Missense variant analysis

Data were obtained from the gnomad database (https://gnomad.broadinstitute.org/) as csv files and processed by in-house MATLAB (Mathworks, MA) programs that are freely available through github (https://github.com/esenlab/variant-analysis) and described in our previous study [16]. Structure files were rendered for publication with PyMOL (Schrödinger, Germany).

Cell culture

HEK293T/17 (ATCC: CRL-11268) cells were incubated in Dulbecco’s Modified Eagle Medium, supplemented with 10% fetal bovine serum, 50 μg streptomycin, and 50 units/mL penicillin, at 37°C and 5% CO2. Cells were passaged onto poly-L-lysine treated 25 mm coverslips. Cells were allowed at least 2h to settle onto the slips before being transfected using Lipofectamine 2000 (Life Technologies) as described in the manufacturer’s instructions. Cells were transfected with pcDNA3.1 containing TRPV1 wild type and mutants. Following transfection, Ca²+ imaging experiments were carried out after 24h, and electrophysiology experiments were completed 24-48h after transfection. For electrophysiology experiments co-transfection of the TRPV1 construct of interest was done alongside a GFP expressing plasmid (JSM-164 NW10 3C-GFP-HS).

Molecular biology

In vivo assembly (IVA) methods of site-directed mutagenesis were done as previously described [16]. In the case of M308I, the construct was obtained from Azenta Life Sciences (South Plainfield, NJ) with our TRPV1.pcDNA3 provided as starting material. For all in-house mutagenesis, our wild-type TRPV1.pcDNA3 and TRPV1-C157A.pcDNA3 vector (Gift from A. Jara-Oseguera) was used as a template with (IVA) overlapping primers that contain the desired mutation (see primers for mutagenesis spreadsheet). The full sequence of each completed construct was confirmed using Sanger sequencing.

Electrophysiology

Currents were recorded at room temperature in symmetric divalent-free solutions (130 mM NaCl, 3 mM HEPES, 0.2 mM EDTA, pH 7.2) using fire polished borosilicate glass pipettes with filament (outer diameter 1.5 mm, inner diameter 0.86 mm; Sutter, MA). Pipettes were heat polished to a resistance of 3.5–7 MΩ using a micro forge. Cells were transfected with channel constructs and replated after 18–24 hours on 12 mm coverslips and placed in divalent-free solution in the chamber. 1 µM and 5 µM capsaicin solutions were prepared from a 5.3 mM capsaicin stock in ethanol. The same stock was diluted in the recording solution to 0.5 mM to make 0.03 µM, 0.1 µM, and 0.3 µM capsaicin solutions. Excised inside-out patches were positioned at the opening of a tube in a “sewer pipe” configuration with 8 tubes connected to a rod controlled by an RSC-200 rapid solution changer (BioLogic, France). Each perfusion barrel was briefly primed by applying pressure through the syringe reservoir before the next barrel was positioned in front of a patch. Capsaicin solutions (0.03, 0.1, 0.3, 1, 5 and 10 µM) were perfused for 40-120s each. For some experiments, the same solutions were perfused into the chamber using open, gravity-driven reservoirs for 100-120s each. TRPV1-M308H pH dependent experiments were conducted using divalent-free HEPES buffer as described above with pH adjusted as indicated in experiments (nominal pH: 7.2; high pH: 8.0; low pH: 6.5). Currents were measured with an Axopatch-200A e-phys amplifier at sweep intervals of 1 second stepping the voltage from resting potential (0mV) to −80 mV (400 ms), then up to +80 mV (400 ms) and back to resting potential.

