To investigate the possibility that LFA-1 ligation might also promote B-cell activation by triggering a significant change in synaptic actin organization, we imaged F-actin at ISs formed by primary mouse B cells on glass surfaces coated with either anti-IgM or anti-IgM plus ICAM-1. F-actin was visualized using GFP-F-Tractin, a dynamic reporter for F-actin (Yi et al., 2012; Murugesan et al., 2016), and two super-resolution imaging modalities: Airyscan (xy resolution ~140 nm) and total internal reflection fluorescence-structured illumination microscopy (TIRF-SIM; xy resolution ~100 nm). Individual video frames of anti-IgM-engaged B cells using both imaging modalities (Figure 1A and B), together with the corresponding videos (Video 1A and B), revealed a thin, bright, highly dynamic outer rim of F-actin (white arrows in Figure 1A and B) that likely corresponds to the branched actin network comprising the dSMAC (Bolger-Munro et al., 2019; Wang et al., 2018; Wang et al., 2017). Both modalities (but especially TIRF-SIM) showed that the F-actin present inside this outer dSMAC rim is composed of a highly disorganized mixture of short-actin filaments/fibers and actin foci (blue brackets in Figure 1A and B), similarly to those observed previously in HeLa cells (Fritzsche et al., 2017). In sharp contrast, individual video frames of anti-IgM + ICAM-1-engaged B cells using both modalities (Figure 1C and D), together with the corresponding videos (Video 2A and B), showed a highly organized network inside the outer dSMAC rim (i.e., in the pSMAC) that is comprised of concentric actin arcs (blue brackets and magenta arrows in Figure 1C and D). The difference in synaptic actin organization between anti-IgM-engaged B cells and anti-IgM + ICAM-1-engaged B cells is very evident in enlarged TIRF-SIM images. While it is challenging to define SMAC boundaries and any pattern of F-actin organization in the pSMAC of B cells engaged with anti-IgM alone (Figure 1E1 and E2), SMAC boundaries and pSMAC F-actin organization are both very distinct in B cells engaged using anti-IgM + ICAM-1 (Figure 1F1 and F2). Consistently, scoring B cells for the presence of any discernable arcs showed that the addition of ICAM-1 increases the percentage of such cells from ~30% to ~70% (Figure 1G). Importantly, static and dynamic imaging showed that the arcs in cells engaged with anti-IgM alone are sparse and transient (Figure 1A and B, Video 1A and B), while those in cells engaged with both anti-IgM and ICAM-1 are dense and persistent (Figure 1C and D, Video 2A and B). In other words, when B cells receiving only anti-IgM stimulation do form discernible arcs (e.g., see those marked by magenta arrows in Figure 1A and B), they are much sparser and less persistent than those formed by cells also receiving ICAM-1 stimulation. Moreover, we could not find any B cells receiving anti-IgM stimulation alone that possessed a robust actin arc network. Consistently, measuring the degree of alignment between actin filaments in the pSMAC portion of B cells stimulated with anti-IgM alone versus both anti-IgM and ICAM-1, which were made using FibrilTool (Boudaoud et al., 2014), revealed a large shift towards more organized pSMAC actin when ICAM-1 was included (Figure 1—figure supplement 1A1–A3; see the figure legend for details). Finally, measuring the percentage of total synaptic F-actin content within each SMAC (Figure 1H), and the percentage of total IS footprint occupied by each SMAC (Figure 1I), showed that the actin arc-containing pSMAC comprises the major actin network at the IS of primary B cells engaged using both anti-IgM and ICAM-1. Together, these results demonstrate that LFA-1 co-stimulation promotes the formation of a pSMAC actin arc network that dominates the B-cell IS.

We next sought to define the origin of the actin arcs that comprise the pSMAC of B cells stimulated using both anti-IgM and ICAM-1. Primary B cells stimulated in this way exhibit small, actin-rich surface spikes at the outer synapse edge (Figure 2A). Importantly, magnified images revealed that the actin within these spikes continues into the cytoplasm in the form of linear actin filaments (Figure 2B1 and B2). Moreover, tracing these linear actin filaments showed that they are contiguous with the pSMAC actin arcs (Figure 2C1 and C2, Video 3A and B). These results argue that linear actin filaments nucleated at the plasma membrane at the outer edge of the synapse give rise to the actin arcs populating the pSMAC. While these results do not identify the specific nucleator involved, they do point to it being a member of the formin family based on the fact that the actin being made is linear and nucleated at the plasma membrane (Goode and Eck, 2007; Breitsprecher and Goode, 2013). Consistent with this conjecture, and with the fact that formins incorporate fluorescent protein-labeled actin monomer into filaments poorly (Yi et al., 2012; Murugesan et al., 2016; Chen et al., 2012), we did not see fluorescent actin arcs in B cells expressing mEOS-labeled G-actin (Video 4).

To test if a formin is indeed responsible for creating the pSMAC actin arc network, we used the pan-formin inhibitor SMIFH2 (Rizvi et al., 2009). Figure 2D1 and D2, together with the line scan in Figure 2E, show that the pSMAC actin arcs present in a representative primary B cell immediately before SMIFH2 addition (blue trace) had largely disappeared 6 min after adding SMIFH2 (magenta trace). Given recent concerns about the specificity of SMIFH2 (Nishimura et al., 2021), we used three different miRNAs to knock down the formin mDia1 in the lymphoma B cell line A20 (Figure 2—figure supplement 1A and B), which also forms pSMAC actin arcs when stimulated using anti-IgG + ICAM-1 (Figure 1J1 and J2, Video 5A). mDia1 was chosen as the miRNA target as it is highly expressed in B cells (ImmGen Database) and is largely responsible for making actin arcs in T cells (Murugesan et al., 2016). Compared to control A20 B cells (Figure 2F1), representative B cells expressing each of the three miRNAs (Figure 2F2–F4) were largely devoid of actin arcs. This difference was supported by quantitating the ratio of pSMAC to dSMAC F-actin (Figure 2G), as well as the amount of F-actin in the pSMAC (Figure 2H). Finally, actin arcs were unaffected by the expression of two different nontargeting miRNAs (Figure 2—figure supplement 1C1–C4 and D1–D4). Together, these results argue that the pSMAC actin arcs are indeed created by a formin, and that the formin mDia1 likely plays a major role.

