Although detailed structural information on TRPV channels is available from X-ray and cryo-EM studies, in most cases the distal N-terminal region preceding the ARD escaped description at an atomic level, either because it was purposefully deleted or because it remained unresolved due to its inherent flexibility (Goretzki et al. 2021). With a spectroscopic approach, we recently showed that the N-terminal regions of human and chicken TRPV4 channels are almost completely disordered (Goretzki et al. 2018, 2022). Interestingly, this was somewhat contradicted by sequence-based disorder prediction web tools that indicated the presence of a significant amount of secondary structure propensity in the TRPV4 IDRs (Goretzki et al. 2022).
Here, to estimate the potential flexibility and disorder of the N-termini of the three remaining group I human TRPV channels and to compare this to NMR spectroscopic data, we used the sequence-based web server ODiNPred (Dass et al. 2020) (Fig. 1). ODiNPred predicts disorder probabilities larger than 0.5 for 86% of residues within TRPV1-IDR, 78% of residues within TRPV2-IDR and 91% of residues within TRPV3-IDR, which is indicative of highly disordered proteins. Nonetheless, on the C-terminal end of the TRPV1- and TRPV2-IDRs, lower disorder probabilities (< 0.5) are predicted for continuous stretches of amino acids in 14 of 100 and 16 of 72 residues, respectively. Such values are indicative of transient structural order within ensembles sampling the potential conformational space. In addition, in the N-terminal part of hsTRPV2-IDR, a second short region with low per-residue disorder probability is predicted. Notably, in hsTRPV3-IDR only 10 of 109 residues show a disorder probability lower than 0.5.
To experimentally characterize the N-terminal regions of the human TRPV1, TRPV2 and TRPV3 channels in solution, we used NMR spectroscopy. In agreement with a predicted low overall secondary structure content, the [1H, 15N]-TROSY-HSQC-spectra of human TRPV1-IDR, hsTRPV2 and TRPV3-IDR show limited signal dispersion in the 1HN dimension (Fig. 2 A-C), indicating a similar chemical environment of all 1HN nuclei and an inherent lack of stable structural elements. By using a set of two- and three-dimensional NMR experiments, the sequence specific resonance assignments for nearly all backbone 1H, 13C and 15N spins could be obtained. In summary, 91.3%, 92.6% and 91.1% of the HN, N´, C´, Cα, Cβ, Hα and Hβ resonances of hsTRPV1-, hsTRPV2- and hsTRPV3-IDR could be assigned, respectively.
All three proteins contain a large number of proline residues, i.e., 14 in TRPV1-IDR, 7 in TRPV2-IDR and 17 in TRPV3-IDR. These do not group in extended proline rich regions as in the N-terminus of TRPV4, but nonetheless frequently cluster in pairs of two or three. For the human TRPV1-IDR, the C´, Cα, Cβ resonances for 12 out of 14 proline residues could be assigned with mean Cβ values of 32.17 ± 0.18 ppm. This leaves only the consecutive proline residues in the triple proline repeat P33*-P34*-P35 unassigned (marked with *).
Likewise, backbone carbon chemical shifts for 6 out of 7 proline residues in hsTRPV2-IDR were assigned with mean Cβ values of 32.12 ± 0.06 ppm, leaving residue 37 in the P37*-P38 double proline motif unassigned.
For the hsTRPV3-IDR, the C´, Caα, Cβ as well as Cγ resonances for 16 of 17 proline residues could be assigned, leaving only P61 in the P61*-P62 motif without chemical shift information. All assigned hsTRPV3-IDR prolines show 13Cβ and 13Cγ values in the range of 32.17 ± 0.08 ppm and 27.47 ± 0.16 ppm, respectively. The mean difference of the proline 13Cβ and 13Cγ chemical shifts is 4.69 ± 0.20 ppm.
Based on the chemical shift values for Cβ and Cγ, it can be assumed that all assigned proline residues in hsTRPV1-IDR, hsTRPV2-IDR and hsTRPV3-IDR are in the trans configuration (Schubert et al. 2002; Shen and Bax 2010). It remains to be seen whether these residues can adopt a stable cis configuration in the presence of ligands, as was observed for TRPV4 (Goretzki et al. 2018).
