Phosphorylation of Ca_v1.2 on S1928 uncouples the L‐type Ca²⁺ channel from the β₂ adrenergic receptor

Abstract

Agonist‐triggered downregulation of β‐adrenergic receptors (ARs) constitutes vital negative feedback to prevent cellular overexcitation. Here, we report a novel downregulation of β₂ AR signaling highly specific for Ca_v1.2. We find that β₂‐AR binding to Ca_v1.2 residues 1923–1942 is required for β‐adrenergic regulation of Ca_v1.2. Despite the prominence of PKA‐mediated phosphorylation of Ca_v1.2 S1928 within the newly identified β₂ AR binding site, its physiological function has so far escaped identification. We show that phosphorylation of S1928 displaces the β₂ AR from Ca_v1.2 upon β‐adrenergic stimulation rendering Ca_v1.2 refractory for several minutes from further β‐adrenergic stimulation. This effect is lost in S1928A knock‐in mice. Although AMPARs are clustered at postsynaptic sites like Ca_v1.2, β₂ AR association with and regulation of AMPARs do not show such dissociation. Accordingly, displacement of the β₂ AR from Ca_v1.2 is a uniquely specific desensitization mechanism of Ca_v1.2 regulation by highly localized β₂ AR/cAMP/PKA/S1928 signaling. The physiological implications of this mechanism are underscored by our finding that LTP induced by prolonged theta tetanus (PTT‐LTP) depends on Ca_v1.2 and its regulation by channel‐associated β₂ AR.

Introduction

Norepinephrine in the brain is important for arousal, behavioral acuity, and learning in novel and emotionally charged situations (Cahill et al, 1994; Berman & Dudai, 2001; Hu et al, 2007; Minzenberg et al, 2008; Carter et al, 2010). It signals via β₁ and β₂AR–G_s–adenylyl cyclase–cAMP–PKA cascades (Sanderson & Dell'Acqua, 2011). The β₂AR uniquely binds directly to the C‐terminus of α₁1.2, the central pore‐forming subunit of Ca_v1.2 (Davare et al, 2001; Balijepalli et al, 2006), and via PSD‐95 and auxiliary TARP subunits to AMPA‐type glutamate receptors (AMPARs) (Joiner et al, 2010; see also Wang et al, 2010). These complexes also contain G_s (Davare et al, 2001; Joiner et al, 2010), adenylyl cyclase (Davare et al, 2001; Efendiev et al, 2010; Joiner et al, 2010; Nichols et al, 2010), and AKAP‐anchored PKA (Davare et al, 2001; Tavalin et al, 2002; Hulme et al, 2003, 2006a; Hall et al, 2007; Oliveria et al, 2007; Joiner et al, 2010; Zhang et al, 2013; Dittmer et al, 2014). Assembly of such complexes brings all components of this cAMP cascade into close proximity with each other (Fig EV1 A and B), which results in localized cAMP signaling and regulation of β₂AR‐associated Ca_v1.2 and AMPAR (Chen‐Izu et al, 2000; Davare et al, 2001; Hulme et al, 2003; Joiner et al, 2010). Spatial restriction of cAMP production, diffusion, and signaling is a key mechanism thought to underlie the specific cAMP effects seen for certain G_s protein‐coupled receptors (G_sPCRs) (Smith et al, 2006; Leroy et al, 2008; Dai et al, 2009; Richter et al, 2013) including β₂AR (Jurevicius & Fischmeister, 1996; Kuschel et al, 1999; Chen‐Izu et al, 2000; Davare et al, 2001; Balijepalli et al, 2006; Nikolaev et al, 2010). This localized signaling is in contrast to the broad non‐target selective signaling by the β₁AR and other G_sPCRs (Xiao et al, 1999b; Steinberg & Brunton, 2001; Balijepalli et al, 2006). Despite much effort to prove this concept, clear evidence in support of this hypothesis as provided here by the effects of acute β₂AR displacement from Ca_v1.2 by peptide and S1928 phosphorylation (see below) has been lacking so far.

Figure EV1. Model of the β₂AR–Cav1.2 and β₂AR–GluA1 complexes and time course of ISO‐induced GluA1 phosphorylation and displacement of β₂AR from Ca_v1.2.

• A
  The β₂AR binds directly to residues 1923–1942 in the C‐terminus of α₁1.2 (red double‐headed arrow). S1928 phosphorylation by PKA displaces the β₂AR. The second PKA site S1700 is located in a segment predicted to interact with the distal C‐terminus (Hulme et al, 2006b). The A kinase anchor protein AKAP5 interacts with the N‐terminus, loop I, and the distal C‐terminus of α₁1.2 (Hall et al, 2007). It links PKA and likely also AC to Ca_v1.2 as it does for GluA1 (Zhang et al, 2013). Peptide 2 (Pep2) mimics residues 1923–1942 and competes with the β₂AR for binding to α₁1.2.
• B
  The β₂AR associates with AMPARs via binding to the third PDZ domain of PSD‐95. PSD‐95 itself binds with its first and second PDZ domain to stargazin (Stg) and its homologues (TARPs, γ2/3/4/5/7/8), which are auxiliary AMPAR subunits. AKAP5 binds to the PSD‐95 homologue SAP97 (Tavalin et al, 2002), which in turn binds to the very C‐terminus of GluA1 (Leonard et al, 1998). It recruits PKA and AC to GluA1 (Zhang et al, 2013). S845 is the PKA site on GluA1. The DSPL peptide mimics residues 894–907 at the very C‐terminus of the β₂AR and competes with the β₂AR for binding to the third PDZ domain of PSD‐95.
• C–G
  Forebrain slices from WT mice (C–F) or S1928KI mice (G) were treated with 10 μM ISO for 0.5–10 min before solubilization, ultracentrifugation, separate IP of GluA1 (C–E) or α₁1.2 (F, G), and IB for pS845, followed by stripping and re‐probing for pS831 and ultimately total GluA1 (C–E) or IB for α₁1.2 and β₂AR (F, G). Shown in (C) are IBs for pS845, pS831, and total GluA1 after GluA1 IP. For quantification of GluA1 phosphorylation, pS845 (D) and pS831 (E) signals from (C) were normalized to total GluA1 signals. Quantification of co‐IP of β₂AR with α₁1.2 following ISO treatment is shown in (F, G). Sample IBs for WT and S1928A KI mice are depicted in bottom panels of Fig 3A and G, respectively. For quantification of coIP, β₂AR signals were normalized to α₁1.2. Data are presented as mean ± SEM. n = 3. **P < 0.01, ***P < 0.001, one‐way ANOVA.
• H
  Mouse brain was homogenized and solubilized with 1% Triton X‐100 before ultracentrifugation and IP with control IgG or β₂AR antibody. IB for pS1928 was followed by stripping and reprobing for total α₁1.2 (top portion of blot; top two panels), for GluA1 (middle portion; third panel), or for β₂AR (bottom portion; bottom panel). As illustrated here, we never observed any pS1928 signal in β₂AR despite clear detection of α₁1.2 suggesting that S1928 phosphorylation completely prevents the β₂AR–α₁1.2 interaction.

