flmb1-cre mice. Data shown as mean 6 SEM. *p , 0.05, **p , 0.005, ***p , 0.0005. less well to CXCL12. Together, the enhanced motility, augmented the bone marrow via the bloodstream. They enter the marginal S1P migratory response, and suboptimal CXCR4 responses could sinus and red pulp of the spleen, where they are designated T0 cells interfere with the Fr. F cell retention and/or recruitment. (B220+CD93+IgM+CD232IgD2). These cells are believed to ac- quire surface IgD following their migration into the splenic white Splenomegaly and a severely disrupted splenic architecture in fl/fl pulp, becoming transitional-type (T) 1 cells, which can further the Grk2 mb1-cre mice differentiate into T2–FO (B220+CD93+IgM+IgD+CD23+) cells and The reduction in Fr. F cells could also result from a failure of B cell eventually FO B cells (21). To assess whether Grk2fl/flmb1-cre splenic maturation in the spleen. Immature B cells, which lack IgD, leave B cells had acquired surface IgD, we first compared the numbers The Journal of Immunology 5 and composition of splenocytes recovered from the Grk2fl/flmb1-cre because of their large spleens, the Grk2fl/flmb1-cre mice had an absolute mice versus those from control mice (Fig. 2A, 2B). The Grk2fl/flmb1-cre increase in FO B cells. Examining IgD versus IgM expression on the mice had markedly enlarged spleens with twice as many hema- B220+CD932 cells revealed that the Grk2fl/flmb1-cre mice had, on a topoietic cells, although they had an unremarkable distribution of percentage basis, fewer IgD+ splenic cells; however; their absolute major cellular subsets. Surprisingly, accompanying the splenic number also exceeded that of Grk2fl/fl mice (Fig. 2C). These data in- B cell expansion in the Grk2fl/flmb1-cre mice were similar in- dicate that a splenic B cell differentiation defect could not account for creases in CD4+ and CD8+ T lymphocytes. Immunostaining the the reduction in the mature IgD+ bone marrow cells. However, the large splenocytes for expression of B220/CD93/IgM/IgD/CD23/CD21 spleen and absolute increase in splenic B cells suggested impaired re- revealed that, on a percentage basis, the Grk2fl/flmb1-cre mice lease of recirculating B cells into the blood of the Grk2fl/flmb1-cre mice. had an increase in T1 cells and T2–FO cells, but a one-third re- Splenic FO B cells exit the white pulp via the MZ in an S1PR1- duction in FO B cells compared with controls (Fig. 2B). However, dependent fashion (22). Thereafter, they enter the red pulp prior to Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021

FIGURE 2. Splenomegaly with an expanded red pulp and disorganized white pulp in the Grk2fl/flmb1-cre mice. (A and B) Flow cytometry results from analysis of splenocytes from Grk2fl/fl and Grk2fl/flmb1-cre mice. (n = 5 versus 5). Data shown as mean 6 SEM. ***p , 0.0005. (C) Representative flow cytometry plots from the analysis of splenocytes gated as indicated for the expression of CD23 versus CD21 and IgD versus IgM. (D) Tiled confocal images of spleen sections. Scale bar, 200 mm. (E) Zoomed images from the boxed regions in (D) comparing the splenic white pulp of spleens from Grk2fl/fl and Grk2fl/flmb1-cre mice. Scale bar, 50 mm. Spleen sections of (D) and (E) were immunostained for CD4 (blue), CD21 [green, (D)], CD169 (orange), Ki67 (pink), and B220 [green, (E)]. 6 GRK2-DEFICIENT B CELLS their release into the splenic sinuses and, eventually, the blood surface IgD+ B cells. To assess whether the Grk2fl/flmb1-cre IgD+ stream, which allows them access to other immune compartments. B cells exited the spleen into the bloodstream, we checked the To assess the localization of B cells within the splenic compart- cellular composition of the blood. We found that the Grk2fl/flmb1-cre ments in the Grk2fl/flmb1-cre mice, we immunostained thick sec- mice had a nearly 2-fold increase in their blood leukocyte counts tions cut perpendicular to the long axis of the spleen with B220, versus control mice (Fig. 4A). Although their blood had a lower CD21, CD4, CD169, and Ki67, collected standard confocal mi- percentage of B cells, the overall B cell numbers in the blood were croscopy images, and tiled the images to give a composite picture similar with controls. Analysis of B cell subsets in the blood of a Grk2fl/fl and Grk2fl/flmb1-cre mouse spleen (Fig. 2D, 2E). We revealed an increase in immature and transitional B cells and a noted a loss of the normal demarcation between the white and red slight reduction of IgD+ cells (Fig. 4B). Checking the cell surface pulps. The white pulp contained usually distorted T cell zones and expression of the blood B cells revealed modest increases in small B cell zones. The reduction in B cell zones undoubtedly con- CXCR4, S1PR1, and ICAM-1 expression but small reductions in tributed to the loss of FO dendritic cells. Ki67 staining revealed CD62L and CD11a (Fig. 4C, 4D). The blood B cells from the constitutive germinal centers present in several Grk2fl/fl B cell follicles; Grk2fl/flmb1-cre mice exhibited even more severe reductions however, none were revealed in the Grk2fl/flmb1-cre splenic follicles in chemokine-directed migration than did the splenic B cells. (Fig. 2E). Some MZ macrophages were tightly associated with an In contrast to the splenic B cells, we did not observe a difference underlying T cell zone without an obvious B cell zone in the mice in nonspecific migration (Fig. 4E). Overall, the blood leukocyte lacking Grk2 in their B cells. The splenic MZs were poorly de- analysis verified the presence of recirculating IgD+ B cells in the lineated from the surrounding red pulp, and the expanded red pulp Grk2fl/flmb1-cre mice and demonstrated the inability of blood contained numerous B cells, scattered knots of