Research ArticleFC RECEPTORS

Mechanisms of inside-out signaling of the high-affinity IgG receptor FcγRI

See allHide authors and affiliations

Science Signaling  24 Jul 2018:
Vol. 11, Issue 540, eaaq0891
DOI: 10.1126/scisignal.aaq0891

Inside-out Fc receptor activation

Important for the activation of integrin adhesion molecules, “inside-out” activation is triggered by cytoplasmic signals that increase the affinity of the integrin for its ligand. Using super-resolution microscopy, Brandsma et al. found that cytokine stimulation similarly augments Fc receptor avidity by increasing the size of FcγRI clusters on the surface of monocytes. Pharmacological inhibitors of actin rearrangement and the phosphatase PP1 reduced receptor clustering. Cytokine-driven FcγRI clustering also correlated with the increased FcγR-dependent cytolytic activity of human neutrophils. Together, these findings suggest that FcR clustering stimulated by cytokines drives the inside-out activation of FcγRI.

Abstract

Fc receptors (FcRs) are an important bridge between the innate and adaptive immune system. Fc gamma receptor I (FcγRI; CD64), the high-affinity receptor for immunoglobulin G (IgG), plays roles in inflammation, autoimmune responses, and immunotherapy. Stimulation of myeloid cells with cytokines, such as tumor necrosis factor–α ( TNFα) and interferon-γ ( IFNγ), increases the binding of FcγRI to immune complexes (ICs), such as antibody-opsonized pathogens or tumor cells, through a process known as “inside-out” signaling. Using super-resolution imaging, we found that stimulation of cells with IL-3 also enhanced the clustering of FcγRI both before and after exposure to ICs. This increased clustering was dependent on an intact actin cytoskeleton. We found that chemical inhibition of the activity of the phosphatase PP1 reduced FcγRI inside-out signaling, although the phosphorylation of FcγRI itself was unaffected. Furthermore, the antibody-dependent cytotoxic activity of human neutrophils toward CD20-expressing tumor cells was increased after stimulation with TNFα and IFNγ. These results suggest that nanoscale reorganization of FcγRI, stimulated by cytokine-induced, inside-out signaling, enhances FcγRI cellular effector functions.

INTRODUCTION

Expressed on immune cells, Fc receptors (FcRs) are required for the cellular effector functions of antibodies, including protection against bacteria and phagocytosis. In humans, the Fcγ receptor (FcγR) family comprises both the activating receptors FcγRI, FcγRIIa, FcγRIIc, and FcγRIIIa as well as one inhibitory receptor, FcγRIIb. FcγRI (CD64), the high-affinity receptor for immunoglobulin G (IgG) (KD of 10−8 to 10−9 M), associates with the FcRγ chain and is constitutively expressed on monocytes, macrophages, eosinophils, and dendritic cells, as well as on neutrophils after activation. The docking mode between IgGs and the different FcγRs, including FcγRI, is very similar and conserved (1). Because of its high affinity, FcγRI is constitutively saturated with monomeric IgG, even after extravasation of immune cells or isolation from the blood (2). Therefore, the in vivo role of FcγRI in immune responses remains unclear. However, several studies have implicated an important role for FcγRI during inflammation, autoimmune responses, and monoclonal antibody immunotherapy in tumor models (35). In addition, FcγRI can efficiently induce major histocompatibility complex (MHC) class II antigen presentation (6).

FcγRI, saturated with prebound IgG, is capable of effective immune complex (IC) binding after cytokine stimulation (7). This phenomenon is termed “inside-out signaling” because ligand binding of the receptor is rapidly enhanced after intracellular signaling without altering receptor expression. This is a well-known process for integrin activation (8), as well as two other FcRs: FcαRI and FcγRIIa (9, 10). Prior studies using the Ba/F3 transfection model, a murine cell line that is dependent on interleukin-3 (IL-3) for survival, indicate that IL-3 can stimulate inside-out signaling (7, 9). In primary human leukocytes, analogous cytokines such as IL-5, interferon-γ (IFNγ), and tumor necrosis factor–α ( TNFα) stimulate ligand binding and receptor function (10). However, the mechanisms by which FcRs increase their ligand binding are largely unknown. Cytokine stimulation induces a small but significant increase in monomeric IgG binding of FcγRI-expressing cells. Strikingly, stimulation with cytokines strongly enhances the binding of IC without altering FcγRI expression (7). This effect is inhibited by okadaic acid (OA), a phosphatase inhibitor of PP2a at low concentrations and PP1 at higher concentrations (11). Furthermore, cytokine stimulation can also enhance antitumor responses of therapeutic antibodies mediated by FcR-bearing immune cells (10, 12).

How cytokine signaling alters FcγRI to increase its binding capacity is still unknown. Potential explanations include changes in FcγRI conformation, dynamics, and clustering. There is increasing evidence indicating that the membrane environment can modulate each of these aspects of receptor behavior (1315). Furthermore, the plasma membrane is organized into microdomains, such as actin corrals, lipid rafts, and protein islands, that play an essential role in promoting signal transduction for a number of membrane proteins (13, 14). Previous work has provided evidence that FcγRI resides in lipid rafts and that disruption of lipid rafts could increase ligand binding (16). Here, we sought to elucidate the mechanism of FcγRI inside-out signaling by cytokine stimulation. To this end, we studied the mobility and nanoscale organization of FcγRI in the plasma membrane using single-particle tracking (SPT) and super-resolution imaging, investigated the role of phosphorylation and the actin cytoskeleton, and measured the effect of cytokine stimulation on antibody-dependent cellular cytotoxicity (ADCC) of human neutrophils.

RESULTS

Inside-out signaling of FcγRI enhances IC binding

For our studies, we made use of the previously characterized FcγRI-expressing Ba/F3 cell line (7) and fluorescently labeled polyclonal anti–dinitrophenyl (DNP) rabbit IgG. This system allowed us to use high-resolution imaging to study the dynamics and distribution of IgG-bound FcγRI in both the presence and absence of IC formed by the addition of the model antigen DNP24-BSA (2,4-dinitrophenyl–bovine serum albumin). We found that IL-3 stimulation significantly enhanced the binding of Ba/F3-FcγRI cells to preformed IC, and this effect could be blocked by the addition of OA (Fig. 1A). FcγRI surface expression was constant under all treatments (Fig. 1, B and C, and fig. S1), confirming that the cytokine-induced binding enhancement was not a result of increased receptor expression. By confocal imaging, we did not observe large-scale changes in receptor organization (fig. S1B).

