Research ArticleANTIGEN RECEPTOR SIGNALING

A PIP2-derived amplification loop fuels the sustained initiation of B cell activation

See allHide authors and affiliations

Science Immunology  17 Nov 2017:
Vol. 2, Issue 17, eaan0787
DOI: 10.1126/sciimmunol.aan0787

Sustaining B cell receptor signaling

It is well established that, upon activation, B cell receptors (BCRs) form microclusters that function as platforms to recruit intracellular molecules that relay signals downstream of the BCRs. Xu et al. have examined how the localization of phosphatidylinositol 4,5-biphosphate (PIP2) is regulated to sustain BCR signaling. Using state-of-the-art microscopy to examine the localization of PIP2 after BCR activation, the authors report that PIP2 is depleted within the BCR microclusters but is enriched outside the microclusters. Furthermore, by developing systems to manipulate PIP2 localization in this context, the authors found that maintenance of this PIP2 gradient is vital to sustain BCR signaling.

Abstract

Lymphocytes have evolved sophisticated signaling amplification mechanisms to efficiently activate downstream signaling after detection of rare ligands in their microenvironment. B cell receptor microscopic clusters (BCR microclusters) are assembled on the plasma membrane and recruit signaling molecules for the initiation of lymphocyte signaling after antigen binding. We identified a signaling amplification loop derived from phosphatidylinositol 4,5-biphosphate (PIP2) for the sustained B cell activation. Upon antigen recognition, PIP2 was depleted by phospholipase C–γ2 (PLC-γ2) within the BCR microclusters and was regenerated by phosphatidic acid–dependent type I phosphatidylinositol 4-phosphate 5-kinase outside the BCR microclusters. The hydrolysis of PIP2 inside the BCR microclusters induced a positive feedback mechanism for its synthesis outside the BCR microclusters. The falling gradient of PIP2 across the boundary of BCR microclusters was important for the efficient formation of BCR microclusters. Our results identified a PIP2-derived amplification loop that fuels the sustained initiation of B cell activation.

INTRODUCTION

B cells recognize antigens through the surface-expressed B cell receptors (BCRs). BCR is a molecular complex composed of a membrane immunoglobulin (Ig) with a heterodimer of Igα and Igβ (1). Upon antigen binding, tyrosine residues within the immunoreceptor tyrosine-based activation motifs of Igα and Igβ are phosphorylated by the Src family kinase, which initiates the recruitment of signaling molecules, such as Syk and phospholipase C–γ2 (PLC-γ2), etc. These allow for the functional assembly of BCR microclusters on the plasma membrane, which are the fundamental platform for the initiation of B cell activation because both the number and intensity of the BCR microclusters determine the strength of B cell activation (14).

Recent advances in live cell imaging studies demonstrated that dynamics of the lipid microdomains on the plasma membrane potently regulate the antigen receptor clustering and activation in both B and T lymphocytes (57). Phosphatidylinositol 4,5-biphosphate (PIP2) is the substrate of PLC-γ2 in a reaction producing diacylglycerol (DAG) and inositol 1,4,5-trisphosphate (IP3) (8); however, PIP2 comprises only ~1% of all phospholipid species on the plasma membrane (9). Thus, it is reasonable to hypothesize that there shall be a highly effective mechanism to regenerate PIP2 to compensate for its consumption by PLC-γ2 in B cell activation (10); however, our understanding of such a PIP2 regeneration system is limited. Specifically, we do not know the dynamics of the spatial-temporal distribution of PIP2 within the B cell immunological synapse. It is not clear how such dynamics shall be coupled to the PIP2 regeneration system for the efficient initiation of B cell activation.

Here, by combining total internal reflection fluorescence microscopy (TIRFM)–supported live cell imaging methods with a variety of newly developed lipid biosensors (1113), we dissected the dynamics, regulation, and function of PIP2 within the immunological synapse during B cell activation. PIP2 was enriched outside the BCR microclusters but was efficiently depleted inside the BCR microclusters upon antigen binding. By purposely manipulating the density of PIP2, we showed that both PIP2 depletion inside and synthesis outside the BCR microclusters were essential for the sustained initiation of B cell activation. These results demonstrated that BCR signaling can be modulated by the alteration of the subcellular localization of PIP2, revealing the importance of a PIP2-derived amplification loop for the sustained initiation of B cell activation.

RESULTS

PIP2 is enriched outside the BCR microclusters upon B cell activation

We first visualized the change in the amount of PIP2 within the B cell immunological synapse. We incubated wild-type DT40 (DT40-WT) B cells expressing a PIP2 biosensor, green fluorescent protein (GFP) fused to the pleckstrin homology domain of PLC-δ (GFP–PLC-δ–PH) (Fig. 1A) (11), on glass-supported planar lipid bilayers (PLBs) presenting anti-chicken IgM antibodies as surrogate antigens. DT40-WT B cells showed the spreading response toward the antigen-containing interface (fig. S1) (1, 4). Further quantification showed that the mean fluorescence intensity (mFI), the total FI (tFI) of the PIP2 biosensor at the bottom plane, and the ratio of the PIP2 biosensor mFI at the bottom plane to that at the equatorial plane were significantly increased after B cell activation (fig. S1, D to G). These data indicated that the level of PIP2 was enriched within the immunological synapse after B cell activation. Next, we performed TIRFM imaging to examine the spatial distribution of the PIP2 and BCR microclusters within the immunological synapse. The results showed that PIP2 was mainly enriched outside the BCR microclusters as supported by four types of distinct mathematical quantifications, as detailed in Materials and Methods (Fig. 1, B to F).

Fig. 1 PIP2 is depleted inside but enriched outside the BCR microclusters.

