Increased cardiac contractility during fight-or-flight response is caused by β-adrenergic
augmentation of CaV1.2 channels
1-4
. In transgenic murine hearts expressing fully PKA phosphorylation-site-deficient
mutant CaV1.2 α1C and β subunits, this regulation persists, implying involvement of
extra-channel factors. Here, we identify the mechanism by which β-adrenergic agonists
stimulate voltage-gated Ca2+ channels. We expressed α1C or β2B subunits conjugated
to ascorbate-peroxidase
5
in mouse hearts and used multiplexed, quantitative proteomics
6,7
to track hundreds of proteins in proximity of CaV1.2. We observed that the Ca2+ channel
inhibitor Rad
8,9
, a monomeric G-protein, is enriched in the CaV1.2 micro-environment but is depleted
during β-adrenergic stimulation. PKA-catalyzed phosphorylation of specific Ser residues
on Rad decreases its affinity for auxiliary β-subunits and relieves constitutive inhibition
of CaV1.2 observed as an increase in channel open probability. Expression of Rad or
Rem, a homolog, also imparted PKA-induced stimulation of CaV1.3 and CaV2.2, revealing
an evolutionarily conserved mechanism that confers adrenergic-modulation upon voltage-gated
Ca2+ channels.
The positive inotropic effect of β-adrenergic agonists on the heart is a classical
physiological phenomenon universally experienced during excitement, exercise, and
fight-or-flight. The effect is mediated by β-adrenergic activation of protein kinase
A (PKA) which leads to increased Ca2+ influx through L-type CaV1.2 channels in cardiomyocytes
1-4
. The commonly accepted model is that PKA increases CaV1.2 current by phosphorylating
CaV1.2 α1C- and/or β2B-subunits (Fig. 1a). However, previously proposed putative regulatory
residues on the C-termini of α1C (Ser1928; Ser/Thr1700/1704)
10,11
and β2B (Ser512 and Ser570)
12
were shown to be dispensable for β-adrenergic stimulation of Ca2+ currents in the
heart
13-16
. Nevertheless, given multiple other Ser/Thr residues on α1C and β2B, it remained
possible that PKA phosphorylation of some combination of these was responsible for
β-adrenergic modulation of CaV1.2 in cardiomyocytes. As shown next, this also is not
the case.
Core CaV1.2 channel subunit phosphorylation is not required for adrenergic regulation
We developed a transgenic approach that enables doxycycline-inducible expression of
FLAG- tagged, dihydropyridine (DHP)-resistant CaV1.2 channels in mice (Fig. 1b)
16
. The transgenic and endogenous CaV1.2 currents are distinguishable by application
of nisoldipine, a Ca2+-channel DHP-antagonist
16
. We mutated all 51 of both conserved and non-conserved Ser and Thr residues within
the 35 intracellular PKA consensus phosphorylation sites of rabbit α1C to Ala (“35-mutant
α1C”; Extended Data Fig. 1a). In cardiomyocytes, the nisoldipine-insensitive 35-mutant
Ca2+ currents were both up-regulated and activated at more negative potentials in
response to isoproterenol or forskolin, to the same extent as were the pseudo-WT (pWT)
α1C channels (Fig. 1c-d; Extended Data Fig. 1b-c).
Similarly, we mutated to Ala all 37 conserved and non-conserved Ser and Thr residues
within 28 PKA-consensus phosphorylation sites of human β2B (“28-mutant β2B”; Extended
Data Fig. 1d). Cardiomyocytes expressing GFP-tagged 28-mutant β2B (Extended Data Fig.
1e-f) displayed isoproterenol or forskolin-induced stimulation of CaV1.2 current amplitude
(Fig. 1e, g), and a hyperpolarizing shift in the voltage-dependence of activation
(Extended Data Fig. 1b), similar to cardiomyocytes isolated from transgenic mice expressing
GFP-tagged WT β2B
17
.
Finally, we crossed 35-mutant α1C with 28-mutant β2B transgenic mice. Anti-FLAG antibody
immunoprecipitation indicated GFP-tagged 28-mutant β2B subunits dominate in the 35-mutant
α1C complex (Extended Data Fig. 1g) in cardiomyocytes isolated from these mice. These
mutant channels also displayed a normal isoproterenol- or forskolin-induced increase
in peak Ca2+ current (Fig. 1f-g) and a hyperpolarizing shift in the V50 of activation
(Extended Data Fig. 1b). These results indicate that β-adrenergic stimulation of CaV1.2
does not involve direct phosphorylation of α1C or β2 subunits.
Identifying the CaV1.2 proteome subdomain
Given the foregoing results, we adapted for application to cardiomyocytes an enzyme-catalyzed
proximity labeling method
5,6
to comprehensively identify components of the CaV1.2 macromolecular complex. We generated
transgenic mice with doxycycline-inducible, cardiomyocyte-specific expression of DHP-resistant-α1C
or β2B proteins with ascorbate-peroxidase (APEX2) and a V5 epitope conjugated to the
N-termini, enabling biotin-labeling of proteins within ~20 nm
18
of the Ca2+ channels. Fusing APEX2 to α1C and β2B did not affect CaV1.2 subcellular
localization and function in cardiomyocytes assessed using cellular electrophysiology,
fractional shortening and immunofluorescence (Extended Data Fig. 2a-d). Incubating
isolated ventricular cardiomyocytes or perfusing whole hearts with a solution containing
biotin-phenol followed by exposure to H2O2 induced robust biotinylation of proteins
in a striated z-disk pattern coinciding with the pattern of transgenic α1C and β2B
subunits (Fig. 2a-b; Extended Data Fig. 2d-e).
Western blots confirmed biotinylation of proteins known to be localized near CaV1.2
at dyadic junctions in cardiomyocytes (Fig. 2c). By contrast, KV1.5 channels were
not biotinylated and enriched (Fig. 2c), implying that these channels may not be as
closely localized to CaV1.2. Triple-stage mass spectrometry (TMT SPS MS3) identified
and quantified hundreds of other biotinylated proteins (Supplementary Data Table 1),
although many likely constitute bystanders rather than physically interacting proteins.
Of these, the 150 proteins with the highest peptide-counts were remarkably similar
for α1C-APEX2 and β2B-APEX2 transgenic mice (Extended Data Fig. 3a). These were primarily
classified as being membrane, cytoskeletal, and sarcomeric proteins (Extended Data
Fig. 3b-c, Supplementary Data Table 2). Some of these proteins, however, were associated
with other compartments likely reflecting the labeling of proteins encountered during
CaV1.2-APEX2 synthesis, maturation, and trafficking.
Isoproterenol-induced changes in CaV1.2 neighbors
It seemed likely that PKA-dependent stimulation of Ca2+ current in the heart involves
recruitment of a distinct activator protein to, or loss of an inhibitory protein from,
the CaV1.2 macromolecular complex. Recently, APEX2-labeling, combined with either
multiplexed quantitative mass spectrometry
7
or quantitative proteomics using a system of spatial references and bystander ratio
calculations
19
, was utilized to analyze ligand-induced changes in the local environment of G-protein-coupled
receptors. Importantly, conjugation of the peroxidase to either of the CaV1.2 subunits
did not interfere with β-adrenergic stimulation of Ca2+ currents (Extended Data Fig.
2f-g), and the β-adrenergic agonist signaling pathways were preserved in the presence
of biotin-phenol and during H2O2-induced labeling, assessed by phospholamban phosphorylation
(Extended Data Fig. 2h-k).
Isolated cardiomyocytes from α1C-APEX2 and β2B-APEX2 mice were pre-incubated with
biotin-phenol for 30-minutes, and during the final 10-minutes, also exposed to either
isoproterenol or vehicle (Fig. 2d). Biotinylated proteins were purified and quantified
using TMT SPS MS3. Relative summed peptide TMT signal-to-noise, indicating relative
protein quantification, changed for several proteins (Fig. 2e-f). We also probed the
effect of isoproterenol on the CaV1.2 proteome signature in Langendorff-perfused whole
hearts (Fig. 2g-h). Isoproterenol-induced recruitment of the PKA catalytic subunit
(PKAcat) to CaV1.2 channels was detected in cardiomyocytes isolated from both α1C-APEX2
and β2B-APEX2 mice (Fig. 2e-f). All three approaches (isolated cardiomyocytes from
α1C-APEX2 and β2B-APEX2 mice, and α1C-APEX2 whole hearts) indicated a 30-50% decrease
in the amount of the small Ras-like G-protein, Rad, following application of isoproterenol
(Fig. 2e-f,h, Extended Data Figs. 4a-c, 5a-b, Supplementary Data Tables 3-5). Analyzing
the overlap of proteins displaying isoproterenol-induced changes in α1C-APEX2 and
β2B-APEX2 experiments, Rad was the only candidate protein displaying this behavior
(Extended Data Fig. 5b-c). In contrast, a 10-minute exposure of cardiomyocytes isolated
from non-transgenic mice to isoproterenol (Fig. 2i) had a minimal effect on proteins
relatively quantified by TMT SPS MS3 and specifically no significant effect on the
amount of Rad as compared to non-treated paired cardiomyocytes (Fig. 2j, Extended
Data Fig. 4d, Supplementary Data Table 6). In this control experiment, proteins were
not proximity labeled. Thus, β-adrenergic stimulation depletes Rad from the neighborhood
of CaV1.2, but not from the cell as a whole.