Currents were filtered at 5 kHz and recorded at 20kHz using an Axopatch 200A amplifier (Axon Instruments, Inc.) and PatchMaster software (HEKA). Data was analyzed in PatchMaster and IGOR (Wavemetrics, OR). Individual patch-clamp recordings with a complete set of stable current measurements in all capsaicin solutions were normalized to the maximal current with 5 µM capsaicin. Normalized data was fitted to the Hill equation to arrive at the EC₅₀ parameter: I = I_max ([ligand]ⁿ/(EC₅₀ⁿ + [ligand]ⁿ)), where n was permitted to float. Statistical significance was determined by two-tailed Student’s t-test. For pH dependent measurements in macroscopic TRPV1 wild type excised patches the approximate open probability was calculated by normalizing the current under each pH condition with the maximal response measured with 5 μM capsaicin. Statistical evaluation of the change in currents due to shifting pH conditions was done with a one-sample t-test applied to the difference in currents for individual patches.

For single channel data sets, the + 80mV segment of the sweep was concatenated in MATLAB (Mathworks, MA). The current baseline was subtracted off using the xarray-graph plug-in with Matlab (https://github.com/marcel-goldschen-ohm/xarray-graph) and channel open probability was determined by partitioning the distribution of events by a cutoff that was half the current for the single channel open state. A notebook in Python was implemented for this purpose. In some instances, the MATLAB plug-in smBEVO was used to correct single channel current drift and evaluate state changes in the current profile. This plug-in is available from the Goldschen-Ohm lab through github (https://github.com/marcel-goldschen-ohm/smBEVO). As described, in the accompanying readme.txt file of the smBEVO plug-in, we optimized the filtering parameters of smBEVO by visual assessment with each data set until the modeled fit approximated behavior for the observed number of single channels.

Temperature-activation experiments were carried out in the whole-cell configuration of the patch clamp using a custom-modified deltaT temperature-control system (Bioptechs, Butler PA). The bottom of the recording chamber in this system is coated with an electrically conductive material that allows to pass current through to cause rapid and homogeneous changes in temperature on the opposite side of the chamber where the cells are located. Our system was modified to achieve more rapid temperature-changes. Temperature can also be cooled below ambient by flowing ice water through a closed metal loop that fits within the chamber. We recorded the temperature close to the pipette tip using a small thermistor bead. For these experiments we used pipettes with resistances between 0.5–1 MΩ and the same solutions as for inside-out recordings. After achieving the whole cell configuration, we lifted cells up and started flowing water through the cooling ring to drop the temperature to 10–15°C. Temperature was then increased up to 60°C and currents were recorded until the patch broke. Currents were recorded in response to pulses to ±100 mV with a duration of 400 ms from a holding potential of 0 mV. Pulses were applied with a frequency of 1 Hz. Data was acquired at 5 kHz and low pass filtered at 1 kHz, using a dPatch amplifier (Sutter, Novato CA). The steady-state current magnitude at the end of each pulse to +100 mV was used to quantify current-temperature relations for each experiment.

Calcium imaging

Cells were grown on 25-mm coverslips and then incubated for 30 min at room temperature with Fluo-4 (AM; Thermofisher) at a concentration of 3 μM. The cells were then rinsed with Hepes buffered Ringer’s (HBR) solution (in mM, 140 NaCl, 4 KCl, 1.5 MgCl2, 5 D-glucose, 10 HEPES, and 1.8 CaCl2 and pH adjusted to 7.4 with NaOH) and allowed to rest for another 30 min in HBR at room temperature. The cells were imaged on a Nikon Eclipse Ti microscope using a 10 × objective. For each slip, a brightfield image, a fluo-4 image, and a 3-min fluo-4 fluorescence movie with exposures of 100-ms and 0.5-s intervals were captured. During the time sequence, HBR is initially perfused throughout the chamber. Perfusion is switched to 500 nM capsaicin in HBR at 30 s, and at the 2 min mark, 3 μM ionomycin is added to the chamber to equilibrate calcium between inside and outside the cell in order to obtain the maximal Fluo-4 response. The HBR and 500 nM capsaicin in HBR were perfused into the chamber using open, gravity-driven reservoirs, and 500 uL of 3 μM ionomycin was Pipetted into the chamber via micro-Pipette. The data obtained during these experiments were analyzed using Image J and Matlab (Mathworks, MA).