(A) Schematic of a PIP2 biosensor. (B to F) Distribution of BCR and PIP2 in the immunological synapse of DT40-WT B cells. (B) TIRFM images; the images in the boxed areas (4.5 μm × 4.5 μm) are magnified (right). Scale bars, 2 μm. (C) FI profiles of BCR and GFP–PLC-δ–PH (or GFP control) along the white lines in (B). (D) Pixel FI of BCR and GFP–PLC-δ–PH (or GFP control) from the boxed areas in (B). (E) FI ratio of GFP–PLC-δ–PH (or GFP control) outside to that inside the microclusters. Each dot represents one microcluster (n = 29). (F) Pearson’s correlation index (PCI) between BCR and GFP–PLC-δ–PH (or GFP control). Each dot represents one cell (n = 20 to 25). Bar represents mean ± SD. (G and H) Merged binary mask of BCR and the PALM image of PIP2 (mEos3.2–PLC-δ–PH, left) or the control mEos3.2–PLC-δ–PH–Mut (right) in the immunological synapse of DT40-WT B cells (3 μm × 3 μm regions) (G). Scale bars, 500 nm. Enrichment of PIP2 within the BCR microclusters (H). Each dot represents a 3 μm × 3 μm region (n = 26 to 29). Bar represents mean ± SD. (I) Two-color time-lapse TIRFM images of BCR and PIP2 [red fluorescent protein (RFP)–PLC-δ–PH] in the immunological synapse of DT40-WT B cells. Images are pseudo-colored (left). Boxed areas (3 μm × 3 μm) are magnified in time sequence. BCR microcluster regions are denoted by white circles in BCR images and by the corresponding black circles in PIP2 images. FI profiles of BCR and PIP2 on the white line in the TIRFM images are given (upper right). Correlated pixel FI plots of BCR and PIP2 are also given (lower right). Scale bar, 2 μm. (J to M) (J) Representative time sequence images (first two rows) of BCR and PIP2 (RFP–PLC-δ–PH) in primary B1-8 primary B cells before and after photoactivation of the caged-NP antigen. The images in the boxed areas (3 μm × 3 μm) are magnified and merged, as shown in the third row (BCR, red; PIP2, green). The correlated pixel FI of BCR and RFP–PLC-δ–PH in the magnified regions are also shown (bottom row). Normalized mFI of BCR (K) and RFP–PLC-δ–PH (L) along time (n = 13 to 14). Scale bar, 2 μm. (M) PCI between BCR and PIP2 along time (n = 11). Bar represents mean ± SEM. a.u., arbitrary units. ns, not significant; ***P < 0.001 in two-tailed t tests. Data are representative of two or three independent experiments.

We also performed super-resolution photoactivated localization microscopy (PALM) imaging of PIP2 by fusing a photoconvertible fluorescent protein mEos3.2 to the PIP2 probe, PLC-δ–PH (Fig. 1G) (14, 15). The PIP2 PALM imaging confirmed that PIP2 was mainly enriched outside the BCR microclusters (Fig. 1G). In contrast, mEos3.2 fused to the mutant PLC-δ–PH–R40L (PLC-δ–PH–Mut) that was deficient in PIP2 binding (16) was equally distributed inside and outside the BCR microclusters (Fig. 1, G and H).

Next, we examined the temporal dynamics of the spatial distribution of PIP2 and BCR microcluster by time-lapse TIRFM imaging. The results showed that PIP2 was enriched in the BCR microcluster region during the initial phase of B cell activation (Fig. 1I). However, during BCR microcluster growth, the density of PIP2 significantly decreased inside the BCR microcluster (Fig. 1I and movie S1). These observations were also validated by using primary splenic B cells from IgHB1-8/B1-8/Igκ−/− transgenic mice expressing the 4-hydroxy-3-nitrophenyl acetyl (NP)–specific μ-B1-8 BCR (B1-8 primary B cells) coupled with a photoactivatable caged-NP antigen system (Fig. 1, J to M, and movie S2) (17).

All of the above experiments were performed by using a PIP2 biosensor, which may affect the localization and function of PIP2. We thus used a PIP2-specific monoclonal antibody to stain PIP2 and confirmed that PIP2 was poorly localized inside and was highly distributed outside the BCR microclusters after B cell activation (Fig. 2, A and B). As a negative control, we showed that both DiO, a general membrane probe, and Lact-C2-GFP (18), a phosphatidylserine (PS) biosensor, were evenly distributed inside or outside the BCR microclusters (Fig. 2, A to C). All of these data indicated that PIP2 is enriched outside the BCR microclusters upon B cell activation.

Fig. 2 Localized hydrolysis of PIP2 by PLC-γ2 is mainly responsible for the PIP2 depletion inside the BCR microclusters.

(A and B) Representative TIRFM images of DiO, PIP2, and BCR on the contact interface in DT40-WT B cells on nonstimulating (A) or antigen-coated (B) coverslips. Cells were stained with DiO and anti-PIP2 antibodies. Correlated pixel FI plot of BCR and DiO or PIP2 (right) in the boxed area (4.5 μm × 4.5 μm) in the left images. Scale bars, 2 μm. (C) Representative TIRFM images of BCR and Lact-C2-GFP on the nonstimulating or antigen-coated contact interfaces in DT40-WT B cells. Correlated pixel FI plot of BCR and PS (right) in the boxed area (4.5 μm × 4.5 μm) in the left images. Scale bars, 2 μm. (D to F) Distribution of BCR and PIP2 in the immunological synapse of DT40–PLC-γ2–KO B cells. (D) TIRFM images; the images in the boxed areas (4.5 μm × 4.5 μm) are magnified (lower right). Scale bar, 2 μm. (E) FI profiles of BCR and GFP–PLC-δ–PH along the white lines in (D). (F) Pixel FI of BCR and GFP–PLC-δ–PH from the boxed areas in (D). (G) Enrichment of PIP2 (mEos3.2–PLC-δ–PH, from PALM imaging) within the BCR microclusters. Each dot represents a 3 μm × 3 μm region (n = 32 to 34). Bar represents mean ± SD. (H) Two-color time-lapse TIRFM images of BCR and PIP2 (RFP–PLC-δ–PH) in the immunological synapse of DT40–PLC-γ2–KO B cells. Images are pseudo-colored (left). Boxed areas (3 μm × 3 μm) are magnified in time sequence. BCR microcluster regions are denoted by white circles in BCR images and by the corresponding black circles in PIP2 images. FI profiles of BCR and PIP2 on the white line in the TIRFM images are given (upper right). Correlated pixel FI plots of BCR and PIP2 (lower right) are also given. Scale bar, 2 μm. (I and J) (I) TIRFM images of PIP2 stained with anti-PIP2 antibodies within the immunological synapse of either DT40-WT or DT40–PLC-γ2–KO B cells. Scale bars, 2 μm. (J) Quantification of mFI of PIP2 within the B cell immunological synapse (n = 40 to 60 cells). Bar represents mean ± SEM. r, correlation. **P < 0.01, ***P < 0.001 in two-tailed t tests. Data are representative of two or three independent experiments.

