Candida albicans hyphae can reach enormous lengths, precluding their internalization by phagocytes. Nevertheless, macrophages engulf a portion of the hypha, generating incompletely sealed tubular phagosomes. These frustrated phagosomes are stabilized by a thick cuff of F-actin that polymerizes in response to non-canonical activation of integrins by fungal glycan. Despite their continuity, the surface and invaginating phagosomal membranes retain a strikingly distinct lipid composition. PtdIns(4,5)P2 is present at the plasmalemma but is not detectable in the phagosomal membrane, while PtdIns(3)P and PtdIns(3,4,5)P3 co-exist in the phagosomes yet are absent from the surface membrane. Moreover, endo-lysosomal proteins are present only in the phagosomal membrane. Fluorescence recovery after photobleaching revealed the presence of a diffusion barrier that maintains the identity of the open tubular phagosome separate from the plasmalemma. Formation of this barrier depends on Syk, Pyk2/Fak and formin-dependent actin assembly. Antimicrobial mechanisms can thereby be deployed, limiting the growth of the hyphae.https://doi.org/10.7554/eLife.34798.001
Billions of microorganisms live on, and in, the human body. Known as the human microbiome, most of these microscopic hitchhikers are harmless. But, for people with a compromised immune system, common species can sometimes cause disease. For example, the yeast Candida albicans, which colonises between 30 and 70% of the population, is normally harmless, but can switch to a disease-causing version that makes branching structures called hyphae. These hyphae grow fast, piercing and damaging the tissues around them.
Immune cells called macrophages usually engulf invading microbes. These cells recognise sugars on the outside of C. albicans, and respond by wrapping their membranes around the yeast, drawing the microorganism in, and sealing it into closed structures called phagosomes. Then, the macrophages fill the phagosomes with acid, enzymes and destructive chemicals, which breaks the yeast down. Yet, C. albicans hyphae grow larger than macrophages, making them difficult to control.
Maxson et al. have now tracked the immune response revealing how macrophages try to control large hyphae. The immune cells were quick to engulf C. albicans in its normal yeast form, but the response slowed down in the presence of hyphae. Electron microscopy revealed that the large structures were only partly taken in. Rather than form a closed phagosome, the macrophages made a cuff around the middle of the hypha, leaving the rest hanging out.
The process starts with a receptor called CR3, which detects sugars on the outside of the hyphae. CR3 is a type of integrin, a molecule that sends signals from the surface to the inside of the immune cell. A network of filaments called actin assemble around the hypha, squeezing the membrane tight. The macrophage then deploys free radicals and other damaging chemicals inside the closed space. The seal is not perfect, and some molecules do leak out, but the effect slows the growth of the yeast. When a phagosome cannot engulf an invading microbe, a state that is referred to as being “frustrated”, the leaking of damaging chemicals can harm healthy tissues and lead to inflammation and disease.
These findings reveal that macrophages do at least try to form a complete seal before releasing their cocktail of chemicals. Understanding how the immune system handles this situation could open the way for new treatments for C. albicans infections, and possibly similar diseases related to “frustrated engulfment” (such as asbestos exposure, where asbestos fibers are also too large to engulf). However, one next step will be to find out what happens to partly engulfed hyphae, and how this differs from the fate of fully engulfed yeast.https://doi.org/10.7554/eLife.34798.002
Candida albicans is a commensal fungus that colonizes the epithelial surfaces of 30–70% of healthy individuals (Perlroth et al., 2007). However, in immune-compromised individuals, C. albicans can cause invasive, life-threatening disease. The mortality rate for infected patients is 46–75%, with candidiasis classified as the fourth most common nosocomial bloodstream infection (Brown et al., 2012). Invasive candidiasis is correlated with a switch of C. albicans from its yeast form to a hyphal form, a shift that can be induced in vitro by nutrient deprivation among other cues (reviewed in Sudbery, 2011). In vivo, C. albicans hyphae are capable of invading epithelium and endothelium; in addition C. albicans is capable of forming recalcitrant biofilms and inducing inflammation (Sudbery, 2011). These conditions activate host defense mechanisms for the control and clearance of C. albicans, mounted predominantly by phagocytic cells of the innate immune system.
Phagocytes can effectively sense, internalize and kill invasive C. albicans. Accordingly, impairment of the phagocytic response, e.g. by elimination of macrophages and neutrophils, is associated with disseminated candidiasis (reviewed in Netea et al., 2015). Phagocytic cells possess receptors that bind the C. albicans cell wall and trigger uptake of the fungus into a phagosome. The C. albicans cell wall is composed mostly (80–90%) of polysaccharides, containing ≈ 60% β-(1,3) and -(1,6) glucans, and ≈ 40% O- and N-linked mannans (Ruiz-Herrera et al., 2006). As such, the main non-opsonic phagocytic receptors for C. albicans are the C-type lectin family of receptors, including Dectin1, the mannose receptor, and DC-SIGN (reviewed in Hardison and Brown, 2012). The phagosome typically matures rapidly after closure, evolving into an acidic, degradative and microbicidal compartment. Acquisition of antimicrobial properties by this compartment depends on its ability to accumulate and retain toxic compounds, including reactive oxygen species (ROS). Superoxide produced by the NADPH oxidase undergoes dismutation into hydrogen peroxide in the acidic luminal environment generated by the V-ATPase, which additionally favors the catalytic activity of various hydrolases. Transporters such as NRAMP-1, that antagonize microbial growth by depleting the phagosome of nutrients, also depend on phagosomal H+ for the extrusion of metal ions.
Unlike most other microbes, C. albicans presents a distinct problem for phagocytes. The hyphal form of C. albicans can grow at a rate of 18.8 μm hr−1 (GOW and Gooday, 1982), quickly exceeding the size of the phagocytes themselves. The challenge is greatest for macrophages, which migrate to infection sites later than the polymorphonuclear cells, and thus encounter growing hyphae (reviewed in Erwig and Gow, 2016). Despite being remarkably plastic, macrophages have difficulty engulfing the much larger C. albicans hyphae, an impasse that no doubt contributes to the pathogenesis of candidiasis.
The aim of the current study was to examine the dynamic and complex process of C. albicans phagocytosis by macrophages. We found that attempts to engulf large hyphae result in the formation of incomplete (frustrated) phagosomes, which nevertheless segregate a section of the hypha, preferentially exposing it to microbiostatic products. The mechanism and fungal components underlying the formation of the diffusion barrier established by the phagocyte when generating the frustrated phagosome was analyzed using a combination of imaging, pharmacological and genetic approaches.
To optimize the phagocytosis of C. albicans, which has a cell wall rich in β-glucans (Gow et al., 2011), we used RAW 264.7 macrophages stably expressing the Dectin1 receptor (RAW-Dectin1; Esteban et al., 2011). Yeast or hyphal forms of C. albicans expressing BFP (Candida-BFP; Strijbis et al., 2013) were used as targets to facilitate their visualization. Under the conditions used to generate them, C. albicans hyphae were considerably longer (>15 μm) than the macrophages (8–10 μm in diameter). After 1 hr of co-incubation with the macrophages the yeast form was fully engulfed (Figure 1A), while a significant number of hyphal C. albicans were only partially internalized (68.5% ± 4.5, while 31.5% ± 4.6 were fully internalized; 1019 events from 12 independent experiments), which was verified using fluorescent concanavalin A to label exposed hyphae (Figure 1B). This was similar to the frustrated engulfment of >20 μm C. albicans hyphae reported earlier (Lewis et al., 2012). Transmission electron microscopy confirmed that most hyphae were only partially internalized (Figure 1C) and, in addition, revealed the existence around the neck of the frustrated phagosome of a low-contrast structure seemingly devoid of membrane-bound organelles (Figure 1C, inset), previously interpreted by Strijbis et al., 2013 as accumulated actin. Indeed, this region corresponded to an actin-rich cuff-like structure (Figure 1D); F-actin was so highly accumulated at the cuff that the remainder of the cellular actin could only be visualized when images were overexposed (Figure 1D, inset). Note that the remainder (i.e. the base) of the frustrated phagocytic cup was virtually devoid of F-actin. 3D visualization verified the continuous accumulation of F-actin around the neck of the tubular phagosomes lining individual hyphae and its sharp delineation of the intracellular and extracellular portions of the fungus (Figure 1E,F,G,H and Video 1). This actin cuff was observed for RAW-Dectin1 cells engulfing C. albicans hyphae up to 100 μm in size (data not shown), and occurred in 96.3% ± 1.9 of the partially internalized hyphae (674 events analyzed in 12 independent experiments). These data support published accounts of actin cuff-like structures seen during the phagocytosis of various filamentous targets (García-Rodas et al., 2011; Gerisch et al., 2009; Heinsbroek et al., 2009; Prashar et al., 2013; Strijbis et al., 2013). The occurrence of frustrated phagocytosis with formation of a pronounced actin cuff was not unique to the RAW-Dectin1 cell line; similar features were seen when murine or human primary macrophages were confronted with C. albicans hyphae (Figure 1—figure supplement 1A and B, respectively). The actin cuff was remarkably stable, lasting for at least 90 min without contracting (Figure 1I). Nevertheless, the actin composing these structures undergoes measurable turnover (treadmilling), since the cuffs underwent gradual disassembly when the cells were treated with latrunculin A, which scavenges actin monomers (last two panels, Figure 1I). These long-lasting yet dynamic cuffs identify the frustrated phagocytic cups generated by macrophages attempting to eliminate C. albicans hyphae.
We proceeded to probe the receptors whose signaling could potentiate the formation of the actin cuff. Because C-type lectin signaling contributes importantly to C. albicans phagocytosis (de Turris et al., 2015; Tafesse et al., 2015; Xu et al., 2009), we analyzed whether Dectin1 accumulated in the membrane at sites where cuffs were evident. Remarkably, while Dectin1 was clearly concentrated in patches elsewhere along the frustrated phagocytic cup, it was poorly detectable by immunostaining near the actin cuff (ratio cuff: cup 0.60 ± 0.04; n = 30 p<0.0001; Figure 2A and inset). The failure to detect accumulation of Dectin1 at these sites was not attributable to masking of the exofacial epitope, possibly resulting from tight apposition to the hyphae, because similar results were obtained when the receptors were tagged with emerald fluorescent protein and visualized directly in live cells (ratio cuff: cup 0.56 ± 0.04; n = 15, p<0.0001; Figure 2B and inset).
In epithelial and endothelial cells, host E- or N-cadherin, respectively, have been reported to contribute to C. albicans internalization (Moreno-Ruiz et al., 2009). This process involved the recruitment of α- and β-catenins and activation of the Arp2/3 pathway for actin nucleation. In agreement with these reports, we observed E-cadherin and β-catenin accumulation at sites of where C. albicans hyphae were being internalized by epithelial A431 cells, with particular accumulation at sites where actin polymerized (Figure 2—figure supplement 1). We considered whether a similar mechanism was responsible for the formation of actin cuffs by macrophages. However, neither E-cadherin nor β-catenin was detectable in RAW-Dectin1 cells or in primary human macrophages by immunoblotting (Figure 2C) or by immunofluorescence (not illustrated). Under comparable conditions, robust signals were obtained when probing A431 cells (Figure 2C). When expressed heterologously in macrophages E-cadherin-GFP was found to line the surface membrane, but was absent from the phagocytic cup (Figure 2D), while β-catenin-GFP was largely soluble and did not accumulate at the cuff (Figure 2E). Thus, E-cadherin and β-catenin are unlikely to mediate phagocytosis of C. albicans in macrophages. Nevertheless, low levels of expression of these proteins (below the level of detection of our assays) or other cadherins may have mediated the internalization. This possibility was assessed by treating the cells with EDTA, which chelates the Ca2+ known to be required for ligand binding by cadherins (reviewed in Brasch et al., 2012). As shown in Figure 2F, omission of Ca2+ had no effect on actin cuff formation in C. albicans-infected RAW-Dectin1 cells.