Cell response counts in TRPV1 truncation constructs

Truncated constructs (TRPV1-preDelM308 and TRPV1–5’sv) were co-expressed with TRPV1 wild type constructs as 25% to 75% plasmid DNA weight ratio: TRPV1 wild type (25%) and PLCδ1-PH-SNAPtag (75% DNA; a plasmid membrane marker for PIP2); TRPV1 wild type (25%) and TRPV1-preDelM308 (75%); TRPV1 wild type (25%) and TRPV1–5’sv (75%). Before counting, the brightness and contrast were adjusted using FIJI’s automatic adjustment. Contrast and brightness adjustment was made with the “auto” feature in contrast and brightness control window at the point in the movie after the capsaicin response was induced, but before administration of ionomycin (around frame 200). This was done to make cells visible and distinguishable. A screenshot was then taken of the region with this brightness/contrast setting. Cells were marked with a pink line on this screenshot as they were counted to avoid any repeat counts or missed cells. Cell brightness was characterized from the beginning of the movie to the end of the movie (after ionomycin addition) for this region. If any cell did not increase in brightness due to either capsaicin or ionomycin addition (this includes cells which began and remained bright), it was crossed out in white and subtracted from the total cell count. Lastly, the cells with positive capsaicin responses were identified. This was done by comparing the brightness of cells in the first frame of the movie and the frame used to establish brightness/contrast settings. Cells with a qualitative change in brightness between these two points were marked as capsaicin responses. The fraction of cells responding to capsaicin was determined by dividing the number of capsaicin-responding cells by the number of total healthy (ionomycin or capsaicin-responding) cells. This process was repeated by two different people to check for consistency in the counting process. To ensure that data was not affected by variability across transfection days, we normalized the ratio of responses in experimental sets (TRPV1 WT with TRPV1-preDelM308 and TRPV1 WT with TRPV1–5’sv) by same day controls (TRPV1 WT with PLCδ1-PH-SNAPtag). Therefore, plotted values in Fig 4D are compared to a maximal TRPV1/PLCδ1-PHh response of 1.0.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Functional characterization of TRPV1-M308 with larger and smaller side chain substitutions by capsaicin dose-response and temperature activation experiments.

(A) Representative traces of capsaicin dose response relations in TRPV1-M308 mutants. Solution exchanges with different concentrations of capsaicin given in μM above traces. (B) Summary data of capsaicin dose response experiments shown as log-linear plots and fitted with a Hill function to acquire EC₅₀ values. The TRPV1-M308I mutant average responses is significantly different from wild type TRPV1 at two capsaicin concentrations, 0.1 and 0.3 μM (p = 0.0117, 0.0124). (C) Whole-cell current-temperature relations at +100 mV showing median temperature dependent activation of TRPV1-M308I (purple), TRPV1-M308V (red) and wild-type TRPV1 (grey) expressed in HEK293T/17 cells. Individual traces were normalized to the mean steady-state current magnitude measured at room temperature. Data points are shown at 2°C intervals and error bars are S.E.M. (TRPV1 WT, n = 11; TRPV1-M308V, n = 11; TRPV1-M308I, n = 5). Significant differences between wild type and TRPV1-M308I are indicated at 37, 39, 41, 43, 45, 47, 49 and 45°C (*; p = 0.033, 0.018, 0.009, 0.006, 0.003, 0.003, 0.006) and TRPV1-M308V at 35°C (*; p = 0.026. Student’s t-test).

https://doi.org/10.1371/journal.pone.0331224.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. TRPV1-M308H modulation by intracellular pH.

