Letter Cite This: Nano Lett. XXXX, XXX, XXX−XXX
pubs.acs.org/NanoLett
Ultrasonic Control of Neural Activity through Activation of the Mechanosensitive Channel MscL Jia Ye,† Siyang Tang,† Long Meng,‡ Xia Li,† Xiaoxu Wen,† Sihan Chen,† Lili Niu,‡ Xiangyao Li,† Weibao Qiu,‡ Hailan Hu,† Mizu Jiang,† Shiqiang Shang,† Qiang shu,† Hairong Zheng,*,‡ Shumin Duan,† and Yuezhou Li*,†
Downloaded via UNIV OF TOLEDO on June 19, 2018 at 21:15:22 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
†
Children’s Hospital and Department of Biophysics, NHC and CAMS Key Laboratory of Medical Neurobiology, Zhejiang University School of Medicine, Hangzhou, Zhejiang 310058, China ‡ Paul C. Lauterbur Research Center for Biomedical Imaging, Shenzhen Key Laboratory for MRI, Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences, Shenzhen, 518005, China S Supporting Information *
ABSTRACT: Externally controlling the excitation of a neuronal subset through ion channels activation can modulate the firing pattern of an entire neural circuit in vivo. As nanovalves in the cell membrane, ion channels can be opened by light (optogenetics) or ultrasonic (sonogenetics) means. A thoroughly analyzed force sensor is the Escherichia coli mechanosensitive channel of large conductance (MscL). Here we expressed MscL in rat hippocampal neurons in a primary culture and showed that it could be activated by low-pressure ultrasound pulses. The gain-of-function mutation, I92L, sensitized MscL’s sonic response, triggering action potentials at a peak negative pressure as low as 0.25 MPa. Further, the I92L MscL reliably elicited individual spikes by timed brief pulses, making excitation programmable. Because MscL opens to tension in the lipid bilayer, requiring no other proteins or ligands, it could be developed into a general noninvasive sonogenetic tool to manipulate the activities of neurons or other cells and potential nanodevices. KEYWORDS: Mechanosensitive channel, neuron, ultrasound, MscL, nanovalve cells.16 To further lay the foundation for such a “sonogenetics”, we transformed primary cultured neurons with MscL (mechanosensitive channel of large conductance) and directly assessed its excitability by ultrasound. The MscL from Escherichia coli is the most thoroughly analyzed mechanosensitive channel in terms of biophysics, genetics, and structures.17−19 It is a homopentamer of 136 amino acid subunits, each forming two transmembrane αhelices (TM1 and TM2) linked by a periplasmic loop.20 It has been purified and shown to retain its mechanosensitivity when reconstituted into a lipid bilayer thereby establishing the forcefrom-lipid principle.21−23 Stretch deforms the bilayer’s internal force profile causing helix movement to open a pore of ∼30 Å in diameter, allowing the passage of ions and small molecules.24−27 We chose to use the MscL for several advantages, besides the clarity of its background research. Its small gene size makes experimentation simple. The existing large collection of gain- or loss-of-function mutant MscL forms a ready-made array of force sensors with graded sensitivity filling different experimental needs.28−31 Chemically modified
I
on channels are pore-forming membrane proteins that allow ions to pass through the channel pore. Ion channels underlie the nerve impulse and act as a molecular switch for controlling the activity of targeted cells within intact neural circuits. To understand neuronal circuits and behavior, external means have been used to activate ion channels grafted into neurons.1−5 Making use of transgenic light-sensitive channels, optogenetics has been successful but is limited to transparent subjects. Ultrasound offers an alternative.6−9 It can easily be delivered through the skull, affecting but not damaging the brain.10−12 The mechanosensitive channel, which is capable of detecting and responding to mechanical stimuli, may be serve as a mechanosensitive nanovalve and confer neural response specifically to ultrasound stimulation. Recently, ultrasoundactivated two-pore-domain mechanosensitive K+ channels (TREK-1, TREK-2, TRAAK) have been reported in Xenopus oocytes.13 Ultrasound, amplified with microbubbles, is also shown to evoke behavioral responses in Caenorhabditis elegans with an expressed mechanosensitive TRP-4 channel.14 Deletion of the DEG/ENaC ion channel MEC-4 subunit in C. elegans abolishes its mechanical response under ultrasound stimuli.15 Ultrasound-induced transcriptional activities in the presence of microbubbles have been investigated in mechanically sensitive Piezo1 ion channel expressed mammalian © XXXX American Chemical Society
Received: March 7, 2018 Revised: June 5, 2018
A
DOI: 10.1021/acs.nanolett.8b00935 Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters
