NEUROSCIENCE

PIEZOs mediate neuronalsensing of blood pressure andthe baroreceptor reflexWei-Zheng Zeng1, Kara L. Marshall1, Soohong Min2, Ihab Daou1, MarkW. Chapleau3,4,Francois M. Abboud3, Stephen D. Liberles2, Ardem Patapoutian1*

Activation of stretch-sensitive baroreceptor neurons exerts acute control overheart rate and blood pressure. Although this homeostatic baroreflex has beendescribed for more than 80 years, the molecular identity of baroreceptormechanosensitivity remains unknown. We discovered that mechanically activatedion channels PIEZO1 and PIEZO2 are together required for baroreception. Geneticablation of both Piezo1 and Piezo2 in the nodose and petrosal sensory gangliaof mice abolished drug-induced baroreflex and aortic depressor nerve activity. Awake,behaving animals that lack Piezos had labile hypertension and increased bloodpressure variability, consistent with phenotypes in baroreceptor-denervated animalsand humans with baroreflex failure. Optogenetic activation of Piezo2-positivesensory afferents was sufficient to initiate baroreflex in mice. These findingssuggest that PIEZO1 and PIEZO2 are the long-sought baroreceptor mechanosensorscritical for acute blood pressure control.

Blood pressure (BP) is tightly regulated toensure that the body is prepared to meetvaried daily activity demands. Mechanismsthat change blood volume control long-term BP regulation. Within seconds and

minutes, BP regulation is initiated primarily bybaroreceptors, a class of stretch-sensitive neuronswithin the nodose and petrosal ganglia withperipheral projections in the walls of the aortaand carotid sinus (1, 2). An increase in BP stretchesbaroreceptor nerve endings to trigger afferentsignals that are transmitted to the central nervoussystem. The consequences of baroreceptor acti-vation are a decrease in heart rate (HR), cardiacoutput, and vascular resistance that counteract

the initial increase in BP (1, 2). Compromisedbaroreceptor function predicts arrhythmias andpremature death in humans with postmyocardialinfarction and heart failure (3, 4).Several ion channels (5–9) have been sug-

gested to contribute to baroreception. How-ever, substantial residual baroreflex is observedwhen these channels are ablated, implicatingthe involvement of other sensory systems. Noneof the candidate ion channels have been directlyactivated by mechanical stimuli in heterologoussystems, which may lack accessory tetheringmolecules to form a mechanosensory complex.Furthermore, whether these channels are actingas sensors or play a role downstream of mechano-

transduction is not clear. PIEZO1 and PIEZO2are mechanically activated ion channels that playcrucial roles in several mechanotransduction pro-cesses (10). PIEZO1 is prominently expressedin the cardiovascular system (11, 12), and PIEZO2is abundant in various populations of sensoryneurons (13–15).We assessed Piezo1 and Piezo2 transcript ex-

pression in nodose and petrosal ganglia, wherebaroreceptor cell bodies are located (1). Theseganglia are fused with each other and with thejugular ganglion in mice. Piezo1 and Piezo2 werehighly expressed in the nodose-petrosal-jugularganglion complex (NPJc) (Fig. 1A). Similar num-bers of cells were identified that highly expressedeither Piezo1 or Piezo2 exclusively (123 and 124cells with each transcript, respectively, Fig. 1B).A small population of neurons expressed both(43 double Piezo-positive cells, or 14.8% of allPiezo-expressing cells, n = 6 mice, Fig. 1B).To test whether Piezo1 and Piezo2 are ex-

pressed in baroreceptors, we performed retro-grade labeling of carotid sensory neurons. Weinjected fluorescent cholera toxin B (CTB) (16)into the carotid sinus beneath the serosal vesselcovering. All CTB-positive neurons detected inthe NPJc from eight mice were quantified forthe presence of Piezo1 or Piezo2 transcript (Fig. 1,C to F). Six out of 95 retrogradely labeled cellswere Piezo1-positive, and eight were Piezo2-positive (Fig. 1B). Piezo-negative cells were likelychemoreceptors, which abundantly innervatethe carotid sinus but do not require mecha-nosensitivity. None of the 95 CTB-labeled cells

