Myo1g is required for efficient adhesion and migration of activated B lymphocytes to inguinal lymph nodes

Cell migration is a dynamic process that involves adhesion molecules and the deformation of the moving cell that depends on cytoskeletal remodeling and actin-modulating proteins such as myosins. In this work, we analyzed the role of the class I Myosin-1 g (Myo1g) in migratory processes of LPS + IL-4 activated B lymphocytes in vivo and in vitro. In vivo, the absence of Myo1g reduced homing of activated B lymphocytes into the inguinal lymph node. Using microchannel chambers and morphology analysis, we found that the lack of Myo1g caused adhesion and chemotaxis defects. Additionally, deficiency in Myo1g causes flaws in adopting a migratory morphology. Our results highlight the importance of Myo1g during B cell migration.

. Although Myo1g has been studied in several B lymphocytes' functions, its participation in the early stages of transmigration to secondary lymphoid organs has not been evaluated. In previous work, we observed that LPS + IL-4 activated Myo1g −/− B lymphocytes had reduced migration in vitro and lower polarization of CXCR5 in the membrane 17,21 . In mature B lymphocytes, the expression of CXCR5 and its function is conserved during the B lymphocyte homing to LNs 2, 23 . Thus, we analyzed whether Myo1g regulates LPS + IL-4 activated B lymphocyte motility on the HEV by IVM of CXCL13-stimulated inguinal LNs. In Fig. 1a,b, we show a schematic representation of HEV venules and their diameter. Activated Myo1g −/− B lymphocytes showed increased cell flux only through venules I and II compared to WT (Fig. 1c,d). Of note, fewer activated B lymphocytes performed slow rolling in the absence of Myo1g with significant changes in venules II-IV (Fig. 1e). Those rolling Myo1g −/− B lymphocytes in venules II-IV showed significantly higher rolling velocities than WT B lymphocytes (Fig. 1f). Higher rolling velocities in these venules translated into less firm adhesion (Fig. 1g) and reduced numbers of activated Myo1g −/− B lymphocytes that transmigrated into inguinal LNs (Fig. 1h). This result agrees with what we previously reported that in the absence of Myo1g −/− LPS + IL-4 activated B lymphocytes reached lymph nodes less efficiently 20,21 . Besides, in an in vitro model of CXCL12-dependent migration, activated Myo1g −/− B lymphocytes had a lower percentage of transmigration (Supplementary Figure S1), suggesting that the same migration defect in vivo could be present with other chemokines. Representative videos can be found as Supplementary Videos S2, S3. B lymphocytes require Myo1g for the expression of specific adhesion molecules upon activation. Due to the previous result and because the adhesion of B lymphocytes to the endothelium depends on the adhesion molecules, we decided to evaluate the surface expression of CD62L, LFA-1, and VLA-4 on WT and Myo1g −/− B lymphocytes. Under basal conditions, we did not observe differences in the expression of these molecules. However, when we activated B cells with LPS and IL-4, Myo1g −/− B lymphocytes showed reduced expression of CD62L, LFA-1, and VLA-4 in comparison with WT B lymphocytes ( Fig. 2a-c). These results could explain the reduced adhesive interactions with the endothelium in HEV II, III, and IV observed in vivo (Fig. 1). Interestingly, we did not observe differences in the expression of CXCR5, CXCR4 and CCR7 (Fig. 2d-f) between resting and LPS + IL-4 activated WT and Myo1g −/− B lymphocytes, suggesting that Myo1g −/− B cells are in principle able to respond to CXCL13 gradient as well as other chemokines. Representative histograms can be found as Supplementary Figure S4.