CD169+ cells, and B cells to appropriately respond to chemokines. Downloaded from many Ki67+ cells. Interestingly, Grk2fl/flmb1-cre B and CD4 T cells The previous analysis of the Grk2fl/flmb1-cre mice noted a appeared to accumulate at various sites along the periphery of the significant reduction of B cells in LNs (14). We also found a sharp spleen (data not shown). Overall the architecture of the Grk2fl/flmb1-cre reduction in Grk2fl/flmb1-cre LNBcells(Fig.4F).TheGrk2fl/flmb1-cre spleens suggests that B cells lacking GRK2 cannot properly access mice consistently had small LNs, which contained small LN follicles, the white pulp, nor can they properly leave the red pulp to enter and the mice had a near absence of Peyer patches (Fig. 4F, the circulation. Supplemental Fig. 1). Again, we found a significant increase in the nonspecific migration of Grk2fl/flmb1-cre LN B cells compared http://www.jimmunol.org/ Enhanced Grk2fl/flmb1-cre B cell–specific migration to S1P but with control cells, although the CD4 T cells from the same mice impaired responses to chemokines exhibited no such difference (Fig. 4G). Like the Grk2fl/flmb1-cre The disrupted splenic architecture in the Grk2fl/flmb1-cre mice B cells from other sites, the LN B cells exhibited reduced specific suggested serious problems in chemoattractant-directed cell mi- migration to chemokines but enhanced responses to S1P (Fig. 4H). gration, as the splenic phenotype resembled that in mice whose CXCR4, CCR7, and CXCR5 receptor expression levels did not fl/fl fl/fl B cells lacked Gai proteins (5). The previous study of Grk2 CD4-cre differ, although the Grk2 mb1-cre LN B cells expressed higher mice had shown an enhanced S1P-triggered CD4 T cell migratory amounts of S1PR1 (Fig. 4I). These results are consistent with poor response; however; a checkerboard analysis showed that the re-

LN homing and/or accelerated egress and the previously observed by guest on October 3, 2021 sponse did not depend upon an S1P gradient (14). The FO and MZ altered migration in response to S1P and chemokines. B cells in the Grk2fl/flmb1-cre mice also had enhanced migratory Impaired Grk2fl/flmb1-cre B cell homing, egress, and responses to S1P. Although the FO and MZ B cell migration to positioning within lymphoid organs CXCL12 and CXCL13 were reduced by 20–25%, the differences did not reach statistical significance (14). We examined the re- Preliminary homing experiments, in which we adoptively trans- sponses of Grk2fl/flmb1-cre and control splenic B cell subsets to S1P, ferred a 1:1 ratio of WT and knockout (KO) B cells to WT mice, CXCL12, CXCL13, and CCL19 (Fig. 3A). We found that all the revealed a poor recovery of KO B relative to WT B cells from LNs B cell subsets exhibited excessive spontaneous migration (no (data not shown). To better assess the homing defect, we transferred fl/fl chemoattractant). Subtracting nonspecific migration from the a 2-fold excess of Grk2 mb1-cre B cells to control cells and chemoattractant-elicited migration gauges the specific migration. sampled the blood, spleen, peripheral LNs, and bone marrow 2 h The Grk2fl/flmb1-cre B cell subsets had heightened S1P-specific later. As expected, we found more KO B cells in the blood and responses, except for the MZ precursors (data not shown) and MZ spleen, yet the LNs and bone marrow had comparable numbers of B cells, which had lower responses to 100 nM S1P. In contrast, cells, consistent with a homing defect (Fig. 5A, left panel). To most Grk2fl/flmb1-cre B cell subsets had suboptimal specific re- assess LN egress, 2 h after adoptive transfer, we blocked further sponses, especially to high chemokine concentrations. The specific LN entrance with an Ab directed against CD62L. Eighteen hours migration to CXCL13 was most affected. All the subsets had ele- later, we enumerated the numbers of control and mutant B cells at vated S1PR1 expression levels, unperturbed CXCR4, CXCR5, and the various sites. Although we transferred a 2-fold excess of KO fl/fl CCR7 expression, and internalized S1PR1 at a slower rate than did B cells, we recovered fewer Grk2 mb1-cre B cells from the LNs WT cells (Fig. 3B, 3C, data not shown). Placing GRK2-deficient and bone marrow, consistent with accelerated egress (Fig. 5A, splenic B cells into an S1P-depleted environment marginally af- middle panel). Despite the 2:1 transfer ratio, we recovered similar fl/fl fected their S1PR1 expression (Fig. 3C). Together, these results numbers of blood B cells, arguing that many Grk2 mb1-cre indicate that the B cell GRK2 deficiency had enhanced cell motility, B cells failed to recirculate. Examining the ratio between the increased responses to low concentrations of S1P, but reduced homing and egress results highlights the increased splenic reten- specific responses to chemokines. As expected, the GRK2 loss re- tion, accelerated LN egress, and poor bone marrow retention (Fig. 5A, duced S1PR1 receptor internalization following S1P exposure. right panel). These results likely explain the lack of B cells in LNs, Peyer patches, and the 5-fold reduction of Fr. F cells in the Leukocytosis but a dearth of B cells in LNs and Peyer patches fl/f bone marrow. of Grk2 mb1-cre mice To better understand the positioning of B cells in the spleen and Although the splenic analysis revealed multiple severe immune LNs, we adoptively transferred labeled Grk2fl/fl and twice as many defects in the Grk2fl/flmb1-cre mice, their spleens had plenty of Grk2fl/flmb1-cre B cells into WT mice. At 4 h posttransfer, we The Journal of Immunology 7 Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021

