The transformation of molecular binding events into cellular decisions is the basis of most biological signal transduction. A fundamental challenge faced by these systems is that reliance on protein–ligand chemical affinities alone generally results in poor sensitivity to ligand concentration, endangering the system to error. Here, we examine the lipid-binding pleckstrin homology and Tec homology (PH-TH) module of Bruton’s tyrosine kinase (Btk). Using fluorescence correlation spectroscopy (FCS) and membrane-binding kinetic measurements, we identify a phosphatidylinositol (3–5)-trisphosphate (PIP3) sensing mechanism that achieves switch-like sensitivity to PIP3 levels, surpassing the intrinsic affinity discrimination of PIP3:PH binding. This mechanism employs multiple PIP3 binding as well as dimerization of Btk on the membrane surface. Studies in live cells confirm that mutations at the dimer interface and peripheral site produce effects comparable to that of the kinase-dead Btk in vivo. These results demonstrate how a single protein module can institute an allosteric counting mechanism to achieve high-precision discrimination of ligand concentration. Furthermore, this activation mechanism distinguishes Btk from other Tec family member kinases, Tec and Itk, which we show are not capable of dimerization through their PH-TH modules. This suggests that Btk plays a critical role in the stringency of the B cell response, whereas T cells rely on other mechanisms to achieve stringency.
(Click on the small image to get a higher-resolution version.)
Figure 1 - Btk in the B cell receptor (BCR) pathway and the PH-TH domain
(A) The B cell signaling pathway.
(B) The domain architecture of Btk.
(C) Sequence alignment of the PH-TH modules for Btk, Itk, and Tec.
Figure 1 continued - Btk in the B cell receptor (BCR) pathway and the PH-TH domain
(D) Schematic depicting the Saraste dimer of the Btk PH-TH domain with two IP6 coordinated to the canonical (blue) and peripheral (magenta) phosphoinositide binding sites (PDB 4Y94). Zoom ins show the residues involved in the dimerization interface (Top) and the peripheral inositol phosphate binding site (Bottom).
(E) Same structure rotated to show the orientation with respect to the membrane.
(F) Comparison between Btk and Tec shows that Tec is missing key residues for the peripheral site and dimerization.
Figure 2 - Detection of 2D dimerization reaction on membrane surfaces by FCS.
(A) In a dual-color FCS setup, TR-labeled lipid (TR-DHPE) and eGFP-labeled PH-TH domain adsorbed to the SLB by PIP3 are simultaneously measured.
(B) For membrane-bound Ras, time-dependent fluorescence intensity fluctuation due to diffusion is recorded (Left), before (red) and after the addition of the RBD-LeuZ crosslinker (blue). The corresponding autocorrelation functions are shown on the Right.
(C) In the case of the PH-TH domain dimerization reaction, the difference in diffusion can be clearly resolved between lower surface density (20 molecules/μm2) and a higher surface density (400 molecules/μm2), reflecting the dimer population increase.
Figure 3 - Diffusion and FRET measurement.
The density-dependent diffusion was measured for the wild-type Btk (A), Tec and Itk (C), Btk dimer interface mutant (B), and the peripheral site mutant (D) on SLBs containing 1% (empty circles) and 4% PIP3
Figure 3 continued- Diffusion and FRET measurement.
(E) The fluorescence lifetime of eGFP tagged to wild-type PH-TH module in presence of mCherry-labeled wild-type PH-TH modules was measured as a function of PH-TH surface density on 4% PIP3 SLBs.
(F) FRET efficiency increases as a function of the PH-TH surface density, consistent with dimerization.
Figure 4 - Btk PH-TH module adsorption kinetics onto PIP3 SLBs by TIRF microscopy.
A. A simple three-step sequential kinetic model for Btk PH-TH domain is tested using mutant constructs that can undergo a subset of reactions.
(B) The membrane adsorption profile measured by TIRF on 4% PIP3 bilayers for each Btk construct.
Figure 4 continued- Btk PH-TH module adsorption kinetics onto PIP3 SLBs by TIRF microscopy.