(A) Wild type TRPV1 macro patch recording shows a large change in current response with intracellular solution exchange to low (6.5) or high pH (8.0) relative to nominal pH (7.2) while capsaicin (Cap) is maintained at 0.1 μM. (B) Single channel recording of TRPV1-M308H in excised patch showing changes in current with simultaneous switching to different capsaicin concentration or pH of solution applied to inside out patch. Zoomed in regions of the current record show single channel opening and closing events (i.-iii.). There is a notable reduction in channel unitary conductance at lower pH and leak current is also noticeably increased at pH 8.0. (C) Single channel recording of TRPV1-M308H with greatest observed effect on P_o upon changing pH of solution applied to inside out patch. (D) Summary data of TRPV1 wild type (black) and M308H (purple) currents in response to 0.1 μM capsaicin and varying inner leaflet pH. Vertical diamond markers indicate macro patches and horizontal diamond markers indicate single channel recordings. Open probability calculated as normalized value to maximal current response with 5 μM capsaicin in TRPV1 wild type, macro patches and as fraction of current data points in “open” state relative to all data points for each condition in single channel, M308H patches – see methods. Thick lines are color coordinated with individual traces shown in A-C. A significant difference between TRPV1 wild type and M308H current responses is observed in repeated measures when switching 0.1 μM capsaicin containing solutions to pH 6.5 from pH 7.2 or from pH 6.5 to pH 8.0 (p = 0.095 in both with Mann-Whitney U-test).

https://doi.org/10.1371/journal.pone.0331224.g003

Live cell staining of TRPV1 surface expression

HEK293T cells were transfected with 50:50 DNA ratios of TRPV1 WT:TRPV1-exCellHalo in a control sample and TRPV1WT:TRPV1Δ1–308-exCellHalo in a test sample. TRPV1Δ1–308-exCellHalo was also transfected in isolation. Cells were stained after 24 hours with the Promega dyes Alexa488-HaloTag and JF549HaloTag, which are respectively cell impermeable and cell permeable. Coverslips were incubated in complete media with 1 μM of each dye type for 30 minutes at 37 degrees C before the media was exchanged and the cells recovered at room temperature for 10 minutes, followed by immediate imaging.

A valine at position 308 in TRPV1 is a frequent missense allele in the human population

Our previous work highlighted a positive correlation between solvent accessible surface area (SASA) of the amino acids and missense variants within the ankyrin repeat domain (ARD) that are present in the human population [16]. When ranked by their fractional SASA scores, the amino acids of the ARD show an increasing trend in the types of amino acids coded by missense variant sequences found in the human population (yellow bar plot, Fig 1A and S1 Fig). This variant analysis of TRPV1 revealed a region between ankyrin repeats 4 and 5, containing a methionine (M308) and an alanine (A256), that is crowded with variant amino acids (see colored side chains of Fig 1B) and contrasts sharply with the reduced number of variants observed in the neighboring fold between ankyrin repeats 5–6 (see grey side chains of Fig 1B). M308 is represented by a black marker (Fig 1A indicated with arrow), and the four variant types associated with that position set it apart as an outlier (missense variants from gnomAD release 4.1.0; Val:121, Thr:8, Ile:2, Arg:2). Nevertheless, we found that M308 is highly conserved in TRPV1 orthologs and across the TRPV sub-family (Fig 1E). The observed missense variants for M308 in the human population appear to be sharply defined by the size of the side chain (Met: 105.1 ml/mole) [19]. A slightly larger residue (107.5 ml/mole) such as isoleucine, is far less common than the valine (91.3 ml/mole) (i.e., there are 121 valine alleles and only 2 isoleucine). To investigate this in greater detail, we compared the missense allele counts for the M308 position in TRPV1 against all 16 methionines present in the ankyrin repeat domain of TRPA1 (Fig 1C). The missense variants in the TRPA1 methionines distribute across hydrophobic amino acids relatively evenly, suggesting that the prevalence of valine for TRPV1 M308 is functionally meaningful. We therefore hypothesized that a functional effect is observed when M308 is exchanged for amino acids larger than valine.