To provide further evidence that the arcs are created by a formin, we imaged A20 B cells following the addition of the Arp2/3 inhibitor CK-666. The rationale for this experiment lies in the recent revelation that the two major consumers of actin monomer in cells, the Arp2/3 complex and formins, are always competing for a limiting pool of actin monomer (Burke et al., 2014; Lomakin et al., 2015; Fritzsche et al., 2016; Hammer et al., 2019). One consequence of this competition is that when one of these nucleators is inhibited, the actin structures created by the other nucleator get more robust because that nucleator now gets more monomer. For example, inhibiting the Arp2/3 complex promotes the formation of formin-dependent actin networks in both yeast and vertebrate cells (Murugesan et al., 2016; Burke et al., 2014; Lomakin et al., 2015; Fritzsche et al., 2016; Hammer et al., 2019). Given this, and assuming that the arcs in B cells are formin-generated, then inhibiting the Arp2/3 complex in B cells should lead not only to a diminution of the branched actin network in the dSMAC, but also to an amplification of the arc network in the pSMAC. Consistently, Figure 2—figure supplement 2A1/A2 (before CK-666 addition) and Figure 2—figure supplement 2A3/A4 (after CK-666 addition) together show that CK-666 addition leads not only to a reduction in the size of the dSMAC (magenta brackets), but also to an increase in arc content in the pSMAC (blue brackets). These changes were supported by measuring the percentage of total synaptic F-actin content residing within each SMAC (Figure 2—figure supplement 2B), which revealed a significant shift away from dSMAC F-actin and toward pSMAC F-actin following CK-666 treatment. This shift was also reflected in measurements of total pSMAC F-actin content (Figure 2—figure supplement 2C), the ratio of pSMAC to cSMAC F-actin content (Figure 2—figure supplement 2D), and the ratio of pSMAC to cSMAC area (Figure 2—figure supplement 2E). Taken together, these data argue strongly that linear actin filaments generated by the formin mDia1 at the outer edge of the synapse give rise to the pSMAC actin arc network.

Having established that ICAM-1 co-stimulation promotes the formin-dependent formation of actin arcs in the pSMAC, we asked how these arcs are organized into concentric structures. Formin-derived linear actin filaments are commonly organized into well-defined structures such as stress fibers, transverse arcs, and the contractile ring in dividing cells by bipolar filaments of the actin-based motor protein myosin 2 (Vicente-Manzanares et al., 2009; Sellers, 2000; Shutova and Svitkina, 2018). We decided, therefore, to test whether myosin 2 co-localizes with the actin arcs and is required for their concentric organization.

To define the localization and dynamics of myosin 2 at the B-cell IS, we used primary B cells isolated from a mouse in which GFP had been knocked into the N- terminus of the myosin 2A heavy chain gene Myh9 (referred to herein as M2A) (Zhang et al., 2012) as M2A is the only myosin 2 isoform expressed in B cells (ImmGen Database). Individual video frames of these cells following transfection with Td-Tomato-F-Tractin and attachment to coverslips coated with anti-IgM and ICAM-1 revealed a dramatic co-localization between M2A and the actin arcs in the pSMAC (Figure 3A1–A3, Video 6). Magnified TIRF-SIM images show that the myosin signals align with actin arcs in a periodic fashion (Figure 3A4) that resembles other myosin 2-rich, linear actin structures like stress fibers and the contractile ring (Beach et al., 2014). Moreover, these myosin signals exhibit the SIM signature for M2A bipolar filaments when M2A is GFP-labeled at its N-terminus (Beach et al., 2017), which is a pair of GFP puncta spaced ~300 nm apart (Figure 3A5; 304 ± 32 nm; n = 230 filaments from 12 cells). The presence of M2A filaments in the medial portion of the synapse was also evident in primary B cells isolated from a mouse in which mCherry had been knocked into the N-terminus of M2A (Figure 3—figure supplement 1A), primary B cells that we genome-edited using CRISPR to place GFP at the N-terminus of M2A (Figure 3—figure supplement 1B), and A20 B cells that we genome edited using CRISPR to place mScarleti at the N-terminus of M2A and then transfected with GFP-F-Tractin (Figure 3—figure supplement 1C1–C3, Video 5B). Finally, 3D-SIM images of A20 B cells that were fixed and stained for M2A and F-actin showed that endogenous M2A also co-localizes with the actin arcs (Figure 3—figure supplement 1D1–D3; note that the signature for M2A filaments using this antibody, which recognizes the C-terminus of M2A, is a single fluorescent punctum that corresponds to the center of an individual M2A filament) (Beach et al., 2014; Beach and Hammer, 2015). The extent of this co-localization was even clearer in enlarged images of immunostained cells (Figure 3—figure supplement 1E1–E3), where line scans showed endogenous M2A coinciding with actin arcs (Figure 3—figure supplement 1F). Together, these results show that the actin arc network in primary B cells receiving ICAM-1 co-stimulation is in fact an actomyosin arc network.