Ca_v1.2 is the most abundant L‐type Ca²⁺ channel in brain and heart (Hell et al, 1993a). Mutations in Ca_v1.2 affect many tissues indicating widespread prominent Ca_v1.2 functions, which include control of cardiac contractility and heart rate as well as autistic‐like behaviors (Splawski et al, 2004). Besides their prominent roles in cardiovascular function, L‐type channels are critical in the brain for long‐term potentiation (Grover & Teyler, 1990; Moosmang et al, 2005) and depression (LTD) (Bolshakov & Siegelbaum, 1994), neuronal excitability (Marrion & Tavalin, 1998; Berkefeld et al, 2006), and gene expression (Dolmetsch et al, 2001; Marshall et al, 2011; Li et al, 2012; Ma et al, 2014). Upregulation of Ca_v1.2 activity by β‐adrenergic signaling is a central mechanism of regulating Ca²⁺ influx into cardiomyocytes (Reuter, 1983; Balijepalli et al, 2006) and neurons (Gray & Johnston, 1987; Davare et al, 2001; Oliveria et al, 2007; Dittmer et al, 2014). The differential global versus local regulation of Ca_v1.2 by β₁AR versus β₂AR might be due to association of the β₂AR but not β₁AR with Ca_v1.2 (Chen‐Izu et al, 2000; Davare et al, 2001; Balijepalli et al, 2006). We now provide clear evidence for this notion by showing that acute displacement of the β₂AR by a peptide and by S1928 phosphorylation prevents phosphorylation and upregulation of Ca_v1.2 by β₂AR stimulation.

The most prominent and heavily regulated PKA phosphorylation site in Ca_v1.2 is S1928 in the C‐terminus of its central α₁1.2 subunit (Hell et al, 1993b, 1995; De Jongh et al, 1996; Davare et al, 1999, 2000; Davare & Hell, 2003; Hulme et al, 2006a; Hall et al, 2007; Dai et al, 2009). However, functional studies argue against S1928 regulating channel activity in the heart (Ganesan et al, 2006; Lemke et al, 2008). Here, we found that the β₂AR binds to α₁1.2 residues 1923–1942 and that S1928 phosphorylation within this segment disrupts this interaction. This mechanism constitutes a particular form of downregulation of β₂AR signaling upon prolonged stimulation that specifically blunts subsequent upregulation of Ca_v1.2 but not AMPAR phosphorylation and activity and is absent in S1928A knock‐in mice.

Results

The β₂AR binds to residues 1923–1942 in the C‐terminus of Ca_v1.2

As the β₂AR C‐terminus mediates binding to Ca_v1.2 (Davare et al, 2001), we utilized amylose‐immobilized maltose‐binding protein (MBP)‐tagged β₂AR C‐terminus in pull‐down experiments to define its binding site in α₁1.2. We first tested affinity purified glutathione S‐transferase (GST) fusion proteins of the N‐terminus of α₁1.2, the three loops between the four homologous membrane domains of α₁1.2, and the three C‐terminal constructs CT1 (aa 1507–1733), CT23 (aa 1622–1905), and CT4 (aa 1909–2171), which cover the whole α₁1.2 C‐terminus (Fig 1A). From these constructs, only CT4 bound to the MBP‐tagged β₂AR C‐terminus, indicating a highly specific interaction (Fig 1B and C). From three fragments that covered CT4 (CTC (aa 1834–1957); CTD (aa 1944–2067); CTE (aa 2054–2171)), only CTC bound to the β₂AR (Fig 1D and E). These results restrict the interaction site to the overlapping region between CTC and CT4 (aa 1909–1957).

Figure 1. Identification of the β₂ AR binding region on Ca_v1.2.

• A
  Schematic of the α₁1.2 subunit of Ca_v1.2 (top) and the α₁1.2‐derived GST fusion proteins covering the C‐terminus (bottom).
• B
  Pull‐down of GST‐tagged α₁1.2 segments (top immunoblot; IB) by immobilized MBP‐β₂AR C‐terminus (residues 326–413 of human β₂AR). GST fusion proteins were detected by an anti‐GST antibody and MBP fusion proteins by an anti‐MBP antibody. Middle IB shows that comparable amounts of the various GST fusion proteins had been added to the resin samples and bottom IB illustrates equal loading of all amylose resin samples with MBP‐β₂AR.
• C
  Quantification of (B).
• D
  Pull‐down of GST‐CTC but not GST‐CTD or GST‐CTE (middle IB) by immobilized MBP‐β₂AR C‐terminus (left) but not MBP alone (right), all of which were present at comparable amounts (bottom and top IBs, respectively). GST fusion proteins were detected by an anti‐GST antibody and MBP fusion proteins by an anti‐MBP antibody.
• E
  Quantification of (D).
• F
  IP of β₂AR in the presence of 10 μM GST (Control; Ctr) or GST‐tagged C‐terminal fragments as indicated. CT4 and CTC specifically displaced α₁1.2 (top of IB) but not GluA1 (middle of same IB) from β₂AR (bottom of same IB). Use of non‐specific IgG (left lane in right panels) indicates specificity of IP.
• G, H
  For quantification of coIPs, α₁1.2 (G) and GluA1 (H) immunosignals from (F) were normalized to β₂AR signals.
• I
  Representative IB showing amounts of the GST fusion proteins that were added in (F), as detected by anti‐GST antibodies.

Data information: Data are presented as mean ± SEM. n = 3 (C, E) or 5 (G, H). ***P < 0.001, one‐way ANOVA.

To test whether CT4 and CTC bind to native Ca_v1.2 and could be used to acutely and specifically disrupt the β₂AR–Ca_v1.2 complex, the β₂AR was immunoprecipitated in the absence and presence of CT4 and CTC and, as negative controls, CT1, CT23, and CTD. CT4 and CTC but not the other polypeptides completely displaced Ca_v1.2 from the β₂AR (Fig 1F, G and I). To eliminate the possibility of nonspecific or secondary effects of the polypeptides on the complex, we monitored within the same samples and same immunoblot lanes co‐immunoprecipitation (co‐IP) of the AMPAR subunit GluA1, which forms a separate complex with the β₂AR. In contrast to Ca_v1.2, this interaction is mediated by PDZ interactions with PSD‐95 (Fig EV1 A and B), and therefore, neither CT4 nor CTC affected the GluA1–β₂AR co‐IP (Fig 1F and H).

To further narrow down the β₂AR binding site of α₁1.2, synthetic peptides covering aa 1906–1925 (Pep 1), aa 1923–1942 (Pep 2), and aa 1939–1959 (Pep 3) whose N‐termini were labeled with fluorescein (FITC), were titrated with the β₂AR C‐terminus, and their binding was monitored by fluorescence polarization. Pep2, but neither Pep1 nor Pep3, showed strong and saturable binding with an apparent K _d of ~1.9 μM (Fig 2A). In addition, only Pep2 displaced α₁1.2 (but not GluA1) from the β₂AR during IP (Fig 2B–D).

Figure 2. The β₂ AR binds to the S1928 phosphorylation site.

• A
  Titration of fluorescence polarization of FITC peptides (100 nM) spanning α₁1.2 aa 1906–1959 with purified MBP‐β₂AR C‐terminus. K _d value was obtained by fitting a nonlinear direct binding curve to Pep2.
• B
  IP of β₂AR in the absence (Control) or presence of 10 μM peptides, as indicated. Pep2 specifically displaced α₁1.2 (top of IB) but not GluA1 (middle, same IB) from β₂AR (bottom, same IB). Use of non‐specific IgG (left lane) indicates specificity of (co)IPs.
• C, D
  For quantification, α₁1.2 (C) and GluA1 (D) IB signals from (B) were normalized to β₂AR signals.
• E
  IP of β₂AR in the absence (Control) or presence of 10 μM Pep2 or Pep2 with S1928 being phosphorylated (phPep2), which did not displace α₁1.2.
• F
  α₁1.2 IB signals were normalized to β₂AR signals.

