Introduction

Hyperpolarization-activated, cyclic nucleotide-sensitive (HCN) ion channels are biophysical anomalies. Despite being structurally related to voltage-gated K⁺ channels, they activate in response to membrane hyperpolarization and pass a mixed Na⁺/K⁺ current. In addition, binding of cyclic nucleotides, particularly cAMP, to a conserved C-terminal cyclic nucleotide binding domain (CNBD) potentiates HCN channels by shifting the voltage-dependence of activation to more depolarized potentials, speeding activation, and slowing deactivation (DiFrancesco and Tortora, 1991; Wainger et al., 2001).

While the details of intramolecular transduction between cAMP binding and channel gating are yet to be fully elucidated, some key aspects are known: 1) the unbound CNBD is inhibitory — truncation of the CNBD potentiates the channel activation, similar to cAMP binding to the intact CNBD (Wainger et al., 2001); 2) the slowing of channel deactivation in response to cAMP binding occurs through a separate mechanism from the shift in activation voltage dependence, and cannot be replicated by truncation of the CNBD (Wicks et al., 2011; Sunkara et al., 2018); and 3) transduction of the signal for the cAMP-dependent shift in channel activation, but not deactivation, requires interactions between the C-linker, the N-terminal HCN domain (HCND), and the S4-S5 linker (Porro et al., 2019; Wang et al., 2020b).

Several proteins interact with HCN channels to modulate their activity (Peters et al., 2022). The inositol 1,4,5-triphosphate receptor-associated proteins, IRAG1 and LRMP/IRAG2 (lymphoid restricted membrane protein), are a family of endoplasmic reticulum (ER) transmembrane proteins that are isoform-specific regulators of HCN4 (Peters et al., 2020). IRAG and LRMP share some sequence homology, particularly in their coiled-coil motifs, and both have been found to regulate IP₃ receptor calcium release (Schlossmann et al., 2000; Geiselhöringer et al., 2004; Prüschenk et al., 2021). However, LRMP and IRAG have opposing effects on HCN4: IRAG1 causes a gain-of-function by shifting HCN4 activation to more depolarized membrane potentials in the absence of cAMP; in contrast, LRMP causes loss-of-function by inhibiting cAMP-dependent potentiation of HCN4 activation (Peters et al., 2020).

In this study, we focused on LRMP and investigated the interaction domains and mechanism by which it inhibits cAMP-dependent shifts in HCN4 activation. Our previous study showed that LRMP differs considerably from TRIP8b, a neuronal protein that also prevents cAMP-dependent shifts in HCN channel activation through direct antagonism of cAMP binding (Santoro et al., 2004; Zolles et al., 2009; Bankston et al., 2017; Saponaro et al., 2018). LRMP doesn’t appear to inhibit cAMP binding to the CNBD, as cAMP retains the ability to slow channel deactivation in the presence of LRMP (Peters et al., 2020). Furthermore, TRIP8b regulates all HCN channel isoforms (Zolles et al., 2009; Santoro et al., 2011), whereas LRMP is specific for the HCN4 isoform (Peters et al., 2020). These observations suggest that LRMP regulates HCN4 by interfering with an isoform-specific step in the signal transduction pathway that links cAMP binding to the shift in activation voltage dependence.

We tested this hypothesis using a combination of patch clamp electrophysiology and FRET. We found that the initial N-terminal 227 residues of LRMP bind to the N-terminus of HCN4 and that the intact HCN4 N-terminus is required for channel regulation by LRMP. Furthermore, we show that two HCN4-specific residues in the C-linker, P545 and T547, are necessary for LRMP to inhibit channel activation, consistent with a model in which LRMP acts via a signal transduction centre formed by the C-linker, N-terminus, and S4-S5 linker.

