Dynamin‐2 controls actin remodeling for efficient complement receptor 3‐mediated phagocytosis

Phagocytosis is the mechanism of the internalization of large particles, microorganisms and cellular debris. The complement pathway represents one of the first mechanisms of defense against infection and the complement receptor 3 (CR3), which is highly expressed on macrophages, is a major receptor for many pathogens and debris. Key to dissecting the mechanisms by which CR3‐mediated phagocytosis occurs, is understanding how the complex actin binding protein machinery and associated regulators interact with actin during phagocytosis, from triggering of receptor, through to phagosome formation and closure.


INTRODUCTION
Phagocytosis is the mechanism of the internalization of large particles, microorganisms, and cellular debris (Depierre et al., 2021;Uribe-Querol & Rosales, 2020). It has a nutritional role in Amoebae and, in the professional phagocytes of the immune system, phagocytosis plays a key role in the development of immune responses via cytokine secretion and presentation of antigen derived et al., 2020a). It is a phagocytic integrin also known as αMβ2 that needs to be activated to bind to extracellular ligands. The actin dynamics play a key role first in the activation of the integrin receptors via inside-out signals and force-dependent conformation changes of the β2 chain, that is, anchored to the cortical actin cytoskeleton (Torre-Gomez et al., 2020a). Second, actin polymerization is induced in downstream signaling pathways that are controlled by the small GTPase RhoA, the formin mDia1, and the Arp2/3 nucleator (Caron & Hall, 1998;Colucci-Guyon et al., 2005;May et al., 2000). In addition, a subtle interplay between the microtubules and the actin cytoskeleton is important for CR3-mediated phagocytosis (Allen & Aderem, 1996;Binker et al., 2007;Lewkowicz et al., 2008). The early steps of CR3 engagement and signaling have been dissected to show how the Syk kinase and focal adhesion proteins act as a molecular clutch to promote phagocytosis (Jaumouille et al., 2019). CR3-mediated phagocytosis was initially reported to rely on particle "sinking," without extensive formation of membranous lamellipodia (Allen & Aderem, 1996;Kaplan et al., 1977). Some membrane extension and ruffling activity have however been associated with CR3-mediated phagocytosis (Jaumouille et al., 2019;Patel & Harrison, 2008). Overall, much less is known about the mechanisms of phagosome formation and closure downstream of CR3 than in the case of other phagocytic receptors such as FcR.
Dynamin-2 is an important mediator of budding and scission of small vesicles (Morlot & Roux, 2013). Dynamin participates in phagosome maturation in Dictyostelium discoideum via F-actin binding and endosome tubulation (Gopaldass et al., 2012). In addition, Dynamin-2 was found to be involved in two different steps of Receptors for Fc portion of Immunoglobulins (FcR)-mediated phagocytosis, both during pseudopod extension and actin polymerization, and during phagosome closure (Marie-Anais et al., 2016b). The role of Dynamin-2 in phagocytosis of zymosan and IgGcoated particles had already been reported previously without pointing to a precise mechanism (Gold et al., 1999). Given the strong requirement for Rho, ROCK, and Myosin II in actin assembly and particle engulfment in CR3-mediated phagocytosis (Olazabal et al., 2002), it could be hypothesized that CR3 phagosome formation and closure could rely on other molecular players than Dynamin-2. Here, we investigated whether Dynamin-2 plays a role in integrin-phagosome formation. Dynamin recruitment at the phagocytic cup during CR3-mediated phagocytosis was investigated in primary human monocyte-derived macrophages (hMDMs) using super resolution microscopy. Concomitant recruitment of F-actin and Dynamin-2 was monitored from the onset of CR3-phagosome formation until closure by live cell imaging with total internal reflection fluorescence microscopy (TIRFM). Actin quantification in phagocy-tosis assays revealed that Dynamin-2 is vital for the formation of the phagocytic cup, and therefore, efficient CR3-mediated phagocytosis.