PIP2 hydrolysis by PLC-γ2 is responsible for PIP2 depletion inside the BCR microclusters

PLC-γ2 is recruited to the BCR microcluster to digest PIP2 into DAG and IP3 (19, 20). Within the immunological synapse of DT40 B cells knocked out for PLC-γ2 (DT40–PLC-γ2–KO) (21), BCR and PIP2 showed a high degree of colocalization (Fig. 2, D to H, and movie S3), suggesting a failed depletion of the PIP2 within the BCR microclusters in the absence of PLC-γ2. However, the mFI of PIP2 within the immunological synapse of DT40–PLC-γ2–KO B cells did not increase after activation (Fig. 2, I and J), and this can be rescued by the exogenously expressed PLC-γ2–WT, but not the lipase-dead mutant PLC-γ2–LD (Fig. 2, I and J) (22). All of these results suggested that the localized hydrolysis of PIP2 by PLC-γ2 is responsible for the PIP2 depletion inside the BCR microclusters. These results also indicated that PIP2 is enriched within the B cell immunological synapse in a PIP2 hydrolysis–dependent manner.

PIP2 synthesis by PIP5K is responsible for PIP2 enrichment outside the BCR microclusters

Type I phosphatidylinositol 4-phosphate 5-kinase (PIP5K) phosphorylates phosphatidylinositol 4-phosphate (PI4P) at the fifth position of the inositol ring to convert PI4P into PIP2 (23). Because all three type I PIP5K family members, PIP5Kα, PIP5Kβ, and PIP5Kγ, are expressed in DT40 B cells (fig. S2A and table S1), we first examined their recruitment within the B cell immunological synapse and found that PIP5Kα and PIP5Kγ, but not PIP5Kβ, were potently polarized and drastically enriched within the B cell immunological synapse (fig. S2B). Thus, we knocked out PIP5Kα or PIP5Kγ or both of them in DT40-WT B cells (fig. S2, C and D). These KO cells showed no changes in the amount of PIP2 at the quiescent state (Fig. 3, A and B). In contrast, PIP2 regeneration upon BCR activation was decreased by ~60% in DT40-PIP5Kα-KO B cells in comparison with DT40-WT B cells, but was not affected in the DT40-PIP5Kγ-KO B cells (Fig. 3, A and B). In DT40 B cells knocked out for both PIP5Kα and PIP5Kγ (DT40-PIP5Kα/γ-DKO), the PIP2 regeneration upon BCR activation was almost completely blocked (Fig. 3, A and B). These results suggested that both PIP5Kα and PIP5Kγ are responsible for the rapid PIP2 regeneration upon BCR activation, although PIP5Kα plays a more important role than PIP5Kγ.

Fig. 3 Localized synthesis of PIP2 by PIP5K is mainly responsible for the PIP2 enrichment outside the BCR microclusters.

(A and B) TIRFM images of PIP2 in DT40-WT, DT40-PIP5Kα-KO, DT40-PIP5Kγ-KO, or DT40-PIP5Kα/γ-DKO B cells on nonstimulating coverslips or after BCR stimulation for 3 min by PLBs containing anti-IgM surrogate antigens. (A) Representative images of the cells on the contact interface. Scale bars, 2 μm. (B) Quantification of PIP2 mFI (n = 28 to 44 cells). Bar represents mean ± SEM. (C to G) Distribution of BCR and PIP2 in the immunological synapse of DT40-PIP5Kα/γ-DKO B cells. (C) TIRFM images; boxed areas (4.5 μm × 4.5 μm) are magnified (lower right). (D) FI profiles of BCR and PIP2 along the white lines in (C). Scale bar, 2 μm. (E) Correlated pixel FI plot of BCR and PIP2 from the boxed areas in (C). (F) FI ratio of PIP2 outside to that inside the microclusters. Each dot represents one microcluster (n = 21 to 25). Bar represents mean ± SD. (G) PCI between BCR and PIP2. Each dot represents one cell (n = 20). Bar represents mean ± SD. (H) Two-color time-lapse TIRFM images of BCR and PIP2 (RFP–PLC-δ–PH) in the immunological synapse of DT40-PIP5Kα/γ-DKO B cells. Images are pseudo-colored (top row). Boxed areas (4.5 μm × 4.5 μm) in the leftmost TIRFM images are magnified and provided as time-lapse images. BCR microclusters are denoted by white circles in BCR images and by the corresponding black circles in PIP2 images. FI profiles of BCR and PIP2 on the white line in the TIRFM images are given (upper right). Correlated pixel FI plots of BCR and PIP2 are also given (lower right). Scale bar, 2 μm. (I to L) Distribution of BCR and GFP-PIP5Kα in the immunological synapse of DT40-WT B cells. (I) TIRFM images; boxed areas (4.5 μm × 4.5 μm) are magnified (lower right). Scale bar, 2 μm. (J) FI profiles of BCR and GFP-PIP5Kα along the white lines in (I). (K) Correlated pixel FI plot of BCR and GFP-PIP5Kα from the boxed areas in (I). (L) PCI between BCR and GFP-PIP5Kα or GFP control (n = 21 to 31 cells). Bar represents mean ± SEM. **P < 0.01, ***P < 0.001 in two-tailed t tests. Data are representative of two or three independent experiments.

Furthermore, PIP2 and BCR exhibited a much higher extent of codistribution after B cell activation (Fig. 3, C to H, and movie S4), suggesting the lack of the sharp spatial-temporal PIP2 density gradient in the absence of PIP5Kα and PIP5Kγ. To determine whether the accumulated PIP2 outside the BCR microclusters was caused by the localized synthesis of PIP2, we imaged the localization of PIP5Kα in the B cell immunological synapse. PIP5Kα was localized outside the BCR microclusters (Fig. 3, I to L). All of these results suggested that strategic positioning of these enzymes could control the localization of PIP2 in activated B cells.

Phosphatidic acid mediated the localization of PIP5Kα to the B cell immunological synapse

Given that blocking PIP2 hydrolysis impaired the PIP2 enrichment in the B cell immunological synapse (Fig. 2, I and J), we proposed a potential loop of PIP2 metabolism between the inside and the outside of BCR microclusters: The product of PIP2 hydrolysis, DAG, can be converted by diacylglycerol kinase (DGK) to phosphatidic acid (PA), a potent activator of PIP5K that not only recruits PIP5K to the membrane but also significantly enhances its enzymatic activity for PIP2 regeneration (Fig. 4A) (24, 25). There are potent DAG and PA production on the plasma membrane in response to antigen stimulation (fig. S3A). In addition, we showed that the potential weak binding of the PA biosensor to PS did not affect the utilization of the PA biosensor to quantify the changes of PA on the plasma membrane (fig. S3A, right).

Fig. 4 DAG-derived PA mediates the recruitment and localization of PIP5Kα to the B cell immunological synapse.