To gain insight into how the arcs become decorated with M2A filaments, we examined time-lapse TIRF-SIM images of GFP-M2A knockin primary B cells expressing Td-Tomato F-Tractin. Individual video frames (Figure 3B1–B6), as well as the corresponding video (Video 7), show that bipolar filaments of M2A begin to appear near the dSMAC:pSMAC boundary in association with the linear actin filaments/bundles exiting the dSMAC (white, yellow, and fuchsia arrowheads mark such myosin filaments at time 0 s in Figure 3B1). As time progresses, these filaments move centripetally and undergo expansion into filament clusters (Figure 3B1–B6; see also Video 7). This expansion, in which individual myosin filaments expand into a small cluster of filaments, is presumably driven by the same sequential amplification pathway described previously for M2A filament assembly in fibroblasts (Beach et al., 2017). Finally, the myosin filaments in these clusters begin to align with the arcs forming at the outer edge of the pSMAC, which then merge with the larger actomyosin arc network in the pSMAC (Figure 3B1–B6). As all this is happening, new myosin filaments keep appearing near the dSMAC:pSMAC boundary to repeat the process (Figure 3B2–B6; follow the blue, green, and purple arrowheads).

Given that ICAM-1 co-stimulation promotes the formation of actin arcs and that the arcs recruit M2A, ICAM-1 co-stimulation should also result in an increase in the amount of M2A at the IS. Consistently, primary GFP-M2A knockin B cells receiving both anti-IgM and ICAM-1 stimulation exhibited a greater amount of synaptic M2A than B cells receiving only anti-IgM stimulation (Figure 3C–E). Of note, this difference remained significant even after normalizing the M2A fluorescence for a small difference in the average cell-spread area under these two conditions (Figure 3—figure supplement 1G1–G2).

The organization of formin-generated linear actin filaments into well-defined structures is typically driven by the contractility of myosin 2 filaments (Vicente-Manzanares et al., 2009; Sellers, 2000). Therefore, we asked if M2A contractility is required for the concentric organization of the pSMAC actin arcs by treating cells with para-nitroblebbistatin (pnBB), a blue light-insensitive version of the cell-permeable, small molecule myosin 2 inhibitor blebbistatin (BB) that blocks myosin 2-based contractility by locking the myosin in its weak actin binding state (Képiró et al., 2014). While control, DMSO-treated cells exhibited concentric actin arcs in their pSMAC as expected (Figure 3F), cells treated with 25 μM pnBB displayed highly disorganized, mesh-like actin arrays in their pSMAC (Figure 3G). Consistently, anisotropy measurements made using FibrilTool revealed a dramatic shift towards more disorganized pSMAC actin when B cells are treated with pnBB (Figure 3H). Together, these results demonstrate that M2A contractility is indeed required for the concentric organization of the pSMAC actin arcs.

We used traction force microscopy in combination with pnBB to ask if integrin-dependent traction forces that B cells exert on a deformable substrate require M2A contractility. As expected (Wang, 2018; Kumari et al., 2019), B cells engaged with substrate coated with anti-IgM and ICAM-1 generated significantly more traction force than B cells engaged with substrate coated with anti-IgM alone (Figure 3—figure supplement 2A1, A2, B1, B2, and D). Importantly, ICAM-1-dependent traction forces were completely abrogated by pretreating the cells with pnBB (Figure 3—figure supplement 2C1, C2, and D), indicating that the generation of integrin-dependent traction forces requires M2A contractility. This requirement likely reflects pulling forces exerted by M2A on the substrate through LFA-1:ICAM-1 pairs, combined with the increase in M2A content at the synapse caused by ligating LFA-1 with ICAM-1, and the contribution that M2A-dependent pulling forces make in keeping LFA-1 in its open, active conformation (Gardel et al., 2010; Case and Waterman, 2015; Comrie and Burkhardt, 2016). These results, together with the fact that integrin clusters are known to accumulate in the pSMAC portion of the B-cell IS (Carrasco et al., 2004; Carrasco and Batista, 2006b), suggest a feed-forward relationship where integrin ligation promotes the formation of pSMAC actomyosin arcs, and the contractile forces exerted by these actomyosin arcs promote further integrin activation and robust adhesion in the pSMAC.

Inward flows of cortical actin networks are thought to drive the transport of antigen receptor clusters to the center of maturing synapses in both T cells and B cells (Wang and Hammer, 2020; Hammer et al., 2019; Blumenthal and Burkhardt, 2020; although see Schnyder et al., 2011; Babich and Burkhardt, 2011). For B cells, the clearest example of this to date is the demonstration that the centripetal flow of the branched actin network comprising the dSMAC propels BCR:antigen clusters towards the cSMAC (Bolger-Munro et al., 2019). As a prelude to asking whether the actomyosin arcs comprising the pSMAC also contribute to antigen centralization, we asked if this contractile network exhibits centripetal flow. Kymograph analyses of actin flow across synapses made by primary B cells expressing GFP-F-Tractin showed that their pSMAC actomyosin arc network indeed flows centripetally at 1.07 ± 0.07 µm/min, or about one-third the rate of centripetal actin flow in the dSMAC (2.89 ± 0.18 µm/min) (Figure 4—figure supplement 1A1–A3). Similar results were obtained for A20 B cells (pSMAC rate: 0.97 ± 0.13 µm/min; dSMAC rate: 3.16 ± 0.35 µm/min) (Figure 4—figure supplement 1B1–B3). Together, these results indicate that the actomyosin arcs could contribute along with the branched actin network in the dSMAC to the inward transport of BCR:antigen clusters.

We used PLBs to determine if the actomyosin arcs do in fact contribute to antigen centralization. As expected, primary B cells expressing GFP-F-Tractin readily formed actin arcs when PLBs contained both anti-IgM and ICAM-1 (Video 8A), but not when they contained anti-IgM alone (Video 8B). Also as expected, primary B cells engaged with PLBs containing fluorescent anti-IgM (magenta) and unlabeled ICAM-1 yielded mature synapses in which concentric actin arcs surrounded antigen accumulated in the cSMAC (Figure 4A1–A3, white arrows). To obtain a holistic view of antigen centralization, we imaged antigen clusters in the dSMAC and pSMAC of primary B cells over time with the aim of correlating their rates of centripetal transport with the distinct rates of centripetal actin flow exhibited by these two IS zones (Video 9). Tracking of single-antigen microclusters showed that they moved inward at 2.36 ± 1.1 µm/min and 1.03 ± 0.3 µm/min across the dSMAC (magenta tracks) and pSMAC (green tracks), respectively (Figure 4B and C). Importantly, these rates are very similar to the rates of centripetal actin flow across the dSMAC and pSMAC, respectively (Figure 4—figure supplement 1A1–A3). Together, these observations argue that the pSMAC actomyosin arc network works together with the dSMAC branched actin network to drive antigen centralization.

