Blockade of TRPV channels by intracellular spermine

ThermoTRPV (TRPV1–4) channels are non-selective, Ca²⁺-permeable, cationic channels gated by a variety of stimuli, including noxious heat (which is reflected in their classification as “thermo”), pH, natural plant-derived compounds (such as capsaicin and cannabinoids), endogenous ligands, and synthetic drugs (Yuan, 2019). They are involved in a variety of sensory functions, including thermo- and pain sensation, and dysfunction of these channels underlies multiple diseases, ranging from various kinds of pain and hyperalgesia (TRPV1; Patapoutian et al., 2009; Marrone et al., 2017) to immune system-related diseases (TRPV2; Perálvarez-Marín et al., 2013), chronic itch and skin Olmsted syndrome (TRPV3; Nilius et al., 2014), and neurological and musculoskeletal disorders (TRPV4; White et al., 2016; Deng et al., 2018), for many of which there is no therapy. Among TRPVs, TRPV1 exhibits a very clear voltage dependence (Caterina et al., 1997; Tominaga et al., 1998), although, unlike canonical voltage-sensitive channels, the S4 helix does not serve as a voltage sensor, and the mechanism of TRPV1 voltage dependence is not well understood (Yang et al., 2020). While current pharmacology of TRPVs is quite extensive, there is still a need to identify novel regulatory mechanisms that could be exploited for more granular and fine-tuning of thermoTRPV activity.

The natural polyamines (PAs) putrescine, spermidine, and spermine, are small organic cations, present in sub-millimolar concentrations in the cytoplasm of virtually all cells and in the extracellular fluid, and have long been recognized as critical for cell growth and proliferation (Pegg, 2016; Igarashi and Kashiwagi, 2019). Intracellular PA levels decrease with age, while pathological alterations of PA levels affect cognitive functions and have been linked to brain ageing and degeneration (Minois et al., 2011; Skatchkov et al., 2014, 2016; Saiki et al., 2019; Sánchez-Jiménez et al., 2019), as well as to cancer (Casero et al., 2018; Li et al., 2020). Other well-studied PA functions are regulation of protein expression via interaction with DNA and RNA molecules (Dever and Ivanov, 2018) and modulation of ion channels, including inward-rectifier potassium (Kir) channels (Lopatin et al., 1994; Méndez-González et al., 2016), cyclic nucleotide-gated channels (Lu and Ding, 1999; Guo and Lu, 2000), ionotropic glutamate receptors (Bowie and Mayer, 1995; Bähring et al., 1997; Bowie et al., 1998; Twomey et al., 2018), and voltage-gated Na⁺ channels (Fleidervish et al., 2008).

TRPVs are widely expressed in excitable membranes of neurons of brain, central and peripheral nervous systems, pancreatic β-cells, and cardiomyocytes, and tight control of channel activity is critical for perception of noxious stimuli and pain. Thus, TRPVs are promising therapeutic targets for human pain conditions. There have been several isolated reports of voltage-dependent PA inhibition of TRP channel family members, including TRPC4 (Kim et al., 2016, 2020), TRPM4 (Nilius et al., 2004), and TRPM7 (Zhelay et al., 2018). It has also been shown that extracellular spermine permeates, potentiates, and weakly activates TRPV1 channels (Ahern et al., 2006), but the effects of intracellular PAs on TRPV sub-family members remain unexplored. Here, we characterize the voltage-dependent inhibition of TRPVs by intracellular polyamines. The data reveal a strong, physiologically relevant action that varies in potency between sub-family members, pointing to the possibility of developing channel-selective modulators of this inhibition.

The experiments were performed on cloned full-length human TRPV1 (Caterina et al., 1997), TRPV3 (Peier et al., 2002; Smith et al., 2002), and TRPV4 (Liedtke et al., 2000; Strotmann et al., 2000) channels in pV10 mammalian expression vector with a C-terminal EGFP tag.

Data presented in Fig. 6 were fit using Microsoft Excel Solver independently for each channel assuming the same values for all parameters (six for a two-site model and four for a simplified one-site model) for each polyamine concentration. Fitting was performed via minimization of RMSD between experimental data and fitting curve using the equation presented on Fig. 5 C.

TRPV3 pore structure (presented in Fig. 5 A) was generated with VMD suite (Humphrey et al, 1996). Pore electrostatic profiles (presented in Fig. 5 D) were generated with MOLEonline Server (https://mole.upol.cz) and UCSF ChimeraX 1.5 suite (Pettersen et al., 2021).

