Independent endothelial functions of PIEZO1 and TRPV4 in hepatic portal vein and predominance of PIEZO1 in mechanical and osmotic stress

Abstract

Background & aims: PIEZO1 and TRPV4 are mechanically and osmotically regulated calcium-permeable channels. The aim of this study was to determine the relevance and relationship of these channels in the contractile tone of the hepatic portal vein, which experiences mechanical and osmotic variations as it delivers blood to the liver from the intestines, gallbladder, pancreas and spleen.

Methods: Wall tension was measured in freshly dissected portal veins from adult male mice, which were genetically unmodified or modified for either a non-disruptive tag in native PIEZO1 or endothelial-specific PIEZO1 deletion. Pharmacological agents were used to activate or inhibit PIEZO1, TRPV4 and associated pathways, including Yoda1 and Yoda2 for PIEZO1 and GSK1016790A for TRPV4 agonism, respectively.

Results: PIEZO1 activation leads to nitric oxide synthase- and endothelium-dependent relaxation of the portal vein. TRPV4 activation causes contraction, which is also endothelium-dependent but independent of nitric oxide synthase. The TRPV4-mediated contraction is suppressed by inhibitors of phospholipase A₂ and cyclooxygenases and mimicked by prostaglandin E₂ , suggesting mediation by arachidonic acid metabolism. TRPV4 antagonism inhibits the effect of agonising TRPV4 but not PIEZO1. Increased wall stretch and hypo-osmolality inhibit TRPV4 responses while lacking effects on or amplifying PIEZO1 responses.

Conclusions: The portal vein contains independently functioning PIEZO1 channels and TRPV4 channels in the endothelium, the pharmacological activation of which leads to opposing effects of vessel relaxation (PIEZO1) and contraction (TRPV4). In mechanical and osmotic strain, the PIEZO1 mechanism dominates. Modulators of these channels could present important new opportunities for manipulating liver perfusion and regeneration in disease and surgical procedures.

Keywords: PIEZO channel; arachidonic acid metabolism; calcium signalling; calcium-permeable channel; endothelial nitric oxide  synthase; endothelium; hepatobiliary system; liver; mechanical force; nitric oxide; non-selective cation channel; osmolality; portal vein; transient receptor potential vanilloid 4 (TRPV4) channel; vascular relaxation; vasculature contraction; vein.

FIGURE 1

PIEZO1 agonism causes nitric oxide synthase (NOS)‐dependent relaxation. Control and gene‐modified mouse portal vein tension data (endothelium intact). (A) Example tension trace obtained from a control mouse contracted with 10 μM phenylephrine (PE) and then exposed to .1, .3, 1, 3 and 10 μM Yoda1 (PIEZO1 agonist) as indicated by the 5 dots. (B) As for (A) but from a PIEZO1^ΔEC mouse. (C) Summary data for Yoda1 responses of the types shown in (A, B) in n = 8 control mice (grey) and n = 9 PIEZO1^ΔEC mice (blue). (D) As for (A) but with a second set of concentration‐response data for Yoda1 in the presence of 100 μM L‐NAME. At the end of the recording, 10 μM SIN‐1 (a nitric oxide donor) was applied to show response to exogenous nitric oxide. Irregularities in the trace after the first Yoda1 applications occurred when the recording chamber was washed out 3 times (3× wash). (E) Summary data for n = 7 experiments (i.e., from 7 mice) of the type shown in (D) for the vehicle control (grey) or L‐NAME (blue). (C, E) Symbols and error bars are mean ± SD. The superimposed dotted lines are the underlying original data. Unpaired (C) and paired (D) t‐tests for PIEZO1^ΔEC compared with control mouse data at the indicated Yoda1 concentration: **p < .01, ***p < .001 and NS where there are no asterisks. n indicates the number of mice.

FIGURE 2

TRPV4 agonism causes contraction. Control mouse portal vein tension data (endothelium intact). (A) Example tension trace and mean summary data for contraction induced by 1 nM GSK1016790A (TRPV4 agonist) applied twice with wash‐out in between ((1) and (2)) (n = 5). (B) Example tension trace for contraction induced by increasing concentrations of GSK1016790A (.2, .4, .6, .8, 1 and 3 nM) as indicated by the 6 dots, with summary data to the right (n = 5). The smooth curve was fitted using the Hill Equation and indicated 50% maximum effect (EC₅₀) at .7 nM. (C) As for (A) but with the second GSK1016790A application in the presence of 300 nM GSK2193874 (TRPV4 antagonist) (n = 5 for the summary data). (D) As for (C) but with GSK1016790A applied after 10 μM phenylephrine (PE). (n = 5 for the summary data). (A–D) Symbols and error bars are mean ± SD. The superimposed dotted lines are the underlying original data. Paired t‐test: (A NS), (C ***p < .001), (D ***p < .001) and NS where there are no asterisks. n indicates the number of mice.

