B cells extract antigens using Arp2/3-generated actin foci interspersed with linear filaments

Antibody production depends on B cell internalization and presentation of antigens to helper T cells. To acquire antigens displayed by antigen-presenting cells, B cells form immune synapses and extract antigens by the mechanical activity of the acto-myosin cytoskeleton. While cytoskeleton organization driving the initial formation of the B cell synapse has been studied, how the cytoskeleton supports antigen extraction remains poorly understood. Here we show that after initial cell spreading, F-actin in B cell synapses forms a highly dynamic pattern composed of actin foci interspersed with linear filaments and myosin IIa. The foci are generated by Arp2/3-mediated branched-actin polymerization and stochastically associate with antigen clusters to mediate internalization. However, antigen extraction also requires the activity of formins, which reside near the foci and produce the interspersed filaments. Thus, a cooperation of branched-actin foci supported by linear filaments underlies B cell mechanics during antigen extraction.


Introduction
Generation of high-affinity, protective antibodies requires B cell uptake of antigens via their antigen-specific B cell receptors (BCRs). After antigen binding, the BCR undergoes endocytosis and delivers the antigen into intracellular processing compartments, where antigenic peptides are loaded onto major histocompatibility II (MHC II) proteins (Hoogeboom and Tolar, 2016). Presentation of the peptide-MHC II complexes on the surfaces of B cells solicits T cell help and promotes B cell proliferation and differentiation into germinal center, memory and plasma cells.
The uptake of soluble antigen by the BCR occurs predominantly via clathrin coated pits (CCPs) (Stoddart et al., 2005). In contrast, the uptake of antigens from antigen-presenting immune or stromal cells requires additional forces generated by the acto-myosin cytoskeleton through the formation of B cell immune synapses (Batista et al., 2001;Fleire et al., 2006;Natkanski et al., 2013). Myosin IIa is thought to be central to generation of these forces, because it is not required for endocytosis of soluble antigens, but promotes force-mediated antigen extraction from presenting cells in vitro and in vivo (Hoogeboom et al., 2018).
Nevertheless, the organization of the actin cytoskeleton in B cell synapses that leads to antigen extraction is incompletely understood. Synaptic actin polymerization is mediated by two mechanisms: the Arp2/3 complex producing branched actin networks, and formins producing linear actin filaments (Blanchoin et al., 2014). It is thought that these distinct actin nucleation pathways are activated in the synapse periphery and create an outer band of branched actin and inner ring of contractile acto-myosin filaments, respectively (Bolger-Munro et al., 2019;Hoogeboom and Tolar, 2016). The constant generation and turnover of the actin results in centripetal flow with central actin clearance. Indeed, in naive B cells interacting with stiff experimental substrates, such as glass or glass-supported planar lipid bilayers, F-actin forms a prominent peripheral ring resembling lamellipodia, and an area with contractile arcs, while the synapse center is clear of actin (Bolger-Munro et al., 2019;Fleire et al., 2006;Freeman et al., 2011;Treanor et al., 2011). However, B cell synapses also contain prominent, but poorly characterized F-actin spots (Keppler et al., 2015;Nowosad et al., 2016;Treanor et al., 2010). Similar spots, termed actin foci, were recently observed in T cell synapses on planar lipid bilayers, where they colocalize with T cell receptor microclusters (Kumari et al., 2015). Nevertheless, the relevance of this actin organization for B cell antigen extraction is unclear as antigen uptake from these stiff substrates is mechanically inefficient and antigens are instead enzymatically digested out through the release of lytic enzymes into the synaptic cleft (Spillane and Tolar, 2017;Yuseff et al., 2011).
In contrast, studies using soft substrates, such as plasma membrane sheets (PMSs), which promote mechanical antigen extraction similar to that occurring with live antigen presenting cells, found that actin formed abundant foci throughout the synapse with little central clearance (Natkanski et al., 2013). These actin foci were highly dynamic and colocalized with membrane invaginations during antigen uptake, suggesting their involvement in antigen extraction. However, how these actin structures cooperate with other actin filaments and with myosin IIa contractility to produce extraction forces is not clear.
One possibility is that myosin IIa causes contraction of formin-generated actin filaments connecting to, or sweeping through the actin foci to generate inward or lateral forces, respectively. This mechanism would be similar to transport of antigen clusters from the synapse periphery to the center by the contractile act-myosin arcs (Bolger-Munro et al., 2019;Murugesan et al., 2016). Alternatively, inward forces on the BCR can be generated by actin polymerization within the actin foci, with the help of membrane-bending proteins orienting the membrane for invagination (Hinze and Boucrot, 2018;Kaksonen et al., 2006).
Here we image the F-actin at the B cell synapse using super-resolution microscopy in 2D and 3D, and using live-cell imaging with high temporal resolution to resolve the structures that are relevant for antigen extraction. We show that the B cell immune synapse contains a highly stochastic and dynamic pattern of actin foci interspersed with actin fibers. The pattern originates from transient spots of Arp2/3 activity generating branched actin in the foci and from closely associated formin activity producing the surrounding linear filaments. The actin foci are more intimately associated with antigen endocytosis, however, manipulation of the actin-polymerization pathways shows that the both Arp2/3 and formins are needed for antigen uptake. We propose that these two actin polymerization mechanisms create a functional foci-filament network which translates local bursts of actin polymerization into an inward force for antigen internalization.