(C) Using the kinetic rate constants obtained from the sequential lipid binding and dimerization model, the equilibrium surface density of dimers was calculated for both model A and B cases at 1% and 4% PIP3 (Top). With the results from C, density-dependent FCS was simulated.
Figure 5 - Fourth-order substrate detection by Btk.
(A) Using the kinetic rate constants derived from the sequential kinetic model, the dimer fraction was calculated for the hypothetical case in which the PH-TH module may dimerize with only one PIP3 (second order) and for the actual case where two are required (fourth order).
(B) An alternative method to achieve fourth-order PIP3 detection, single lipid binding followed by tetramerization to activation (Lower), was considered and compared with the observed mechanism (Upper).
(C) The proposed activation mechanism for Btk.
Figure 5 continued- Fourth-order substrate detection by Btk.
(D) Calcium flux for Btk-deficient chicken B cells transfected with variants of Btk and activated with BCR antibody (Left) and relative integrated calcium response for WT, gain-of-function (E41K), kinase dead (D521N), dimer interface mutant, and peripheral site mutant (Right).
(Click on the small image to get a higher-resolution version.)
Supplementary Figure 1 - TIRF calibration
TIRF intensity was calibrated against surface density measured by FCS with the dimer interface mutant.
Supplementary Figure 2 - Membrane adsorption of the canonical site mutant, N24D/R28C.
When PIP3 binding atthe canonical site is disrupted by the N24D/R28C mutation, there is negligible membrane recruitment, suggesting that the first kinetic step seen in the wild type is due to PIP3 binding at the canonical site.
[PIP3] = 4%, [WT] = 10 nM, [N24D/R28C] = 20 nM.
Supplementary Figure 3 - k1 is the same for all constructs.
The initial velocity was estimated by fitting the initial rate for each construct at 50 nM, 200 nM, and 400 nM solution protein concentration. The initial rates for all constructs agree, which is consistent with our assumptions about the sequential steps and independent binding (black = WT, orange = dimer mutant, green = peripheral site mutant).
Supplementary Figure 4 - Fitting of rate constants to experimental TIRF adsorption data.
Rate constants were fit for two of the models considered (Figure 4). Adsorption data for each mutant was fit sequentially, based on the observations from the FCS data.
(A) After the initial rate was fit, the adsorption data from the peripheral site mutant was fit to find k-1.
(B) Adsorption data for the dimer interface mutant was fit to find k2 and k-2.
(C) The WT adsorption data was fit to find k3a, k-3a, k3b, k-3b.
Supplementary Figure 5 - Simulated amount of each species for double binding or tetramer case.
The time course for different species plotted for 40 minutes in either case. The kinetic rate constants calculated from Model A and kcat from Src (1) were used.
(A) Shows the double binding-dimer case and (B) shows the tetramer case.
Supplementary Figure 6 - Effects of other anionic lipids.
Btk PH-TH module dimerization for various lipid bilayer compositions, including PIP3.
Supplementary Figure 7 - Effect of IP6 on PH-TH module dimerization.
(Left) On 4% PIP3 bilayers, the wild type PH-TH and the peripheral site mutant shows identical dimerization behavior with or without IP6.
(Right) On 1% PIP3 bilayers, the presence of IP6 does not promote dimerization.
Supplementary Figure 8 - Expression level-dependent B-cell activation for Btk variants.
The calcium flux upon activation by anti-BCR antibody are shown for wild type, D521N (kinase dead, negative control), E41K (gain-of-function, positive control), Y42R/F44R (dimer interface mutant), and R49S/K52S (peripheral
site mutant). The dimer interface and peripheral site mutants show calcium flux comparable to that of D521N, demonstrating that both dimerization and the peripheral PIP3 binding are important for Btk activation.
Supplementary Figure 9 - E41K mutation to the PH-TH module in vitro leads to loss of specificity and weaker binding affinity.
Here we show the surface density of GFP labeled PH-TH measured by FCS for either WT or E41K on both 4% PIP3 and 4% PIP2 bilayers. The solution concentration of PH-TH in each case was kept constant at 20 nM, and measurements were made after 40 minutes of incubation with the supported lipid bilayers.