When expressed in COSm6 cells and activated in cell-attached patches by the TRPV3-specific agonist 2-APB (present at 1 mM in the patch pipette), recombinant C-terminal EGFP-tagged TRPV3 typically elicited large currents, up to a few nanoamps per patch. In our initial studies, we first observed a consistent kink in cell-attached current-voltage relationships around the reversal potential (Fig. 1 A). TRPV3-specific conductance, measured in response to a linear 2 s long voltage ramp (−100 to 100 mV), decayed in the vicinity of 0 mV and recovered at higher de- and hyper-polarizing potentials (Fig. 1 A), which hinted at a potential voltage-dependent inhibition by intracellular factors around 0 mV. As previous studies indicated, inhibition of multiple types of channels, including a few other TRP family members, by intracellular polyamines, we tested the effect of the intracellular polyamine spermine on human TRPV3 channel activity. Following patch excision, TRPV3 currents rapidly linearized, losing the kink in the I-V curve around 0 mV (Fig. 1 B). Application of spermine to the cytoplasmic side of inside-out patches at 10 or 100 µM restored a kink, and I-V and G_rel-V curves (Fig. 1 B) became qualitatively similar to those obtained in cell-attached configuration (Fig. 1 A), suggesting that endogenous polyamines, or other organic cations, inhibit the channel in the intact cell.

We examined TRPV3 conductance in excised patches in the presence of 10, 100, and 1,000 µM spermine or of the shorter polyamines spermidine and putrescine at 100 µM (Fig. 2). The ratio of TRPV3 currents in the presence versus absence of spermine (G_rel) was used as a measure of efficiency and voltage-dependence of spermine inhibition (Fig. 1). Inhibition of TRPV3 channels by spermine was clearly concentration-dependent (Fig. 2, A and D). Spermidine also inhibited TRPV3 channels in excised inside-out patches (Fig. 2, B and E) with similar characteristics, although compared to spermine, the inhibition was less potent (Fig. 2 E). The effect of 100 µM intracellular putrescine was negligible (Fig. 2, C and E). Overall, these observations are in line with data on natural polyamine block of Kir channels (Lopatin et al., 1994, 1995; Nichols and Lee, 2018; Lee and Nichols, 2023) and of CNG channels (Lu and Ding, 1999; Guo and Lu, 2000). In the latter, there is good evidence that spermine acts as a permeant blocker, and that relief of block at positive voltages is due to voltage-driven permeation of spermine to the extracellular side of the membrane. To examine the possibility that a similar mechanism underlies relief of TRPV3 inhibition at positive voltages, we tested the effect of the AMPAR antagonist (Koike et al., 1997; Twomey et al., 2018), 1-naphthyl-acetyl spermine (NASPM). NASPM is a spermine analog, with a bulky aromatic group at one end of the molecule (Fig. 3 A, inset). At 100 µM, NASPM also exhibited strong voltage-dependent inhibition at negative potentials but, in contrast to the natural polyamines, the block was not relieved at positive potentials (Fig. 3). At 10 µM and more so at 1 µM, however, there was clear evidence of relief at the most positive potentials (Fig. 3, A and D), which suggests weak permeation. Given that the head group is too large to permeate the TRPV pore resolved in static cryo-EM structures (Singh et al., 2018; Zubcevic et al., 2018a; Deng et al., 2020), this would suggest that there may be plasticity of the TRPV3 selectivity filter, as previously suggested (Ferreira and Faria, 2016) for TRPV1 (Chung et al., 2008) and TRPV2 (Zubcevic et al., 2018b).

We then characterized the action of the most potent of the natural PAs, spermine, on human TRPV1 and TRPV4 (Fig. 4, A and B), using the same stepped voltage protocol as above. Owing to lack of high potency agonists, we did not evaluate polyamine inhibition of human TRPV2 (Juvin et al., 2007; Neeper et al., 2007). In excised patches, agonist-activated I-V relationships for excised TRPV4 were again almost linear, without noticeable rectification at either positive or negative membrane potentials (Fig. 4 B), whereas TRPV1 demonstrated a strong outward rectification (Fig. 4 A), in agreement with previous findings (Yang et al., 2020). Independent of this intrinsic outward rectification, spermine added to the cytoplasmic side at concentrations 10, 100, and 1,000 µM caused a similar inhibition of both hTRPV1- and hTRPV4-specific currents, again with a maximum efficacy near the reversal potential (Fig. 4, A and B).