Fig. 1 IL-3 stimulation enhances IC binding that is not correlated with changes in the lateral mobility of FcγRI.

(A) Flow cytometric analysis of the binding of ICs to Ba/F3-FcγRI cells after IL-3 stimulation. Mean fluorescence intensity (MFI) data are means ± SD representative of three independent experiments. ns, not significant. (B) Flow cytometric analysis of FcγRI expression on Ba/F3-FcγRI cells stimulated as indicated. Dotted line represents the isotype control. Histogram plots are representative of four independent experiments. (C) Flow cytometric analysis of FcγRI expression on Ba/F3-FcγRI cells after IL-3 stimulation. MFI data are means ± SD from three independent experiments. (D and E) SPT by microcopic analysis of the movement of QD655-labeled anti-DNP IgG in live Ba/F3-FcγRI cells stimulated with IL-3, with and without antigen (DNP24-BSA), to induce IC. Particle trajectories (D) are representative of at least 10 experiments. Diffusion coefficients for QD655-labeled IC mobility (E) were calculated using mean square displacement analysis of individual SPT trajectories. For each trajectory, the diffusion coefficient (D) was calculated for control, IL-3–stimulated, and antigen-bound conditions. Data are means ± 95% confidence interval from 85 to 200 cells per condition pooled from all experiments. (F and G) FRAP analysis of the mobility of FcγRI-eYFP in Ba/F3 cells stimulated with IL-3 as indicated. Normalized and corrected mean relative fluorescence data are pooled from five independent experiments and (F) fit by nonlinear, two-phase regression (black line). The FcγRI half-time (G) was calculated per cell. Data and median from 70 to 81 cells per condition are pooled from all experiments. Scale bars, 1 μm. **P < 0.01, ***P < 0.001, and ****P < 0.0001 by one-way analysis of variance (ANOVA) and Tukey’s post hoc test (A and E), or Student’s t test (C and G).

Cytokine stimulation enhances FcγRI clustering in the plasma membrane

To better understand how cytokine stimulation may be associated with spatiotemporal changes in receptor behavior, we characterized the mobility of FcγRI using SPT (Fig. 1D). We found that receptor mobility was reduced when the addition of multivalent antigen (DNP24-BSA) stimulated the formation of IC on the plasma membrane surface, consistent with cross-linker–induced receptor aggregation (17). Treatment of the cells with IL-3 increased FcγRI diffusion, both in resting and antigen-bound receptors (Fig. 1E), but this was insensitive to OA pretreatment. Using the Kolmogorov-Smirnov test, we found that the diffusion coefficient of all conditions was significantly different from each other (P < 0.001), except for IL-3 versus IL-3/OA (both with and without antigen). We also used fluorescence recovery after photobleaching (FRAP) to measure the ensemble mobility of FcγRI–enhanced yellow fluorescent protein (eYFP) in Ba/F3 cells. In these experiments, stimulation with IL-3 did not increase the diffusional behavior or recovery half-time of the receptor (Fig. 1, F and G). Thus, changes in FcγRI mobility were not correlated with IL-3–stimulated enhanced IC binding.

Because small aggregates of immune receptors are capable of robust signaling without inducing changes in mobility (18), we used direct stochastic optical reconstruction microscopy (dSTORM), a localization-based super-resolution imaging technique that provides ~20-nm resolution to gain insight into the nanoscale organization of FcγRI (19). Surface FcγRI on Ba/F3-FcγRI cells were labeled with fluorescent Alexa Fluor 647 (AF647)–labeled anti-DNP IgG. Super-resolution images of FcγRI distribution were reconstructed from independent localizations of AF647-IgG on the basal side of the Ba/F3-FcγRI cells with and without IL-3 or antigen stimulation (Fig. 2A and fig. S2A). To quantify the images, we used the DBSCAN (density-based spatial clustering of applications with noise) clustering algorithm (20) to classify localizations into clusters based on their relative local spatial density in multiple, nonoverlapping individual 4-μm2 regions of interest (ROIs). This approach allowed us to identify individual clusters and make comparisons between the distributions of clusters across different conditions (Fig. 2B). We found that the equivalent cluster radius, calculated as the radius of a circle with the same area as the boundary of all fits within a cluster, was a useful representation of cluster size (Fig. 2C). Other methods of dSTORM data analysis, such as changing the maximal distance between neighboring cluster points to an ε of 35 nm instead of 50 nm or using the Getis-G clustering method instead of DBSCAN, gave similar results for all conditions.

Fig. 2 Cytokine stimulation promotes FcγRI clustering.

(A to C) dSTORM super-resolution microscopic analysis of fluorescently labeled anti-DNP IgG on Ba/F3-FcγRI cells stimulated as indicated. FcγRI-bound IgG clusters were identified by DBSCAN analysis (B) of the data in (A). Images (A and B) are representative of three independent experiments. Data and median from multiple independent ROIs on 20 to 30 cells per condition are pooled from all experiments. (D to G) Two-color dSTORM super-resolution microscopic analysis of fluorescently labeled anti-DNP IgG in Ba/F3-FcγRI cells stimulated with or without IL-3. Images (D) are representative of two independent experiments. Circles indicate examples of observed overlap between AF647 (magenta) and Cy3B (green). The two-color pair correlation function (E), g(r), indicates the mean number of particles at a distance between r and r + dr from a point of reference (middle green spot). g(r) is normalized such that randomly distributed particles give g(r) = 1 (top). If proteins are clustered, then the pair correlation will show a peak corresponding to the distribution of molecule separations (bottom). Two-color dSTORM images were analyzed using pair correlation analysis (F) and H-SET collapsed data (G). Data from >6 ROIs across three to four cells per condition (symbols) were pooled to generate pair correlation curves (solid lines). The dashed black line indicates a pair correlation of g(r) = 1, which is considered a random distribution. Scale bars, 500 nm (A), 200 nm (B), and 1 μm (D). **P < 0.01, ***P < 0.001, and ****P < 0.0001 by Kruskal-Wallis test and Dunn’s multiple comparison test.