Data information: Data are presented as mean ± SEM. n = 4. ***P < 0.001, one‐way ANOVA.

Unlabeled synthetic Pep2 and synthetic Pep2 with S1928 being phosphorylated (PhPep2) were added during IP of β₂AR to test whether phosphorylation of S1928 affects β₂AR binding. While Pep2 removed Ca_v1.2 (but, once more, not GluA1), PhPep2 had no effect suggesting that S1928 phosphorylation impairs β₂AR binding (Fig 2E and F).

S1928 phosphorylation displaces the β₂AR from Ca_v1.2

To evaluate whether S1928 phosphorylation displaces the β₂AR from Ca_v1.2, we monitored Ca_v1.2 phosphorylation and β₂AR–Ca_v1.2 association in forebrain slices upon stimulation with the βAR agonist isoproterenol (ISO). S1700 has recently emerged as a PKA phosphorylation site that is important for upregulation of Ca_v1.2 activity in heart (Fuller et al, 2010; Hell, 2010; Fu et al, 2013). As phosphorylation of S1700 and S1928 increased (Fig 3A–C), association of Ca_v1.2 with the β₂AR decreased (Fig 3D and E). Strikingly, no such decrease was observed in slices from S1928A KI mice, even though ISO induced S1700 phosphorylation in these mice (Fig 3G–K). Displacement of the β₂AR from Ca_v1.2 is unique for S1928 phosphorylation, as the β₂AR–GluA1 interaction was not disrupted by ISO application (Fig 3D, F, J and L), which induced phosphorylation of S845 (Fig EV1C and D), a well‐established PKA site on GluA1 (Roche et al, 1996). IP of Ca_v1.2 followed by IB of β₂AR confirmed their dissociation upon ISO treatment in WT but not S1928A KI mice (Fig 3A, bottom; Fig EV1F and G).

Figure 3. S1928 phosphorylation displaces β₂ AR from Ca_v1.2.

Forebrain slices from WT (A–F) and S1928A KI mice (G–L) were treated with vehicle (water) or 10 μM isoproterenol (ISO) for 0.5–10 min before solubilization, ultracentrifugation, IP of α₁1.2 (A–C, G–I) or β₂AR (D–F, J–L), and sequential IB for pS1928, pS1700, and α₁1.2, for GluA1, or for β₂AR, of corresponding regions of the blots, as indicated. All the α₁1.2 IPs in (A–C) and (G–I) were from the same samples (which were split in half for parallel IP) as the β₂AR IPs in (D–F) and (J–L), respectively (for quantification of coIP of β₂AR with α₁1.2 see Fig EV1F and G).

• A–F
  In WT, the time‐dependent increase in S1928 and S1700 phosphorylation (A–C) paralleled the decrease in coIP of β₂AR with α₁1.2 (A bottom, Fig EV1F) and of α₁1.2 with β₂AR (D–F). For quantification of α₁1.2 phosphorylation (B, C), pS1928 and pS1700 signals were normalized to α₁1.2. For quantification of coIP (E, F), α₁1.2 and GluA1 signals were normalized to β₂AR.
• G–L
  In S1928A KI mice, ISO induced S1700 phosphorylation (G, I) but did not disrupt the α₁1.2–β₂AR interaction (G bottom, J–L, Fig EV1G). For quantification of α₁1.2 phosphorylation (H, I), pS1928 and pS1700 signals were normalized to α₁1.2. For quantification of coIP (K, L), α₁1.2 and GluA1 signals were normalized to β₂AR.

Data information: Data are presented as mean ± SEM. n = 4 (B, C, H, I, K and L) or 6 (E, F). **P < 0.01, ***P < 0.001, one‐way ANOVA.

To test whether the ISO‐induced dissociation of the β₂AR from Ca_v1.2 results in their spatial separation, we co‐expressed α₁1.2 with the HA tag within an extracellular loop (α₁1.2‐HA) and β₂AR with the FLAG tag at its extracellular N‐terminus (FLAG‐β₂AR) in cultured hippocampal neurons. Line scan analysis of the fluorescence distribution of the surface labeled α₁1.2‐HA and FLAG‐β₂AR (see Appendix Supplementary Methods) showed that the median distance between neighboring α₁1.2‐HA and FLAG β₂AR clusters significantly increases from 0.24 μm (25–75% interquartile range: 0.15–0.44 μm) to 0.34 μm (IQR: 0.21–0.54 μm) after 5 min of ISO treatment (Fig 4). Because the distribution of distances failed two normality tests (see Appendix Supplementary Methods), we used the nonparametric Mann–Whitney Rank test for statistical analysis, which resulted in a two‐tailed P‐value of < 0.0001. The Kolmogorov–Smirnov cumulative distributions test yielded a P‐value < 0.0001. In addition, Pearson's correlation analysis yielded a coefficient 0.36 ± 0.03 (mean ± SEM) for control and 0.29 ± 0.02 for ISO treated neurons with P = 0.037. Furthermore, we calculated the fraction of overlap between Ca_V1.2‐HA and FLAG‐β₂AR puncta. We obtained a Mander's coefficient of 0.41 ± 0.03 (mean ± SEM) for control and 0.31 ± 0.03 for ISO treated neurons with P = 0.02. These results are consistent with the idea that ISO stimulation displaces the β₂AR from Ca_v1.2.

Figure 4. S1928 phosphorylation separates β₂ AR from Ca_v1.2.

Hippocampal cultures were transfected with FLAG‐β₂AR and α₁1.2‐HA at 6 days in vitro (DIV), treated with vehicle (water) or 1 μM isoproterenol (ISO) for 5 min at 18 DIV, fixed and surface labeled for HA and FLAG.

• A, B
  Representative immunofluorescent images obtained by wide‐field microscopy at lower and higher resolutions (scale bar, 5 μm). Arrows in (A) indicate the samples enlarged in (B).
• C
  Quantification of distance between centers of HA and FLAG puncta (**P < 0.001, Mann–Whitney rank sum test). The bars represent 5^th (lower end) and 95^th percentile (higher end).

ISO‐triggered dissociation of the β₂AR from Ca_v1.2 prevents subsequent Ca_v1.2 phosphorylation

Could the displacement of the β₂AR from Ca_v1.2 upon S1928 phosphorylation be a novel mechanism that specifically downregulates this powerful β adrenergic regulation of Ca²⁺ influx into neurons? To test this idea at the molecular level, forebrain slices from WT mice were treated with vehicle or ISO for 5 min, followed by washout of ISO for various time periods before re‐application of ISO. 20‐min and even 10‐min but not 3‐min washout reversed the ISO‐induced displacement of the β₂AR from α₁1.2 (Fig 5A, lane 5 vs. 7; Fig EV2A–C). S1700 and S1928 phosphorylation returned to baseline already after 3‐min washout (Fig 5D, lanes 2–4; Fig 5E and F). As expected, phosphorylation of GluA1 on S845 behaved similarly (Fig 5D and G); phosphorylation of S831, which can be mediated by PKC and CaMKII but not PKA (Roche et al, 1996; Halt et al, 2012), served as a negative control that is inert to PKA stimulation (Fig 5D and H).

Figure 5. ISO‐induced displacement of the β₂ AR from Ca_v1.2 blunts subsequent Ca_v1.2 phosphorylation.

