Supplementary AppendixThis appendix has been provided by the authors to give readers additional information about their work.Supplement to: Chesler AT, Szczot M, Bharucha-Goebel D, et al. The role of PIEZO2 in human mechanosensation.N Engl J Med 2016;375:1355-64. DOI: 10.1056/NEJMoa1602812

Supplementary appendix for: The role of PIEZO2 in humanmechanosensationAuthors: Alexander T. Chesler 1†, Marcin Szczot1*, Diana Bharucha-Goebel2,3*, Marta Čeko1*,Sandra Donkervoort2*, Claire Laubacher1, Leslie H Hayes2, Katharine Alter4, Cris Zampieri4,Chistopher Stanley4, A. Micheil Innes5, Jean K. Mah6, Carla M. Grosmann7, Nathaniel Bradley2,David Nguyen2, A. Reghan Foley2, Claire E. Le Pichon2, Carsten G. Bönnemann2†.

Table of Contents:Expanded MethodsFig S1: Compound heterozygous mutations in PIEZO2 as confirmed by conventional sequencingFig S2: Human PIEZO2 is highly similar at the amino acid level to the mouse orthologFig S3: Homologous mutations inactivate mouse Piezo2Fig S4: Antibody detection of heterologously expressed epitope-tagged mPiezo2 variantsFig S5: Detection of PIEZO2 variants by RT-PCR from skin biopsiesFig S6: Functional imaging reveals PIEZO2-independent detection of slow brushing of hairy skin.Fig S7: ROIs and fMRI images from control subjectsFig S8: PIEZO2 is required for proprioception in humansFig S9: Additional representative healthy control reaching dataFig S10: Erratic speed of reaching in blindfolded PIEZO2-deficient subjectsTable S1: Further Clinical Characteristics of PS-1 and PS-2Table S2: Haplotype data for PS-1 and PS-2Table S3: PCR primers usedLegends for Supplementary MoviesPage 2Page 10Page 11Page 12Page 13Page 14Page 16Page 18Page 19Page 20Page 22Page 24Page 25Page 26Page 27

Expanded MethodsPatient recruitmentThis study was approved by the Institutional Review Board of the National Institute ofNeurological Disorders and Stroke (NINDS) and National Institutes of Health (NIH). Writteninformed consent was obtained by a qualified medical investigator. DNA samples, obtainedfrom blood and skin biopsies, were collected based on standard procedures. Medical historywas obtained and clinical evaluations were performed as part of the standard neurologicevaluation.Exome sequencingWhole exome sequencing (WES) on DNA from whole blood samples obtained from the patientswas performed at the NIH Intramural Sequencing Center (NISC), using a Nimblegen SeqCap EZExome UTR Library version 3.0 and Illumina HiSeq 2500 sequencing instruments. Exome datawas analyzed using Varsifter and Seqr. Variants present in dbSNP, NHLBI EVS, ExAC wereexcluded, and the remainder were ranked based on potential to be damaging (data available uponrequest). The mutations were then confirmed by Sanger sequencing (NM 022068.3). Thegenomic regions were amplified by PCR and Sanger sequenced in forward and reverse directionsusing an ABI 3730xl DNA Analyzer for capillary electrophoresis and fluorescent dye terminatordetection.RNA isolation from skin, RT-PCR, and TA cloningFlash frozen skin tissue punches were lysed in Qiazol using a Qiagen Tissue Lyser (3 x 3 [email protected] 30 Hz). Total RNA was prepared using miRNeasy Mini Kit (Qiagen Cat#217004) accordingto the manufacturer’s instructions. 0.3-0.5 µg RNA was used as input for each RT reaction, using