FIGURE 3. Increased nonspecific migration and decreased CXCL13 but enhanced S1P-triggered migration of Grk2fl/flmb1-cre B cells. (A) Chemotaxis assays of CXCL12 (nanograms per milliliter), CCL19 (nanograms per milliliter), CXCL13 (nanograms per milliliter), and S1P (nanomolars). Left panel, Nonspecific migration; right panel, specific migration of T1, FO, and MZ B cells. (B) Flow cytometry examining S1PR1, CXCR4, CXCR5, and CCR7 expression on B cell subsets from Grk2fl/fl and Grk2fl/flmb1-cre mice. FACS profile shown only for S1PR1. (C) Time-dependent S1PR1 internalization following exposure to S1P (1 mM) and recovery of S1PR1 expression following placement of splenic B cells in media lacking S1P (n = 3 versus 3). Grk2fl/fl, black lines; Grk2fl/flmb1-cre, gray lines. Experiments were repeated three to four times. Data shown as mean 6 SEM. *p , 0.05, **p , 0.005, ***p , 0.0005. found nearly one-third fewer GRK2-deficient B cells had entered chimeras by infusing CD45.1 bone marrow with either CD45.2 the splenic white pulp, whereas 2.6-fold more resided in the red Grk2fl/fl or CD45.2 Grk2fl/flmb1-cre bone marrow into irradiated pulp (data not shown). At 18 h, the splenic follicles had one-fourth CD45.1 mice. Eight weeks after bone marrow reconstitution, we fewer GRK2-deficient B cells, whereas 4.3-fold more resided in checked the localization of the CD45.1 and CD45.2 cells by flow the red pulp (Fig. 5B). Examining LNs from the same animals, we cytometry and immunohistochemistry (Supplemental Fig. 2). As found fewer GRK2-deficient B cells in the LN follicles but more we had noted following the B cell adoptive transfer, the chimeric Grk2fl/flmb1-cre B cells than control cells at the B–T cell border mice demonstrated that the GRK2-deficient B cells largely resided (Fig. 5C). To confirm these findings, we made 1:1 bone marrow in the red pulp, whereas the control B cells localized in the B cell 8 GRK2-DEFICIENT B CELLS Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021

FIGURE 4. Leukocytosis, but a dearth of B cells in LNs and Peyer patches from Grk2fl/flmb1-cre mice. (A) Flow cytometry analysis of the blood from control and Grk2fl/flmb1-cre mice. Shown are total cell count and composition of blood leukocytes on a percentage basis. (B) Percentage of B cells in the indicated subsets from blood of Grk2fl/fl and Grk2fl/flmb1-cre mice. (C) Flow cytometry profiles of S1PR1 expression on indicated blood cells. (D) Flow cytometry results analyzing expression of S1PR1 and homing receptors on blood B cells. (E) Chemotaxis assays to CXCL12 (nanograms per milliliter), CCL19 (nanograms per milliliter), and CXCL13 (nanograms per milliliter). Left panel, Nonspecific migration; right panel, specific migration of bloodB cells and CD4 T cells. (F) Cell recovery and subset distribution in LNs and Peyer patches (PP) from Grk2fl/fl and Grk2fl/flmb1-cre mice. (G) Spontaneous migration of LN B cells and CD4 T cells. (H) Specific migration of LN B cells and CD4 T cells to indicated chemokines (nanograms per milliliter) or S1P (nanomolars) cells prepared from Grk2fl/fl and Grk2fl/flmb1-cre mice. (I) Flow cytometry results assessing S1PR1, CXCR4, CXCR5, and CCR7 expression on LN B cells from control and Grk2fl/flmb1-cre mice. All results are from B cells purified from a minimum of five Grk2fl/fl and Grk2fl/flmb1-cre mice. Data shown as mean 6 SEM. *p , 0.05, **p , 0.005, ***p , 0.0005. follicles. The B cell follicles had 20-fold fewer Grk2-deficient they have difficulty in crossing the endothelial and pericyte barriers B cells than control B cells. In contrast, the MZs exhibited a to enter LNs. To address the cause of the homing defect, we ex- strong preference for GRK2-deficient B cells. In LNs, the GRK2- amined the behavior of fluorescently labeled, adoptively transferred deficient B cells poorly populated LNs, and those cells that en- Grk2fl/fl and Grk2fl/flmb1-cre B cells in HEVs by intravital mi- tered LNs largely localized at the B–T border rather than in the croscopy. Imaging between 10 and 70 min following i.v. transfer, fl/fl fl/fl center of the follicle (Supplemental Fig. 2). These results con- we found that similar numbers of Grk2 and Grk2 mb1-cre firm the previous data illustrating the defective positioning of the B cells arrived and adhered to HEVs (Fig. 5D, Supplemental GRK2-deficient B cells in lymphoid organs. Video 1). Overall, their movements on the HEVs were similar (Fig. 5D, below); however, many of the Grk2fl/flmb1-cre B cells Blood-borne GRK2-deficient B cells adhere to high endothelial remained adherent to the HEV, either failing to cross the endo- venules but exhibit defective transmigration thelial barrier or only crossing the endothelial barrier without fully The LN homing defect suggests that either the GRK2-deficient exciting the HEV. We tracked 15 Grk2fl/fl-and27Grk2fl/flmb1-cre– B cells adhere poorly to high endothelial venules (HEVs) or that adherent B cells in HEVs. We found that 84.6% of the Grk2fl/fl The Journal of Immunology 9 Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021

