F-actin flashes on phagosomes mechanically deform contents for efficient digestion in macrophages

Phagocytosis is the uptake process of relatively large target particles by specialized innate immune cells and amoeba. Phagocytosis in macrophages is characterized by dramatic F-actin protrusions that are orchestrated by opsonic receptor signaling. The major opsonic receptors in macrophages are Fcγ receptors that recognize bound IgG and complement receptor 3 (CR3) that binds to deposited complement protein fragment C3bi on target particles. Studies so far have revealed striking differences between the modes of phagocytosis induced by the opsonic receptors in macrophages, including a strong requirement for the RhoA GTPase, but not Rac1 and Cdc42, for CR-mediated phagocytosis (Caron and Hall, 1998). Our laboratory has chosen to delve into the poorly understood CR-mediated phagocytosis and shown that membrane ruffles actively capture C3bi-opsonized targets (Patel and Harrison, 2008), challenging the previous dogma that C3bi particles passively sink into the cell. Once internalized, the subsequent process of phagosome maturation is better understood for FcγR-mediated phagocytosis in macrophages. During engulfment of IgG-coated particles, F-actin disassembly occurs around the phagosome to facilitate interactions with the endocytic pathway, which drives content acidification and degradation.

Although F-actin is believed to be an impediment to phagosome maturation, the presence of transient accumulations of F-actin on phagosomes has been observed during uptake of bacteria and IgG-opsonized particles. These ‘F-actin flashes’ were first described in Madin–Darby canine kidney (MDCK) epithelial cell phagosomes containing Listeria monocytogenes (Yam and Theriot, 2004). Manipulation of bacteria surface and effector proteins determined that this was a host cell-driven event. F-actin flashes were also observed in macrophages overloaded with IgG-opsonized latex beads (Liebl and Griffiths, 2009). Macrophages with increased phagocytic load had a greater incidence of F-actin flashing, effectively reducing the rate of maturation (as shown by lowered LAMP2 recruitment to phagosomes). In addition, the authors determined that phagosome membrane curvatures presented a surface that was favorable for actin nucleation (Liebl and Griffiths, 2009). Other observed F-actin accumulations on phagosomes included F-actin comet tails on CR-phagosomes in macrophages that could potentially propel phagosomes towards lysosomes (Bohdanowicz et al., 2010).

In this investigation, we have focused on the mechanism and function of F-actin flashes on phagosomes. F-actin flashes transiently encompass the phagosome and are cyclical in nature over the first hour of internalization. F-actin flashes predominate on integrin-based CR-phagosomes and involve recruitment of focal adhesion proteins and RhoA signaling. Interestingly, the rigidity of the phagosome contents influences F-actin flashing dynamics on phagosomes. F-actin flashes show strong myosin II recruitment, and myosin II activity drives phagosome content deformation. Interestingly, phagosomes that exhibit F-actin flashes show accelerated degradation at later time points. Thus, we have evidence showing for the first time that an active ‘chewing’ mechanism may be at play during intracellular particle destruction in macrophages.

Our laboratory studies CR-mediated phagocytosis and we have previously shown a prominent role for F-actin in mediating membrane ruffle capture of C3bi-opsonized targets. To image F-actin, we generated RAW 264.7 cells stably transfected with fluorescent LifeAct (RAW-LifeAct) constructs (Riedl et al., 2008). Transfected cells were treated with phorbol myristate acetate (PMA) to activate CR (Wright and Silverstein, 1982), exposed to C3bi-opsonized sheep red blood cells (sRBCs) and imaged with epifluorescence microscopy. We saw the characteristic F-actin cup at the time of particle capture, after which the LifeAct-CFP signal dissipated when the sRBC was internalized (Fig. 1A), as expected (Vieira et al., 2002). However, to our surprise, a strong F-actin signal was recruited to the internalized phagosome, which then disappeared and reappeared multiple times over the first hour of particle internalization and finally subsided (Fig. 1A–C; Movie 1). To better visualize the dynamics of the cyclical accumulations of F-actin on CR-phagosomes, we measured LifeAct-CFP signal intensity over time on the phagosome membrane (Fig. 1B). These F-actin flashes were also observed in mouse bone marrow-derived macrophages (BMDMs) transiently transfected with LifeAct-RFP (Fig. 1C). The recurrent phagosome recruitment of F-actin was very reminiscent of F-actin flashes described around internalized Listeria in epithelial cells and in macrophages after uptake of IgG-opsonized latex beads (Liebl and Griffiths, 2009; Yam and Theriot, 2004). We fed RAW-LifeAct cells with IgG-sRBCs to compare the frequency of F-actin flashes between IgG- versus C3bi-opsonized targets. The F-actin flashes were specific to C3bi-opsonized targets, where approximately one-third of phagosomes exhibited flashing, versus IgG-sRBCs which showed less than 5% frequency (Fig. 1D), as reported for IgG-particle-overloaded macrophages (Liebl and Griffiths, 2009). During imaging, we also occasionally observed asymmetric F-actin comet tails on phagosomes that typically appeared once on a phagosome and resulted in particle translocation, as described previously (Bohdanowicz et al., 2010). We quantified these distinct events, which we determined to be rarer than F-actin flashes on CR-phagosomes (Fig. 1E). We next investigated F-actin recruitment to complement-opsonized Escherichia coli compartments in macrophages, given that F-actin flashing was observed on internalized bacteria in epithelial cells (Yam and Theriot, 2004). We saw a strong accumulation of F-actin around internalized C3bi-E. coli 40 min after internalization (Fig. 1F).

Our DIC analysis revealed a striking fate of the sRBCs within the F-actin flashing phagosomes (Fig. 1A,C,G; Movie 1). The appearance of F-actin on phagosomes resulted in near-instant sRBC deformation, with fission events and sRBC lysis often occurring (Fig. 1A,C,G,H; Movie 1). These data strongly indicated that the presence of F-actin on phagosomes was mechanically altering the cargo. sRBC integrity was rarely altered in non-flashing phagosomes, showing a strong connection between the appearance of F-actin flashes and cargo manipulation.