Materials and Methods

DNA Constructs

The LRMP construct in PCMV6-Kan/Neo, HCN1 in pcDNA3 (generously provided by Dr. Eric Accili), HCN4-2 in pcDNA3.1, HCN4-S719X in pCDNA3.1, and HCN4 Δ1-25 in pcDNA6 (also known as HCN4s, generously provided by Dr. Richard Aldrich) have been described previously (Proenza et al., 2002; Liao et al., 2012; Liu and Aldrich, 2011; Peters et al., 2020). HCN2 was subcloned from pcDNA4 into pcDNA3.1 for this study. Other constructs were synthesized by Twist Biosciences or using site-directed mutagenesis either in-house or by Applied Biological Materials. The HCN4 Δ1-65, HCN4 Δ1-130, and HCN4 Δ1-185 deletion clones were made using a site-directed mutagenesis kit (New England Biolabs) and a codon-optimized HCN4 plasmid in the pTwist-CMV-WPRE-Neo vector synthesized by Twist Biosciences. For FRET experiments, recombinant fusions of mHCN4 and mLRMP were constructed by introducing Cerulean (CER) or Citrine (CIT) fluorescent proteins using PCR-based cloning. The C-termini of HCN4 constructs were tagged with CER, while the C-termini of LRMP constructs were tagged with CIT. All clones used in this study are the murine sequences. All new constructs were confirmed by DNA sequencing (Barbara Davis Center BioResource Core, ACGT, or Plasmidsaurus). Detailed information about constructs can be found in supplemental table S1.

Cell Lines

HEK 293 cells (ATCC) were grown in a humidified incubator at 37°C and 5% CO₂ in high glucose DMEM with L-glutamine supplemented with 10% FBS, 100 U/mL penicillin, and 100 µg/mL streptomycin. Cells were transfected 48 hours prior to experiments and were plated on either protamine-coated glass coverslips (for patch clamp experiments) or poly-d-lysine coated glass-bottom dishes (for FRET experiments).

Patch clamp experiments were performed in either transiently transfected HEK293 cells, an HCN4 stable line in HEK293 cells (Zong et al., 2012), or seven new stable cell lines in HEK293 cells: HCN2, HCN4 Δ1-65, HCN4 Δ1-130, HCN4 Δ1-185, HCN4 Δ1-200, HCN2 A467P/F469T (HCN2 AF/PT), and HCN4 P545A/T547F (HCN4 PT/AF). Stable cell lines were made by transfecting HEK293 cells with the respective plasmids using Lipofectamine 2000 (Invitrogen) according to the manufacturer’s instructions. 48 hours post-transfection, 200 ug/mL of G418 disulfate (Alfa Aesar) was added to the cell culture media in place of pen-strep to select for stably transfected cells. Single cell clones were tested using whole-cell patch clamp and the clonal lines that exhibited the largest and most consistent currents were grown into stable cell lines.

Transient transfection of HCN4 constructs and/or LRMP was performed using Fugene6 (Promega) according to the manufacturer’s instructions. Transfections of all constructs that did not include fluorescent tags were performed with the addition of eGFP as a co-transfection marker.

Patch Clamp Electrophysiology

Cells were plated on sterile protamine-coated glass coverslips 24-48 hours prior to experiments. Cells on coverslip shards were transferred to the recording chamber and perfused (∼0.5 to 1 mL/min) with extracellular solution containing (in mM): 30 KCl, 115 NaCl, 1 MgCl₂, 1.8 CaCl₂, 5.5 glucose, and 5 HEPES. Transiently transfected cells were identified by green fluorescence.

Patch pipettes were pulled from borosilicate glass to a resistance of 1.0-3.0 MOhm when filled with intracellular solution containing (in mM): 130 K-Aspartate, 10 NaCl, 1 EGTA, 0.5 MgCl₂, 5 HEPES, and 2 Mg-ATP. 1 mM cAMP was added to the intracellular solution as indicated. All recordings were performed at room temperature in the whole-cell configuration. Data were acquired at 5 KHz, and low-pass filtered at 10 KHz using an Axopatch 200B amplifier (Axon Instruments), Digidata 1440A A/D converter and Clampex software (Molecular Devices). Pipette capacitance was compensated in all recordings. Membrane capacitance and series resistance (Rs) were estimated in whole-cell experiments using 5-mV test pulses. Only cells with Rs <10 MOhm were analyzed. Data were analyzed in Clampfit 10.7 (Molecular Devices).

Channel activation was determined from peak tail current amplitudes at -50 mV following 3-s hyperpolarizing pulses to membrane potentials between -50 mV and -170 mV from a holding potential of 0 mV. Normalized tail current-voltage relationships were fit by single Boltzmann equations to yield values for the midpoint activation voltage (V_1/2) and slope factor (k). All reported voltages are corrected for a calculated +14 mV liquid junction potential between the extracellular and intracellular solutions.