RESULTS
Dynamin-2 is recruited concomitantly with actin during CR3 phagosome formation and closure in macrophages Direct stochastic optical reconstruction microscopy (dSTORM) was used to visualize primary human macrophages (hMDMs) in a ''3D'' arrangement as described previously (Marie-Anais et al., 2016a;Mularski et al., 2018). For this, hMDMs sediment onto coverslips coated with red blood cells (RBCs) opsonized with IgM followed by complement (C'-IgM-RBCs), and noncovalently bound to a poly-L-lysine activated surface (shown schematically in Figure 1a). Cells were allowed to phagocytose the opsonized RBCs for 10-15 min prior to fixation and staining to detect both F-actin and Dynamin-2 by dSTORM ( Figure 1b). F-actin is accumulated in focal adhesion structures as well as at the site of nascent phagosomes (Figure 1b, left panel). Dynamin-2 is enriched in these F-actin rich phagocytic cups, although not exactly at the same place as F-actin accumulations (Figure 1b, middle panel). A tessellationbased colocalization analysis (Levet et al., 2015) was performed yielding a colocalization value for each single molecule localization of each species separately, minimizing false positive colocalization. The colocalization parameter, the Spearman coefficient, can have values from −1 (anticorrelation) through 0 (no correlation and therefore a low probability of colocalization) to 1 (perfect correlation and therefore a high probability of colocalization). The analysis allowed for a comparison of the actin-Dynamin and Dynamin-actin colocalization in the cell body (blue dashed line, Figure 1b) and the phagocytic cups (white dashed line, Figure 1b). The Spearman coefficients of actin with Dynamin and Dynamin with actin from analysis of regions of interest in the cell body are close to zero (means of 0.024 and 0.15, respectively), indicating a low probability of colocalization ( Figure 1c). The Spearman coefficients for actin-Dynamin and Dynamin-actin colocalization from regions of interest around phagocytic cups yield positive means (means of 0.30 and 0.28, respectively), indicating a higher probability of colocalization in the phagocytic cups than in the cell body ( Figure 1c). This result seems to indicate that Dynamin-2 is part of the CR3-phagosome formation process.
To better describe the role Dynamin-2 plays in CR3dependent phagocytosis, we used TIRFM to observe the dynamics of recruitment of F-actin and Dynamin-2 in RAW264.7 macrophages transfected to transiently FIGURE 1 Dynamin-2 is recruited with actin during CR3 phagosome formation and closure. STORM visualization of hMDMs undergoing ''3D'' phagocytosis shows accumulation of F-actin and Dynamin-2. Schematic representation of a cell undergoing ''3D'' phagocytosis (a). Primary human macrophages are allowed to sediment on coverslips coated with loosely bound C'-IgM-RBCs and then fixed, permeabilized, and labeled. Actin was labeled with phalloidin-Alexa 647 and Dynamin-2 with Alexa-555. Cells were visualized with STORM (b), asterisks indicate the position of the RBCs, bar = 2.5 μm. Image analysis determined Spearman coeffcients for actin (red circles) and dynamin (green circles) in regions of interest in cell body (b, blue dashed line, nine cell body regions) and phagocytic cups (b, white dashed line, 14 phagocytic cups) (c). Phagosome closure assay was performed using RAW 264.7 macrophages transiently expressing both LifeAct-mCherry and Dynamin-2-GFP. Shown schematically in d, cells are allowed to phagocytose opsonized particles that are non-covalently bound to the coverslip, allowing the site of closure to be observed using TIRFM. TIRFM images of a representative cell (of 10 replicate experiments) expressing LifeAct-mCherry (e, top row) and Dynamin-2-GFP (e, middle row) and brightfield images (e, bottom row) were captured every 0.5 s for 10 min at 37 • C. The RBC phagocytosed is clearly visible in brightfield images at 0 and 60 s (b, bottom row, red circles), but pulled from the focal plane inside the cell by 90 s. Bar = 5 μm.