Extraction of antigen is mediated by actin foci interspersed with linear filaments
To understand the structure and dynamics of the F-actin cytoskeleton in B cell synapses specifically during antigen uptake, we imaged primary B cells from Lifeact-expressing mice (Riedl et al., 2010) incubated with PMSs loaded with a surrogate antigen, anti-Igκ F(ab')2.
Total internal reflection fluorescence (TIRF) microscopy time-lapses showed that after initial cell spreading, between 5 and 20 minutes after synapse formation, the Lifeact-labeled F-actin was distributed throughout the synapse in a range of small structures including actin foci and short fibers ( Figure 1A). 3D imaging verified that during these 20 min, B cells were extracting the antigen as previously described ( Figure 1B). TIRF microscopy with a 100 ms time resolution showed that the F-actin was highly dynamic and that the foci and fibers were appearing stochastically, and often, merged, split or interconverted between each other (Movie 1, left panel). To analyze the actin structures, we segmented the actin foci and fibers from the Lifeact images using round and linear filtering elements, respectively ( Figure 1C, We therefore used the automated segmentation as an unbiased quantification tool to analyze all imaged cells. Segmentation of actin foci and fibers from time-lapse microscopy of the primary mouse B cells showed that foci numbers per area remained stable at about 5 per 10 µm 2 during the 20 min of imaging ( Figure 1D, E), but that most individual foci persisted for less than 5 s ( Figure 1F). A similar pattern of foci and interconnecting fibers was also observed in the human Ramos cell line, a commonly used B cell lymphoma line, although these cells were larger and had more filopodia emanating from the sides of the synapse ( Figure 1G).
To generate a high-resolution view of these F-actin structures, we used the super-resolution localization technique direct stochastic optical reconstruction microscopy (dSTORM) using phalloidin-AlexaFluor647 staining of fixed cells ( Figure 1H). Super-resolution reconstruction of the F-actin showed numerous dense actin foci ( Figure 1I), interspersed by a sparse network of fibers ( Figure 1J). This pattern filled the body of the synapse and was surrounded by thin lamellipodia ( Figure 1K).
Dual color live-cell TIRF microscopy indicated that although the actin structures did not overtly colocalise with antigen, some actin foci did transiently coincide with antigen clusters (Movie 2), in line with previous data. To understand the structure of the F-actin at the sites of antigen extraction in more detail, we used 3D super-resolution imaging of F-actin and overlaid the super resolution actin image with epifluorescence stacks of antigen images ( Figure 1L, Movie 3). Antigen clusters that were being internalized could be identified, because they were lifted several hundred nanometers from the PMSs. Such lifted clusters of antigen were surrounded by dense F-actin foci from the sides ( Figure 1M, N). Measurement of the height of all the actin foci showed that more than 20% of them extended more than 200 nm into the cell ( Figure 1O). This suggests that foci are the actin structures associated with antigen movement into the cell and thus likely correspond to actin spots previously associated with membrane invaginations (Natkanski et al., 2013). The 3D STORM reconstruction also indicates that the actin foci interact with antigen clusters from the sides or from below, suggesting that they are involved in the transport of the antigen to the interior of the cell.