Actin can also be tethered to the phagocytic cup via integrins (Freeman et al., 2016). Integrins can be directly or indirectly involved in the phagocytosis of opsonized particles, apoptotic cells and a variety of other targets (reviewed in Dupuy and Caron, 2008) and link with actin filaments via talin and vinculin (reviewed in Shattil et al., 2010). However, canonical integrin activation and ligand binding require divalent cations (reviewed in Leitinger et al., 2000), and would therefore be inhibited by their chelation with EDTA. Moreover, actin cuffs formed normally in CALDAG-GEF1−/− macrophages (Figure 3—figure supplement 1), consistent with the notion that cuff formation was independent of canonical activation of integrins, which involves Rap1 (reviewed in Hogg et al., 2011). There is, however, one atypical instance where integrin activation can occur in the absence of divalent cations. The α chain of the integrin complement receptor 3 (CR3, also referred to as Mac1), is unique in that it contains a lectin-like domain (LLD) that binds carbohydrates in a divalent cation-independent manner (Thornton et al., 1996). The LLD is separate from the I-domain –the conventional ligand-binding domain of integrins (reviewed in Ross, 2002)– and, interestingly, binds fungal β-glucan (Ross et al., 1985; Vetvicka et al., 1996). We therefore proceeded to test whether CR3, which consists of αM (CD11b) and β2 (CD18) subunits, is present in the region of the actin cuff. As illustrated in Figure 3, both CD11b and CD18 accumulated in the region of the actin cuff in RAW-Dectin1 cells that had partially internalized C. albicans hyphae (CD11b ratio cuff: cup 4.75 ± 0.29; n = 30, p<0.0001; CD18 ratio cuff: cup 4.79 ± 0.28; n = 30, p<0.0001; Figure 3A,B and insets). Moreover, talin, vinculin and paxillin were also localized to the cuff (Figure 3C,D and insets; Figure 3—figure supplement 1E and inset), as was HS1, the homologue of cortactin in leukocytes (Figure 3E and inset). Like cortactin, HS1 is thought to regulate actin nucleation and branching (Daly, 2004).
The preceding findings support a model whereby ligation of β-glucan by the LLD causes outside-in activation of CR3 directly (O'Brien et al., 2012; Vetvicka et al., 1996), or in conjunction with Dectin1 signaling (Huang et al., 2015; Li et al., 2011), resulting in Arp2/3-dependent actin nucleation. This model was tested using the M1/70 antibody, which binds to CD11b between its β-propeller and thigh domains (residues 614–682; Osicka et al., 2015) and effectively blocks the binding of CR3 to β-glucan (Xia et al., 1999). Cells pretreated with M1/70 failed to show accumulation of CR3 around partially internalized C. albicans hyphae, and their ability to form actin cuffs was markedly impaired (Figure 3H); actin cuffs were much less prominent or missing altogether when CR3 was blocked (Figure 3F versus G). The number of fully internalized C. albicans did not differ between conditions (Figure 3H). We concluded that binding of the CR3 integrin to C. albicans was critical for the establishment of long-enduring actin cuffs observed during frustrated phagocytosis of the hyphae.
Dectin1 and CR3 both bind β-glucans (Brown and Gordon, 2001; Brown et al., 2002; Ross et al., 1985; Vetvicka et al., 1996), and have been reported to cooperate during phagocyte responses to fungal pathogens (Huang et al., 2015; Li et al., 2011). Dectin1 has also been reported to cooperate with TLR2, TLR4 (Ferwerda et al., 2008; Netea et al., 2006; Netea et al., 2002) and mannose receptors (Astarie-Dequeker et al., 1999; Bain et al., 2014; Lewis et al., 2012; McKenzie et al., 2010; Netea et al., 2006) in the recognition of C. albicans. We therefore sought to clarify the receptors and ligands involved in actin cuff formation.
Untransfected RAW 264.7 cells express negligible levels of Dectin1 (Brown et al., 2003; Esteban et al., 2011; Taylor et al., 2004), providing a means to assess the contribution of this receptor to actin cuff formation. As shown in Figure 4A, RAW 264.7 cells rarely formed actin cuffs compared to RAW-Dectin1 cells, suggesting that initial engagement of the hyphae by Dectin1 was essential. The requirement for Dectin1 in C. albicans phagocytosis (Marakalala et al., 2013; Taylor et al., 2007) could be bypassed when the hyphae were serum-opsonized, enabling opsonin receptors to establish the initial contact with the fungus (Figure 4A). Thus, while not accumulating in the region of the cuff, Dectin1 binding to the hyphae (which is evident by its accumulation in the frustrated phagocytic cup; Figure 2A and B) is required for the subsequent activation of F-actin polymerization by CR3.
We also studied cooperativity by using soluble ligands to competitively block defined receptors, and scoring the frequency of actin cuff formation (Figure 4B). Soluble mannan, a ligand for mannose receptor, had no effect on actin cuff formation by RAW-Dectin1 cells. Accordingly, we did not find mannose receptors in the membrane lining the actin cuff (data not shown). Laminarin, a soluble β-glucan ligand for Dectin1 (Brown and Gordon, 2001; Brown et al., 2002) impaired phagocytosis and actin cuff formation when present prior to and during phagocytosis, but not if added after the hyphae had adhered to the RAW-Dectin1 cells (Figure 4B). These findings support the notion that Dectin1, but not mannan receptors, cooperate with CR3 to generate the actin cuffs.
C. albicans cell wall components include β-(1,3)-glucans, β-(1,6) glucans, O- and N-linked mannans and chitin (Netea et al., 2008; Ruiz-Herrera et al., 2006). These can contribute to the recognition of C. albicans by phagocytes (reviewed in Netea et al., 2008), and potentially also to actin cuff formation. To clarify the contribution of individual wall components we used gene replacement and conditional expression (GRACE) strains (Roemer et al., 2003) with specific depletion targeting chitin, mannan, and β(1,6)-glucan biosynthetic pathways upon incubation with doxycycline (Table 1; O'Meara et al., 2015). Repression of pathways involved in chitin, mannan and β(1,6)-glucan synthesis using doxycycline did not affect actin cuff formation (Figure 4C,D and data not shown), implying that these components are dispensable. We next assessed the role of β(1,3)-glucan through pharmacological inhibition of Fks1 with caspofungin (Douglas et al., 1997), as genetic depletion of Fks1 results in defects in hyphae formation (Ben-Ami et al., 2011). Remarkably, the ability to form actin cuffs was greatly reduced in C. albicans grown and allowed to form hyphae in the presence of caspofungin (Figure 4E and F). The inhibitory effect of caspofungin on actin cuff formation was dose-dependent (Figure 4F), reaching ≈ 80% at 5 ng mL−1 caspofungin, a dose that reduced the β(1,3)-glucan content of the wall by 55.3%, as assessed by aniline blue staining. Actin cuff formation around caspofungin-treated hyphae could not be rescued by serum opsonization (Figure 4—figure supplement 1), suggesting that β(1,3)-glucan is the ligand that promotes actin cuff assembly via CR3. Interestingly, Aspergillus fumigatus hyphae (routinely exceeding 80 µm in length) were also able to illicit actin cuff formation by RAW-Dectin1 cells (Figure 4G). A. fumigatus hyphae, while displaying some unique cell wall components compared to C. albicans hyphae, also have cell wall-associated β(1,3)-glucan (Erwig and Gow, 2016). We concluded that ligation of fungal β(1,3)-glucan by CR3 is required for actin cuff formation during frustrated phagocytosis of long hyphae.
Despite the paucity of Dectin1 and mannose receptors (Figure 2A and B), phosphotyrosine was markedly concentrated at the cuff (Figure 5A), possibly as a consequence of CR3 activation. While there is disagreement over the requirement of Src-family kinases (SFKs) for the interaction of phagocytes with fungal targets (Elsori et al., 2011; Herre et al., 2004; Le Cabec et al., 2002; Mansour et al., 2013; Underhill et al., 2005), there is evidence that Syk, as well as Pyk2 and Fak, two related tyrosine kinases, participate in CR3-mediated phagocytosis (Li et al., 2006; Paone et al., 2016; Zhao et al., 2016). The contribution of individual kinases to the tyrosine phosphorylation was explored next.
Phosphorylated SFKs accumulated along the frustrated phagocytic cup (Figure 5A) where Dectin1 was also found (Figure 2A and B), but were not particularly enriched in the region of the actin cuff (ratio cuff: cup 0.98 ± 0.03; n = 17, p=0.61). SFK inhibition by PP2 following adherence of the hyphae to RAW-Dectin1 cells had no effect on actin cuff formation (Figure 5E). In contrast, the phosphorylated (active) forms of Pyk2 and Fak were enriched solely at the actin cuff (pPyk2 ratio cuff: cup 23.69 ± 1.20; n = 46, p<0.0001, pFak ratio cuff: cup 22.56 ± 1.01; n = 34, p<0.0001; Figure 5C and D). Moreover, inhibition of Pyk2/Fak activity by PF573228 following adherence of the hyphae to the cells abolished actin cuff formation, with no effect on internalization (Figure 5E). Also, as reported by Strijbis et al., 2013, we observed phosphorylation of Syk with accumulation at the actin cuff (ratio cuff: cup 21.55 ± 1.75; n = 22, p<0.0001; Figure 5—figure supplement 1). As expected, inhibition of Syk by piceatannol after C. albicans adherence blocked actin cuff formation (Figure 5E). These data provide evidence that, along with Syk, Pyk2/Fak play a role in the interaction between macrophages and C. albicans, and are important for actin cuff formation during frustrated phagocytosis of hyphae.
Interestingly, the interaction of Pyk2 with β2 integrins activates Vav1 (Gakidis et al., 2004; Kamen et al., 2011), a GEF for Rho-family GTPases that is also essential for the phagocytosis and control of C. albicans by macrophages (Strijbis et al., 2013). Accordingly, Rac1 and/or Cdc42 were seemingly involved in the marked polymerization of actin at the cuff. This was indicated by the recruitment of PAK(PBD), a biosensor of the active (GTP-bound) form of these GTPases (Benard et al., 1999), that accumulated at the cuff to levels ≥4 fold higher than along the cup. F-actin accumulation at the cuff was sensitive to the formin inhibitor SMI-FH2, but not to the Arp2/3 inhibitor CK-666 (Figure 5G). Together, these data suggest that activation of Syk and Pyk2/Fak by CR3 leads to activation of Rho-family GTPases, culminating in formin-mediated actin assembly, a process akin to focal adhesion formation (reviewed in Vicente-Manzanares et al., 2005).
Phospholipids undergo striking changes during the course of conventional phagocytosis. PtdIns(4,5)P2 that is normally found in the plasma membrane is converted to PtdIns(3,4,5)P3 at sites of receptor engagement, and is subsequently degraded by lipases and phosphatases, becoming undetectable in sealed phagosomes. PtdIns(3,4,5)P3 can be detected for up to a minute following sealing, but then disappears abruptly as PtdIns(3)P appears; the latter is detectable on early phagosomes for about 10–15 min (reviewed in Levin et al., 2015). These drastic switches are thought to reflect and possibly dictate the identity and developmental stage of the maturing phagosome. It has been observed that the frustrated tubular phagosomes of heat-killed filamentous Legionella pneumophila are accompanied by a sharp separation of plasmalemmal and phagosomal phosphoinositide species (Naufer et al., 2018; Prashar et al., 2013). Additionally, atypical phosphoinositide dynamics can occur in sealed phagosomes containing filamenting C. albicans (Heinsbroek et al., 2009) or during CR3-mediated phagocytosis of opsonized targets (Bohdanowicz et al., 2010). Therefore we analyzed the phosphoinositides in frustrated phagosomes of C. albicans hyphae. We used the genetically-encoded fluorescent biosensor PLCδ-PH-GFP to monitor the distribution of PtdIns(4,5)P2. Remarkably, while PtdIns(4,5)P2 was present as expected in the surface membrane facing the extracellular milieu, it was undetectable in the invaginated section that constituted the frustrated phagosome (Figure 6A). In stark contrast, PtdIns(3,4,5)P3 –which was visualized using AKT-PH-GFP– was found solely in the open phagosomal cup (Figure 6B), where it co-existed with PtdIns(3)P, detected using the PX-GFP sensor (Figure 6C). In addition to the localization of PtdIns(3,4,5)P3 in the cup reported in a previous collaborative study (Strijbis et al., 2013), we detected additional enrichment of PtdIns(3,4,5)P3 in the actin cuff region (ratio cuff: cup 1.39 ± 0.10; n = 30, p=0.0006). In contrast, PtdIns(3)P was comparatively excluded from the actin cuff (ratio cuff: cup 0.823 ± 0.05; n = 30, p=0.0025). The segregation of these phosphoinositides persisted for the duration of our observations (up to 90 min after frustrated phagosome formation; not illustrated).