(A) Schematic of the potential PIP2 synthesis–feedback loop from the product of PIP2 hydrolysis DAG; DAG is converted into PA catalyzed by DGK and then PA activates PIP5K to regenerate PIP2. (B to D) Quantification of DGKζ in either DT40-WT or DT40–PLC-γ2–KO B cells. (B) Representative confocal images of equatorial and bottom planes of the DT40-WT and DT40–PLC-γ2–KO cells expressing GFP-DGKζ on fibronectin-coated coverslips (0 min) and PLBs tethering anti-IgM surrogate antigens. Scale bars, 2 μm. (C) Ratio of the mFI of GFP-DGKζ (plasma membrane localization to cytosol localization) in DT40-WT and DT40–PLC-γ2–KO B cells, or in DT40–PLC-γ2–KO B cells with the exogenous expression of PLC-γ2–WT, PLC-γ2–LD, or PLC-γ2–R564A (n = 28 to 52 cells). PM, plasma membrane. (D) Ratio of the mFI of GFP-DGKζ (bottom localization to equatorial localization; n = 28 to 52 cells). Bar represents mean ± SEM. (E to H) (E) TIRFM images of GFP-PIP5Kα and PA (RFP-Spo20) in DT40-WT B cells on PLBs tethering anti-IgM surrogate antigens. Boxed areas (4.5 μm × 4.5 μm) are magnified (right). Scale bar, 2 μm. (F) Correlated pixel FI plot of GFP-PIP5Kα and PA (top) and BCR and PA (bottom) from the boxed areas in (E). (G) FI profiles of GFP-PIP5Kα, PA, and BCR along the white lines in (E). (H) PCI between BCR and PA, or between GFP-PIP5Kα and PA (n = 28 to 40 cells). Bar represents mean ± SEM. (I and J) Quantification of immunological synapse recruitment of PIP5Kα in DT40-WT or DT40-DGKζ-KO B cells. (I) Representative images of the equatorial and bottom planes of the DT40-WT or DT40-DGKζ-KO B cells expressing GFP-PIP5Kα on fibronectin-coated coverslips (0 min) and PLBs tethering anti-IgM surrogate antigens. Scale bars, 2 μm. (J) Ratio of mFI of GFP-PIP5Kα (bottom localization to equatorial localization; n = 18 to 26 cells). Bar represents mean ± SEM. (K to N) (K) TIRFM images of BCR and GFP-PIP5Kα in DT40-DGKζ-KO B cells on PLBs tethering anti-IgM surrogate antigens. Scale bar, 2 μm. (L) FI profiles of BCR and GFP-PIP5Kα along the white line in (K). Correlated pixel FI plot (M) of BCR and GFP-PIP5Kα from the magnified areas in (K). (N) PCI between BCR and PIP5Kα at indicated time points (n = 16 to 27 cells). Bar represents mean ± SEM. (O and P) (O) TIRFM images of PIP2 stained with anti-PIP2 antibodies within the immunological synapse of DT40-WT or DT40-DGKζ-KO B cells. Scale bars, 2 μm. (P) Quantification of mFI of PIP2 within the immunological synapse (n = 30 to 46 cells). Bar represents mean ± SEM. (Q to T) DT40-DGKζ-KO B cells expressing GFP–PLC-δ–PH on PLBs tethering anti-IgM surrogate antigens at 3 min. (Q) Representative TIRFM images. Images in the boxed areas (4.5 μm × 4.5 μm) are magnified. Scale bar, 2 μm. (R) FI profiles of BCR and PIP2 along the white line in (Q). The r value between two signals is given. (S) Correlated pixel FI plot of BCR and PIP2 in the boxed area in (Q). (T) PCI between BCR and PIP2. Each dot represents one cell (n = 15 cells). Bar represents mean ± SD. **P < 0.01, ***P < 0.001 in two-tailed t tests. Data are representative of two or three independent experiments.

A variety of DGK members are expressed in DT40 B cells, including DGKδ, ε, θ, and ζ (fig. S3B). We knocked out DGKδ, ε, θ, and ζ (fig. S3, C to J). In these four types of DGK-KO B cells, only DT40-DGKζ-KO B cells induced more DAG accumulation than DT40-WT B cells, in response to antigen simulation (fig. S4, A and B). These data indicated the critical role of DGKζ in the conversion of DAG to PA after B cell activation. PA generation was significantly impaired in DT40-DGKζ-KO B cells, and this effect could be rescued by the expression of an exogenous DGKζ-WT but not by a kinase-dead mutant (DGKζ-KD) (fig. S4, C and D) (26). These results suggested that BCR signaling–induced PA generation is dependent on DGKζ.

Because of the key function of DGKζ in PA generation, we further examined the spatial-temporal dynamics of DGKζ within the B cells in B cell activation. Time-lapse imaging revealed the efficient translocation of DGKζ from the cytosol to the plasma membrane in response to antigen stimulation [Fig. 4, B (left), C, and D]. Next, we examined the function of DAG in mediating the plasma membrane translocation and synaptic recruitment of DGKζ (27) using DT40–PLC-γ2–KO B cells, which displayed a severely impaired production of DAG in comparison with DT40-WT B cells (fig. S4, E and F). The BCR-induced plasma membrane translocation and synaptic recruitment of DGKζ was impaired in DT40–PLC-γ2–KO B cells, but not in rescued cells with exogenously expressed PLC-γ2–WT [Fig. 4, B (right), C, and D, and fig. S4G]. Furthermore, the subsequent PA production was also severely impaired in DT40–PLC-γ2–KO B cells (fig. S4, H and I). These results demonstrated the crucial importance of DAG in maintaining the correct spatial dynamics of DGKζ. To further validate this conclusion, we visualized the plasma membrane localization of either DGKζ-WT or DGKζ-KD mutant in DT40-DGKζ-KO B cells. The results confirmed that the amount of DAG on the plasma membrane highly determined the efficiency of the plasma membrane localization of DGKζ (fig. S4, J and K).

To directly assess whether the function of PIP5Kα is affected by the spatial-temporal dynamics of PA, we visualized the synaptic localization of PA and PIP5Kα by TIRFM imaging. We found that PA and PIP5Kα showed high extent of codistribution within the immunological synapse of DT40-WT B cells (Fig. 4, E to H). Because PA is converted from DAG by DGKζ, it is not surprising to observe that the synaptic recruitment of PIP5Kα was impaired in DT40-DGKζ-KO B cells (Fig. 4, I and J). In addition, distinct from the observation in this report that PIP5Kα is mainly localized outside the BCR microclusters in DT40-WT B cells (Fig. 3, I to L), PIP5Kα was equally distributed inside or outside the BCR microclusters in DT40-DGKζ-KO B cells (Fig. 4, K to N, and fig. S4L). Thus, these data demonstrated that PA determined the recruitment and localization of PIP5Kα in the B cell immunological synapse.