RESEARCH

Zeng et al., Science 362, 464–467 (2018) 26 October 2018 1 of 4

1Howard Hughes Medical Institute, Neuroscience Department,Dorris Neuroscience Center, The Scripps Research Institute, LaJolla, CA 92037, USA. 2Howard Hughes Medical Institute,Department of Cell Biology, Harvard Medical School, Boston,MA 02115, USA. 3Abboud Cardiovascular Research Center,Department of Internal Medicine and Molecular Physiologyand Biophysics, Carver College of Medicine, University ofIowa, Iowa City, IA 52242, USA. 4Veterans Affairs MedicalCenter, Iowa City, IA 52242, USA.*Corresponding author. Email: ardem@scripps.edu

Fig. 1. Expression of Piezo1 and Piezo2 transcripts in NPJc. (A) Z-projectionof NPJc tissue after fluorescent in situ hybridization with probestargeting Piezo1 (red) and Piezo2 (cyan). Nuclei are labeled with 4′,6-diamidino-2-phenylindole (DAPI, blue). Arrows mark double Piezo-positivecells. (B) Quantification of transcript labeling area as a fraction of totalcell area (n = 290 cells, six mice). Each dot represents one cell. Numbers

in parentheses indicate number of cells. P1, Piezo1; P2, Piezo2. (C to F) NPJccell bodies back-labeled by carotid sinus CTB injections [(C) and (D),green] and Piezo transcript [(E) and (F), red and cyan]. Arrows indicatea Piezo2-positive cell in (C) and (E) and a Piezo1-positive cell in (D)and (F). (G) Quantification of Piezo transcript labeling area in CTB-positivecells (n = 95 cells, eight mice). Piezo-negative cells are not shown.

on June 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

were double Piezo-positive. These data suggestthat a subset of neurons that innervate thecarotid sinus (which include mechanoreceptorsand chemoreceptors) express either Piezo1 orPiezo2 (Fig. 1G). We hypothesized that thesecells could function as baroreceptors.We therefore crossed Piezo floxed mice to the

Phox2bCre line, which express Cre recombinasein epibranchial placode-derived ganglia (e.g.,nodose and petrosal) but not in neural crest–derived ganglia (jugular, trigeminal, and dorsalroot) (17). We first analyzed the baroreflex inanesthetized mice in response to phenylephrine(PE). PE induces rapid vasoconstriction (6), which

elevates BP. Increased BP then triggers barore-ceptor activity and induces a reflex decrease inHR. PE-induced baroreflex changes were com-pared in conditional double-knockoutmice (dKO;Phox2bCre+;Piezo1f/fPiezo2f/f) and Cre-negativewild-type littermates (WT). Infusion of PE intothe jugular vein produced a dose-dependent andtransient increase in systolic BP and a conse-quent decrease in HR, reflecting baroreflex con-trol (6) (Fig. 2A). The PE-induced HR reduction[−29 ± 20 versus −234 ± 24 beats per minute(bpm), P < 0.001] and decreased baroreflex sen-sitivity (−0.6 ± 0.4 versus −5.0 ± 0.5 Dbpm/DmmHg, P < 0.001) were essentially abolished

in the dKOmice (Fig. 2, A to D). PE-induced sys-tolic BP increase in dKO mice was significantlyhigher than in WT littermates (55.7 ± 3 versus45.7 ± 6 mmHg, P <0.05) (Fig. 2, A and B). HRresponse to sodium nitroprusside–induced acutebaroreceptor unloading was also absent in dKOmice (fig. S1, A to C). By contrast, Phox2bCre+;Piezo1f/f (P1cKO) and Phox2bCre+;Piezo2f/f (P2cKO)single-knockout mice showed no difference inPE-induced change of baroreflex compared withWT littermates (Fig. 2, B to D). We focused re-maining analyses primarily on dKO mice.Wenextmeasured aortic depressor nerve (ADN)