Rad is a member of the Rad, Rem, Rem2, and Gem/Kir (RGK) family of Ras-like GTP-binding
proteins, known for their capacity to inhibit all high-voltage-activated Ca2+ channels
8,9
, and is a potential PKA target
20
. Moreover, Rad-knockout mice display an increased basal Ca2+ current and the Ca2+
channels activate at lower voltages, mimicking the effects of β-adrenergic receptor
stimulation
21,22
. Other studies, however, led to expectations that Rad is not directly involved in
adrenergic regulation of CaV1.2 since over-expression of Rad
23
or Rem
24
in cultured cardiomyocytes markedly attenuated basal CaV1.2 current that was not rescued
by β-adrenergic stimulation. Nevertheless, our proximity labeling results elevated
Rad as the leading candidate for the critical missing link that enables PKA regulation
of CaV1.2.
PKA regulation of CaV1.2 requires phosphorylation of Rad
The robust heterologous reconstitution of PKA regulation of CaV1.2 currents has been
long pursued as a crucial step in identifying and validating the mechanism. We now
find that Rad was the missing ingredient. To prove this experimentally, we co-expressed
Rad with α1C and β2B subunits in HEK293T cells, limiting Rad expression by using 1:3
to 1:6 cDNA ratio of Rad:CaV1.2 subunits in order to avoid eliminating the Ca2+ current.
Perforated, whole-cell patch clamp was used to preserve the normal intracellular milieu
and signaling cascades, and minimize current run-down. In HEK293T cells transfected
with only α1C + β2B, superfusion of forskolin over 1-3 minutes had no impact on Ba2+
current (Fig. 3a-b, g-h). In contrast, applying forskolin to cells expressing α1C
+ β2B + Rad increased the maximal conductance (Gmax) by as much as 4.5-fold and by
a mean of 1.5-fold, and shifted the V50 for activation (Fig. 3c-d, g-h; Extended Data
Fig. 6a-b). The forskolin-induced increase in current was inversely proportional to
the basal current density, as observed in cardiomyocytes (Extended Data Figs. 1c,
6c). In HEK293T cells expressing 35-mutant α1C + 28-mutant β2B + Rad, applying forskolin
increased Gmax by as much as 3.1-fold and by a mean of 1.9-fold, and caused a hyperpolarizing
shift in the V50 for activation (Fig. 3i, Extended Data Fig. 6d-e). For both WT and
phospho-site mutant α1C and β2B subunits, the forskolin-induced enhancement of Ba2+
current in Rad-transfected cells was greatest at hyperpolarized potentials and fell
as the test potential approached the reversal potential (Extended Data Fig. 6f-g),
consistent with observations in cardiomyocytes
25
.
We used single-channel recordings to determine the mechanism of PKA/Rad modulation
of CaV1.2. In the absence of Rad, sweeps with no openings or blank sweeps are rare,
while most sweeps exhibit either intermediate or high levels of openings (Fig. 3j,
Extended Data Fig. 6h, k). In HEK293T cells transfected with α1C + β2B, co-expression
of PKAcat subunit had no effect on open probability (PO) (Fig. 3j). When Rad is expressed,
the fraction of blank sweeps is increased, low activity mode predominates, and the
PO is reduced (Fig. 3k, Extended Data Fig. 6i). By comparison, if the PKAcat subunit
is co-expressed with Rad, the fraction of blank sweeps is reduced, the high activity
mode resurges, and PO is increased by 10.6 ± 2.9-fold compared to transfection without
PKA (Fig. 3k, Extended Data Fig. 6j, l). These results suggest that Rad potently dampens
CaV1.2 current while phosphorylation of Rad allows channels to operate as though they
were devoid of Rad.
We identified 14 consensus PKA phosphorylation sites in Rad (Extended Data Fig. 7a),
which were mutated to Ala. This mutant Rad effectively inhibited CaV1.2 currents;
however, the cAMP-PKA mediated upregulation of CaV1.2 current was lost (Fig. 3g).
In lysates from forskolin-stimulated HEK293T cells transfected with GFP-Rad, we observed
phosphorylation of Ser25, Ser38 and Ser300 by mass spectrometry (Extended Data Fig.
7b). These residues were previously identified in the hearts of mice as phosphorylation
targets
26
(Extended Data Fig. 7c). We were unable to detect either non-phosphorylated or phosphorylated
peptides containing Ser272, notwithstanding prior biochemical studies identifying
Ser272 as a PKA target
20
. Alanine-substitutions of Rad at Ser25, Ser38, Ser272 and Ser300 (4-SA mutant, Extended
Data Fig. 7d) prevented the forskolin-induced increase in Gmax and hyperpolarizing
shift in the I-V curve (Fig. 3e, g-h). In contrast to transfection with WT Rad, co-transfection
of PKA with 4-SA mutant Rad failed to increase PO (PKA to no PKA: 1.15 ± 0.56-fold;
Fig. 3l).
A C-terminal polybasic region of 32 amino acids in Rad is involved in plasma membrane
targeting via binding to negatively charged phospholipids such as phosphatidylinositol
4,5-bisphosphate (PI(4,5)P2 or PIP2)
27
. Deletion of the C-terminus of RGK GTPases prevents their inhibition of Ca2+ channel
function
9
. We found that Ala-substitutions at Ser272 and Ser300 (2-SA mutant) within the C-terminal
polybasic membrane region prevented both the forskolin-induced increase in current
amplitude and the hyperpolarizing shift in the I-V curve (Fig. 3f-h; Extended Data
Fig. 6c).
Rad binding to β is required for PKA regulation of CaV1.2 channels
Rad can inhibit CaV1.2 via β-dependent and β-independent (α1C-dependent) mechanisms
28
. Either Ala-substitutions of Rad at residues R208 and L235, or Ala-substitutions
at residues D244, D320 and D322 of β2B (Extended Data Fig. 8a-b) attenuate Rad binding
to β subunits
28,29
. These mutations prevented the forskolin-induced increase in Gmax and the hyperpolarizing
shift in the I-V curve (Fig. 4a; Extended Data 8c-d). Thus, the Rad-β subunit interaction
is essential for cAMP-PKA regulation of CaV1.2.
We utilized flow-cytometry Förster resonance energy transfer (FRET) 2-hybrid assay
30
to probe potential PKA-mediated changes in the binding of β2B subunit and WT Rad.
At baseline, robust binding is detected between Cerulean-tagged β2B subunit and Venus-tagged
WT Rad, consistent with previous studies
31
(Kd,EFF = 7957 ± 418; Fig. 4b, Extended Data Fig. 9c). Co-expression of PKAcat subunit,
however, markedly weakened this interaction (Kd,EFF = 145231 ± 3084). In contrast,
co-expression of the PKAcat subunit in cells expressing fluorophore-tagged β2B and
4-SA mutant Rad had no effect on FRET binding (Kd,EFF = 4349 ± 138 versus Kd,EFF =
4346± 197 with and without PKA catalytic domain respectively; Fig. 4c, Extended Data
Fig. 9c). These results suggest that phosphorylation of Rad is required for dissociation
of the Rad-β2B interaction. In a similar manner, PKA phosphorylation of Rad also reduced
FRET binding to both β3 and β4 (Extended Data Fig. 9a-c).
Rad and Rem confer PKA regulation to CaV1.3 and CaV2.2 channels
In adrenal chromaffin cells and the sinus node cells of the heart, L-type CaV1.3 channels
are robustly stimulated by PKA
32,33
. In HEK293T cells transfected with only CaV1.3 α1D + β2B, superfusion of forskolin
over 1-3 minutes had no impact on Ba2+ current (Fig. 4d-f). In contrast, in cells
expressing α1D + β2B + Rad, applying forskolin increased Gmax by as much as 2.3-fold
and by a mean of 1.9-fold, and shifted the V50 for activation (Fig. 4d-f). We also
expressed the N-type CaV2.2 α1B subunit, which is widely expressed in neurons, with
β2B and Rad in HEK293T cells. Forskolin increased Gmax through CaV2.2 when co-expressed
with Rad by a mean of 2.2-fold and shifted the V50 for activation (Fig. 4g-i). For
both CaV1.3 and CaV2.2, attenuating binding of Rad to β prevented the forskolin-induced
modulation of CaV2.2 current (Extended Data Fig. 8e-f). We also expressed in HEK293T
cells the CaV2.2 α1B subunit with β2B and Rem, another member of the RGK GTPase family.