Given the above observations, we hypothesized that PIP2 regeneration may be defective when PA production is impaired. BCR-induced PIP2 regeneration was severely impaired in the immunological synapse of DT40-DGKζ-KO B cells (Fig. 4, O and P). These results indicated the presence of a positive feedback loop for PIP2 synthesis from PIP2 hydrolysis. Furthermore, the colocalization between the BCR microclusters and PIP2 in DT40-DGKζ-KO B cells was much higher compared with that in DT40-WT B cells (Fig. 4, Q to T). All of these analyses demonstrated that PA determined PIP2 regeneration by mediating the recruitment of PIP5Kα outside the BCR microcluster.

The PIP2 synthesis feedback loop is achieved because of the fast Brownian motility of DAG

There is a crucial missing piece in this PIP2 synthesis feedback loop considering that PLC-γ2 is localized within the BCR microclusters (19), whereas PIP5Kα is localized outside the BCR microclusters (Fig. 3, I to L). It is intriguing how PIP2 metabolic products, such as DAG from PLC-γ2 and DAG-derived PA from DGKζ, can physically recruit and regulate the function of PIP5Kα at the region outside the BCR microcluster. To address this question, we first imaged the distribution of DAG and DGKζ within the B cell immunological synapse. DAG was found to be randomly distributed at the region inside and outside the BCR microclusters (Fig. 5A). In marked contrast, DGKζ was mainly localized outside the BCR microclusters (Fig. 5B). All of these results prompted us to hypothesize that DAG may have high Brownian mobility and thus can freely diffuse from the BCR microcluster region (where they are originally produced by the membrane-proximal PLC-γ2) into the region outside the BCR microclusters (where they may efficiently engage with DGKζ). In contrast, as an anionic lipid, PIP2 interacts with cationic macromolecules on the membrane (2830) and may have different fatty acid chains from DAG (31, 32), two properties that may prohibit its free diffusion within the B cell immunological synapse. To examine this hypothesis that DAG and PIP2 may exhibit distinct lateral Brownian motilities on the plasma membrane, we used the mEos3.2-tagged PIP2 biosensor PLC-δ–PH and the DAG biosensor protein kinase Cθ (PKCθ)–C1, and performed single-molecule imaging (2, 33, 34). Mathematical quantification revealed that DAG exhibited much faster Brownian mobility than PIP2 (Fig. 5, C to H). Specifically, we observed a notable ~45% increase in the median diffusion coefficient of DAG compared with PIP2 (Fig. 5, E and F).

Fig. 5 The fast Brownian motility of DAG in B cell immunological synapse.

(A and B) Representative TIRFM images of DT40-WT B cells expressing PKCθ-C1–GFP (A) or GFP-DGKζ (B) on PLBs containing anti-IgM surrogate antigens (left). FI profiles of BCR and DAG or GFP-DGKζ (middle) along the white line in the left images. Correlated pixel FI plot of BCR and DAG or GFP-DGKζ (right) in the boxed area in the left images. Scale bars, 2 μm. (C to H) Single-molecule tracking of DAG (PKCθ-C1–mEos3.2) or PIP2 (mEos3.2–PLC-δ–PH) in DT40-WT B cells on PLBs tethering anti-IgM surrogate antigens. (C) Trajectories were pseudo-colored according to the value of the diffusion coefficients (n is given above the figures). (D) Mean square displacement (MSD) plots. Bar represents mean ± SEM. (E) Diffusion coefficients with the median indicated by black bars. (F) Cumulative probability of diffusion coefficients of DAG and PIP2. (G) The probability distribution of diffusion coefficients and fitting to two populations showing slow (red line) and fast (green line) diffusion. The mean diffusion coefficients of fast and slow population are given above the plots. (H) The percentage of the slow population in DAG or PIP2. (I to K) (I) DAG or PIP2 trajectories are randomly colored and are plotted on the background of BCR microclusters. The right panels are the magnified regions from the boxed areas (left) with time series. Scale bars, 2 μm. (J) The percentage of trajectories crossing the BCR microcluster borders. (K) The percentage of the trajectories showing the direction from inside to outside the BCR microclusters. n = 6 cells. Bar represents mean ± SEM. **P < 0.01, ***P < 0.001 in two-tailed t tests. Data are representative of three independent experiments.

Furthermore, we plotted the trajectories of DAG and PIP2 on the BCR microclusters as background. All trajectories can thus be sorted into three types: those always localized within the BCR microclusters, those always localized outside the BCR microclusters, and those crossing the border of the BCR microclusters (Fig. 5I and movie S5). Statistical quantification showed that the percentage of crossing-border trajectories was 66 ± 5% in DAG, which was drastically higher than that in PIP2 (32 ± 5%) (Fig. 5J). We next sorted the crossing-border behaviors of either DAG or PIP2 single molecules according to the following directions: either from the inside zone to the outside zone of the BCR microclusters or from the outside zone to the inside zone of the BCR microclusters. The percentage of inside-to-outside diffusing DAG was 61 ± 3% and was much higher than that of PIP2 (38 ± 2%) (Fig. 5K), indicating that DAG is more inclined to diffuse from the inside zone to the outside zone than PIP2. All of these results showed that this PIP2 synthesis–feedback loop is likely achieved by the fast Brownian motility of DAG.

PIP2 depletion inside the BCR microclusters is essential for B cell activation

Our data have demonstrated the existence of a PIP2 density gradient mediated by the PIP2 synthesis–feedback loop: low PIP2 density within the BCR microclusters and high density outside the BCR microclusters. We next determined whether the PIP2 gradient is necessary for the initiation of B cell activation. We first designed a PIP2 manipulation system that allows us to change the density of PIP2 inside the BCR microclusters, as illustrated in Fig. 6A. PLC-γ2–LD fused to either Ins54p or a mutant version of PIP5K (donated as PIP5Km) (11) will locate within the BCR microclusters after B cell activation (via PLC-γ2–LD) and will catalyze the dephosphorylation or synthesis of PIP2 (via Ins54p or PIP5Km, respectively) (11, 19, 22). As a proof of principle, it was evident that these two types of fusion proteins were efficiently translocated from the cytosol to the BCR microclusters after B cell activation (fig. S5, A to C). TIRFM imaging showed that DT40-WT B cells expressing PLC-γ2–LD–PIP5Km were more severely impaired in the formation of BCR microclusters than the cells expressing PLC-γ2–LD fused with a kinase-dead mutant of PIP5Km (PLC-γ2–LD–PIP5Km–KD) (Fig. 6, B to D). In contrast, the expression of PLC-γ2–LD–Ins54p did not reduce the number of BCR microclusters, and instead, the intensity of the BCR microclusters was enhanced compared with cells expressing PLC-γ2–LD fused with a phosphatase-dead mutant of Ins54p (Ins54p-D281A) (Fig. 6, B to D). These data demonstrated that the failure of sustaining low PIP2 density inside the BCR microclusters impaired the formation of BCR microclusters, whereas PIP2 depletion within the BCR microclusters facilitated such formation.

