Moitrayee Bhattacharyya
Moitrayee Stamp


Flexible linkers in CaMKII control the balance between activating and inhibitory autophosphorylation

Moitrayee Bhattacharyya, Young Kwang Lee, Serena Muratcioğlu, Baiyu Qiu, Priya Nyayapati, Howard Schulman, Jay T. Groves and John Kuriyan

ELife 2020;9:e53670     (local copy)

Abstract

The many variants of human Ca2+/calmodulin-dependent protein kinase II (CaMKII) differ in the lengths and sequences of disordered linkers connecting the kinase domains to the oligomeric hub of the holoenzyme. CaMKII activity depends on the balance between activating and inhibitory autophosphorylation (on Thr 286 and Thr 305/306, respectively, in the human α isoform). Variation in the linkers could alter transphosphorylation rates within a holoenzyme and the balance of autophosphorylation outcomes. We show, using mammalian-cell expression and a single-molecule assay, that the balance of autophosphorylation is flipped between CaMKII variants with longer and shorter linkers. For the principal isoforms in the brain, CaMKII-α, with a ~30 residue linker, readily acquires activating autophosphorylation, while CaMKII-β, with a ~200 residue linker, is biased towards inhibitory autophosphorylation. Our results show how the responsiveness of CaMKII holoenzymes to calcium signals can be tuned by varying the relative levels of isoforms with long and short linkers.

Figures from the paper

(Click on the small image to get a higher-resolution version.)
Figure 1 from paper

Figure 1 - Structural organization and Ca2+/CaM-dependent activation of CaMKII.


(A) CaMKII is organized as a holoenzyme with kinase domains connected to a central dodecameric/tetradecameric hub by a regulatory segment and a flexible linker, referred to as the kinase-hub linker. All the domains are labeled and this color scheme used will be maintained throughout. The kinase-hub linker is the principle difference between the four CaMKII isoforms: α/β/γ/δ.
(B) Crystal structure of the autoinhibited kinase domain from human CaMKII-δ with a small molecule inhibitor (PDB ID: 2VN9) bound (Rellos et al. 2010). The regulatory segment places Thr 306 optimally for cis-phosphorylation, while Thr 286 at the base of the kinase can only be phosphorylated in trans.
(C) Schematic diagram showing the design principle for all the constructs used in this study.
(D) Depiction of the four possible phosphorylation states of CaMKII.
Figure 2 from paper

Figure 2 - Mammalian expression-based single-molecule Total Internal Reflection Fluorescence (TIRF) assay.


(A) Schematic diagram showing the experimental setup. Biotinylated mEGFP-CaMKII overexpressed in HEK 293T cells was pulled down directly from diluted cell lysate, allowing visualization at a single-molecule resolution. The immobilization onto glass substrates functionalized with streptavidin relies on the interaction between biotinylated CaMKII and streptavidin. Autophosphorylation status of CaMKII holoenzyme can be measured using phosphospecific primary antibodies and Alexa-labeled secondary antibodies.
(B) Representative single-molecule TIRF images showing mEGFP-CaMKII holoenzymes (green dots), phosphorylation at Thr 286 (red dots) and phosphorylation at Thr 305/306 (purple dots) from left to right. A 3-color merge of these images reports on the fraction of CaMKII holoenzymes that are phosphorylated at Thr 286 and/or Thr 305/306.

Figure 2 from paper

Figure 2 continued - Mammalian expression-based single-molecule Total Internal Reflection Fluorescence (TIRF) assay.


(C) Fraction of CaMKII-α that shows detectable phosphorylation at Thr 286 is plotted for different Ca2+/CaM concentrations ranging from 0.02 µM to 5 µM. The cartoon on the bottom panel depicts two extreme cases, where only a few holoenzymes are phosphorylated or where most holoenzymes are phosphorylated.
(D) Distribution of intensity for pThr 286 (561 nm), at different Ca2+/CaM concentrations, for CaMKII-α holoenzymes with detectable phosphorylation. The area under each plot is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization). The cartoon in the bottom panel shows that a right-shift in the peak value of the intensity histogram represents a higher extent of phosphorylation within a CaMKII holoenzyme and vice versa.

Figure 3 from paper

Figure 3 - Autophosphorylation status in activated CaMKII-α and CaMKII-β/β*.