The experimentally obtained chemical shifts of the distal N-termini of hsTRPV1, hsTRPV2 and hsTRPV3 were also used for an initial secondary chemical shift-based structural analysis. The POTENCI web server (Nielsen and Mulder 2018) was used for the prediction of random coil chemical shifts at our experimental conditions. The predicted chemical shifts were compared with those experimentally obtained to reveal potential regions of structural order. For all three TRPV channel constructs, the measured and predicted Cα, Cβ, C´, N´, HN, Hα, and Hβ chemical shift values agree remarkably well (Fig. 3 A-C, i-vii). The mean differences between the experimental and POTENCI-predicted random coil chemical shift values for hsTRPV1-IDR are: ΔCα = -0.08 ± 0.24 ppm, ΔCβ = -0.08 ± 0.27 ppm, ΔC´ = 0.00 ± 0.26 ppm, ΔN´ = -0.03 ± 0.64 ppm, ΔHN = 0.01 ± 0.10 ppm, ΔHα = 0.02 ± 0.05 ppm, and ΔHβ = 0.02 ± 0.03 ppm. Likewise, the mean differences for hsTRPV2-IDR are: ΔCα = -0.02 ± 0.17 ppm, ΔCβ = -0.09 ± 0.22 ppm, ΔC´ = 0.06 ± 0.18 ppm, ΔN´ = -0.03 ± 0.51 ppm, ΔHN = 0.01 ± 0.07 ppm, ΔHα = 0.02 ± 0.04 ppm, and ΔHβ = 0.02 ± 0.02 ppm. For hsTRPV3-IDR the mean differences between experimental and predicted chemical shifts are: ΔCα = 0.18 ± 0.36 ppm, ΔCβ = 0.06 ± 0.21 ppm, ΔC´ = 0.20 ± 0.38 ppm, ΔN´ = 0.00 ± 0.61 ppm, ΔHN = 0.06 ± 0.09 ppm, ΔHα = 0.03 ± 0.11 ppm, and ΔHβ = 0.04 ± 0.04 ppm.
The notion that the human TRPV1, TRPV2 and TRPV3 distal N-termini are highly disordered is further supported by the sequence-specific secondary structure propensity method (Marsh et al. 2006). As recommended for intrinsically disordered proteins, we used the SSP method to combine Cα, Cβ and Hα chemical shift values into single residue specific scores (Fig. 3 A-C, viii). In contrast to the analysis with the ODiNPred server, which indicated that hsTRPV1-IDR and hsTRPV2-IDR but not hsTRPV3-IDR contain extended regions able to form transient structures (Fig. 1), the SSP scores predict both the TRPV1 and TRPV2-IDR sequences to be highly disordered (mean SSP scores of 0.000 ± 0.097 and 0.015 ± 0.075 for hsTRPV1-IDR and hsTRPV2-IDR, respectively). Globally, the mean SSP score of the hsTRPV3-IDR is also close to 0 (-0.009 ± 0.156). However, a stretch of residues (P103-R113) in the C-terminal region of the hsTRPV3-IDR shows a significantly higher mean SSP score (0.355 ± 0.041), suggesting the potential formation of a helical structural element. The presence of an α-helical structure in this region is supported by recent cryo-EM structures of human TRPV3 (e.g. Zubcevic et al. 2019). Interestingly, AlphaFold (Jumper et al. 2021) also predicts a helical structure for this region.
By averaging the calculated SSP scores, an overall secondary structure content of only 7.7% for hsTRPV1-IDR, 6.0% for hsTRPV2-IDR and 12.0% for hsTRPV3-IDR can be estimated.
Although partial structural information on the N-terminal IDRs of group I TRPV channel members has been reported (e.g. Zubcevic et al. 2019; Pumroy et al. 2019; Nadezhdin et al. 2021), for the most part these regions are missing from structural studies. NMR spectroscopy is an optimal tool to investigate highly flexible and intrinsically disordered proteins. Together with our previous study on TRPV4 (Goretzki et al. 2022) we supplement the available structural information on group I TRPV channels with a detailed view on the extent of intrinsic disorder in their distal N-termini at atomic resolution. Interestingly, while there is some agreement with sequence-based structure predictions, the disorder determined in vitro tends to be more extensive than what is seen in silico. The inherent flexibility of the N-termini of group I TRPV channels is likely a conserved molecular feature underpinning their importance for channel regulation but making them challenging targets for structural studies.