Forskolin increased Gmax through CaV2.2 when co-expressed with Rem by 1.6-fold and
shifted the V50 for activation (Fig. 4g-i). Thus, PKA-modulation of CaV channels is
not idiosyncratic as currently believed; rather, it is emerging to be a universal
mechanism transferable to all CaV channels that bind β subunits.
Discussion
The core α1C and β2B subunits, previously hypothesized to contain the PKA target sites
required for β-adrenergic agonist-induced stimulation of CaV1.2, do not. Rather, successful
reconstitution of regulation in a heterologous expression system required an additional
component, which we now identify as Rad. The cAMP-PKA regulation on CaV1.2 required
both the phosphorylation by PKA on the C-terminus of Rad and the interaction of Rad
with the β subunit. Multiple-alignment analysis of mouse Rad and other species shows
conservation of the four phosphorylation sites (Extended Data Fig. 10a). The required
interaction with the β subunit is consistent with our recent findings showing that
disrupting the α1C-β interaction prevented PKA regulation of CaV1.2 in heart
17
.
Analysis of Rad and other members of the RGK GTPase family indicate that their C-terminal
phosphorylation sites are highly similar (Extended Data Fig. 10b). Short stretches
of basic and hydrophobic amino acids are known to interact with the membrane
27
and phosphorylation of residues within these stretches alters their electrostatic
character, thereby reducing membrane affinity.
34
We found that phosphorylation of two Ser residues within the C-terminal polybasic
region of Rad released the Ca2+ channels from Rad-mediated inhibition, likely mediated
by reducing the affinity of Rad with the membrane and the CaVβ-subunit (Fig. 4j-k).
This mechanism of regulation is modular and transferable as CaV1.3 channels and neuronal
CaV2.2 channels are also imparted with forskolin-PKA mediated upregulation by Rad
and Rem. The activation of Ca2+ channels via release of inhibition by PKA phosphorylation
is reminiscent of PKA phosphorylation of phospholamban, an inhibitor of SERCA
35
.
Whereas phosphorylation of Ser1928 is definitively not required for β-adrenergic regulation
of CaV1.2 in the heart, Ala-substitution of Ser1928 prevents β-adrenergic stimulation
of Ca2+ channels in hippocampal neurons and hyperglycemia-induced stimulation of Ca2+
currents in arterial smooth muscle cells
36,37
. Perhaps, there are tissue-specific differences in CaV1.2 regulation. Phosphorylation
on the α1C subunit may also affect the trafficking and channel clustering in neurons
and cardiomyocytes
38,39
.
Our results indicate that proximity labeling using APEX2 is feasible in heart and
combined with multiplexed TMT proteomics can identify a dynamically evolving network
of interactions induced by β-adrenergic stimulation. This study establishes the utility
of proximity labeling in animals and provides an important foundation for future studies
that will investigate how diseases, such as heart failure, change the proteomic subdomain
of the excitation-contraction coupling machinery.
Augmented Ca2+ entry loads the cardiomyocyte with additional Ca2+ and enhances RyR
opening via Ca2+-induced Ca2+ release. This increases cardiac contractility. Dysregulation
of CaV1.2 activity can result in cardiac arrhythmias, heart failure and sudden death.
Supporting an important role of RGK GTPases in humans, a Rem2 variant was identified
as a genetic modifier in long QT syndrome 2
40
, and a Rad variant was recently linked to Brugada syndrome
41
. Our results identify potential targets and interaction sites for the therapeutic
modulation of β-adrenergic regulation of Ca2+ currents in the heart and other tissues.
For instance, disrupting the interaction between Rad and β subunit can be inotropic
by increasing Ca2+ entry in the heart. Conversely, interfering with the α1C-β interaction
17
, blocking PKA phosphorylation of Rad or potentially enhancing the interaction of
Rad with the plasma membrane could attenuate the sympathetic nervous system activation
of cardiac Ca2+ entry and inotropy more specifically than achievable with β-blockers.
Methods:
Clone construction and cell culture.
All mouse N-terminal-GFP-tagged mouse Rad (accession # XM_006531206) constructs in
a pEGFP-C1 vector were generated by gene-synthesis (Gene Universal). The human CaV2.2
(pSAD442-1) was a gift from Diane Lipscombe (Addgene plasmid # 62574; http://n2t.net/addgene:62574
; RRID:Addgene_62574). All cDNA clones were authenticated by sequencing.
HEK293T cells (ATCC, CRL-3216) were cultured in DMEM with 10% FBS, 1% Pen/Strep, and
transfected with 3 μg rabbit α1C (accession # X15539) or human α1B (for CaV2.2 experiments)
or rat α1D (for CaV1.3 experiments, accession # AF3070009), 3 μg human β2B (NM-201590.3)
and 0.5 μg N-terminal-GFP-tagged mouse Rad using Lipofectamine 2000 (Thermo Fisher
Scientific). The media was changed 4-6 hours after transfection. The cells were split
onto coverslips coated with attachment factor protein (Gibco). Electrophysiological
recordings were carried out at room temperature 24-48 hours after transfection. The
cells were authenticated by ATCC and tested for mycoplasma contamination.
Transgenic mice generation:
The Institutional Animal Care and Use Committee at Columbia University approved all
animal experiments. Male and female mixed-strain mice, 6-weeks to 4-months of age
were used. Sample sizes exceeded the number of samples determined by power calculations,
which were based on effect sizes previously reported
16,17,42,43
. Number of animals was always greater than 3 per genotype. The investigators were
blinded to group allocation during data acquisition and analysis.
The transgenic constructs were generated by fusing rabbit α1C cDNA or human β2B cDNA
to the modified murine α-myosin heavy chain (MHC) tetracycline-inducible promoter
vector (gift of Jeffrey Robbins and Jeffrey Molkentin, University of Cincinnati, Cincinnati,
OH)
44,45
. A 3X FLAG-epitope was ligated in-frame to the N-terminus of α1C. The α1C subunit
was engineered to be dihydropyridine (DHP)-insensitive with the substitutions T1066Y
and Q1070M
46,47
. GFP was ligated to the N-terminus of β2B. The creation of GFP-WT β2B transgenic
mice was described previously
17
. The V5 epitope and APEX2 cDNA
5,48
, created by gene synthesis, were conjugated to the N-terminus of DHP-resistant-α1C
and WT β2B.
The 35-α mutant and the 28-β mutant cDNA were generated by site-directed mutagenesis.
The optimal PKA phosphorylation motif is a tetrapeptide with Arg at the 2nd and 3rd
positions (termed −2 and −3) prior to the phosphorylated Ser or Thr, and a large hydrophobic
residue immediately thereafter (R-R-X-S/T-Φ)
49-51
. The requirement for a positive charge is highest for residues at -2 and -3, but
can be found for residues as far as position -6 in PKA target sites
52
. Sites with Arg in positions between −4 and −1 are strongly preferred and to a lesser
extent His or Lys
50,53
. We identified all potential intracellular PKA phosphorylation sites (Extended Data
Fig. 1a, d) in rabbit α1C and human β2B using both manual sequence analysis and several
web-based PKA phosphorylation prediction tools, including pKaPS
54
, DISPHOS
55
, GPS
56
, NETPHOS
57
and SCANSITE
58
. Each phosphorylation site was mutated to Ala. We also mutated additional Ser and
Thr within several amino acid residues C-terminal to the Arg or Lys, in order to ensure
that we fully captured each phospho-regulatory site. 51 residues in rabbit α1C were
replaced with Ala at 35 potential phospho-regulatory domains in the 35-mutant α1C
construct, and 37 residues were replaced with Ala at 28 putative phospho-regulatory
domains of β2B. We excluded those sites predicted to be extracellular or within the
plasma membrane.
Transgenic mice with non-targeted insertion of these tetracycline-regulated cDNA (Fig.
1b) were bred with cardiac-specific (α-MHC), doxycycline-regulated, codon-optimized
reverse transcriptional trans-activator (rtTA) mice
59
(obtained via the Mutant Mouse Resource and Research Center) to generate double-transgenic
mice. For the α1C-APEX2 and β2B-APEX2 mice, transgene expression did not require doxycycline
due to a low basal binding of rtTA protein to the Tet operator sequences (so-called
“leak”)
59
. These expression levels result in Ca2+ current levels similar to native conditions
in heart. The results presented were consistent across all founder lines and gender,
and therefore were pooled.
Isolation of adult cardiac myocytes:
Mice ventricular myocytes were isolated by enzymatic digestion using a Langendorff
perfusion apparatus as previously described
16,17,42,43,60
. Cardiomyocytes were isolated from 8-12 week-old non-transgenic and transgenic mice.
Proximity labeling biotinylation:
Proximity labeling was performed as described
48
with minor modifications. Isolated ventricular cardiomyocytes were incubated in labeling
solution with 0.5 μM biotin-phenol (Iris-biotech) for 30 minutes. For some experiments,
during the final 10 minutes of labeling, 1 μM isoproterenol (Sigma I5627) was added.