(A) Comparison of the extent of autophosphorylation (intensity histogram) within CaMKII-α and CaMKII-β holoenzyme at the activating site (Thr 286) (left panel) and the inhibitory site (Thr 305/306) (right panel). The insets show the fraction of holoenzymes that exhibit any detectable phosphorylation for the corresponding phosphosite in CaMKII-α and CaMKII-β.
(B) Comparison of the extent of autophosphorylation (intensity histograms) between CaMKII-α and CaMKII-β* holoenzyme at the activating site (Thr 286) (left panel) and the inhibitory site (Thr 305/306) (right panel). The insets show the fraction of holoenzymes that exhibit any detectable phosphorylation for the corresponding phosphosite in CaMKII-α and CaMKII-β*. The area under each intensity histogram is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization).
Figure 3 from paper

Figure 3 continued - Autophosphorylation status in activated CaMKII-α and CaMKII-β/β*.


(C) Autophosphorylation status of CaMKII after activation in HEK 293T cells using ionomycin. Fractions of CaMKII-α and CaMKII-β* that show detectable phosphorylation at Thr 286 is plotted for different conditions.
(D) Fractions of CaMKII-α and CaMKII-β* that show detectable phosphorylation at Thr 305/306 is plotted for different conditions. +/- depicts the presence or absence of ionomycin and/or phosphatase inhibitors.
Figure 4 from paper

Figure 4 - Inhibitory autophosphorylation in CaMKII-α-CaMKII-β* heterooligomers.


A. Schematic diagram showing that co-expression of GFP-CaMKII-α and mCherry-CaMKII-β* leads to the formation of heterooligomers.
(B) Bar graph showing the fraction of holoenzymes that show detectable phosphorylation at the inhibitory site (Thr 305/306), which increases as the ratio of CaMKII-β* increases.
(C) Intensity histogram for the homooligomers and heterooligomers. As the ratio of CaMKII-β* increases, there is a right-shift in the peak value of the intensity histogram. The area under each intensity histogram is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization)

Figure 5 from paper

Figure 5 - A simplified schematic diagram showing the key pathways for autophosphorylation at the activating and inhibitory sites, in the absence or presence of Ca2+/CaM.


While Thr 305/306 can get phosphorylated both in cis in the absence of Ca2+/CaM or in trans in the presence of Ca2+/CaM, autophosphorylation of Thr 286 can only happen in trans in the presence of Ca2+/CaM. Ca2+/CaM shows a rapid association and dissociation until CaMKII gets phosphorylated at Thr 286, when its affinity for Ca2+/CaM increases by about 1000-fold. A detailed description of all the different reactions and conditions that form the basis of our kinetic model is discussed in the Appendix.

Figure 6 from paper

Figure 6 - Effect of λ-phosphatase on the phosphorylation status of CaMKII when the kinase is active.


(A) Schematic diagram showing the experimental set up. CaMKII was activated in the presence of 0, 200, 400, or 800 units of λ-phosphatase for 45 min.
(B-C) Bar graph showing the fraction of CaMKII-α and CaMKII-β* holoenzymes that shows detectable phosphorylation at the activating site (Thr 286) and the inhibitory site (Thr 305/306), respectively, in the presence of λ-phosphatase.
(D) Intensity distribution of pThr 286 (561 nm) signal for CaMKII-α (left panel) and CaMKII-β* (right panel) holoenzymes with detectable phosphorylation in the presence of λ-phosphatase.
(E) Intensity distribution of pThr 305/306 (640 nm) signal for CaMKII-α (left panel) and CaMKII-β* (right panel) holoenzymes with detectable phosphorylation in the presence of λ-phosphatase. The area under each plot in (D-E) is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization).

Figure 7 from paper

Figure 7 - Effect of λ-phosphatase on the dephosphorylation kinetics when kinase activity is switched off.


(A) Schematic diagram showing the experimental set up. CaMKII was activated, followed by a wash to remove the components of the activation buffer, and then saturating amounts of λ-phosphatase/PP1α (400-800 units) were added for 0, 3, 15, or 30 min.
(B) Plot showing the fraction of α and β* holoenzymes that exhibits detectable phosphorylation at the inhibitory site (Thr 305/306) upon treatment with λ-phosphatase for defined time-points. The fractions at 3, 15, or 30 minutes for α and β* are normalized by the corresponding activated version that has not been exposed to any λ-phosphatase (0-minute time-point, whose value is set to 1.0).
(C) Same as (B) but for the activating site (Thr 286).
(D) Intensity distribution for pThr 286 (561 nm) for CaMKII-α (left panel) and CaMKII-β* (right panel) holoenzymes with detectable phosphorylation after 0, 3, 15, or 30 minutes of λ-phosphatase treatment. The area under each plot is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization).