Forebrain slices from WT mice were treated with vehicle (water; lanes 1, 5; numbers on bottom) or 10 μM ISO for 5 min, followed, if indicated, by 1‐min, 3‐min, or 20‐min washout of ISO (lanes 3–7) and a second application of ISO for 5 min (lane 6), before solubilization and ultracentrifugation.

• A
  β₂AR was IPed before IB for α₁1.2 (top part of IB), GluA1 (middle part of IB), and β₂AR (bottom part of IB), as indicated. ISO‐induced displacement of the β₂AR from α₁1.2 (lanes 2–6) lasted at least 3 min but not 20 min (compare lanes 6 and 7).
• B, C
  For quantification, α₁1.2 (B) and GluA1 (C) IB signals from (A) were normalized to β₂AR.
• D
  α₁1.2 and GluA1 were concurrently IPed from same samples as in (A) by simultaneous addition of anti‐α₁1.2 and anti‐GluA1 antibodies before probing and stripping/re‐probing upper part of IB for pS1928, pS1700, and total α₁1.2 (top three panels) and middle part for pS845, pS831, and total GluA1 (bottom three panels). ISO‐induced displacement of the β₂AR from α₁1.2 (see A) rendered α₁1.2 (but not GluA1) refractory to re‐phosphorylation of S1928 and S1700 upon a second ISO application of α₁1.2 (compare lanes 5 and 6).
• E–H
  For quantification, pS1700 (E) and pS1928 (F) IB signals from (D) were normalized to total α₁1.2, and pS845 (G) and pS831 (H) signals from (D) to total GluA1.

Data information: Data are presented as mean ± SEM. n = 3 (B, C) or 4 (E–H). **P < 0.01, one‐way ANOVA.

Figure EV2. Time course of β₂AR re‐association with α₁1.2 and of de‐phosphorylation of α₁1.2 and GluA1 after ISO treatment.

Forebrain slices from WT mice were treated with vehicle or 10 μM ISO for 5 min, followed by ISO washout for 3, 10, 30, and 90 min if indicated, before solubilization and ultracentrifugation.

• A
  IP of β₂AR was followed by IB for total α₁1.2 (top part of blot), GluA1 (middle part), or β₂AR (bottom part), as indicated. Re‐association of the β₂AR with α₁1.2 took more than 3 min but was fully accomplished 10 min after ISO washout.
• B, C
  For quantification of coIP, GluA1 (B) and α₁1.2 (C) signals from (A) were normalized to β₂AR.
• D
  α₁1.2 and GluA1 were simultaneously IPed before probing and re‐probing upper part of blot for pS1928, pS1700, and total α₁1.2 and middle part for pS845 and total GluA1.

Data information: Data are presented as mean ± SEM. n = 4. *P < 0.05, one‐way ANOVA.

We tested whether the acute displacement of the β₂AR from α₁1.2 affects re‐phosphorylation and re‐stimulation of Ca_v1.2 after 3‐min washout. In fact, re‐application of ISO after a 3‐min washout was not able to induce a second round of phosphorylation of S1700 or S1928 (Fig 5D, lane 6). In contrast, in mock washout samples ISO was fully effective in inducing phosphorylation of these residues following initial application of vehicle instead of ISO before the 3‐min washout (Fig 5D, lane 5). The ISO‐induced displacement of the β₂AR from α₁1.2 was specific for Ca_v1.2, as coIP of GluA1 with the β₂AR was not affected within same samples that were analyzed for Ca_v1.2 coIP (Fig 5A and C). Re‐phosphorylation of S845 during the second ISO treatment was also not blunted by the first ISO application (Fig 5D, lane 6; Fig 5G). Accordingly, displacement of the β₂AR from Ca_v1.2 selectively downregulates the signaling pathway from the β₂AR to Ca_v1.2 without affecting another target, GluA1, which also forms a signaling complex with the β₂AR. To test whether endocytosis of the β₂AR is responsible for lack of re‐phosphorylation of α₁1.2 by the second ISO application, we blocked endocytosis with two different drugs, dynasore and pitstop. Neither affected the loss of α₁1.2 phosphorylation by the second ISO pulse (Fig EV3) arguing against this possibility.

Figure EV3. Blunting of ISO‐induced Ca_v1.2 phosphorylation by ISO pre‐treatment is not affected by endocytosis blockers.

Forebrain slices from WT mice were pre‐treated for 30 min with 100 μM dynasore (lanes 4, 5; denoted at bottom of panel) or 30 μM pitstop (lanes 6, 7), treated with vehicle (water; lanes 4, 6) or 3 μM ISO for 5 min, followed by 3‐min or 20‐min washout of ISO (or of vehicle in lanes 4, 6) and a second application of ISO for 5 min (lanes 2–7), before solubilization and ultracentrifugation.

• A
  α₁1.2 and GluA1 were simultaneously IPed before probing and re‐probing upper part of IB for pS1928, pS1700, and total α₁1.2, and middle part for pS845 and total GluA1. Inhibiting endocytosis with either dynasore or pitstop did not affect blunting of S1700 and S1928 phosphorylation during a second ISO application (compare lanes 5 and 7 with lane 2). Note that re‐phosphorylation upon the second ISO application was blunted after 3‐min washout but restored after 20‐min washout subsequent to the first ISO treatment (compare lanes 1–3 with each other).
• B–D
  For quantification, pS1700 (B) and pS1928 (C) IB signals from (A) were normalized to total α₁1.2, and pS845 (D) IB signals to total GluA1.

Data information: Data are presented as mean ± SEM. n = 4. ***P < 0.001, one‐way ANOVA.

Strikingly, in S1928A KI mice, re‐phosphorylation of S1700 during the second ISO application after 3‐min washout was not decreased at all as compared to single ISO applications (Fig 6A, lane 6 vs. lanes 2 and 5). In fact, the second ISO treatment appears to have increased S1700 phosphorylation more strongly than the first treatment. These results suggest that additional, as yet to be identified, mechanisms exist that enhance phosphorylation of S1700 during repetitive activation of β₂AR bound to Ca_V1.2. For instance, like PKA, the phosphatase PP2B/calcineurin is linked to Ca_V1.2 via AKAP5 to counteract Ca_V1.2 phosphorylation by PKA (Oliveria et al, 2007; Fuller et al, 2014; Murphy et al, 2014) but released upon elevated Ca²⁺ influx via Ca_V1.2 (Li et al, 2012; Murphy et al, 2014). Because β adrenergic stimulation will increase Ca²⁺ influx via Ca_V1.2 as occurring under basal conditions due to neuronal network activity (Hall et al, 2013), it is conceivable that PP2B is displaced from Ca_V1.2 for 3 min or longer, allowing for stronger phosphorylation of S1700 in S1928A KI neurons upon ISO application that is repeated within 3 min.

Figure 6. ISO pre‐treatment does not blunt Ca_v1.2 phosphorylation by subsequent ISO treatment in S1928A KI mice.

Forebrain slices from S1928A KI mice were treated with vehicle (lanes 1, 5) or 10 μM ISO for 5 min, followed, if indicated, by 1‐min, 3‐min, or 20‐min washout of ISO (lanes 3–7) and a second application of ISO for 5 min (lane 6), before solubilization and ultracentrifugation.

• A
  α₁1.2 and GluA1 were concurrently IPed from same samples by simultaneous addition of anti‐α₁1.2 and anti‐GluA1 antibodies before probing and stripping/re‐probing upper part of IB for pS1928, pS1700, and total α₁1.2 and middle part for pS845, pS831, and total GluA1. In S1928A KI mice, S1700 re‐phosphorylation after a 3‐min washout of ISO was not blunted (lane 6) in contrast to WT mice but rather augmented (compare to lane 5).
• B–E
  For quantification, pS1700 (B) and pS1928 (virtually absent) (C) IB signals from (A) were normalized to total α₁1.2, and pS845 (D) and pS831 (E) signals from (A) to total GluA1.