To identify the mechanism by which the actomyosin arcs drive antigen centralization, we imaged F-actin and anti-IgM in the medial portion of forming synapses at high magnification using TIRF-SIM. Anti-IgM microclusters were seen to move across the pSMAC towards the cSMAC (which in the following images was in the down direction) while embedded in an arc network moving in the same direction (Video 10). White lines in Video 10 and in the corresponding still images in Figure 4D1–D6 mark actin arcs that were sweeping an individual anti-IgM microcluster inward (Figure 4E1–E6). Figure 4F shows the trajectory of this microcluster (temporally color-coded) as it moved towards the cSMAC. Finally, a kymograph of this trajectory (Figure 4G) shows that several actin arcs contributed to the inward movement of this microcluster (areas bracketed in white), and that pauses in movement (areas bracketed in pink) occurred where no actin signal was immediately adjacent to the microcluster. Together, these results argue that individual actin arcs move individual BCR:antigen microclusters inward via a sweeping mechanism that likely depends on frictional coupling between the actin arc and microcluster (Yu et al., 2010; Ditlev et al., 2019; Smoligovets et al., 2012). While arcs can slip past microclusters, the overall incidence of such slippage must be fairly small as the rate of inward antigen transport across the pSMAC (Figure 4C) is not significantly slower than the rate of inward actin arc flow across the pSMAC (Figure 4—figure supplement 1A3).

B cells engaged with membrane-bound antigen at low density fail to centralize antigen unless their integrin LFA-1 is also engaged with ICAM-1 in the target membrane (Carrasco et al., 2004). As a prelude to investigating the myosin dependence of this integrin co-stimulatory effect, we sought to recapitulate these findings using primary B cells and PLBs containing varying amounts of mobile, fluorophore-labeled anti-IgM antibody in the presence or absence of unlabeled ICAM-1. Using this approach, we determined an amount of anti-IgM antibody that would not elicit robust antigen centralization in the absence of ICAM-1, but would in its presence. B cells exhibited robust antigen centralization/cSMAC formation over 10 min without the need for ICAM-1 when the PLB was loaded using a solution containing anti-IgM at a concentration of 2 µg/ml (hereafter referred to as ‘high-density antigen’) (Figure 5—figure supplement 1A1–A3). By contrast, B cells formed antigen microclusters across their synaptic interface but failed to centralize them over 10 min when the PLB was loaded using a solution containing anti-IgM at a concentration 0.15 µg/ml (hereafter referred to as ‘low or limiting density antigen’) (Figure 5—figure supplement 1B1–B3). Importantly, when unlabeled ICAM-1 was included in these low-density antigen bilayers, B cells now exhibited robust antigen centralization/cSMAC formation (Figure 5—figure supplement 1C1–C3). This co-stimulatory effect was supported by scoring antigen distribution as centralized, partially centralized or noncentralized (Figure 5—figure supplement 1D1–D3 and E). It was also supported by scoring the percent of total synaptic antigen present within the cSMAC, which was defined by a circular area encompassing 20% of the entire synaptic interface and centered around the center of mass of the fluorescent antigen-containing pixels within the interface (Figure 5—figure supplement 1F). Finally, it was supported by measuring the size of antigen clusters as a function of their distance from the center of the cSMAC (defined as above) (Figure 5—figure supplement 1G). Specifically, B cells engaged with PLBs containing antigen at the limiting density and no ICAM-1 exhibited small antigen clusters (~0.3 µm²) located roughly evenly across the synaptic interface (Figure 5—figure supplement 1G, black trace), while B cells engaged with PLBs containing ICAM-1 in addition to antigen at the limiting density exhibited large antigen clusters (up to 3 µm²), the largest of which were located at the center of the cSMAC (Figure 5—figure supplement 1G, green trace). Of note, the total amount of antigen present at the synaptic interface was also greater for cells engaged with low-density anti-IgM + ICAM-1 than for cells engaged with low-density anti-IgM alone (Figure 5—figure supplement 1H). Together, these results recapitulated a central aspect of the integrin co-stimulatory effect described by Carrasco et al., 2004, and they established the specific conditions we used next to test the myosin dependence of this co-stimulatory effect.

To score the myosin dependence of the integrin co-stimulatory effect, we measured the ability of primary B cells treated with either vehicle control (DMSO) or pnBB to centralize antigen and form a cSMAC when engaged for 10 min with PLBs containing ICAM-1 and anti-IgM at the limiting density. While DMSO-treated cells exhibited robust antigen centralization/cSMAC formation (Figure 5A1–A3), pnBB-treated cells failed to centralize antigen/create a cSMAC (Figure 5B1–B3). Consistently, the actin arcs that surround centralized antigen in DMSO-treated cells (Figure 5C1–C3, white arrows) were absent in pnBB-treated cells (Figure 5D1–D3). The fact that myosin inhibition abrogates the integrin co-stimulatory effect was further supported by scoring antigen distribution in control and pnBB-treated cells as centralized, partially centralized, or noncentralized (Figure 5E), byscoring the percent of total synaptic antigen present within the cSMAC (Figure 5F), and bymeasuring the size of antigen clusters as a function of their distance from the center of the cSMAC (Figure 5G). Of note, the total amount of antigen present at the synaptic interface was also greater for cells treated with DMSO than for cells treated with pnBB (Figure 5H). Together, these results show that the ability of integrin ligation to promote antigen centralization and cSMAC formation when antigen is limiting requires myosin contractility. This in turn argues that the contractile actomyosin arc network created downstream of integrin ligation plays an important role in the mechanism by which LFA-1 co-stimulation promotes B-cell activation.