Capsaicin sensitivity of M308 mutants points to side chain constraints in TRPV1 ARD

To test our theory that substitution of the M308 fold is functionally tolerated up to a specified side-chain volume, we replaced the methionine with Ala, Val, Ile, and Trp and experimentally determined the capsaicin dose-response relation, and the associated EC₅₀, for each mutant (Fig 2A,B).Mutating M308 to tryptophan resulted in a non-functional channel. Alanine and valine substitutions could not be distinguished from wild type in their capsaicin sensitivity measured by patch-clamp while the isoleucine substitution at M308 incurred a significant increase in sensitivity to capsaicin. Therefore, our findings suggest that the fold between ankyrin repeats 4–5 imposes a unique space constraint on the side chain at position 308. Channels with an Ala, Val, or Met side-chain at position 308 behave just like WT, but insertion of hydrophobic residues Ile and Trp that are larger than methionine [19] have a strong effect on TRPV1 channel function, respectively increasing the channel’s sensitivity to stimuli or completely compromising function.

Temperature dependent activation of M308 mutations is consistent with their capsaicin sensitivity

We tested the temperature activation profiles of TRPV1 wild type, TRPV1-M308V and TRPV1-M308I to examine whether the isoleucine substitution also increased sensitivity to heat. Although there was considerable variability in the data, we observed a clear trend where M308I currents increased more steeply than those of WT or M308V channels (Fig 2C). These observations provide further support to the hypothesis that there are steric constraints for the side-chain at position 308, and that even a modest increase in side-chain volume has the functional effect of increasing channel activity.

Protonation of M308H modulates TRPV1 function

We observed that substitutions at M308 with side chains smaller than methionine retained WT channel function, whereas substitution with the slightly larger, and branched isoleucine increased channel activity in response to capsaicin, which leads us to hypothesize that covalent modification of the M308 side-chain could result in channel activation or sensitization. We first tested whether covalent modification of a cysteine introduced at position 308 in the C157A background construct of TRPV1, which is insensitive to MTS reagent modification [6], had any effect on channel function. Although the TRPV1-C157A/M308C construct expressed as a functional channel with a dose response profile to 0.3 and 5 μM capsaicin that was indistinguishable from the C157A background, we did not detect any effects on the currents measured upon exposure to MTSEA (S2 Fig) or MTSET (S3 Fig). This observation suggests that M308C either reacts with the reagents but this does not alter channel function or the site is not accessible to these reagents, the latter being consistent with the low SASA for this position measured from structures. We reasoned that whereas inaccessible to large cysteine-reactive molecules, the side-chain at position 308 might be accessible to protons, and its protonation might be sufficient to induce a change in channel activity. We therefore proceeded to mutate the M308 to a histidine, which is protonatable across a physiological pH range. It was previously shown that in addition to a strong sensitivity to extracellular protons [20], the TRPV1 channel exhibits increased activity with elevated intracellular pH [21]. Consistent with these findings, we show that WT TRPV1 expressed in HEK293T inside-out patches and activated with sub-saturating capsaicin has slightly reduced or unchanged current at pH 6.5 but increased currents at pH 8 (Fig 3A). The TRPV1-M308H channel is functional, but it suffers from inconsistent and poor expression in HEK293T cells. We did not detect any grossly altered capsaicin sensitivity for the M308H mutant evaluated at pH 7.2 (S4 Fig). Therefore, nearly all measurements of channel activity for this mutant are single-channel recordings (expanded views of the single channel behavior [i.,ii.,iii.] are shown below the full extent of the recording in Fig 3B). In contrast with WT channels, we observed that the open probability of M308H channels increased at pH 6.5, which promotes histidine protonation, and decreased at pH 8, which promotes deprotonation (Fig 3B,C,D). Although most of the data collected for M308H was of single channels, we collected one complete recording of a multi-channel patch expressing M308H, which had a pH-dependent response profile consistent with what we observed in single channel patches (S5 Fig). This trend was apparent in all patches, even though some single M308H channels exhibited increased activity at all pH values relative to WT channels. Overall, these findings suggest that protonation/deprotonation of the histidine side chain at position 308H results in detectable changes in channel activity that are absent in WT channels. These observations establish that steric or electrostatic perturbations at position M308 may influence channel activity, but that a methionine is not required for normal channel function, raising the question of why M308 is so highly conserved.