FIGURE 3

Endothelium‐dependence of the TRPV4 response. Control mouse portal vein tension data. (A) Typical trace with endothelial cells intact (+EC) showing responses to 10 μM phenylephrine (PE), .3, 1, 3 and 10 μM acetylcholine (ACh) and then, after wash out, 1 nM GSK1016790A. (B) Typical trace for a portal vein segment without endothelial cells (−EC). (C) As for (A, B), data for all experiments of this type. GSK1016790A contraction as a % of contraction evoked by 60 mM K⁺ plotted against ACh relaxation as a % of the maximum relaxation (pre‐PE tone). Each data point is for a segment of vein (n = 39). The straight line was fitted mathematically, indicating Pearson's correlation coefficient (r) .85. On the right, as for the left graph, but excluding the data in which there was no response to ACh (n = 21). r = .87. (D) For +EC, example trace (left) and summary data (right) for 1 nM GSK1016790A responses before and after incubation with 100 μM L‐NAME (n = 5). (E) As for (D) but using 500 nM apamin (Apa) and 100 nM charybdotoxin (ChTx) (n = 5). (F) As for (D, E) but using L‐NAME, Apa and ChTx (n = 8). Summary data are mean ± SD. Paired t‐test: (D–F) (NS). n indicates the number of mice.

FIGURE 4

COX and PLA₂ dependence of the TRPV4 response. Control mouse portal vein tension data (endothelium intact). (A) Example trace (left) and summary data (right) for 1 nM GSK1016790A responses before and after incubation with 1 μM SC‐560 (n = 5). (B) Similar to (A), but incubated with 1 μM SC‐560 and 10 μM celecoxib (n = 6). (C) Similar to (A), but incubated with 10 μM bromoenol lactone (n = 8). (D) Example trace (left) and summary data (right) for responses to increasing concentrations of prostaglandin E₂ (PGE₂) (.1, .3, 1, 3, 10 and 30 μM) applied twice and indicated by 6 dots each. Summary data are for n = 8 and mean ± SD. Individual data are superimposed. The two colours are for the first (1) and second (2) applications of PGE₂ with wash out in between. Paired t‐test: (A *p < .05), (B, C ***p < .001) and NS where there are no asterisks. n indicates the number of mice.

FIGURE 5

PIEZO1 responses do not involve TRPV4. Control mouse portal vein tension data (endothelium intact). (A) Example trace (left) and summary data (right) for responses to Yoda1 (.1, .3, 1, 3 and 10 μM, indicated by the 5 dots in the left trace) before and after incubation with 300 nM GSK2193874 (TRPV4 antagonist) (n = 5). (B) As for (A), but using Yoda2 instead of Yoda1 (.1, .3, 1, 3, 10 and 30 μM) (n = 5). (C) As for (A), but using 1 μM SC‐560 instead of GSK2193874 (n = 5). (A–C) Summary data are mean ± SD and the individual data are superimposed. Paired t‐tests: NS. n indicates the number of mice.

FIGURE 6

Mechanical and osmotic strain amplify PIEZO1 function. Control mouse portal vein tension data (endothelium intact). (A) Concentration‐response data for Yoda1‐induced relaxation in normal (open symbol, .8 mN tension and 282 mOsm·kg⁻¹), hyper‐stretch (light purple symbol, 2.2 mN tension and 282 mOsm·kg⁻¹), hypo‐tonicity (dark purple symbol, .8 mN tension and 255 mOsm·kg⁻¹) and combined hyper‐stretch and hypo‐tonicity (blue symbol, 2.2 mN tension and 255 mOsm·kg⁻¹) conditions (n = 5, 5, 5 and 6 respectively). The normal condition data are reproduced from Figure 5A and are shown only as mean values. (B) In the combined hyper‐stretch and hypo‐tonicity condition, example trace (left) and summary data (right) for responses to Yoda1 (.1, .3, 1, 3 and 10 μM, indicated by the 5 dots on the traces) before and after incubation with 300 nM GSK2193874 (TRPV4 antagonist) (n = 6). The Yoda1‐only data are reproduced from (a) and shown only as mean values. ANOVA (a) at the indicated Yoda1 concentration: **p < .01 and ***p < .001 for hyper‐stretch plus hypo‐tonicity compared with normal and NS, where there are no asterisks. n indicates the number of mice.

FIGURE 7

Mechanical and osmotic strain suppress TRPV4 function. Control mouse portal vein tension data (endothelium intact). (A) In the hyper‐stretch or hypo‐tonicity condition, example traces (left and middle) and summary data (right) for responses to 1 nM GSK1016790A (n = 8 each). The summary data are for the first (1) and second (2) GSK1016790A responses, with wash‐out in between. The normal condition data are reproduced from Figure 2A for direct comparison. (B–E) In the normal, hyper‐stretch or hypo‐tonicity condition, example traces and summary data for PE (10 μM) responses without (−) and with (+) 300 nM GSK2193874 (n = 8 normal, n = 5 hyper‐stretch, n = 5 hypo‐tonicity). Summary data are shown as mean ± SD. ANOVA: (A ***p < .001; hyper‐stretch or hypo‐tonicity compared with normal for (1) and (2)) and (E NS). n indicates the number of mice.