Toll-like receptor ligands sensitize B-cell receptor signalling by reducing actin-dependent spatial confinement of the receptor

Integrating signals from multiple receptors allows cells to interpret the physiological context in which a signal is received. Here we describe a mechanism for receptor crosstalk in which receptor-induced increases in actin dynamics lower the threshold for signalling by another receptor. We show that the Toll-like receptor ligands lipopolysaccharide and CpG DNA, which are conserved microbial molecules, enhance signalling by the B-cell antigen receptor (BCR) by activating the actin-severing protein cofilin. Single-particle tracking reveals that increased severing of actin filaments reduces the spatial confinement of the BCR within the plasma membrane and increases BCR mobility. This allows more frequent collisions between BCRs and greater signalling in response to low densities of membrane-bound antigen. These findings implicate actin dynamics as a means of tuning receptor signalling and as a mechanism by which B cells distinguish inert antigens from those that are accompanied by indicators of microbial infection.


Supplementary Figure 2 | Immobilized anti-MHCII antibodies do not alter cytoskeletal organization in B cells.
Primary B cells that had been cultured overnight with 5 ng ml -1 BAFF or with BAFF + 5 µg ml -1 LPS were adhered to anti-MHCII-coated coverslips and stained with Alexa488-phalloidin. F-actin at the ventral and dorsal surfaces of the cell was visualized by confocal microscopy. Scale bar, 5 µm. The ratio of F-actin fluorescence intensity per unit area at the ventral and dorsal surfaces was determined for >10 cells in each experiment and the mean + s.e.m. for 3 experiments is shown. For both the BAFF-and LPS-cultured B cells, the F-actin densities on the ventral and dorsal sides of the cells remained nearly identical when the cells were plated on anti-MHCII antibodies. This, in combination with the absence of cell spreading and F-actin clearance on the ventral side of the cell that was in contact with the anti-MHCII antibodies, indicates that this is a non-stimulatory condition. This is in contrast to the dramatic spreading and F-actin clearance induced by plating cells on anti-Ig antibodies 1, 2 .

Supplementary Figure 3 | Single-particle labeling of BCRs on the B cell surface.
(a) B cells were labeled on ice with 5 ng ml -1 Cy3-labeled anti-IgM Fab fragments. The cells were adhered to anti-MHCII-coated coverslips and then imaged by total internal reflection microscopy (TIRF) at 33 frames s -1 for 10 s with laser settings that allowed photobleaching to be observed. A still image from a representative video is shown (left panel). Fluorescence intensities in the indicated regions are compared to that for monodispersed soluble Cy3-labeled anti-IgM Fabs that were imaged using the same settings (right panel). Single-particle labeling is indicated by a single quantized fluorescence decrease (e.g. region #1, black trace) of similar magnitude to that for monodispersed Cy3-anti-IgM Fabs (red trace). Although the majority of spots contained a single Cy3-labeled Fab, some contained >1, e.g. region #2 (blue trace) contained two.
(b) This photobleaching analysis was performed on >300 random particles. When cells were labeled with 5 ng ml -1 Cy3-anti-IgM Fabs, ~75% of identified particles contained only one Fab. When cells were labeled with 1 ng ml -1 Cy3-anti-IgM Fabs (the concentration used in Qdot SPT experiments), ~95% of particles contained only one Fab. The linear relationship between the Fab concentration and the number of particles containing multiple Cy3 molecules (~5% for 1 ng ml -1 Fab; ~25% for 5 ng ml -1 ) is consistent with labeling of mostly single BCRs.
(c) B cells were labeled with 1 ng ml -1 biotin-anti-IgM Fabs plus streptavidin-655 nm Qdots before imaging the ventral surface. The number of Qdots at the ventral surface was determined before and after washing the cells with a low-pH stripping buffer. Acid-dissociable Qdots are assumed to be on the cell surface.
Note that anti-IgG Fabs did not label primary splenic B cells and that anti-IgM Fabs did not label the mIgG + A20 B-lymphoma cells, demonstrating the specificity of the reagents used for SPT of BCRs. Qdots alone did not bind to B cells. Moreover, when B cells were labeled with Qdots, washed, and then adhered to anti-MHCII-coated coverslips, we did not observe any Qdots that were not associated with B cells. Figure 4 | Estimation of the error in determining particle position (positional accuracy). Ex vivo B cells were labeled with 1 ng ml -1 Cy3-labeled anti-IgM Fab fragments or with 1 ng ml -1 biotinylated anti-IgM Fab fragments plus streptavidin-655 nm Qdots. A20 B-lymphoma cells were labeled with 1 ng ml -1 biotinylated anti-IgG Fab fragments plus streptavidin-655 nm Qdots. The cells were adhered to anti-MHCII-coated coverslips before being imaged in real time. Single-particle tracking (SPT) using u-track software was performed as described 3 . Particle positions were determined by fitting Gaussian kernels to local maxima of fluorescence intensities. The positional accuracy was calculated from the width of the Gaussian curve that was fit to the particle intensity profile 3 . For each labeling condition, this value was determined for >300 random particles. Each point represents the estimated positional accuracy for an individual particle. Bars indicate mean values. The mean positional accuracy for particles visualized by labeling primary B cells with biotinylated anti-IgM Fab fragments plus Qdots was 10.5 nm. This labeling condition was employed for all experiments with the following exceptions. For experiments involving A20 B-lymphoma cells (Fig. 5i, Fig. 6) BCRs were labeled with biotinylated anti-IgG Fab fragments plus Qdots (mean positional accuracy of 18.9 nm). For the experiments shown in Supplementary Fig. 15, B cells were labeled with Cy3-anti-IgM Fab fragments (mean positional accuracy of 17.8 nm).