We observed that particle deformation often occurred simultaneously with F-actin recruitment on CR-phagosomes and we hypothesized that phagosome membrane proteins may have mechanosensing properties that detect an engulfed particle's stiffness. We predicted that a disparity between opsonized particles of varying malleability would manifest in the form of changes in the temporal dynamics of F-actin flashes. We compared F-actin flashes on phagosomes containing similarly sized C3bi-sRBCs versus C3bi-beads to represent malleable and rigid particles, respectively. F-actin flashes on phagosomes were monitored using live-cell epifluorescence imaging of RAW 264.7 cells stably transfected with LifeAct-RFP.

The time taken for a single F-actin flash on a phagosome was ∼3 min for both opsonized sRBCs and beads (Fig. 2A). The total period of F-actin flashing events was also similar between C3bi-sRBCs and C3bi-beads, ranging at approximately 30 min for both (Fig. 2B). Interestingly, the relative rigidity of the target particle influenced the frequency dynamics of F-actin flashes on phagosomes. Despite each F-actin flash and total duration of all flashes being similar (Fig. 2A,B), the more malleable opsonized particles (sRBCs) had much shorter time intervals between F-actin flashes on phagosomes (Fig. 2C), which resulted in twice as many flashes over the first hour compared with opsonized rigid latex beads (about six versus three, respectively; Fig. 2D). The time from F-actin cup formation until the first F-actin flash on phagosomes was also compared. Phagosomes containing malleable C3bi-sRBCs took ∼9 min to flash, whereas the phagosomes containing stiff C3bi-beads showed flashes within ∼3.5 min (Fig. 2E); however, this difference was not significant. Interestingly, the prevalence of F-actin flashes was lower on phagosomes with C3bi-beads (13.2±0.6%) compared with phagosomes containing C3bi-sRBCs (36.3±5.7%) (Fig. 2F). Together, the results show that C3bi-sRBCs triggered more instances of F-actin flashes on phagosomes, each having more F-actin recruitment events than phagosomes containing latex beads. A comparative summary of the differential dynamics of F-actin flashes on phagosomes containing C3bi-sRBCs versus C3bi-beads is shown in Fig. 2G,H.

We wanted to understand the mechanism behind F-actin recruitment to phagosomes during CR-mediated phagocytosis. The prominent deformation of sRBCs during F-actin flashes in phagosomes, and varying flashing kinetics on particles of different stiffness, prompted us to investigate whether these were a result of the mechanosensing properties of focal adhesions. CR3 is an integrin and integrins are the core of focal adhesion structures that transduce extracellular mechanical cues to the actin cytoskeleton (Gardel et al., 2010). We first probed F-actin-positive phagosomes for CR3 and saw marked enrichment around phalloidin-positive phagosomes at 45 min after particle internalization (Fig. 3A). Phalloidin staining for endogenous F-actin along phagosomes confirmed that F-actin flashes were not an artifact of LifeAct expression (Fig. 3A). We quantified the number of CR3-positive phagosomes for both F-actin-positive and F-actin-negative phagosomes at early (20 min) and late (45 min) times of internalization (Fig. 3B). Although most phagosomes showed detectable levels of CR3, the presence of strong CR3 immunolabeling on the phagosomes corresponded with F-actin recruitment at both time points (Fig. 3B). A small population of phalloidin-negative phagosomes enriched in CR3 was also observed (Fig. 3B). We next investigated talin, which acts as an interface between integrins and the actin cytoskeleton, and promotes integrin clustering and activation (Calderwood and Ginsberg, 2003; Janoštiak et al., 2014). Using immunofluorescence, we imaged RAW 264.7 macrophages after ingesting particles and saw strong talin enrichment around F-actin-positive phagosomes, but not on F-actin-negative phagosomes (Fig. 3A,B). To investigate the dynamics of talin recruitment to F-actin flashing phagosomes, we transfected mEmerald-talin into RAW 264.7 cells stably expressing LifeAct-RFP and performed live-cell confocal imaging during C3bi-phagocytosis. The pattern and timing of mEmerald-talin recruitment strongly overlapped with LifeAct-RFP accumulation on phagosomes (Fig. 3C; Movie 2). We quantified the effects of this coincident recruitment on the dynamics of F-actin flashes and found that the presence of mEmerald-talin did not influence the duration of flashes on phagosomes (Fig. 3D) but did induce an extended delay between F-actin flashes compared with cells expressing LifeAct-RFP alone (Fig. 3E). To look at downstream effectors, we also probed phagosomes for total and active focal adhesion kinase (FAK, Tyr397) using immunofluorescence and found that both showed strong recruitment around F-actin-positive phagosomes exclusively (Fig. 3A,B).

Actin comet tails have been described on CR-phagosomes and are generated by phosphatidylinositol 4,5-bisphosphate [PI(4,5)P₂] synthesis, which in turn is thought to drive the polymerization of actin on the membrane (Bohdanowicz et al., 2010). To investigate whether PI(4,5)P₂ was present on flashing phagosomes, we utilized a GFP-C1-PLCδ-PH construct as a probe to monitor PI(4,5)P₂ activity (Stauffer et al., 1998). RAW 264.7 cells stably transfected with LifeAct-RFP were transiently transfected with GFP-C1-PLCδ-PH by electroporation. Live-cell epifluorescence imaging was performed for 1 h post-C3bi-sRBC-binding with 15 s intervals between images. Although we observed an increase in GFP-C1-PLCδ-PH signal at the site of C3bi-sRBC entry, subsequent imaging of flashing and non-flashing phagosomes revealed no significant enrichment of PI(4,5)P₂ on phagosome membranes (Fig. 3F). Similar results were observed with the tandem PH probe (GFP-C1-PLCδ-2PH; data not shown).