Consistent with other reports of antigen-induced receptor cross-linking (17), the formation of IC increased FcγRI cluster size in all conditions (Fig. 2C and fig. S2B). Cluster size was dependent on antigen dose (fig. S2C), verifying that these observed changes were antigen-specific. In line with the increased antigen binding seen by flow cytometry, antigen-induced clusters were larger in IL-3–stimulated cells at all doses (fig. S2C). IL-3 stimulation before the addition of antigen increased the size of receptor clusters, and these increased further with antigen exposure (Fig. 2C). This IL-3–induced increase in cluster size was reduced by pretreatment with OA, the same treatment that reversed the effect of IL-3 on IC binding (Figs. 1A and 2C). Similar results were obtained when AF647-IgG IC was preformed in solution before addition to the Ba/F3-FcγRI cells (fig. S2D) or when AF647-conjugated human IgG1 was used instead of rabbit IgG (fig. S2E). These results demonstrate that cytokine stimulation leads to an enhanced clustering of FcγRI.

When performing dSTORM imaging, a minimum cluster size was generated from a single fluorophore because of repeated localizations and the precision of the measurement. Therefore, to confirm that the small shift in cluster radius observed with IL-3 stimulation was due to multiple proteins in a cluster, we performed two-color dSTORM. In these experiments, FcγRI was labeled stochastically with either AF647-labeled or Cy3B-labeled anti-DNP IgG (Fig. 2D), and receptor proximity was quantified using localization-based two-color pair correlation analysis (Fig. 2, E to G) (21, 22). The pair correlation of dual-labeled surface FcγRI in the absence of IL-3 was close to 1 but showed a small increase at short distances (<200 nm), suggesting that a fraction of receptors existed in small clusters in the absence of cross-linking (Fig. 2F, black line). Upon stimulation with IL-3, the correlation at short distances markedly increased (Fig. 2F, red line). To confirm that this difference was not due to an artifact of multiple localizations of the same fluorophore, we used H-SET (hierarchical single-emitter hypothesis test) analysis that collapses clusters of observations of blinking fluorophores into single estimates of their true locations (23). Analysis of the collapsed data showed the same relative increases of pair correlation at short distances (Fig. 2G). The higher pair correlation seen for FcγRI in the presence of IL-3 was consistent with IL-3 stimulation and increase in FcγRI clustering, as seen in the single-color dSTORM results (Fig. 2C). Together, the super-resolution imaging experiments demonstrated that FcγRI was found in small clusters on the plasma membrane, and the clustering was increased with IL-3 stimulation. This change in spatial organization with IL-3 suggests that cytokine stimulation sensitizes the cells to engage antigen or IC by enhanced preclustering of FcγRI.

An intact actin cytoskeleton is required for increased FcγRI clustering

Because the intracellular tail (CY) of the FcγRI α subunit interacts with the actin-binding proteins filamin A, periplakin, and protein 4.1G (2426), actin rearrangement may facilitate cytokine-induced changes in receptor clustering. Using IgG-coated beads, we measured FcγRI-mediated rosette formation in the presence of latrunculin A (LatA), an inhibitor of actin polymerization (Fig. 3). Consistent with previous data (7), we detected a nearly twofold increase in the percentage of rosettes formed with Ba/F3-FcγRI cells after IL-3 stimulation (Fig. 3, A and B). In contrast, pretreatment of the cells with LatA before IL-3 stimulation significantly reduced the rosette formation, similar to OA. LatA did not abrogate the rosette formation completely, indicating that, at least to some extent, IC can still bind without an intact cytoskeleton. LatA treatment did not alter FcγRI surface expression (fig. S3) or change FcγRI cluster size in control cells or prevent the formation of larger aggregates with antigen (Fig. 3C). However, LatA inhibited IL-3 enhancement of FcγRI cluster size both before and after the addition of antigen. These results suggested that cytokine-enhanced clustering of FcγRI required an intact actin cytoskeleton, whereas IC-induced clustering may be independent of the actin cytoskeleton.

Fig. 3 An intact actin cytoskeleton is required for increased clustering of FcγRI.

(A and B) Microscopic analysis of antibody (Ab)–opsonized bead binding to Ba/F3-FcγRI cells stimulated with IL-3, LatA (+ LatA), or OA as indicated. Rosettes were defined as cells bound with ≥5 beads. Images (A) are representative of three independent experiments, with arrows indicating rosettes. Quantified data (B) are pooled means ± SD at the indicated antibody concentrations. (C) dSTORM super-resolution microscopy analysis of fluorescently labeled anti-DNP IgG in Ba/F3-FcγRI cells stimulated as indicated. Data and median from 9 to 12 cells per condition are pooled from two independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by Kruskal-Wallis test and Dunn’s multiple comparison test.

Inhibiting PP1 phosphatase activity reduced FcγRI inside-out signaling without altering FcγRI phosphorylation

Inside-out signaling depends on phosphatase activity, which is inhibited by treatment with OA (7) that also blocks FcγRI nanoscale reorganization (Fig. 2). However, OA inhibits the phosphatase activity of PP2a at low concentrations and PP1 at higher concentrations (11). To determine whether PP1 may be involved in FcγRI inside-out signaling, we treated cells with the inhibitor tautomycetin (TC), which has a ~40× higher selectivity for PP1 over PP2a. We found that TC was a much more potent inhibitor of IC binding by IL-3–stimulated Ba/F3-FcγRI cells than OA, which inhibited IC binding at 10 nM, whereas OA required concentrations of 100 nM or more to show an effect (Fig. 4A). Although these differences may be influenced by compound potency, these data suggested the involvement of PP1 in cytokine-stimulated FcγRI clustering.

Fig. 4 An intact actin cytoskeleton is required for increased clustering of FcγRI.

(A) Microscopic analysis of antibody-opsonized bead binding to Ba/F3-FcγRI cells stimulated with IL-3, TC, or OA as indicated. Rosettes were defined as cells bound with ≥5 beads. Quantified data are means ± SD pooled from three independent experiments. (B to D) Phos-tag SDS-PAGE gel and Western Blot analysis (left) or quantification (right) of phosphorylation on FcγRI α subunit immunoprecipitated from Ba/F3-FcγRI cells (B and C) or FcγRI 4S/2T>A (D) after stimulation with IL-3, OA, or TC, as indicated. Short exposure blots (top), long exposure blots (middle), and control blots (bottom) are representative of three independent experiments. Increased band size indicates phosphorylation of FcγRI α subunit (p-FcγRI, phosphorylated FcγRI). Numbers indicate the protein marker size in kilodalton. Quantified data are relative mean band intensity values ± SEM and pooled from three independent experiments. (E) Microscopic analysis of antibody-opsonized bead binding to Ba/F3 cells expressing FcγRI WT (wild type), FcγRI 4S/2T>A, or FcγRI-ΔCY (without intracellular domain) stimulated with or without IL-3. Data are representative of three independent experiments. ***P < 0.001 and ****P < 0.0001 by Student’s t test (A) or one-way ANOVA and Bonferroni’s multiple comparison test (E).