Supplementary Figure 5 | Single-state analysis of BCR diffusion.
Primary B cells were treated as described below, labeled with biotinylated anti-IgM Fab fragments plus streptavidin-655 nm Qdots, and adhered to anti-MHCII-coated coverslips before performing SPT of BCRs on the ventral surface of the cells. Trajectories of individual mIgM-containing BCRs were generated from live-imaging videos taken at 33 frames per s -1 . Single-state diffusion coefficients (D) for individual tracks were calculated using a maximum likelihood estimation approach developed by Das et al. 4 , as described in the Methods. The data are presented as cumulative frequency curves and the median values are indicted by the dots on the curves. For each condition, >500 tracks were analyzed.
(a) B cells were cultured overnight in 5 ng ml -1 BAFF, with 5 µg ml -1 LPS being added for the last 0-16 h of culture. These samples are identical to those for which MSS analysis is shown in (b) B cells that had been cultured overnight with either BAFF or BAFF + 5 µg ml -1 LPS were treated with DMSO or 1 µM latrunculin B (LatB) for 3 min. These samples are identical to those for which MSS analysis is shown in Fig. 3 and HMM analysis is shown in Supplementary Table 1.
(c) B cells that had been cultured overnight with either BAFF + 5 µg ml -1 LPS were treated with 5 µM of the M/W cofilin-blocking peptides or the control Q peptide for 5 min at 37 o C. These samples are identical to those for which MSS analysis is shown in Fig. 5b- (d) B cells that had been cultured overnight with BAFF + 5 µg ml -1 LPS were incubated with or without the RhoA-activating peptide for 2-3 h. These samples are identical to those for which MSS analysis is shown in Fig. 5h and HMM analysis is shown in Supplementary Table 1.

Supplementary Figure 6 | Representative distributions of BCR confinement diameters.
SPT of mIgM-containing BCRs was performed as in Fig. 2. Confinement diameters were determined for individual IgM-containing BCRs that were classified as exhibiting confined/subdiffusive motion by the MSS algorithms. Confinement diameter values for individual BCRs were sorted into bins that are centered around multiples of 100 nm, e.g. 100 + 50 nm. The graphs display the relative frequency of BCR confinement diameters in all of the videos in which cells in a specific treatment group were analyzed within a single experiment. The total number of BCR trajectories per condition was >600 and the median confinement diameter values are indicated by the triangles. Comparing the 0 anti-Ig lanes shows the increased basal pERK levels caused by overnight culture with LPS or CpG DNA. The anti-Ig stimulation lanes show that TLR priming enhanced BCR-induced ERK phosphorylation at low concentrations (0.1 µg ml -1 ) of anti-Ig and that the increased basal level of pERK caused by TLR priming is less than that caused by acute stimulation of BAFF-cultured B cells with 1-10 µg ml -1 anti-Ig.
Full blot for Figure 4b: Extracts were prepared from ex vivo primary B cells (lanes 1,2), B cells that had been cultured overnight with 5 ng ml -1 BAFF (lanes 3,4), B cells that had been cultured overnight with BAFF + 5 µg ml -1 LPS (lanes 5,6), or B cells that had been cultured overnight with BAFF + 0.5 µg ml -1 CpG DNA (lanes 7,8). The cell extracts were then separated into soluble fractions containing G-actin (lanes labeled G) and insoluble fractions containing F-actin (lanes labeled F). Blots were probed with a -actin antibody.
(b) The percent of mIgM BCRs that exhibited free diffusion, as opposed to confined diffusion.
(c) Median confinement diameters for the confined mIgM BCRs (d) Median diffusion coefficients for both confined and free mIgM-containing BCRs.
Each dot represents the trajectories from a single video. Horizontal lines are mean values for >50 videos per condition; >3000 trajectories per condition were analyzed. ***P < 0.001, Student's unpaired 2-tailed t-test.
The mean positional accuracy for particles labeled with Cy3-anti-IgM Fab fragments was 17.8 nm (see Supplementary Fig. 4). The results obtained using this approach were similar to those obtained using biotinylated anti-IgM Fab fragments plus Qdots. Given the limited penetration depth of TIRF, these results confirm that only BCRs on the cell surface were tracked when we used Qdot labeling to visualize BCR trajectories.