The sharp boundary between the PtdIns(4,5)P2-rich surface membrane and the tubular membrane endowed with PtdIns(3,4,5)P3 and PtdIns(3)P coincided with the location of the actin cuff, suggesting that the latter may function as a diffusion barrier. However, the restricted localization of the phosphoinositides may have resulted from the strategic positioning of synthetic (i.e. kinases) and degradative (i.e. phosphatases or lipases) enzymes. To more definitively assess the existence of a diffusion barrier, we analyzed the distribution and dynamics of molecules that do not undergo rapid metabolic transformation, including lipid-anchored and transmembrane proteins, which had been reported to segregate in frustrated phagosomes. As shown in Figure 6D, LC3 –a small protein covalently linked to PtdEth– was found in the frustrated phagosome (Kanayama and Shinohara, 2016; Martinez et al., 2015; Sprenkeler et al., 2016; Tam et al., 2016), yet did not reach the surface membrane. Similarly, both wild-type Rab7 (Figure 6E) and constitutively-active Rab7 (not illustrated) are confined to the frustrated phagosomal tube and partially excluded from the actin cuff (Rab7 ratio cuff: cup 0.68 ± 0.05; n = 30, p<0.0001), as was LAMP1 (ratio cuff: cup 0.59 ± 0.03; n = 30, p<0.0001; Figure 6F), a late-endosomal/lysosomal membrane-spanning glycoprotein. The exclusion from the actin cuff was better appreciated by 3D visualization of LAMP1 (Figure 6G,H and Video 2). Because metabolic conversion to other species could not account for the segregation of the latter probes to the invaginated section of the membrane, we considered it more likely that restricted diffusion accounted for the observations.
It was nevertheless possible that molecules like LC3, Rab7 or LAMP were inserted through fusion into the tubular part of the membrane, where they could conceivably remain immobile. To exclude this possibility, we assessed their mobility measuring fluorescence recovery after photobleaching (FRAP). The constitutively-active form of Rab7, Rab7(Q67L), was used for these experiments; because this variant is unable to exchange nucleotides, it does not associate stably with GDI and remains membrane associated (Méresse et al., 1995), eliminating the confounding effects of fluorescence recovery from a cytosolic pool. Rapid recovery was observed following photobleaching of a ≈3 µm spot within the phagosomal cup. In four independent experiments, half-maximal recovery was attained after 3.3 s (Figure 7B). Similar analyses were performed using GFP-tagged LAMP1 (Figure 7A,B), which also recovered within seconds (t1/2 = 7.9 sec). Between 75–80% of the fluorescence was recovered in both instances, implying that the majority of the Rab7(Q67L) and LAMP1 molecules were mobile.
The retention of Rab7(Q67L) and LAMP1 in the cup for many minutes despite their ability to move laterally in the plane of the membrane implies that they are unable to cross the junction with the surface membrane. The existence of a diffusion barrier was confirmed by expressing the N-terminal domain of Lyn (Lyn11) tagged with GFP. This region of the protein becomes myristoylated and palmitoylated, targeting it to the plasma membrane and, to a lesser extent, to early endosomes. Following frustrated phagocytosis of hyphae, Lyn11-GFP is found both at the membrane and in the phagosomal cup, where its density is lower, likely because of dilution caused by insertion of unlabeled endomembranes. We analyzed comparatively small phagosomes to enable photobleaching of Lyn11-GFP in the entire cup (Figure 7C). Strikingly, the fluorescence of the cup failed to recover, despite the persistence of abundant Lyn11-GFP in the adjacent plasmalemma. In three independent experiments only 19% of the original fluorescence reappeared, possibly via fusion with Lyn11-GFP-containing early endosomes. Failure to recover was not attributable to immobility of Lyn11-GFP in the membrane, which displayed very fast and nearly complete recovery following photobleaching (Figure 7D,E). These data confirm that the region of the actin cuff acts as a lateral diffusion barrier, separating the inner leaflet of the plasma membrane from that of the open phagocytic cup.
It is noteworthy that while the barrier curtails the diffusion of lipids and proteins anchored to lipids on the inner leaflet of the membrane, exofacial lipids and lipid-associated proteins readily traverse the junction between the membrane and the tubular phagosome. This was demonstrated by incorporation of rhodamine-labeled PtdEth to the surface membrane following stabilization of the frustrated phagosome. The labeled lipid, which inserts into the outer leaflet of the plasmalemma, reached the entire membrane of the frustrated phagocytic cup within ≈5 min (Figure 7—figure supplement 1A). Similarly, fluorescent cholera toxin B subunit, which binds to exofacial ganglioside GM1, promptly entered the phagocytic cup (Figure 7—figure supplement 1B). Thus, the actin-dependent diffusion barrier selectively restricted the mobility of components of the inner leaflet, including transmembrane proteins, while exofacial lipids remained able to traverse the junction.
How is the diffusion barrier generated? We speculated that the molecular crowding resulting from tight clustering of integrins and their ancillary proteins could restrict the diffusion of membrane-associated components across the cuff. To test this possibility, we investigated whether sufficient molecular crowding could be generated to exclude other membrane components from regions of integrin clustering. To this end, we used antibody-induced cross-linking, a strategy shown earlier to induce the formation of CR3 patches on the plasma membrane (Fukushima et al., 1996; Pavan et al., 1992; Zhou et al., 1993). Whether exclusion could be induced by molecular crowding was assessed analyzing the distribution of CD2-CD45-GFP (Figure 8B), a transmembrane protein having a short, 7 nm ectodomain (Cordoba et al., 2013). As shown in Figure 8A, prior to cross-linking both CD2-CD45-GFP and CR3 were distributed diffusely throughout the membrane, overlapping extensively at the resolution of the confocal microscope. The CD2-CD45-GFP fluorescence intensity in CR3-positive regions compared to the average CD2-CD45-GFP fluorescence intensity of the entire plasma membrane averaged 0.69 ± 0.01 (585 CR3-positive regions in 20 cells from three different experiments). After antibody treatment, CR3 clustered into large, dense patches. Strikingly, CD2-CD45-GFP was largely (81%) and significantly (p>0.0001) excluded from such patches, where the fluorescence was only 0.13 ± 0.01 of the plasmalemmal average (measured in 472 CR3 patches in 15 cells from three experiments). Importantly, the exclusion was not alleviated by treatment with latrunculin A, the fluorescence of the patches averaging 0.12 ± 0.01 of the plasmalemmal average (measured in 445 patches in 18 cells from three experiments), implying that the actin cytoskeleton is not involved in the domain segregation. CD2-CD45-GFP exclusion did not differ between these two conditions (p=0.66). We concluded that integrins could be sufficiently clustered to exclude other membrane components. By forming a continuous and thick ring around the neck of the frustrated phagosome, the molecular crowding of clustered integrins could generate a diffusional barrier.
While actin is not essential to constrain the diffusion across patches of antibody-aggregated integrins, it is nevertheless required to maintain the integrins clustered in response to the glucan during frustrated phagocytosis. As such, an intact actin cuff is required to establish and maintain the barrier to phosphoinositides or transmembrane proteins. This was validated in cells that had formed a stable frustrated phagosome around C. albicans hyphae and were then treated with latrunculin A, which was shown earlier (Figure 1I) to cause gradual disassembly of the cuff. PtdIns(4,5)P2 –which in untreated cells is excluded from the phagocytic cup (Figures 6A and 9A)– gained access to the entire cup when actin was disassembled by latrunculin (Figure 9B). The PtdIns(4,5)P2 present in the cup, expressed relative to the plasmalemma, increased 4.88 times after latrunculin treatment (Figure 9C). Conversely, LAMP1 –that is restricted to the cup in untreated cells (Figures 6F and 9D)– was able to reach the surface membrane following treatment with latrunculin (Figure 9E). After latrunculin treatment, the ratio of LAMP1 present in the cup decreased 3.99 times (Figure 9F). Clearly, while clustering of CR3 is sufficient to form a diffusional barrier (Figure 8), the actin cuff formed during phagocytosis of C. albicans hyphae likely contributes to the stability of the barrier between CR3 and C. albicans β(1,3)-glucans, presumably by maintaining integrins in their active conformation (Kaizuka et al., 2007; Lavi et al., 2007; Lavi et al., 2012) during frustrated phagocytosis.
Despite remaining unsealed, frustrated phagosomes acquired markers of endosomes and lysosomes, implying that they had undergone at least partial maturation. It was therefore conceivable that the cells established the diffusion barrier in an effort to generate a microbicidal compartment, despite their inability to form a sealed vacuole. Acidification of the lumen, secretion of antimicrobial enzymes and peptides and deployment of the NADPH oxidase are among the principal mechanisms used by leukocytes to eliminate pathogens. We first tested the ability of frustrated phagosomes to generate and maintain an acidic lumen, using the fluorescent acidotropic dye LysoBrite Red dye. As expected, the dye accumulated in lysosomes; however, it was never found to concentrate inside the frustrated phagosome (Figure 10A), suggesting that vacuolar ATPases are not functional on its membrane and/or that the junction separating the lumen from the extracellular milieu is permeable to H+. That the latter interpretation is correct was suggested by determinations of permeability of the junction using dextrans of varying size. For these experiments lysosomes were loaded with either 10 kDa or 70 kDa fluorescent dextran and then exposed to C. albicans hyphae. The dextrans were delivered into fully formed (sealed) phagosomes, where they were clearly retained (Figure 10—figure supplement 1A). The smaller (10 kDa) dextran, however, was not detectable inside frustrated phagosomes; the reduced overall staining of the cells (cf. main panel and inset in Figure 10B) suggests that secretion of lysosomes did occur, but that the dextran must have escaped the confines of the frustrated phagosome. In contrast, the 70 kDa dextran was readily visible along the frustrated phagosome, implying that its diffusion into the external medium was limited. Thus, a size-selective filter determined the extent to which solutes were retained within the frustrated phagosome. The cut-off of this filter must be greater than ≈50 kDa, because cathepsin D, a globular protein of ≈28 kDa, managed to escape the frustrated phagosome (Figure 10D), yet was routinely detected in sealed phagosomes (Figure 10—figure supplement 1B). Therefore, the incomplete phagocytic cup formed around partially internalized hyphae would be expected to have limited degradative capacity towards C. albicans. These data are in accord with the findings of (Prashar et al., 2013) that showed frustrated L. pneumophila phagosomes to retain large molecular weight dextrans, but not protons or lysosomal enzymes, despite acquisition of the V-ATPase and fusion with lysosomes.
Though unable to retain luminal macromolecules over extended periods of time, the partial barrier to diffusion at the mouth of the frustrated phagosome, together with the geometrical constraint posed by the length and narrowness of the luminal space, are expected to delay the exit of molecules secreted into the phagosome. Rapidly reacting molecules may therefore be able to exert microbicidal/microbiostatic effects under these circumstances. Such is the case of reactive oxygen species produced by the NADPH oxidase. Indeed, we were able to detect preferential deposition of formazan, a product of the reaction of superoxide with nitroblue tetrazolium (NBT), inside frustrated phagosomes (Figure 10E). Heat-killed or paraformaldehyde-killed C. albicans hyphae were utilized for these experiments, eliminating the need to account for superoxide production by live C. albicans (see Materials and methods).