The absence of Myo1g alters the 3D migratory behavior of activated B lymphocytes. For the evaluation of cell motility, there are 2D and 3D migration models. In both models, cells' intrinsic factors participate, such as cell contractility, adhesion, and cytoskeleton rearrangements. Both models resemble conditions in vivo. However, 2D migration involves a process dependent on integrins and protrusion-adhesion-contraction. In contrast, in the case of 3D migration, a treadmilling effect has been seen where the actin flow is continuous 24,25 . To analyze the motility of B lymphocytes in a confined microchannel, we used a custom-made microfluidic chamber. These devices, coated with ICAM-1 or Fibronectin, have been used to imaging lymphocytes "walking" through brief contacts with the microchannel walls during migration 26 . Microchannels connected two chambers with different widths (5, 10, and 15 μm), and CXCL13 was used as a chemoattractant to promote migration of cells from one chamber to the other (Fig. 3a). We observed that displacement could be better observed in 10 μm wide channels, suggesting that 15 μm channels were too broad and 5 μm channels too narrow for cell displacement. When analyzing LPS + IL-4 activated B lymphocytes that traveled through 10 μm microchannels, we observed that activated B lymphocytes "walk" interacting with the Fibronectin-coated microchannel walls, as reported for similar devices 26,27 . When we analyzed activated Myo1g −/− B lymphocytes, we observed reduced contact with the microchannel walls, affecting their speed (Fig. 3b,c). Figure 3d shows a series of representative time-lapse photographs of activated Myo1g −/− and WT B lymphocytes displacement. In summary, these experiments indicated that Myo1g might function in controlling the interaction of activated B lymphocytes to the substrate for confinement-optimized motility. We observed similar behavior with CXCL12 as a chemoattractant (Supplementary Figure S5). Previous work has demonstrated that in the absence of Myo1g, T cells adhere less efficiently to ICAM-1 and migrate faster than WT T cells 27 , suggesting that the absence of Myo1g reduces the adhesive abilities of T and B lymphocytes. Similarly, the absence of Myo1g decreases the adhesion of B lymphocytes to other substrates 20,21 . As a whole, these results could explain why we observed reduced adherence to HEV as well as lower transmigration of Myo1g −/− B lymphocytes shown in Fig. 1. Representative videos can be found as Supplementary Videos S6, S7.
Actin cytoskeletal remodeling upon activation depends on Myo1g. Lamellipodia and filopodia favor the initial interaction between B cells and the endothelium. Furthermore, the B cells must adopt a migratory phenotype and perform spreading on the endothelial cells to increase the contact area before starting to migrate 28 . Besides, B lymphocytes must modify their morphology during chemotactic migration, allowing amoeboid-like movements. Myo1g is known to regulate actin cytoskeletal dynamics and the plasma membrane's elasticity, critically involved in cytoskeletal changes 11,20,29,30 . Therefore, we analyzed the distribution of Myo1g under conditions that favor a migratory phenotype. We incubated LPS + IL-4 activated B lymphocytes under a CXCL13 chemotactic gradient using a Zigmond chamber and analyzed migration for 30 min. Figure 4a shows how Myo1g accumulates at the leading edge of migrating activated B cells. Next, we evaluated whether  (Fig. 4c,d). To measure the cell's curvature, we used the length/width ratio (elliptic factor). An elliptic factor > 2 indicates an elongated morphology, which is characteristic of a moving lymphoid cell. On Fibronectin-coated slides, activated Myo1g −/− B lymphocytes had a lower elliptical factor compared with activated WT B lymphocytes. Of note, no difference was observed on Poly-l-Lysine (Fig. 4e). These results suggest that in the absence of Myo1g, activated B lymphocytes www.nature.com/scientificreports/ have a reduced ability to modify their cytoskeleton and adopt a migratory phenotype. The differences observed in the assays described above when Fibronectin or Poly-l-Lysine coated slides were used also suggest the participation of integrins. Myo1g has been shown to regulate membrane tension and thus form membrane projections 31 , we decided to evaluate the capacity to generate membrane projection in non-activated and activated B lymphocytes. We activated B cells on coverslips coated with Fibronectin. Figure 5a shows that both resting and LPS + IL-4 activated Myo1g −/− and WT B lymphocytes can generate filopodia-like structures. However, the number of membrane projections and their lengths were significantly lower in Myo1g −/− B lymphocytes in both conditions (Fig. 5b,c). Besides, evaluating curvature of B lymphocytes in both conditions using the elliptical factor, Myo1g −/− B lymphocytes had a lower index, suggesting rounder shape (Supplementary Figure S8). These results suggest that B lymphocytes have defects in cytoskeletal dynamics in the absence of Myo1g. As we also observed that Myo1g participates in regulating the surface expression of adhesion molecules such as CD62L, VLA-4, and LFA-1

Discussion
In this work, we used an in vivo model to analyze the participation of Myo1g during the migration of activated B lymphocytes. We observed that in the absence of Myo1g, LPS + IL-4 activated B lymphocytes had reduced capacity to interact with the endothelium of postcapillary venules II, III, and IV in inguinal LNs. The deficiency was manifested by increased rolling velocities and reduced adhesion and transmigration. In the absence of Myo1g, activated B lymphocytes reached fewer lymph nodes, indicating this myosin's relevance in B cells' homing 20,21 .