Arp2/3 and formins generate the distinct actin structures of the B cell synapse
To determine the mechanisms by which the actin foci and fibers are generated, we blocked actin polymerization in PMS-attached B cells either with CK666, which specifically inhibits the activity of the Arp2/3 complex, or with SMIFH2, which inhibits formins. Superresolution imaging showed that CK666 treatment resulted in a loss of actin foci, which were instead replaced by abundant actin fibers ( Figure 2A). In contrast, treatment with SMIFH2 primarily dispersed the fibers in between foci ( Figure 2A). Quantification of fast timelapse imaging of CK666-treated B cells confirmed reduction in foci numbers ( Figure 2B) and showed that the residual fibers were dynamic, growing, shrinking and curving rapidly (Movie 4). On the other hand, in SMIFH2-treated B cells, foci numbers were increased, although their lateral movement was arrested and their dynamics stunted (Movie 4, Figure 2B). These results suggested that the actin foci are branched-actin structures generated by the Arp2/3 complex, while the interspersed fibers are linear actin filaments or filament bundles produced by formins. To confirm the inhibitor treatment, we targeted the Arp2/3 complex and formins genetically. The Arp2/3 complex was acutely targeted by CRISPR in Ramos cells expressing Cas9 using three different guide RNAs (gRNAs) specific for ARPC2, an essential component of the complex. All three gRNAs greatly reduced the average levels of the ARPC2 protein in the targeted cell population ( Figure 2C). All three gRNAs reduced the numbers of actin foci and promoted production of actin fibers across the synapse and in filopodia ( Figure 2D, E). Thus, ARPC2 is essential for production of actin foci in the B cell synapse.
Formins are a family with multiple members that are in many aspects functionally redundant.
We targeted two formins highly expressed in B cells including Ramos cells (Brazão et al., 2016;Klijn et al., 2015), DIAPH1 and FMNL1. DIAPH1 was successfully targeted in Ramos cells with one gRNA and FMNL1 with two gRNAs ( Figure 2C). We also generated Ramos cells lacking both DIAPH1 and FMNL1 by re-targeting the DIAPH1-targeted cells with two different FMNL1 gRNAs. Imaging F-actin and quantification of actin foci revealed that targeting of the formins resulted in little change of the synaptic actin pattern ( Figure 2F), although quantification showed a subtle decrease in the number of actin foci in cells targeted with the DIAPH1 gRNA, and a small increase in cells targeted with FMNL1 or both DIAPH1 and FMNL1 gRNAs. Therefore, neither DIAPH1 nor FMNL1 are required for the formation of actin foci, and they are redundant in production of the filaments outside of the foci.