Because PP1 is a serine/threonine phosphatase, and the CY domain of FcγRI contains four serine and two threonine residues, we next investigated whether FcγRI is directly phosphorylated. Neither IL-3 stimulation nor phosphatase inhibition influenced the abundance of p-FcγRI as observed by Western blot and Phos-tag SDS–polyacrylamide gel electrophoresis (PAGE) analysis of FcγRI α subunit immunoprecipitants (Fig. 4, B and C). Similarly, although the p-FcγRI was no longer present when the intracellular serines and threonines were mutated to alanines (FcγRI 4S/2T>A mutant) (Fig. 4D), cytokine stimulation of the FcγRI 4S/2T>A mutant resulted in equal rosette formation compared to FcγRI wild type. These data suggested that serine/threonine phosphorylation within the CY domain of FcγRI was not required for inside-out signaling. Rosette formation was not increased after IL-3 stimulation when a truncated version of FcγRI that lacked the CY domain was used (Fig. 4E). Together our data indicated that although the intracellular domain of FcγRI was required for inside-out signaling, the 4S/2T residues within this domain did not play a role.

Inside-out signaling increases FcγRI clustering and effector functions in human myeloid cells

The above experiments were performed using the Ba/F3 cell line expressing FcγRI. To confirm that cytokine-regulated FcγRI clustering also occurs in primary immune cells, we quantified receptor organization in human monocytes from healthy donors. Human monocytes endogenously express FcγRI, and treatment with IFNγ and TNFα induces inside-out signaling of FcγRI in these cells (7). Similar to our cell line results, we found that stimulation of monocytes with these cytokines did not alter FcγRI expression or diffusion (fig. S4, A and B). By using single-color dSTORM, we found that the cluster sizes observed in human monocytes were comparable to those measured in Ba/F3-FcγRI cells. Stimulation with IFNγ and TNFα increased FcγRI clustering, both before and after the addition of antigen (Fig. 5A), which was inhibited by pretreatment with OA. These data confirmed that cytokine stimulation increased FcγRI clustering in human monocytes.

Fig. 5 Cytokines increase FcγRI clustering and IgG-mediated ADCC in human myeloid cells.

(A) dSTORM super-resolution microscopic analysis of fluorescently labeled anti-DNP IgG on human monocytes stimulated as indicated. Mean cluster radius data and median from five to eight cells per donor are pooled from two independent experiments. (B) Specific lysis of CD20-expressing tumor cell lines in the presence of anti-CD20 antibody (1 μg/ml; rituximab) and human neutrophils, simulated as indicated. Data are representative of two independent experiments. *P < 0.05, **P < 0.01, and ****P < 0.0001 by Kruskal-Wallis test and Dunn’s multiple comparison test (A) and Student’s t test (B).

Finally, we investigated whether cytokine stimulation also enhanced FcγR effector functions. To study this, we measured ADCC of neutrophils in response to different CD20-expressing tumor cell lines in the presence of the therapeutic IgG anti-CD20 antibody rituximab. Two activating FcγRs are expressed by neutrophils (FcγRI and FcγRIIa), both of which are under inside-out control. Neutrophils induced tumor cell lysis of Ramos, Daudi, and EL4-CD20 cells in the presence of an anti-CD20 antibody (Fig. 5B). Cytokine-stimulated neutrophils had significantly increased cytotoxic capacity, especially when EL4-CD20 cells were used as targets (Fig. 5B). Expression of both FcγRI and FcγRIIa on neutrophils did not change during these experiments (fig. S4C). These results in primary human cells suggest that cytokine-induced FcγR activation can also augment immune cell function.

On the basis of our results, we propose a model in which FcγRI exists as monomers or small multimers bound by monomeric IgG in unstimulated immune cells. After cytokine stimulation, PP1 activity and actin polymerization together lead to enhanced FcγRI clustering in the plasma membrane (Fig. 6). This may lead to increased FcγRI-IC binding capacity and, ultimately, enhanced FcγRI effector functions.

Fig. 6 Mechanisms of FcγRI inside-out signaling.

(A) In unstimulated cells, FcγRI is bound by monomeric IgG because of its high affinity. FcγRI might exist as monomers or possibly small multimers, as indicated by the two-color dSTORM data. (B) Cytokine stimulation of FcγRI-expressing cells first induces “outside-in” signaling when a cytokine binds its cytokine receptor. Subsequently, inside-out signaling is initiated, resulting in increased clustering of FcγRI in the plasma membrane. The phosphatase PP1 may be essential for this process, consistent with evidence that inside-out signaling can be blocked by OA and TC. Increased actin polymerization is likely to facilitate the clustering of FcγRI, although these clusters retain the same mobility as without cytokine stimulation. Together, this leaves the cell in a primed state that has a stronger IC binding capacity. (C) The presence of antigens or IC can induce clustering of FcγRI. These cluster sizes are approximately similar to the inside-out signaling–induced clusters of FcγRI. However, the mobility of these clusters is decreased because antigen or IC binding initiates cross-linking of FcγRI that leads to ITAM signaling and FcγRI effector functions. (D) Antigens or IC can bind more efficiently to FcγRI on cells stimulated with cytokines. Both the inside-out signaling and the antigen increase the FcγRI clusters, resulting in large FcγRI clusters in the plasma membrane and stronger effector functions like ADCC. FcγRI in these clusters is also less mobile compared to stimulated cells without antigen.

DISCUSSION

Our results provide evidence of nanoscale reorganization of FcγRI in response to cytokine stimulation in both a cell line and primary human myeloid cells. The increase in receptor clustering was prevented when cells were treated with inhibitors of PP1 and actin polymerization, indicating an important role for phosphatase and cytoskeletal activity in FcγRI inside-out signaling. The specific dependence of cytokine-induced FcγRI activation on actin polymerization suggests that this process may induce cytoskeleton rearrangements that facilitate FcγRI clustering and sensitize the cells to bind opsonized pathogens. Actin might also facilitate the binding or recruitment of other proteins, which, in turn, can enhance FcγRI clustering. The CY domain of FcγRI interacts with the actin-binding protein filamin A. This interaction between filamin A and FcγRI is reduced upon large IC binding to FcγRI, which may uncouple FcγRI from the actin cytoskeleton to allow efficient phagocytosis (27). In addition, the human interactome identified a significant interaction between filamin A and PP1 (28), suggesting that filamin A could facilitate interaction of FcγRI and PP1. However, FcγRI is not expressed in these cells (HeLa cells); therefore, no direct evidence for an interaction between FcγRI and PP1 exists.