Inhibition of each of TRPV1, 3, and 4 channels by intracellular spermine was rapid, with no measurable time-dependence, and shared similar voltage- and concentration-dependence, with most efficient blocking occurring around the reversal potential, resulting in the characteristic “dip” on the G_rel-V relationship (Fig. 1 B, Fig. 2 D, and Fig. 4). Inhibitory potency was not identical, the rank order being most potent for TRPV3, less so for TRPV1, and weakest for TRPV4. Fitting of G_rel-V plots with the sum of two Boltzmann distributions (Fig. 2 D and Fig. 4) also revealed the strongest concentration-dependent shift of midpoint potential (V_mid) for TRPV3 (Table 1). For each channel subtype, inhibition by 100 µM spermine was marginal to moderate at what would be a typical cellular resting membrane potential (−70 mV), but inhibition was ∼80–90% for each channel at around 0 mV, which would effectively inhibit non-selective cationic TRPV currents during the depolarization phase of an action potential.

Thus, PA inhibition of thermoTRPVs, similarly to Kir channels and especially to CNG channels, is voltage and concentration dependent, and follows the same dependence on the length/charge of PAs. The voltage-dependence implies a binding site within the electric field deep in the TRPV pore, potentially at the selectivity filter or in the cytoplasmic cavity (Fig. 5 A). Another hallmark of PA inhibition of thermoTRPVs, particularly obvious in the case of TRPV1, is a clear second descending phase for all spermine concentrations (Figs. 4 and 6), also typical for CNG channels (Guo and Lu, 2000), but not for AMPARs (Bowie et al., 1998), which hints at a similar mechanism for TRPV and CNG channel block.

To explore a mechanism of spermine block of thermoTRPV channels, we therefore followed the modeling approach developed for PA inhibition and permeation of CNG channels (Guo and Lu, 2000). To account for the tri-phasic behavior of G_rel-V relationships, most obvious for TRPV1 and TRPV4 (Figs. 4 and 6), the model assumes two PA binding sites with different affinity, resulting in two blocked states without direct transitions between them (Fig. 5, B and C). PA can permeate when bound in conformation a, but not in conformation b (Fig. 5, B and C). We applied this model to the full concentration- and voltage-dependent spermine data on each of hTRPV1, 3, and 4, allowing the controlling rate constants and effective charges to vary between each channel, resulting in reasonably well predicted G_rel-V relationships for 10–1,000 μM spermine in each case (Fig. 6, solid lines, and Table 2). The equilibrium constants K_a and K_b, which reflect the steady state binding equilibria at the two sites, were quite similar for each channel. The constant K, which reflects the relative probability of spermine bound in conformation a permeating the channel versus going back to the intracellular solution, was approximately one order of magnitude lower for TRPV3 than TRPV1 or TRPV4, indicating higher permeability in the latter. Although the data sets were not sufficient to warrant formal modeling of the block by spermidine or putrescine, the same model is also likely to predict the results, with lower binding affinities, and potentially higher permeation rates for these smaller polyamines.

We also tested a one-site model, assuming only one PA binding site (conformation a in Fig. 5 C) described by the same equations as in Fig. 5 C without the term K_b (labeled in blue). Although this simplified model fit the data reasonably well in the case of TRPV3, it did not describe the inhibition of TRPV1 and TRPV4 (Fig. 6, dotted lines), and was therefore rejected as too simple.

Expressed in both the central and peripheral nervous systems, TRPV1–4 are involved in the regulation of a multitude of physiological processes, including heat sensation, thermoregulation, itch, pain perception, and nociception, but also in immune response and (neuro)inflammation, osmoregulation, mechano-transduction, and bone formation. TRPV channel open probability, conductance, and ion selectivity, are extensively modulated by a number of parameters, including lipids (Lukacs et al., 2013; Cao et al., 2013), phosphorylation (Chung et al., 2008), and hydroxylation (Karttunen et al., 2014), as well as the extent of activation and even the type of agonist (Chung et al., 2008). We demonstrate here that each of TRPV1, 3, and 4 is effectively inhibited by physiological concentrations (Watanabe et al., 1991; Bowie and Mayer, 1995) of the endogenous intracellular PA spermine. We showed that this inhibition is concentration- and voltage-dependent, and is maximal around the reversal potential. Given extensive TRPV representation in excitable membranes, such non-linear voltage-dependent inhibition could be a powerful modulator, in particular during action potentials. Although a number of natural and synthetic agonists and inhibitors have been identified for TRPVs, so far, no voltage-dependent modulators are available. The steeply voltage-dependent PA block that we identify here thus introduces a potentially very dynamic and flexible regulatory mechanism.