Dynamics of Arp2/3 and formins accounts for the actin architecture of the B cell synapse
To observe the role of Arp2/3 and formins in actin dynamics directly, we transduced Ramos cells with constructs of ARPC2-mRuby or DIAPH1-mRuby and analyzed their localization in phalloidin-stained cells interacting with anti-IgM loaded PMSs. ARCP2-mRuby localized predominantly in round or slightly elongated spots that corresponded to phalloidin-labeled actin foci ( Figure 3A). ARPC2-mRuby also closely followed the dynamics of actin in foci visualized in time-lapse imaging of Ramos cells co-expressing Lifeact-GFP ( Figure 3B, Movie 6). The ARPC2-mRuby-positive actin foci were surrounded by short, ARPC2-mRubynegative actin fibers, which were frequently seen dynamically emanating from the foci and sometimes transiently connecting to other foci ( Figure 3C). Simultaneous labelling of the Ramos B cell plasma membrane using the lipid dye DiD indicated that while in the cell periphery the fibers grew into filopodia, in the center of the synapse, the short fibers did not DIAPH1-mRuby localization was more diffuse than that of ARPC2, but some DIAPH1-mRuby spots were also visible and they often overlapped with actin foci ( Figure 3D).
However, live cell imaging indicated that DIAPH1 spots followed the dynamics of actin foci less precisely than ARPC2 spots did ( Figure 3E, Movie 7). The spots of DIAPH1-mRuby were also seen producing actin fibers growing out of them, including filopodia ( Figure 3F,

Movie 7).
Quantification confirmed that ARPC2-mRuby was significantly enriched in actin foci, while it was depleted in fibers ( Figure 3G). In addition, dual segmentation of the ARPC2 spots and actin foci showed that a significant fraction of actin foci directly colocalized with ARPC2 spots and vice versa ( Figure 3H). DIAPH1-mRuby was also enriched in actin foci as compared to fibers ( Figure 3G) and DIAPH1-mRuby spots significantly colocalized with actin foci ( Figure 3I). However, its enrichment in foci and rates of colocalization were significantly lower than that of ARPC2.
Thus, actin foci contain predominantly branched actin marked with ARPC2, and they represent the dominant branched-actin structures of the B cell synapse. Nevertheless, the foci also associate with DIAPH1, which does not contribute to foci generation, but produces short linear filaments that surround and interconnect them.

Myosin IIa activity is not required for foci formation
To understand the role of myosin IIa contractility in regulation of the actin architecture and particularly the actin foci, we analyzed the relationship of F-actin to activated myosin IIa using antibodies specific for the phosphorylated myosin II light chain. This imaging showed that most of the active myosin IIa did not overlap with the most prominent F-actin structures, although it was localized in close proximity to both actin foci and actin fibers (

BCR signaling regulates actin foci dynamics
To investigate the relationship of the actin architecture to BCR signaling triggered by antigen binding, we stimulated Ramos cells with anti-IgM coated PMSs and analyzed localization of total tyrosine-phosphorylated proteins as well as phosphorylated BLNK and CD19, two critical adaptors that propagate signaling to the cytoskeleton and downstream activation pathways ( Figure 4A, B). All phosphotyrosine signals had low colocalization with actin, however, we detected significantly higher enrichment of total phosphotyrosine and phospho-CD19 signals in actin foci than in actin fibers ( Figure 4B). Phosphorylated BLNK was only weakly enriched in foci, but was still highly excluded from fibers ( Figure 4B). Thus, actin foci are the predominant actin structure associated with BCR signaling.
However, we observed that the actin architecture composed of foci and interspersed fibers did not require antigen-induced stimulation of the BCR, as it was similar in B cells interacting with PMSs loaded with non-stimulatory anti-MHC I or anti-MHC II antibodies

Actin foci dynamically colocalize with CCPs
As mentioned above, actin foci dynamically interacted with antigen clusters during extraction, and had a close relationship with BCR signaling, suggesting that ARPC2 is proximal to antigen uptake. We observed that ARPC2 clusters, colocalized with antigen clusters at the time of extraction and endocytosis (Movie 11), while DIAPH1 did not (Movie 12). To understand the relationship of the actin foci to CCPs, the major known endocytic mechanism for antigen uptake, we simultaneously imaged F-actin and the clathrin light chain A (CLTA), tagged with mCherry in Ramos cells. Although most actin foci and CCPs did not colocalise with each other ( Figure 4C), some CCPs were located in the vicinity of actin structures. Quantification showed that CLTA-mCherry signal was significantly enriched in actin foci as compared to actin fibers ( Figure 4D). Although only about 3% of foci directly colocalized with CCPs, and 7% of CCPs colocalized with actin foci, these rates were significantly above those expected from random co-localization of these structures ( Figure   4E). Thus, while most actin foci are independent of CCPs, a small number of them coincides with a fraction of CCPs at any given time, consistent with the transient nature of actin polymerization during clathrin-mediated endocytosis.