For many other immune receptors, receptor clustering by microdomains is important for their regulation and function. One well-known example is the role of clustering in integrin inside-out signaling, where even a subtle increase in clustering can lead to substantial increases in cellular adhesion (29). Furthermore, clustering facilitates antigen binding to FcεRI, regulates B cell receptor (BCR) activity, and enhances FcγRIIa binding to ligand (15, 17, 30). Besides clustering, conformational changes in the extracellular (EC) domain of receptors can be involved in receptor activation. For example, the activation of integrins requires a distinct conformational change from a bent to an extended conformation, essential for higher ligand-binding affinity (8). In case of the BCR, a more open conformation of the receptor is induced for its activation (31). Here, we focused on altered lateral mobility and clustering of FcγRI in the plasma membrane as possible mechanism of FcγRI inside-out activation. Alternatively, conformational changes in FcγRI may also alter its affinity for IgG-IC. The crystal structure of FcγRI in complex with the IgG Fc fragment indicates that the EC3 domain undergoes a conformational shift after IgG is bound (32, 33). Future research might elucidate whether the increased clustering of FcγRI after cytokine stimulation coincides with a conformational change of the receptor.

We demonstrated that the CY domain of FcγRI was necessary for inside-out signaling but that the serine/threonine motifs within this domain did not play a role. In contrast, in FcγRI-transfected P388D1 cells, the CY domain is constitutively serine-phosphorylated as determined by an antiphosphoserine Western blot (34). Because this approach is not quantitative, we instead determined the degree of phosphorylation of FcγRI using a Phos-tag SDS-PAGE gel, which allowed for direct quantification of protein phosphorylation. We found that a phosphorylated fraction of FcγRI in unstimulated cells was absent in FcγRI 4S/2T>A mutant cells, indicating that this signal represents serine/threonine phosphorylation. Edberg et al. also show that the CY domain serines are transiently dephosphorylated upon FcγRI cross-linking (outside-in signaling), and this dephosphorylation is prevented by OA treatment (34). These data contrast with inside-out signaling, where OA inhibited enhanced IC binding of FcγRI but did not directly influence FcγRI phosphorylation. Furthermore, the 4S>A mutant generated by Edberg et al. results in reduced phagocytosis compared to FcγRI wild type (34), whereas our 4S/2T>A mutant had no influence on IC binding after IL-3 stimulation (Fig. 5D). IL-3 stimulation led to inside-out activation of FcγRI, as measured by increased IC binding in the rosette assay, which was performed at temperatures that prevent outside-in signaling. However, phagocytosis is dependent on FcγRI outside-in signaling, leading to immunoreceptor tyrosine-based activation motif (ITAM) phosphorylation of the FcR γ chain (35). Together, these observations suggest that outside-in and inside-out signaling are distinct processes and that dephosphorylation of these four serines, although important for outside-in signaling leading to phagocytosis, is not required for cytokine-induced inside-out signaling.

Receptor phosphorylation is not the only posttranslational modification that can influence protein function. Because direct phosphorylation of the FcγRI α subunit was excluded by our data, other modifications such as ubiquitination and methylation might also affect FcγRI function or localization. For other FcRs, including FcεRI, FcγRIIa, and FcγRIIIa, ligand-induced ubiquitination is essential for receptor internalization and degradation, providing a negative feedback on FcR activity (36). For FcγRI, ubiquitination may play a role as well because this receptor is continuously internalized and recycled to the plasma membrane within minutes under steady-state conditions. Therefore, in future studies, it would be very interesting to monitor the ubiquitination of FcγRI in resting, inside-out activated, and IC-bound cells.

One promising clinical application of cytokine stimulation is to enhance the activity of neutrophils (or other FcγRI-expressing immune cells) during administration of antitumor therapeutic antibodies (Fig. 6B) (10). Administration of cytokines in combination with therapeutic antibodies against several tumor antigens increases the efficacy of these antibodies both in vitro and in vivo (12, 37, 38). However, many of the tested cytokines in these in vivo studies are associated with processes that take hours or even days. Granulocyte colony-stimulating factor (G-CSF), for example, is associated with increased FcγRI expression on neutrophils and increased recruitment of effector cells (38). The inside-out signaling of FcγRI we describe here occurs rapidly—on the order of minutes—to promote FcγRI-IC binding. Therefore, we expect that cytokine stimulation in combination with antibody therapy would increase the binding capacity of FcγRI-expressing cells to tumor cells directly after cytokine administration (minutes), whereas increasing FcγRI expression and recruiting more effector cells from the bone marrow occur later (hours or days). Together, this may lead to more effective antitumor responses, especially when the timing of cytokine administration is taken into account. A thorough understanding of the regulatory mechanisms of FcγRI activation may aid manipulation of immune responses using cytokines during infections, vaccinations, antibody immunotherapy, or autoimmune diseases.

MATERIALS AND METHODS

Cell lines

Ba/F3 (murine pro-B cell line), Ramos, Daudi, and EL4-CD20 cells were cultured in RPMI 1640 (Gibco) supplemented with 10% fetal calf serum (FCS), penicillin/streptomycin, and murine IL-3 [provided by P. Coffer, University Medical Center (UMC) Utrecht] (7, 9). The retroviral vector pMX human FcγRI was described previously (24). In addition, the human FcγRI in this vector was replaced with a fusion protein of human FcγRI and eYFP. Using site-directed mutagenesis, the four serines (Ser328, Ser331, Ser339, and Ser340) and two threonines (Thr312 and Thr374) in the intracellular domain of FcγRI were mutated to alanine (4S/2T>A mutant); to generate FcγRI-ΔCY, Thr312 was first mutated to alanine, and then Glu316 was mutated to a stop codon. Amphotropic viral particles produced in human embryonic kidney 293T cells were used to transduce Ba/F3 cells. Ba/F3-FcγRI (with mutations) and Ba/F3-FcγRI-eYFP cells were sorted on a FACSAria (BD Biosciences) for FcγRI expression.