express LifeAct-mCherry and Dynamin-2-GFP. Cells were exposed to coverslips coated with C'-IgM-RBCs non-covalently bound to poly-L-lysine activated surface. After transfected cells were allowed to sediment to the surface, phagocytic cup extension and closure around the bound RBCs were observed in the TIRF region, as described in detail in (Mari-Anais et al., 2016a, 2016bMularski et al., 2018) and shown schematically in Figure 1d. A time-course of a representative macrophage (of 10 replicate experiments) undergoing CR3 mediated phagocytosis is shown in Figure 1e (movie available in Video S1). The RBC phagocytosed is clearly visible in brightfield images at 0 and 60 s (E, bottom row, red circles), but pulled from the focal plane inside the cell by 90 s. F-actin and Dynamin-2 are recruited concomitantly throughout the phagocytic event, from membrane extension around the RBC, through to fusion and scission, before disassembly. Dynamin-2 was recruited with actin at the tips of membrane extensions surrounding the RBC and accumulated at the site of phagosome closure and scission.
These results confirm that Dynamin-2 is present at different stages of CR3 phagosome formation and suggest a role for Dynamin-2 in phagosome completion from the plasma membrane.

Phagocytosis, but not association, is impaired by Dynamin-2 inhibition
To determine the role of Dynamin-2 in CR3-mediated phagocytosis, we inhibited its function prior to performing phagocytosis assays, as in (Mari-Anais et al., 2016a, 2016b. hMDMs were treated with DMSO (control) or with two different inhibitors of dynamin activity: dynasore and pyrimidyn-7. After treatment, cells were centrifuged in medium containing C'-IgM-RBCs for 2 min to synchronize phagocytosis. Macrophages were either fixed at this point to assess the efficiency of RBC association (0 min), or, were allowed to phagocytose C'-IgM-RBCs for 60 min at 37 • C to evaluate phagocytic efficiency. For both time points, following fixation, macrophages were FIGURE 2 Pharmacological inhibition of Dynamin-2 impairs phagocytosis but not association. hMDMs were treated with DMSO (control, a and e), dynasore monohydrate (80 μM, b and f) or Pyrimidyn-7 (4 μM). hMDMs were then allowed to phagocytose C'-IgM-RBCs for 0 (a and b) and 60 min (e and f), fixed and stained to detect bound but not internalized RBCs with Cy3-coupled anti-rabbit IgG antibodies. Internalized RBCs were detected using phase contrast. Average number of associated RBCs per cell (c and d) and internalized RBCs (g and h) were calculated in 25 cells per donor at 0 and 60 min, respectively. The means of three independent experiments are plotted (* p < 0.01). Bar = 10 μm. stained for RBCs (Figure 2a-e). The number of external (detected by fluorescence) and internal (detected by fluorescence and/or phase contrast) RBCs was quantified. The average number of associated (bound + internalized) RBCs per cell and internalized RBCs per cell were calculated in 25 cells per donor at 0 and 60 min, respectively. The average number of associated RBCs per cell from three independent experiments on different donors (Figure 2c,d), revealed no significant difference between cells treated with DMSO, dynasore, or pyrimidyn-7, in their ability to engage with C'-IgM-RBCs. The average number of internalized RBCs per cell (Figure 2g,h), however, demonstrate that phagocytic efficiency is significantly reduced in cells treated with dynasore or pyrimidyn-7 relative to control cells (**p < 0.01). This result indicates that inhibiting Dynamin-2 leads to the impairment of uptake without interfering with integrin-receptor binding to complement.