FRET Hybridization Assays

HEK293 cells expressing HCN4-CER and LRMP-CIT fusion proteins were examined 48 h after transfection using a Zeiss LSM 710 confocal laser scanning microscope. An area of 500– 2,500 µm² was selected from the overall field of view. Images were taken through a 40× water objective. CER and CIT were excited with separate sweeps of the 458- and 514-nm laser lines of an argon laser directed at the cell with a 458/514-nm dual dichroic mirror. Relative to full power, the excitation power for the imaging sweeps was attenuated to 1% for CER and 0.5% for Citrine.

Bleaching was performed by using multiple (20–60) sweeps of the CIT laser at full power. Bleaching was usually complete within 1-2 m. Emitted light was collected between 449 and 488 nm for CER and 525 and 600 nm for CIT. With this setup, there was no contamination of the relevant CER signal from the CIT. For each experiment, the photomultiplier tube gain was adjusted to ensure that the maximum pixel intensity was not >70% saturated. Fluorescence intensity was then measured by drawing regions of interest (ROIs) around the cytoplasmic portion of the cell in ImageJ (Schneider et al., 2012). Masks were occasionally used to eliminate bright fluorescent puncta within the cell (this was a rare occurrence in the CER signal). Percent FRET (E) was calculated as:

where I_CERpost is the CER intensity after bleaching and I_CERpre is the CER intensity before bleaching.

Statistical Analysis

All statistical analysis was performed using JMP 15 software (SAS Institute). Tests for differences in the average midpoint of activation for a given HCN channel construct in the presence of LRMP and/or 1 mM cAMP were performed with a 2-way ANOVA. The main independent variables were the absence or presence of LRMP and the absence or presence of 1 mM cAMP in the pipette solution. Differences in the effects of cAMP in the absence or presence of LRMP were analyzed using an interaction term between the main independent variables. P < 0.05 was used as the cut-off for a significant effect in all cases.

Results

The N-terminus of LRMP is necessary and sufficient to regulate HCN4

We previously showed that LRMP significantly reduces the cAMP-dependent depolarizing shift in HCN4 activation and that it does not regulate HCN1 or HCN2 (Peters et al., 2020). We next sought to identify a subdomain within LRMP that is responsible for this regulation. We began with a truncated LRMP construct with a citrine fluorescent protein replacing the C-terminal ER transmembrane and lumenal domains (LRMP 1-479Cit; Figs. 1A and 1B; Table 1). We found that LRMP 1-479Cit inhibited HCN4 to an even greater degree than the full-length LRMP, likely because expression of this tagged construct was improved compared to the untagged full-length LRMP, which was detected by co-transfection with GFP. A key feature of LRMP in our original study is that it does not prevent binding of cAMP to the CNBD of HCN4. This was also the case for LRMP 1-479Cit, as indicated by the significant slowing of deactivation by cAMP even in the presence of LRMP 1-479Cit (P = 0.0036; Fig. 1C). These results indicate that the ER transmembrane and luminal domains of LRMP are not required for regulation of HCN4 and they support the idea that LRMP limits cAMP potentiation of HCN4 by interfering with a downstream step in the cAMP signal transduction pathway.

Figure 1:

Table 1:

To further resolve which regions of LRMP are required to regulate HCN4, we tested a series of additional truncated LRMP constructs (Fig. 2) for their ability to prevent cAMP-dependent shifts in HCN4 activation. We first split LRMP into two fragments: the LRMP 1-227 construct contains the N-terminus of LRMP with a cut-site near the N-terminus of the predicted coiled-coil sequence, while LRMP 228-539 contains the remainder of the protein. We found that LRMP 1-227 recapitulated the effects of full-length LRMP, while LRMP 228-539 had no effect on HCN4 gating (Figs. 2A, 2B, and 2E; Table 2). However, when we further split the N-terminal domain of LRMP into two fragments, neither LRMP 1-108 nor LRMP 110-230 regulated HCN4 (Figs. 2C, D, and 2E; Table 2). Thus, the first 227 residues of LRMP are sufficient to regulate HCN4, with residues in both halves of the LRMP N-terminus necessary for the regulation.

Figure 2:

Table 2:

The N-terminus of HCN4 is required for regulation by LRMP

We next examined the domains of the HCN4 channel that are necessary for regulation by LRMP. Since LRMP regulates only the HCN4 isoform, we focused on the large non-conserved regions in the distal N- and C-terminals as potential sites for LRMP regulation. We first examined the N-terminus by testing the ability of LRMP to regulate a series of HCN4 channels with progressively larger truncations (Δ1-25, Δ1-65, Δ1-130, Δ1-185, and Δ1-200). The four smaller deletions all produced functional channels with normal cAMP-dependent shifts in activation (Fig. 3), while the HCN4 Δ1-200 construct produced insufficient current amplitude for analysis.