Actin foci and fibers jointly regulate antigen uptake
Overall, our data suggested that actin polymerization in foci is important for antigen extraction. We found that internalization of soluble surrogate antigen anti-Igκ F(ab')2 or soluble transferrin by naive mouse B cells was only very modestly sensitive to inhibition of either of the Arp2/3 complex or formins, or to complete block in actin polymerization by cytochalasin D (

Both Actin foci and fibers contribute to mechanical activity
To understand how inhibition of Arp2/3 or formins compromises antigen extraction, we analyzed the mechanical activity of the B cells in the immune synapses. Pulling on the antigen during extraction can be visualized by imaging the deformation of the antigen-coated PMSs using the lipid dye DiI (Natkanski et al., 2013). Ramos cells expressing Lifeact-GFP produced PMS deformations that appeared as spots, which we previously showed are short, vertical membrane invaginations (Natkanski et al., 2013), and also laterally elongated structures, ridges or tubules ( Figure 6A, Movie 13). The shape of the pulled structures corresponded to the shape of the actin structures associated with them. Spots typically coincided with actin foci ( Figure 6B, at 15 and 30s), while shorter tubules associated with elongating foci ( Figure 6B, at 45 s) and longer tubules or ridges with longer actin fibers ( Figure 6C). We quantified the rates and location of the PMS deformations by tracking the appearance and movement of these membrane structures with respect to the actin cytoskeleton. Approximately 40% of these pulling events could be assigned to actin foci or fibers using this analysis ( Figure 6D). About 30% occurred with actin fibers, while less than 10% associated with foci ( Figure 6D). Most mechanical activity thus comes from areas containing actin fibers. However, correcting for the much smaller area of the cells occupied by foci showed that the actin foci were three times more efficient at applying forces on the PMSs than fibers ( Figure 6E). Thus, both foci and fibers apply pulling forces on the antigenpresenting membrane, but with different geometry and efficiency.
To understand the role of Arp2/3 and formins in pulling on the antigen-presenting membrane, we analyzed PMS deformations induced by the Ramos cells in the presence of CK666 or SMIFH2. We found that the overall number of membrane pulling events was not affected by either of the inhibitors compared to the DMSO control ( Figure 6F). However, Ramos cells treated with CK666 often generated elongated membrane structures in the synapse, with some cells producing excessively long and convoluted membrane tubules and ridges ( Figure   6G, Movie 14-16). To quantify this effect, we classified the DiI images of the PMS in immune synapses as containing only pulled spots, a mixture of spots and short tubules, and a predominance of tubules or ridges ( Figure 6G). The analysis showed that the pulling activity of CK666-treated Ramos cells resulted in less synapses containing only spots and more synapses dominated by tubules ( Figure 6H). These data are consistent with formin-generated fibers predominantly contributing to long-range antigen movement along the plasma membrane. Overall, while actin polymerization either through Arp2/3 or formins exerts mechanical forces in the immune synapse, coordinated function of both of these actin polymerization mechanisms is required to extract antigen in a manner that leads to efficient endocytosis.