SuperScript First-Strand Synthesis System for RT-PCR (Invitrogen Cat#11904-018) and 1 µl ofrandom hexamer primers per reaction. PCR sequences can be found in Table S2. PCR was doneusing Phusion High Fidelity PCR Kit (NEB Cat#E0553L). No other detectable bands wereamplified. PCR products were either purified for direct sequencing with nested primers, or theywere blunt end ligated and subcloned using TOPO blunt end PCR product cloning kit(ThermoFisher Cat#451245) and One Shot TOP10 competent E. Coli cells (ThermoFisherCat#C4040-10). Colonies were grown at 42ºC for up to 24 hrs, plasmids were amplified using aQiaprep Spin Miniprep kit (Qiagen Cat# 27104), and sequenced using standard procedures.Whole cell recordingHuman Embryonic Kidney (HEK) cells were cultured in glutamine-supplemented DMEM media(Lonza, Walkersville, MD, USA) with 10% fetal bovine serum (Gibco, Gaithersburg, MD,USA), 100 U/ml penicillin and 100 U/ml streptomycin (Gibco) in an incubator at 37 C in 5%CO2 atmosphere. The day before transfection, cells were replated on 24-well plates. Cells weretransfected at 70-90% confluency with 1000 ng with Piezo2 (full or truncated version) and 1 ngof EGFP (or just EGFP for sham recordings), encoding plasmids on a pCDNA3.1 backboneusing lipofectamine 2000 (ThermoFisher, Waltham, MA, USA). The next day, cells werereplated on coverslips that were covered with 0.1 mg/ml laminin (Sigma-Aldrich, St. Louis, MO,USA).All recordings were performed 48-72 hours after transfection. Transiently transfected cells wereidentified by EGFP fluorescence. Whole cell recordings were performed in voltage clamp mode,using Multiclamp 700B amplifier (Molecular Devices, Sunnyvale, CA, USA), at a holdingpotential of -40 mV. Signals were digitized with Digidata 1550 acquisition system (Molecular

Devices), and recorded on a PC computer running Clampex 10.3 (Molecular Devices).Extracellular solution consisted of (in mM): 133 NaCl, 3 KCl, 1 MgCl2, 10 HEPES, 2.5 CaCl2,10 Glucose, and 18.9 Sucrose, pH was adjusted to 7.3 with NaOH. Patch pipettes (resistance of2-4 MΩ) were filled with intracellular recording solution consisting of 133 CsCl, 10 HEPES, 5EGTA, 1 CaCl2, 1 MgCl2, 4Mg-ATP, 0.4 Na-GTP, and 10 Cs-gluconate; pH was adjusted to 7.3with CsOH. All chemicals were from Sigma-Aldrich. Cells exhibiting a stationary leak currentabove 200 pA were excluded from the analysis.Mechanical stimulation of the cells was performed by heat polished blunt pipette with a tipdiameter of 3-5 μm. Pipette was mounted on a P841.20 piezoelectric translator (PhysikInstrumente, Karlsruhe, Germany) that was positioned at the angle of 60 relative to the cellcontaining coverslip. The pipette was placed directly next to the target cell and positioned totouch the cell membrane without causing any visible indentation. Then a series of incrementalindentation stimuli was applied (typically 1-10 μm, with 2 μm increments), and the whole cellcurrent was recorded during this stimulation. To avoid accumulation of channels in theinactivated state, subsequent stimuli were separated by 10 s intervals. Responses were analyzedusing Clampfit software (Molecular Devices).Cell stainingHEK293 cells were transiently transfected and cultured on coverslips as above. The cellsurface of live cells was labeled by incubating in WGA-Alexa594 (Molecular Probes) for 5mins followed by three 5-minute washes in PBS prior to fixation with 4% paraformaldehydeand subsequent permeablization with 0.01% Triton X. Tagged mPiezo2 was then detected with

a mouse anti-HA primary antibody (1:100, Cell Signaling) followed by a donkey anti-mouse488 secondary (1:750, Molecular Probes).Sensory testingVibration task: A series of 10 sinusoidal vibration stimuli (150 Hz, 2 s) was appliedperpendicular to the skin surface, once to the left volar forearm, and once to the glabrous aspectof the left index finger using an in-house mechanical vibration device with a 2 mm probe in apaired (A-B) on-off pseudorandom order. Participants were asked to report whether they felt thevibration during A or B (forced-choice). To prevent auditory cues from the device, participantswore noise-cancelling headphones. To measure vibration probe displacement, a lightaccelerometer sensor ( 1 g, ADXL335, Sparkfun, USA) was attached to the vibrating probe.Acquired data were filtered with a digital lowpass filter, with a 3 dB cutoff at 250 Hz. Measuredtotal amplitude of the displacement vector 17.3 µm.Von Frey touch sensitivity: Sensitivity to punctate touch was tested on the left volar forearmusing von Frey monofilaments (Touch-Test Sensory Evaluators, North Coast Medical, GilroyCA) that exert forces between 0.008-300 g upon bending (equivalent to monofilament size 1.656.65). Using the method of limits, five threshold determinations were made, with a series ofascending and descending stimulus intensities. The final touch detection threshold was thegeometric mean of these five series.