We next investigated whether RhoA was present on flashing phagosomes. RhoA is crucial in CR-mediated phagocytosis where it influences both actin assembly and myosin II activity (Caron and Hall, 1998; Colucci-Guyon et al., 2005; Olazabal et al., 2002; Wiedemann et al., 2006). We found that RhoA immunostaining was enriched on phalloidin-positive phagosomes, but not on phalloidin-negative phagosomes (Fig. 4A,B). To determine whether the phagosome-associated RhoA was active, we exploited a RhoA biosensor in which the GBD domain of rhotekin is fused to GFP (GFP-rGBD) (Benink and Bement, 2005). RAW 264.7 cells stably expressing LifeAct-RFP were co-transfected with GFP-rGBD and imaged with spinning disk confocal microscopy during uptake of C3bi-sRBCs. The timing of enrichment of GFP-rGBD was near-simultaneous with LifeAct-RFP recruitment to phagosomes in the image snapshots acquired (Fig. 4C; Movie 3). Expression of the RhoA biosensor did not markedly influence F-actin flashing dynamics on phagosomes as the length of individual flashes was similar (Fig. 4D), with a small but insignificant increase in the time between flashes (Fig. 4E) compared with LifeAct-RFP-transfected cells alone. To understand more fully how active RhoA was being removed from phagosomes, we probed phagosomes for the RhoA-inactivating protein, p190RhoGAP (also known as ARHGAP5) (Amin et al., 2016; Ridley et al., 1993). Immunofluorescence analysis revealed that F-actin-positive but not F-actin-negative phagosomes were strongly labeled for p190RhoGAP (Fig. 4A,B).

We next tested for the functional requirement of the RhoA signaling pathway. We used a commercially available inhibitor of ROCK1, which is a downstream signaling protein of RhoA known to influence actomyosin contractility (Janoštiak et al., 2014). RAW 264.7 cells were pretreated with 10 μM of Y-27632 before being challenged with C3bi-opsonized target particles for 30 min prior to fixation and immunostaining for external particles and phalloidin to label F-actin flashes on phagosomes. In fixed cells, the overall frequency of F-actin flashing on phagosomes was intrinsically lower than in live cells, probably because of the transient nature of flashes on phagosomes (Fig. 4F). Regardless, treatment of cells with Y-27632 significantly decreased the frequency of F-actin flashes on phagosomes by ∼4.4-fold (Fig. 4F).

To investigate actin remodeling proteins, we probed for the presence of Arp2/3, as well as the Arp2/3-binding protein Wiskott–Aldrich syndrome protein (WASP), on ingested phagosomes in RAW 264.7 cells. We used an antibody to p34 (also known as ARPC2), a subunit of Arp2/3, and saw strong enrichment of both p34 and WASP on phalloidin-positive phagosomes, which was absent on non-flashing phagosomes (Fig. 4A,B). To delve more deeply into the role of Arp2/3 in F-actin flashing on phagosomes, we utilized the commercially available inhibitor CK-666, which we added to RAW 264.7 cells expressing LifeAct-RFP 15 min after particle uptake. After 45 min of particle ingestion, cells were fixed and imaged. Significantly fewer LifeAct-RFP-positive phagosomes were observed in CK-666-treated cells compared with DMSO-treated cells (Fig. 4H). We next examined the small subset of flashing phagosomes that persisted during CK-666 treatment to understand the role of Arp2/3 in flashing dynamics and particle deformation. We performed live-cell spinning disk confocal microscopy of LifeAct-RFP-transfected cells during CR-mediated phagocytosis and CK-666 treatment. The change in F-actin flashing dynamics on phagosomes in the presence of the inhibitor was striking. Frequently, LifeAct-RFP was recruited to phagosomes in a single flash that persisted for over 10 min on average (Fig. 4G,I; Movie 4), with many protracted flashes lasting over 20 min, compared with ∼3 min in untreated cells (Fig. 2A). The average time between flashes was also higher, averaging 6 min (Fig. 4I) compared with ∼2 min in untreated cells (Fig. 2C). These two flashing phenotypes predominated in CK-666-treated macrophages (Fig. 4J). We also investigated the impact of Arp2/3 inhibition on sRBC deformation using DIC imaging. In the presence of CK-666, sRBC deformation was only observed in ∼30% of LifeAct-RFP-positive phagosomes (Fig. 4K), compared with ∼60% of flashing phagosomes in untreated cells (Fig. 1H). Thus, Arp2/3 activity is important both for F-actin flashing dynamics on phagosomes and for cargo manipulation.

Because of the mechanical changes observed in flashing phagosome content, we next investigated whether the non-muscle myosin IIA (NMIIA) was present and important for these contractions. NMIIA is a myosin isoform found exclusively in hematopoietic cells (Maupin et al., 1994) and is required for formation of the phagocytic cup during CR-mediated phagocytosis (Olazabal et al., 2002). Using immunofluorescence, strong recruitment of NMIIA to F-actin-positive phagosomes in macrophages was observed during CR-mediated phagocytosis (Fig. 5A). To confirm the role of myosin in C3bi-sRBC deformation during F-actin flashing, we inhibited the activity of myosin IIA through blebbistatin treatment. Blebbistatin has a high affinity for the myosin-ADP-Pi complex, preventing it from releasing a phosphate group, effectively disabling it from inducing actomyosin contractions without altering its actin-binding properties (Kovács et al., 2004). We examined the effect of blebbistatin on F-actin flashing and C3bi-sRBC deformation in RAW 264.7 cells stably transfected with LifeAct-RFP (Fig. 5B). The frequency of F-actin flashes on phagosomes was slightly, but not significantly, reduced by blebbistatin treatment (Fig. 5C). We monitored sRBC deformation using DIC microscopy. The term ‘deformation’ was inclusive of sRBC lysis, fission and non-lytic deformation. In untreated cells, F-actin flashing phagosomes displayed deformation of sRBCs in 98.9±1.1% of cases. This was significantly reduced to 67.6±4.9% of phagosomes following blebbistatin treatment (Fig. 5D).