Fig. 6 Both PIP2 depletion inside and synthesis outside the BCR microclusters are essential for the initiation of B cell activation.

(A) Schematic presentation for the targeted PIP2 manipulation system inside the BCR microclusters. PLC-γ2–LD fused to Ins54p (or PIP5Km) is recruited to early BCR microclusters after antigen recognition. (B to D) Quantification of antigen microclusters in DT40-WT B cells and DT40-WT B cells with the expression of the indicated type of plasmids, as shown in (C). (B) TIRFM images of antigen microclusters in DT40-WT B cells with the expression of PLC-γ2–LD–Ins54p or PLC-γ2–LD–PIPKm. Scale bars, 2 μm. (C) Number of microcluster in DT40-WT B cells with the expression of the indicated type of plasmids (n = 29 to 39 cells). Bar represents mean ± SD. (D) Integrated FI of the antigen microclusters in cells as in (C) (n > 500 microclusters). Bar represents mean ± SEM. (E and F) (E) Number of antigen microclusters in DT40–PLC-γ2–KO B cells with the exogenous expression of PLC-γ2–LD–Ins54p or PLC-γ2–LD–Ins54p-D281A with or without the addition of PA micelles (n = 32 to 50 cells). Bar represents mean ± SEM. (F) Integrated FI of microclusters in cells as in (E) (n > 500 microclusters). Bar represents mean ± SEM. (G) Schematic representation of the targeted PIP2 manipulation system outside the BCR microclusters. Before rapamycin addition, the FKBP fused to Ins54p (or PIP5Km) will localize to the cytosol. Instead, the FKBP fused to Ins54p (or PIP5Km) will be recruited to the CD45-(ET)-FRB after the addition of rapamycin. (H to J) (H) Representative TIRFM images of DT40-WT B cells on PLBs tethering anti-IgM surrogate antigens with the expression of the indicated type of plasmids in the CD45-targeted PIP2 manipulation system. Scale bars, 2 μm. (I) Number of BCR microclusters in the B cells expressing the indicated type of plasmids in the CD45-targeted PIP2 manipulation system. Each dot represents one cell (n = 19 to 35 cells). Bar represents mean ± SD. (J) Integrated FI of the BCR microclusters in the cells as in I (n > 216 microclusters). Bar represents mean ± SEM. LD, PLC-γ2–LD; Rap, rapamycin. *P < 0.05, **P < 0.01, ***P < 0.001 in two-tailed t tests. Data are representative of at least two independent experiments.

In the above experiments, we noticed that the expression of PLC-γ2–LD fusion protein impaired the formation of the BCR microclusters in comparison with the DT40-WT control B cells (Fig. 6, B to D). This observation was further supported by the experiments using PLC-γ2–LD–R564A–Ins54p (Fig. 6D), which lacks Blnk binding ability and thus cannot be targeted to the BCR microcluster (35). We thus reconstitute the PLC-γ2–LD–derived targeting system in DT40–PLC-γ2–KO B cells by the addition of exogenous PA micelles to compensate for its lack of PA generation (figs. S4, H and I, and S5, D and E). PLC-γ2–LD–Ins54p more efficiently restored BCR microcluster formation than PLC-γ2–LD–Ins54p–D281A (Fig. 6, E and F). Thus, PIP2 depletion inside the BCR microclusters is essential for B cell activation.

PIP2 synthesis outside the BCR microclusters is essential for B cell activation

We also developed a system that can manipulate the density of PIP2 outside the BCR microclusters. CD45 is a good candidate docking molecule for localization outside the BCR microclusters (36, 37). However, if we directly fused PIP2-synthesizing or -hydrolyzing enzymes to the intracellular domain of CD45, these enzymes will be directly localized to the plasma membrane and thus constitutively increase or decrease PIP2 density. To solve this problem, we used a rapamycin-supported inducible system: Rapamycin can simultaneously bind to FK506 binding protein (FKBP) and FKBP-rapamycin binding (FRB) domains to form a ternary complex (Fig. 6G) (11). As a proof of principle, we found that CD45-(extracellular and transmembrane, ET)-FRB and PIP2 showed high extent of colocalization and both showed low codistribution with the BCR microclusters (fig. S5, F and G). The addition of rapamycin drastically induced PIP2 depletion only in FKBP-Ins54p– and CD45-(ET)-FRB–coexpressing 293T cells but not in FKBP-Ins54p-D281A– and CD45-(ET)-FRB–coexpressing 293T cells (fig. S5, H and I). Thus, a rapamycin-inducible CD45-(ET)-FRB–targeted PIP2 manipulation system was successfully reconstituted.

We placed the DT40-WT B cells that were reconstituted with the CD45-(ET)-FRB–targeted PIP2 manipulation system on antigen-coated PLBs at time 0 min. In control group 1, cells were fixed at 3 min. In control group 2, dimethyl sulfoxide (DMSO) was added at 3 min and the cells were fixed at 7 min. In the experimental group, rapamycin was added at 3 min and the cells were fixed at 7 min. TIRFM imaging showed that only weak FIs of these exogenous enzymes (Ins54p, PIP5Km, or inactive versions) were detected without rapamycin addition (Fig. 6H). In the presence of rapamycin, the distribution of FKBP-fused enzymes was similar to that of CD45-(ET)-FRB (Fig. 6H), representing the dimerization of FRB and FKBP upon rapamycin addition. Further analyses showed that the groups lacking rapamycin exhibited comparable amounts of BCR microclusters and BCR microcluster FIs (Fig. 6, H to J). B cells with CD45-targeted localization of Ins54p showed drastically reduced numbers of BCR microclusters and decreased FI of each BCR microcluster in comparison with the cells with the enzyme-dead version, Ins54p-D281A (Fig. 6, H to J). Conversely, B cells with CD45-targeted PIP2 synthesis showed enhanced activation compared with the B cells expressing the enzyme-dead version, PIP5Km-KD (Fig. 6, H to J). These results suggested that PIP2 synthesis outside the BCR microclusters is essential for B cell activation.