Pinprick pain: Punctate pain threshold was tested on the left volar forearm using von Freymonofilaments (Touch-Test Sensory Evaluators, North Coast Medical, Gilroy CA). Using themethod of limits, five threshold determinations were made, with a series of ascending anddescending stimulus intensities. The final pain threshold was the geometric mean of these fiveseries.Pressure pain threshold: Using a calibrated, hand-held, algometer, with a 1 cm diameter tip(Force Dial FDK/FDN Series Push Pull Force Gauge, Wagner Instruments, USA), pressure wasapplied to participants’ left thumbnail. The pressure pain threshold was determined in three trialsof increasing pressure intensity (rate 0.5 kg/cm2s, inter-stimulus interval (ISI) 10 s).Two-point discrimination: A series of stimuli was applied to the left palm using a two-pointdiscriminator device (three-point aesthesiometer, Lafayette Instrument, IN, USA), with eitherone or two tips of the device (separated by 10-13 mm) touching the skin in a pseudo-randomizedorder. Participants were asked to determine if one or two ends were touching the skin (forcedchoice).Thermal detection and pain thresholds: A Peltier-based thermode (30x30 mm; Medoc AdvancedMedical Systems, Ramat Yishai, Israel) was applied in full contact to the left volar forearm toassess thermal sensitivity. Stimulus temperature was decreased and increased, respectively, at arate of 1 C/s from a baseline temperature of 32 C, and the participants were asked to verballyindicate the slightest sensation of cooling (cool detection threshold), warming (warm detectionthreshold) and at the shift from hot to painfully hot (heat pain threshold). Three assessments were

performed for each of the three stimulus types (ISI 4-10 s for detection thresholds, 30 s forpain threshold).Proprioception: A hand-held goniometer was used to assess the smallest detectable verticalangular motion of proximal and distal joints (toe, ankle, knee, finger, wrist, elbow, shoulder).The experimenter performed vertical angular 10 movements of the joints guided by thegoniometer, and with joints fixated laterally (to avoid proprioceptive cues). Participants wererequired to report if the felt vertical motion was upwards or downwards (forced-choice). Theprocedure was repeated on each joint in 10 decrements until participants’ response was atchance.Brushing task: Participants received two blocks of brushing: one on the left palm and the otheron the left volar forearm. A trained experimenter brushed each participant in the proximo-distaldirection across 6 cm of skin with a 3-inch diameter goat-hair watercolor brush. Audio cuesallowed the experimenter to brush at a constant velocity. Each block contained ten slowbrushing trials (3 cm/s). Each trial consisted of 6 s of brushing (ISI 8-10 s to prevent significantCT fiber fatigue 27. Detection of brushing stimuli was assessed in a forced-choice manner.fMRI acquisitionExperiments were limited to consenting adults (PS-1, and age/gender matched controls).Participants completed an fMRI session, during which two runs of brushing stimuli (forearm,palm), and two runs of vibration stimuli (forearm, palm) were applied. Each stimulus blockcontained ten slow brushing trials (3 cm/s). Each trial consisted of 6 s of brushing (interstimulus

interval was 8-10 s to prevent significant fiber fatigue 27). Detection of brushing stimuli wasassessed in a forced-choice manner. Throughout the session, participants wore earplugs and theirheads were immobilized. Images were acquired on a Siemens Skyra 3T scanner (Siemens,Erlangen, Germany) with a 20-channel head and neck coil. A whole-brain T1-weightedanatomical scan was acquired with an MPRAGE sequence (TR 1900 ms; TE 2.07 ms; FOV 256 mm; image matrix 256 x 256; number of slices 192; voxel size 1 x 1 x 1 mm).Whole-brain functional images were acquired using a blood oxygenation level-dependent(BOLD) protocol with a T2*-weighted echo planar imaging (EPI) sequence (TR 2000 ms; TE 29 ms; flip angle 70 ; FOV 224 mm; image matrix 64 x 64; number of slices 38; voxelsize 3.5 x 3.5 x 3.5 mm; total volumes 210 after discarding the first three volumes to allow forsteady-state magnetization).fMRI preprocessing and analysisfMRI data were preprocessed and analyzed in FSL 28. Briefly, preprocessing included slicetiming correction, six-parameter rigid body correction for head motion, alignment to the middlevolume of each scan, co-registration to the skull-stripped T1-weighted anatomical image usingboundary-based registration, spatial normalization to the MNI space using a 12-parameter affineregistration, intensity normalization, and spatial smoothing with a 5 mm Gaussian kernel.GLM analysis: For each subject, a voxel-wise general linear model (GLM) was conducted usingFSL’s FEAT (FMRI Expert Analysis Tool). Slow and fast brushing blocks were modeled using aboxcar function convolved with the canonical hemodynamic response (double-Gamma) function.The six motion parameters were included as regressors of no interest. Activation outside the