activity during a rise of BP induced by PE. We

Zeng et al., Science 362, 464–467 (2018) 26 October 2018 2 of 4

Fig. 2. Baroreflex is abolished in nodose and petrosal ganglia–specificdKO mice. (A) Cardiovascular recordings show PE-induced baroreflex inWTmice but no baroreflex in dKO littermates. BP, raw blood pressuresignal. SYS, systolic blood pressure derived from raw BP. (B to D) Changesin systolic BP (B), HR (C), and baroreflex (D) (10 s after intravenousinjection of PE) in knockout (KO) mice. Number of animals shownin bars (B) also apply for (C) and (D). Piezo1 KO indicates P1cKO mice;Piezo2 KO indicates P2cKO mice; Piezo1 Piezo2 KO indicates dKO mice.

All WT are littermates. (E) Traces show BP and ADN activity induced byPE and sodium nitroprusside injection in a WT and a dKO mouse. SNP,sodium nitroprusside. (F) Statistical analysis of drug-induced ADN activityin WT (n = 16) and dKO (n = 11) mice. (G) Raw BP and ADN activityexample before and after PE injection. Expanded time scale showed burstsof ADN activity in phase with individual arterial pulses in WT. No integratedactivity is observed in dKOmice. *P<0.05, ***P<0.001, and n.s. is statisticallynot significant by unpaired Student’s t test; data are means ± SEM.

RESEARCH | REPORTon June 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

observed a lack of drug-induced ADN activity indKO mice compared to WT mice (−131.9 ± 163.9versus 5558 ± 1234 normalized area under curveof integrated ADN activity, P < 0.001; Fig. 2, E toG). The dKOmice had no appreciable responsesduring both phasic and tonic phases of PE-induced ADN activity (fig. S1, D and E). This isnot due to gross anatomical deficits, because weobserved comparable baroreceptor ending den-sities within the aortic arch of dKO andWTmice(fig. S2).Impaired baroreceptor function leads to dys-

regulation of BP, including volatile hypertensionand increased BP variability in humans (18–20).We examined daily BP variability in freely mov-ing, conscious mice using a telemetric sensor (6).The dKO mice showed significantly increasedmean arterial pressure (MAP) during their ac-tive time (gray shading, 6 p.m. to 6 a.m.) com-pared with WT littermates (112 ± 0.4 versus 95 ±0.5 mmHg, P < 0.001) (Fig. 3, A and B, and figs.S3, A and B, and S4). The HR of dKO mice wasslightly increased during active times comparedwith that of WT mice (583 ± 3 versus 566 ±3 bpm, P < 0.001), whereas the HR remainedunchanged during inactive times (6 a.m. to 6 p.m.,532 ± 3 versus 536 ± 3 bpm, not significant)(Fig. 3B). No difference in locomotor activitywas observed between dKO and WT mice (fig.S3C), ruling out the possibility that activitycaused the increased BP and HR in dKO mice.We scanned telemetry data for spontaneous

changes in systolic BP and pulse interval (PI)consistent with a baroreflex relationship. Thismethod (sequence technique) is used to nonin-vasively assess baroreflex function (21, 22). Thespontaneous baroreflex sensitivity (sBRS) is de-fined as the slope of changes in systolic BP versusPI from 1 hour of recording. sBRS was severelyreduced in dKO mice (2.0 ± 0.1 versus 3.8 ±0.2 ms/mmHg forWT,P<0.001, Fig. 3C). Sinoaorticbaroreceptor denervatedmice also show residualsBRS (22), and this may be due to compensationfrom other sensory systems.We compared the BP variability of WT and

dKO mice. The systolic BP values of dKO micewere distributed in a broader range than those ofWT littermates (Fig. 3D). Variability was greatlyenhanced in dKO mice (7.9 ± 0.3 versus 6.1 ±0.3 mmHg in WT, P < 0.001, Fig. 3E). We quan-tified the range of BP variability of mean, sys-tolic, and diastolic BP within each group in a72-hour period. Maximum values of BP fromdKO mice were significantly higher than thosefrom WT littermates, whereas minimum valueswere significantly lower (Fig. 3F). Lastly, homo-vanillic acid concentrations in dKO mouse urinewere significantly higher than those inWT urine(13.9 ± 0.05 versus 12.1 ± 0.06 mg/ml, fig. S3D),suggesting an increase in hormone norepineph-rine concentration, as in human baroreflexfailure patients (18). There were no significantBP variability and sBRS differences in P1cKO

(fig. S5) and P2cKO (fig. S6) single-knockoutmice comparedwithWT littermates. P2cKOmiceshowed a subtle hypotensive BP distribution(fig. S6).