Inhibition of Dynamin-2 decreases F-actin recruitment to the phagocytic cup
Phagocytosis is an active process requiring F-actin polymerization to provide the force with which membranes are deformed to envelop the phagocytic target (Mularski & Niedergang, 2017;Torre-Gomez et al., 2020a). As CR3-mediated phagocytosis is impaired when Dynamin-2 is inhibited, F-actin quantification was performed at the site of phagocytosis in cells treated with DMSO (control) or dynasore. Phagocytosis was synchronized by centrifugation of hMDMs with C'-IgM-RBCs for 2 min. Macrophages were then allowed to phagocytose C'-IgM-RBCs for 15 min, then fixed and stained for RBCs and F-actin. Quantification was performed on four different macrophage preparations from four different donors. In control cells, many F-actin rich phagocytic cups under C'-IgM-RBCs were easily located (Figure 3a ). When

FIGURE 3
Pharmacological inhibition of Dynamin-2 decreases the F-actin recruitment to the phagocytic cup. hMDMs were treated with DMSO (control) or dynasore monohydrate (80 μM). hMDMs were allowed to phagocytose C'-IgM-RBCs for 15 min, then fixed and stained to detect bound but not internalized RBCs with Cy3-coupled anti-rabbit IgG antibodies. Cells were then re-fixed, permeabilized and F-actin was stained with phalloidin-Alexa Fluor 488. Images for DMSO (a) and dynasore-treated (b) cells were acquired using an inverted wide-field microscope (Leica DMI6000) with a 100× (1.4 NA) objective. Bar = 10 μm. Actin enrichment was quantified in phagocytic cups (a, b, red arrows): maximum fluorescence intensities of selected regions of interest were background corrected by subtracting the maximum intensity value from a membrane region in the same cell with no phagocytic event. Ratio values obtained by dividing the fluorescence intensities in the phagocytic cups by fluorescence intensities in the cell body were calculated (c). This ratio was expressed as a percentage of F-actin enrichment relative to control cells (d). The results of four independent experiments are plotted (n = 15 actin cups/condition, **** p < 0.01). Bar = 10 μm, inset box = 12 μm.
Dynamin-2 was inhibited, although RBCs were associated with the membrane, there were fewer phagocytic cups (Figure 3b). Regions of interest were selected where C'-IgM-RBCs were attached and a phagocytic cup was visible. F-actin recruitment was quantified as a ratio between the fluorescence in the phagocytic cup and the fluorescence in a cell body region without particle, as described in the Materials and Methods section. The index of F-actin enrichment at sites of phagocytosis was 1.8 ± 0.7 (Figure 3c) in the control cells treated with DMSO. Macrophages pre-treated with dynasore monohydrate exhibited a significantly lower F-actin ratio cup to body index: 0.75 ± 0.4 (Figure 3c), representing 50% decrease in the F-actin recruitment at the site of CR3-phagocytosis in dynasore-treated macrophages as compared with control cells (Figure 3d). Therefore, Factin enrichment in CR3 phagocytic cups is decreased when Dynamin-2 is inhibited by dynasore monohydrate, implicating Dynamin-2 in the actin assembly of the CR3 phagocytic cup.

DISCUSSION
In this study, we reveal that Dynamin-2 is crucial for CR3 integrin-mediated F-actin recruitment and phagosome completion. The CR3-mediated phagocytosis was often described as a "sinking" process relying on the formation of reduced phagocytic cups (Allen & Aderem, 1996;Kaplan et al., 1977). As pointed out recently (Jaumouille et al., 2019;Patel & Harrison, 2008), when analyzed using live cell imaging methods, it becomes visible that this type of phagocytosis does also generate some membrane extensions. dSTORM and live TIRFM experiments used in this study also showed membrane protrusions containing F-actin that close over the phagocytic target ( Figure 1). As CR3 is an integrin, formed by the αMβ2 pair, the actin structures formed around the target particle are similar to focal adhesions, in that they contain talin, vinculin, and associated proteins (reviewed in [Torre-Gomez et al., 2020a]). Integrin activity is highly regulated by conformational changes. During the "inside-out" phase of integrin activation, there is an actomyosin-dependent stretching of talin that allows the release of RIAM and subsequent binding to vinculin (Torre-Gomez et al., 2020b;Vigouroux et al., 2020), which reinforces the association to the β2 tail and its activation. Interestingly, it seems that Dynamin-2 was not implicated in regulating these actin-dependent steps of integrin activation in human macrophages that are required for efficient ligand binding prior to phagocytosis.