Phagosomes fuse with both early and late endosomes and ultimately lysosomes to form phagolysosomes (Fairn and Grinstein, 2012). To detect whether F-actin flashes on phagosomes influence fusion with components of the early endocytic pathway, we immunostained cells during phagocytosis for early endosome antigen 1 (EEA1) (Steele-Mortimer et al., 1999). Spinning disk confocal microscopy imaging revealed delayed recruitment of EEA1 to phalloidin-positive phagosomes (Fig. 6A). During the first 10 min after internalization, flashing phagosomes had significantly less EEA1 recruitment than non-flashing phagosomes (Fig. 6B). At 10 min after C3bi-sRBC-internalization, 5.6±5.6% of phagosomes enveloped with F-actin colocalized with EEA1, versus 33.1±3% of phagosomes when F-actin was absent (Fig. 6B). A delay in phagolysosome formation in FcγR-phagosomes recruiting F-actin has been previously demonstrated (Liebl and Griffiths, 2009). To investigate this during CR-mediated phagocytosis, we probed phagosomes for both F-actin and LAMP1, a marker for both lysosomes and late endosomes (Vieira et al., 2002). The number of LAMP1-positive phagosomes (both F-actin flashing and non-flashing) was determined at 30 min after C3bi-sRBC-binding (Fig. 6C). Quantification of immunofluorescence images revealed that phagosomes without F-actin had a significantly higher frequency of LAMP1 recruitment to the membrane (87.3±2.7%) compared with F-actin-flashing phagosomes (40.3±4.8%) (Fig. 6D).

Our DIC imaging analysis revealed pronounced deformation of sRBCs in F-actin-flashing phagosomes. Correlative transmission electron microscopy (TEM) also revealed physical changes in sRBCs in phagosomes that were surrounded by a dense F-actin meshwork, compared with non-flashing phagosomes (Fig. 7A). In this assay, we identified a macrophage containing one phagosome with F-actin flashing and two phagosomes that were not flashing, using epifluorescence microscopy (Fig. 7A). TEM analysis of the macrophage showed an extensive cytoskeletal network devoid of organelles surrounding the F-actin-flashing phagosome (Fig. 7A). The magnitude of the F-actin network surrounding the phagosome has been proposed to preclude association with endocytic organelles (Liebl and Griffiths, 2009). Interestingly, the C3bi-sRBC within the F-actin-positive phagosome was much more electron dense than the sRBCs in the non-flashing phagosomes (Fig. 7A), indicating a potential compaction or rupturing of the sRBC contents.

We determined that F-actin flashes delayed phagosome maturation; however, it remains to be determined whether these pre-contractions later aid the degradation of phagosome cargo. We hypothesized that deformation of particles leading to fragmentation would increase the surface area of the phagosome contents. This would augment the efficiency of degradation by proteolytic enzymes after lysosome fusion. To address this, we employed a proteolysis assay that allowed us to bind a self-quenching dye (DQ Green BSA) to C3bi-sRBCs through carbodiimide crosslinks. DQ-BSA only fluoresces after proteolysis. We fed LifeAct-RFP-expressing RAW 264.7 cells with DQ-C3bi-sRBCs and monitored DQ fluorescence up to 60 min post phagocytic cup formation (Fig. 7B). After about 30 min, DQ Green BSA fluorescence became significantly higher (∼3.4-fold) in F-actin-flashing phagosomes (Fig. 7B,C), suggesting acute degradation. It remained higher than in non-flashing phagosomes until the end of the assay (60 min), when the signal was ∼6.5-fold greater in F-actin-flashing phagosomes than in non-flashing phagosomes (Fig. 7C).

This study uncovers a novel tactic within the antimicrobial arsenal employed by macrophages, involving mechanical crushing of ingested opsonized particles. This is achieved by cyclical F-actin flashes on phagosomes prior to maturation. Unlike actin comet tails previously observed on IgG- and CR-phagosomes (Bohdanowicz et al., 2010; Liebl and Griffiths, 2009), F-actin flashes do not result in translocation of the phagosome within the cytoplasm and instead appear to be responsive to phagosome content stiffness. This behavior is reminiscent of the process behind durotaxis, whereby focal adhesion turnover is mediated by substrate rigidity (Lo et al., 2000; Pelham and Wang, 1997). Migrating cells experience greater focal adhesion turnover if adhered to a soft substrate, probably a result of increased FAK activity (Plotnikov and Waterman, 2013; Schober et al., 2007). This is in line with our observation that phagosomes containing malleable cargo experience twice as many F-actin flashes than phagosomes containing latex beads.

We determined that F-actin flashes on phagosomes predominately occur on C3bi-opsonized cargo. This implicates the CR3 integrin in initiating F-actin flashes on phagosomes and, indeed, CR3 is enriched on F-actin flashing phagosomes. A small subpopulation of CR-positive and phalloidin-negative phagosomes was also identified, which might represent phagosomes in-between F-actin flashes. Ideally, we would like to investigate the activation status of CR3 on phagosomes; however, the only available antibody for detecting high affinity CR3, the CBRM1/5 monoclonal antibody, is specific to the human protein and occludes the ligand-binding site (Diamond and Springer, 1993). The enrichment of talin on the flashing phagosome subset suggests that integrins are selectively engaged when phagosomes are flashing. Talin, a crucial link between integrins and actin filaments, promotes integrin clustering and activation by directly binding to the β-tail of integrins (Calderwood et al., 1999; Ratnikov et al., 2005). Importantly, talin is enriched at sites of C3bi-particle binding and required for internalization (Allen and Aderem, 1996; Jaumouillé et al., 2019; Lim et al., 2007). Using live-cell imaging, we observed a cyclical recruitment of mEmerald-talin coincident with LifeAct-RFP accumulation on phagosomes, indicating that talin is probably a major upstream driver of F-actin polymerization on phagosomes. Interestingly, we observed that overexpression of talin increased the time interval between F-actin flashes, possibly as a result of cytosolic sequestration of downstream signaling elements required for F-actin flash reassembly on phagosomes. The relatively small population of flashing phagosomes that we observed during FcγR-mediated phagocytosis could be induced by a smaller, yet significant, recruitment of CR receptors to the phagosome, known to assist with FcγR-mediated phagocytosis (Jongstra-Bilen et al., 2003).