To initiate labeling, H2O2 (Sigma H1009) was added to a final concentration of 1 mM
for 1 minute. Exactly 1 minute after H2O2 treatment, the labeling solution was decanted
and cells were washed three times with cold quenching solution containing (in mM)
10 Sodium ascorbate (VWR 95035-692), 5 Trolox (Sigma 238813), and 10 Sodium azide
(Sigma S2002). After cells were harvested by centrifugation, the quenching solution
was aspirated and the pellet was flash-frozen and stored at −80 °C until streptavidin
pull-down.
For biotinylation in Langendorff-perfused hearts, mice were injected with 5 mg/kg
Propranolol-HCl (Sigma PHR1308) to suppress adrenergic stimulation during the isoflurane
anesthesia and cardiectomy. Hearts were retrograde perfused with Krebs’ solution for
10 minutes prior to addition of biotin-phenol for 15 minutes. During the final 5 minutes,
1 μM Isoproterenol (Sigma 1351005) or vehicle was added to the perfusate. Electrocardiograms
were monitored throughout the experiment to ensure viability of the preparation and
an isoproterenol-induced increase in heart rate.
The cells or whole heart tissue were lysed with a hand-held tip homogenizer in a solution
containing (in mM), 50 Tris (tris(hydroxymethyl)aminomethane) , 150 NaCl, 10 EGTA,
10 EDTA, 1% Triton X-100 (v/v), 0.1% SDS (w/v), 10 Sodium ascorbate, 5 Trolox, and
10 Sodium azide, phosphatase inhibitors (Sigma 4906845001), protease inhibitors (Sigma
4693159001), Calpain inhibitor I (Sigma A6185) and Calpain inhibitor II (Sigma A6060).
Biotin labeling of the samples was confirmed after size-fractionation of proteins
on SDS-PAGE, transfer to nitrocellulose membranes, and probing with Streptavidin-HRP
(Thermofisher, S911, Lot# 1711896, 0.6mg/ml). The response to isoproterenol was assessed
by immunoblotting with an anti-phospho-phospholamban (Ser16/Thr17) antibody (Cell
Signaling, #8496, Lot#1; 1:1000 dilution).
Immunoprecipitation and Immunoblots:
Cardiomyocytes were lysed with a hand-held tip homogenizer in a 1% (v/v) Triton X-100
buffer containing (in mM): 50 Tris-HCl (pH 7.4) 150 NaCl, 10 EDTA, 10 EGTA and protease
inhibitors. The lysates were incubated on ice for 30 minutes, centrifuged at 14,000
RPM at 4°C for 10 minutes and supernatants collected. Proteins were size-separated
on SDS-PAGE, transferred to nitrocellulose membranes, and probed with anti-V5 antibody
(Thermofisher, R960-25; 1:5000 dilution), a custom-made polyclonal anti-α1C antibody
(Yenzym, 1:1000 dilution)
61,62
, a custom-made polyclonal anti-β antibody (epitope: mouse residues 120-138: DSYTSRPSDSDVSLEEDRE;
Yenzym, 1:1000 dilution), an anti-JPH2 antibody (Pierce, PA5-20642, lot# NG1583142;
1:1000 dilution), an anti-calmodulin antibody (Millipore Sigma 05-173; 1:1000 dilution),
a custom-made anti-RyR2 antibody (1:5000 dilution)
63
, and an anti-KV1.5 antibody (Alomone, APC-150, Lot# APC004AN0850; 1:1000 dilution).
Anti-FLAG antibody (Sigma, F7425, Lot#078M4886V) immunoprecipitations were performed,
as previously described
17
, overnight in a lysis buffer consisting of (in mM): 50 Tris-HCl pH 7.4, 150 NaCl,
0.25% Triton X-100 (v/v), 10 EDTA, 10 EGTA, 0.01 Calpain inhibitor I, 0.01 Calpain
inhibitor II, and Complete protease inhibitors (1 per 7 ml, Roche). Antibody-protein
complexes were collected using protein A conjugated agarose (Amersham) for 2 h, followed
by 3 washes in lysis buffer. Proteins were size-separated by SDS, transferred to nitrocellulose
membranes and probed with HRP-conjugated anti-FLAG antibody (Sigma, A8592), a custom-made
anti-β antibody and HRP-conjugated secondary goat anti-rabbit antibody. Phospholamban
phosphorylation was detected with an anti-phospho-PLB antibody. Detection of luminescence
was performed with a CCD camera (Carestream Imaging). The uncut gels are shown in
Supplementary Fig. 1.
Immunofluorescence:
Isolated cardiomyocytes were first exposed to biotin-phenol and H2O2 as described
above. After quenching, the cells were fixed for 15 minutes in 4% paraformaldehyde,
washed with glycine/PBS (phosphate-buffered saline) twice, treated with PBST (0.1%
Triton X-100 (v/v) in PBS) for 5 minutes, and blocked with 3% BSA (w/v) in PBS for
1 hour. Indirect immunofluorescence was performed using dilutions of 1:500 of anti-V5
antibody (Thermofisher, R960-25) and 1:200 of Alexa594-labeled goat-anti-mouse antibody
(Thermofisher, A11032, Lot#2069816), and 1:800 of streptavidin-Alexa Fluor 488 conjugate
(Thermofisher, S32354, Lot# 1719656). Images were acquired using a confocal microscopy.
Processing of biotinylated proteins for mass spectrometry:
Proteins were prepared as previously described
7
with some modifications. A detailed protocol of the procedures used was published
recently
64
. Proteins were precipitated with Trichloroacetic acid (TCA; Sigma T9159) and then
centrifuged at 21,130 × g at 4 °C for 10 minutes. The pellet was washed with −20 °C
cold acetone (Sigma 650501), vortexed, and centrifuged at 21,130 × g at 4 °C for 10
minutes. Following centrifugation, acetone was aspirated and the pellet was acetone-washed
again three more times. After the last washing step, the pellet was resuspended in:
8M urea, 100 mM sodium phosphate pH 8, 100 mM NH4HCO3, and 1% SDS (w/v) and rotated
at room temperature until fully dissolved. Re-suspended proteins were centrifuged
at 21,130 × g at room temperature for 10 minutes and the cleared supernatant was transferred
to a new microcentrifuge tube. To reduce disulfides, 10 mM TCEP-HCl (Thermo Fisher
Scientific PG82089) in Milli-Q water titrated to pH 7.5 with NaOH was added. To alkylate
free Cys, freshly prepared 400 mM iodoacetamide (Thermo Fisher Scientific 90034) stock
solution in 50 mM ammonium bicarbonate was added to the supernatant to a final concentration
of 20 mM, immediately vortexed, and incubated in the dark for 25 minutes at room temperature.
After alkylation, freshly prepared DTT (dithiothreitol) stock solution was added to
50 mM final concentration to quench alkylation. Water was added to each sample to
reach a final concentration of 4 M urea and 0.5% (w/v) of SDS.
A 100 μL suspension equivalent per sample of streptavidin magnetic beads (Thermo Fisher
Scientific #88817) was washed twice with 4 M urea, 0.5% SDS (w/v), 100 mM sodium phosphate
pH 8 and was added to each sample (about 1 mg), diluting each sample with an equal
amount of water to reach a final concentration of 2 M urea, 0.25% SDS (w/v), 50 mM
sodium phosphate pH 8 during pulldown. The tubes were rotated overnight at 4 °C. Following
streptavidin pull-down, the magnetic beads were washed three times with 4 M urea,
0.5% SDS (w/v), 100 mM sodium phosphate pH 8, and three times with the same buffer
without SDS. The beads were transferred to new tubes for the last wash step. Before
final pulldown of the beads for mass spectrometry analysis, 5% of beads resuspension
was removed for streptavidin-HRP blotting.
On-bead digestion and TMT labeling:
The samples were prepared as previously described
7
. Liquid reagents used were HPLC quality grade. Washed beads were re-suspended in
50 μL of 200mM EPPS (4-(2-Hydroxyethyl)-1-piperazinepropanesulfonic acid) buffer pH
8.5, 2% acetonitrile (v/v) with 1 μL of LysC stock solution (2 mg/ml, Wako), vortexed
briefly and incubated at 37°C for 3 hr. Then, 50 μL of trypsin stock (Promega #V5111)
diluted 1:100 (v/v) in 200 mM EPPS pH 8.5 was added. After mixing, digests were incubated
at 37°C overnight and beads were magnetically removed. Peptides were directly labeled
after digest. For this, acetonitrile was added to a concentration of 30% (v/v) and
peptides were labeled with TMT 10-plex or 11-plex reagents (Thermo Fisher Scientific
#90406 and #A34807) for 1 hr. Reactions were quenched with hydroxylamine at a final
concentration of 0.3% (v/v) for 15 minutes and 1% of labeled peptides were analyzed
for efficiency of label incorporation and relative ratios by mass spectrometry. After
quenching, peptide solutions were acidified with formic acid, trifluoroacetic acid
(TFA) was added to a concentration of 0.1% and peptides were desalted and fractionated
by high pH reversed phase chromatography (Thermo Fisher Scientific #84868). After
loading of labeled peptides onto pre-conditioned columns and a single wash with water,
excess unincorporated TMT label was removed by washing reversed phase columns once
with 0.1% trimethylamine (TEA) buffer containing 5% acetonitrile. Samples were fractionated
under alkaline conditions into 12 fractions with increasing concentrations of acetonitrile:
10%, 12.5%, 15%, 17.5%, 20%, 25%, 30%, 35%, 40%, 50%, 65% and 80%. Fractions 1 and
7, 2 and 8, 3 and 9, 4 and 10, 5 and 11, 6 and 12 were pooled to obtain 6 final pooled
fractions for subsequent analysis. Pooled fractions were dried to completion and further
purified and desalted by acidic C18 solid phase extraction (StageTip). Labeled peptides
were finally re-suspended in 1% formic acid (v/v) and 3% acetonitrile (v/v).