brain was masked, and a 100 s high pass filter was applied to the fMRI time-series. For directedsearches, a region of interest (ROI) was drawn by hand in FSL for S1 and S2 guided by theHarvard-Oxford cortical atlas and the group BOLD response to hand brushing. The posteriorinsula ROI was created directly from the Harvard-Oxford cortical atlas. The anterior insula ROIwas created from patient’s activation to slow back of hand brushing. Whole-brain activationscontrasting brushing rest were considered significant at z 2.3 (p 0.01) voxel-wise threshold,cluster-corrected at p 0.05 using random-field theory (RFT). For directed searches, peak zvalues were extracted from each ROI for each subject, and values of z 2.3 (p 0.01)considered significant activation.Kinematic Reaching taskIn the kinematic reaching task, subjects were directed to alternate between touching the target(placed at a 50 cm distance from their sternum) and their nose using index finger at a selfselected pace. The 3D motion-capture system MX run with Nexus software (Vicon MotionSystems, Denver, CO) was used to quantify the finger-to-nose test. Data were collected at 100Hz. Passive reflective markers were placed on the hands and index finger as well as a rod thatwas used as the target. Three trials of left and right hand finger-to-nose movement were collectedwith three repetitions (nose-target-nose) in each trial. Trials were performed with eyes opened orwith a sleeping mask over the face to prevent visual cueing. In some rare instances ( 1% of timepoints collected) the sensor position couldn’t be read due to steric hindrance by other body parts,those data points were therefore omitted in the analysis. Spatiotemporal data were analyzed usingcustom MATLAB R2013A (MathWorks, Cambridge, MA) scripts. To calculate nose position,

Cartesian coordinates of the finger at time points when it was with contact with it were linearlyinterpolated between each other to span the entire recording session.To better understand and visualize target contact, the coordinate system was transformed so thatthe target position was equivalent to (0,0,0), and the y versor was directed at the nose position atany given time moment. In this setting we calculated the finger-to-target distance:𝑟(𝑡) 𝑥 𝑡' 𝑦(𝑡)' 𝑧(𝑡)'To calculate the absolute movement velocity, the following formula was used:𝑣(𝑡) Δ𝑥Δt' Δ𝑦Δt' ΔzΔt'The skilled phase was defined to start when r(t) decreased below 15 cm, and to end when itincreased again above 15 cm. Distance covered during target approach phase was calculated as asum of differential distances travelled in single phase:1234 𝑥 ' 𝑦 ' 𝑧 '𝑑 15167897The ballistic phase was defined as movement when distance both from the target and nose wasgreater than 15 cm, and distance travelled during the entire reach phase was calculated with thesame formula. Incomplete phases (i.e trial start with finger located at the target or trial endwithout completing ongoing movement) were excluded from analysis.To establish the position of subjective target contact (Figure S10) we calculated thetarget distance at minimal speed during the skilled movement phase (speed drop 5cm/s, 15 cm of the target, separated by 50 ms).

Figure S1. Compound heterozygous mutations in PIEZO2, as confirmed by conventionalsequencing for PS-1 (A-B) and PS-2 (C).

Figure S2. Human PIEZO2 is highly similar at the amino acid level to the mouse ortholog. (A)Human vs mouse protein alignment. (B-C) Local sequence alignment of human PIEZO2 andmouse PIEZO2 show the position and amino acid conservation of R1575 and R1685 (bold) (Band C respectively). Amino acids that differ between human variants and homologous mousesequences are shown in red. (D) The protein sequences that surround the conserved residuesare highly conserved. Red box is the location of R1575. Blue box is the location of R1685.

Figure S3. Homologous mutations inactivate mouse Piezo2. (A) A snake diagramillustrating the predicted membrane topology of PIEZO2 protein, and approximate location ofthe mutations within a single large intracellular loop. The loop occurs before the presumptivepore domain. (B) Representative whole cell patch clamp recordings from HEK293 cellsexpressing either full-length mouse PIEZO2 or versions of the molecules containingmutations at homologous sites (Fig. S2). Individual cells were mechanically stimulated withincreasing indentations (2 µm steps) with a piezo-driven glass probe.