Reagents

Antibodies used were as follows: anti-FcγRI, either unlabeled rabbit IgG1 (clone EPR4624, Abcam) or AF488- or AF647-labeled [mIgG1 (mouse IgG1) clone 10.1, BioLegend]; eFluor 450–labeled anti-CD14 (61D3, eBioscience); fluorescein isothiocyanate (FITC)–labeled goat anti-rabbit IgG (Jackson ImmunoResearch); goat anti-rabbit IgG horseradish peroxidase (HRP)–conjugated (Pierce); and rabbit IgG anti-DNP (anti-DNP IgG, polyclonal; Vector Labs). Anti-DNP IgG was biotinylated and subsequently conjugated to Qdot 655 (QD655) (Invitrogen) as described for IgE (17). Anti-DNP IgG and anti-trinitrophenyl (TNP) human IgG1 (39) were fluorescently labeled with AF647 or Cyanine 3B (Cy3B) (Life Technologies) following the same protocol as for biotinylation. Where indicated, 1 μM OA (Enzo Life Sciences) or 1 to 1000 nM TC (Tocris) was added for 30 min, or LatA (0.1 μg/ml; Life Technologies) was added 10 min before IL-3 stimulation. Phos-tag Acrylamide was from Wako Chemicals. The anti-human CD20 antibody rituximab (Roche) was purchased from the pharmacy of UMC Utrecht. Eight-well Lab-Tek chambers (Nunc) were coated with poly-l-lysine [1 mg/ml in 10% 1× phosphate-buffered saline (PBS), 90% water] for 30 min at room temperature (RT). Thrombin and fibrinogen were both from Enzyme Research Laboratories.

FcγRI (CD64) expression and IC binding

Ba/F3-FcγRI cells were cytokine-starved overnight in RPMI 1640 with 1% FCS (RPMI 1640–1% FCS). The next day, cells were stimulated with IL-3 (in RPMI 1640–1% FCS) for 1 hour at 37°C. For confocal microscopy, cells were added to eight-well Lab-Tek chambers, fixed with 4% paraformaldehyde (PFA; Sigma-Aldrich), and stained with AF488-labeled anti-FcγRI. Images were collected on a Zeiss LSM 510 two-photon confocal microscope (Zeiss Axiovert 200 M inverted microscope with X,Y-motorized stage) with a 63× oil immersion objective using an argon laser.

For flow cytometry, 1 × 105 Ba/F3-FcγRI cells per well were added to a 96-well plate, washed with cold PBS, and stained with AF647-labeled anti-FcγRI for 1 hour at 4°C. Afterward, cells were washed with PBS and fixed with 1% PFA. Expression of FcγRI was measured on a FACSCanto II (BD Biosciences). FcγRI expression after IL-3 stimulation was routinely measured. For monocyte flow cytometry, 2 × 105 peripheral blood mononuclear cells (PBMCs) per well were added to a 96-well plate and stimulated with TNFα and IFNγ (500 and 400 U/ml, respectively) for 1 hour in RPMI 1640–1% FCS at 37°C. Afterward, cells were washed once with cold PBS and stained with eFluor 450 anti-CD14 and AF647 anti-FcγRI for 1 hour at 4°C. Cells were washed with PBS and fixed with 1% PFA. Expression of FcγRI on CD14high monocytes cells was measured on a FACSCanto II (BD Biosciences).

Binding of Ba/F3-FcγRI to monomeric anti-DNP IgG was measured by plating 1 × 105 Ba/F3-FcγRI cells per well in a 96-well plate, washing the cells with cold PBS, and incubating them with different concentrations of this antibody for 1 hour at 4°C. Afterward, cells were washed with PBS and incubated with a FITC-labeled anti-rabbit IgG antibody for 45 min at 4°C. Cells were washed with PBS and fixed with 1% PFA. IgG binding was measured on a FACSCanto II (BD Biosciences). IC binding was assessed with preformed IC (AF647-labeled anti-DNP IgG/DNP24-BSA mixed at 3:1 ratio) and measured on a HyperCyt Autosampler (Intellicyt).

Primary monocytes

Deidentified blood was obtained from healthy donors [University of New Mexico (UNM) Hospital Blood and Tissue Bank and Mini Donor Dienst UMC Utrecht], and PBMCs were isolated by Ficoll-Paque gradient separation (GE Healthcare). PBMCs were allowed to rest for 1 hour in RPMI 1640–1% FCS at 37°C in (uncoated) eight-well Lab-Tek chambers. Next, PBMCs were stimulated with human TNFα and IFNγ (500 and 400 U/ml, respectively) for 1 hour in RPMI 1640–1% FCS at 37°C. During the last 10 min of incubation, IV.3 antigen binding fragments [F(ab′)2] (5 μg/ml) and 3G8 F(ab′)2 (1 μg/ml) were added to block the other FcγR (40). Wells were washed with RPMI 1640–1% FCS to remove all unbound cells, leaving the adherent cells (monocytes) in the wells. Next, monocytes were stained for SPT or super-resolution imaging.

Sample preparation for SPT

Cytokine-starved Ba/F3-FcγRI cells were stimulated with IL-3, labeled with a low concentration of QD655-labeled anti-DNP IgG (2 nM) for 5 min at RT, and subsequently saturated with unlabeled anti-DNP IgG (100 nM) for 5 min at RT. This resulted in the QD655 labeling of single receptors (2 to 20 per cell). Cells were washed with PBS, resuspended in Hanks’ buffer, and plated in eight-well Lab-Tek chambers.

SPT image registration and processing

SPT was performed as described previously (17, 41, 42). Images were acquired at 20 frames/s using an Olympus IX71 inverted microscope with a 1.2–numerical aperture (NA) 60× water objective lens combined with an extra ×0.6 magnification. An objective heater (Bioptechs) maintained samples at 34° to 35°C. A mercury lamp with a 436/10-nm band-pass (BP) excitation filter provided wide-field excitation. Emission was collected by an electron multiplying charge-coupled device (CCD) camera (Andor iXon 887) using a DuoView image splitter (Optical Insights) to image the QD655 (655/40 BP) probe. All data reported were collected after focusing on the apical surface of the cells. Image processing was performed using MATLAB (MathWorks) functions in conjunction with the image processing software DIPImage (Delft University of Technology). Single-molecule localization and trajectory elongation were performed as previously described (22). The diffusion coefficient (D) was calculated on the basis of the mean square displacement over all QD655 tracks within an experiment (17).