FIGURE 5. Impaired Grk2fl/flmb1-cre B cell homing, egress, and positioning within lymphoid organs when compared with B cells from Grk2fl/fl mice. (A) Adoptive transfer of a 1:2 ratio of Grk2fl/f and Grk2fl/flmb1-cre B cells to WT mice. Number of cells recovered from the blood, spleen, inguinal LN (iLN), and popliteal LNs (pLN) 2 h after transfer (left) or 18 h after administration of an Ab to CD62L, which had been injected 2 h after cell transfer (middle). Also shown is the ratio between the homing and retention results (right). (B and C) Confocal images of spleen and LN sections 18 h posttransfer of differentially labeled Grk2fl/fl (pink) and Grk2fl/flmb1-cre (blue) B cells to WT mice. Sections immunostained for B220 (green), CD169 (red), and Lyve-1(yellow). The distribution of cells shown to right of the sections. (D) Captured two-photon laser-scanning microscopy (TP-LSM) image of an HEV with adoptively transferred Grk2fl/fl (red) and Grk2fl/flmb1-Cre (green) B cells. HEVs outlined by injection of Evans blue (gray). Visualized tracks are shown (top). Below is analysis of B cell behavior in the HEVs. Calculated mean speed and displacement are shown. (E) Grk2fl/flmb1-cre B cells interfere with the transendothelial migration of WT B cells. Grk2fl/fl or Grk2fl/fl mb1-cre B cells were pretransferred to WT mice. One hour later, labeled WT B cells were transferred, and 2 h later, labeled LN WT B cells were enumerated. (F) Captured intravital TP-LSM LN image showing Grk2fl/fl and Grk2fl/flmb1-cre (2-fold excess) B cells. Cells transferred the day prior to imaging. Tracks were generated spanning imaging duration (top). (Figure legend continues) 10 GRK2-DEFICIENT B CELLS

B cells that adhered to the HEV surface transmigrated with a 100% that of controls, although surprisingly, they were elevated in FO success rate. In contrast, only 54.6% Grk2fl/flmb1-cre–adherent and T0/T1 B cells from the same mice (Fig. 6B). Exposing T0/T1 B cells attempted to transmigrate, with a success rate of 37.5%, B and FO B cells to S1P reduced pERM proteins in Grk2fl/fl resulting in an overall transmigration frequency of 20.3%. Some B cells and GRK2-deficient B cells, although the latter exhibited GRK2-deficient B cells remained confined to the perivascular an exaggerated and more pronounced downregulation. The Grk2fl/fl space between the and pericyte layer (Supplemental MZ B cells behaved like the other B cell subsets, whereas the Video 1). During these experiments, we noted fewer WT B cells GRK2-deficient MZ B cells had a biphasic response to S1P with an had transmigrated than expected, based on our previous studies. initial increase in pERM levels followed by a prolonged downreg- To test whether the GRK2-deficient B cells might impair WT ulation (Fig. 6B). B cell entrance into LN, 1 h prior to adoptively transferring Grk2fl/fl Chemoattractant receptor signaling predominately raises intra- B cells fluorescently marked (blue) we transferred either Grk2fl/flmb1-cre cellular Ca2+ levels via Gbg subunit–mediated increases in phos- (green) or Grk2fl/fl B cells (red). Two hours later, we assessed the pholipase C activity (24). In contrast, to the other CXCL13-induced number of Grk2fl/fl B cells (blue) in LNs. We found that the prior signaling pathways we had assessed, the loss of Grk2 enhanced the transfer of Grk2fl/flmb1-cre B cells had severely compromised the intracellular Ca2+ responses. As expected, S1P elicited much more homing of Grk2fl/fl B cells (Fig. 5E). potent responses in the KO B cells (Fig. 6C). The lack of GRK2 in To track B cells within the LN parenchyma we transferred a 2:1 B cells raised the basal intracellular Ca2+ (KO 363 6 3 versus ratio of Grk2fl/flmb1-cre and Grk2fl/fl B cells and imaged the fol- control 283 6 4) and slightly enhanced the increase intracellular lowing day (Fig. 5F). The imaging revealed that the GRK2- Ca2+ following cross-linking the B cell Ag receptor (data not deficient B cells that had entered LN tended to remain near the shown). The Grk2fl/flmb1-cre B cells intracellular Ca2+ responses Downloaded from HEVs rather than migrating into the LN follicles. Tracking the to chemoattractants remained dependent upon Gai proteins, as per- control and GRK2-deficient B cells did not reveal any statisti- tussis toxin blocked both the S1P- and CXCL13-elicited increases cal difference in their behavior, except that the GRK2-deficient (Fig. 6C). To assess the contribution of S1PR1 to the S1P-induced B cells had a reduced displacement, consistent with their ten- increases in intracellular Ca2+, we pretreated the Grk2fl/flmb1-cre and dency to remain close to the HEVs. To analyze the motility of FO Grk2fl/fl splenic B cells with Ex 26, a specific S1PR1 antagonist (25), B cells, we transferred a 4:1 ratio of Grk2fl/flmb1-cre and Grk2fl/fl prior to the S1P challenge (Fig. 6D). The Ex 26 pretreatment B cells. This allowed enough Grk2-deficient B cells to access the blocked the S1P-induced increases in intracellular Ca2+ in both the http://www.jimmunol.org/ follicle for us to track them (Fig. 5G, Supplemental Video 2). We control and GRK2-deficient B cells. The Ex 26 pretreatment slightly found that the Grk2-deficient B cells moved faster with greater reduced the CXCL13-induced intracellular Ca2+ response in the displacement than did the control cells within the follicle. These control B cells and consistently enhanced the CXCL13 response in results show that the loss of GRK2 in B cells impairs trans- the GRK2-deficient B cells. endothelial migration and interferes with the normal trafficking Ex 26 reverses the CXCL13 chemotaxis defect and partially through LN follicles. restores normal in vivo B cell positioning Enhanced S1P signaling but reduced CXCL13 signaling, 2+ To determine whether blocking S1PR1 signaling would affect except for the intracellular Ca response in Grk2-deficient B cell migratory responses, we again used Ex 26 (25), (Fig. 7). by guest on October 3, 2021 B cells Using B cells from the chimeric mice allowed for testing B cells As we had noted the most significant impairment in the migration to from the same mice in the same assays, distinguishing their ge- CXCL13, we compared the effect of CXCL13 and S1P on the notype based on CD45.1 (WT) versus CD45.2 (GRK2-deficient). induction of downstream signaling intermediates. Chemoattractant Pretreatment of WT B cells with Ex 26 increased the percentage receptor signaling leads to Gi-dependent increases in AKT and of cells migrating to CXCL13, whereas it marginally affected the ERK phosphorylation; but