Data information: Data are presented as mean ± SEM. n = 4. **P < 0.01, ***P < 0.001, one‐way ANOVA.

The ISO‐induced displacement of the β₂AR from α₁1.2 is completely reversible, as 10‐min washout of ISO resulted in full coIP of α₁1.2 with the β₂AR (Fig 5A, lane 7; Fig EV2A–C), which is preceded by dephosphorylation of S1928 and also S1700 (Fig EV2D). A 20‐min washout also restored the capability of the β₂AR to induce S1700 and S1928 phosphorylation (Fig EV3A and B, compare lanes 2 and 3).

ISO‐triggered dissociation of the β₂AR from Ca_v1.2 prevents subsequent stimulation of L‐type channel activity

To functionally test whether β₂AR stimulation affects subsequent regulation of Ca_v1.2 by a second, closely timed pulse of β₂AR stimulation, we sought to record single‐channel L‐type currents from cultured neurons in the cell‐attached patch clamp mode as in our previous work (Davare et al, 2001). Other Ca²⁺ channels were blocked by adding specific inhibitors (ωCTxGVIA and ωCTxMVIIC) to the patch pipette solution. We determined open probability (Po) from all channels within each patch (NPo) in recordings from neurons with either vehicle or ISO added to the patch pipette. Figure 7A shows original traces with single‐channel activity elicited by depolarizing pulses from −80 mV to several test potentials. These data were fit with a linear function that revealed a slope conductance for these channels of 27 ± 2 pS (Fig 7A), which corresponds to the expected slope conductance for an L‐type Ca_v1.2 channel under similar experimental conditions (Yue & Marban, 1990). As expected, ISO significantly increased NPo (Fig 7B–D) without affecting single‐channel amplitudes (see Appendix Fig S1) (Davare et al, 2001). Inclusion of the potent L‐type channel blocker nifedipine abrogated virtually all currents either with or without ISO present (Fig 7D). Hence, recordings reflect L‐type currents under either condition. The ISO effect was prevented by the highly specific PKA‐inhibitory PKI peptide, which carried 11 Arg residues to render it membrane permeant (Lu et al, 2007, 2011) (Fig 7D), thus confirming that the ISO‐induced upregulation of L‐type current is via PKA.

Figure 7. ISO‐induced displacement of the β₂ AR from Ca_v1.2 blunts subsequent stimulation of L‐type channel activity.

• A
  Representative single‐channel recordings from hippocampal neurons at 7–14 DIV with 500 nM BayK‐8644, 1 μM ω‐conotoxin GVIA, and 1 μM MVIIC in patch pipette upon depolarization from −80 to −30, −20, −10, and 0 mV. The right panel depicts the single‐channel current–voltage relationship. Mean amplitude of unitary currents for different membrane potential studied are −2.07 pA (−30 mV), −1.75 (−20 mV), −1.53 (−10 mV), and −1.25 (0 mV; n = 5 patches per test potential). Solid line represents best‐fit of data using a linear equation (R ² = 0.92) revealing a slope conductance for these channels of 27 ± 2 pS.
• B–D
  Representative single‐channel traces and summary plot upon depolarization from −80 to 0 mV under control conditions and in the presence of ISO in the patch pipette. Cultures were pre‐incubated for 15 min with 10 μM 11R‐PKI if indicated. The patch pipette contained either vehicle for control, nifedipine (nif; 1 μM), ISO (1 μM), or ISO plus nifedipine, which blocked all currents. The ISO‐induced increase in NPo was prevented by 11R‐PKI.
• E–H
  Representative single‐channel currents upon depolarization from −80 to 0 mV and summary plot after pre‐treatment of whole cultures with ISO. Cultures were pre‐incubated with vehicle (H₂O, mock wash, E) or 1 μM ISO for 5 min (F, G) and washed for 3 (E, F) or 10 min (G) before forming the cell‐attached patch with ISO present in the patch pipette. The upregulation of NPo to ˜0.4 (cf. C, D) occurred only if neurons were pretreated with vehicle instead of ISO (E, H; mock wash) or if ISO washout duration was 10 min (G, H) but not if washout was only 3 min (F, H).

Data information: Data are presented as mean ± SEM. NPo was determined for each recording and pooled under each condition for comparison (n = 10–14 patches; *P < 0.05, one‐way ANOVA with Tukey post hoc test). Arrows throughout the figure indicate the 0‐current level (i.e., closed channel).

Most critically, when ISO was first applied to the bath for 5 min before washout and subsequent formation of a patch, the ISO included in the patch only upregulated L‐type current when the washout was at least 10 min long (Fig 7E–H). If washout was only 3 min, channel activity remained low during the cell‐attached recording with ISO in the patch pipette (Fig 7F and H). As expected, pre‐treatment with vehicle followed by a 3‐min washout (mock wash; Fig 7E and H) did not affect upregulation of channel activity by ISO in the patch pipette. Accordingly, sequential stimulation of L‐type currents by two ISO applications was only effective if the interim time period was long enough to match the time frame required for the β₂AR to re‐associate with Ca_v1.2 (Fig 5A, lanes 6 and 7, and C; Fig EV2A and C) and re‐phosphorylate it (Fig EV3A, lane 3 vs. lane 2).

Binding of the β₂AR to residues 1923–1942 is required for β adrenergic stimulation of α₁1.2 phosphorylation and Ca_v1.2 activity

To exclude the possibility that covert effects other than displacement of the β₂AR from Ca_v1.2 might be responsible for loss of sensitivity of channel activity to a second pulse of ISO, the β₂AR was acutely displaced from Ca_v1.2 by Myr‐Pep2, a myristoylated version of Pep2, which mimics the binding site of aa 1923–1942 on the α₁1.2 subunit and displaces the β₂AR from Ca_v1.2 (Fig 2). Myristoylation renders peptides membrane permeant. We first determined at which concentration Myr‐Pep2 effectively disrupts the β₂AR–Ca_v1.2 interaction by adding increasing amounts to brain extracts during the IP of the β₂AR. 0.1–10 μM Myr‐Pep2 increasingly displaced Ca_v1.2 from the β₂AR, with 10 μM being apparently 100% effective without affecting the β₂AR–GluA1 association (Fig EV4A–C).

Figure EV4. Characterization of Myr‐Pep2 and Myr‐Pep2scr.

Forebrain slices from WT mice were pre‐incubated for 30 min with vehicle (water), 0.1–10 μM Myr‐Pep2, or 10 μM Myr‐Pep2scr (i.e., scrambled Myr‐Pep2).