Figure 3:

When the first 25 residues in the HCN4 N-terminus were truncated, LRMP still prevented cAMP from shifting HCN4 activation, just as in the WT HCN4 channel (Fig. 3A and 3E; Table 1). Truncation of residues 1-62 led to a partial LRMP effect where cAMP caused a significant depolarizing shift in the presence of LRMP, but the activation in the presence of LRMP and cAMP was hyperpolarized compared to cAMP alone (Fig. 3B, C and 3E; Table 1). In the HCN4Δ1-130 construct, cAMP caused a significant depolarizing shift in the presence of LRMP; however, the midpoint of activation in the presence of LRMP and cAMP showed a non-significant trend towards hyperpolarization compared to cAMP alone (Fig. 3C and 3E; Table 1). Finally, truncation of the first 185 residues, which removes most of the non-conserved region of the HCN4 N-terminus, completely abolished LRMP regulation of the channel (Fig. 3D and 3E; Table 1); when LRMP was present cAMP caused a significant depolarizing shift in the HCN4 Δ1-185 activation, and the midpoint of activation in the presence of both LRMP and cAMP was not significantly different from the midpoint in the presence of cAMP alone. These results suggest that the non-conserved region in the N-terminus of HCN4 are a primary site for functional regulation by LRMP.

We also investigated LRMP regulation of two C-terminal truncations in HCN4: HCN4 S719X, which removes the C-terminus distal to the CNBD (Liao et al., 2012), and HCN4 V604X, which additionally removes the CNBD. Truncation of the distal C-terminus did not prevent LRMP regulation. In the presence of both LRMP and cAMP the activation of HCN4-S719X was still significantly hyperpolarized compared to the presence of cAMP alone (Figs. 4A and 4B; Table 1). And the cAMP-induced shift in HCN4-S719X in the presence of LRMP (∼7 mV) was less than half the shift in the absence of LRMP (∼18 mV). HCN4-V604X, which truncates the channel between the C-linker and CNBD, shifts channel activation to more depolarized potentials and prevents cAMP-dependent regulation (Figs. 4C and 4D; Table 1). This is similar to the effects of the homologous HCN2-V526X mutant (Wainger et al., 2001). LRMP did not alter the gating of HCN4-V604X in the absence of cAMP, and the lack of cAMP binding limits the investigation of any LRMP inhibition of cAMP-dependent potentiation (Figs. 4C and 4D). Together, these results show significant LRMP regulation of HCN4 even when the distal C-terminus is truncated, consistent with a minimal role for the C-terminus in the regulatory pathway.

Figure 4:

Table 3:

The N-terminus of LRMP associates with the N-terminus of HCN4

We next used FRET hybridization assays to examine potential interactions between small fragments of LRMP and HCN4 in a cellular environment. This approach has been used to define interactions between the N- and C-termini of EAG channels as well as between Calmodulin and Ca_V1.2 (Gianulis et al., 2013; Erickson et al., 2003). Fragments of LRMP tagged on the C-terminus with Citrine and fragments of HCN4 tagged on the C-terminus with Cerulean were co-expressed in HEK293 cells and FRET was assessed using the acceptor photobleaching method (Fig. 5; Table 3) (Bastiaens and Jovin, 1996; Wouters et al., 1998; Klipp et al., 2020).

Figure 5:

A Citrine-tagged construct corresponding to the functionally active domain of LRMP (LRMP 1-230) did not significantly FRET with the full-length HCN4 N-terminus (HCN4 NF, residues1-260; Fig. 5A). However, these fragments are large and may orient the fluorophores at a distance outside of the range for FRET, which is ∼20-80Å. Indeed, when we expressed LRMP 1-230 with smaller fragments of the HCN4 N-terminus — HCN4 N1 (residues 1-125) and HCN4 N2 (residues 126-260) — we measured significant FRET compared to control cells co-transfected with Cer-HCN4 fragments and Citrine alone (i.e., without any LRMP sequence; Fig. 5A). Smaller fragments of LRMP — LRMP L1 (residues1-108) and LRMP L2 (residues 110-230) also exhibited significant FRET with the whole HCN4 N-terminus and with HCN4 N-terminal fragments (Fig. 5B). No significant FRET was observed between LRMP fragments and the HCN2 N-terminus, which shares the conserved HCND with HCN4 (Figs. 5B). None of the LRMP fragments tested exhibited significant FRET with the HCN4 CNBD compared to control experiments (Fig. 3A and 3B; Table 3). Ultimately these data suggest that LRMP interacts with the non-conserved distal N-terminus of HCN4.