Figure 8 from paper

Figure 8 - Effect of addition of Ca2+/CaM on the rates of dephosphorylation at the activating site.


(A) Schematic diagram showing the experimental set up. CaMKII was activated, followed by wash to remove the components of the activation buffer and then saturating amounts of λ-phosphatase were added for 3, 15, or 30 min, in the presence and absence of Ca2+/CaM.
(B) Plot showing the fraction of CaMKII-&aplha; holoenzymes with detectable phosphorylation at the activating site (Thr 286, right panel) after 0 minute (activated control) and 3, 15, or 30 minutes of treatment with saturating amounts of λ-phosphatase in the absence (green trace) and presence (pink trace) of Ca2+/CaM. The fractions at 3, 15, or 30 minutes are normalized with respect to activated CaMKII that has not been exposed to any λ-phosphatase (0-minute, whose value is set to 1.0).
(C) Intensity distribution for pThr 286 (561 nm) for CaMKII-α holoenzymes with detectable phosphorylation, upon 0, 3, 15, or 30 minutes of λ-phosphatase treatment in the presence of 5 µM Ca2+/CaM. The area under each plot is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization).

Figure 9 from paper

Figure 9 - Recovery of phosphorylation at the activating site by subthreshold concentrations of Ca2+/CaM.


(A) Schematic diagram showing the experimental set up. CaMKII was activated, followed by a wash to remove the components of the activation buffer and then saturating amounts of λ-phosphatase were added for 3-5 min. The sample is then washed to remove the λ-phosphatase, followed by further treatment with subthreshold concentrations of Ca2+/CaM (25 nM) for 30 minutes and the autophosphorylation status was measured.
(B) Cartoon representation of the different species generated after each treatment. Each species is color-coded and the color schemes are maintained throughout the plots.
(C) Fraction of CaMKII-α holoenzymes with detectable phosphorylation at Thr 286 is plotted for each species.
(D) Intensity distribution of pThr 286 (561 nm) for only those CaMKII-α holoenzymes that show any detectable Thr 286 phosphorylation, for the different species of interest as described in (B).


Supplemental figures from the paper

(Click on the small image to get a higher-resolution version.)
Supplemental Figure 1 from paper

Figure 1 Figure supplement 1 - Amino acid sequences for the kinase-hub linker in the four isoforms of human CaMKII.




Supplemental Figure 2 from paper

Figure 2 Figure supplement 1 - Validation of phosphospecific antibodies.


Mutation/deletion of epitopes for pThr 286 (left) and pThr 305/306 (right)-specific antibodies leads to no detection of phosphosignal at the respective sites.

Supplemental Figure 3 from paper

Figure 3 Figure supplement 1 - Inhibitory autophosphorylation status in activated CaMKII-α.


(A) Schematic diagram showing the experimental set up. CaMKII was activated, followed by a wash to remove the components of the activation buffer. The pre-activated CaMKII (species A) is then treated with Mg2+-ATP for 30 minutes to generate species B.
(B) Fraction of CaMKII-α holoenzymes with detectable phosphorylation at Thr 305/306 is plotted for each species.
(C) Intensity distribution of pThr 305/306 (640 nm) for species A and species B with detectable phosphorylation. The area under each intensity histogram is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization).

Supplemental Figure 4 from paper

Figure 3 Figure supplement 2 - Autophosphorylation status in activated CaMKII-γ* and CaMKII-δ*.


Comparison of the extent of autophosphorylation (intensity histogram) among CaMKII-γ*, CaMKII-δ*, CaMKII-α, and CaMKII-β* holoenzymes at the activating site (Thr 1015 286) (left panel) and the inhibitory site (Thr 305/306) (right panel). The insets show the fraction 1016 of holoenzymes that exhibit any detectable phosphorylation for the corresponding phosphosite in 1017 CaMKII-γ* and CaMKII-δ*. The area under each intensity histogram is scaled by the fraction of 1018 holoenzymes that show no detectable phosphorylation under that condition (see Methods for 1019 details of normalization).