Whole cell proteomics: Cell lysis, protein digest and TMT labeling mass spectrometry
analysis:
Cells were lysed by homogenization (QIAshredder cartriges, Qiagen) in lysis buffer
(2% SDS, 150 mM NaCl, 50 mM Tris pH 7.4). Lysates were reduced with 5 mM DTT, alkylated
with 15 mM iodoacetamide for 30 minutes in the dark, alkylation reactions quenched
with freshly prepared DTT added to a concentration of 50 mM and proteins precipitated
by methanol/chloroform precipitation. Digests were carried out in 1M urea freshly
prepared in 200 mM EPPS pH 8.5 in presence of 2% acetonitrile (v/v) with LysC (Wako,
2 mg/ml, used 1:75 w/w protease:substrates during digest) for 3 hours at room temperature
and after subsequent addition of trypsin (Promega #V5111, stock 1:100 w/w protease:substrates)
over night at 37°C. Missed cleavage rate was assayed from a small aliquot by mass
spectrometry. For whole proteome analysis, digests containing approximately 60 μg
of peptide material were directly labeled with TMT reagents (Thermo Fisher Scientific).
Labeling efficiency and TMT ratios were assayed by mass spectrometry, while labeling
reactions were stored at −80°C. After quenching of TMT labeling reactions with hydroxylamine,
TMT labeling reactions were mixed, solvent evaporated to near completion and TMT labeled
peptides purified and desalted by acidic reversed phase C18 chromatography. Peptides
were then fractionated by alkaline reversed phase chromatography into 96 fractions
and combined into 24 samples.
Mass spectrometry analysis:
Data collection followed a MultiNotch MS3 TMT method
65
using an Orbitrap Lumos mass spectrometer coupled to a Proxeon EASY-nLC 1200 liquid
chromatography (LC) system (both Thermo Fisher Scientific). The capillary column used
was packed with C18 resin (35 cm length, 75 μm inner diameter, matrix 2.6 μm Accucore
(Thermo Fisher Scientific)). Peptides of each fraction were separated for 4 hours
over acidic acetonitrile gradients by LC prior to mass spectrometry (MS) analysis.
The scan sequence started with an MS1 scan (Orbitrap analysis; resolution 120,000;
mass range 400–1400 Th). MS2 analysis followed collision-induced dissociation (CID,
CE=35) with a maximum ion injection time of 150-300 ms and an isolation window of
0.4 m/z. In order to obtain quantitative information, MS3 precursors were fragmented
by high-energy collision-induced dissociation (HCD) and analyzed in the Orbitrap at
a resolution of 50,000 at 200 Th. Further details on LC and MS parameters and settings
used were described recently
66
.
Peptides were searched with a SEQUEST (v.28, rev. 12) based software against a size-sorted
forward and reverse database of the M. musculus proteome (Uniprot 07/2014) with added
common contaminant proteins. For this, spectra were first converted to mzXML. Searches
were performed using a mass tolerance of 20 ppm for precursors and a fragment ion
tolerance of 0.9 Da. For the searches maximally 2 missed cleavages per peptide were
allowed. We searched dynamically for oxidized methionine residues (+15.9949 Da) and
where indicated for phospho-modification of S,T and Y residues (+79.9663 Da). We applied
a target decoy database strategy and a false discovery rate (FDR) of 1% was set for
peptide-spectrum matches following filtering by linear discriminant analysis (LDA)
26,66
. The FDR for final collapsed proteins was 1%. MS1 data were calibrated post search
and searches performed again. A modified version of the Ascore algorithm was used
to quantify the confidence assignment of phosphorylation sites. Phosphorylation localized
to particular residues required Ascore values > 13 (p ≤ 0.05) for confident localization
26
. Quantitative information on peptides was derived from MS3 scans. Quant tables were
generated requiring an MS2 isolation specificity of >70% for each peptide and a sum
of TMT signal to noise (s/n) of >200 over all channels for any given peptide and exported
to Excel and further processed therein. Details of the TMT intensity quantification
method and further search parameters applied were described previously previously
67
. Proteomics raw data and search results were deposited in the PRIDE archive
68,69
and can be accessed under ProteomeXchange accession nunbers: PXD014499, PXD014500,
PXD014501.
The relative summed TMT s/n for proteins between two experimental conditions (referred
to as “enrichment” in the text) was calculated from the sum of TMT s/n for all peptides
quantified of a given protein. For gene ontology (GO) term enrichment, the BINGO package
in Cytoscape was used
70
. Scaled quantification data were subjected to two-way clustering (JMP software package)
and changes in enrichment were analyzed using Graphpad Prism 8 (Graphpad Software).
Statistical significance was determined by multiple t-tests without correction for
multiple comparisons and α=0.05. Data for relative protein quantification accompany
this paper in the form of Supplementary Data Tables 1,3-6, and can be accessed online.
Purification of GFP-conjugated Rad and phosphoproteomic analysis:
HEK293T cells were cultured in DMEM with 10% FBS and 1% Pen-Strep. Cells were transfected
with GFP-Rad using lipofectamine 2000 as described above. Media was changed 4-6 hours
after transfection. After 24 hours, cells were treated with trypsin and spun down
for 5 minutes at 1000 rpm. Cells were then resuspended in PBS with 10 μM forskolin
for 5 minutes. After washing cell pellets three times with phosphate-buffered saline
(PBS), cells were frozen at −80°C. Cold cell pellets were lysed in PBS with 0.1% triton
X-100 (v/v, Sigma) and a phosphatase inhibitor mixture (PhosSTOP, Roche) by pipetting
up and down several times. Lysates were homogenized by passing them through QIAshredder
cartridges (Qiagen) and incubated with GFP-trap agarose beads (Chromotek, Germany)
for 4 hours at 4°C with constant rotation. Beads were washed 3 times with PBS with
0.1% (v/v) Triton and three times with detergent-free PBS and subjected to on-bead
digest with trypsin (Promega #V5111), LysC (Wako) or ArgC (Promega #V1881, ArgC digestion
buffer 50 mM Tris-HCL pH 7.8, 5 mM CaCl2, 2 mM EDTA, 2% acetonitrile (v/v)) separately
as described above overnight at 37°C. After acidification, peptides were purified
by reversed phase C18 chromatography and subjected to MS/MS analysis. For this, the
same parameters as above for MS1 and MS2 scans were used with an isolation window
of 1.2 Da and taking neutral loss of 97.9763 Da into account with multi-stage activation
(MSA) set for MS2 scans. Analysis of phospho-site localization was performed as described
above.
Fractional shortening:
Freshly isolated myocytes were perfused with a Tyrode’s solution containing 1.8 mM
CaCl2. Myocytes were field stimulated at 1-Hz. Nisoldipine (300 nM) dissolved in Tyrode’s
solution was then superfused. Fractional shortening of sarcomere length was measured
using the SarcLen module of Ionoptix.
Whole cell patch clamp electrophysiology:
Isolated cardiomyocytes or HEK cells on glass 8×8 mm coverslips were placed in Bioptechs
Delta T Dishes filled with solution containing (in mM): 112 NaCl, 5.4 KCl, 1.7 NaH2PO4.
1.6 MgCl2, 20.4 HEPES (pH 7.2), 30 Taurine, 2 DL-Carnitine, 10.3 Creatine, 5.4 Glucose.
The petri dishes were mounted on the stage of an inverted microscope and served as
a perfusion chamber. After establishing a seal and achieving whole-cell configuration,
external solutions were changed by fast local perfusion method.
For cardiomyocytes, pipette resistance was between 1-3 MΩ. Membrane currents were
measured by conventional (ruptured) whole-cell patch-clamp method using a MultiClamp
700B or Axopath200B amplifier and pCLAMP 10.7 software (Molecular Devices). Capacitance
transients and series resistance were compensated. Voltage was corrected for liquid
junction potential (−10 mV) during analysis. Leak currents were subtracted by a P/4
protocol. The parameters of voltage-dependent activation were obtained using a modified
Boltzmann distribution: I(V)= Gmax * (V-Erev)/[1 + exp(V-V50)/Vc)], where I(V) is
peak current, Gmax is maximal conductance, Erev is reversal potential, V50 is the
midpoint, and Vc is the slope factor.