Mutants in the HCN4 C-linker disrupt LRMP’s functional effects

Prior work has shown that transduction of cAMP-binding to shifts in channel activation require a tripartite interaction of the N-terminal HCND, the C-linker, and the S4-S5 linker (Porro et al., 2019; Wang et al., 2020b; Kondapuram et al., 2022). Since LRMP interacts with the HCN4 N-terminus and disrupts cAMP-dependent potentiation downstream of the cAMP binding site, we hypothesized that it may act via this cAMP transduction centre. Although most elements of the transduction centre are conserved among HCN channel isoforms, the C-linker conformation differs subtly between structures of HCN1 and HCN4 (Fig. 6A; Lee and MacKinnon, 2017; Saponaro et al., 2021). We identified two HCN4-specific residues, P545 and T547 in the C-linker, that are close in proximity to the transduction centre that could account for these changes (Fig. 6A). When we mutated these residues to the analogous HCN2 amino acids, the HCN4-P545A/T547F channel responded normally to cAMP with an ∼10 mV shift in activation voltage (Figs. 6B and 6C; Table 1). However, HCN4-P545A/T547F was insensitive to LRMP (Figs. 6B and 6C; Table 1), indicating that the unique HCN4 C-linker is necessary for regulation by LRMP. The HCN4 C-linker alone was not sufficient to confer regulation by LRMP onto HCN2 in the reverse mutant (HCN2 A467P/F469T), consistent with our data showing that the HCN4 N-terminus is also required (Figs. 6D and 6E; Table 1). Thus, LRMP appears to regulate HCN4 by altering the interactions between the C-linker, S4-S5 linker, and N-terminus at the cAMP transduction centre.

Figure 6:

The HCN4 N-Terminus and cAMP transduction centre confer LRMP regulation to HCN2

Given that LRMP requires the unique HCN4 N-terminus to regulate the channel, we investigated whether this region is sufficient to confer LRMP regulation to HCN2 channels, which are normally LRMP-insensitive. We found that a previously described HCN4-2 chimera containing the HCN4 N-terminus and transmembrane domains (residues 1-518) with the HCN2 C-terminus (442-863) (Liao et al., 2012) was partially regulated by LRMP (Fig. 7A and 7B). In the presence of LRMP and 1 mM cAMP, the HCN4-2 activation was shifted towards more hyperpolarized potentials compared to the presence of 1 mM cAMP alone and the magnitude of the cAMP-dependent shift in activation was reduced from ∼20 mV in the absence of LRMP to ∼9 mV when LRMP was present. Because this construct still contains the HCN2 C-linker sequence, we next made a targeted chimera of HCN2 that contains the distal HCN4 N-terminus (residues 1-212) and the HCN2 transmembrane and C-terminal domains with 5 point mutants in non-conserved residues of the S5 segment and C-linker elbow (M338V/C341V/S345G/A467P/F469T). The resulting HCN4-2 VVGPT channel has a similar voltage-dependence of activation as HCN2 and a normal response to cAMP in the absence of LRMP (Table 1; Fig. 7C and 7D). Importantly, the HCN4-2 VVGPT channel is insensitive to cAMP in the presence of LRMP (Fig. 7C and 7D), indicating that the HCN4 N-terminus and cAMP-transduction centre residues are sufficient to confer LRMP regulation to HCN2.

Figure 7:

Discussion

LRMP is one of only two isoform-specific regulators of HCN4 to be discovered. Importantly, this study of LRMP’s mechanisms of action also reveals unique features of HCN4 function that could contribute to its physiological roles and could inform development of isoform-specific drugs. Although LRMP prevents cAMP-dependent shifts in HCN4 activation, it does not act by preventing cAMP from binding to the channel. Instead, we show here that the N-terminal domains of LRMP and HCN4 are required for both physical interaction and regulation. Our data further show that LRMP acts by disrupting transduction between cAMP binding and the shift in voltage-dependence in a manner that depends on HCN4-specific residues in the C-linker that are in close proximity to a previously described cAMP transduction centre formed by the C-linker, N-terminus, and S4-S5 linker.