We hypothesized that cyclical recurrences of F-actin networks on phagosomes are rooted in signaling pathways typically associated with focal adhesion signal transduction. The presence of mechanosensitive focal adhesion components and an actomyosin network at the phagosome membrane could allow tension to be applied to the CR-phagosome in a controlled manner. FAK binds directly to talin (Chen et al., 1995), is responsive to mechanical forces (Seko et al., 1999), and is interestingly an essential component for CR-mediated phagocytosis in neutrophils (Kasorn et al., 2009). Curiously, we did not detect the phosphoinositide PI(4,5)P₂, a known activator of FAK (Cai et al., 2008), on the membrane of F-actin-positive CR-phagosomes. However, reducing FAK levels promotes RhoA activation (Ren et al., 2000). Therefore, the absence of PI(4,5)P₂ on phagosomes would support an increase in RhoA activity during waves of F-actin recruitment, and may indeed be a factor in the continuous polymerization of actin after phagocytic cup formation. F-actin flashes on phagosomes were enriched in RhoA, which is activated in part by the β₂ tail of CR3 (Wiedemann et al., 2006) and is necessary for CR-mediated phagocytosis (Caron and Hall, 1998). RhoA signaling also drives the actomyosin pulses (Munjal et al., 2015) in epithelial cells during morphogenesis (Martin et al., 2009; Rauzi et al., 2010). The actomyosin waves in epithelial cells during development are very similar, although slightly quicker in duration than those seen on phagosomes (∼100 versus 210 s on phagosomes containing C3bi-sRBCs), suggesting that similar mechanisms may be at play. RhoA GAP/GEF temporal activities set the cycles of actomyosin assembly and disassembly during the actomyosin pulses in epithelial cells (Mason et al., 2016). In agreement, the RhoA biosensor, GFP-rGBD, revealed cyclical patterns of recruitment that coincided with F-actin enrichment on flashing phagosomes. Active RhoA was not observed on non-flashing phagosomes, suggesting targeted and localized RhoA activation on flashing phagosomes. The cyclical nature of RhoA activity and its removal from the phagosome could be explained, in part, by the presence of p190RhoGAP on flashing phagosomes. Interestingly, active RhoA inactivates myosin phosphatase through ROCK, thereby promoting myosin-based contractions (Kimura et al., 1996). We saw reduced F-actin recruitment to phagosomes following ROCK inhibition, probably due to the upstream role of ROCK in activating myosin II (Amano et al., 1996).

The oscillatory nature of F-actin flashes on phagosomes is probably a combination of positive and negative feedback loops influencing F-actin dynamics. Inhibition of Arp2/3 activity decreased the prevalence of F-actin flashes on phagosomes, similar to the inhibitory effect of F-actin coating on fusing insulin granules (Ma et al., 2020). Where F-actin flashes persisted in the presence of CK-666, the flashes were abnormally long with slower turnover cycles on phagosomes. This could be caused by the drug freezing previously assembled branched actin in place on phagosomes, or a switch to formin-based F-actin polymerization, which has been reported (Henson et al., 2015; Yang et al., 2012). Within these unusual flashing phagosomes, the physical manipulation of cargo was reduced. Under normal circumstances, F-actin depolymerization from the phagosome might occur from negative feedback loops induced by tension changes across the membrane. Tension changes modulate the binding of cofilin to F-actin (Hayakawa et al., 2011), potentially disassembling F-actin on the phagosome. RhoA activity downregulation might also be triggered by a build-up of focal adhesion components along the phagosome (Ren et al., 2000), preventing further actin polymerization. Once disassembled, repeated cycles of F-actin assembly could be a result of slip bonds between C3bi, integrin α_Mβ₂ and talin, shown to exist for integrin binding to ECM (Jiang et al., 2003). The tension generated by F-actin flashes on phagosomes is cyclical, which would enhance the strength of the bond. In an analogous scenario, repeated application of force on fibronectin-α₅β₁ bonds prolongs their lifespans in a phenomenon termed cyclical mechanical reinforcement (Kong et al., 2013). Stronger links could be produced in phagosomes containing malleable cargo. This would explain the relatively high number of F-actin flashes we observed on phagosomes containing sRBCs, to promote ‘mastication’ of the cargo. With malleable cargo, deformation of the substrate could lead to disengagement of integrins and reattachment to new ligands as the particle is manipulated in space. The ultimate cessation of F-actin flashing on phagosomes is probably a result of phagosome membrane remodeling that removes the crucial concentration of integrins or talin driving the signal, which remains to be tested experimentally.

In F-actin-flashing phagosomes, we noticed a delay in phagolysosome maturation, as evidenced by reduced LAMP1 staining on phalloidin-positive phagosomes. This is similar to observations in IgG-particle-overloaded macrophages, but opposite to what was seen in phagosomes associated with F-actin comet tails, which exhibited enhanced phagosome association with degradative organelles (Bohdanowicz et al., 2010; Liebl and Griffiths, 2009). It is not surprising that the presence of a dense F-actin coat precludes association with lysosomes, a strategy utilized by some internalized bacteria that recruit ‘actin cages’ to promote their growth (Lerm et al., 2006; Meresse et al., 2001). F-actin flashes on phagosomes strongly correlated with deformation of the ingested particles. The resulting pressure exerted from the actomyosin contractions on phagosomes was powerful enough to induce lysis of sRBCs in phagosomes. F-actin flashes on phagosomes could be analogous to chewing prior to digestion (in this case, phagosome maturation). It is also tempting to speculate that early manipulation also results in fragments of the larger particle becoming available to the macrophage to sample for antigen presentation. Indeed, we were able to capture fission events happening in concert with F-actin flashes, possibly representing an intracellular version of trogocytosis (Ahmed et al., 2008). Vesicles containing mechanically disrupted fragments would contain antigens within their native confirmation, necessary for cross-presentation (Batista and Harwood, 2009). After content sampling, the remaining ‘chewed’ particles, although initially experiencing a delay in maturation, would later be digested more efficiently with enzymes as a result of the increase in exposed surface area resulting from the mechanical forces.