Finally, we were curious if the robust centralization of antigen that occurs in the absence of LFA-1 ligation when the density of antigen is high is also dependent on myosin contractility, at least to some extent. Indeed, we found that treatment with para-amino BB (paBB), a newer, slightly more water-soluble version of BB (Várkuti et al., 2016), attenuated antigen centralization significantly even when the density of antigen was high (Figure 5—figure supplement 2; see legend for details), although the magnitude of the inhibition was smaller than for B cells engaged with limiting antigen plus ICAM-1 (compare the results in Figure 5—figure supplement 2 to the results in Figure 5). We conclude, therefore, that M2A contractility potentiates antigen centralization when antigen density is high as well as when antigen density is low enough that LFA-1 co-stimulation becomes important for IS formation. That said, additional experiments should help define how myosin contributes to antigen centralization in B cells receiving only strong anti-IgM stimulation.

To measure the contribution that actomyosin arcs might make to BCR-dependent signaling, we determined the effect that pnBB has on the distribution and synaptic content of phosphorylated CD79a (P-CD79a), an early signaling molecule responsible for signal transduction downstream of BCR-antigen interaction (Batista et al., 2010; Tanaka and Baba, 2020). Consistent with results above and with the known properties of CD79a, DMSO-treated primary B cells engaged for 10 min with PLBs containing ICAM-1 and limiting antigen and then fixed/stained for P-CD79a exhibited robust cSMAC formation, with P-CD79a and anti-IgM concentrated in the cSMAC (Figure 6A1–A4). Also as expected, pnBB-treated B cells failed to form a clear cSMAC, resulting in CD79a and anti-IgM spread across the synapse (Figure 6B1–B4). Importantly, quantitation showed that pnBB-treated cells also exhibited a significant reduction relative to control cells in synaptic P-CD79a content (Figure 6C). This defect was also seen after only 5 min on PLBs (Figure 6—figure supplement 1A), and the defects at both time points were not due to differences between BB-treated cells and control cells in synaptic CD79a content (Figure 6—figure supplement 1B).

To extend these results, we determined the effect that pnBB has on the distribution and synaptic content of phosphorylated CD19, an important co-receptor for the BCR that is responsible for PI3K activation (Harwood and Batista, 2011; Tuveson et al., 1993; Keppler et al., 2015; Depoil et al., 2008). DMSO-treated primary B cells engaged with PLBs as above exhibited robust cSMAC formation, with P-CD19 enriched at the outer edge of the IgM concentrated in the cSMAC (Figure 6D1–D4). This enrichment of P-CD19 at the pSMAC/cSMAC boundary was confirmed by line scans of the fluorescence intensities for F-actin, anti-IgM, and P-CD19 (Figure 6G, see boxed pSMAC regions). In contrast to control cells, pnBB-treated B cells failed to concentrate anti-IgM at the center of the synapse, and P-CD19 staining was now spread across the synaptic interface (Figure 6E1–E4 and H). Moreover, quantitation showed that pnBB-treated cells also exhibited a significant reduction relative to control cells in synaptic P-CD19 content (Figure 6F) that was not due to a difference in synaptic CD19 content (Figure 6—figure supplement 1C). Together, these results indicate that the actomyosin arcs promote BCR-dependent signaling.

Recent studies have presented evidence that GC B cells differ markedly from naïve B cells with regard to the distribution and fate of antigen at mature synapses. Rather than concentrating antigen at the center of the synapse and using actomyosin force to extract it there, GC B cells accumulate antigen in clusters at the periphery of the synapse and use actomyosin force to extract it there (Hammer et al., 2019; Nowosad et al., 2016; Kwak et al., 2018). These and other results argue that GC B cells differ dramatically from naïve B cells with regard to the organization of actomyosin at their synapse. We wondered, however, if actomyosin arcs could be detected in mouse GC B cells using our imaging approaches. Consistently, TIRF-SIM imaging of mouse GC B cells isolated from the GFP-M2A knockin mouse that were stained with CellMask Deep Red and plated on coverslips coated with anti-IgM, anti-IgG, and ICAM-1 revealed a subset of cells exhibiting enrichment of M2A filaments in the medial, pSMAC portion of the synapse (Video 11), just as in naïve B cells. Moreover, these myosin filaments move centripetally (Video 11) and co-localize with pSMAC actin arcs in phalloidin-stained samples (Figure 7A1–A3, white arrows), just as in naïve B cells. Importantly, scoring showed that about one-third of GC B cells exhibited robust accumulation of M2A filaments in the pSMAC when engaged with anti-IgM/IgG-coated glass (Figure 7B). Similarly, about one-third of GC B cells engaged for 10 min with PLBs containing fluorophore-labeled anti-IgM/IgG and unlabeled ICAM-1, and then fixed and stained with phalloidin, exhibited robust accumulation of M2A filaments in the pSMAC (Figure 7C1–C4 and D,). Importantly, these actomyosin arcs can be seen to surround antigen accumulated at the center of the synapse (see the white arrows in Figure 7C1, C2, and C4). This finding, together with the fact that the myosin moves centripetally during IS formation (Video 11), suggests that actomyosin arcs can contribute to antigen centralization in GC B cells as well as in naïve B cells.