• A–C
  After incubation with Myr‐Pep2, slices were solubilized before ultracentrifugation, IP of β₂AR, and IB for α₁1.2, GluA1, and β₂AR. Increasing amounts of Myr‐Pep2 progressively displaced α₁1.2 but not GluA1 from the β₂AR, with 10 μM resulting in near complete dissociation. For quantification, α₁1.2 (B) and GluA1 (C) immunosignals were normalized to β₂AR signals.
• D–F
  After incubation with Myr‐Pep2scr, slices were treated with ISO (10 μM, 5 min) before solubilization, ultracentrifugation, IP of β₂AR, and IB for α₁1.2, GluA1, and β₂AR. In contrast to Myr‐Pep2, Myr‐Pep2scr did not displace α₁1.2 (top; compare lanes 1 and 3) nor GluA1 (middle, same blot) from β₂AR (bottom, same blot). ISO treatment resulted in dissociation of the α₁1.2–β₂AR interaction, as seen before. For quantification, α₁1.2 (E) and GluA1 (F) immunosignals were normalized to β₂AR signals.
• G–J
  After incubation with Myr‐Pep2scr, slices were treated with ISO (10 μM, 5 min) before solubilization, ultracentrifugation, simultaneous IP of α₁1.2 and GluA1, and IB for pS1928, pS1700, and pS845. In contrast to Myr‐Pep2 (Fig 8D), Myr‐Pep2scr did not blunt ISO‐induced phosphorylation of α₁1.2 on S1700 nor on S1928 (top three panels reflecting sequential probings of top portion of blot) nor GluA1 phosphorylation on S845 (bottom two panels reflecting sequential probings of middle portion of same blot). For quantification, pS1928 (H) and pS1700 (I) immunosignals were normalized to α₁1.2, and pS845 (J) signals to GluA1 signals, respectively.

Data information: Data are presented as mean ± SEM. n = 3. **P < 0.01, ***P < 0.001, one‐way ANOVA.

Forebrain slices were incubated for 30 min with vehicle, 10 μM Myr‐Pep2, Myr‐Pep2scr, or Myr‐DSPL. MyrPep2scr is a scrambled version of MyrPep2 and served as negative control. Myr‐DSPL consists of the 14 aa at the very C‐terminus of the β₂AR, which interacts with the third PDZ domain of PSD‐95 (Joiner et al, 2010) (Fig EV1B). PSD‐95 in turn is linked to a subset of AMPARs via its binding to the auxiliary AMPAR subunits known as γ subunits or TARPs, including stargazin (Stg/γ₂). Myr‐DSPL specifically disrupts the interaction of the AMPAR subunit GluA1 with the β₂AR (Joiner et al, 2010) and served as a second negative control. In our experiments, Myr‐Pep2 displaced α₁1.2 but not GluA1 from the β₂AR (Fig 8A, compare lane 5 with 3; Fig 8B and C). In contrast, Myr‐DSPL removed GluA1 but not α₁1.2 from the β₂AR (Fig 8A, compare lane 1 with 3; Fig 8B and C). As before, ISO on its own caused a strong reduction in the coIP of α₁1.2, but not GluA1, with the β₂AR (Fig 8A, compare lane 4 with 3; Fig 8B and C). In combination with ISO, Myr‐Pep2 (compare lane 6 with 4) but not Myr‐DSPL (compare lane 2 with 4) caused a virtually complete displacement of the β₂AR from Ca_v1.2. As a second control, Myr‐Pep2scr had no effect on the association of the β₂AR with Ca_v1.2 whether slices were treated with ISO or not (Fig EV4D and E).

Figure 8. β₂ AR binding to α₁1.2 is required for β‐adrenergic stimulation of α₁1.2 phosphorylation and L‐type currents.

Forebrain slices (A–F) were pre‐incubated for 30 min with vehicle (water) or 10 μM Myr‐Pep2 or Myr‐DSPL. Like Myr‐DSPL (Joiner et al, 2010), Pep2 was myristoylated at its N‐terminus (Myr‐Pep2) to make it membrane permeant. Slices were then treated with ISO (10 μM, 5 min) or vehicle (water) before solubilization, ultracentrifugation, and IP of β₂AR (A–C) or simultaneously α₁1.2 and GluA1 with a combination of corresponding antibodies within same samples (D–F).

• A
  Myr‐Pep 2 displaced α₁1.2 (lane 5 vs. 3, top of blot) but not GluA1 (middle, same blot) from β₂AR (bottom, same blot); the inverse was true for Myr‐DSPL (lane 1 vs. 3).
• B, C
  For quantification, α₁1.2 (B) and GluA1 (C) immunosignals from (A) were normalized to β₂AR signals.
• D
  Myr‐Pep2 blunted ISO‐induced phosphorylation of α₁1.2 S1700 (lane 6 vs. 4, top of blot) but not GluA1 S845 (middle, same blot); the inverse was true for Myr‐DSPL (lane 2 vs. 4).
• E, F
  For quantification, pS1700 (E) and pS845 (F) immunosignals from (D) were normalized to α₁1.2 and GluA1 signals, respectively.
• G
  Representative cell‐attached recordings from hippocampal neurons as in Fig 7. In interleafed experiments, cultures were pre‐incubated for 30 min with 10 μM Myr‐Pep2 or scrambled Myr‐Pep2 (Myr‐Pep2scr). The patch pipette contained either vehicle (H₂O; control) or 1 μM ISO. The ISO‐induced increase in NPo was prevented by Myr‐Pep2 but not Myr‐Pep2scr. Arrows indicate the 0‐current level (i.e., closed channel).
• H
  Summary plot for (G). For statistical analysis, the NPo value was determined for each recording and pooled under each condition for comparison.

Data information: Data are presented as mean ± SEM. (B, C, E, F): n = 4. **P < 0.01, ***P < 0.001, one‐way ANOVA. (H): n = 10‐12 patches; *P < 0.005, one‐way Tukey post hoc test.

Importantly, the ISO‐induced increase in phosphorylation of α₁1.2 on S1700 (Fig 8D, lane 4 vs. 3, and E) was blocked by Myr‐Pep2 (lanes 6 vs. 5) but not Myr‐DSPL (lanes 1 vs. 2). The exact opposite was true for phosphorylation of GluA1 on S845 (Fig 8D and F). Accordingly, specific displacement of the β₂AR from Ca_v1.2 affects Ca_v1.2 but not GluA1 phosphorylation and vice versa. Furthermore, Myr‐Pep2scr did not affect phosphorylation of either α₁1.2 or GluA1 (Fig EV4G–J), confirming the specific actions of Myr‐Pep2.

To define the functional consequences of disrupting the β₂AR–Ca_v1.2 interaction, in interleaved experiments cultured neurons were pre‐incubated for 30 min with Myr‐Pep2 or Myr‐Pep2scr. Subsequent recording with ISO in the patch pipette indicated that Myr‐Pep2 but not Myr‐Pep2scr completely blocked the upregulation of channel function by ISO compared to vehicle controls (Fig 8G and H). We conclude that dissociation of the β₂AR–Ca_v1.2 complex by Myr‐Pep2 prevents upregulation of Ca_V1.2 channel phosphorylation and activity.

The β₂AR–Ca_v1.2 interaction is required for PTT‐LTP

Prolonged stimulation of the Schaffer collateral pathway at 5–10 Hz, which mimics the naturally occurring theta frequency (7 Hz), induces LTP (PTT‐LTP) if at the same time β₂AR (but not β₁AR) are stimulated (Thomas et al, 1996; Hu et al, 2007; Qian et al, 2012). This potentiation develops over a period of 15 min with the first 5–10 min showing an initial depression (Thomas et al, 1996; Hu et al, 2007; Qian et al, 2012). Because β₂AR stimulation prominently augments Ca²⁺ influx through Ca_v1.2 at postsynaptic sites (Hoogland & Saggau, 2004), we tested whether Ca_v1.2 in general and specifically its upregulation by the β₂AR is required for PTT‐LTP. In fact, PTT‐LTP was completely absent in conditional knockout mice in which Cav1.2 had been deleted in glutamatergic forebrain neurons when compared to WT littermate controls (Fig 9A–C). Analysis of input–output relation and paired pulse facilitation indicated that synaptic transmission is normal in both genotypes (Fig EV5). In contrast to the 5 Hz PTT‐LTP stimulus paradigm, LTP induced by a 100 Hz/1 s tetanus depends on NMDARs and not L‐type channels. In this case, potentiation is very strong immediately after the tetanus in part due to presynaptic mechanisms but typically relaxes to a significantly lower level over ~5 min. This 100 Hz LTP was normal in conditional Ca_V1.2 knockout mice (Fig 9D–F). Accordingly, respective NMDAR‐dependent synaptic plasticity mechanisms can be engaged in a normal manner in the Ca_V1.2 knockout mice when PTT‐LTP is absent. Strikingly, Myr‐Pep2 but not Myr‐Pep2scr blocked PTT‐LTP in wild‐type mice (Fig 9G–I).