Instead, Dynamin-2 was critical for the "outside-in" signal downstream of CR3 leading to actin polymerization. This complex pathway involves the tyrosine kinase Syk, the small GTPases of the Rho family RhoG, and the actin nucleators VASP, Arp2/3 and Formin, as well as an actomyosin activity to orchestrate this integrinmediated phagosome formation (Colucci-Guyon et al., 2005;Jaumouille et al., 2019;Olazabal et al., 2002;Torre-Gomez et al., 2020a). Here, we provide novel evidence that Dynamin-2 controls actin polymerization downstream of the phagocytic CR3. We used both a dynamin inhibitor, which might have some off-target effects in the absence of Dynamin (Preta et al., 2015), as well as transient expression of the Dynamin-2 K44A mutant (Orth et al., 2002), that is, unable to bind and hydrolyze GTP (not shown). Dynamin-2 has been shown to interact directly with actin in vitro, and its interaction with short F-actin filaments was found to promote its GTPase activity and its oligomerization, which was confirmed in cellulo (Gu et al., 2014). It was recently further deciphered how Dynamin bundles actin filaments in vitro and how this mechanism could play a role in the context of cell-cell fusion (Zhang et al., 2020). The assembly-stimulated GTPase activity of Dynamin triggers rapid GTP hydrolysis and helix disassembly. Of note, the nucleoside diphosphate kinase (NDPK), an important regulator to fuel GTP to Dynamin, is recruited together with Dynamin-2 at the site of phagocytosis, thus ensuring efficient phagocytic uptake in various experimental models (Farkas et al., 2019). This mechanism of actin bundling could also be relevant in the context of CR3-phagocytosis. We have indeed shown previously that the integrity of the F-actin network is necessary for sustained recruitment and function of Dynamin-2 downstream of the phagocytic receptors for the FcR and, reversely, that Dynamin-2 activity was necessary for an efficient assembly of the F-actin cup upon FcR ligation (Mari-Anais et al., 2016a, 2016b. This is also in line with observations that actin polymerization precedes Dynamin recruitment at clathrin-coated pits in the last phases of clathrin-mediated endocytosis before the scission and release of vesicles (Grassart et al., 2014;Merrifield et al., 2002). Given the specificities of the CR3 integrin signaling and morphological features of the phagocytic cups, that Dynamin-2 plays a crucial role during CR3-phagosome shaping is particularly noteworthy.
Interestingly, the macropinocytic capture that shares many morphological features with CR3-phagocytosis has been described to be independent from Dynamin activity (Le Roux et al., 2012;Liberali et al., 2008). In the context of CR3-phagocytosis, F-actin could help control the membrane tension and therefore complement the constriction activity of Dynamin-2 for scission, as proposed (Morlot & Roux, 2013). Thus, Dynamin might contribute to the mechano-sensing and responding strategies used by the phagocytic cells to adapt their response to the physical properties of their targets, which is highly relevant to clear pathogens, debris, and tumor cells (Mularski & Niedergang, 2017).
Human peripheral blood mononuclear cells were isolated from whole blood of healthy donors (Etablissement Français du Sang Ile-de-France, Site Trinité, INSERM agreement #18/EFS/030 ensures all donors have provided written informed consent to provide anonymous samples) by density gradient sedimentation using Ficoll-Plaque (GE Healthcare). Cells were allowed to adhere in 6 or 24 well plates at 37 • C for 2 h in adhesion medium (RPMI 1640-GlutaMAX supplemented with 2 mM Lglutamine and 100 μg/mL streptomycin/penicillin) before cultivating them in complete medium (RPMI 1640-Glu-taMAX supplemented with 10% FCS, 2 mM L-glutamine, and 100 μg/mL streptomycin/penicillin). Every 2 days, complete medium was exchanged. The adherent monocytes were left to differentiate into macrophages (referred to as human monocyte-derived macrophages (hMDMs)) in the text as described previously (Jubrail et al., 2020) and used for experiments from day 5 to day 10 of differentiation.