Figure 1. Functional expression of MscL channels in neurons. (A) Immunocytochemistry of cultured hippocampal neurons infected with a lentiviral vector (control), histidine-tagged (Histag) WT-MscL channels (WT-MscL), or I92L mutant channels (I92L). The scale bar is 20 μm. mCherry fluorescent protein was coexpressed in all infected neurons. The MscL protein was detected by a Histag antibody. (B) Representative currents recorded from excised inside-out patches from WT-MscL- or I92L-MscL-infected and control neurons in response to a 7 s ramp of gradually increasing the negative pressure from 0 to −120 mmHg (bottom trace) at Vm = −65 mV. (C) Gating thresholds for WT and I92L MscL channels (mean ± SEM, n = 23−29 patches from 4 cell cultures, ***, p < 0.001 by unpaired two-tailed Student’s t test). (D) Representative singlechannel activity of WT and I92L showing differences in fully open versus low-conducting substate currents. C, S, and O mark the current levels in the closed, substate, and fully open states, respectively. (E) Current−voltage curves for WT- and I92L-MscL single-channel activity (mean ± SEM, n = 20−25 patches from 4 cell cultures). Conductance was calculated from the slope of the linear regression fits. (F) Normalized current−pressure relationship of WT and I92L MscL channels recorded from inside-out patches at Vm = −65 mV (mean ± SEM, n = 15−18 patches from 4 cell cultures). Solid lines represent fits to a Boltzmann equation. (G) Representative currents recorded from inside-out patches of neurons expressing MscL in response to 500 ms pulses of successively increasing (10 mmHg) negative pressure steps (−100 to −140 mmHg for WT-MscL and −30 to −70 mmHg for I92L as indicated) at Vm = −65 mV. (H) Sample traces of I92L currents activated by repetitive (100 ms interval) −50 mmHg pressure pulses for various times as indicated.
negative pressures as low as 0.25 MPa, in the absence of microbubbles. We further demonstrated that the I92L mutant precisely elicited ultrasound-evoked spike trains with a millisecond temporal precision and fidelity up to 5 Hz. This work adds to the foundation of sonogenetics, suggesting that MscL-based tools can be used for the noninvasive ultrasonic control of neuronal activity, and potential nanodevices. Results and Discussion. Infection of primary cultured rat hippocampal CA1/CA3 neurons with a MscL-containing lentivirus led to the expression of MscL channels on a neuronal membrane, detected by immunostaining for the histidine-tagged MscL (Figure 1A). Single-channel recordings from excised inside-out patches from MscL-infected neurons, but not control neurons, displayed the large mechanosensitive
MscL can be triggered by light or pH, making it a versatile tool to receive additional stimuli if necessary.32−36 Its large pore can be used to pass small molecules besides ions and different filters can be fitted there in the future. It has been reported that the bacterial MscL can function in any membrane including mammalian CHO and HEK293 cells,37 and most recently, in cultured neurons,38 but will not interact with cellular ligands or proteins to complicate interpretation. The ultrasound-induced accumulation of impermeable dye was observed in MscLexpressed REP cells, which was dependent on the functional linkage of the microbubbles with an intact actin cytoskeleton.39 In this study, we have engineered a MscL channel into primary cultured neurons. The I92L mutation sensitizes the MscL’s sonic response and can be activated by ultrasound at peak B
DOI: 10.1021/acs.nanolett.8b00935 Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters
Figure 2. I92L mutant enables ultrasound-evoked neuron spiking. (A) Schematic illustration of the ultrasound device, which generated the acoustic pressure to stimulate neurons in the chamber during the electrophysiological recording or fluorescence measurement. (B) Representative inward currents in a voltage-clamped neuron expressing I92L evoked by 1 s of ultrasound stimulation (gray area) at Vm = −65 mV. The peak negative pressure applied is marked above each trace. Inset: enlargement of the square area of I92L current induced by 0.25 MPa (bottom trace) showing the multiple openings. C, S, and O = closed, substate, and fully open state, respectively. (C) Ultrasound-evoked I92L current is dependent on the peak negative pressure (mean ± SEM, n = 20−24 patches from 5 cell cultures). (D) Representative traces of calcein released from preloaded neurons though I92L channels activated by 20 repetitive 1 s ultrasound stimuli (0.45 MPa) at an interstimulus interval of 1 s (gray, once every 100 s). (E) Ultrasound-induced calcein release from I92L-infected but not WT-MscL-infected or control neurons (mean ± SEM, n = 23−25 cells from 3 cultures). (F) Ultrasound stimulation for 1 s (gray, peak negative pressure indicated above each trace) evoked spikes in current-clamped neurons expressing I92L MscL. (G) The numbers of ultrasound-evoked spikes in I92L-infected neurons are peak negative pressure-dependent (mean ± SEM, n = 25−28 patches from 5 cell cultures). (H) Voltage traces showing membrane depolarization in I92L-infected neurons evoked by a 1 s ultrasound stimulus (gray, 0.45 MPa) in the presence of TTX (1 μM), a blocker of voltage-gated Na+ channels. (I) Summary of ultrasound-induced membrane depolarization in I92L-infected, WT MscL-infected, and control neurons in the presence of TTX (mean ± SEM, n = 15−20 patches from 3 cell cultures).