Discussion
Our data show that after initial cell spreading, the B cell synapse contains a stochastic pattern of foci and fibers. These structures are generated locally in the synapse by transient activities of the Arp2/3 complex and formins, respectively and are important for antigen extraction.
Despite their different mechanism of production, the foci and fibers are closely related as they are often generated from each other during their constant turnover. This is also illustrated by colocalization of both Arp2/3 and formins with the actin foci, leading to production of both branched actin and linear filaments from the same sites. This actin behavior produces network-like connectivity that is distinct from the actin structure of CD4 (Kumari et al., 2015;Murugesan et al., 2016) and CD8 (Ritter et al., 2016) T cell synapses and also differs from B cell synapses formed with stiff substrates, where peripheral lamellipodial actin, and acto-myosin contractile rings occupy distinct domains of the synapse (Bolger-Munro et al., 2019;Fleire et al., 2006).
We propose that when B cells extract antigens from live antigen-presenting cells, actin foci and fibers form nodes and edges in a loosely connected network, whose function is required for efficient antigen extraction. In this model, stochastic Arp2/3 activation produces actin foci by local bursts of branching of preexisting short filaments. Branched actin polymerization may either push on the membrane to generate protrusions, similarly as it happens in podosomes (van den Dries et al., 2019), or synergize with BCR-dependent signaling and endocytic proteins to produce an invagination of the plasma membrane and extract the antigen. Formin-generated filaments may stabilize the foci by surrounding and connecting them, which, along with crosslinking and tensioning by myosin IIa, could help to direct the foci-generated forces perpendicularly to the plasma membrane. In addition, the linear filament polymerization out of the foci may help to extend the range of the inward movement or enhance the extraction force directly. This model is supported by our observation and manipulation of the two major regulators of actin polymerization, the Arp2/3 complex and formins. The foci are selectively marked by stoichiometric incorporation of ARPC2 and they disperse after inhibition of Arp2/3 function, consistent with a branched-actin structure. The importance of the Arp2/3 complex in foci formation is consistent with previous data linking activators of the Arp2/3 complex, WAS and WIPF1, with foci formation in T cells (Janssen et al., 2016;Kumari et al., 2015) and B cells (Keppler et al., 2015). In contrast, formins, such as DIAPH1, localize less specifically to the foci and their activity instead generates interspersed actin fibers. Thus, both Arp2/3 and formins are active throughout the B cell synapse during antigen uptake without large-scale segregation. Our data thus extend previous knowledge of the importance of Arp2/3 complex in the generation of the lamellipodial actin ring and centripetal membrane flow important for B cell adhesion, BCR signaling and B cell activation (Bolger-Munro et al., 2019).
Our data also show that Arp2/3-generated actin foci are intimately associated with antigen extraction, as seen in instances of their presence at the sites of antigen endocytosis and their close, albeit stochastic, relationship to BCR tyrosine-phosphorylated proteins and CCPs. Our model thus suggests that forces that lead to antigen extraction are dependent on new actin polymerization in actin foci. This hypothesis is supported by previous observation that Factin is continuously accumulating on individual membrane invaginations during B cell antigen uptake (Natkanski et al., 2013). If the plasma membrane is instead pulled inward by myosin IIa contractility, F-actin should accumulate before the membrane starts invaginating and then move inward during invagination. In addition, our 3D superresolution imaging shows that actin is associated with the sides of the antigen clusters during internalization, where polymerization could contribute to membrane invagination.
Nevertheless, our data also indicate that the branched actin polymerization in the foci is not sufficient to extract antigen as inhibitions of formin and myosin IIa activities also inhibit antigen uptake. Since formin inhibition altered foci dynamics, we hypothesize that formins together with myosin IIa provide stability to actin polymerization in the foci that helps the branched actin to expand in between the future base and tip of the invagination, generating and inward force, as opposed to a lateral force and a comet-like movement along the plasma membrane. A similar stabilizing role for myosin contractility was suggested for function of CCPs in other cells types (Chandrasekar et al., 2014). Alternatively, myosin contractility or fiber dynamics may help to dislodge the antigen from the presenting membrane before endocytosis through actin foci. Finally, it is possible that formin-mediated actin polymerization out of the foci extends the range of movement of the foci during late stages of antigen endocytosis. The latter two points are supported by the observations of cells after pharmacological inhibition of the Arp2/3 complex, when the residual fibers are able to transport antigen clusters and pull large membrane ridges or tubules from the presenting membrane both by their contractile motion and by their growth. Future studies will be needed to dissect the role of these mechanisms further and also to determine the exact roles of specific formin family members for these processes as our knock outs of DIAPH1 and FMNL1 were not sufficient to fully recapitulate formin inhibitor treatment.
The actin organization and mechanism of antigen extraction studied here pertains primarily to the synapses formed by naive and memory B cells, where antigen extraction is directly followed by endocytosis (Nowosad et al., 2016). Another mechanism to extract antigen, not directly coupled to endocytosis, exists in germinal center B cells. In these cells, peripheral actomyosin-rich pod like protrusions extract antigen and transport it by outward actin flow along the plasma membrane away from the antigen-presenting surface (Kwak et al., 2018;Nowosad et al., 2016). The regulation of actin in these structures has not yet been investigated, but may also involve Arp2/3-generated branched actin as they resemble miniature lamellipodia. Arp2/3-dependent actin polymerization is therefore likely a general requirement for antigen extraction from antigen-presenting cells.
Unexpectedly, our data show that the formation of the actin foci in naive B cells does not require antigen stimulation. The actin foci may thus be generated continuously as a general property of the B cell cortical cytoskeleton. In agreement with this observation, the majority of actin foci in the synapse do not associate with antigen clusters, or with CCPs. However, a small fraction of actin foci did associate with antigen clusters and to a lower degree with CCPs. Since BCR signaling regulates actin dynamics (Tolar, 2017), including foci dynamics shown here, as well as CCP dynamics (Natkanski et al., 2013;Stoddart et al., 2002) we speculate that a local synergy between BCR signaling, actin dynamics and CCP formation at BCR clusters either enhances the rates of formation of endocytic structures, or stabilizes otherwise abortive events to promote efficient antigen uptake. More detailed observations of BCR signaling during interaction of BCR clusters with foci and CCPs will be needed in the future.
Our data establish a new view on the architecture of B cell synapses and on the molecular basis driving its formation. This information can be used to better understand the role of Factin not just in B cell antigen endocytosis, but also in receptor signaling at the plasma membrane. The organization of the BCR, and other signaling receptors on B cell surfaces critically depend on actin structure (Mattila et al., 2016) and defects in actin polymerization contribute to the dysregulation of signaling in primary immunodeficiencies and in autoimmunity (Tolar, 2017). Our studies should guide future investigations of the regulation Arp2/3 and formin pathways in B cells in this context.