Because the frustrated phagosome appeared to retain some antimicrobial function, we assessed the effect of the frustrated phagosome environment on the fate of partially internalized hyphae. We could not detect significant loss of viability of the partially internalized C. albicans, as assessed by propidium iodide staining. We reasoned that the antimicrobial effectors may not suffice to kill the fungus, yet their effects may manifest as an observable change in the rate of hyphal extension, which can average 0.31 µm min−1 on serum agar (GOW and Gooday, 1982). When measured in RPMI medium without serum (wthout macrophages present) C. albicans hyphae grew at a rate of 0.22 µm min−1 ±0.03. Remarkably, the extension rate of partially internalized hyphae, which displayed an actin cuff, was significantly reduced (0.11 µm min−1 ±0.01). This reduced growth rate was not different (p=0.742) to that of fully internalized hyphae (0.108 µm min−1 ±0.011). In the same experiments, neighboring C. albicans hyphae not in contact with macrophages grew at a rate of 0.198 µm min−1 ±0.014, indistinguishable from that measured in the absence of macrophages. Therefore, while small molecular weight contents can eventually diffuse out of the frustrated phagosome, they are nevertheless retained sufficiently to limit the growth of partially internalized C. albicans. This microbiostatic effect on partially internalized C. albicans hyphae could be ablated by blocking macrophage CR3 with the M1/70 antibody (see Figure 3F,G and H) before phagocytosis (Figure 10—figure supplement 2), reiterating the importance of CR3 ligation to β(1,3)-glucan for the generation and maintenance of this atypical phagocytic environment.
Most of the microbicidal and degradative properties of the phagosome depend on the release and containment of lysosomal hydrolases, antimicrobial peptides and reactive oxygen species in close proximity to the internalized microorganism. However, when phagocytes are faced with exceptionally large targets, their internalization can become retarded or frustrated altogether. The inability to complete phagocytosis, as in the case of long asbestos fibers (Donaldson et al., 2010) or bacterial biofilms (Costerton et al., 1999; Scherr et al., 2014; Thurlow et al., 2011) can potentiate harmful inflammation.
C. albicans hyphae can attain lengths of ≥50 μm (GOW and Gooday, 1982), overwhelming the comparatively diminutive phagocytes that are unable to ingest them whole. Accordingly, attempts to internalize such hyphae are frustrated and inflammatory in nature (Branzk et al., 2014; Goodridge et al., 2011; Lewis et al., 2012; Rosas et al., 2008). Nevertheless, our study shows that macrophages endeavor to seal the frustrated phagocytic compartment, in an effort to maximize their antimicrobial effect and minimize the release of inflammatory agents. To this end, they generate de novo a strikingly effective diffusion barrier by a process that involves activation of integrins that induce the formation of a thick F-actin cuff at the neck of the tubular phagosomes. The formation of actin-rich structures was reported previously during infection of macrophages with C. albicans (García-Rodas et al., 2011; Heinsbroek et al., 2009; Strijbis et al., 2013) and other rod/filament shaped microbes (Gerisch et al., 2009; Prashar et al., 2013), but neither their mechanism of assembly nor their functional significance were fully understood, which motivated our studies.
As expected, Dectin1 –the major phagocytic receptor for fungal β-glucan (Brown and Gordon, 2001; Brown et al., 2002; Taylor et al., 2007)– was present along the phagocytic cup, lining the internalized portion of the hyphae. Indeed, the RAW-Dectin1 cell line used for some of our studies was created to allow efficient internalization of fungal zymosan (Esteban et al., 2011), and has been used to study C. albicans-macrophage interactions (Strijbis et al., 2013). Dectin1 signaling leads to robust production of reactive oxygen species by the NADPH oxidase in response to fungal ligands (Brown et al., 2002; Goodridge et al., 2011; Underhill et al., 2005), accounting for our observation that superoxide is detected within the frustrated phagocytic cup. Remarkably, however, Dectin1 did not concentrate in the region of the membrane adjacent to the actin cuff where the diffusion barrier was established. Instead, integrin αMβ2 (CR3, or CD11b/CD18) was found to accumulate in this region.
Engagement of CR3 at the cuff is required for the formation of the underlying actin cuff, a process likely mediated by talin and vinculin, which were also found accumulated at the site. The entire assembly appears central to the establishment of the diffusion barrier, which is lost when blocking CR3 binding with the M1/70 antibody and also when latrunculin is used to disassemble actin filaments. CR3 is unique amongst integrins in that its α domain contains a lectin-like domain (LLD) capable of binding fungal β-glucan (Ross et al., 1985; Vetvicka et al., 1996). This LLD, located at the membrane-proximal C terminus (between residues 943–1047; Lu et al., 1998) is distinct from the traditional ligand-binding I domain. Importantly, the LLD can bind β-glucan in a Ca2+-independent manner while the integrin is in the inactive, bent conformation (Thornton et al., 1996). Binding of glucan has been shown to induce a semi-active conformation of the integrin that is predicted to facilitate outside-in signaling (O'Brien et al., 2012; Vetvicka et al., 1996). We found that Dectin1 did not have a direct role in actin cuff formation, being instead required for adhesion and the initiation of phagocytosis. Interestingly, when Dectin1 expression is low, the deposition of opsonins contained in serum –including complement and possibly also anti-Candida antibodies– suffice to engage CR3 and promote actin cuff formation directly activating this process, as predicted from earlier observations (Boxx et al., 2010; Kozel et al., 1987; Vetvicka et al., 1996). Dectin1 signaling can initiate inside-out activation of CR3 (Li et al., 2011). However, conventional Rap1-dependent inside-out signaling mediated by CalDAG-GEF1 was dispensable for actin cuff formation, as were divalent cations. Therefore, it is likely that CR3 binds fungal β-glucan in a manner that does not require inside-out activation by Dectin1.
We showed that mannan, a ligand for the mannose receptor, had no effect on actin cuff formation. The utilization of a curated set of C. albicans GRACE strains confirmed that mannan was dispensable and further excluded chitin and β(1,6)-glucan as ligands for cuff formation. Importantly, caspofungin inhibition of β(1,3)-glucan synthesis blocked formation of the cuff, in a dose dependent manner. These observations are in accord with involvement of the LLD of CR3, which was previously demonstrated to ligate β-glucan (Mueller et al., 2000; Thornton et al., 1996; Vetvicka et al., 1996). Clearly, β-glucan is also a ligand for Dectin 1 (Brown and Gordon, 2001; Brown et al., 2002; Palma et al., 2006), and initial engagement of this receptor is required for formation of actin cuffs around unopsonized hyphae. However, β(1,3)-glucan appears to play a distinct role in CR3-mediated actin cuff formation, as the effects of caspofungin could not be rescued by serum opsonization, with the caveat that caspofungin treatment may have affected complement deposition on C. albicans (Boxx et al., 2010; Kozel et al., 1987), although we regard this as unlikely because the fungal cell wall is rich in other polysaccharides and proteins that can serve to attach complement.
Our observations implicate clustered CR3 as an initiator of actin polymerization and a key constituent of the diffusion barrier. The signals mediating this effect include activation of Syk, which had been reported earlier (Strijbis et al., 2013), and also of Pyk2 and Fak. The latter related kinases were enriched at the cuff and dual inhibition by PF573228 blocked cuff formation. Along with activated Syk, Pyk2 and Fak have been shown to interact with β2 integrins, including CR3 (Duong and Rodan, 2000; Fernandez and Suchard, 1998; Han et al., 2010; Hildebrand et al., 1995; Kamen et al., 2011; Mócsai et al., 2002; Raab et al., 2017; Rubel et al., 2002; Wang et al., 2010; Yan and Novak, 1999). Pyk2, in particular, is required for paxillin and Vav1 activation during integrin engagement and CR3-mediated phagocytosis (Kamen et al., 2011). Paxillin was proposed to act as a scaffold, bridging integrin-initiated complexes with Rho-GTPases (Deakin and Turner, 2008), and Vav1, previously identified as important for the phagocytosis of C. albicans (Strijbis et al., 2013), also links β2 integrins to the activation of Cdc42, Rac1 and RhoA (Gakidis et al., 2004). These signaling events appear conserved during the frustrated phagocytosis of C. albicans hyphae, as we detected paxillin and active Rac1/Cdc42 at the actin cuff, and Vav1 was found to be enriched in actin cuff-like structures around C. albicans (Strijbis et al., 2013).
The activation of the Rho family GTPases is linked to both formin- and Arp2/3- mediated actin dynamics. The actin cuffs formed around C. albicans hyphae were singularly sensitive to SM1-FH2 –and therefore dependent on formin-mediated linear actin polymerization– and not to inhibitors of the Arp2/3 complex that promotes branched actin polymerization. Interestingly, Rac1 and Cdc42 interact with actin-nucleating formins of the mDia family (Lammers et al., 2008). Collectively, our findings support a model whereby CR3 initiates signaling through Syk and Pyk2/Fak, leading to activation of Vav1 and Rho GTPases, culminating in formin-dependent actin nucleation.
Our studies show that the integrin/actin cuff separates the open phagocytic cup from the plasma membrane, appearing to act as a boundary that segregates distinct and mobile membrane domains. There was a clear segregation of phosphoinositides between the plasma membrane (PtdIns(4,5)P2) and the open cup (PtdIns(3,4,5)P3 and PtdIns(3)P). In principle, such separation could stem from the differential and strategic localization of kinases and/or phosphatases in the two membranes and in the junctional complex. However, we also observed slowly convertible (LC3) or non-convertible lipid-anchored proteins (constitutively-active Rab7) and transmembrane proteins (LAMP1) to be retained in the phagocytic cup, unable to reach the surface membrane. FRAP studies confirmed that these molecules were freely mobile within the phagocytic cup, pointing to the integrin/actin cuff as the diffusion barrier.
While antibody-induced clustering of CR3 was sufficient to form an actin-independent diffusional barrier, actin was required to maintain the barrier function of the cuff during phagocytosis of hyphae. Actin can stabilize integrins in their active conformation (Kaizuka et al., 2007; Lavi et al., 2007; Lavi et al., 2012), and such stabilization is likely required to maintain the cuff during extended frustrated phagocytosis. Actin-dependent diffusional barriers have been invoked in other systems (Golebiewska et al., 2011; Nakada et al., 2003; Prashar et al., 2013), although the pickets that anchor the cytoskeletal fence and restrict the diffusion of membrane components had not been previously identified.
We also analyzed the functional consequences of the establishment of the cuff and diffusion barrier. By segregating the two domains, the barrier enabled the open phagocytic cup to undergo an atypical maturation, despite the fact that scission from the surface membrane never occurred. This enabled targeting and activation within the frustrated phagosome of the NADPH oxidase, which generated toxic superoxide in the immediate vicinity of the portion of the hypha that had been engulfed. Thus, a crucial means for C. albicans control during infection (Sasada and Johnston, 1980; Brothers et al., 2013) remains operational in the frustrated phagosomes.
The observation that the tubular phagosomes were rich in LAMP1, a prototypic lysosomal marker, suggests that lysosomal hydrolases must have been secreted also into the phagosomal lumen. These were, however, not well retained by the phagosomes because, despite the tight seal that separated the inner leaflet of the membrane, the junction separating the aqueous compartments (i.e. the lumen from the extracellular space) was not perfectly tight. While 70 kDa dextran was retained within the phagosome, 10 kDa dextran was not, resembling the findings in frustrated L. pneumophila phagosomes (Prashar et al., 2013) and indicating the establishment of a sieve that excluded molecules with a hydrodynamic radius greater than ≈ 6–8 nm (Nicholson and Tao, 1993). Cathepsin family members (radius ≈ 2.4 nm; Fazili and Qasim, 1986) and similarly-sized hydrolases would therefore eventually escape the lumen. Nevertheless, because a partial seal is formed, their rate of loss might be slowed, allowing hydrolases and other antimicrobial molecules to act on the partially internalized C. albicans hyphae before exiting the open cup. Fast-acting antimicrobial agents, like ROS, released in close proximity to the target would be expected to be at least partially effective. Consistent with this hypothesis, partially internalized C. albicans hyphae exhibited a reduced growth rate compared to external hyphae. Importantly, this growth restriction was abolished upon antibody blockade of CR3 and loss of actin cuff formation, presumably a result of increased leakage of phagosome contents. In this case, agents such as ROS, would not reach sufficient concentration to manifest the microbiostatic effect. However, leakage of phagosomal contents or ROS does not explain the failure of the frustrated phagosomes to kill the fungus, because C. albicans yeast and hyphae survive also within fully sealed phagosomes.