The PIP2-derived amplification loop accounts for the enhanced activation of B cells expressing disease-associated PLC-γ2 mutants at low temperature

Last, we examined the clinical implications of this PIP2-derived amplification loop. Patients with PLC-γ2–associated antibody deficiency and immune dysregulation (PLAID) disease exhibit immune dysregulation and cold-induced urticaria. B cells expressing PLC-γ2 mutants from PLAID patients exhibit much higher BCR activation compared with healthy controls (38). These phenomena only occur at low temperatures (20° to 30°C) but not at physiological temperature (37°C) (38, 39). Because this report showed that the PIP2-derived amplification loop is required for highly efficient activation of B cells, we investigated the contribution of PIP2-derived amplification loop to the phenotype of PLAID-associated PLC-γ2 mutants. We first examined whether the amount of PIP2 on the plasma membrane would be affected after low-temperature treatment in quiescent DT40–PLC-γ2–KO B cells with the exogenous expression of PLAID-associated PLC-γ2 mutants with either exon 19 deleted (Δ19-expressing B cells) or exons 20 to 22 deleted (Δ20-22–expressing B cells) (38). Using PIP2 intracellular staining, we observed that the amount of plasma membrane–localized PIP2 was increased by 21.6 and 20.9% in Δ19- and Δ20-22–expressing B cells, respectively, at 20°C in comparison with the situation at 37°C (Fig. 7A and fig. S6). After BCR activation, the amount of PIP2 in B cell immunological synapse was also increased at 20°C compared with 37°C in Δ19- and Δ20-22–expressing B cells, respectively (Fig. 7A). Instead, the PIP2 amount was higher at 37°C compared with 20°C in PLC-γ2–WT–expressing B cells in response to antigen stimulation (Fig. 7A). Further correlation analyses indicated better spatial separation of PIP2 and BCR microclusters within the immunological synapse of Δ19-expressing B cells at 20°C compared with the situation at 37°C (Fig. 7B). The hyperenzyme activity exhibited by the Δ19 mutant at 20°C accelerated PIP2 hydrolysis, leading to the significantly increased PIP2 density outside the BCR microclusters (39). The number of BCR microclusters (Fig. 7C) and integrated microcluster intensity (Fig. 7D) were significantly higher in Δ19-expressing B cells compared with PLC-γ2–WT–expressing B cells at 20°C. Similar observations were also acquired when comparing the activation of Δ20-22– and PLC-γ2–WT–expressing B cells (Fig. 7, C and D).

Fig. 7 The PIP2 metabolism–derived amplification loop accounts for the enhanced activation of B cells expressing PLAID-associated PLC-γ2 mutants at low temperature.

(A) Quantification of PIP2 mFI at the contact interface between fibronectin-coated coverslips (nonstimulatory) or PLBs tethering anti-chicken IgM surrogate antigens and DT40-WT, DT40–PLC-γ2–KO, and DT40–PLC-γ2–KO B cells with the expression of PLC-γ2-WT, PLC-γ2–Δ19, or PLC-γ2–Δ20-22, respectively. PIP2 molecules were stained with anti-PIP2 antibodies (n = 41 to 52 cells). Bar represents mean ± SEM. (B) PCI between BCR and PIP2 in DT40–PLC-γ2–KO B cells expressing PLC-γ2–Δ19 on PLBs tethering anti-chicken IgM surrogate antigens under either low or physiological temperature (n = 17 cells). Bar represents mean ± SD. (C and D) Quantification of antigen microclusters in DT40–PLC-γ2–KO B cells expressing either PLC-γ2–Δ19 or PLC-γ2–Δ20-22 under low or physiological temperature. (C) Microcluster number. Each dot represents one cell (n = 22 to 31). Bar represents mean ± SD. (D) Integrated FI of each microcluster (n > 800 microclusters). Bar represents mean ± SEM. (E to H) (E) TIRFM images of PIP2 and BCR in human primary B cells expressing PLC-γ2–WT or PLC-γ2–Δ19 under low or physiological temperature. Correlated pixel FI plot of BCR and PIP2 from the magnified areas (3 μm × 3 μm) (right). Scale bars, 2 μm. (F) Quantification of mFI of PIP2 (n = 20 to 24 cells). Bar represents mean ± SEM. (G) Microcluster number. Each dot represents one cell (n = 25 to 38). Bar represents mean ± SD. (H) Integrated FI of each microcluster (n > 166 microclusters). Bar represents mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001 in two-tailed t tests. Data are representative of at least two independent experiments.

We validated these observations using human primary B cells from peripheral bloods of healthy donors. TIRFM imaging and correlation analysis showed that PIP2 was mainly enriched outside the BCR microclusters at 37°C within the immunological synapse of human primary B cells expressing PLC-γ2–WT. In contrast, PIP2 was uniformly distributed inside and outside the BCR microclusters at 20°C (Fig. 7E). However, in Δ19-expressing human primary B cells, PIP2 was mainly enriched outside the BCR microclusters at 20°C compared with the situation at 37°C (Fig. 7E). Moreover, Δ19-expressing primary B cells exhibited 29.9% more PIP2 at 20°C in comparison with that at 37°C (Fig. 7F). A similar conclusion was reached when examining the formation of BCR microclusters (Fig. 7, G and H). Thus, the PIP2-derived amplification loop accounts for the enhanced activation of B cells expressing PLAID-associated PLC-γ2 mutants at low temperature.

DISCUSSION

Here, we reported the identification of a PIP2-derived signaling amplification loop for the initiation of B cell activation (fig. S7). Specifically, we observed that there is a highly dynamic spatial-temporal change of PIP2 within the immunological synapse during B cell activation: PIP2 is efficiently depleted inside the BCR microclusters but is regenerated outside the BCR microclusters. Both events are important for the sustained initiation of B cell activation. Mechanistically, the hydrolysis of PIP2 inside the BCR microclusters induced a positive feedback mechanism for its synthesis outside the BCR microclusters.