with leucine may lead to increased mechanosensitivity through TM1−TM2 interaction. As expected, the I92L mutant channel showed a much lower activation threshold of −31 ± 10 mmHg (Figure 1C). Consistently, the leftward shift of the dose− response curve relative to that of the WT channel revealed a pressure for half-maximal activation of the I92L mutant (P50 = −60 ± 8 mmHg) lower than that of the WT (P50 = −150 ± 11 mmHg) (Figure 1F). Although the conductance was comparable in I92L (1.58 ± 0.21 nS) and the WT (1.90 ± 0.15 nS), the I92L mutant presented flicker activity and lowconducting substates, unlike the predominant full-opening occupancy in the WT (Figure 1D,E). Moreover, the I92L mutant responded to pulsed negative pressure stimuli as short as 25 ms, showing repeated activation at 100 ms intervals (Figure 1H). These results demonstrate that MscL channels can be functionally expressed in primary cultured neurons. The I92L mutant shows a high mechanical sensitivity and responds to subtle stretch stimuli.
currents in response to an increasing linear ramp of negative pressure (Figure 1B). The threshold of activation for the wildtype (WT) MscL channel was −97 ± 10 mmHg (Figure 1C), comparable to that previously determined in mammalian cells expressing MscL.37 Pulsed negative pressure activated MscL channels with a similar threshold, displayed rapid activation and a subsequent decrease to a steady-state current (Figure 1G). In contrast, negative pressure-induced currents were absent in control neurons (Figure 1B), which are consistent with the study by Soloperto et al.,38 indicating that the mechanical currents recorded from MscL-expressing neurons were due to the activity of the MscL channel rather than endogenous expression of other mechanically gated channels. In order to confer the extra mechanosensitivity to cultured neurons, we generated a sensitive variant of the MscL that isoleucine at position 92 was substituted with leucine (I92L). I92, located in the TM2 transmembrane α-helix, is a critical residue for the gating of the MscL.26 Substitution of isoleucine C
DOI: 10.1021/acs.nanolett.8b00935 Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters
Figure 3. Precision of spike trains elicited in I92L-expressing neurons by ultrasound pulses. (A) Representative current-clamp recordings from a neuron expressing I92L in response to a 0.36 MPa ultrasound stimuli of various durations (indicated in gray). (B) Summary of latency to the peak of the first spike evoked by ultrasound (mean ± SEM, n = 19 patches from 5 cell cultures). (C) Representative current-clamp recordings from a neuron expressing I92L in response to 100 ms ultrasound stimuli at various peak negative pressures (gray). (D) Latency to the peak of the first spike is dependent on a peak negative pressure (mean ± SEM, n = 21 patches from 5 cell cultures). (E) Representative voltage traces showing spikes in a current-clamped neuron expressing I92L evoked by 1 and 5 Hz trains of ultrasound pulses (gray dashes) at 0.36 MPa (50 ms duration). (F) Relative latency to spike peak as in part E (mean ± SEM, n = 9 patches from 3 cell cultures). (G) Summary of the spikes (out of 10 possible spikes) evoked in current-clamped neurons expressing I92L by 1, 5, and 10 Hz trains of ultrasound pulses at 0.36 and 0.45 MPa (50 ms duration) (mean ± SEM, n = 22 patches from 4 cell cultures).