Based on the preceding considerations, we hypothesize that the integrin/actin cuff is generated and maintained by macrophages as a means to sustain antimicrobial functions in open tubular phagosomes formed around C. albicans hyphae and possibly other targets. It is tempting to speculate that the unique conditions established by the diffusion barrier might provide additional benefits to the phagocyte. In dendritic cells, decreased phagosomal proteolysis associated with reduced phagosome acidification protects antigens for enhanced presentation (Mantegazza et al., 2008; Rybicka et al., 2012; Savina et al., 2006). In addition, the frustrated yet maturing phagosome may enable activation of endomembrane Toll-like receptors (TLRs). TLR3 and TLR9 both localize to intracellular compartments and recognize C. albicans nucleic acids and chitin, respectively, contributing to a protective cytokine response to the fungus (Nahum et al., 2011; Wagener et al., 2014). Interestingly, Dectin1 can collaborate with plasmalemmal TLRs (TLR2 or TRL4) to enhance signaling and augment cytokine production (Ferwerda et al., 2008; Netea et al., 2006; Underhill, 2007) and a similar synergy may apply to endomembrane TLRs. Indeed, Dectin1 recognition is required for TLR9 localization to the C. albicans phagosome and TLR9-dependent gene expression (Khan et al., 2016). Thus, the unique structure described here may play an important role in the control of fungal infection and possibly also in the management of biofilms and other large targets by phagocytes.
Mammalian expression vectors were obtained from the following sources: Emerald-Dectin1 (plasmid #56291; Addgene, Cambridge, MA), PAK-PBD-YFP (Srinivasan et al., 2003), E-cadherin-GFP (plasmid #67937; Addgene), β-catenin-GFP(plasmid #16071; Addgene), Talin-GFP (Franco et al., 2004), AKT-PH-GFP (Marshall et al., 2001), PLCδ-PH-GFP (Botelho et al., 2000), PX-GFP (Kanai et al., 2001), LC3-GFP (Kabeya et al., 2000), Rab7-GFP (Bucci et al., 2000), Rab7(Q67L)-RFP (D’'Costa et al., 2015), Lamp1-GFP (Martinez et al., 2000), Lyn11-GFP (Teruel et al., 1999), cathepsin D-RFP (Yuseff et al., 2011), LifeAct-RFP or -GFP (Riedl et al., 2008), F-tractin-GFP (Belin et al., 2014), CD2-CD45-GFP (Cordoba et al., 2013).
Primary antibodies were purchased from the following vendors: HA (catalogue #MMS-101P; Covance, Princeton, NJ), pTyrosine (catalogue #05–321; EMD Millipore, Billerica, MA), pFAK-Y397 (catalogue #3283S; Cell Signaling, Berverly, MA), pPYK2-Y402 (catalogue #3291S; Cell Signaling), pSFK-Y418 (catalogue #44660G; Invitrogen, Carlsbad, CA), pSYK-Y525/526 (catalogue #2771S; Cell Signaling), Talin (catalogue #T3287; Sigma-Aldrich, St. Louis, MO), Vinculin (catalogue #MAB3574; EMD Millipore), HS1 (catalogue #4557S; Cell Signaling), LAMP1 (catalogue # 1D4B-s, Developmental Studies Hybridoma Bank, Iowa City, IA), actin (catalogue #A4700; Sigma-Aldrich), CD11b (catalogue #557394; BD Biosciences, Franklin Lakes, NJ), CD18 (catalogue #557437; BD Biosciences), rat IgG2B isotype control (catalogue #MAB0061; R and D systems, Minneapolis, MN), paxillin (catalogue #P13520; Transduction Laboratories, Lexington, KY), GAPDH (catalogue #MAB374; EMD Millipore), E-cadherin (catalogue #610181; BD Biosciences), β-catenin (catalogue #610153; BD Biosciences). Unconjugated and Alexa488, Cy3, Cy5, HRP-conjugated secondary antibodies against mouse, goat, rat, rabbit IgGs were obtained from Jackson ImmunoResearch Labs (West Grove, PA).
A list of all C. albicans strains tested is provided in Table 1. C. albicans strain SC5314 expressing BFP (Candida-BFP; Strijbis et al., 2013) was grown at 30°C in YPD (BD Biosciences). C. albicans cell wall mutant strains were obtained from the GRACE collection of tetracycline-repressible mutant strains (O'Meara et al., 2015; Roemer et al., 2003). Depletion of target gene expression was achieved by adding 0.5 μg mL−1 doxycycline (DOX) to the growth medium. To induce hyphae of C. albicans, overnight cultures were subcultured 1:1000 in RPMI-1640 medium and incubated at 37°C for 1–3 hr, as indicated in the text. In some cases, caspofungin (Sigma-Aldrich) was used to pharmacologically inhibit β(1,3)-glucan synthesis. C. albicans overnight cultures were subcultured into RPMI-1640 containing 10, 5, 2.5, 1.25 and 0 ng mL−1 caspofungin for 2 hr at 30°C. Cultures were then moved to 37°C for 1 hr to induce hyphae in the presence of caspofungin. To measure the effect of caspofungin on β(1,3)-glucan levels, C. albicans hyphae-infected wells were stained with 0.05% aniline blue (EVANS et al., 1984; Lee et al., 2016) overnight.
A. fumigatus strain AF293 (clinical isolate) was grown on YPD agar (Bioshop, Burlington, ON) plates at 30°C. Conidia were harvested in PBS containing 0.01% Tween-80. For experiments, resuspended conidia were diluted 1:10 in RPMI-1640 containing 0.01% Tween-80, and allowed to form hyphae at 30°C overnight. Hyphae were then washed twice with PBS 0.01% Tween-80, and diluted 1:10 or 1:100 into RPMI-1640 containing 0.01% Tween-80.
The RAW 264.7 cell line was obtained from and authenticated by the American Type Culture Collection (ATCC, Manassas, VA). The RAW-Dectin1-LPETG-3xHA cell line (RAW-Dectin1) was provided by Dr. Karin Strijbis and authenticated for Dectin1-HA expression and Dectin1-mediated phagocytic ability by flow cytometry (Esteban et al., 2011). Prior to experimentation, these cell lines were validated in our laboratory by assessing their morphology, phagocytic ability and expression of plasma membrane markers. RAW 264.7 and RAW-Dectin1 cells were grown in RPMI-1640 medium containing L-glutamine (MultiCell, Wisent, St. Bruno, QC) and 10% heat-inactivated fetal calf serum (FCS; MultiCell, Wisent), at 37°C under 5% CO2. The A431 cell line was obtained from and authenticated by the American Type Culture Collection (ATCC). Prior to experimentation, this cell line was revalidated by assessing its expression of plasma membrane markers, and responsiveness to epidermal growth factor (EGF). A431 cells were grown in DMEM medium containing L-glutamine (MultiCell, Wisent) and 10% heat-inactivated FCS, at 37°C under 5% CO2. All cell lines tested negative for mycoplasma contamination by DAPI staining.
Bone marrow-derived macrophages (BMDM) were obtained from the femoral bones of CALDAG-GEF1−/− (Bergmeier et al., 2007) or +/+ (C57BL/6) mice, and differentiated for 5–7 days in DMEM containing L-glutamine, 10% heat-inactivated FCS, 100 U mL−1 penicillin, 100 μg mL−1 streptomycin, 250 ng mL−1 amphotericin B (MultiCell, Wisent) and 10 ng mL−1 mM-CSF (PeproTech, Rocky Hill, NJ), at 37°C and 5% CO2. To obtain M2 human monocyte-derived macrophages, peripheral blood mononuclear cells were isolated from the blood of healthy donors by density-gradient separation with Lympholyte-H (Cedarlane, Burlington, ON). Human monocytes were then separated by adherence, and incubated in RPMI-1640 containing L-glutamine, 10% heat-inactivated FCS, 100 U mL−1 penicillin, 100 μg mL−1 streptomycin, 250 ng mL−1 amphotericin B and 25 ng mL−1 hM-CSF (PeproTech) for 7 days.
Mammalian cell lines or primary cells were seeded on 18 mm coverslips in 12-well plates at 2 × 105 cells mL−1. For infections with C. albicans, the medium was aspirated from the wells and replaced with 1 mL of C. albicans that had been induced to form hyphae. Plates were centrifuged for 1 min at 1500 rpm, then incubated with the following cell types at 37°C and 5% CO2 for phagocytosis to proceed: RAW-Dectin1: 1 hr; BMDM: 15 min; human M2 macrophages: 20 min; A431 cells: 3 hr. In some cases, 30 min prior to infection, C. albicans hyphae were opsonized in human serum to promote the deposition of complement, although deposition of donor-specific anti-Candida antibodies may have also occurred, further favoring phagocytosis.
Where indicated, cells were treated with either vehicle or 1 μM Latrunculin A (Sigma-Aldrich) after 1 hr Candida-BFP infection, or pretreated 30 min with 4 mM EDTA (Bioshop), followed by infection with Candida-BFP in the presence of EDTA. For inhibition of actin polymerization or kinases, after 30 min incubation with Candida-BFP, monolayers were treated with either vehicle, 50 μM CK-666 (Calbiochem, La Jolla, CA), 10 μM SMI-FH2 (Calbiochem), 10 μM PP2 (Calbiochem), 50 μM piceatannol (Sigma Aldrich), or 50 μM PF573228 (Tocris, Oakville, ON), for 30 min. Following infection, monolayers were washed three times with PBS and fixed with 4% paraformaldehyde (PFA). In some cases, wells were treated with various fluorescent reagents before or after phagocytosis, as described below.
For infections with A. fumigatus, RAW-Dectin1 cells were incubated with 1 mL diluted A. fumigatus hyphae, and plates centrifuged for 5 min at 1500 rpm. Plates were incubated for 2 hr at 37°C and 5% CO2 for phagocytosis to proceed. RAW-Dectin1 cells were pretreated 10 min and infected in the presence of 100 mM L-cysteine (Sigma-Aldrich) to prevent gliotoxin-mediated inhibition of phagocytosis (Schlam et al., 2016). Following infection, monolayers were washed three times with PBS and fixed with 4% paraformaldehyde (PFA). In some cases, wells were treated with various fluorescent reagents before or after phagocytosis, as described below.
After phagocytosis, external C.albicans were labeled for 20 min at room temperature using a solution of 5 μg mL−1 fluorescent conjugated concanavalin A (ThermoFisher Scientific, Waltham, MA). To stain actin filaments, cells were permeabilized 5 min with 0.1% Triton X-100 and incubated 30 min with a 1:1000 dilution of fluorescent phalloidin (Thermofisher Scientific) or acti-stain (Cytoskeleton, Inc., Denver, CO). In some cases, C. albicans and A. fumigatus were stained with 10 μg mL−1 calcofluor white (Fluorescent Brightener 28; Sigma-Aldrich).
For transient transfection, RAW-Dectin1 cells were plated on 18 mm glass coverslips at a concentration of 2 × 105 cells mL−1 16–24 hr prior to experiments. FuGENE HD (Promega, Madison, WI) transfection reagent was used according to the manufacturer’s instructions. RAW-Dectin1 cells were transfected at a 3:1 ratio using 1.5 μL FuGENE HD and 0.5 μg DNA per well, and used for experiments 16 hr after transfection.
In some cases, DNA transfections were performed using the Neon transfection system (Life Technologies, Carlsband, CA) according to the manufacturer’s protocol. RAW-Dectin1 cells were lifted, washed and resuspended to a concentration of 4 × 106 cells mL−1 and 100 μL of the suspension were mixed with 5 μg DNA. Electroporation was done using a single 20 ms pulse of 1750 V. Cells were then immediately transferred to RPMI-1640 containing L-glutamine and 10% heat-inactivated FCS, before seeding on coverslips at concentration of 2 × 105 cells mL−1. Cells were used for experiments 16 hr after electroporation.
After phagocytosis, fixation and concanavalin A staining (as indicated), monolayers were permeabilized in PBS containing 0.1% Triton X-100 for 5 min and blocked in PBS containing 5% skim milk and 0.1% Triton X-100 for 30 min at room temperature. Samples were incubated with primary antibodies for 30 min at room temperature. Primary antibody dilutions were: HA (1:1000), pTyrosine (1:100), pFAK-Y397 (1:100), pPYK2-Y402 (1:100), pSRC-Y418 (1:100), pSYK-Y525/526 (1:100), talin (1:500), vinculin (1:500), HS1 (1:250), LAMP1 (1:20), actin (1:100), E-cadherin (1:100), β-catenin (1:100), CD11b (1:100), CD18 (1:100), paxillin (1:100). After rinsing with PBS, samples were incubated 30 min at room temperature with Alexa488, Cy3 or Cy5-conjugated secondary antibodies at a 1:10,000 dilution. Where indicated, fluorescent phalloidin at a 1:1000 dilution was included with secondary antibodies. Samples were rinsed and viewed in PBS by confocal microscopy.