it leads to decreases in ezrin, radixin, responses to CXCL12 and CCL19. The drug treatment reduced and moesin (ERM) phosphorylation (14, 23). We used flow the WT T1 B and FO B cell responses to S1P but minimally af- cytometry to assess the levels of pAKT and pERK in Grk2fl/flmb1-cre fected the MZ B cells, as S1PR3 predominately mediates MZ and Grk2fl/fl splenic B cell subsets following exposure to CXCL13 or B cell chemotaxis. Treating the GRK2-deficient B cells with Ex S1P (Fig. 6A, 6B). The Grk2fl/fl and GRK2-deficient splenic B cell 26 reduced the nonspecific migration and raised the percentage subsets had similar basal levels of pAKT and pERK (data not cells specifically migrating in response to all the chemokines, shown). CXCL13 increased pAKT levels in control B cells, but although the FO B cell responses to CXCL12 and CCL19 did not the GRK2-deficient B cells responded less well. S1P exposure reach statistical significance. The Ex 26 pretreatment blocked the induced weak increases in pAKT levels in the control B cells, exaggerated T1 and FO B cells responses to S1P. whereas the loss of GRK2 raised the response to that observed To test whether Ex 26 would affect the positioning of GRK2- with CXCL13. The pERK responses mirrored the pAKT responses, deficient B cells in lymphoid organs we treated the CD45.1 as the loss of GRK2 reduced CXCL13 responses and strongly in- (WT):CD45.2 (Grk2fl/flmb1-cre)chimericmiceovernightwith creased S1P responses. Similarly, the lack of GRK2 enhanced the Ex 26. The following day we examined the location of the WT and S1P-induced increases in pBTK and reduced the CXCL13-induced GRK2-deficient B cells in tissue sections from the spleen and increases (Supplemental Fig. 3). We found pERM levels were lower LNs. The Ex 26 treatment resulted in a clear shift of the GRK2- in the Grk2fl/flmb1-cre blood B cells but not in blood CD4 T cells. deficient B cells from the red pulp into the white pulp, although Basal pERM levels were similar in Grk2fl/flmb1-cre MZ B cells to the ratio between WT and GRK2-deficient B cells varied among

Analysis of B cell behavior in the cortical ridge and LN follicle. Mean speed and displacement are shown (bottom). (G) Captured intravital LN TP-LSM image of LN showing Grk2fl/fl and Grk2fl/flmb1-cre (4-fold excess) B cells. Cells were transferred the day prior to imaging (top). Tracking results of B cells confined to the LN follicle. Shown are mean speed and displacement (bottom). Scale bars, 200 mm(B) and 50 mm(C, D, F, and G). Each experiment was repeated a minimum of three times. Data shown as mean 6 SEM. *p , 0.05, **p , 0.005, ***p , 0.0005. The Journal of Immunology 11 Downloaded from http://www.jimmunol.org/ by guest on October 3, 2021

FIGURE 6. The loss of GRK2 in B cells enhances S1P signaling but decreases CXCL13 signaling, except for the intracellular Ca2+ response. (A) Chemoattractant-induced pAKT and pERK in B cell subsets. Grk2fl/fl or Grk2fl/flmb1-cre B cells were exposed to CXCL13 (1000 ng/ml) or S1P (1 mM) for indicated times and the levels of pAKT or pERK were determined by flow cytometry. Data are expressed as percentage of basal. (B) pERM levels in blood and splenic B cells from Grk2fl/fl or Grk2fl/flmb1-cre mice. Level of pERM in blood B cell subsets and CD4 T cells determined by flow cytometry (left). Basal pERM levels in splenic B cell subsets and following exposure to S1P (1 mM). (C) Peak intracellular Ca2+ levels following chemoattractant exposure and impact of pertussis toxin (200 nM). Grk2fl/fl or Grk2fl/flmb1-cre splenic B cells were exposed to various concentrations of CXCL13 (nanograms per milliliter) or S1P (nanomolars). (D) Effect of an S1PR1 antagonist on S1P- and CXCL13-induced increases in intracellular Ca2+. Plot of intracellular Ca2+ levels over time in splenic B cells from Grk2fl/fl or Grk2fl/flmb1-cre mice exposed to S1P or CXCL13. Where indicated, the B cells were preincubated with Ex 26 for 1 h prior to the addition of S1P or CXCL13. Results are from 2 to 10 mice per group. All experiments were repeated a minimum of three times. Data shown as mean 6 SEM. *p , 0.05, **p , 0.005, ***p , 0.0005. the different B cells follicles, with some being well populated with outputs (26). Consequently, disruption of one of these mech- GRK2-deficient B cells and others less so (Supplemental Fig. 4). anisms can have far-reaching and, sometimes, unanticipated Similarly, in the LNs, more GRK2-deficient B cells populated consequences. This is the case with the loss of GRK2 in B cells. A the LN follicle, although WT cells predominated in the LN severe phenotype arose, apparently largely because of a failure of follicle center (Supplemental Fig. 4). Because we only treated GRK2 to phosphorylate S1PR1 receptors. Inappropriate S1PR1 with a single injection of Ex 26, a more prolonged treatment signaling likely usurps intermediates needed regimen may have further normalized the distribution of the by other chemoattractant receptor signaling pathways. It also led GRK2-deficient B cells. to compensations that may have exacerbated the phenotypes. This report revealed a major impact of the loss of GRK2 in Discussion B cells on CXCR5 signaling. In the end, the misbalance between Outputs from chemoattractant-triggered signaling pathways con- chemokine and S1PR1 signaling in the GRK2-deficient B cells trol cellular migration and behavior. Not surprisingly, numerous explains many of the abnormalities in the B cell compartment of interacting mechanisms exist to either enhance or dampen these these mice. 12 GRK2-DEFICIENT B CELLS Downloaded from http://www.jimmunol.org/