Fluorescence recovery after photobleaching

For FRAP experiments, Ba/F3-FcγRI-eYFP cells were washed twice in PBS and seeded in a fibrin matrix: fibrinogen (2.5 mg/ml) and thrombin (1 × 10−4 U/μl) in RPMI 1640 without phenol red supplemented with 1% FCS/2 mM l-glutamin on a μ-Dish (35 mm, high; ibidi). The cells were incubated overnight in this matrix; to prevent dehydration of the matrix, RPMI 1640 without phenol red with 1% FCS/l-glutamin was added. The next day, FRAP measurements were performed on these cells (no IL-3) or after IL-3 stimulation for 30 min (IL-3). FRAP experiments were performed on a Zeiss LSM 710 confocal microscope with a 63× oil objective lens and an environmental chamber for temperature (37°C) and CO2 (5%) control. An argon laser provided the 488-nm excitation. Ten prebleach images were acquired, after which a small area (~1 μm2) spanning the membrane was bleached for 0.2 s to obtain a bleach of ~50%. The fluorescence in this region was monitored by acquiring images at 7.3 frames/s for 30 to 35 s per cell. For each condition, >70 cells were measured.

FRAP data analysis

The fluorescence intensity of the bleached area was corrected for loss of fluorescence during the measurement (by subtracting the background fluorescence intensity and correcting for the overall fluorescence intensity) and normalized (by setting the mean fluorescence before bleaching to 1; this corrects for differences in cell fluorescence between measurements). The relative mean fluorescence intensity of the bleached area of all cells per condition was plotted, and a nonlinear, two-phase association (GraphPad Prism 6 Software) was used to fit the experimental data. To determine the FcγRI half-time, the relative fluorescence intensity of the bleached area of each imaged cell was plotted individually, and a nonlinear, one-phase association (GraphPad Prism 6 Software) was used to fit the data and calculate the half-time.

Super-resolution imaging

Cytokine-starved Ba/F3-FcγRI cells were stimulated with IL-3 and labeled with AF647-labeled anti-DNP IgG (2 μg/ml), DNP24-BSA (1 μg/ml), preformed IC (AF647-labeled anti-DNP IgG/DNP24-BSA mixed at 3:1 ratio), or AF647-labeled anti-TNP human IgG1, as indicated. This anti-TNP antibody is cross-reactive with DNP and binds DNP24-BSA with a similar affinity as rabbit anti-DNP IgG (as measured with a DNP24-BSA binding enzyme-linked immunosorbent assay). For two-color super-resolution imaging, cells were labeled with a mix of AF647-labeled anti-DNP IgG (0.667 μg/ml) and Cy3B-labeled anti-DNP IgG (1333 μg/ml) at RT after IL-3 incubation. After labeling, cells were washed with PBS and incubated in Hanks’ buffer with or without antigen (DNP24-BSA at 1 μg/ml, unless other concentrations are indicated) for 10 min at 37°C to induce IC. Cells were washed with Hanks’ buffer, plated in eight-well Lab-Tek chambers, and allowed to adhere for 10 min at 37°C. Cells were then fixed with 4% PFA and 0.2% glutaraldehyde for 1 to 2 hours. Before super-resolution imaging, 200 μl of fresh super-resolution buffer [50 mM tris, 10 mM NaCl, 10% glucose, glucose oxidase (168.8 U/ml), catalase (1404 U/ml), 10 mM cysteamine hydrochloride (pH 8.0)] was added to the well. Labeling of human monocytes followed the same protocol as for Ba/F3-FcγRI, with some adjustments: Labeling and incubation with antigen were both done at 37°C, and monocytes were fixed immediately after incubation with IC.

dSTORM imaging was performed using an inverted microscope (IX71; Olympus America) equipped with an oil immersion objective 1.45-NA total internal reflection fluorescence objective (U-APO 150×; Olympus America) (19). A 637-nm diode laser (HL63133DG, Thorlabs) was used for AF647 excitation, and a 561-nm frequency-doubled diode laser (Spectra-Physics Cyan Scientific) was used for Cy3B excitation. A quad-band dichroic and emission filter set (LF405/488/561/635-A, Semrock) was used for sample illumination and emission. Emission light was separated onto different quadrants of an Andor iXon 897 electron-multiplying CCD camera (Andor Technology) using a custom-built two-channel splitter with a 585-nm dichroic (Semrock) and additional emission filters (692/40 and 600/37 nm). The sample chamber of the inverted microscope (IX71, Olympus America) was mounted in a three-dimensional piezo stage (Nano-LPS; Mad City Labs) with a resolution along the xyz axes of 0.2 nm. Sample drift was corrected throughout the imaging procedure using a custom-built stage stabilization routine. Images were acquired at 57 frames/s in TIRF (total internal reflection fluorescence), and between 10,000 and 20,000 frames were collected for each image reconstruction.

Super-resolution image reconstruction and data analysis

dSTORM images were analyzed and reconstructed with custom-built MATLAB functions as described previously (43, 44). For each image frame, subregions were selected on the basis of local maximum intensity. Each subregion was then fitted to a pixelated Gaussian intensity distribution using a maximum likelihood estimator. Fitted results were rejected on the basis of log-likelihood ratio and the fit precision, which was estimated using the Cramér-Rao lower-bound values for each parameter, as well as intensity and background cutoffs.

Analysis of dSTORM FcγRI cluster data was performed using the density-based DBSCAN algorithm (20) implemented in MATLAB (45) as part of a package of local clustering tools (https://stmc.unm.edu/). Parameters chosen were a maximal distance between neighboring cluster points of ε = 50 nm and a minimal cluster size of six observations. Cluster boundaries were produced with the MATLAB “boundary” function using a default methodology that produced contours halfway between a convex hull and a maximally compact surface enclosing the points. The cluster areas within these boundaries were then converted into the radii of circles of equivalent area for a more intuitive interpretation. ROIs of size 2 μm × 2 μm (4 μm2) were selected from the set of images from which statistics for the equivalent radii were collected per ROI.