Figure S4. Antibody detection of heterologously expressed epitope-tagged mPiezo2 variants.(A) Schematic depicting the location of the 3XHA tag to an intracellular loop. (B) Mechanicallyevoked currents (bottom traces) with a series of membrane indentations (top trace) from the3XHA tagged mPiezo2. (C, left column) Co-staining of cell surface (magenta) and HA (green)in HEK293 cells transiently transfected with HA-tagged mPiezo2 constructs (scale bar 50 µm).(C, middle columns) Higher magnification view of boxed areas on the left showing anti-HA(green), cell surface marker (magenta), and the merged image (scale bar 5 µm). (C, rightcolumn) Close-up of boxed areas in merged images showing the overlap between both markers.

Figure S5. Detection of PIEZO2 variants by RT-PCR from skin biopsies. (A) Schematic ofthe PIEZO2 coding region with the relative positions and sizes of the two PCR fragments. (B)RT-PCR for the two fragments amplified from PS-1’s mother and PS-1 showing that the patienthas diminished PIEZO2 expression in skin (technical replicates and controls are found in Figure1). (C-F) Direct sequencing of an RT-PCR reaction from total RNA extracted from skinbiopsies. (C) Sequencing with nested primers of the sample of PS-2’s mother demonstrates thatwhile the predominant nucleotide at position 5053 corresponds to that found in the intact allele(c, blue trace), the missense allele is also detectable (t, red trace). (D) PS-2’s father carries boththe intact (5054C; blue trace) and missense mutation (5054G, black trace). (E-F) Sequencing oftwo biological replicates of samples for PS-2 showing that (E) the 5054C G allele is mostprevalent. (F) However, repeated sequencing revealed that the nonsense allele is also detectable.

Figure S6. Functional imaging reveals PIEZO2-independent detection of slow brushing of hairyskin. (A) During brushing on the forearm, healthy controls (top panel; n 7) showed significantfMRI activation (z 2.3 (p 0.01) voxel-wise threshold, cluster-corrected at p 0.05) in S1, S2 andthe posterior insula, while the patient did not (see Fig. S7B-C). However, the patient showed aspatially distinct response in the anterior insula not seen in controls (Fig. S7A). (B) A directedsearch of somatosensory areas (see Fig. S7A) reveals atypical activation of the anterior insula inthe Piezo2-deficient subject. (C) During brushing on the palm, healthy controls (top

panel; n 7) show a robust fMRI response in contralateral S1 and bilateral S2, while the patient(bottom panel) showed no significant response in a whole-brain analysis, nor (D) in a directedsearch of somatosensory areas. All activation maps represent slow brushing rest (z 2) and arerendered on the standard MNI brain. See Fig. S7C for individual control maps. x, y and z refer tothe MNI coordinates (mm) corresponding to the left–right, anterior–posterior and inferior–superior axes, respectively. In directed searches, peak z-values were extracted for each subjectfrom a-priori ROIs identified in previous studies of slow brushing (with the exception of theaINS, see Methods). Values of z 2.3 (p 0.01) were considered significant activation. S1 primary somatosensory cortex, S2 secondary somatosensory cortex, aINS anterior insularcortex, pINS posterior insular cortex.

Figure S7. ROIs and fMRI images from control subjects. (A) Anatomical locations and ROIsused for analyzing the fMRI data. (B) BOLD responses in PS-1 to brushing on forearm at MNIcoordinates 54, -18, and 12. (C) The individual BOLD responses for 6 subjects used for groupresponse in Fig. 3. Activation maps represent slow brushing rest (z 2), and are rendered on thestandard MNI brain.

Figure S8. PIEZO2 is required for proprioception in humans. Forced choice testing for limbposition. A given joint was moved 10º in isolation, and the subject asked to report the vectorof movement (up or down) while blindfolded. All control subjects (n 4; grey) answeredcorrectly on every trial while PIEZO2-deficient subjects (red) performed no better thanchance for every joint tested. Grey bars are healthy age and gender matched controls relativeto PS-1 (circles; n 4). Red bars are the two PIEZO2 subjects (PS-1, diamond and PS-2,hexagon). All trials: p 0.0001, Crawford test.