Our discovery of F-actin flashes on phagosomes reveals a unique strategy employed by macrophages to destroy unwanted materials. Enzymatic degradation of contents within phagolysosomes is the dogma for clearance of unwanted extracellular contents and our work has added a mechanical ‘chewing’ activity imposed on phagosome cargo for digestion. The mastication of phagosome contents may also represent a unique strategy for obtaining native peptides from phagosomes for antigen presentation.

Dulbecco's modified Eagle's medium (DMEM) and fetal bovine serum (FBS) were purchased from Wisent (St-Bruno, QC). RBC lysis buffer and macrophage colony-stimulating factor (M-CSF) were purchased from Millipore Sigma (Oakville, ON). Sheep red blood cells (sRBCs) and rabbit anti-sRBC IgG were obtained from MP Biomedical (Solon, OH); rabbit anti-sRBC immunoglobulin M (IgM) was from Cedarlane Labs (Burlington, ON). The 3.87 μm crosslinked polystyrene divinylbenzene beads P(S/2% DVB) were purchased from Bangs Laboratories (Fishers, IN). E. coli strain MG1655 was from the University of Toronto Scarborough teaching laboratories. Paraformaldehyde (PFA) was from Canemco & Marivac (Lakefield, QC). Rat anti-mouse CD11b antibody (M1/70) and Alexa Fluor rhodamine phalloidin were purchased from ThermoFisher Scientific (Waltham, MA). Rabbit polyclonal E. coli antibody was from Novus Biologicals (Centennial, CO). Mouse monoclonal anti-RhoA (26C4), mouse monoclonal anti-FAK (D-1), rabbit polyclonal anti-phospho-FAK (Tyr397), mouse monoclonal anti-RhoGAP p190-B antibody (G-11), mouse monoclonal anti-WASP (B-9), rat monoclonal anti-LAMP1 (1D4B) and goat polyclonal anti-EEA1 (N-19) antibodies were all purchased from Santa Cruz Biotechnology (Dallas, TX). Rabbit polyclonal anti-p34-Arc/ARPC2 antibody was purchased from EMD Millipore Canada Ltd (Etobicoke, ON). Rabbit polyclonal anti-non-muscle myosin heavy chain II-A antibody (Poly19098) was purchased from BioLegend (San Diego, CA). Mouse monoclonal anti-talin antibody (8d4) was purchased from Millipore Sigma. Fluorescent secondary antibodies were purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). CK-666, blebbistatin, ROCK inhibitor (Y-27632), DAPI, phorbol 12-myristate 13-acetate (PMA), mouse serum, C5-deficient serum and all other reagents were purchased from Millipore Sigma (Oakville, ON).

The RAW 264.7 mouse macrophage cell line (TIB-71) was purchased from American Type Culture Collection (Manassas, VA). The cells were grown in DMEM containing 10% heat-inactivated FBS. RAW 264.7 macrophages (5×10⁵) were plated on 35 mm dishes 2 days prior to phagocytosis assays. Primary bone marrow-derived macrophages (BMDMs) were obtained by bone marrow extraction from the femurs and tibias of male and female C57BL/6 mice [aged 3 months, in accordance with the Canadian Council on Animal Care (CCAC)], as described (Khandani et al., 2007). Transfection of macrophages was carried out using FuGENE HD (Promega; Madison, WI) in 35 mm glass-bottom dishes, as recommended by the manufacturer. Electroporation-based transfection of BMDMs were performed with a Neon transfection system (Thermo Fisher Scientific, Waltham, MA), following the manufacturer's instructions. RAW 264.7 macrophages stably transfected with LifeAct-CFP (cerulean) constructs were a gift from Dr Ray Truant (McMaster University, Hamilton, ON) and those stably transfected with LifeAct-RFP constructs were produced in the laboratory by Dr He Song Sun. The GFP-C1-PLCδ-PH construct was deposited by Tobias Meyer (Addgene plasmid # 21179) and the GFP-C1-PLCδ-2PH construct was a gift from Sergio Grinstein (Hospital for Sick Children, Toronto, ON). Plasmids also obtained from Addgene were mEmerald-talin (plasmid #54266) and GFP-rhotekin GBD (rGBD) (plasmid # 26732) (Benink and Bement, 2005).

sRBCs (100 μl of a 10% suspension) were washed with PBS and then opsonized with 2 mg/ml rabbit anti-sheep IgG or IgM for 1 h on a rotator at room temperature. IgM-opsonized sRBCs were washed with PBS++ (0.5 mM CaCl₂ and MgCl₂) and then incubated with C5-deficient human serum, as described (Jongstra-Bilen et al., 2003). Polystyrene beads were washed with PBS and opsonized with human serum IgM (2 mg/ml; Millipore Sigma) prior to C5-deficient serum treatment. Overnight cultures of E. coli were opsonized with 100 µl of mouse serum for 1 h at 37°C. RAW 264.7 macrophages or BMDMs were plated 2 days prior to phagocytosis assays onto glass-bottom dishes or coverslips and the complement receptors activated using 150 nM PMA in serum-free DMEM for 7 min prior to incubation with opsonized sRBCs. Phagocytosis was allowed to proceed for specified times for live imaging, immunofluorescence or TEM analysis. For bacteria ingestion experiments, bacteria were added to cell cultures and spun onto cells at 300 g for 5 min at 8°C. For immunofluorescence assays, cells underwent phagocytosis for 15 min prior to washing of unbound particles or bacteria, after which phagocytosis was allowed to proceed for the indicated time periods. For ROCK inhibition, cells were pretreated with 10 μM Y-27632 ROCK inhibitor for 8 h prior to phagocytosis and fixation. To inhibit myosin, cells were pretreated with 100 μM blebbistatin prior to live-cell imaging. Drugs were maintained in culture during the phagocytosis assays.