TRPV1 splice variant with start codon at M308 has a dominant-negative effect on full-length TRPV1 activity

The existence of mRNA for a splice variant of TRPV1 with a reading frame starting at position M308 has been described and found to have a dominant-negative effect on the magnitude of currents measured from cells where this shorter variant is co-expressed with WT. [18] However, it is unclear if the effect is due to a reduction in channel expression or channel function. A physiologically relevant role for this shorter splice variant could provide an explanation for why the methionine at position 308 is so highly conserved across species even though other amino acids at this position (i.e., Ala, Val, Cys – see S3 Fig) result in a channel with no or subtle changes to its function under the experimental conditions that we explored. To address this question, we first assessed whether the shorter isoform of TRPV1 gave rise to functional channels using Ca²+-imaging (Fig 4B). We focused on two different constructs, one that exactly recapitulates the splice variant that was previously identified (TRPV1–5’sv, Fig 4A and C, structure with missing segments shown in purple), and another where only the first 307 residues of the N-terminus are deleted, which we term TRPV1-preDelM308 (Fig 4A). We conducted baseline Ca²+ imaging experiments with only the truncated constructs to confirm previously published results that these constructs are non-functional. [18] To test the dominant negative potential of each truncated construct, we counted the number of capsaicin responding HEK293T cells that co-express the indicated combination of TRPV1 wild type and TRPV1-preDelM308 or TRPV1–5’sv relative to a global calcium response from ionomycin and calculated response ratios to TRPV1 wild-type co-expressed with a bystander construct (pcDNA3.1 with PLCδ1-PH-SNAPtag) to balance DNA amounts in the transfection step. We found that the addition of TRPV1–5’sv to TRPV1 wild type in the transfection step significantly reduced the number of cells that responded to capsaicin (Fig 4D). The TRPV1-preDelM308 isoform added to TRPV1 wild type also had a reduced number of cells responding to capsaicin compared to WT alone but did not surpass significance at the p = 0.05 level at the given DNA ratio.

TRPV1 splice variant with start codon at M308 is not detected on cell surface

From our previous experiment that examined the activity of TRPV1 co-expressed with shortened isoforms, the question remained whether a reduction in TRPV1 activity was caused by a loss in surface expression or if a heteromeric channel was expressed with altered function. We next determined the cellular expression pattern of a shortened isoform in the absence and presence of co-expressed full-length TRPV1. To accomplish this we utilized a fluorescent reporter construct for TRPV1 surface expression in which a self-labeling HaloTag domain was fused into the extracellular face of the pore of the channel, so that surface-expressed channels can be specifically labeled using a HALO-reactive membrane-impermeable fluorescent probe [22]. A labeling strategy with cell permeable and impermeable HaloTag ligands could therefore distinguish between channels reaching the cell surface and channels retained in intracellular compartments. This labeling strategy with HaloTag ligands showed that full-length TRPV1-exCellHalo is present at the plasma membrane and in cellular compartments, as would be expected for the cellular expression trafficking of the full-length channel (Fig 4E). By co-expressing our HALO-tag enabled shortened TRPV1 isoform (TRPV1Δ1–308-exCellHalo) with and without TRPV1 wild type we could discern if the truncated construct’s dominant negative effect arises through a heteromeric complex at the plasma membrane or if it is linked to upstream steps in channel expression. The TRPV1Δ1–308-exCellHalo accumulates in cell compartments but does not accumulate at the cell surface even when co-expressed with TRPV1 wild-type construct (Fig 4E). Together with the TRPV1 Ca²+ imaging data to establish the dominant-negative effect of isoform co-expression on activity, the extracellular label findings implicate a mechanism by which the shorter isoforms suppress the cell surface expression of full-length TRPV1 homomeric channel complex.