Two-color image analysis

Dual-color images were acquired by imaging AF647 and Cy3B sequentially. AF647 was imaged first to prevent photobleaching by Cy3B. To correct for shifts due to chromatic aberrations, channels were aligned using multicolor beads (TetraSpeck; Invitrogen). Using the piezo stage, we placed a single TetraSpeck bead at 36 locations (6 × 6) uniformly distributed across the image window. The emission position in both channels was fitted and recorded. This transform was then used to convert fits from the Cy3B channel into the AF647 channel. A channel registration data set was always taken within 2 hours of any data acquisition. This was necessary to ensure that the registration transform was relevant and that no alignment drift occurred. Two-color super-resolution data sets were analyzed by localization-based pair correlation analysis similar to previously described methods (21, 4649). Briefly, multiple subregions within the centers of each cell (of size 3 μm × 3 μm, excluding the lateral membrane region) were selected, and opposite color localizations were collected radially in 5-nm bins after being angularly averaged over a set of ROIs. The computation of the pair correlation function was based on MATLAB code originally developed by S. Veatch (21). In some of the analyses, the data were first run through H-SET, a top-down hierarchical clustering algorithm implemented in MATLAB that collapses clusters of observations of blinking fluorophores into single estimates of their true locations (localizations) (23). Briefly, for a cluster of observations to be collapsed into a single localization, a hypothesis test is performed with the null hypothesis that all observations come from the same fluorophore. The null hypothesis is not rejected if the P value, calculated using a log-likelihood ratio statistic, is larger than a specified level of significance (0.01 was used here).

Rosette assay

The rosette assay using Dynabeads was adapted from van der Poel et al. (7). DNP24-BSA Dynabeads were opsonized with anti-DNP IgG using the indicated concentrations. Ba/F3-FcγRI cells were stimulated with IL-3 and, where indicated, incubated with inhibitors before IL-3 stimulation. Ba/F3-FcγRI cells were combined with beads and incubated for 1 hour at 4°C on a shaker. Rosette formation was evaluated using microscopy, and cells bound to ≥5 beads were defined as rosettes.

Phos-tag SDS-PAGE

To measure FcγRI phosphorylation, cytokine-starved Ba/F3-FcγRI cells stimulated with IL-3 (1 ng/ml) for 45 min at 37°C (where indicated, IL-3 stimulation was preceded with treatment by inhibitors). Afterward, cells were lysed in cold triton lysis buffer (50 mM tris-HCl, 150 mM NaCl, 1% triton with cOmplete EDTA-free protein inhibitor cocktail and PhosSTOP cocktail, Roche). Lysates were incubated with protein G beads coupled to mouse anti-FcγRI α subunit antibody m22 for 4 hours at 4°C. Subsequently, beads were washed in Triton lysis buffer and boiled in reducing sample buffer. Samples were loaded onto a 12% polyacrylamide separating gel (no Phos-tag) and a 7.5% polyacrylamide separating gel containing 100 μM Phos-tag (Wako Chemicals) and 200 μM MnCl2 (50, 51). After electrophoresis, the Phos-tag gel was washed in a 1 mM EDTA solution, followed by a wash with EDTA-free solution, before transferring the proteins to a polyvinylidene difluoride membrane. Membranes were blocked with BSA and incubated with rabbit anti-FcγRI α subunit monoclonal antibody (clone EPR4624) for 2 hours at RT. After washing, membranes were incubated with goat anti-rabbit HRP–conjugated secondary antibodies for 1 hour at RT. FcγRI was visualized using ECL prime (GE Healthcare), imaged with a ChemiDoc MP (Bio-Rad), and quantified using Image Lab Software. For quantification, nonsaturated images were used.

Antibody-dependent cellular cytotoxicity

ADCC of 51Cr-labeled target cells was described previously (52). Briefly, 1 × 106 target cells were labeled with 100 μCi (3.7 MBq) 51Cr for 2 hours. After extensive washing, cells were adjusted to 105/ml. Blood was obtained from healthy donors at the UMC Utrecht. The neutrophil or PMN (polymorphonuclear leukocyte) fraction was isolated from blood by Ficoll/Histopaque separation (GE Healthcare; Sigma-Aldrich). PMNs were pretreated for 15 min at 37°C with TNFα (250 U/ml). Effector cells (40:1 effector-to-target ratio), the monoclonal antibody rituximab at 1 μg/ml, medium, and tumor cells were added to round-bottom microtiter plates (Corning Incorporated). After 4 hours of incubation at 37°C, 51Cr release into the supernatant was assessed by measuring the radioactivity of supernatant samples in counts per minute (cpm). Percentage of specific lysis was calculated using the following formula: [(experimental cpm − basal cpm)/(maximal cpm − basal cpm)] × 100, with maximal lysis determined in the presence of 3% Triton and basal lysis in the absence of antibodies and effector cells.

Statistical analysis

Statistical analysis was performed using GraphPad Prism 6 Software. An unpaired Student’s t test was used to compare mean values between two groups. The Kruskal-Wallis test that does not assume a normal distribution was used to compare the cluster size distribution. Statistical analysis for other multiple comparisons was performed using one-way ANOVA with the indicated post hoc tests.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/540/eaaq0891/DC1

Fig. S1. FcγRI surface expression does not change after IL-3 stimulation.

Fig. S2. FcγRI clustering increases after cytokine stimulation.

Fig. S3. FcγRI surface expression does not change after LatA pretreatment.

Fig. S4. FcγRI expression does not change after cytokine stimulation of human myeloid cells.

REFERENCES AND NOTES

Acknowledgments: We thank C. van der Poel for generating the cell lines described in this paper. In addition, we thank M. Plantinga for providing us with human cytokines, K. Crookston for assistance with the isolation of PBMCs, and P. Coffer for providing murine IL-3. We acknowledge use of the Spatiotemporal Modeling Center (STMC) Super-Resolution Core, the UNM Cancer Center Fluorescence Microscopy Shared Resource, and the UNM Comprehensive Cancer Center. Funding: This work was supported by grants from the NIH (R01GM100114 to D.S.L., P50GM085273 to the New Mexico STMC, and P30CA118100 to the UNM Comprehensive Cancer Center), a EFIS-IL (European Federation of Immunological Societies–Immunology Letters) World Fellowship (to A.M.B.), and KiKa grant 227 (to T.t.B.). Author contributions: J.H.W.L. and D.S.L. conceived the project and supervised all research. A.M.B., S.L.S., D.S.L., and J.H.W.L. designed the experiments and wrote the manuscript. A.M.B., S.L.S., G.L.A.B., and T.t.B. performed the experiments. S.L.S., M.J.W., C.C.V., and A.M.B. designed and performed the computational analysis. D.S.L., K.A.L., and G.V. provided expertise and specialized reagents and equipment. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article