FIGURE 7. The S1PR1 antagonist inhibits S1P-directed migration but enhances CXCL12-, CXCL13-, and CCL19-directed migration of Grk2fl/flmb1-cre B cells. Standard chemotaxis assays using splenic B cells from chimeric mice (CD45.1 WT and CD45.2 Grk2fl/flmb1-cre). Cells distinguished by flow cytometry using CD45.1- and CD45.2- specific Abs. Nonspecific migration (left) and specific migration to indicated chemoattractants of T1, FO, and MZ B cells. The cells were treated or not with Ex 26 for 1 h prior to exposure to chemoattractants. Results are from splenic B cells from four chimeric mice, each

assayed in duplicate. Experiment was repeated twice with comparable results. Data shown as mean 6 SEM. *p , 0.05, **p , 0.005, ***p , 0.0005. by guest on October 3, 2021

Because B cell lymphopoiesis in adult mice begins in the bone the precise reason for this is unclear. The levels of signaling marrow, we first assessed the impact of the loss of GRK2 on bone available S1P in the splenic red pulp is low (27), arguing that a marrow B cell development. The noted increase in Fr. D cells may retention mechanism is unlikely to be the sole explanation. S1PR1 be secondary to the loss of Fr. F cells and the premature release signaling supports plasma cell egress from the spleen into the of immature Fr. E cells. We noted a difference in nonspecific blood (28) and FO B cell egress from the white pulp to the red migration and a slight reduction in migration to CXCL12. The pulp, but its involvement in the later stages of lymphocyte egress enhanced motility of the GRK2-deficient immature B cells may from the spleen are unknown. The reduced white pulp in the have facilitated their premature release into the bloodstream. spleens of the Grk2fl/flmb1-cre mice reflects the loss of B cells Consistent with this idea, blood from the Grk2fl/flmb1-cre mice had from the white pulp, which in-turn led to the FO dendritic cell an excess of immature B cells. The most intriguing bone marrow loss. The failure of B cells to populate the white pulp is secondary phenotype was the major of loss Fr. F cells, mature B cells that to excessive S1PR1 signaling retaining the cells near the MZ and have recirculated to and lodged in the bone marrow. This loss did reduced CXCL13-directed migration, which normally draws cells not arise from a lack of mature IgD+ B cells in blood but rather into the B cell follicle (22). The poor population of the splenic results from mature B cells that have failed to access and/or be B cell follicles could be partially reversed by treating the mice retained in their designated microenvironment. Future bone mar- with an S1PR1 antagonist, Ex 26. Based on our analysis, the drug row imaging experiments may help explain why these cells fail to would be not only expected to reduce the S1PR1 signal, but also to lodge and be retained in the bone marrow. improve signaling through CXCR5. The splenic architecture of the Grk2fl/flmb1-cre mice was se- Impaired chemokine receptor signaling along with excessive verely compromised. Whereas dysregulated chemokine and S1P S1PR1 signaling also explains many of the LN phenotypes in the signaling provides an explanation for some of the splenic abnor- Grk2fl/flmb1-cre mice. The lack of B cells in the LNs results from malities, other abnormalities will require additional study to be a homing defect along with their accelerated egress. Imaging fully understood. The Grk2fl/flmb1-cre animals had spleens 2-fold in the HEVs revealed no obvious adherence defect but revealed a larger than control animals. Furthermore, the spleens had an ex- problem in B cell transendothelial migration. This result suggests cess of both B and T cells, many of which accumulated in the red that the loss of Grk2 in B cells had not impaired chemokine-induced pulp. This suggests that the GRK2-deficient B cells had interfered integrin activation. Some adherent GRK2-deficient B cells migrated with the normal lymphocyte egress pathways in the spleen, thereby on the endothelium for extended periods, adopting a normal po- causing a problem for T cell egress. This may reflect a difficulty in larized morphology, but often failed to transmigrate. Other GRK2- cell migration from the splenic cords into the sinusoids; however, deficient B cells migrated through the endothelial barrier only to get The Journal of Immunology 13 stuck in the perivascular space. This explains the reduced homing phosphorylated Bruton tyrosine kinase in the Grk2fl/flmb1-cre of WT B cells in the presence of GRK2-deficient B cells. The B cells, whereas CXCL13 triggered a poor response. However, the B cell transmigration defect may reflect the inability of GRK2- intracellular Ca2+ responses behaved differently, as the GRK2- deficient B cells to properly sense the chemokines that direct deficient B cells had heightened responses to CXCL13 and S1P. B cells into the LN or, conversely, to properly limit integrin ac- Typically, S1P elicits a rather anemic intracellular Ca2+ response; tivation. Alternatively, the inappropriate S1PR1 signaling may however, the loss of GRK2 boosted it to the level typically ob- have altered the physical properties of the cells, interfering with served following CXCL13 stimulation of control B cells. Not the normal membrane deformation required for transmigration. only did the GRK2 deficiency not impair the CXCL13-induced Of those B cells that did enter the LN parenchyma, many did not intracellular Ca2+ response, we consistently found a 1.5–2-fold migrate away from the HEVs into the LN follicle. Rather, they increase in the peak response, compared with controls. We do remained localized near the HEVs, which presumably improved not have a good explanation for this disparity. We did note that their likelihood of leaving the LN by entering the nearby cortical the Grk2fl/flmb1-Cre splenic B cells had a higher basal level of lymphatics, thereby explaining the accelerated egress noted in intracellular Ca2+ and a 20% increase in CD38 expression (I.