Fig 2: Arhgef6 -/- MZB migrate faster, detach more under flow, and respond strongly to S1P. a Representative track plots of wild-type (WT) and Arhgef6 -/- MZB bells, untreated or with S1P (200 nM), migrating under flow (3 dyn cm-2) on ICAM-1 (2.5 µg ml-1) or VCAM-1 (2.5 µg ml-1). b Quantification of migration index, velocity, straightness, and percentage of cells that migrated >10 µm for WT (gray bars) and Arhgef6 -/-(white bars) MZB on ICAM-1 or VCAM-1 (WT: n = 4 mice, Arhgef6 -/-: n = 5 mice; 5 experiments total). Data are expressed as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, by t test (for migration index and % displacement on ICAM-1), one-way ANOVA (for straightness and % displacement, comparison of values to WT), and two-way ANOVA (velocity on ICAM-1, migration index on VCAM-1). c Detachment of WT and Arhgef6 -/-MZB. Fraction of remaining cells calculated as a % relative to number of cells after start of flow on ICAM-1 (1.5 µg ml-1), VCAM-1 (1.5 µg ml-1), or ICAM-1 and VCAM-1 together (0.75 µg ml-1 each) under flows of 3 dyn cm-2 (0–10 min), 9 dyn cm-2 (10–20 min), and 15 dyn cm-2 (20–30 min) (n = 5 mice per genotype, 5 experiments total). d Static adhesion of WT and Arhgef6 -/- MZB treated with S1P (200 nM) on ICAM-1 (2.5 µg ml-1), relative to total adhesion on poly-lysine. Bars are mean plus SEM. **P < 0.01, two-way ANOVA for S1P treatment difference. Data are from three experiments with two mice (one per genotype) each. ns not significant

Fig 3: MZB are not flushed out by shear flow but migrate up it. a Detachment over time of marginal zone B cells (MZB) and follicular B cells (FOB). Percent attachment based on number of cells after flow (3 dyn cm-2) start (t = 10 s), mean ± SEM on ICAM-1 (2 µg ml-1), VCAM-1 (2 µg ml-1), or ICAM-1 and VCAM-1 (1 µg ml-1 each) mixed (n = 3). b Representative track plots of MZB migration up the flow on ICAM-1 (7 µg ml-1). Flow strengths indicated in dyn cm-2 by the number over the track plot; flow direction indicated by arrow. Tracks with end points above the horizontal axis are shown in black; tracks below are shown in red. c Quantification of velocity, migration index, and straightness for the indicated flow strengths. All three graphs show different parameters from the same set of experiments; bars show mean ± SEM. Data are from three experiments with 1 mouse each. Symbols in one condition group denote the three replicates and represent the average of 100–200 cells from each individual mouse. *P < 0.05, **P < 0.01, one-way ANOVA with Dunnett post hoc tests comparing all flow strengths to 0 dyn cm-2

Fig 4: Model of MZB migration up flow with S1P counterforce. a MZBs in spleen shuttle from the marginal zone (MZ) surrounding the follicle (FO) to the inside of the follicle, to carry antigens from blood to the follicular dendritic cells and back again. CXCL13 is required for MZB entry into the follicle, while S1PR1 is required for MZB exit from the follicle. The marginal zone is rich in ICAM-1 while the red pulp contains both ICAM-1 and high levels of VCAM-1. An unknown force drives MZBs toward the follicle. b Flow: Shear flow induces MZB migration up the flow if ICAM-1 is present. S1P signaling through S1P receptors inhibits MZB migration up flow and may trigger LFA-1 integrin re-distribution. S1PR1 is rapidly internalized, thus S1PR3 may be the main counterforce. Upon eventual internalization of S1PR3, MZBs would migrate up the flow. MZBs that reach the sinus (elongated green cells) would bind strongly to MAdCAM-1 and CXCL13, preventing their migration against flow but protecting them from washing out. MZBs would then transmigrate through the sinus into the follicle. No flow: without flow, MZBs migrate for shorter distances in random directions and MZB follicular shuttling is impaired. S1pr3 -/-: without S1PR3, MZBs do not respond to S1P and are not inhibited from migrating up the flow toward the follicle. Arhgef6 -/- or LFA-1 block: MZBs that undergo a loss of adhesion, through Arhgef6 mutation or LFA-1 blockade, are susceptible to washing out of the marginal zone toward the red pulp. S1P binding to receptors would accelerate this process. Impaired migration up flow causes reduced shuttling into the follicle. c VCAM-1 in the red pulp. MZBs can adhere to VCAM-1 under flow but do not migrate on it. MZBs with reduced adhesion via LFA-1 are conveyed by flow to the red pulp. Left panel: Arhgef6 -/- MZBs that reach the red pulp can adhere to it and resist being washed out by flow. Right panel: MZBs with blocked aLFA-1 are only able to partially resist the force of the flow by adhering to VCAM-1

Fig 5: MZBs migrate up the flow on ICAM-1 but not on VCAM-1. a Representative track plots of MZB migration up the flow (8 dyn cm-2; direction indicated by arrow) with coating (ICAM-1 or VCAM-1) amounts shown above. b Quantification of velocity, straightness, migration index, and % of cells with displacement >10 µm for the ICAM-1 to VCAM-1 (IV) ratios shown above in a. All four graphs show different parameters from the same set of experiments; bars show mean ± SEM. Data are from three experiments with one mouse each. Symbols in one condition group denote the three replicates and represent the average of 100–200 cells from each individual mouse. c Microscopy of ICAM-1 (green), VCAM-1 (magenta), and the merge (white) in the spleen, showing follicles (FO), marginal zone (MZ), and red pulp (RP). Data represent two experiments with one spleen each. Scale bar; 100 µm. **P < 0.01, ***P < 0.001, ****P < 0.0001, one-way ANOVA with Dunnett post hoc tests comparing all ICAM-1/VCAM-1 mixtures to ICAM-1 alone (5:0)