Following phagocytosis for the indicated time periods, any remaining external sRBCs were lysed with ddH₂O prior to fixation. Cells were fixed using 4% PFA for 20 min. C3bi-beads were used instead of C3bi-sRBCs to determine the signaling elements involved in F-actin flashes, as the human IgM bead opsonin did not overlap with the primary antibodies used. Prior to permeabilization, any external polystyrene beads were labeled with anti-human secondary antibodies. Cells were permeabilized with 0.25% Triton X-100 for 20 min prior to blocking with 10% bovine serum albumin (BSA). Primary antibody staining in 1% BSA for 1 h included anti-E. coli (1:200), anti-CR3 (1:100; with milk block), anti-talin (1:250), anti-FAK (1:500), anti-phospho-FAK (Tyr397) (1:500), anti-RhoA (1:150), anti-RhoGAP p190-B (1:100), anti-WASP (B-9) (1:500), anti-p34-Arc/ARPC2 (1:500), anti-non-muscle myosin heavy chain IIA (1:250), anti-EEA1 (1:250) and rat monoclonal anti-LAMP1 (1:250). After washing, cells were incubated with phalloidin (1:500) and AffiniPure Cy2 secondary antibodies (1:500), followed by DAPI staining and mounting. Images were acquired using a Quorum WaveFX-X1 spinning disk confocal system (Quorum Technologies, Guelph, ON) through MetaMorph image acquisition software (Molecular Devices, Sunnyvale, CA).

LifeAct-CFP or LifeAct-RFP stably transfected RAW 264.7 macrophages were plated 2 days prior to imaging in 35 mm glass-bottom dishes (5×10⁵ cells per dish). Serum starvation and opsonization with sRBCs or polystyrene beads were carried out according to the cell culture protocol described earlier. RAW 264.7 macrophages were activated with 150 nM PMA in serum-free DMEM 7 min prior to being challenged with 50 μl of IgG- or C3bi-sRBCs. Throughout the duration of each live imaging session, the cells were kept at 37°C and 5% CO₂. The objectives used included a 40× lens (oil immersion) for cell population imaging and a 63× lens (oil immersion) for single-cell imaging. Four positions were defined per 35 mm glass-bottom dish, with 15 s intervals between rounds of imaging. Live-cell imaging was conducted using a Zeiss AxioObserver Z1 (Carl Zeiss Canada, Toronto, ON) inverted epifluorescence microscope using AxioVision (version 4.9.1.0, 64 bit) software for image capture. Differential interference contrast (DIC) images were taken in addition to epifluorescence channels. On average, 240 images were taken over the course of an hour at each position on cell culture dishes. In addition, RAW 264.7 macrophages stably expressing LifeAct-RFP were transfected with mEmerald-talin or GFP-rGBD, or treated with CK-666, and imaged using a Quorum WaveFX-X1 spinning disk confocal system (Quorum Technologies, Guelph, ON) and MetaMorph image acquisition software (Molecular Devices, Sunnyvale, CA). Four to six positions were acquired at 45 s intervals for 45 min using a 63× objective lens. All other parameters were reproduced from the movies acquired with the AxioObserver Z1.

To look at the ultrastructure of F-actin accumulation on C3bi-sRBC-containing phagosomes, 2×10⁵ RAW 264.7 macrophages stably transfected with LifeAct-RFP were plated in 6-well plates 2 days before the phagocytosis assay. The opsonization protocol for C3bi-sRBCs, as well as the live-cell epifluorescence imaging protocols described earlier, was used before sample processing for TEM. During this period of live-cell imaging, phagosomes showing F-actin recruitment were monitored before being fixed with 2% glutaraldehyde containing 0.5% saponin and 1% tannic acid for 1 h. The cells were then washed three times with 0.1 M sodium cacodylate buffer before adding 1% osmium tetroxide and 1.25% potassium ferrocyanide in sodium cacodylate buffer and incubating for 1.5 h. The samples were washed three times with 0.1 M sodium cacodylate buffer, followed by two washes with ddH₂O. The samples were then stained with 4% uranyl acetate for 30 min. Prior to dehydration, the samples were washed twice with ddH₂O. Dehydration was achieved using ethanol ranging from 70% to 100%. For the infiltration step, a 1:1 mixture of Epon and ethanol was applied for 45 min, followed by two incubations with 100% Epon for 75 and 105 min. To embed the samples, the coverslip was inverted into a silicone mold containing Epon, which was then left overnight at 70°C. Thin sections (80 nm) were cut using a microtome before being placed on a layer of Formvar film. Finally, the processed samples were stained with uranyl acetate and lead citrate. Images were acquired with a Hitachi H7500 TEM (Hitachi High Technologies Canada, Rexdale, ON).