-Y. these mice. The B cells that managed to migrate from the HEVs to Hwang, unpublished observations). Excessive S1PR1 signaling may the LN follicle either failed to enter the follicle or tended to remain have increased CD38 expression, which could elevate intracellular along the edge. These cellular behaviors are best explained by an Ca2+ responses. inability of the GRK2-deficient B cells to properly sense CXCL13 In conclusion, the loss of GRK2 in B cells disrupted the normal emanating from the LN follicle. The decreased tendency to migrate S1P-induced S1PR1 desensitization/resensitization cycle. Despite to the follicle center could also be explained by an enhanced exposure to saturating concentrations of S1P, GRK2-deficient sensitivity to 7a,25-dihydroxyxcholesterol, the GPR183 (Ebi2) blood B cells maintain S1PR1 expression, whereas control B cells Downloaded from ligand (29); however, we could find no in vitro evidence to support do not. Strikingly, the loss of GRK2 led to a nearly 10-fold in- that hypothesis, as LPS-activated B cells migrated less well than crease in the magnitude of S1PR1 signaling as assessed by changes did control cells to 7a,25-dihydroxyxcholesterol (I.-Y. Hwang, in pAKT, pERK, and intracellular Ca2+ levels. The enhanced unpublished observations). signaling likely contributes to the hypermigratory responses of the Not only did the Grk2fl/flmb1-cre mice have small underpopu- GRK2-deficient B cells to S1P. Although the heightened receptor lated LNs, they had only two to four visible Peyer patches versus expression may explain the increased S1PR1 signaling, a role for http://www.jimmunol.org/ the usual 8–10 found in control mice, and we recovered 20-fold GRK2 in limiting signal transduction by some form of negative fewer cells. Deleting GRK2 from bone marrow B cell progenitors feedback remains plausible. Although GRK2 predominately tar- should not impair early Peyer patch development, suggesting that geted S1PR1, the loss of GRK2 in B cells also affected signaling loss of Peyer patch cells results from a major defect in B cell through chemokine receptors. The reversal of the GRK2-deficient recruitment or retention. CXCR4, CXCR5, and CCR7 all partic- B cell CXCL13 migratory defect by an S1P antagonist in vitro and ipate in B cell homing to Peyer patches (30). The same issues as the partial reversal of the B cell FO homing defect in vivo argues outlined with the entrance of B cells into LNs, nondirectional that GRK2-mediated S1PR1 desensitization allows normal chemo- S1PR1 signaling along with inadequate chemokine receptor sig- kine receptor signaling. Future studies will focus on understanding by guest on October 3, 2021 naling, likely limit the recruitment of GRK2-deficient B cells from the mechanism by which signaling through S1PR1 impairs chemo- the blood into Peyer patches. Because the Peyer patch cell population kine responses and vice versa. Overall, the misbalanced S1PR1 and normally contains 70–80% B cells, a recruitment/retention defect homeostatic chemokine receptor signaling in the Grk2fl/flmb1-cre would lead to a major loss of the cells populating Peyer patches. mice causes a surprisingly severe B cell–trafficking defect that Our migration assay results differ from those previously markedly disrupts normal immune organ architecture and function. reported, which also analyzed Grk2fl/flmb1-cre mice on a B6 background (14). Like their report, we found that FO B cells had a Acknowledgments heighted response to S1P; however, we found that MZ B cells We thank Dr. Anthony Fauci for continued support. responded less well to 50 and 100 nM S1P. The reason for this discrepancy is unclear. Compared with the usual 2–5% nonspecific Disclosures migration, as many as 10–15% of the GRK2-deficient B cells The authors have no financial conflicts of interest. migrated in the absence of any chemoattractant. Interestingly, treating the B cells with Ex 26 substantially reduced this non- specific migration, arguing that in GRK2-deficient cells S1PR1 References signals, in part constitutively, and that Ex 26 may function not 1. Lian, J., and A. D. Luster. 2015. Chemokine-guided cell positioning in the lymph node orchestrates the generation of adaptive immune responses. Curr. Opin. Cell only as a neutral antagonist, but also as an inverse agonist. Others Biol. 36: 1–6. have noted constitutive S1PR1 signaling, and a previous study 2. Schulz, O., S. I. Hammerschmidt, G. L. Moschovakis, and R. Fo¨rster. 2016. showed that the constitutive activity by S1PR1 receptors enhanced Chemokines and chemokine receptors in lymphoid tissue dynamics. Annu. Rev. Immunol. 34: 203–242. platelet-derived growth factor–induced cell migration. A partial 3. Okada, T., and J. G. Cyster. 2006. B cell migration and interactions in the early S1PR1 agonist, SB649146, behaved as an inverse agonist, inhib- phase of antibody responses. Curr. Opin. Immunol. 18: 278–285. 4. Kehrl, J. H. 2016. The impact of RGS and other G-protein regulatory proteins on iting the migration (31). Treating the GRK2-deficient T1 and FO Gai-mediated signaling in immunity. Biochem. Pharmacol. 114: 40–52. B cells with Ex 26 largely blocked their migration response to S1P 5. Hwang, I. Y., C. Park, T. Luong, K. A. Harrison, L. Birnbaumer, and J. H. Kehrl. and reversed their decreased specific migration to chemokines. 2013. The loss of Gnai2 and Gnai3 in B cells eliminates B lymphocyte com- partments and leads to a hyper-IgM like syndrome. PLoS One 8: e72596. Intriguingly, Ex 26 also enhanced the specific migratory response 6. Hwang, I. Y., C. Park, K. Harrison, C. Boularan, C. Gale´s, and J. H. Kehrl. 