Zeng et al., Science 362, 464–467 (2018) 26 October 2018 3 of 4

Fig. 4. Piezo2-positive sensory neurons acutely control blood pressure. (A) Schematicdepiction of the optogenetic strategy. The carotid sinus and vagus nerves were illuminated to activateChR2-expressing Piezo2-sensory neurons. The optical fiber was placed on area 1, vagus nerve trunk;area 2, superior laryngeal branch; and area 3, carotid sinus. (B) Traces of cardiovascular effectsafter focal vagus nerve illumination (light blue shading) in anesthetized Piezo2Cre−;ChR2-eYFP(WT, black trace) and Piezo2Cre+;ChR2-eYFP mice (Piezo2Cre+, gray, blue, and pink traces).Numbers on the left (1 to 3) correspond to areas in (A). Blood pressure was measured by a pressuretransducer cannulated in the left carotid artery. BP, carotid arterial pressure. (C) Light-inducedchanges in BP and HR were calculated over the 10 s (n = 7 to 18 mice, as indicated; ***P < 0.001,and n.s. is statistically not significant by unpaired Student’s t test; data are means ± SEM).

Fig. 3. Increased BP variability in conscious nodose and petrosal ganglia–specific dKO mice.(A) Continuous measurements of MAP and HR over 72 hours, binned by hour. The differences betweengroups were significant during the night (gray shading, two-way analysis of variance, data are means ±SEM). (B) Average MAP and HR during the day (6 a.m. to 6 p.m.) and night (6 p.m. to 6 a.m.).(C) sBRS, expressed as change in PI (ms) per change in systolic BP (mmHg), was significantly reducedin dKO mice (n = 17 mice) compared with WTmice (n = 15 mice). (D) Frequency distribution histogramof the systolic BP over 72 hours. Red arrows indicate wider distribution of BP in dKO mice. (E) BPvariability reported as standard deviation from the 72-hour period. (F) Maximum and minimum BP values.P values are indicated in the bars. *P <0.05, **P <0.01, ***P < 0.001, and n.s. is statistically not significant.Unpaired Student’s t test was used unless indicated otherwise; data are means ± SEM. n = 7 to 17 mice.

RESEARCH | REPORTon June 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

We next investigated whether stimulatingPiezo2-positive neurons can induce the baroreflexin adult mice. We crossed Piezo2GFP-IRES-Cre(Piezo2Cre) knockin mice with Cre-dependentchannelrhodopsin-2 (ChR2) reporter mice togenerate Piezo2Cre+;ChR2-eYFP mice (13) andrecorded the cardiovascular response to activat-ing different regions of Piezo2-positive vagalsensory nerves by optogenetics (Fig. 4A; eYFP,enhanced yellow fluorescent protein gene). Wedid not observe cardiovascular changes duringlong optogenetic stimulation (5-ms pulses, 50 Hz,10 s) of the vagal nerve trunk (area 1 in Fig. 4A; BP,−5.3 ± 1.0%, and HR, −2.0 ± 1.0%; not significant)(Fig. 4, A to C). Next, we focused on specificallyactivating baroreceptor afferents. For aortic baro-receptors, we stimulated the superior laryngealnerve branch, which carries afferent inputs fromthe ADN (area 2 in Fig. 4A). For carotid baro-receptors, we exposed the carotid sinus regionand directly stimulated the local nerve terminals(area 3 in Fig. 4A). Light stimulations at bothlocations caused an immediate decrease in bothBP and HR (area 2: BP, −55.6 ± 2.0%, and HR,−50.5 ± 2.0%; area 3: BP, −37.5 ± 3.5%, and HR,−32.3 ± 3.7%; P <0.001) compared with the un-stimulated baseline (Fig. 4, B and C). A promi-nent consequence of baroreceptor activation israpid inhibition of efferent sympathetic activ-ity (1). We found that light-induced decrease inHR was markedly attenuated after administra-tion of the b-adrenergic receptor–blocker pro-pranolol, indicating that the reflex bradycardiawasmediated primarily by inhibition of cardiacsympathetic nerve activity (fig. S7). Piezo2Cre−;ChR2-eYFPmice (WT) did not show any changesduring optogenetic stimulation in all three re-gions (BP, −0.6 ± 0.6%, and HR, 0.4 ± 0.8%; notsignificant; Fig. 4, B and C).This study demonstrates that the mechani-