Figure 9. β₂ AR binding to α₁1.2 is required for PTT‐LTP .

Graphs depict fEPSP initial slopes recorded from hippocampal CA1 before and after either a 5 Hz/3 min (A–C, G–I) or 100 Hz/1 s tetanus (D–F). Arrowheads mark onset of tetani and bars perfusion with 1 μM ISO and 10 μM of Myr‐Pep2 or Myr‐Pep2scr. Inserts show sample traces immediately before (left) and ~30 min after (right) tetani.

• A–C
  Litter‐matched WT but not conditional Ca_v1.2 KO mice showed PTT‐LTP.
• D–F
  WT as well as Ca_v1.2 KO mice showed NMDAR‐dependent 100 Hz LTP.
• G–I
  Myr‐Pep2 but not Myr‐Pep2scr blocked PTT‐LTP.

Data information: Data are presented as mean ± SEM. **P < 0.01, one‐way ANOVA.

Figure EV5. Basal synaptic transmission is normal in conditional forebrain Ca_v1.2 KO mice.

1. Increasing stimulus strength increased fEPSP initial slope to the same degree in Ca_v1.2 KO mice as in their WT littermates. Plotted are initial slopes of fEPSP against presynaptic fiber volley amplitudes. Inserts on top show sample traces for WT and KO mice for increasing stimulus strengths.
2. Quantification of slopes of graphs in (A).
3. Paired pulse facilitation (PPF) did not differ between WT and KO mice. Shown are summary graphs of the ratios of second to first pulse response (fEPSP initial slopes) for increasing interval lengths. The first pulse was set in each individual recording to equal 100%.

Data information: Data are presented as mean ± SEM.

Discussion

The importance of tight control over β₂AR signaling is exemplified by the existence of a complex set of distinct mechanisms for its downregulation upon prolonged activation (Shenoy & Lefkowitz, 2011), which include receptor phosphorylation by G protein‐coupled receptor kinases (GRKs) (Nobles et al, 2011) and the consequent phosphorylation‐triggered recruitment of arrestins for receptor uncoupling from Gs and endocytosis (Lohse et al, 1990; von Zastrow & Kobilka, 1992; Ferguson et al, 1996; Goodman et al, 1996; Cao et al, 1999) as well as the activity‐dependent, PKA‐mediated switching of β₂AR coupling from Gs to Gi (Daaka et al, 1997; Xiao et al, 1999a). Our surprising discovery of the role of S1928 phosphorylation in displacing the β₂AR unveils a novel negative feedback regulatory mechanism that targets the pervasive regulation of Ca_v1.2 by the β₂AR to prevent excessive Ca²⁺ influx into neurons. This mechanism is devoted to highly specific downregulation of β₂AR signaling to Ca_v1.2 but not AMPARs, revealing how tightly controlled the activity of Ca_v1.2 must be to ensure proper function.

Our first important finding is that phosphorylation of S1928 in α₁1.2 displaces the β₂AR from the C‐terminus of Ca_v1.2. S1928 is the most prominent PKA phosphorylation site in Ca_v1.2 as determined by biochemical methods, and S1928 phosphorylation is robustly induced by β adrenergic signaling (Hell et al, 1993b, 1995; De Jongh et al, 1996; Davare et al, 1999, 2000; Davare & Hell, 2003; Hulme et al, 2006a; Hall et al, 2007; Dai et al, 2009). However, its physiological role has remained an enigma, as it does not appear to significantly augment channel function in the heart (Ganesan et al, 2006; Lemke et al, 2008), which is mediated in part by phosphorylation of S1700 (Fuller et al, 2010; Hell, 2010; Fu et al, 2013, 2014). We now identify S1928 phosphorylation as a novel negative feedback mechanism for Ca_v1.2 regulation by β₂AR signaling. This is the first example of termination of G_sPCR activity by dissociation of a receptor–substrate complex and therefore introduces a new paradigm for the regulation of cell signaling by this widely expressed class of receptors.

Our second important finding is that dissociation of the β₂AR–Ca_v1.2 interaction by either S1928 phosphorylation during an initial ISO treatment or by Myr‐Pep2 prevents upregulation of channel phosphorylation and activity by subsequent ISO application. Could ISO‐induced displacement of the β₂AR from Ca_v1.2 result in endocytosis of the β₂AR, making it inaccessible to ISO and therefore to regulation? Evidently that is not the case, as abrogation of the re‐phosphorylation of S1700 and S1928 during a second ISO application within 3 min of the initial one was not affected by dynasore or pitstop (Fig EV3), two different endocytosis inhibitors whose efficacy is well established in our hands (Hall et al, 2013). Accordingly, preventing β₂AR endocytosis, which in some cell lines is a general mechanism of downregulating signaling through the β₂AR (von Zastrow & Kobilka, 1992; Cao et al, 1999; Shenoy & Lefkowitz, 2011), does not affect this Ca_v1.2‐specific form of downregulation. We conclude that it is the displacement of the β₂AR from Ca_v1.2 per se that is responsible for loss of subsequent signaling and not endocytosis of this receptor. This conclusion is also in accordance with the finding that the β₂AR can fully re‐associate with Ca_v1.2 within 10 min (Figs 5A and B, and EV2A and C). Re‐association of the β₂AR with Ca_v1.2 was paralleled by the ability of a second ISO application to induce re‐phosphorylation of S1700 and S1928 (Fig EV3A, lane 3, and C). This finding underlines the notions that ISO‐induced displacement of the β₂AR from Ca_v1.2 is not permanent and that the functionality of this interaction is reinstated within minutes.

Our third important finding is that displacement of the β₂AR from Ca_v1.2 is a specific process that downregulates signaling from the β₂AR to Ca_v1.2 without affecting β₂AR‐mediated regulation of GluA1, which also forms a signaling complex with the β₂AR. This mechanism is fundamentally different from the arrestin‐mediated downregulation of β₂AR signaling by endocytosis and by uncoupling from Gs (Shenoy & Lefkowitz, 2011). Most strikingly, downregulation of Ca_v1.2 stimulation is highly specific for Ca_v1.2, whereas arrestin‐mediated effects are cell‐wide affecting all β₂AR signaling. On a molecular level, arrestin is recruited to stimulated β₂ARs upon their phosphorylation by GRKs, whereas the PKA‐mediated phosphorylation of S1928 acts to displace the β₂AR from Ca_v1.2.