We reported recently that another class 1 myosin (Myo1e) regulates adhesion and migration of B lymphocytes by mobilizing FAK and allowing the activation of the FAK-PI3K-Rac1 pathway 32 . Myo1e has an SH3 domain that allows for protein-protein interactions 13,16,30,32 . However, Myo1g lacks an SH3 domain. Therefore, the mechanism by which Myo1e and Myo1g participate in the migration of B lymphocytes should be different. Myo1c, another short tail class I myosin present in B lymphocytes' microvilli, takes part in B lymphocytes' ability to modify their cytoskeleton 33 . These results suggest that although short-tailed class I myosins (such as Myo1c and Myo1g) lack an SH3 domain, they are relevant for the actin cytoskeleton's plasticity in B lymphocytes by a still poorly defined mechanism. We used microfluidic devices to characterize further activated Myo1g −/− B lymphocytes; these devices allowed us to analyze 3D migration in vitro 26,27 . The results showed that activated Myo1g-deficient B lymphocytes moved faster within a 10 μm channel. Previous work has demonstrated that in the absence of Myo1g, T cells adhere less efficiently to ICAM-1 and migrate faster than WT T cells 27 , suggesting that the absence of Myo1g reduces the adhesive abilities of T and B lymphocytes. In 3D migration models, round morphology favors displacement, so activated Myo1g −/− B lymphocytes may move more easily. Another possibility in the 3D model is that activated Myo1g −/− B lymphocytes have reduced expression of VLA-4, a Fibronectin ligand, the substrate that covers our device. Then, the interaction could be weak allowing an increased speed in migration. However, VLA-4 is not the only Fibronectin' receptor present on B lymphocytes, so other receptors should be evaluated to rule out their participation 34 . Similarly, the absence of Myo1g decreases the adhesion of B lymphocytes to other substrates 20,21 . Myo1g also regulates the recycling of lipid raft-associated receptors 21 .
Myo1g has a PH-like domain that allows it to interact with lipid rafts 19,22 , and several adhesion molecules such as selectins, integrins are present in these lipid domains. Therefore, problems with mobilization or recycling www.nature.com/scientificreports/ could be associated with lower adhesion molecules' expression in activated Myo1g −/− B lymphocytes. However, it remains to be investigated whether Myo1g can also regulate the signaling pathways promoting these adhesion molecules' de novo expression. It has been observed that Myo1g regulates actin cytoskeletal dynamics and the elasticity of the plasma membrane. These factors favor the generation of lamellipodia and filopodia, which are critically involved in these processes 11,29 . Our results showed that Myo1g is concentrated at the poles of the cells when adopting a migratory phenotype. During cell migration, the changes generated in the cytoskeleton require a force to be exerted on the membrane. This force is provided by actin filament dynamics, promoting cell shape modifications and forming membrane structures such as filopodia and lamellipodia 10,12,35 . These mechanical forces exerted on the membrane are regulated by the membrane's tension regulated by various proteins, including Myo1g 14,29,36 . In previous work, we reported that in the absence of Myo1g, LPS + IL-4 activated B lymphocytes had decreased membrane tension 17 . In addition to regulating elasticity, membrane tension must control the strength between the cytoskeleton and adhesion molecules through a clutch that joins the actin filaments with the integrins 10,37,38 . Moreover, it has been reported that membrane tension acts through the PLD2-mTORC2 pathway to inhibit actin nucleation via the WAVE2 complex. These molecules could also participate in the signaling that favors rearrangements in the cytoskeleton [39][40][41] . Therefore, we propose that when membrane tension of activated B lymphocytes is reduced (by the absence of Myo1g), the cells have less capacity to produce stable membrane structures to convert them into a migratory phenotype. In conclusion, here we present pieces of evidence about the role of Myo1g in regulating the adhesion of B lymphocytes to HEV, affecting the formation of membrane projections necessary for a migratory phenotype. Thus, the absence of Myo1g decreases B lymphocytes' interaction with the endothelium, which, together with the lower expression of adhesion molecules, reduces activated B lymphocytes' capacity to enter inguinal LNs. Additionally, in this work, we show evidence for the role of Myo1g in B lymphocyte migration in vitro and in vivo. Myo1g regulates activated B lymphocytes' adhesion to the endothelium of HEV and membrane tension that affects the cell membrane's elasticity, the formation of membrane projections, and the acquisition of a migratory phenotype. Finally, our findings give information about how Myo1g regulates cytoskeleton rearrangements during the B lymphocyte CXCL13-dependent migration process. Though our in vivo model was focussed on LN homing, lymphocyte migration is a conserved process. Hence, the phenomenon triggered by the absence of Myo1g could be present when Myo1g −/− B lymphocytes migrates toward other sites, for example, Peyer' Patches, where homing is CXCL13-dependent 42 . It will be interesting to analyze the functional effect of the activated Myo1g −/− B lymphocytes' irregular migration. In Fig. 6, a schematic illustration is shown that combines the results of this work with previous reports about the function of Myo1g in murine B lymphocytes. In this way, a model is proposed in which Myo1g could regulate different functions during the migration of B lymphocytes to the inguinal lymph node.