Supplementary Table 1 | Two-state hidden Markov model (HMM) analysis of BCR diffusion.
Primary B cells were treated as described below, labeled with biotinylated anti-IgM Fabs plus streptavidin Qdots, and adhered to anti-MHCII-coated coverslips before performing SPT of BCRs on the ventral surface of the cells. Trajectories of individual mIgM-containing BCRs were generated from live-imaging videos taken at 33 frames s -1 and analyzed using a twostate HMM algorithm 4 . This model assumes that individual trajectories consist of slow and fast segments and calculates the distribution of diffusion coefficients for each state as well as the probability of transitions between the two states. Diffusion coefficients for the slow and fast states (D slow , D fast ) are shown along with the frequency of transitions between the two states (k slow  fast , k fast  slow ). K eff , the ratio of k slow  fast divided by k fast  slow , indicates whether the predominant mode of switching is to the fast state (K eff >1) or to the slow state (K eff <1). The ranges in parentheses indicate 95% confidence intervals. For each condition, >500 BCR trajectories were analyzed.
The samples analyzed using this HMM algorithm are the same as those in the following figures, which show the corresponding MSS analysis: a) Fig. 2h-j. B cells were cultured overnight in 5 ng ml -1 BAFF, with 5 µg ml -1 LPS being added for the last 0-16 h of culture. Single-state analysis for these samples is shown in Supplementary  Fig. 5a. b) Fig. 3. B cells that had been cultured overnight with either BAFF or BAFF + LPS were treated with DMSO or 1 µM latrunculin B (LatB) for 3 min. Single-state analysis for these samples is shown in Supplementary Fig. 5b. c) Fig. 5b-d. B cells that had been cultured overnight with BAFF + LPS were treated with 5 µM of the M/W cofilin-blocking peptides or the control Q peptide for 5 min. Single-state analysis for these samples is shown in Supplementary Fig. 5c. Table 1 Condition # of tracks analyzed D slow (10 -3 µm 2 s -1 ) D fast (10 -2 µm 2 s -1 ) k slow à fast (s -1 ) k fast à slow (s -1 ) K eff = k slow à fast /k fast à slow Fig. 2h- Fig. 5h. B cells that had been cultured overnight with BAFF + LPS were incubated with or without the RhoA-activating peptide for 2-3 h. Single-state analysis for these samples is shown in Supplementary Fig. 5d.

Condition
Confinement diameter mean ± s.e.m. (nm) Fig. 2c- Primary B cells were treated as described below, labeled with biotinylated anti-IgM Fab fragments plus streptavidin-655 nm Qdots, and adhered to anti-MHCII-coated coverslips before performing SPT of BCRs on the ventral surface of the cells. The trajectories were analyzed using MSS analysis as described in the Methods. For BCRs classified as exhibiting confined/subdiffusive motion, the MSS algorithms calculated the maximum diameter of the confinement region. The values in this table are the mean + s.e.m. for median confinement diameters from >10 videos per condition; >500 trajectories were analyzed per condition.
The samples for which confinement diameters are reported above are the same as those in the following figures: a) Fig. 2h-j. B cells were cultured overnight in 5 ng ml -1 BAFF, with 5 µg ml -1 LPS being added for the last 0-16 h of culture. b) Fig. 3. B cells that had been cultured overnight with either BAFF or BAFF + 5 µg ml -1 LPS were treated with DMSO or 1 µM latrunculin B (LatB) for 3 min. c) Fig. 5b-d. B cells that had been cultured overnight with BAFF + 5 µg ml -1 LPS were treated with 5 µM of the M/W cofilin-blocking peptides or the control Q peptide for 5 min. d) Fig. 5h. B cells that had been cultured overnight with BAFF + 5 µg ml -1 LPS were incubated with or without the RhoA-activating peptide for 2-3 h.
When B cells were fixed with 4% paraformaldehyde, BCRs were rendered immobile. The median confinement diameter calculated for BCRs on these fixed cells reflects the size of the Qdot (15-20 nm) as well as vibrational movement of the Qdot. This value provides an estimate of the accuracy of the confinement diameter measurements. For almost all of the conditions shown in this table, the confinement diameter value for immobile BCRs on fixed cells was <25% of the measured confinement diameter value. Changes in the confinement diameter values caused by various treatments were sufficiently large that this value for immobile BCRs does not affect the conclusions drawn from the data.