An intramolecular transduction centre between the C-Linker, HCND, and S4-S5 linker links cAMP binding to shifts in activation.

The binding of cAMP to the CNBD of HCN channels causes two distinct effects – a depolarizing shift in the voltage-dependence of activation and a slowing of channel deactivation (Wicks et al., 2011; Sunkara et al., 2018). The slowed deactivation appears to be caused by stabilization of the open state in the presence of cAMP (Wicks et al., 2009, 2011; Hummert et al., 2018). In contrast, voltage-clamp fluorimetry experiments and computational modelling suggest that shifts in channel activation are caused by a hyperpolarized voltage-dependence of S4 movement in the presence of cAMP (Magee, 2017; Hummert et al., 2018). The two separate cAMP effects allows slowing of deactivation to be used as a proxy for cAMP binding even in the absence of cAMP-dependent shifts in activation (Liao et al., 2012; Porro et al., 2019; Peters et al., 2020).

Recent studies indicate that the cAMP-dependent shift in activation requires an intramolecular cAMP transduction centre formed by interactions between the C-linker, N-terminal HCND, and S4-S5 linker (Weißgraeber et al., 2017; Porro et al., 2019; Wang et al., 2020b; Saponaro et al., 2021; Kondapuram et al., 2022). The inhibitory effects of the unbound CNBD are stabilized by an interaction between a conserved lysine in the C-linker elbow (K542 in HCN4; Fig. 8A) with the HCND of the adjacent subunit at M233 (Porro et al., 2019; Kondapuram et al., 2022), likely through a downstream hydrophobic contact between the HCND and the voltage sensor (Porro et al., 2019; Elbahnsi et al., 2023). Another interaction, between the C-linker residue E556 and R232 on the HCND of the same subunit, is required for the cAMP-dependent shift in activation (Porro et al., 2019; Wang et al., 2020b). The K542A/E556A double mutation eliminates the cAMP-dependent shift in activation, but not the slowing of deactivation (Porro et al., 2019). Interestingly, the S4-S5 linker does not appear to be required for voltage-dependent gating in HCN channels (Flynn and Zagotta, 2018; Cowgill et al., 2019). However, in HCN4 a Mg²⁺ coordination site between residues H406 and D410 on the S4-S5 linker and H552 and E566 on the C-linker is required for the cAMP-dependent shift in activation (Fig 6A; Fig. 8A; Saponaro et al., 2021).

Figure 8:

Proposed model: LRMP disrupts cAMP regulation of HCN4 activation at the cAMP transduction centre.

In our proposed model for how LRMP disrupts the cAMP-dependent shift in HCN4 activation (Fig. 8), LRMP is bound to HCN4 through an interaction between the distal N-terminus of HCN4 and the N-terminus of LRMP. Within the HCN4 N-terminus, the HCND, which is conserved across isoforms, is known to be required for functional channel expression (Tran et al., 2002; Wang et al., 2020b). However, little is known about the function of the more distal HCN4 N-terminus, which is not resolved in channel structures (Lee and MacKinnon, 2017; Saponaro et al., 2021) and not conserved between HCN channel isoforms. We found that truncation of the non-conserved HCN4 N-terminus abolishes regulation by LRMP without affecting other aspects of HCN4 function. This finding was further corroborated by FRET experiments showing interactions between LRMP and the HCN4 N-terminus, but not the CNBD or more distal C-terminal regions.

We hypothesize that the interaction with the N-terminus of HCN4 allows LRMP to allosterically disrupt the transduction between cAMP binding and channel activation. Unlike the neuronal protein TRIP8b, which directly competes for the cAMP binding site to regulate all HCN channel isoforms (Bankston et al., 2017; Saponaro et al., 2018), LRMP doesn’t appear to act at the cAMP binding site, as cAMP still slows deactivation in the presence of LRMP (Fig. 1C). In support of this model, we did not measure significant FRET between Citrine-tagged LRMP fragments and the Cerulean-tagged CNBD (Fig. 5). Because of the known interactions of the C-linker with the N-terminal HCND (Porro et al., 2019; Wang et al., 2020b; Kondapuram et al., 2022), we looked for unique residues in the HCN4 C-linker that could be necessary for functional regulation by LRMP. Interestingly, two C-linker residues, P545 and T547, between the sites of interaction with the HCND are conserved in HCN4 across mammalian species, but are not conserved in HCN1 or HCN2 (Fig. 6A). Given that these residues are the only ones in the region that differ between isoforms, they appear to be responsible for the different orientations of the C-linker in the HCN1 and HCN4 cryoEM structures (Lee and MacKinnon, 2017; Saponaro et al., 2021).