Given these results, we asked if our PLB-engaged mouse GC B cells can centralize antigen. In partial agreement with previous findings (Nowosad et al., 2016; Kwak et al., 2018), ~45% of synapses exhibited small to medium-sized antigen clusters distributed to varying degrees in the synapse periphery (Figure 7E1, E2, and F). In addition, ~20% of synapses exhibited antigen microclusters spread throughout the synaptic interface (Figure 7E3 and F). Importantly, the remaining ~35% of synapses exhibited highly centralized antigen (Figure 7E4 and F). Images of these synapses showed a pSMAC-like accumulation of GFP-M2A surrounding much of the centralized antigen (Figure 7—figure supplement 1A1; see also Video 12). Conversely, images of synapses containing either peripheral antigen clusters or microclusters showed no obvious pattern to the distribution of GFP-M2A (Figure 7—figure supplement 1A2 and A3; see also Video 13). Moreover, the synapses containing peripheral antigen clusters exhibited less total GFP-M2A signal than the synapses with centralized antigen (Figure 7—figure supplement 1B), and the signals appeared more transient (compare Video 13 to Video 12). These results, together with the images in Figure 7C1–C4, argue that GC B cells with centralized antigen (about one-third of cells) are the ones that make actomyosin arcs (again, about one-third of cells). We conclude, therefore, that GC B cells can make actomyosin arcs and that they likely use this structure to centralize antigen, although the degree to which they do this is considerably less than for naïve B cells. We note, however, that our conclusions regarding GC B cells require additional supporting data that include testing the ICAM-1 dependence of actomyosin arc formation and quantitating the contributions that this contractile structure makes to GC B cell traction forces, signaling, and antigen centralization.

Primary B cells were isolated from the spleens of 6–12-week-old C57BL/6 mice (Jackson Laboratories #002595) and GFP-M2A (Myh9) KI mice (gift of R. Adelstein, NHLBI/NIH) of either sex using negative selection B cell isolation (StemCell Technologies). Euthanasia was performed in accordance with protocols approved by the National Human Genome Research Institute Animal Use and Care Committee at the National Institutes of Health. The A20 murine IgG⁺ B cell line was purchased from ATCC (ATCC TIB-208), verified by responsiveness to anti-IgG stimulation, and confirmed to be free of mycoplasma. B cells were cultured in complete medium (RPMI-1640, 10% heat-inactivated fetal calf serum [FCS], 2  mM l-glutamine, 1  mM sodium pyruvate, 50  µM 2-mercaptoethanol, and 1X Antibiotic-Antimycotic) at 37°C with 5% CO₂. Primary B-cell complete media also contains 5 ng/ml of BAFF (R&D Systems).

GFP- and tdTomato-tagged F-Tractin were gifts from Michael Schell (Uniformed Services University, Maryland). Alexa Fluor-conjugated phalloidins were purchased from Thermo Fisher. Anti-mDia1 antibody was purchased from Thermo Fisher (PA5-27607). HRP-conjugated mouse anti-β-actin antibody was purchased from Santa Cruz (SC-47778 HRP). Rabbit anti-CD79a (#3351), anti-PCD79a (#5173), anti-CD19 (#3574), and anti-PCD19 (#3571) were purchased from Cell Signaling Technologies. Anti-M2A was purchased from MilliporeSigma (#M8064). CK-666 and SMIFH2 were purchased from MilliporeSigma and used at final concentrations of 100 μM and 25 μM, respectively. pnBB and paBB were purchased from Cayman Chemicals and used at a final concentration of 25 μM. DMSO vehicle control was purchased from MilliporeSigma. CellMask Deep Red Plasma Membrane Stain was purchased from Thermo Fisher. Alexa Fluor 488- (#111-545-003), 594- (#111-585-003), and 647- (#111-605-003) conjugated goat, anti-rabbit secondary antibodies were purchased from Jackson ImmunoResearch. Goat anti-mouse IgG Fcγ fragment-specific antibody (#115-005-008) and goat anti-mouse IgM, µ-chain-specific antibodies (#115-005-020) were purchased from Jackson ImmunoResearch. Anti-rabbit-HRP (#32260) was purchased from Thermo Fisher.

GC B cells were generated and sorted using a previously described protocol (Hwang et al., 2015). Briefly, 6–12-week-old M2A-GFP KI mice were immunized with sheep’s red blood cells. After 8–10 days, total B cells from the spleens and lymph nodes were isolated using the Negative Selection B cell isolation kit (StemCell Technologies) according to the manufacturer’s instructions. Dead cells were stained using Zombie Yellow viability stain (BioLegend) and Fc receptors were blocked with the mouse TruStain FcX antibody (#156604). Cells were immunostained with anti-mouse CD38 (#102719), B220 (#103235), and GL-7 (#144617) purchased from BioLegend. GC B cells were sorted on a BD Aria III FACs sorter (Beckton Dickinson) for GFP+, Zombie Yellow^-, B220⁺, CD38^low and GL-7⁺ cells, and were used immediately.

A20 B cells and primary B cells were transfected as previously described (Wang et al., 2017). Briefly, ex vivo primary B cells were first cultured for 12 hr in complete media supplemented with 5 ng/ml BAFF (R&D Systems) and 2.5 µg/ml Escherichia coli O111:B4 LPS (MilliporeSigma) (LPS was included to promote cell survival during nucleofection). 2 × 10⁶ B cells were then nucleofected with 2 μg of plasmid DNA using Nucleofector Kit V (Lonza) and rested for at least 16–24 hr using complete media containing 5 ng/ml BAFF and lacking LPS. We refer to both rested, transfected cells and ex vivo nonmanipulated cells as naïve B cells because neither had been activated by antigen.