Cells were grown in six well plates at a concentration of 4 × 105 cells per well. After infections, wells were lysed in Laemmli buffer (Bio-Rad, Mississauga, ON). Samples were run on a 7% SDS-PAGE gel for separation and the gel was transferred to a polyvinylidene difluoride (PVDF) membrane. Membrane was blocked in PBS containing 5% skim milk and 0.05% Tween-20 for 30 min at room temperature, followed by primary antibody staining for 1 hr at room temperature, in blocking buffer. Primary antibodies dilutions: E-cadherin (1:10,000), β-catenin (1:1000), GAPDH (1:20,000; loading control). After washing membrane in PBS containing 0.05% Tween-20, samples were incubated 30 min at room temperature with HRP-conjugated secondary antibodies at a 1:3000 dilution. Blots were visualized using the ECL Prime Western Blot detection reagent (GE Healthcare, Mississauga, ON) on an Odyssey Fc (LI-COR, Lincoln, NE).
5 μg Rhodamine-PtdEth (L-α-phosphatidylethanolamine-N-(lissamine rhodamine B sulfonyl) ammonium salt; Avanti Polar Lipids, Inc., Alabaster, AL) was dried under N2 and resuspended in 10 μL methanol. After vortexing, 900 μL 3 mg mL−1 bovine serum albumin (BSA) was added. This was then diluted 1:1 in cold serum-free RPMI-1640 containing 25 mM HEPES (HPMI; MultiCell, Wisent). Following phagocytosis, the medium was aspirated and replaced with 500 μL of the prepared Rhodamine-PtdEth/HPMI, and incubated at 4°C for 10 min. The adherent cells were rinsed three times with cold HPMI, heated to 37°C 10 min and imaged live.
Following phagocytosis, cells were rinsed three times with cold HPMI. Cholera toxin subunit B, Alexa488 conjugate (Thermofisher Scientific) was added to a final concentration of 1 μg mL−1 and cells incubated at 4°C for 10 min. The adherent cells were rinsed three times with cold HPMI, heated to 37°C 10 min and viewed live.
Following phagocytosis, acidic intracellular compartments were stained using a 1:5000 dilution of the acidotropic LysoBrite Red dye (AAT Bioquest, Sunnyvale, CA) for 5 min at 37°C. Monolayers were rinsed three times with PBS, placed in HPMI, and imaged live.
RAW-Dectin1 cells were pulsed overnight with 20 μg mL−1 Alexa647-conjugated 10 kDa dextran or 25 μg mL−1 tetramethylrhodamine-conjugated 70 kDa dextran. Cells were washed and incubated with Candida-BFP, as described above, for 1 hr. Following phagocytosis, monolayers were rinsed three times with PBS, placed in HPMI, and viewed live.
This assay required PFA-inactivated Candida-BFP hyphae, as metabolically active C.albicans reduced nitroblue tetrazolium (NBT) to formazan, confounding the results. Candida-BFP hyphae were made as described above, followed by fixation in 8% PFA for 20 min. PFA-inactivated hyphae were rinsed and used to infect RAW-Dectin1 cells for 1 hr, in the presence of 0.5 mg mL−1 NBT (Sigma-Aldrich). Following infection, cells were washed three times with PBS and fixed with 4% PFA. Formazan precipitate, created in response to superoxide anion produced by phagosomal NADPH oxidase, was visualized by bright field microscopy.
To block CR3, adherent RAW-Dectin1 cells were incubated with 10 μg blocking antibody to CD11b (monoclonal M1/70) or rat IgG2B isotype control (MAB0061; R and D systems) for 30 min at room temperature. After warming to 37°C, cells were incubated with Candida-BFP hyphae as described above. Following phagocytosis, monolayers were rinsed and fixed in 3% PFA as mentioned above. After permeabilization and blocking, described above, blocking antibody was detected using a Alexa488-conjugated secondary antibody against rat IgG (1:10,000; Jackson ImmunoResearch Labs), and viewed by confocal microscopy.
For sugar blocking experiments, adherent RAW-Dectin1 cells were pretreated 30 min with 100 μg mL−1 mannan (Sigma Aldrich) or laminarin (Sigma Aldrich), followed by infection with C.albicans in the continued presence of each sugar. In the case of laminarin, Candida-BFP hyphae were allowed to adhere to RAW-Dectin1 cells 10 min, followed by the addition of 100 μg mL−1 laminarin for the remainder of the 1 hr infection.
RAW-Dectin1 cells transiently transfected with CD2-CD45-GFP were washed with PBS and cooled to 10°C in 1X HBSS (MultiCell, Wisent). To crosslink surface CR3, cells were sequentially incubated with: (1) 5 μg mL−1 anti-CD18 primary (monoclonal M18/2); (2) goat anti-rat secondary and (3) donkey anti-goat tertiary antibodies, for 30 min each at 4°C, in PBS containing Ca2+, Mg2+ and 0.1% glucose. Labeled cells were then incubated 30 min at 37°C in the presence of 30 μM DYNGO 4a, to patch crosslinked CR3 without its internalization. Following this treatment, cells were treated 10 min with or without 1 μM latrunculin A, in the presence of 1 μM N-ethylmaleimide to prevent internalization of crosslinked CR3 following actin depolymerization. Monolayers were fixed in 3% PFA, and extracellular CR3 was labeled using a fluorescent anti-goat antibody. Plasma membrane clusters of CR3 were analyzed in Volocity software for exclusion of CD2-CD45-GFP. Exclusion was calculated from confocal slices as the ratio of the GFP intensity per pixel in the CR3-positive patches to the average intensity in the plasma membrane.
RAW-Dectin1 cells were incubated with Candida-BFP hyphae for 10 min to allow adherence, then fixed in 4% PFA after 0, 10, 20, 30, 40, 50 or 60 min. Following phagocytosis, the cells were fixed in 3% PFA as above, external Candida-BFP were labeled with fluorescent concanavalin A, and permeabilized and stained with fluorescent phalloidin, as described above. Samples were imaged by confocal microscopy, and the length of individual Candida-BFP hypha measured in μm. Hyphal extension rate was calculated by linear regression analysis in GraphPad Prism software (GraphPad Software, Inc., La Jolla, CA).
Confocal images were acquired using a Yokogawa CSU10 spinning disk system (Quorum Technologies Inc., Guelph, ON) or a Leica SP8 laser scanning system (Leica). Images were acquired using a 63x/1.4 NA oil objectives or 25x/0.8 NA water objective (ZEISS, Germany), as indicated, with an additional 1.5x magnifying lens. For live experiments, cells were maintained at 37°C using an environmental chamber (Live Cell Instruments, Korea). Routine analyses and colocalizations were done using Volocity software (Perkin Elmer, Woodbridge, ON). 3D data visualization was done using Imaris (Bitplane, Concord, MA) software.
For colocalization analyses, Volocity software was used to calculate positive product of the differences of the mean (Li et al., 2004) channels, which were then overlayed on merged images for visualization.
For fluorescent intensity calculations, background subtracted intensities per unit area for expressed fluorescent protein constructs or endogenous proteins (immunofluorescence) were measured in Volocity software. Ratios were calculated comparing relative intensities in the actin cuff compared to phagocytic cup, or phagocytic cup compared to membrane, as indicated in the text.
FRAP experiments of GFP or RFP-tagged proteins, transiently expressed in RAW-Dectin1 cells, were conducted on an A1R point-scanning confocal system (Nikon Instruments, Japan). For FRAP, Candida-BFP hyphae-infected cells were imaged in HPMI at 37°C. Images were acquired using a 60x/1.4 NA oil objective (Nikon), 1.2-AU pinhole, resonant scanning mode, and 16x line averaging. For a complete 2 min FRAP acquisition at 1.9 fps, after 5 s of initial imaging, a region of interest 3 μm in diameter was bleached for 1.06 s using the 405 laser at 100% power, followed by imaging for fluorescence recovery. Images were exported and analyzed for fluorescence intensity using Volocity software.
After background subtraction, fluorescence intensity units were normalized (see Figure 6 legend) using Microsoft Excel software, and transformed to a 0–1 scale, to correct for differences in bleaching depth and allow for comparison of up to 30 individual FRAP curves per condition. Graphpad Prism software was used to fit the FRAP curves to a single exponential, plotted as fractional recovery over time.
Candida-BFP hyphae-infected RAW-Dectin1 cells were washed with cold PBS, and fixed with 2% glutaraldehyde in 0.1 M sodium cacodylate buffer, pH 7.3. For improved specimen preparation, samples were then subjected to a zymolyase digestion protocol (Bauer et al., 2001) designed to weaken the fungal cell wall and allow for adequate structural preservation. Samples were then post-fixed in 1% osmium tetroxide in 0.1 M sodium cacodylate buffer, pH 7.3, dehydrated in a graded ethanol series followed by propylene oxide, and embedded in Quetol-Spurr resin. Ninety nm sections were cut on a Leica Ultracut ultramicrotome and stained with uranyl acetate and lead citrate. Samples were imaged on a FEI Tecnai 20 transmission electron microscope, equipped with an AMT 16000 digital camera.
Fitness and virulence costs of Candida albicans FKS1 hot spot mutations associated with echinocandin resistanceThe Journal of Infectious Diseases 204:626–635.https://doi.org/10.1093/infdis/jir351
Characterization of rac and cdc42 activation in chemoattractant-stimulated human neutrophils using a novel assay for active GTPasesJournal of Biological Chemistry 274:13198–13204.https://doi.org/10.1074/jbc.274.19.13198
Mice lacking the signaling molecule CalDAG-GEFI represent a model for leukocyte adhesion deficiency type IIIJournal of Clinical Investigation 117:1699–1707.https://doi.org/10.1172/JCI30575
Class I and class III phosphoinositide 3-kinases are required for actin polymerization that propels phagosomesThe Journal of Cell Biology 191:999–1012.https://doi.org/10.1083/jcb.201004005
Localized biphasic changes in phosphatidylinositol-4,5-bisphosphate at sites of phagocytosisThe Journal of Cell Biology 151:1353–1368.https://doi.org/10.1083/jcb.151.7.1353
Influence of mannan and glucan on complement activation and C3 binding by Candida albicansInfection and Immunity 78:1250–1259.https://doi.org/10.1128/IAI.00744-09
Thinking outside the cell: how cadherins drive adhesionTrends in Cell Biology 22:299–310.https://doi.org/10.1016/j.tcb.2012.03.004
Dectin-1 mediates the biological effects of beta-glucansThe Journal of Experimental Medicine 197:1119–1124.https://doi.org/10.1084/jem.20021890
Dectin-1 is a major beta-glucan receptor on macrophagesThe Journal of Experimental Medicine 196:407–412.https://doi.org/10.1084/jem.20020470
Identification of the FKS1 gene of Candida albicans as the essential target of 1,3-beta-D-glucan synthase inhibitorsAntimicrobial Agents and Chemotherapy 41:2471–2479.
PYK2 is an adhesion kinase in macrophages, localized in podosomes and activated by beta(2)-integrin ligationCell Motility and the Cytoskeleton 47:174–188.https://doi.org/10.1002/1097-0169(200011)47:3<174::AID-CM2>3.0.CO;2-N
Integrin-dependent phagocytosis: spreading from microadhesion to new conceptsJournal of Cell Science 121:1773–1783.https://doi.org/10.1242/jcs.018036
Protein kinase Cδ is a critical component of Dectin-1 signaling in primary human monocytesJournal of Leukocyte Biology 90:599–611.https://doi.org/10.1189/jlb.0610376
Purification and some properties of buffalo liver cathepsin BThe Journal of Biochemistry 100:293–299.https://doi.org/10.1093/oxfordjournals.jbchem.a121715
Syk activation is required for spreading and H2O2 release in adherent human neutrophilsThe Journal of Immunology : Official Journal of the American Association of Immunologists 160:5154–5162.