The positive feedback nature of the PIP2-derived amplification loop is achieved by the unique Brownian mobility of PIP2 metabolic products. DAG, the product of PIP2 hydrolysis within the BCR microclusters, exhibits high Brownian mobility, which ensures its efficient interaction with DGKζ outside the BCR microclusters. PA, converted from DAG by DGKζ, drastically facilitates the localization of PIP5Kα outside the BCR microclusters and, in turn, promotes the localized PIP2 generation. Different from DAG, the mobility of PIP2 is more confined and mainly localized outside the BCR microclusters. The difference in chemical structure between DAG and PIP2 may explain why these two lipid molecules show different Brownian mobility features. PLC-derived DAG may have different fatty acid chains with the PIP5K-derived PIP2 (31, 32), which may cause the difference in diffusion (40, 41). Moreover, DAG lacks the inositol trisphosphate head group present in PIP2. The neutrality of DAG may decrease its interaction with other charged molecules, consequently allowing for faster and free mobility within the heterogeneous B cell immunological synapse.

The positive feedback nature of the PIP2-derived amplification loop relies on the two sources of PIP2. The first source originates from the basal and low level of PIP2 on the plasma membrane, which is hydrolyzed by PLC-γ2 at early stages of B cell activation. This allows for the initial formation of the BCR microclusters. The second source of PIP2 originates from the regenerated PIP2 by the DAG-PA-DGKζ-PIP5Kα/γ module outside the BCR microclusters, which further promotes the formation of BCR microclusters.

The PIP2-derived amplification loop may be affected in specific diseases. Two PLC-γ2 mutants identified from PLAID patients, Δ19 and Δ20-22, exhibit low temperature (20° to 30°C)–induced hyperactivation (38, 39). In this report, we found that low temperature (20°C) induced increased PIP2 generation, promoting enhanced BCR microclustering in B cells expressing the Δ19 or Δ20-22 mutants. Therefore, we predict that the increased amount of PIP2 produced by the enhanced activity of mutant PLC-γ2 may induce a hyper-feedback loop for BCR microcluster formation, resulting in the abnormal activation of BCR signaling under low temperature.

Note that our studies have some limitations: First, most of our studies were performed in the B cell lines. Although the core experiments were repeated in primary B cells, knockout mice model experiments are needed in future studies. Second, we have used PKCθ-C1 and PLC-δ–PH biosensors to probe DAG and PIP2 molecules, respectively, in the imaging experiments. Both DAG and PIP2 lipids that were detected by these two biosensors shall be precisely termed free lipids (6, 29, 30) in comparison with those already bound by the endogenous binding proteins and thus are inaccessible to PKCθ-C1 and PLC-δ–PH biosensors. These free lipids are just the relevant form available for cellular function, thus reporting the potentially functional populations. Third, because PIP2 and DAG are truly small molecules, ultraresolution microscopic imaging or electron microscopy shall be used to precisely define their dynamics in future studies.

Collectively, we identified a PIP2-derived amplification loop, where PIP2 consumption and synthesis occur in concert at different microdomains within the B cell immunological synapse. In this amplification loop, the hydrolysis of PIP2 inside the BCR microclusters potently induced a positive feedback mechanism for its synthesis outside the BCR microclusters, which facilitates B cell activation. Thus, determining the molecular mechanism underlying PIP2-derived regulation will further enhance our understanding of the molecular mechanism of the activation of other types of immune receptors.

MATERIALS AND METHODS

Study design

We used TIRFM-supported live cell and molecular imaging methods combined with a variety of lipid biosensors and localization-specific PIP2 manipulation tools. We dissected the dynamics, regulation, and function of PIP2 within the immunological synapse during B cell activation. For this purpose, imaging experiments were performed by using laboratory cell lines and primary B cells on glass-supported PLBs tethering surrogate or real antigen systems. Usually, more than 20 biological replicates per group were included in each experiment. Experiments were repeated once or twice. Sample sizes were based on power analysis or on our previous experience. All data points were included, and no outliers were excluded. All end points were prospectively selected. It was not possible to blind the study because of the need to identify the control and the experimental group. The investigator who analyzed the imaging data was not blinded, but randomly chose cells in each group. In this study, TIRFM, confocal microscopy, PALM imaging and the corresponding image process, and quantitative analysis methods were used and detailed in Supplementary Materials and Methods.

SUPPLEMENTARY MATERIALS

immunology.sciencemag.org/cgi/content/full/2/17/eaan0787/DC1

Materials and Methods

Fig. S1. PIP2 is enriched in the immunological synapse upon B cell activation.

Fig. S2. Expression of PIP5Kα, PIP5Kβ, and PIP5Kγ in DT40 B cells.

Fig. S3. DGKζ is essential for converting DAG into PA after B cell activation.

Fig. S4. BCR signaling–induced PA generation is converted from the PLC-γ2–derived DAG by DGKζ.

Fig. S5. The validation of the PIP2 density manipulation system.

Fig. S6. Cold treatment induces the regeneration of excessive amount of PIP2 in B cells expressing PLAID-associated PLC-γ2 mutants.

Fig. S7. Model depicting the PIP2-derived amplification loop for the robust initiation of B cell activation.

Table S1. Sequence of primers used in reverse transcription polymerase chain reaction for detecting the expression of PIP5K and DGK isoenzymes in DT40 B cells.

Movie S1. The depletion of PIP2 inside the BCR microclusters and the enrichment of PIP2 at the region outside the BCR microclusters in DT40-WT B cells.

Movie S2. PIP2 is depleted inside the BCR microclusters and enriched at the region outside the BCR microclusters in B1-8 primary B cells on a caged-NP antigen system.

Movie S3. The depletion of PIP2 inside the BCR microclusters and the enrichment of PIP2 at the region outside the BCR microclusters are impaired in DT40-PLC-γ2-KO B cells.

Movie S4. Enrichment of PIP2 at the region outside the BCR microclusters in a PIP5K-dependent manner.

Movie S5. DAG is more often than PIP2 to cross the BCR microcluster border.

Raw data Excel file

References (4248)

REFERENCES AND NOTES

Acknowledgments: We thank T. Kurosaki and H. Shinohara [World Premier International Research Center Initiative (WPI) Immunology Frontier Research Center, Osaka University, Japan] for providing DT40 cell lines. Funding: This work was supported by funds from National Science Foundation China (81730043, 81422020, and 81621002) and the Ministry of Science and Technology of China (2014CB542500-03). Author contributions: Chenguang Xu performed experiments and statistical analyses. H.X. contributed to single-molecule analysis. X.G. and L.L. prepared materials for the experiments. H.G., H.Q., and Chenqi Xu gave suggestions. W.L. supervised the project. Chenguang Xu and W.L. conceived the project and wrote the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article