cally activated ion channels PIEZO1 and PIEZO2are together required for arterial baroreceptor activ-

ity and function. Baroreflex is critical to maintainshort-term BP homeostasis in mammals. Thelong-term changes observed in HR and BP thataccompany baroreflex failure are complex. Acuteelimination of baroreceptor function (e.g., sino-aortic denervation) causes immediate, large in-creases in BP and HR (23, 24). Over time, themean BP decreases but remains labile hyper-tensive, and BP variability is markedly increasedand persists (18–20, 24, 25). We observed a sig-nificant increase in MAP during the active periodof the Piezo dKO mice that falls just under thedesignation for hypertension (26), and dKOmice also developed increased blood pressurevariability. These data show that losing PIEZO1and PIEZO2 function recapitulates the pheno-type observed in animal models (24, 25) andhumans with baroreflex failure (18–20). However,we cannot exclude the possibility that sensorymechanisms beyond the baroreceptors withinthe vagus contribute to the observed increasedblood pressure.

REFERENCES AND NOTES

1. E. A. Wehrwein, M. J. Joyner, Handb. Clin. Neurol. 117, 89–102(2013).

2. H. R. Kirchheim, Physiol. Rev. 56, 100–177 (1976).3. A. Mortara et al., Circulation 96, 3450–3458 (1997).4. M. T. La Rovere, Lancet 351, 478–484 (1998).5. H. A. Drummond, M. P. Price, M. J. Welsh, F. M. Abboud,

Neuron 21, 1435–1441 (1998).6. Y. Lu et al., Neuron 64, 885–897 (2009).7. A. T. Chesler et al., N. Engl. J. Med. 375, 1355–1364

(2016).8. O. C. Lau, B. Shen, C. O. Wong, X. Yao, Nat. Commun. 9, 1244

(2018).9. P. Thakore, S. D. Brain, D. J. Beech, Nat. Commun. 9, 1245

(2018).10. S. E. Murthy, A. E. Dubin, A. Patapoutian, Nat. Rev. Mol.

Cell Biol. 18, 771–783 (2017).11. S. Wang et al., J. Clin. Invest. 126, 4527–4536 (2016).12. K. Retailleau et al., Cell Reports 13, 1161–1171 (2015).13. K. Nonomura et al., Nature 541, 176–181 (2017).14. S. H. Woo et al., Nat. Neurosci. 18, 1756–1762 (2015).15. S. S. Ranade et al., Nature 516, 121–125 (2014).16. K. L. Marshall et al., Cell Reports 11, 851–858 (2015).

17. A. Pattyn, X. Morin, H. Cremer, C. Goridis, J. F. Brunet,Development 124, 4065–4075 (1997).

18. D. Robertson et al., N. Engl. J. Med. 329, 1449–1455(1993).

19. K. Heusser, J. Tank, F. C. Luft, J. Jordan, Hypertension 45,834–839 (2005).

20. T. Ketch, I. Biaggioni, R. Robertson, D. Robertson, Circulation105, 2518–2523 (2002).

21. J. Oosting, H. A. Struijker-Boudier, B. J. Janssen, J. Hypertens.15, 391–399 (1997).

22. P. Martinka et al., Am. J. Physiol. Regul. Integr. Comp. Physiol.288, R767–R776 (2005).

23. I. C. Wenker et al., J. Neurosci. 37, 4565–4583 (2017).24. F. L. Rodrigues, M. de Oliveira, H. C. Salgado, R. Fazan Jr.,

Exp. Physiol. 96, 853–862 (2011).25. C. S. Ito, A. M. Scher, Circ. Res. 48, 576–591 (1981).26. P. A. James et al., JAMA 311, 507–520 (2014).