Although total cellular PA content is estimated to be in the millimolar range, most intracellular PAs are bound to negatively charged macromolecules, including DNA, RNA, ATP, lipids, and proteins. Free spermine has been estimated to be in the 10–100 µM range, and other PAs at slightly higher concentrations, depending on the cell type (Watanabe et al., 1991; Bowie and Mayer, 1995). Although normally stable, intracellular PA content may be significantly altered in pathological conditions, including cancer (Li et al., 2020; Miska et al., 2021) and Alzheimer’s disease (Morrison and Kish, 1995; Seidl et al., 1996; Inoue et al., 2013). As we show, at a likely physiological concentration of 10 µM, spermine provides significant inhibition of each of TRPV1, 3, and 4, with maximal block of 20–80% at membrane potentials around 0 mV. Additional, albeit weaker, contributions of other endogenous intracellular PAs, spermidine, and putrescine, typically present at concentrations similar to or higher than, spermine, will further add to the physiological action.

The behavior of each of TRPV1, 3, and 4 in the presence of spermine was similar to that previously reported for TRPC channels (Kim et al., 2016, 2020), which may imply a common (or a similar) mechanism throughout the TRP channel superfamily. CNG channels and ionotropic glutamate receptors are also inhibited by endogenous PAs with similar efficacy (Lu and Ding, 1999; Guo and Lu, 2000; Bowie and Mayer, 1995; Bähring et al., 1997; Bowie et al., 1998) and are also weakly permeable for PAs (Guo and Lu, 2000), as is TRPV1 (Ahern et al., 2006) and, probably, TRPV2 (Zubcevic et al., 2018b; Elbaz et al., 2016). Intracellular PA block of Kir channels (Lopatin et al., 1994; Lopatin et al., 1995) is typically much more potent, and there is much less evidence of permeation, potentially as a consequence of the much wider selectivity filter in TRPVs (Guo and Lu, 2003; Xu et al., 2009; Deng et al., 2018; Deng et al., 2020), although there is evidence for some degree of block relief at positive voltages in both Kir2 and Kir4 channels (Guo and Lu, 2003; Kucheryavykh et al., 2007; Nichols and Lee, 2018).

There is no obvious relief from block by the bulky analog NASPM at 100 µM and much less complete relief at 10 or 1 μM than is seen for spermine. This is consistent with relief from spermine block being the result of spermine permeation, with NASPM being more restricted in traversing the narrowest regions of the pore, but some relief is consistent with previous data indicating that TRPV1 and TRPV2 are capable of conducting relatively large organic cations (Ahern et al., 2006; Chung et al., 2008; Li et al., 2011; Puopolo et al., 2013; Munns et al., 2015; Elbaz et al., 2016; Zubcevic et al., 2018b), including spermine in the case of TRPV1 (Ahern et al., 2006). The weaker block relief in TRPV3 at high depolarizing potentials implies a lower permeability of TRPV3 to spermine compared to the other TRPVs. Together with variable steepness of the G_rel-V relationships of TRPVs in the blocking phase, this feature suggests there will be different specific contributions of each TRPV sub-type to action potential properties.

The overall similarity of the G_rel-V relationships in TRPVs (Fig. 2 D and Fig. 4) and CNG channels, as well as their similar dependence on PA length and charge (Lu and Ding, 1999; Guo and Lu, 2000), suggests a common mechanism of PA block and relief. A similar mechanism of interaction is also consistent with the demonstration that both CNG channels and, at least, TRPV1 channels are capable of conducting polyamine molecules (Ahern et al., 2006). This prompted us to consider the same kinetic model suggested earlier for CNG channels (Lu and Ding, 1999; Guo and Lu, 2000). With two blocking sites, from only one of which the blocker may continue to permeate, the model provides a reasonably accurate description of spermine block, relief, and permeation for each of TRPV1, 3, 4 channels. The apparent binding affinities for spermine are similar in each channel, there is a much slower apparent exit rate of spermine to the outside in TRPV3. The model however does not imply any specific locations for spermine binding within the TRPV pore. As recent molecular dynamics analyses of strong inward rectifier Kir2 channels have suggested (Chen et al., 2020; Jogini et al., 2023; Lee and Nichols, 2023), polyamines may occupy a diffuse space within channel pores. Available structures of TRPV1, 3, and 4 (Fig. 5 D) indicates multiple regions of electronegativity along TRPV pores. The overall higher electronegativity of the entire length of TRPV3 conductive pore, particularly near the lower gate (Fig. 5 D), correlates with the higher polyamine sensitivity of this channel, and may contribute to its lower polyamine permeation rate.

Crina M. Nimigean served as editor.

The work was funded by National Institutes of Health grants R35 HL140024 to C.G. Nichols and R01 NS099341 to P.Yuan.

The authors declare no competing financial interests.

Author contributions: G. Maksaev conceived the project; G. Maksaev, P. Yuan, and C.G. Nichols designed the studies; G. Maksaev performed the electrophysiological experiments; G. Maksaev and C.G. Nichols developed the kinetic models; G. Maksaev, P. Yuan, and C.G. Nichols prepared the manuscript.