Mouse GFP-M2A and Scarleti-M2A template plasmids were gifts from Jordan Beach (Loyola University, Chicago). Mouse M2A sgRNAs were synthesized by Synthego and used according to the manufacturer’s instructions. Briefly, sgRNAs were mixed with Cas9 (IDT) to form ribonucleoproteins and then added together with 0.5 μg of template plasmid to 2 × 10⁶ cells suspended in the solution for Nucleofector Kit V. Following nucleofection, the cells were cultured in complete media for 24 hr before fluorescence-activated cell sorting (FACS) for GFP or Scarleti expression using the Aria III (Becton Dickinson).

miRNAs targeting the 3′ UTR of mouse mDia1 were designed as previously described (Alexander and Hammer, 2016) using BLOCK-iT RNAi Designer (Thermo Fisher), synthesized (Gene Universal), and fused to the C-terminus of mNeonGreen-F-Tractin using In-Fusion cloning (Takara). As a control, a version of this plasmid containing an miRNA sequence that has been verified as nontargeting in mouse (Cai et al., 2006) was used. A20 B cells were transfected with 2 μg of F-Tractin-mNeonGreen vector control, F-Tractin-mNeonGreen-mDia1-miRNAs, F-Tractin-mNeonGreen-nontargeting miRNA, or a combination of F-Tractin-mNeonGreen and 300 nM of mirVana-negative control and cultured in complete media for 16 hr. Cells were then lysed and immunoblotted using an antibody to mDia1 (1:250) and an HRP-conjugated antibody to β-actin (1:5000) to confirm knockdown. Cells that had received the miRNA were identified based on the expression of F-Tractin-mNeonGreen and then quantified after staining with phalloidin. F-Tractin-mNeonGreen-positive cells were also used in a cell spreading assay as described below.

Eight-well Labtek chambers (Nunc) were coated with 15 μg/ml of anti-IgM and/or anti-IgG with or without 0.5 μg/ml of mouse histidine-tagged ICAM-1 (Sino Biological) for 1 hr at room temperature. B cells were resuspended in modified HEPES-buffered saline (mHBS) (Wang et al., 2017) and adhered to functionalized glass for 15 min at 37°C before live imaging or fixing with 4% paraformaldehyde for staining (see Materials and methods). Where inhibitors were used, cells were pretreated for 30 min with 100 μM CK-666, 25 μM SMIFH2, 25 μM pnBB or paBB, or dH₂O/DMSO vehicle control in mHBS at 37°C. Cells were then added to functionalized Labtek chambers in mHBS containing the same concentrations of inhibitors or vehicle control as the pretreatment.

Liposomes were prepared as described previously (Yi et al., 2012; Murugesan et al., 2016; Hong et al., 2017). Briefly, 0.4 mM 1,2-dioleoyl-sn-glycero-3-phosphocholine, biotin–CAP-PE, 1,2-dioleoyl-sn-glycero-3-[(N-(5-amino-1-carboxypentyl)iminodiacetic acid)succinyl] (DGS)–NTA and 1,2-dioleoyl-sn-glycero-3-phosphocholine (Avanti Polar Lipids, Inc) were mixed at 1:3:96 molar % ratio. Lipids were dried under a stream of argon and then desiccated in a vacuum chamber. Unilamellar liposomes were generated from lyophilized lipids hydrated in Tris-buffered saline via extrusion through a 50 nm pore membrane using a mini-extruder kit (Avanti Polar Lipids, Inc). PLBs were assembled in Sticky-Slide VI^0.4 Luer closed chambers (Ibidi) as previously described (Comrie et al., 2015a). 25 × 75 mm glass coverslips (Ibidi) were cleaned using Piranha solution (1:3 ratio of sulfuric acid and 30% hydrogen peroxide). After depositing liposomes onto the flow channels, the channels were washed with HBS buffer containing 1% BSA. A solution containing mono-biotinylated, Alexa Fluor 647-labeled anti-IgM antibody (0.15 µg/ml [300 molecules/μm²] for the limiting antigen condition and 2 µg/ml [4000 molecules/μm²] for the high antigen condition) and streptavidin (Sigma-Aldrich) were added to the flow chambers with or without 0.5 µg/ml unlabeled histidine-tagged ICAM-1. Anti-IgM antibody (µ-chain specific) was monobiotinylated and labeled with Alexa Fluor 647 (Thermo Fisher) as described previously (Carrasco et al., 2004). The uniformity and lateral mobility of PLBs were assessed using FRAP as described previously (Yi et al., 2012). Photobleached circles with a diameter of 4 µm typically recovered within 60 s. B cells were resuspended in modified HEPES-buffered saline and allowed to engage PLBs at 37°C and imaged immediately, or fixed with 4% paraformaldehyde after 5 and 10 min for immunostaining.

Polyacrylamide gels (PA, 0.23  kPa shear modulus, 40  μm thickness) were prepared on glass coverslips with embedded 40 nm fluorescent beads (TransFluoSpheres [633/720], Thermo Scientific), as described previously (Jaumouillé et al., 2019). B cells were resuspended in mHBS with 2% FCS and added to PA gels. Images of B cells that had engaged PA gels for 20 min were captured. A no-stress reference image of the PA gels with beads was captured after lifting cells from the PA gel by adding 1% sodium dodecyl sulfate in 1× PBS to the imaging chamber at a final concentration of 0.04%. Particle image velocimetry was used to calculate bead displacements relative to the reference position, and the corresponding contractile energy was quantified using ImageJ plugins as previously described (Jaumouillé et al., 2019; Martiel et al., 2015). Traction forces were reported as the mean magnitude of traction stress within the cell relative to the cell surface area.

Fixed cells were permeabilized with 0.2% Triton-X-100 and blocked for 30 min at room temperature using PBS containing 2% BSA. Cells were incubated with primary antibodies (1:200) overnight at 4°C and then secondary antibodies (1:250) with Alexa Fluor-conjugated phalloidins for 1 hr at room temperature. Antibodies and phalloidins were diluted in blocking buffer. All washes were performed with 1× PBS.