The pipette solution contained (in mM): 40 CsCI, 80 Cesium Gluconate, 10 BAPTA (1,2-bis(o-aminophenoxy)ethane-
N,N,N′, N′-tetraacetic acid), 1 MgCl2, 4 Mg-ATP, 2 CaCl2, and 10 HEPES (4-(2-hydroxyethyl)-1-piperazineethanesulfonic
acid), adjusted to pH 7.2 with CsOH. After the isolated cardiomyocytes were dialyzed
and adequately buffered with 10 mM BAPTA in the internal solution, cells were superfused
with (in mM): 140 TEA-Cl (Tetraethylammonium chloride), 1.8 CaCl2, 1 MgCl2, 10 glucose,
and 10 HEPES, adjusted to pH 7.4 with CsOH. To measure peak currents, the cell membrane
potential was held at −50 mV and stepped to +0 mV for 350 ms every 10 s. To evaluate
the current-voltage (I-V) relationship in cardiomyocytes, the same protocol was repeated
with steps between −40 mV to +60 mV in 10 mV increments. Nisoldipine (Santa Cruz)
was stored protected from light at −20 °C as 3 mM stock in ethanol. The final dilution
of nisoldipine to 300 nM was in the extracellular recording solution immediately prior
to the experiment. Isoproterenol (Sigma I5627) and Forskolin (Santa Cruz) were prepared
daily and diluted in extracellular solution.
For HEK293T cell experiments involving the Forskolin-induced stimulation of CaV1.2,
CaV1.3 and CaV2.2 currents, perforated whole-cell patch clamp technique was implemented
to minimize current run-down and preserve the intracellular milieu. Amphotericin B
(Sigma A9528) was initially dissolved in DMSO (20 mg/ml) and used in the pipette solution
at a final concentration of 200 μg/ml. The tip of pipette was filled with amphotericin-free
solution (in mM): 80 Cesium Gluconate, 40 CsCI, 10 HEPES, 10 BAPTA, 1 MgCI2, 1 Mg-ATP,
pH adjusted to 7.2 with CsOH. The pipette was backfilled with (in mM): 125 CsCI, 10
HEPES, 4 CaCI2, 1 MgCI2, pH-7.2 with CsOH; containing amphotericin, 200 μg/ml. CaCI2
(4 mM) was added to the patch electrode solution to enable the detection of conversion
from perforated to ruptured configuration. The external solution contained (in mM):
130 tetraethylammonium methanesulfonate, 10 HEPES, 1 MgCI2,10 (with Rad expression)
or 2 (without Rad expression) BaCI2, 5 Glucose. For experiments with HEK293T cells,
in addition to step protocols, we used a ramp protocol with a 200-ms voltage ramp
from −60 to +60 mV (0.6 V/s) applied every 10-seconds to monitor the I-V relationship.
All experiments were performed at room temperature, 22 ± 1° C. Cells were selected
based on co-transfection of a vector containing GFP in the absence of Rad, and GFP-conjugated
Rad for the experiments with Rad transfected. For both cardiomyocytes and HEK293T
cells, cells without a stable baseline (potentially due to run-down or run-up) were
not studied.
Voltage steps protocol used in cardiomyocytes studies evaluated IIpeak = Ipeak (V),
which was recalculated in CLAMPFIT to G=G(V) as G=I/(V-Erev). For HEK cells experiments,
we used a ramp protocol (Extended Data Fig. 8b). The experimental records were I=I(t),
where t=time. We transformed these records to I=I(V), which was then further transformed
to G=G(V) in CLAMPFIT (see Extended Data Fig. 8b). The parameters of voltage-dependent
activation were obtained using a modified Boltzmann distribution. A nonlinear fit
with Boltzmann sigmoidal function was done in Prism (GraphPad) with goodness of fit:
R2>0.99.
Single channel patch clamp electrophysiology:
Cell-attached single-channel recordings were performed at room temperature as previously
described
71,72
. Patch pipettes (5–10 MΩ) were pulled from ultra-thick-walled borosilicate glass
(BF200-116-10, Sutter Instruments), and coated with Sylgard. Currents were filtered
at 2 kHz. The pipette solution contained (in mM): 140 tetraethylammonium methanesulfonate;
10 HEPES; 40 BaCl2; at 300 mOsm/L, adjusted with tetraethylammonium methanesulfonate;
and pH 7.4 adjusted with tetraethylammonium hydroxide. To maintain the membrane potential
at 0 mV, the bath contained (in mM): 132 potassium glutamate; 5 KCl; 5 NaCl; 3 MgCl2;
2 EGTA (ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid); 10 glucose;
20 HEPES; at 300 mOsm/L adjusted with glucose; and pH 7.4 adjusted with NaOH. Cell-attached
single-channel currents were measured during 200 ms voltage ramps between −80 to +70
mV (portions between −50 and 40 mV displayed and analyzed). For each patch, we recorded
80-120 sweeps with a repetition interval of 10 seconds.
The sample size for the Po-V relationships:
α1C
n = 10 (3 transfections)
933 sweeps
α1C +PKA
n = 5 (2 transfections)
450 sweeps
Rad
n = 5 (3 transfections)
372 sweeps
Rad+PKA
n = 8 (3 transfections)
527 sweeps
Rad Mut
n = 6 (4 transfections)
388 sweeps
Rad Mut+PKA
n = 5 (3 transfections)
439 sweeps
The modal analysis are histograms of single-trial average Po values obtained from
one-channel patches. For Po-V analysis, we conservatively corrected for shifts in
voltage. The number of sweeps is as follows:
α1C alone:
281 sweeps (from 3 one-channel patches)
Rad:
372 sweeps (from 5 one-channel patches)
Rad + PKA:
311 sweeps (from 3 one-channel patches)
Flow Cytometric FRET 2-hybrid assay:
For flow cytometric FRET assay, HEK293 cells (ATCC CRL1573) were cultured in 12 well
plates and transfected with polyethylenimine (PEI) 25 kDa linear polymer (Polysciences
#2396602). Briefly, 1.5 μg of Cerulean (Cer) and Venus (Ven) -tagged cDNA pairs were
mixed together in 100 μl of serum-free DMEM media and 6 μl of PEI is added to each
sterile tube. FRET experiments were performed 1-day post-transfection. Protein synthesis
inhibitor cycloheximide (100 μM) was added to cells 2-hrs prior to experimentation
to halt synthesis of new fluorophores to allow existing fluorophores to fully mature.
For FRET measurements, we utilized an LSR II (BD Biosciences) flow cytometer equipped
with 405 nm, 488 nm, and 633 nm lasers for excitation and 18 different emission channels.
Forward and side scatter signals were detected and used to gate for single, and healthy
cells. For determining FRET efficiency, we measured three distinct fluorescence signals:
(1) S
Cer corresponding to cerulean emission is measured through the BV421 channel (excitation,
405 nm; emission, 450/50), (2) S
Ven corresponding to venus emission measured via FITC channel (excitation, 405 nm;
dichroic, 505LP; emission, 525/50), and (3) S
FRET corresponding to FRET emission via the BV510 channel (excitation, 405 nm; dichroic,
505LP; emission, 525/50). These raw fluorescence measurements are subsequently used
to obtain Vendirect (venus emission due to direct excitation), Cerdirect (cerulean
emission due to direct excitation), and VenFRET (venus emission due to FRET excitation).
Flow cytometric signals were collected at medium flow rate (2k-8k events/sec). Fluorescence
data was exported as FCS 3.0 files for further processing and analysis using custom
MATLAB functions (MathWorks).
For each experimental run on the flow cytometer, we performed several control experiments:
(1) Background fluorescence level for each fluorescent channel (BG
Cer, BG
Ven, and BG
FRET) was obtained by measuring fluorescence from cells exposed to PEI without any
fluorophore containing plasmids. (2) Cells expressing Ven fluorophore alone were utilized
to measure spectral crosstalk parameter R
A1, corresponding to bleed-through of Ven fluorescence into the FRET channel. (3)
Cells expressing cerulean fluorophore alone were used to measure spectral crosstalk
parameter R
D1 and R
D2, corresponding to bleed-through of cerulean fluorescence into FRET and Ven channels
respectively. (4) FRET measurements also require determination of instrument-specific
calibration parameters gVen / gCer and f
Ven / f
Cer, corresponding to ratios of fluorescence excitation and emission for Ven to cerulean
fluorophores. These parameters also incorporate fluorophore-dependent aspects including
molar extinction (for g) and quantum yield (for f) as well as instrument specific
parameters including laser power, attenuation by optical components, as well as photodetection,
amplification, and digitization of fluorescence. For determination of these parameters,
we utilized Cer-Ven dimers with four different linker lengths (5, 32, 40, and 228).
(5) Co-expression of cerulean and Ven fluorophores provided estimates for concentration-dependent
collisional FRET.