Calpain-mediated proteolysis of talin regulates adhesion dynamicsNature Cell Biology 6:977–983.https://doi.org/10.1038/ncb1175
Na+/H+ exchange activity during phagocytosis in human neutrophils: role of Fcgamma receptors and tyrosine kinasesThe Journal of Cell Biology 132:1037–1052.https://doi.org/10.1083/jcb.132.6.1037
Vav GEFs are required for beta2 integrin-dependent functions of neutrophilsThe Journal of Cell Biology 166:273–282.https://doi.org/10.1083/jcb.200404166
Self-organizing actin waves as planar phagocytic cup structuresCell Adhesion & Migration 3:373–382.https://doi.org/10.4161/cam.3.4.9708
Candida albicans morphogenesis and host defence: discriminating invasion from colonizationNature Reviews. Microbiology 10:112–122.https://doi.org/10.1038/nrmicro2711
C-type lectin receptors orchestrate antifungal immunityNature Immunology 13:817–822.https://doi.org/10.1038/ni.2369
Actin and phosphoinositide recruitment to fully formed Candida albicans phagosomes in mouse macrophagesJournal of Innate Immunity 1:244–253.https://doi.org/10.1159/000173694
The insider's guide to leukocyte integrin signalling and functionNature Reviews Immunology 11:416–426.https://doi.org/10.1038/nri2986
Pyk2 is required for neutrophil degranulation and host defense responses to bacterial infectionThe Journal of Immunology 186:1656–1665.https://doi.org/10.4049/jimmunol.1002093
The PX domains of p47phox and p40phox bind to lipid products of PI(3)KNature Cell Biology 3:675–678.https://doi.org/10.1038/35083070
Roles of autophagy and autophagy-related proteins in antifungal immunityFrontiers in Immunology 7:47.https://doi.org/10.3389/fimmu.2016.00047
Dectin-1 Controls TLR9 Trafficking to Phagosomes Containing β-1,3 GlucanThe Journal of Immunology 196:2249–2261.https://doi.org/10.4049/jimmunol.1401545
Specificity of interactions between mDia isoforms and Rho proteinsJournal of Biological Chemistry 283:35236–35246.https://doi.org/10.1074/jbc.M805634200
Phosphoinositides in phagocytosis and macropinocytosisBiochimica Et Biophysica Acta (BBA) - Molecular and Cell Biology of Lipids 1851:805–823.https://doi.org/10.1016/j.bbalip.2014.09.005
Dectin-1 activation controls maturation of β-1,3-glucan-containing phagosomesJournal of Biological Chemistry 288:16043–16054.https://doi.org/10.1074/jbc.M113.473223
Synaptotagmin VII regulates Ca(2+)-dependent exocytosis of lysosomes in fibroblastsThe Journal of Cell Biology 148:1141–1150.https://doi.org/10.1083/jcb.148.6.1141
The L412F variant of Toll-like receptor 3 (TLR3) is associated with cutaneous candidiasis, increased susceptibility to cytomegalovirus, and autoimmunityJournal of Allergy and Clinical Immunology 127:528–531.https://doi.org/10.1016/j.jaci.2010.09.031
pH of endophagosomes controls association of their membranes with Vps34 and PtdIns(3)P levelsThe Journal of Cell Biology 217:329–346.https://doi.org/10.1083/jcb.201702179
An integrated model of the recognition of Candida albicans by the innate immune systemNature Reviews Microbiology 6:67–78.https://doi.org/10.1038/nrmicro1815
Immune sensing of Candida albicans requires cooperative recognition of mannans and glucans by lectin and Toll-like receptorsJournal of Clinical Investigation 116:1642–1650.https://doi.org/10.1172/JCI27114
The role of toll-like receptor (TLR) 2 and TLR4 in the host defense against disseminated candidiasisThe Journal of Infectious Diseases 185:1483–1489.https://doi.org/10.1086/340511
Lectin site ligation of CR3 induces conformational changes and signalingJournal of Biological Chemistry 287:3337–3348.https://doi.org/10.1074/jbc.M111.298307
Global analysis of fungal morphology exposes mechanisms of host cell escapeNature Communications 6:6741.https://doi.org/10.1038/ncomms7741
The tyrosine kinase Pyk2 contributes to complement-mediated phagocytosis in murine macrophagesJournal of Innate Immunity 8:437–451.https://doi.org/10.1159/000442944
Patching and capping of LFA-1 molecules on human lymphocytesHistochemistry 98:253–258.https://doi.org/10.1007/BF00271039
Filamentous morphology of bacteria delays the timing of phagosome morphogenesis in macrophagesThe Journal of Cell Biology 203:1081–1097.https://doi.org/10.1083/jcb.201304095
Membrane complement receptor type three (CR3) has lectin-like properties analogous to bovine conglutinin as functions as a receptor for zymosan and rabbit erythrocytes as well as a receptor for iC3bThe Journal of Immunology : Official Journal of the American Association of Immunologists 134:3307–3315.
Role of the lectin domain of Mac-1/CR3 (CD11b/CD18) in regulating intercellular adhesionImmunologic Research 25:219–228.https://doi.org/10.1385/IR:25:3:219
The final steps of integrin activation: the end gameNature Reviews Molecular Cell Biology 11:288–300.https://doi.org/10.1038/nrm2871
Rac and Cdc42 play distinct roles in regulating PI(3,4,5)P3 and polarity during neutrophil chemotaxisThe Journal of Cell Biology 160:375–385.https://doi.org/10.1083/jcb.200208179
A versatile microporation technique for the transfection of cultured CNS neuronsJournal of Neuroscience Methods 93:37–48.https://doi.org/10.1016/S0165-0270(99)00112-0
Staphylococcus aureus biofilms prevent macrophage phagocytosis and attenuate inflammation in vivoThe Journal of Immunology 186:6585–6596.https://doi.org/10.4049/jimmunol.1002794
Collaboration between the innate immune receptors dectin-1, TLRs, and NodsImmunological Reviews 219:75–87.https://doi.org/10.1111/j.1600-065X.2007.00548.x
Soluble beta-glucan polysaccharide binding to the lectin site of neutrophil or natural killer cell complement receptor type 3 (CD11b/CD18) generates a primed state of the receptor capable of mediating cytotoxicity of iC3b-opsonized target cellsJournal of Clinical Investigation 98:50–61.https://doi.org/10.1172/JCI118777
Activated dectin-1 localizes to lipid raft microdomains for signaling and activation of phagocytosis and cytokine production in dendritic cellsJournal of Biological Chemistry 284:22005–22011.https://doi.org/10.1074/jbc.M109.009076
Dominique Soldati-FavreReviewing Editor; University of Geneva, Switzerland
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]
Thank you for submitting your work entitled "An integrin-based diffusion barrier separates membrane domains enabling formation of microbicidal frustrated phagosomes" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Neil Gow (Reviewer #1).
Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.
As you will see from their individual comments below, the reviewers concur that this study reports some advances the field of host-fungus interactions and add supporting data on role for integrin in the establishment or maintenance of the actin cuff. The manuscript is clearly written, and the work is well executed and clearly illustrated with beautiful imaging data.
However, the originality of the manuscript is in part compromised because some of the concepts were preceded by a study from Dr. Terebiznik's lab. While the current work needs to be more cross-referral to this, there is still novelty in the current submission leaning on the identification of Mac-1-dependent diffusion barriers formed between the hyphae cell wall and the plasma membrane of the macrophage. The experimental data supporting this model requires important revisions in order for it to be convincing (more quantitative data). Furthermore, data in support of the microbicidal properties of these frustrated phagosomes is missing. The roles of Dectin-1 and Integrin receptors in the formation of actin cuffs and barriers should investigated in more detail whereas it is optional to expand the study into inflammatory response.
While we are rejecting the paper as a result of the need for a substantial revision (longer than 2 months), we are broadly supportive of this manuscript and would consider a newly submitted form of this paper that would be treated as a revised manuscript.
This paper addresses a well appreciated but univestigated conundrum. How do immune phagocytes interact with target cells that are much larger than they can envelop? The authors show that macrophages that are challenged with elongated hyphal cells of the fungus Candida albicans wrap around the end of a hypha and form an actin cuff that partially seals the phagocytic cup at the mouth of the tubular phagosome. This structure retains some of the ability to establish an antimicrobial environment within the unsealed tubular phagosome – for example allowing NAPPH oxidase activation, but not phagosome acidification. The actin cuff is shown to represent a diffusion barrier that segregates phosphoinositides between the phagocytes cell membrane and the open cup. Some elegant experiments investigate how this is segregation is achieved and regulated. Overall this is an interesting, novel and well executed study that clearly advances the field of host-fungus interactions.
I have a number of questions, suggestions and presentational issues.
1) The fungus studied is Candida albicans but is most commonly referred to as Candida. It would not be simply pedantic to request that it be referred to as C. albicans throughout since most Candida species cannot form elongated hyphae.
2) It would be interesting to know if similar actin cuff seals are formed when phagocytes encounter hyphae of Aspergillus fumigatus (interesting because this fungus has quite marked differences in cell wall composition).
3) The hyphae used are not very long (1 h or 2 h hyphae were used) and the macrophages shown seem to be invaginating one end. Macrophages will also wrap around the trunk of a very large hypha. Are actin cuffs formed here- and what is their architecture? It would be useful to include some images of longer hyphae of 4-6 hours incubation in RPMI.
4) Various C. albicans mutants are available with alterations in the cell wall that would help determine what surface components of the fungus are important for inducing the cuffs.
5) In many of the figures it is not possible to clearly evaluate the shape and position of the phagocyte relative to the fungus. It would be very helpful to show a DIC or phase image of the same interacting cells in order to really see the orientation of the hypha and phagocyte. I have no doubt that the descriptions in the text are accurate, but I have to somewhat take it on trust since I cannot see the outline of the two interacting cells clearly in all cases. (For example see Figure 3B,C; Figure 4B,E,F; Figure 5B,C,F; Figure 6E; Figure 8A,D and others).
6) The authors show that the CTL Dectin-1 is not present at the actin cuff, but phosphotyrosine accumulated at this site suggesting that some other pattern recognition receptor may drive actin assembly. It would be useful to verify that glucan phosphate or laminarin did not block actin cuff formation.
7) Since the mannose receptor is likely to engage with hypha mannan, (which is located in the outer cell wall) it would be interesting to know if this is located at this site.
8) I'm not sure that the TEM shown in Figure 1C can be seen as evidence of showing the actin cuff unless the actin is perhaps stained using colloidal gold.
9) My88-/- cells would be useful in assessing the possible involvement of TLRs.
10) The actin cuffs appear to be elongated in some images and much shorter or almost like spots in others. Do they retain their shape over time or contract to a spot?
In this manuscript M Maxson and collaborators have characterized the phagocytosis of Candida albicans hyphae, which for long hyphae, cannot progress beyond the stage of phagocytic cup, leading to the formation of a frustrated phagosome. The manuscript is clearly written, and the findings illustrated with beautiful imaging data. The authors propose that actin cuffs in the frustrated phagosomes, which have been described in previous reports on the phagocytosis in Candida albicans (Bain et al., 2014 and Lewis et al., 2012), may act as diffusion barriers and contribute to membrane fences that allow the accumulation of ROS in the frustrated phagosomes. They also propose that these actin cuffs are the product of the atypical activation of MAC1receptors in the macrophages by ligands expressed on the cell wall of Candida and that MAC1 may restrict the free diffusion of lipids and membrane proteins along the membrane. The proposed model is sound, and it may explain possible microbicidal actions phagocytes may attempt when pathogens cannot be completely engulfed. However, there are many important points in the manuscript that require clarification and revision to better support the authors' model.
The authors claim that the frustrated phagosomes are microbicidal. Yet, there is no data on fungal survival to support this conclusion. Are the macrophage-trapped hyphae growing longer or dying as time passes?
The contribution of Candida hyphae to the formation of the frustrated phagosomes and actin cuffs must be assessed. Changes in Candida cell wall composition have been previously reported to affect macrophage actin dynamics (Bain et al., 2014). However, the authors are not considering or discarding a role for the yeast in the proposed model.
For the imaging data, even though the regions of interest are nicely indicated with brackets, since different markers don't always localize to same regions along the frustrated phagosome, having additional panels with channels merged would make it easier to visualize what the authors are illustrating.