Figure S9. Additional representative healthy control reaching data (A-B) Individual data for 2additional control subjects for path length (right) and speed (left). Total reach ( 50 cm) was fromnose to object, (left to right). The skilled phase was defined as all movement within 15 cm oftarget. The task was performed with and without visual cues (compare grey vs green or red). Forall subjects there were 3 trials, n 11 reaches (single trails shown in graphs). (A-B top panels) 3Dreconstructions were used to create trajectory maps for the full reach (left) and skilled reach(middle); 2D projections are shown on three faces. Total path length for each skilled phase wasquantified and plotted (right) for eyes open (grey) and blindfolded (green). (A-B bottom panels)The speed of reaching is highly correlated with the distance from target, smoothly decreasingduring the approach to the goal whether or not visual cues are present (grey vs green).

Figure S10. Erratic speed of reaching in blindfolded PIEZO2-deficient subjects. (A-C) Left:Speed vs distance for the entire trajectory of the reaching task. As demonstrated by arepresentative control (A), the speed of reaching is highly correlated with the distance fromtarget, smoothly decreasing during the approach to the goal whether or not visual cues arepresent (grey vs green). PIEZO2-deficient subjects (B-C) have more variance in speed duringtarget approach, which increases upon visual deprivation (red vs grey). Middle: Speed (cm/s)plotted relative to distance from target (15 cm) for each trial. Right: Standard deviation of thetarget distance at subjective contact was plotted to compare subjects’ consistency in repeatedlyreaching the same position (cm).

Supplementary Table 1: Further Clinical Characteristics of Patient 1 and Patient 2Patient IdentifiersUrinary symptomsPulmonary functiontestingSensory nerveconduction study *Patient 1Patient 2Nocturnal enuresis (until 12 years)FVC: 2.24 L (72%ile) FEV1: 1.93L (69%ile)Sural nerve:Amp 6 μV; CV 47 m/s[Low-normal amplitude; normal CV]Urinary urgency (ongoing)FVC: 1.05 L (57%ile) FEV1: 0.90L (54%ile)Sural nerve:Technically limited due to prior footequinovarus.Median nerve:Amp 17μV; CV 52 m/s[Normal]Median nerve:Amp 18 μV; CV 43 m/s[Normal amplitude; borderline slow CV]Ulnar nerve:Amp 10 μV; CV 50 m/s[Low-normal amplitude; normal CV]Motor nerve*conductionstudyPeroneal nerve (at EDB):Amp 4.9 mV; Latency 3.7 msCV 52 m/s (Ankle-Fibular head)[Normal]Tibial nerve (at AH):Amp 3.7 mV; Latency 3.2 msCV 41 m/s[Normal]Median nerve (at APB):Amp 7.0 mV; Latency 2.9 msCV 58 m/s (Wrist-Elbow)[Normal]Median nerve (at APB):Amp 5.5 mV; Latency 3.3 msCV 50 m/s[Normal]Ulnar nerve (at ADM):Amp 8.6 mV; Latency 2.1 msCV 61 m/s (Wrist-Below elbow)[Normal]* Nerve conduction tests were performed during the diagnostic evaluation, and were performed at thecurrent age as listed in the table for each patient. μV microvolts; mV millivolts; ms milliseconds;m/s meters/secondAH abductor hallucis muscle; Amp amplitude; APB abductor pollicis brevis muscle; CV conduction velocity; EDB extensor digitorum brevis muscle; FEV1 forced expiratory volume in 1second; FVC forced vital capacity; L liters

Table S2: Haplotype data for PS-1 and PS-2


Supplementary Movie 1. Video of PS-1 demonstrating severe sensory ataxia in the absence ofvisual cues including inability to maintain standing position and a severe gait ataxia. Note thepatient was secured by harness.Supplementary Movie 2. Video of PS-1 demonstrating pseudoathetosis in the outstretchedarms, eyes open.Supplementary Movies 3-4. Videos of PS-1 and PS-2 performing the reaching task in theabsence of visual cues.Supplementary Movies 5-6. Video reconstructions of the movement of subjects’ hands andindex fingers during typical kinematic reaching trial. Reflective markers on the right hand arerepresented as tetrahedral apexes (orange), index finger marker position is additionallyvisualized with spherical marker. Crosses represent nose (red) and target position (green). Traildescribes pathway during preceding 1 second of the movement. All axis length equals 0.5 m.Each movie represents a single trial, first conducted with eyes open (EO) then blindfolded(EC). All animations are presented in slow motion (1/5 actual speed). Movie S5. Healthycontrol, Movie S6. PS-1.

(ThermoFisher Cat#451245) and One Shot TOP10 competent E. Coli cells (ThermoFisher Cat#C4040-10). Colonies were grown