For our experiments, R
A1 ~ 0.11, R
D1 ~ 2.8, and R
D2 ~ 0.006. We observed only minor day-to-day variation in these measurements. For
each cell, spectral cross-talk was subtracted as follows:
C
e
r
direct
=
R
D
1
⋅
S
Cer
,
V
e
n
direct
=
R
A
1
⋅
(
S
Ven
−
R
D
2
⋅
S
Cer
,
and
V
e
n
FRET
=
S
FRET
−
R
A
1
⋅
(
S
Ven
−
R
D
2
⋅
S
Cer
)
−
R
D
1
⋅
S
Cer
.
Following spectral unmixing, we obtained g
Ven / g
Cer and f
Ven / f
Cer from data for Cer-Ven dimers by determining the slope and intercept for the following
linear relationship:
V
e
n
F
R
E
T
C
e
r
direct
=
g
Cer
g
V
e
n
•
V
e
n
d
i
r
e
c
t
C
e
r
direct
−
f
V
e
n
f
C
e
r
For typical experiments, g
Ven / g
Cer = 0.0194 and f
Ven / f
Cer = 2.3441. Having obtained these calibration values, donor-centric FRET efficiencies
were computed as,
E
D
=
V
e
n
FRET
V
e
n
FRET
+
f
V
e
n
f
C
e
r
•
C
e
r
direct
For Cer-Ven dimers, FRET efficiencies of ~ 0.55, 0.38, and 0.05 were obtained for
linker lengths 5, 32, and 228 respectively, matching previous work
30
. The relative proportion of Cerulean and Venus fluorophores in each cell was determined
as N
Cer = Cerdirect / (1 - E
D) and N
Ven = Vendirect / (g
Ven / g
Cer × f
Ven / f
Cer ). To construct FRET 2-hybrid binding curves, we imposed a 1:1 binding isotherm
as in previous studies
30
. For each FRET pairs, K
d,EFF and E
D,max and 95% - confidence intervals were obtained by constrained least squares fit.
For Fig. 4b-c and Extended Data Fig. 9a-b, we show all the individual cells from two
different transfections. We fit these data with the equation: ED = Emax*[Ven-Rad]free/([Ven-Rad]free
+ kdeff) using least squares algorithm (https://www.mathworks.com/help/stats/nlinfit.html).
The Kdeff fit value based on all the data points is shown as the bar in Extended Data
Fig. 9c. The error is calculated by the fitting algorithm as a 95%-confidence interval
on the fit parameter: https://www.mathworks.com/help/stats/nlparci.html).
Statistical Analysis:
Results are presented as mean ± SEM. For comparisons between two groups, Student’s
t-test was used. Statistical analyses were performed using Prism 8 (Graphpad Software).
For multiple group comparisons, a one-way ANOVA followed by either Dunnett’s or Tukey’s
post-hoc test were performed using Prism 8. Differences were considered statistically
significant at values of P < 0.05.
Extended Data
Extended Data Fig. 1.
Putative PKA phosphorylation sites in α1C and β2B subunit.
(a) The 35 putative PKA phosphorylation sites in rabbit α1C. The 51 residues in red
are either predicted phosphorylation sites or within the immediate region of the predicted
phosphorylation sites. All 51 residues were replaced with Ala in the 35-mutant α1C
transgenic mice. (b) Combined bar and column scatter plot of Boltzmann function parameters
V50. Mean ± SEM. ** P <0.01; *** P <0.001; **** P <0.0001 by paired two-tailed t-test.
pWT α, n= 19; 35-α mutant, n= 14; 28-β mutant, n= 16; 35-α mutant X 28-β mutant, n=
24. Specific P values can be found in the Source Data associated with this figure.
(c) Graph of isoproterenol and forskolin-induced increase in nisoldipine-resistant
current stratified by total basal current density before nisoldipine. (d) The 28 putative
PKA phosphorylation sites in the N-terminal (NT), Hook, GK and C-terminal (CT) domains
of β2B. The 37 residues in red are either predicted phosphorylation sites or within
the immediate vicinity of predicted phosphorylation sites, and were mutated to Ala
in the 28-mutant GFP-tagged β2B transgenic mice. (e) Fluorescence imaging of isolated
cardiomyocytes expressing GFP-tagged 28-β mutant. Representative of images from more
than 5 biologically independent mice. (f) Anti-β antibody immunoblot of cleared lysates
from doxycycline-fed 35-mutant α1C transgenic mice or 35-mutant α1C X GFP-tagged 28-mutant
β2B expressing mice hearts. Representative of immunoblots obtained from at least 3
biologically independent mice. (g) Anti-FLAG antibody (upper) and anti-β antibody
(lower) immunoblots of anti-FLAG antibody immunoprecipitations from cleared lysates
of hearts from pWT, 35-α and three 35-α X GFP-tagged-28-β expressing mice. Representative
images of two independent experiments. For source gel data, see Supplementary Fig.
1.
Extended Data Fig. 2.
Trafficking and function of APEX2-conjugated CaV1.2 channels subunits in heart.
(a) Exemplar current-voltage relationship of Ca2+ currents from α1C-APEX2 mice cardiomyocytes
acquired in absence (black trace) and presence of nisoldipine (red trace). Inset scale
bars: horizontal 100 ms, vertical 10 pA/pF. Representative of 5 experiments. (b) Time
course of changes in sarcomere length after superfusion of nisoldipine-containing
solution. Representative of 7 experiments (c) Percent-shortening in the presence of
nisoldipine. Mean ± SEM. ****, P <0.0001 by unpaired two-tailed t-test. n=12, 7 cardiomyocytes,
respectively. (d) Immunofluorescence of cardiomyocytes isolated from α1C-APEX2 and
β2B-APEX2 expressing mice exposed to biotin-phenol and H2O2 or no H2O2. Nuclear labeling
with DAPI stain. Scale bar = 5 μm. Representative of 5 and 8 cardiomyocytes from 2
and 3 mice respectively. (e) Streptavidin-horseradish peroxidase (HRP) blot of lysates
of isolated ventricular cardiomyocytes. Blot representative of 5 similar experiments.
(f) Exemplar whole-cell CaV1.2 currents recorded from cardiomyocytes of α1C-APEX2
transgenic mice. Black trace: 300 nM nisoldipine; blue trace: 200 nM isoproterenol
+ nisoldipine. Representative of 9 cells from 2 biologically independent mice. (g)
Same as (f) except from β2B-APEX2 transgenic mice. Black trace: control; blue trace:
200 nM isoproterenol. Representative of 7 experiments from 2 biologically independent
mice. (h-j) Anti-phospho-phospholamban antibody immunoblot of isolated cardiomyocytes
exposed to 1 μM isoproterenol. For cardiomyocytes isolated from α1C-APEX2 and β2B-APEX2
mice, the cardiomyocytes were exposed to isoproterenol after incubation with biotin-phenol.
Blots representative of 3 independent experiments from at least 5 biologically independent
mice for each genotype. (k) Same as (h-j) except whole heart exposed to 1 μM isoproterenol
for 5 minutes after infusion of biotin-phenol. Blots representative of at least 5
biologically independent mice for no isoproterenol and at least 5 biologically independent
mice for isoproterenol. For source gel data, see Supplementary Fig. 1.
Extended Data Fig. 3.
Analysis of proteins quantified by mass spectrometry in cardiomyocytes isolated from
α1C-APEX2 and β2B-APEX2 mice.
(a) Proteins with a ratio of >2 (measured by normalized TMT signal/noise) in the experimental
conditions compared to a no-labeling control (no H2O2) are sorted by spectral counts.
The 150 proteins with the highest peptide counts are displayed in the color-coded
table. α1C-APEX2 and β2B-APEX2 data were collected in biological duplicate experiments.
The full table with 3883 proteins quantified by multiplexed SPS MS3 TMT mass spectrometry
is reported in Supplementary Data Table 1. (b) Prefuse force directed map of proteins
from (a). Peptide counts were used as weight. Proteins mapping to the GO-terms “Z
disc” are colored in green, to “membrane” in yellow and to both in purple. The α1C-APEX2
and β2B-APEX2 are colored in blue. (c) Gene-ontology (GO) term (cellular localization)
enrichment for proteins in (a). The full table is reported in Supplementary Data Table
2.
Extended Data Fig. 4.
Two-way hierarchical clustering of scaled data from Fig. 2.
(a) Dendrogram showing two-way hierarchical clustering of scaled TMT signal to noise
(s/n) data for streptavidin-purified proteins from α1C-APEX2 expressing cardiomyocytes
after vehicle or after isoproterenol stimulation. Scaled relative TMT protein quantification
data for 1951 proteins from biological quintuplicate α1C-APEX2. Clustering used Ward’s
minimum variance method. (b) Dendrogram showing two-way hierarchical clustering of
scaled relative quantification data for 1936 proteins from biological triplicate β2B-APEX2
experiments. Heterogeneity between cardiomyocyte preparations from different mice
is apparent. (c) Dendrogram showing two-way hierarchical clustering of scaled relative
quantification data for 2610 proteins from whole-organ α1C-APEX without or with perfusion
of isoproterenol. Prominent heterogeneity in relative protein quantification between
hearts is apparent. Position of Rad is indicated by a red line. In this experiment,
the individual hearts are not paired. (d) Dendrogram showing two-way hierarchical
clustering of scaled TMT signal to noise (s/n) data from non-transgenic (NTG) cardiomyocytes
under isoproterenol stimulation or with vehicle. Scaled data of 4622 quantified proteins
from biological quadruplicate experiment are displayed. Pairing of samples is apparent.