To measure the intensity of proteolysis in phagosomes, an adaptation of a protocol developed by the Botelho research group was followed (Kim et al., 2014). sRBCs (200 μl) were re-suspended in 500 μL 1× PBS and 25 mg/ml of carbodiimide crosslinker. The mixture was placed on a rotator for 15 min. Excess carbodiimide was washed off three times with 1 ml of PBS (centrifugation at 3000 g for 5 min). The sRBC pellet was then re-suspended in 500 μl of PBS containing 1 mg of DQ Green BSA (conjugated with BODIPY FL fluorophores) (Thermo Fisher Scientific, Waltham, MA). The mixture was left on a rotator at 4°C for 12 h, while being protected from light. Excess DQ Green BSA was washed off three times with 500 ml of PBS (centrifugation at 3000 g for 5 min). The final pellet was re-suspended in around 200 μl of PBS and opsonized with C3bi using the same protocol as previously described. RAW 264.7 macrophages were challenged with DQ-C3bi-sRBCs and imaged using epifluorescence microscopy under the same parameters as previously described for live-cell imaging. Quantification for the proteolysis assay was based on densitometry measurements over 60 min. Three flashing and non-flashing phagosomes were monitored for DQ fluorescence. Images from the green channel (DQ Green BSA) at each time point were converted to grayscale using Adobe Photoshop CS6 (Adobe Systems, San Jose, CA). Densitometry measurements were taken using ImageJ (version 1.48, 64 bit; NIH, Bethesda, MD). Mean background density was subtracted from the mean density of the phagosome lumen for both flashing and non-flashing phagosomes. The average density value of the three flashing and non-flashing phagosomes was taken before each value was normalized to the initial value at time 0. The resulting numbers represented the fold-change in DQ Green BSA intensity from the initial measurements.

Images acquired by epifluorescence microscopy were quantified using AxioVision (version 4.9.1.0, 64 bit; Carl Zeiss Canada) and ImageJ (version 1.48, 64 bit; NIH, Bethesda, MD). Spinning disk confocal image acquisition was carried out using MetaMorph; image processing was done using Volocity for fixed cell analyses (contrast adjustments and file exportation). Spinning disk confocal movies were acquired using MetaMorph and processed using Fiji/ImageJ (version 2.0) with the plugins Bio-Formats, Z Projection and Annotation to Overlay. TEM images were acquired, processed and exported using iTEM (version 5.2; Olympus Soft Imaging Solutions, Muenster, Germany). Figures were prepared using Adobe Illustrator CS6 (Adobe Systems, San Jose, CA).

For quantification purposes, F-actin flashing during live-cell phagocytosis was defined as brief and symmetrical increases in F-actin polymerization at the phagosome membrane, occurring post phagocytic cup formation. The F-actin flashes observed using DIC imaging were characteristically transient and oscillatory (more than one round of F-actin recruitment) in stationary phagosomes that were often associated with target particle deformation. To quantify F-actin flash dynamics and intensity, an region of interest (ROI) was defined around an internalized particle from image frames acquired using live epifluorescence microscopy that was then processed using Volocity (version 6.1.1, 64 bit; Perkin Elmer, Waltham, MA). Quantification of frames began when the C3bi-sRBC became bound to the macrophage. The arbitrary fluorescence intensity unit (AFU) from the appropriate color channel was taken for every frame in the defined ROI. The frequency of F-actin flashing on phagosomes was calculated by dividing the number of flashing phagosomes by the total number of phagosomes.

Unless otherwise indicated, for quantifications, 50 phagosomes were assayed in three biological replicates. For the immunofluorescence assays, F-actin-positive and F-actin-negative phagosomes were scored for enrichment of candidate proteins. Total phagosomes assayed were RhoA and WASP (20 phagosomes), talin (50 phagosomes), FAK and phospho-FAK (Tyr397) (90 phagosomes), CR3 (>150 phagosomes for each time point), p190RhoGAP (>250 phagosomes) and p34 (>350 phagosomes). EEA1-positive phagosomes were tabulated at 5 min intervals post-internalization. The percentage of EEA1-positive phagosomes among total phagosomes over time (1 h) was then calculated. The number of LAMP1-positive phagosomes was determined for both F-actin flashing and non-flashing phagosomes to calculate and compare the proportion of LAMP1-positive phagosomes in both cases. Three independent experiments were performed for these assays.

For Arp2/3 inhibition assays, phagocytosis was performed for 15 min to ensure particle internalization prior to addition of 150 µM CK-666. LifeAct-RFP-transfected cells were then imaged live; DMSO- and CK-666-treated cells were left for an additional 30 min prior to fixation. For ROCK inhibition assays, the percentage of phalloidin-positive phagosomes for each condition (treated and untreated) was determined by counting phagosomes using the ‘Cell Counter’ plug-in for ImageJ (version 1.48, 64 bit; NIH, Bethesda, MD). Three independent experiments were conducted. Live-cell DIC imaging was used to assess qualitatively the deformation of C3bi-sRBCs through visual inspection in blebbistatin-treated cells. A C3bi-sRBC was considered deformed if its circularity was noticeably altered (≥10% change by eye) and occurred synchronously with F-actin flashing. Slight fluid-like motions on the surface of phagosome membranes during F-actin flashes were not sufficient to be considered noticeable alterations of C3bi-sRBC circularity. Cells were imaged every 30 s for 2 h, at multiple positions on the glass coverslips. Deformation frequency (deformed C3bi-sRBCs divided by non-deformed C3bi-sRBCs) was compared between flashing and non-flashing phagosomes. The frequency of C3bi-sRBC deformation phenotypes was calculated by dividing the number of flashing phagosomes exhibiting one of the three phenotypes by the total number of flashing CR-phagosomes.

An unpaired two-tailed Student's t-test was used to determine statistical significance between two groups (P-value <0.05 was considered statistically significant). For comparing three groups or more, a non-parametric one-way ANOVA was used, followed by Tukey's test for multiple comparisons. The mean±s.e.m. of each group is displayed in both cases. Prism (version 5.0) was used to conduct the statistical analyses (GraphPad Software, La Jolla, CA).

We thank Bob Temkin in the Centre of the Neurobiology of Stress (University of Toronto Scarborough) for assistance with TEM analysis. We thank Mauricio Terebiznik (University of Toronto Scarborough), Roberto Botelho (Ryerson University) and Sergey Plotnikov (University of Toronto) for critical reading of the manuscript.

The peer review history is available online at https://jcs.biologists.org/lookup/doi/10.1242/jcs.239384.reviewer-comments.pdf