Mice
C57BL/6 mice and heterozygous Lifeact-EGFP mice on a C57BL/6 background were used as a sources of primary mouse B cells. All mice were between 1-6 months old and were male and female. Mice were bred and treated in accordance with guidelines set by the UK Home Office and the Francis Crick Institute Ethical Review Panel.

B cell purification
Primary mouse B cells were isolated from spleens by mashing tissue through a 70 µm cell strainer, followed by red blood cell lysis using ACK Lysing Buffer (Gibco), and negative selection using anti-CD43 microbeads (Miltenyi Biotec).

Cell lines
HEK293T and Ramos cell lines were provided, authenticated and mycoplasma-tested by the Francis Crick Institute Cell Services.

Anti-BCR antibodies used to stimulate B cells on PMSs and in solution were goat anti-mouse
Igκ F(ab')2(Southern Biotech) for mouse splenic B cells and goat F(ab')2anti-humanFc5µ (Jackson Immunoresearch) for Ramos cells. The antibodies were biotinylated with 20-fold molar excess of NHS-LC-LC-biotin (Pierce) and labelled with either Cy3 or Cy5 Monoreactive dyes (GE Healthcare), or with AlexaFluor 405 or AlexaFluor 647 NHS esters (ThermoFisher) in Sodium Carbonate buffer. Excess dye was removed using Zeba 7K MWCO desalting columns (Pierce, ThermoFisher).