Extended Data Fig. 5.
Isoproterenol-induced change in Rad detected by mass spectrometry.
(a) MS2 spectrum and TMT quantification parameters of a Rad peptide changed upon isoproterenol
treatment of murine hearts. The MS2 spectrum used for identification of the peptide:
IFGGIEDGPEAEAAGHTYDR mapping to Rad is displayed. y and b ion m/z identified in the
spectrum and their deviation from theoretical m/z are displayed on the left. Precursor
mass was measured as 778.71 Da with a charge of +3. Peptide modifications were +229.16
Da for TMT on the peptide N-terminus and lysine residues, +57.02 Da for cysteine alkylation
and +15.99 for methionine oxidation. Ion injection times, isolation specificity, sum
of signal to noise (SN) over all TMT channels, TMT raw intensities, adjusted intensities
and final SN intensities used for relative quantification as well as SPS ion m/z are
displayed. (b) Table showing gene names of proteins with P < 0.05 for the three approaches:
cardiomyocytes isolated from α1C-APEX and β2B-APEX mice, and α1C-APEX hearts. Yellow-highlighted
genes are in common amongst groups, but note for Mast2, the fold-change is not consistent.
D3Z0N2* is a single peptide with the sequence ESFDSQSLINNQSK and its abundance is
reduced by isoproterenol-treatment in non-transgenic cardiomyocytes to 27% likely
by post-translational modification (P =0.000002, see Fig. 3 and Supplementary Data
Table 6). Data are mean fold-change for 5 pairs of biologically independent pairs
of α1C-APEX2 cardiomyocyte samples, 3 pairs of biologically independent pairs of β2B-APEX
cardiomyocyte samples, and 10 α1C-APEX2 hearts, 5 without isoproterenol and 5 with
isoproterenol. Non-adjusted unpaired two-tailed t-test. (c) Venn diagram of the data
from (b). Proteins: Rad = Rrad; PKA catalytic subunit= Prkaca; Acss1 = acyl-CoA synthetase
short chain family member 1. Rad is the only protein consistently changed amongst
the three approaches.
Extended Data Fig. 6.
Rad is required for Forskolin-induced activation of heterologously expressed CaV1.2
channels.
(a) Exemplar whole-cell CaV1.2 currents elicited from step depolarizations recorded
from HEK293T cells expressing Rad. Pulses every 10-s before (black traces) and during
Forskolin (blue traces). Representative of at least 10 cells. (b) Methodology used
for generating G/V curves. i. Upper: 200-ms voltage ramp from −60 mV to + 60 mV was
applied every 10-s. Lower: Current traces – average of 3 traces before Forskolin (black)
and 3 traces after Forskolin (blue). ii. Conversion of time scale to applied voltage.
iii. Convert to G-V relationship. (G). Fold-change was calculated at Gmax. (c) Graph
of Forskolin-induced fold-change in current stratified by basal current density. (d)
Exemplar traces of Ba2+ currents in absence and presence of Rad elicited by voltage
ramp every 10-s. Black traces before and blue traces after Forskolin. no Rad:7 cells;
Rad:16 cells. (e) Boltzmann function parameter V50. Mean ± SEM. ** P <0.01 by paired
two-tailed t-test. n= 7, 16, left to right. (f-g) Graphs of ratio of current after
Forskolin to current before Forskolin for cells transfected without and with Rad.
Representative of analyses for 3 cells for each condition. (h-l) The distribution
of sweep-by-sweep average P
O (single trial P
O). (h) In absence of Rad, sweeps with no openings or blank sweeps are rare (10%),
while most sweeps exhibit either intermediate-high levels of openings. (k) Pale blue
lines are conditional Po-voltage relationship obtained for sweeps exhibiting high
activity. Blue line is Boltzmann fit. (i) Fraction of blank sweeps is increased with
expression of Rad. (j) If PKA catalytic domain is also co-expressed with Rad, the
fraction of blank sweeps is reduced and there is a resurgence of the high activity
mode. (l) Same as (k) but with Rad and PKA expression.
Extended Data Fig. 7.
PKA phosphorylation sites in mouse Rad.
(a) Serine/threonine residues (purple-highlighted) mutated to alanine in the 14-SA
mutant. (b) Mass spectrometry identification of phosphorylated residues on Rad enriched
with an anti-GFP nanobody matrix from HEK cells expressing GFP-Rad and treated with
forskolin. The number of spectral counts is plotted against the position of the phosphorylated
amino acids in Rad. 534 aggregated phosphopeptides were detected from two independent
experiments. (c) The database entry of phosphorylation sites identified previously
in Rad is displayed. The highest level of Rad phosphorylation was detected in the
heart. Peptides with phosphorylated Ser residues (bold, red) on positions 25, 38 and
300 (mapped to Rad expression constructs used in this study) were detected. (https://phosphomouse.hms.harvard.edu/site_view.php?ref=IPI00133102).
(d) Serine residues mutated to alanine in 4-SA mutant indicated by arrows.
Extended Data Fig. 8.
Binding of Rad and β2B is required for regulation of Forskolin-induced stimulation
of voltage-gated Ca2+ channels.
(a) Alanine substitutions of Rad at residues R208 and L235 (yellow-highlighted), and
(b) alanine substitutions at residues D244, D320 and D322 (yellow-highlighted) of
β2B were created to attenuate Rad binding to β as described previously
28,29
. (c) Ba2+ current of CaV1.2 channels elicited by voltage ramp every 10 s from −60
mV to +60 mV over 200 ms. Black traces before and blue traces after Forskolin. Representative
of 20 and 15 cells, from top to bottom. (d) Boltzmann function parameter V50. Mean
± SEM. ** P <0.001 by paired two-tailed t-test. Data for WT Rad is same as Fig. 3h.
Specific P values can be found in the Source Data associated with this figure. N=
16, 19, 13, from left to right. (e) Fold-change (Forskolin vs. No Forskolin) in Gmax.
Mean ± SEM. P < 0.0001 by one-way ANOVA; **** P <0.0001 by Dunnett’s test. Data for
WT Rad and WT β2B is same as Fig. 4e. n= 7, 7, 9 cells, from left to right. (f) Fold-change
(forskolin vs. no forskolin) in Gmax. Mean ± SEM. P < 0.001 by one-way ANOVA; ***
P < 0.001 by Dunnett’s test. Data for WT Rad and WT β2B is same as Fig 4h. n= 11,
7, 8 cells, from left to right.
Extended Data Fig. 9.
Phosphorylation-dependent dissociation of Rad and β3 and β4 subunits.
(a-b) FRET 2-hybrid binding isotherms were determined for Cerulean (Cer)-tagged β3
and β4, and N-terminal Venus (Ven)-tagged WT or 4-SA mutant Rad. FRET efficiency (ED)
is plotted against the free concentration Ven-WT or Ven-4SA-mutant Rad. Solid line
fits a 1:1 binding isotherm. Co-expression of the PKA catalytic subunit weakened binding
in WT Rad-expressing cells, but not 4-SA mutant Rad-expressing cells. (c) Bar graph
summarizing mean Kd,EFF for β2B, β3 and β4, and WT and 4-SA mutant Rad, expressed
without and with catalytic PKA subunit. Mean ± 95% confidence intervals (CI). The
error bars on the Kd,EFF are a 95%-CI for the pooled non-linear fits based on the
Jacobians computed. The sample size for each condition is 1,580-10,364 cells acquired
via two independent transfections and then pooled. The distribution of data in this
graph is reflected in Fig. 4b-c, and in (a) and (b) of this figure.
Extended Data Fig. 10.
ClustalW alignment of Rad sequences and RGK GTPases.
(a) Conserved phosphorylation sites for mouse Ser25, Ser38, Ser272 and Ser300 are
shown. Blue highlights basic amino acids (Arg, Lys and His), and red highlights Ser
and Thr. (b) Phosphorylation sites for mouse Rad Ser25, Ser38, Ser272 and Ser300 are
indicated (arrows). The C-terminal phosphorylation sites are conserved. The equivalent
of Ser25 phosphorylation site is conserved in Rem1. The equivalent of Ser38 phosphorylation
site is probably conserved in Gem. Blue highlights basic amino acids (Arg, Lys and
His), and red highlights Ser and Thr.
Supplementary Material
sup_guide
sup_tab1
sup_tab2
sup_tab3
sup_tab4
sup_tab5
sup_tab6
sup_fig1