Our fourth important finding is that PTT‐LTP depends on Ca_v1.2 and its association of the β₂AR. PTT‐LTP is induced by prolonged stimulation at 5 Hz, which approximates the naturally occurring θ rhythm in the hippocampus (Mizuseki et al, 2009). Prolonged stimulation at the naturally occurring theta tetanus induces LTP (PTT‐LTP) if at the same time β adrenergic signaling is engaged (Thomas et al, 1996; Hu et al, 2007; Qian et al, 2012). PTT‐LTP is thought to be important for contextual learning under demanding situations (Hu et al, 2007). The β₂AR–Ca_v1.2 signaling cascade might thus be important for such learning.

The finding that the association of the β₂AR with Ca_v1.2 is critical for β adrenergic regulation of Ca_v1.2 has important further functional implications. Accordingly, the β₂AR must be localized in the immediate vicinity of Ca_v1.2 for effective signaling. This signaling is clearly mediated by the cAMP/PKA cascade as the PKA‐specific inhibitory PKI peptide prevented the ISO‐induced upregulation of L‐type currents (Fig 7D). Loss of cAMP signaling from the β₂AR to Ca_v1.2 upon their dissociation constitutes the first clear evidence for the hypothesis that cAMP signaling by certain GsPCR, especially the paradigmatic β₂AR (Kuschel et al, 1999; Chen‐Izu et al, 2000; Davare et al, 2001; Balijepalli et al, 2006; Nikolaev et al, 2010), is mediated by cAMP production that is localized within nanodomains, that is, domains smaller than 100 nm in diameter. The reasoning for this notion is that the average distance between more or less evenly distributed β₂ARs on the cell surface will not allow for regions devoid of β₂ARs that are larger than 100 nm; likely, such regions are much smaller. We also exclude that ISO‐triggered endocytosis is playing a role in functional uncoupling of the β₂AR from regulating Ca_v1.2 phosphorylation (Fig EV3). The localized regulation of Ca_v1.2 via cAMP signaling is consistent with earlier finding that in addition to AKAP150‐anchored PKA (Hall et al, 2007; Oliveria et al, 2007; Dittmer et al, 2014), G_s and adenylyl cyclase are also associated with Ca_v1.2 (Davare et al, 2001; Balijepalli et al, 2006).

Downregulation of β adrenergic augmentation of Ca_v1.2 activity might provide a brake necessary to ensure cell integrity, which could be jeopardized by the otherwise overpowering effects of a sustained increase in Ca²⁺ influx. In contrast, continued upregulation of GluA1 phosphorylation by prolonged β adrenergic stimulation might not be as detrimental because these receptors primarily conduct Na⁺ rather than Ca²⁺.

In conclusion, we demonstrate that S1928 phosphorylation of Ca_v1.2 upon β₂AR stimulation results in a temporary dissociation of the β₂AR from Ca_v1.2 with an equally fleeting but complete loss of Ca_v1.2 regulation by the β₂AR. This novel potent negative feedback mechanism adds to the surprisingly diverse arsenal of tools the cell developed to curb overactivation of βAR signaling and Ca_v1.2.

Materials and Methods

Animals

All procedures followed NIH guidelines and had been approved by the IACUC at UC Davis. S1928A KI mice were described by Lemke et al (2008) and conditional forebrain Ca_v1.2 KO mice by White et al (2008). All mice used in this study were between 8 and 12 weeks old.

Reagents, peptides, and antibodies

Isoproterenol bitartrate salt, ICI118551, CGP20712, and microcystin LR were from Sigma. Dynasore was from Tocris and Pitstop from Abcam. Protein A‐covered beads were from Repligen and Amylose beads from New England Biolabs. Polyvinyldifluoride (PVDF) membranes were from Millipore. Horseradish peroxidase‐coupled (HRP) protein A, ECL, and ECL plus reagents were from GE Healthcare. The chemiluminescent Femto substrate, EGTA, EDTA, Tween‐20, Triton X‐100, and tris(hydroxymethyl)aminomethane (Tris) were from Fisher Thermo Scientific. Other reagents were from the typical suppliers and of the usual quality.

All synthetic peptides (Appendix Table S1) were purchased from China Peptides (Shanghai, China). Origin and other details of antibodies are given in Appendix Table S2.

Fluorescence polarization (FP)

Fluorescein (FITC)‐labeled synthetic peptides (1 μM final conc.) were added to serial dilutions (8 times twofold, i.e., each time 1:1) of recombinant MBP‐tagged β₂AR C‐terminus in FP buffer (50 mM HEPES, 100 mM KCl, 1 mM MgCl₂, 0.05 mM EGTA, 5 mM NTA, pH 7.4) in black 384‐well polystyrene plates (Corning). FP was measured with a Synergy 2 (BioTek) plate reader with polarization filters to determine parallel and perpendicular fluorescence intensities of exciting (485/20λ) and emitted light (528/20λ). Data were acquired with Gen5 software. FP was calculated as P = (I _v − g*I _h)/(I _v + g*I _h); I _v and I _h constitute the vertical and horizontal fluorescence intensities, respectively, and g the correction factor for fluorescein. Data were analyzed with GraphPad Prism 5 for curve fitting and K _d determination by fitting binding curves to the equation Y = B*X/(K _d + X); B is the maximal FP value that would be reached at saturation as determined by extrapolation of the fitted curve (Lim et al, 2002).

Preparation of brain slices and use for biochemical analysis

Mice (8–12 weeks) were decapitated and brains placed into ice‐cold artificial cerebrospinal fluid (ACSF; in mM: 127 NaCl, 26 NaHCO₂, 1.2 KH₂PO₄, 1.9 KCl, 2.2 CaCl₂, 1 MgSO₄ and 10 D‐glucose, 290–300 mOsm/kg, saturated with 95% O_2, and 5% CO₂; final pH 7.3). About one‐third of the rostral and caudal ends of the brain were trimmed off. A total of 350‐μm‐thick forebrain slices containing hippocampus were prepared with a vibratome (Leica VT 1000A). Slices were equilibrated in oxygenated ACSF for 1 h at 30°C before transfer to incubation chambers, equilibration for 30 min at 32°C and treatment with vehicle (H₂O), ISO (10 μM). Slices were extracted with IP buffer containing protease and phosphatase inhibitors as above before IP of β₂AR, α₁1.2, and anti‐GluA1 and IB as above.

Methods for immunoprecipitation (IP) and immunoblotting (IB) as well as pull‐down and fluorescence microscopy are standard, and details are given in Appendix Supplementary Methods. The Ca_v1.2 glutathione S‐transferase (GST)‐fusion proteins of α₁1.2 are described in Hall et al (2006, 2007, 2013) and listed in Appendix Table S3. The maltose‐binding protein (MBP)‐tagged C‐terminus (CT) of human β₂AR (residues 326–413) is as in Joiner et al (2010).

Electrophysiology

Cell‐attached patch clamp recordings were performed as previously (Davare et al, 2001) with 500 nM (S)‐(−)‐BayK‐8644, 1 μM ω‐conotoxin GVIA, and 1 μM ω‐conotoxin MCVIIC in the patch pipette. Hippocampal slice recordings were basically as described (Qian et al, 2012). Exact details are given in Appendix Supplementary Methods.

Author contributions

TP, VDB, WAC, FH, YKX, GGM, CYC, MFN, and JWH designed experiments; TP, HQ, VDB, ZAM, DC, JLP, EAH, ORB, REW, CYC, and MFN performed experiments; TP, HQ, VDB, JLP, EAH, CYC, and MFN analyzed data; TP, MFN, and JWH wrote the manuscript.

Conflict of interest

The authors declare that they have no conflict of interest.

Supporting information

Appendix

Expanded View Figures PDF

Review Process File

The EMBO Journal (2016) 35: 1330–1345