In subsection “Signals driving actin cuff formation”, referring to Figure 2AB, the authors indicate, a lack of Dectin-1 at the actin cuffs. However Emerald-Dectin-1 can be seen in the cuff in Figure 2B, which is in good agreement with Strijbis et al., (2013) that also showed that Dectin-1 is enriched in the cuff region of the phagocytic cup, as well as in areas of enrichment outside the cuff region. To support their claim, the authors must present quantitative data on Dectin-1 distribution along the frustrated phagosome. Also, it is not indicated, but the imaging shown in Figure 2, and other figures, seems to be a single confocal z-plane. It may also be appropriate to include extended focus projections, considering the size of the hyphae, which could be included as supplementary files.
Clarifying the distribution of Dectin1 in the frustrated phagosomes, is critical because it affects the interpretation of data in Figure 2D and other results in the manuscript.
Figure 2D showing PAK-PBD and tyrosine phosphorylation in the actin cuffs are likely expected results since Strijbis et al., 2013, have shown that the GEF for CDC42 and RAC1, VAV1, as well as the tyrosine kinases Syk and BTK, are all recruited to cuffs. Quantitative data for the recruitment of these molecules to the frustrated phagosomes and data proving that their activities are involved in the formation or sustaining the cuffs will certainly help the proposed model.
Figure 3D shows that on average, for a given field, 6 hyphae are observed in frustrated phagosomes with actin cuffs and this is consistent throughout the manuscript. However, it is not clear how many individual events were analyzed. How frequently are these frustrated phagosomes observed? Do all hyphae engaged by macrophages have these structures associated with them?
Figure 4. For accumulation of receptors provide quantitative data. According to the authors, MAC1 activation signals for the formation of the actin cuffs. However, in Figure 4A, CD11b did not localize to the thicker section of the cuffs. Please clarify and indicate the frequency of this phenotype.
Figure 4G shows a shorter cup in the presence of CD11b blocking antibodies. The treatment could cause an inhibition of phagocytosis. Authors must demonstrate that phagocytosis is not affected by the treatment.
Figure 4G/H In cells treated with the M1/70 antibody, actin is still present in the frustrated phagosome. Is this a cuff? In fact, Concavalin A labeling is limited to the phalloidin boundary. This suggests the presence of a cuff/ barrier, even after CD11b treatment.
Is dectin1 distribution along the frustrated phagosome affected after CD11b is blocked? Blocking CD11b in dectin1 knockout cells will help address whether dectin1 contributes to this barrier.
Figure 5, subsection “Phospholipid segregation between the plasma membrane and the cuff-delimited phagosomal cup” states "Astonishingly, while PtdIns(4,5)P2 was present as expected in the surface membrane facing the extracellular milieu, it was undetectable in the invaginated section that constituted the frustrated phagosome (Figure 5A)". This and other characteristics of the cuff and the frustrated phagosomes reported in Figure 5 and Figure 8 are similar to those reported by Prashar et al., (2013). This must be acknowledged in the manuscript. It must be acknowledged that PtdIns(3,4,5)P3 distribution in the frustrated phagosome has been previously described by Strijbis et al. In Figure 5D, how do the authors explain the presence of LC3 in the actin cuff?
Subsection “The actin cuff forms a diffusional barrier to the movement of proteins and lipids” "The sharp boundary between the PtdIns(4,5)P2-rich surface membrane and the tubular membrane endowed with PtdIns(3,4,5)P3 and PtdIns(Bauer et al., 2001)P coincided with the location of the actin cuff, suggesting that the latter may function as a diffusion barrier".
Subsection “The actin cuff forms a diffusional barrier to the movement of proteins and lipids” Similarly, both wild-type Rab7 (Figure 5E) and constitutively-active Rab7 (not illustrated) are confined to the frustrated phagosomal tube, as was LAMP1 (Figure 5F)……..we considered it more likely that restricted diffusion accounted for the observations"
Differently to what the authors describe in the paragraph from above, the phagosomal markers are penetrating the barrier as in the case of Figure 5B and E and surpassing the cuffs in Figure 5C, E and F. As indicated before for other results the authors must present quantitative data on the distribution of the markers and extended focus projections to support the proposed model.
Figure 7 strongly supports authors' claim However, if latranculin A treatment allowed Lamp-1 and PtdIns(4,5)P2 to migrate to previously excluded zones, why concanavalin A is not following and labels the full extension of the hyphae? That will be the expected if the diffusion barriers in the cuffs are dismantled. Quantitative data is missing, and full cell imaging would be preferable to better illustrate the authors' claim.
Figure 8 The characteristics of the frustrated phagosome are remarkably similar to long cups described by Prasher et al., 2013 and Strijbis et al., 2003. This must be cited in the text. The accumulation of NBT in Figure 8E is puzzling since small molecules can diffuse out across the actin cuff barrier. The authors must address the possibility that the NBT is accumulating inside the hyphae. NBT can accumulate in fungi Camile et al., 2008. Thus, authors must prove a link between ROS and microbicidal conditions in the long open phagosomes.
The manuscript by Maxson et al., describes and analyzes partially sealed phagocytic cups in macrophages which form around Candida albicans hyphae. It demonstrates that an actin-rich cuff at the distal margin of the cup supports a lateral segregation between the components of the plasma membrane and the contiguous inner leaflet of the phagocytic cup. This lateral segregation excludes the PI(4,5)P2 from the cup and confines the PI(Bauer et al., 2001)P and PIP3 to the cup. The actin cuff contains the integrin CR3 and associated integrin signaling molecules. The barrier effectively limits the movement of inner leaflet probe molecules into or out of the cup, without significantly diminishing diffusion within the plasma membrane or the cup. The actin cuff can be disrupted by the actin depolymerizing drug latrunculin A. The barrier limits escape by diffusion of large macromolecules delivered into the lumen of the cup, but not the diffusive loss of smaller molecules, including dextrans and protons (pH). Nonetheless, reactive oxygen species can be delivered into the unclosed cups. The experimental work is carefully done and the morphological evidence in support of the claims is beautiful. However, much if not most of the conclusions reported here were first described in a large study by Prashar et al., (2013), using an analogous experimental model. Examining the interactions between macrophages and filamentous bacteria, that study demonstrated (Astarie-Dequeker et al., 1999) the actin cuff (called a "jacket") that segregates two domains of contiguous membrane (plasma membrane and phagocytic cup), (Bain et al., 2014) the effect of actin depolymerization on the maintenance of that segregation, (Bauer et al., 2001) the nature of the diffusion barrier with respect to extracellular probes and (Belin et al., 2014) the retention of diffusible molecules in the cup lumen (different sizes of dextrans, pH, hydrolytic enzymes). Moreover, unlike the present manuscript which shows that ROS can be generated in the unclosed cups, the previous paper measured the effect of partial phagocytosis on microbicidal activities against filamentous bacteria (albeit somewhat incompletely). The present manuscript but does not measure microbicidal activity against C. albicans.
Assuming this manuscript can be rewritten to acknowledge appropriately the demonstrated precedents and concepts of that earlier work, the present study does add some interesting new data which supports a role for integrin in the establishment or maintenance of the actin cuff. The role of integrin in the maintenance of the barrier remains indirect, however. Perhaps the actin ring is the main ingredient of the barrier, and anything which organizes actin into a ring can support the formation of diffusion barriers such as described here, and elsewhere for analogous structures (Golebiewska et al., 2011; Welliver et al., 2011. To better define the nature of the diffusion barrier, the diffusion measurements (e.g. Figure 6) or probe localization studies (e.g. Figure 5) described in the manuscript should be extended to analyze the roles of actin (latrunculin A) and integrin (M1/70) to barrier maintenance. This could provide a mechanistic underpinning to the diffusion measurements.
[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]
Thank you for resubmitting your work entitled "Integrin-based diffusion barrier separates membrane domains enabling formation of microbiostatic frustrated phagosomes" for further consideration at eLife. Your revised article has been favorably evaluated by Ivan Dikic (Senior editor), a Reviewing editor, and three reviewers.
This revised manuscript contains extensive additional experimental work and text that address concerns noted in earlier reviews. The new work is thorough and adequately supportive of the conclusions. The manuscript now establishes a diffusion barrier in the inner leaflet of incompletely closed phagocytic cup membranes, comprised of CR3 integrins and maintained by an actin-rich cuff. Further, it identifies signaling molecules that contribute to cuff formation, and establishes the incompletely closed cups as microbiostatic. Related studies are adequately cited. The discussion provides a thoughtful analysis of the biology and its implications. Yet there are some minor remaining issues that need to be addressed in the text before acceptance, as outlined below:
The authors addressed all the critiques from the reviewers.
They reported additional results on the phagocytosis of C. albicans hyphae, improved the quality of the imaging, and provided numerical and statistical data to support their most relevant experiments.
However, below there are several points that we believe could be discussed in the manuscript:
1) In the abstract it is stated “[…]in response to non-canonical activation of integrins by fungal β(1,3)-glucans." We suggest using the singular "glycan". There is more than one form of fungal β(1,3)- glucan, but this is not relevant to this article.
2) B-glucans are involved in the activation of complement and the deposition of C3b fragments on the surface of C. albicans. (Boxx et al., 2010., Kozel et al., 1987, Vetvicka et al., 1996). What could be the contribution of C3b opsonization on the formation of the actin cuffs and the properties of the membrane and luminal diffusion barriers?
3) Previous reports on the phagocytosis of C. albicans showed the formation of PIPs in phagosomes containing this yeast (Heinsbroek, 2009). This was attributed to pathogenic mechanisms. On the other hand, and coinciding with the authors' interpretation, Naufer et al., 2018 recently reported that PI(Bauer et al., 2001)P co-exists with phagolysosomal markers in the phagocytic cup of heat-killed filamentous bacteria. Maybe the authors could mention this in their Discussion section.
4) Please clarify: In Figure 4, human serum was used to complement opsonize hyphae. How did the authors account for the presence of antibodies against C. albincans in the serum? Antibodies could be expected as this yeast is part of the human microbiome
5) Caspofungin treatment could prevent the deposition of complement in hyphae, and its recognition by antibodies and Dectin-1. This could explain the results in Figure 4—figure supplement 1.
6) The authors showed that ROS production in the phagocytic cup retarded the elongation of the hyphae. This was alleviated by blocking CR3 with monoclonal antibodies to impede the formation of cuffs, a procedure that favours a ROS leaching. Since ROS are generally considered microbicidal, perhaps, ROS failing to reach lethal concentrations in the open cup could be the cause of the effect reported by the authors. Therefore, assuming a microbiostatic mechanism is probably incorrect.https://doi.org/10.7554/eLife.34798.041
- Sergio Grinstein
- Michelle E Maxson
- Teresa R O'Meara
- Xenia Naj
- Leah E Cowen
- Leah E Cowen
- Leah E Cowen
- Leah E Cowen
- Leah E Cowen
- Leah E Cowen
- Leah E Cowen
- Leah E Cowen
- Leah E Cowen
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We sincerely thank Dr. Karin Strijbis (Utrecht University, The Netherlands), for providing RAW-Dectin1-LPETG-3xHA cells and C. albicans strain SC5314 expressing BFP, and Dr. Wolfgang Bergmeier (University of North Carolina), for providing tissue from CalDAG-GEF1 knockout mice. We thank Merck and Genome Canada for making the C. albicans GRACE collections available. MEM was the recipient of a Heart and Stroke Pfizer Research Fellowship. TRO is supported by a National Institutes of Health (NIH) Ruth L Kirschstein National Research Service Award (NRSA, AI115947-01) from the NIAID. LEC is supported by Canadian Institutes of Health Research Operating Grants (PJT-153403, PJT-148548, MOP-86452, and MOP-119520) and Foundation Grant (FDN-154288), the Natural Sciences and Engineering Council (NSERC) of Canada Discovery Grants (06261 and 462167), an NSERC EWR Steacie Memorial Fellowship (477598), and a National Institutes of Health NIAID R01 (1R01AI127375-01). SG is supported by Canadian Institutes of Health Research grant FDN–143202.
- Dominique Soldati-Favre, Reviewing Editor, University of Geneva, Switzerland
© 2018, Maxson et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.