Antibodies
All antibodies used in this study were:

Preparation of samples for imaging
Plasma membrane sheets were generated as described (Nowosad and Tolar, 2017  which produces a 360° spinning beam with adjustable TIRF illumination angle. Acquired datasets were analyzed using ImageJ and Matlab (Mathworks). All images were aligned and, where necessary, corrected for background, flatfield illumination and photobleaching.

Super-resolution imaging
To label F-actin for super-resolution imaging, samples prepared as above blocked for 1 hr  (Ovesný et al., 2014), drift-corrected and displayed using average shifted histograms. 3D STORM localizations were subsequently grouped into 50 nm artificial z-stacks. To produce 3D STORM and epifluorescence image composites, a single epifluorescence image with the appropriate z-coordinate was paired with each 50 nm 3D STORM image. Images were aligned in ImageJ to produce an epifluorescence and 3D STORM composite z-stack. Side view reconstructions were produced from composite zstacks using maximal intensity projections of a region of a cell along the x or y axis. . We compared the total number of foci detected in the timelapses and also the centers of their positions using a 5-pixel and 1-frame tolerance.

Image analysis of actin foci and fibers
"True positive rate" of the automated segmentation was calculated compared to each human researcher as the percent of the human-detected foci that were also detected by the computer.
"False positive rate" of the automated segmentation was calculated as the percent of foci detected by the computer but not detected by the human. As a comparison, foci positions obtained by the human researchers were also compared to each other using the same calculations. We found that the differences between the computer and the humans were not significantly different from differences between the human researchers themselves. Thus, the automated segmentation provided an unbiased high-throughput foci detection tool with quality similar to manual segmentation.

Image quantification of enrichment and colocalization
Foci and fibers were first segmented in automatically detected cells in all images as described above. The enrichment of fluorescence signal of interest in actin foci and fibers was calculated from the fluorescent intensity in these structures, normalized to mean intensity in the synapse. Coincidence of actin foci with ARPC2 spots, DIAPH1 spots or CCPs and vice versa was calculated as the fraction the foci that overlapped with the center of the other structures. As a control, centers of the same number and size of structures were placed without overlap at random location in the same cell. A mean of ten of the randomization runs was taken as the random value for each cell.

Quantification of B cell mechanical activity
Mechanical activity in the synapse was visualized using DiI-labeled PMSs as described (Natkanski et al., 2013). Briefly, PMSs were labeled using DiI and imaged in TIRF timelapses for dynamic membrane deformations that appear as local increase in fluorescence intensity. To quantify these membrane pulling events, bright DiI structures were located and tracked using previous algorithms (Tolar et al., 2009) Figure 6G and Movies 14-16.

Quantification of substrate-bound antigen internalization
For quantification of antigen uptake from PMSs, cells were incubated with inhibitors for 15 min at 37°C before incubation with PMSs in the continued presence of the inhibitors for 30 min. Large-scale image datasets and their processing was performed as described (Nowosad et al., 2016). In brief, experimental images were cropped, aligned, background subtracted and corrected for flatfield and spectral bleedthroughs. Cells were segmented using B220 staining (primary B cells) or CD19 (Ramos cells). CellScore (Nowosad et al., 2016) was used to gate on live B cells. To detect extracted antigen in the z-stack images, each image was bandpassfiltered and antigen clusters were identified in planes above the synapse by a user-specific global threshold. Total fluorescence of extracted antigen was calculated as the sum of pixel intensities of the background-corrected extracted clusters.           DiI are visualized as dynamically appearing and moving spots, tubules and ridges in the DiI signal. Related to Figure 6A. Elapsed time is in minutes and seconds.