All live-cell imaging was performed at 37°C in mHBS supplemented with 2% FCS. TIRF-SIM and 3D-SIM imaging were performed on a GE DeltaVision OMX SR microscope (Cytiva) equipped with a 60 × 1.42 NA oil objective (Olympus). For 3D-SIM, z-stacks were acquired at 0.125 µm increments. Raw data were reconstructed using Softworx software (Cytiva) with a Wiener filter constant of 0.002–0.003. Airyscan imaging was performed using an LSM 880 Zeiss confocal microscope equipped with Airyscan and using a Plan-Apochromat 63 × 1.4 NA oil objective. Airyscan image reconstruction was performed using Zeiss ZEN imaging software. TFM was imaged using a Nikon Eclipse Ti2 microscope equipped with a 60 × 1.2  NA water objective. Linear adjustments to images were made using ImageJ 1.53 (NIH).

All image analyses were performed using ImageJ (NIH). To draw ROIs for measurements of the fraction of total IS footprint occupied by each SMAC, the content of F-actin in each SMAC, and the anisotropy of actin filaments within the pSMAC, we relied on the distinctive appearance of actin in each SMAC. This was straightforward in TIRF-SIM images of B cells stimulated with both anti-IgM and ICAM-1, where the thin outer dSMAC was comprised of moderately bright pixels with not much fluctuation in intensity, the medial pSMAC was comprised of bright actin arcs with intervening dim signals, and the central cSMAC was comprised mostly of dim signals. For B cells engaged with anti-IgM alone, the thin outer dSMAC was still readily identifiable, the central cSMAC was identifiable in lower mag images as a central circle with less signal than the area between it and the dSMAC, and the medial pSMAC corresponded to the area between the dSMAC and the cSMAC. Fluorescence intensities within the SMAC regions were quantified using ROIs and reported as the total background-corrected fluorescence within the ROI, which was quantified as described (Burgess et al., 2010) using the following equation: Integrated density – [(area of ROI) × (mean background fluorescence per unit area)], where the integrated density is equal to [(area of ROI) × (mean fluorescence per unit area within the ROI)]. Mean background fluorescence was determined using the same ROI size at three separate positions less than 3 μm away from the cell. The myosin fluorescence intensity in 3D-SIM images was quantified using a maximum projection image of the image stacks where the cell ROI was determined based on the F-actin threshold and the background-corrected myosin fluorescence within the cell ROI was reported. The FibrilTool plugin for ImageJ was used to measure actin arc morphology based on the intensity gradients between pixels as described previously (Murugesan et al., 2016; Boudaoud et al., 2014). Briefly, the pSMAC regions in TIRF-SIM images were divided into 10–12 trapezoid-shaped ROIs of similar size to measure the anisotropy of arcs in the radially symmetric pSMAC. The values obtained range from 0, when the orientation of the structures is random, to 1, when the structures show higher orientation in the same direction. The velocity of centripetal actin flow was assessed by assembling kymographs from TIRF-SIM videos using the Kymograph Builder plugin from ImageJ, as previously described (Murugesan et al., 2016). Briefly, the dSMAC and pSMAC regions were identified by the relatively abrupt slope change for F-actin flow, and slope angles were used to quantify the rates of actin movement. The size of each antigen cluster and their relative distance from the cSMAC center were quantified using an ImageJ macro. First, the perimeter of the synaptic interface was determined based on thresholds for F-actin, and an ROI that encompassed the interface area was drawn (the synaptic ROI). The coordinates for each pixel contained in the ROI were determined and the linear distance of each pixel from the center of mass of the total synaptic antigen (defined as cSMAC center) was determined. The longest distance was defined as the furthest distance to travel from the outermost edge of the cell. A binary image of the antigen channel combined with the ImageJ watershed algorithm was used to segment individual antigen clusters within the synaptic ROI. The area of each antigen cluster was quantified using the Analyze Particles function in ImageJ. The relative distance of each antigen cluster was reported as the distance between the center of mass of the antigen cluster and the cSMAC center after normalizing to the furthest distance from the cell edge to the cSMAC center. To quantify the antigen fluorescence in the cSMAC, a circular ROI corresponding to 20% of the total synaptic area (based on the average area of the cSMAC at the synaptic interface) was drawn such that the center of the circle lies at the same coordinates as the center of mass of the total antigen signal. Antigen fluorescence within this circle was quantified and presented as a percent of the total synaptic antigen fluorescence. The fluorescence intensities of the signaling molecules CD79a, P-CD79a, CD19, and P-CD19 were all reported as the total fluorescence intensity within the synaptic ROI. All fluorescence intensities were corrected for background as described above. Fluorescence intensity profiles were obtained by drawing a 10 μm line across the center of the synaptic interface and using the ImageJ function ‘Plot Profiles’ to obtain fluorescence intensity values across the line. The intensity profiles of several cells were combined, and the average fluorescence intensity ± standard deviation was reported. The speeds of antigen cluster movement were quantified using the ImageJ plugin TrackMate as previously described (Tinevez et al., 2017) where a combination of automated and manual tracking was performed. Prior to quantification, the perimeter of the cell was identified by oversaturating the signal for GFP-M2A, and the anti-IgM fluorescence signal outside of the cell was removed so that only antigen clusters formed by that cell were quantified. Antigen clusters were determined using a blob diameter of 0.2 μm², and tracks were obtained using a threshold of 2000 units with sub-pixel localization. Mean antigen cluster movement speeds were reported as distance traveled over time. Kymographs of moving antigen clusters were created using the ImageJ plugin Kymograph Builder.

All statistical analyses were performed using Prism 9 (GraphPad). Statistical comparisons of dot plots were performed using unpaired, two-tailed t-tests, and data are represented as mean ± standard deviation. Statistical comparisons of bar charts were performed using paired, two-tailed t-tests, and data are represented as mean ± standard error of the mean, unless otherwise stated. The following annotations are used to indicate significance: *p<0.05, **p<0.01, ***p<0.001, and ****p<0.0001.