Lymphocyte isolation and flow cytometry. Splenic mononuclear cells were isolated by Ficoll-paque
Plus (GE Healthcare, Little Chalfont, UK) density gradient separation, and B220 + cells were enriched by panning, using plastic dishes coated with α-Thy-1 monoclonal antibody ascites (NIM-R1) (Chayen and Parkhouse, 1982). For B-cell activation, 2 × 10 6 cells were incubated 48 h at 37 °C, and 5% CO 2 in 1 ml RPMI 1640 (Life Technologies, Grand Island, NY) supplemented with 10% fetal bovine serum (Thermo Fisher Scientific, Waltham, MA). Stimulation was induced by 48 h incubation with LPS from Escherichia coli O55: B5 (20 μg/ml) (Sigma-Aldrich, St Louis, MO) plus the addition of IL-4 (10 U/ml) (R&D Systems, Minneapolis, MN). This methodology was carried out following a previously published protocol 32 . For immunostaining, the Fc receptors were blocked using 10% goat serum. Cell suspensions were washed with PBS containing 1% bovine serum albumin (BSA) (Thermo Fisher Scientific) and 0.01% NaN 3 (PBA). One million cells were stained 15 min using the antibodies described in the following section. After incubation, the cells were washed with PBA and fixed with 1% formaldehyde in PBS (0.5% albumin, 0.01%NaN3, 100 ml PBS). Doublets were excluded by gating on FSC-H versus  www.nature.com/scientificreports/ Migration-phenotype assay. A Zigmond chamber (Neuroprobe, Gaithersburg, MD) was used to quantify the migratory phenotype. Briefly, 1 × 10 6 LPS + IL-4 activated B lymphocytes from WT and Myo1g −/− were suspended in 0.5 ml RPMI 1640 supplemented with 10% fetal bovine serum immediately plated onto glass coverslips. The glass coverslips were previously coated with Fibronectin (2.5 μg/ml) (Sigma-Aldrich) or Polyl-Lysine (Sigma-Aldrich), and the cells were incubated 30 min at 37 °C and 5% CO 2 to allow attachment. The coverslips, with the cells attached, were gently washed with PBS. One of the grooves in the Zigmond chamber was filled with supplemented medium (∼70 μl), and the other with CXCL13 (2.5 μg/μl) dissolved in the supplemented medium. Subsequently, cells were incubated for 1 h at 37 °C and 5% CO 2 to allow sensing of the CXCL13 gradient. Migration tracks of at least 50 activated lymphocytes of WT and Myo1g −/− mice, in three independent experiments, were analyzed using ImageJ software (National Institutes of Health, Bethesda, MD). This methodology was carried out following a previously published protocol 21 .
Membrane-projections assay. 1 × 10 6 LPS + IL-4 activated B lymphocytes from WT and Myo1g −/− mice were suspended in 0.5 ml RPMI 1640 supplemented with 10% fetal bovine serum immediately plated onto glass coverslips coated with Fibronectin (2.5 μg/ml) (Sigma-Aldrich). Cells were incubated one h at 37 °C and 5% CO 2 to allow adhesion. Subsequently, cells were fixed and stained with TRITC-Phalloidin for 20 min. At least 50 activated B lymphocytes of WT and Myo1g −/− mice, in three independent experiments, were analyzed using ImageJ software (National Institutes of Health, Bethesda, MD). This methodology was carried out following a previously published protocol 21 .
Trans-well assay. 1.5 × 10 4 LPS + IL-4 activated WT and Myo1g −/− B lymphocytes were placed in the upper compartment of 5 μm pore-trans-well chamber (Corning) in RPMI 1640 (Life Technologies) without Fetal Calf Serum (FCS). Ten minutes before seeding the cell suspension in the top compartment, we added FCS or CXCL12 (100 ng/ml) to the lower well. The assembled chamber was incubated 4 h at 37 °C and 5% CO 2 . After incubation, the cells in the lower compartment were counted, and we calculated the percentage of migrating cells from the number of initial cells seeded in the top chamber.
Statistical analysis. Data are presented as the arithmetic mean with standard deviations; Student's t-test was used for evaluating statistical differences. A p-value of < 0.05 was considered statically significant. The p values are represented as *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001, and the number of samples or cells (n) used are mentioned in each figure legend. This analysis was carried out following a previously published protocol 21 .

Data availability
Raw data used and analyzed during the current study are available from the corresponding author on reasonable request.