MPa induced a further decrease of current to ∼4 pA/pF, providing an acoustic pressure-dependent response for the I92L mutant (Figure 2B,C). This response profile, including substate openings (Figure 2B inset), resembled that induced by negative pressure stimulation (Figure 1D,G). In contrast, ultrasound-evoked currents were not detected in the WT MscL-infected or control neurons following 1 s, 0.45 MPa stimulation, a result consistent with I92L mutant having a much lower activation threshold to mechanical force. Moreover, we have exposed fresh (∼1 day after hippocampal cells were cultured) and up to 5 days old hippocampal cells to the same ultrasound stimulus and we did not detect any neuronal response (Figure S2). From these results, other mechanosensitive background components in neurons could be ruled out and I92L would dominate the response to ultrasound in vivo. The open pore of the MscL channel allows the passage of molecules such as the fluorescent dye calcein in molecular weight ∼600 Da.33−36,40 We then measured the flux of calcein in live cells as an indication of MscL activation. Repetitive ultrasound stimulation (0.45 MPa, every other second, 20 times per 100 s) induced up to 20% calcein release within 10 min in the I92L-expressing neurons, whereas no calcein release
It is noteworthy that lentiviral expression of MscL did not alter the electrical properties or survival of infected neurons. The membrane resistance and resting potential of the neurons expressing WT or I92L MscL were similar to those of the control neurons (Figure S1A,B). The cellular viability were not altered by expressing WT or I92L MscL, as the percentage of MscL-expressing neurons that took up the NucGreen dead cell probe reagent was close to that of the controls (Figure S1C). These results are consistent with the recent report that the expression of the MscL channel has minimal side effects in infected neurons.38 A surface-acoustic-wave-based chip was designed to transmit standing ultrasound waves along the surface of a recording chamber to the neurons (Figure 2A). The peak negative pressure of the ultrasound applied to the neurons was adjustable in the range from 0.12 to 0.45 MPa. Whole-cell recordings from I92L-expressing neurons held at −65 mV in voltage-clamp mode showed that 1 s continuous ultrasound with a peak negative pressure of 0.45 MPa evoked large depolarizing currents reaching a peak value of 26 ± 2.1 pA/pF. Decreasing the peak negative pressure to 0.36 MPa reduced the inward current to 15 ± 2.8 pA/pF, and ultrasound at 0.25 D
DOI: 10.1021/acs.nanolett.8b00935 Nano Lett. XXXX, XXX, XXX−XXX
Letter
Nano Letters
the energy accumulated with time is required for the channel activation. In summary, we can now use ultrasound to open the MscL to control neuronal activities. The force-from-lipid principle, discovered from the very research of the MscL itself, means that it can be activated in any membrane independent of other proteins or ligands. Its single small gene can easily be targeted to a specific neuron subset in vivo. Further, we can tune its sensitivity based on its rather simple submolecular structure through mutations; e.g., here, the gain-of-function I92L excites hippocampal neurons at 0.25 MPa, which is much lower than the pressure (1 MPa) known to penetrate skull and brain tissue with very little impedance or tissue damage.14,41 Different or additional mutations can generate a graded panel of sensors, with members suited to different needs. For example, the frequency of ultrasound-evoked spikes may be improved by the molecular engineering of MscL variants with fast gating kinetics combined with strong currents and rapid inactivation. In addition, MscL modifications with designed ion selectivity and pore size may achieve a more accurate manipulation. Backed by its thoroughly understood biophysical mechanism and molecular structure at atomic resolution, the MscL adds to our arsenal to advance neuroscience. MscL-based sonogenetics may provide a feasible method for the noninvasive control of neuronal activity, with a versatility for applications ranging from neuroscience to biomedical engineering. Unlike ionspecific channels, MscL rapidly passes molecules smaller than ∼1000 in molecular weight.42−45 Doener et al.,37 for example, have used transgenic MscL to deliver phalloidin into mammalian cells. Thus, ultrasound should be able to deliver material into any cytoplasm with MscL installed in the membrane. Future will tell whether such drug delivery can become “sonotherapy”. However, whether MscL could be well tolerated in mammals in vivo would be investigated when the therapeutical applications are explored. Methods. Ultrasound Application. Surface-acoustic-wave (SAW)-based chips were used to generate ultrasound.46 The chips consisted of a pair of interdigital transducers (IDTs) patterned onto a piezoelectric substrate of 127.8° Y-rotated, Xpropagating single-crystal lithium niobate (LN). Aluminum electrodes (200 nm thick) were deposited on a 1 mm thick LN substrate. Vibration of the substrate was estimated to be