2015. of control B cells to CXCL13, suggesting that constitutive S1PR1 An essential role for RGS protein/Gai2 interactions in B lymphocyte-directed signaling may limit CXCL13-directed migration in normal B cells. cell migration and trafficking. J. Immunol. 194: 2128–2139. 7. Erlandson, S. C., C. McMahon, and A. C. Kruse. 2018. Structural basis for In general, the pERK and pAKT assays correlated well with the G protein-coupled receptor signaling. Annu. Rev. Biophys. 47: 1–18. migration assay, as the Grk2fl/flmb1-cre B cells had reduced 8. Homan, K. T., and J. J. Tesmer. 2014. Structural insights into G protein-coupled receptor kinase function. Curr. Opin. Cell Biol. 27: 25–31. responses to CXCL13 and heightened responses to S1P. Like 9. Komolov, K. E., Y. Du, N. M. Duc, R. M. Betz, J. Rodrigues, R. D. Leib, the pERK and pAKT responses, S1P elicited a strong increase in D. Patra, G. Skiniotis, C. M. Adams, R. O. Dror, et al. 2017. Structural and 14 GRK2-DEFICIENT B CELLS

functional analysis of a b2- complex with GRK5. Cell 169: 21. Henderson, R. B., K. Grys, A. Vehlow, C. de Bettignies, A. Zachacz, T. Henley, 407–421.e16. M. Turner, F. Batista, and V. L. Tybulewicz. 2010. A novel Rac-dependent 10. McGovern, K. W., and K. A. DeFea. 2014. Molecular mechanisms underlying checkpoint in B cell development controls entry into the splenic white pulp beta-arrestin-dependent chemotaxis and actin-cytoskeletal reorganization. and cell survival. J. Exp. Med. 207: 837–853. Handb. Exp. Pharmacol. 219: 341–359. 22. Arnon, T. I., R. M. Horton, I. L. Grigorova, and J. G. Cyster. 2013. Visualization 11. Kozasa, T. 2004. The structure of GRK2-G beta gamma complex: intimate as- of splenic marginal zone B-cell shuttling and follicular B-cell egress. Nature sociation of G-protein signaling modules. Trends Pharmacol. Sci. 25: 61–63. 493: 684–688. 12. Stoffel, R. H., R. R. Randall, R. T. Premont, R. J. Lefkowitz, and J. Inglese. 23. Hwang, I. Y., C. Boularan, K. Harrison, and J. H. Kehrl. 2018. Gai signaling 1994. Palmitoylation of G protein-coupled receptor kinase, GRK6. Lipid mod- promotes marginal zone B cell development by enabling transitional B cell ification diversity in the GRK family. J. Biol. Chem. 269: 27791–27794. ADAM10 expression. Front. Immunol. 9: 687. 13. Fong, A. M., R. T. Premont, R. M. Richardson, Y. R. Yu, R. J. Lefkowitz, and 24. Boyer, J. L., G. L. Waldo, and T. K. Harden. 1992. Beta gamma-subunit activation D. D. Patel. 2002. Defective lymphocyte chemotaxis in beta-arrestin2- and of G-protein-regulated phospholipase C. J. Biol. Chem. 267: 25451–25456. GRK6-deficient mice. Proc. Natl. Acad. Sci. USA 99: 7478–7483. 25. Cahalan, S. M., P. J. Gonzalez-Cabrera, N. Nguyen, M. Guerrero, E. A. Cisar, 14. Arnon,T.I.,Y.Xu,C.Lo,T.Pham,J.An,S.Coughlin,G.W.Dorn,andJ.G.Cyster. N. B. Leaf, S. J. Brown, E. Roberts, and H. Rosen. 2013. Sphingosine 1-phosphate 2011. GRK2-dependent S1PR1 desensitization is required for lymphocytes to over- receptor 1 (S1P(1)) upregulation and amelioration of experimental autoimmune come their attraction to blood. Science 333: 1898–1903. encephalomyelitis by an S1P(1) antagonist. Mol. Pharmacol. 83: 316–321. 15. Park, C., I. Y. Hwang, and J. H. Kehrl. 2018. The use of intravital two-photon 26. Kinzer-Ursem, T. L., and J. J. Linderman. 2007. Both ligand- and cell-specific and thick section confocal imaging to analyze B lymphocyte trafficking in lymph parameters control ligand agonism in a kinetic model of g protein-coupled re- nodes and spleen. Methods Mol. Biol. 1707: 193–205. ceptor signaling. PLOS Comput. Biol. 3: e6. 16. Hardy, R. R., Y. S. Li, D. Allman, M. Asano, M. Gui, and K. Hayakawa. 2000. 27. Ramos-Perez, W. D., V. Fang, D. Escalante-Alcalde, M. Cammer, and S. R. Schwab. B-cell commitment, development and selection. Immunol. Rev. 175: 23–32. 2015. A map of the distribution of sphingosine 1-phosphate in the spleen. Nat. 17. Cariappa, A., C. Chase, H. Liu, P. Russell, and S. Pillai. 2007. Naive recircu- Immunol. 16: 1245–1252. lating B cells mature simultaneously in the spleen and bone marrow. Blood 109: 28.Kabashima,K.,N.M.Haynes,Y.Xu,S.L.Nutt,M.L.Allende,R.L.Proia, 2339–2345. and J. G. Cyster. 2006. Plasma cell S1P1 expression determines secondary 18. Cariappa, A., I. B. Mazo, C. Chase, H. N. Shi, H. Liu, Q. Li, H. Rose, H. Leung, lymphoid organ retention versus bone marrow tropism. J. Exp. Med. 203: B. J. Cherayil, P. Russell, et al. 2005. Perisinusoidal B cells in the bone marrow 2683–2690. participate in T-independent responses to blood-borne microbes. Immunity 23: 29. Gatto, D., and R. Brink. 2013. B cell localization: regulation by EBI2 and its Downloaded from 397–407. oxysterol ligand. Trends Immunol. 34: 336–341. 19. Beck, T. C., A. C. Gomes, J. G. Cyster, and J. P. Pereira. 2014. CXCR4 and a 30. Okada, T., V. N. Ngo, E. H. Ekland, R. Fo¨rster, M. Lipp, D. R. Littman, and cell-extrinsic mechanism control immature B lymphocyte egress from bone J. G. Cyster. 2002. Chemokine requirements for B cell entry to lymph nodes and marrow. J. Exp. Med. 211: 2567–2581. Peyer’s patches. J. Exp. Med. 196: 65–75. 20. Pereira, J. P., Y. Xu, and J. G. Cyster. 2010. A role for S1P and S1P1 in immature- 31. Waters, C. M., J. Long, I. Gorshkova, Y. Fujiwara, M. Connell, K. E. Belmonte, B cell egress from mouse bone marrow. [Published erratum appears in 2010 PLoS G. Tigyi, V. Natarajan, S. Pyne, and N. J. Pyne. 2006. Cell migration activated by One. DOI: 10.1371/annotation/2ae645ec-9413-4f7d-b51f-eb0678fa2f1b.] PLoS platelet-derived growth factor receptor is blocked by an inverse agonist of the

One 5: e9277. sphingosine 1-phosphate receptor-1. FASEB J. 20: 509–511. http://www.jimmunol.org/ by guest on October 3, 2021