Despite their homology, cAMP signal transduction in HCN4 appears less robust than in HCN2. The cAMP-dependent shift in HCN4 (∼14 mV) is smaller than in HCN2 (∼20 mV; Table 1); in HCN4, but not HCN2, the transduction of cAMP is sensitive to divalent cations (Saponaro et al., 2021; Peters et al., 2023); and cAMP-dependent shifts in activation can be disrupted in HCN4 by regulatory factors such as cyclic-dinucleotides and LRMP (Lolicato et al., 2014), It’s likely that the difference in orientation of the C-linker relative to the S4-S5 linker accounts for differences in cAMP signal transduction in HCN4 compared to HCN2. Indeed, mutating the non-conserved residues to the corresponding HCN2 amino acids (P545A/T547F) abolishes LRMP regulation of HCN4 (Fig. 6). Importantly, chimeric HCN2 channels that include the distal HCN4 N-terminus as well as mutations in non-conserved residues in S5 and the C-linker are rendered LRMP sensitive (Fig. 7). These experiments not only show that LRMP acts via this region, but also suggest that the orientation of the C-linker elbow is critical for isoform-specific regulation in HCN4.

Potential physiological implications

The first half of LRMP’s cytosolic domain (residues 1-230) that make up the N-terminus of the protein is necessary and sufficient to interact with and regulate HCN4. Because the C-terminus of LRMP is embedded in the ER, the N-terminal region of LRMP would naturally be in closer proximity to HCN4 in the plasma membrane. LRMP also interacts with and regulates Ca²⁺ release through inositol triphosphate (IP₃) receptors in the ER membrane, likely via a site in the coiled-coil region (Prüschenk et al., 2021). Together these results suggest the intriguing possibility of coordination between the activity of IP₃Rs and HCN4 and the formation of ER-plasma membrane junctions in cells where LRMP and HCN4 are co-expressed. For example, in sinoatrial myocytes HCN4 and SR Ca²⁺ release, including through IP3 receptors, are both known to regulate pacemaking (DiFrancesco, 2010; Peters et al., 2021; Capel et al., 2021). A potential interaction with LRMP (or IRAG1), could serve to coordinate these important processes. Thus, future studies to identify tissues where LRMP and HCN4 are co-expressed are of clear importance.

Limitations

Unfortunately, the available HCN4 structures do not resolve the distal N-terminus (Shintre* et al., 2018; Saponaro et al., 2021), and the structure of LRMP has yet to be resolved. This lack of structural information hindered our decisions about specific cut sites for LRMP and HCN4 constructs and restricts our ability to predict the precise residues that are involved in the described interactions. For example, we found that LRMP interacts with isolated fragments representing each half of the HCN4 N-terminus. This could be explained by a diffuse interaction, multiple sites, or our cut site overlapped the interaction site. The partial disruption of LRMP’s functional regulation in the HCN4 Δ1-62 and Δ1-130 deletions suggest that multiple or diffuse interactions are more likely, but we are unable to give a definitive conclusion. Ultimately, these questions will require a structure the LRMP-HCN4 interaction sites.

Summary

Overall, these data support a model for LRMP regulation of HCN4 where LRMP interacts with the HCN4 N-terminus (Fig. 8) to allosterically disrupt cAMP signal transduction between the C-linker, N-terminus, and S4-S5 linker (Porro et al., 2019; Wang et al., 2020b). Our data suggest that LRMP disrupts the C-linker from transferring the cAMP-dependent conformational change in the CNBD to this transduction-centre, thereby preventing activation potentiation. The isoform specificity is determined by both the unique distal N-terminus and the unique orientation of the C-linker of HCN4. While a potential physiological role for LRMP regulation of HCN4 remains unknown, our data show that LRMP is a useful biophysical tool to study the intramolecular signal transduction between cAMP binding and the shift in HCN4 activation.

Acknowledgements

This work was funded by NIH grants R01HL088427 and R01GM140004 to CP and R35GM137912 to JB. CHP was funded by an American Heart Association Postdoctoral Fellowship 830889. The authors gratefully acknowledge the contributions of Abby Camenisch and Karin Nunley.

The authors declare they have no conflict of interest.