Supplemental Figure 5 from paper

Figure 3 Figure supplement 3 - Autophosphorylation status in activated CaMKII-β/β* variants.


(A) Comparison of the extent of phosphorylation at the activating site (left panel) and the inhibitory site (right panel) for CaMKII-β and CaMKII-β’E holoenzymes with detectable phosphorylation.
(B) Comparison of the extent of phosphorylation at the activating site (left panel) and the inhibitory site (right panel) for CaMKII-β and CaMKII-β’E* holoenzymes with detectable phosphorylation.
(C) Comparison of the extent of phosphorylation at the activating site (left panel) and the inhibitory site (right panel) for CaMKII-α, CaMKII-β and CaMKII-α* holoenzymes with detectable phosphorylation. The area under each intensity histogram is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization).

Supplemental Figure 6 from paper

Figure 5 Figure supplement 1 - Results from simulations of a simple kinetic model for autophosphorylation in CaMKII using Berkeley Madonna.


Plot showing the production of all species bearing pThr 286 or pThr 305/306 over simulation time, when (A) the linker-length is short with faster rates of trans-autophosphorylation and when (B) the linker-length is long and the rates of trans-autophosphorylation are 10-fold slower.

Supplemental Figure 7 from paper

Figure 7 Figure supplement 1 - Effect of phosphatases on dephosphorylation kinetics when kinase activity is switched off.


(A) Fraction of CaMKII-α/β* holoenzymes with detectable phosphorylation at Thr 305/306 (left) and Thr 286 (right) after 0, 3, 15, or 30 minutes of treatment with λ-phosphatase, when the kinase activity is switched off.
(B) Same as (A), but PP1α phosphatase was used.
(C) Intensity distribution for pThr 286 (561 nm) for CaMKII-α (left panel) and CaMKII-β* (right panel) holoenzymes with detectable phosphorylation at Thr 286 after 0, 3, 15, or 30 minutes of PP1α treatment. The area under each plot is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization).

Supplemental Figure 8 from paper

Figure 7 Figure supplement 2 - Measurement of dephosphorylation kinetics in solution.


(A) Schematic diagram depicting the experimental design to measure dephosphorylation kinetics in solution using λ-phosphatase.
(B) Fraction of CaMKII-α/β* holoenzymes with detectable phosphorylation at Thr 305/306 (left) and Thr 286 (right) after 0, 3, 15, or 30 minutes of treatment with λ-phosphatase in-solution, when the kinase activity is switched off by staurosporine.
(C) Intensity distribution for pThr 286 (561 nm) for CaMKII-α (left panel) and CaMKII-β* (right panel) holoenzymes with detectable phosphorylation at Thr 286 after 0, 3, 15, or 30 minutes of λ-phosphatase treatment. The area under each plot is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization).

Supplemental Figure 9 from paper

Figure 7 Figure supplement 3 - Inhibition of CaMKII kinase activity by staurosporine.


Fraction of CaMKII-α/β* holoenzymes with detectable phosphorylation at Thr 286 (left) and Thr 305/306 (right) after 60 minutes of treatment with the activation buffer (see Methods) in the absence and presence of 100 µM staurosporine.

Supplemental Figure 9 from paper

Figure 8 Figure supplement 1 - Effect of addition of Ca2+/CaM on the rates of dephosphorylation at the autonomy site.


(A) Fraction of CaMKII-β* holoenzymes with detectable phosphorylation at Thr 286, upon 3 minutes of λ-phosphatase treatment in the absence and presence of 5 µM Ca2+/CaM.
(B) Intensity distribution of pThr 286 (561 nm) for CaMKII-β* holoenzymes with detectable phosphorylation, under the same conditions as in (A). The area under each plot is scaled by the fraction of holoenzymes that show no detectable phosphorylation under that condition (see Methods for details of normalization).


Appendix

(Click on the figure to download the see the high resolution version.)
Appendix Figure 1 from paper

Appendix-figure 1: A summary of the reactions considered in the kinetic model.


K denotes the kinase (CaMKII) and C denotes CaM. Phosphorylation on Thr 286 or Thr 305/306 is denoted by 286 and 306, respectively.

 Source code file 1:(click here)


Source code file In-house Matlab programs that are used for data analyses are provided as an open source package. A readme file and a test dataset is included for clarity.