ACKNOWLEDGMENTS

We thank D. Morgan, S. Ma, and K. Nonomura for assistance andD. Ginty for the suggestion to assess the role of PIEZO2 inbaroreceptors. Funding: This work was supported by NIHgrants R01 DE022358 and R35 NS105067 to A.P. W.-Z.Z. wassupported by a postdoctoral fellowship from the GeorgeHewitt Foundation for Medical Research. S.D.L. was supportedby NIH grants DP1 AT009497 and OT2 OD023848. M.W.Cand F.M.A were supported by NIH grant P01 HL14388. A.P.and S.D.L. are investigators of the Howard Hughes MedicalInstitute. Author contributions: W.-Z.Z., K.L.M., and A.P.designed experiments and wrote the paper. K.L.M. performedin situ hybridization and baroreceptor innervation analysis.W.-Z.Z. performed drug-induced baroreflex assessment,telemetry sensor implantation, BP variability, and sBRS analysis.W.-Z.Z. and I.D. performed optogenetics experiments. S.M.performed ADN activity recordings in the S.D.L. laboratory.M.W.C and F.M.A advised and trained W.-Z.Z., contributedto technical approaches, and edited the manuscript.Competing interests: The authors declare no competinginterests. Data and materials availability: All dataare available in the main text or the supplementarymaterials.

SUPPLEMENTARY MATERIALS

www.sciencemag.org/content/362/6413/464/suppl/DC1Materials and MethodsSupplementary TextFigs. S1 to S7References (27, 28)

29 June 2018; accepted 7 September 201810.1126/science.aau6324

Zeng et al., Science 362, 464–467 (2018) 26 October 2018 4 of 4

RESEARCH | REPORTon June 28, 2020

http://science.sciencemag.org/

Dow

nloaded from

PIEZOs mediate neuronal sensing of blood pressure and the baroreceptor reflex

and Ardem PatapoutianWei-Zheng Zeng, Kara L. Marshall, Soohong Min, Ihab Daou, Mark W. Chapleau, Francois M. Abboud, Stephen D. Liberles

DOI: 10.1126/science.aau6324 (6413), 464-467.362Science 

, this issue p. 464; see also p. 398Scienceand blood pressure.failure. In mice, selective activation of PIEZO2-expressing ganglion neurons triggered immediate increases in heart ratepressure regulation and heart rates in mice. These changes were very similar to those seen in patients with baroreflex Conditional double knockout of PIEZO1 and PIEZO2 in these neurons abolished the baroreflex and disrupted bloodthe baroreflex, a homeostatic mechanism that helps to keep blood pressure stable (see the Perspective by Ehmke).

found that both ion channels are expressed in sensory neurons of a ganglion complex that contribute toet al.skin. Zeng PIEZO1 and PIEZO2 are two mechanically activated ion channels that are highly expressed in lungs, bladder, and

Heart rate and blood pressure control

ARTICLE TOOLS http://science.sciencemag.org/content/362/6413/464

MATERIALSSUPPLEMENTARY http://science.sciencemag.org/content/suppl/2018/10/24/362.6413.464.DC1

CONTENTRELATED

http://stm.sciencemag.org/content/scitransmed/10/431/eaap8674.fullhttp://stm.sciencemag.org/content/scitransmed/4/132/132ra54.fullhttp://science.sciencemag.org/content/sci/362/6413/398.full

REFERENCES

http://science.sciencemag.org/content/362/6413/464#BIBLThis article cites 28 articles, 5 of which you can access for free

PERMISSIONS http://www.sciencemag.org/help/reprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAAS.ScienceScience, 1200 New York Avenue NW, Washington, DC 20005. The title (print ISSN 0036-8075; online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science. No claim to original U.S. Government WorksCopyright © 2018 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of

on June 28, 2020

http://science.sciencemag.org/

Dow

nloaded from