External RBCs are defined as particles positive for the fluorescent labeling and not detectable by phase contrast due to their physical properties. Internal RBCs appear weakly stained with the labeled F(ab')2 antirabbit IgG because the cells are fixed prior to labeling. In addition, internal RBCs are detectable by phase contrast as holes in the cells.
External and internalized RBCs were counted in 25 randomly chosen cells, and the mean number of phagocytosed RBCs per cell was calculated. The number of cell-associated (bound + internalized) RBCs, was also counted, to calculate the mean number of associated RBCs per cell. Images were acquired using an inverted wide-field microscope (Leica DMI6000) with a 100× (1.4 NA) objective and a (MicroMAX Princeton Instruments) camera. Actin enrichment was quantified in phagocytic cups: maximum fluorescence intensities of a selected region of interest were background corrected by subtracting the mean value from a membrane region in the same cell with no phagocytosis event. Ratio values obtained by dividing the fluorescence intensities in the phagocytic cups by fluorescence intensities in the cell body were calculated as in (Braun et al., 2007).

Electroporation of RAW264.7 cells
Cells were transfected by electroporation with the Elec-troBuffer kit from Cell Projects Ltd (EB-110). One 100 mm plate of cells, grown to sub-confluence, and 20 μg of plasmid (10 μg per plasmid for co-transfection) were used for each electroporation. Cells were electroporated in 4 mm cuvettes (Biorad) at 250 V, 900 μF in an electroporation apparatus (X pulser Bio-Rad Laboratories) then immediately resuspended in complete culture medium and plated. Cells were used for experiments 15-18 h post electroporation.

Phagosome closure assay
Functionalized glass bottom dishes for phagosome closure assays were prepared according to the method described in (Marie-Anais et al., 2016a, 2016bMarion et al., 2012;Mularski et al., 2018). Briefly, C'-IgM-SRBCs were centrifuged onto 35 mm glass bottom dishes (Mat-Tek) pre-treated with 0.01% poly-L-lysine in PBS for 30 min at RT. After washing once with 10% BSA in PBS, dishes were incubated in 10% BSA to saturate exposed poly-L-lysine. Dishes were then washed and filled with pre-warmed serum-free microscopy medium (RPMI 1640-GlutaMAX without phenol red, supplemented with 10 mM HEPES, 2 mM L-glutamine, 1 mM sodium pyruvate, 50 μM β-mercaptoethanol), and main-tained at 37 • C. Transfected cells were then allowed to sediment onto C'-IgM-RBCs adhered to the dish surface.

Total internal reflection fluorescence microscopy (TIRFM)
TIRFM was performed using a Till PHOTONICS iMIC microscope equipped with an oil immersion objective (Apo N 100×, NA1.49 Olympus America Inc.), a heating chamber maintained at 37 • C and two cameras: a cooled iXonEM camera and an iXon3 897 Single Photon Detection EMCCD Camera (Andor Technology). The critical angle was determined prior to imaging by scanning through incident angles from 0 • to 5 • to maximize evanescent wave induced fluorescence (Mari-Anais et al., 2016a, 2016b. Excitation was performed with 491 and 561 nm lasers. For the phagosome closure assay, images were acquired at 50 ms per frame in bright field illumination, TIRF mode, and in epifluorescence mode with polychrome illumination at 3 μm increment. TIRFM image stacks were processed using ImageJ Color Profiler software (NIH).