Author response:
Reviewer #1 (Public review):
Chaiyasitdhi et al. set out to investigate the detailed ultrastructure of the scolopidia in the locust Müller's organ, the geometry of the forces delivered to these scolopidia during natural stimulation, and the direction of forces that are most effective at eliciting transduction currents. To study the ultrastructure, they used the FIB-SEM technique, to study the geometry of natural stimulation, they used OCT vibrometry and high-speed light microscopy, and to study transduction currents, they used patch clamp physiology.
Strengths:
I believe that the ultrastructural description of the locust scolopidium is excellent and the first of its kind in any insect system. In particular, the finding of the bend in the dendritic cilium and the position of the ciliary dilation are interesting, and it would be interesting to see whether these are common features within the huge diversity of insect chordotonal organs.
Thank you very much for your comments. We indeed plan to extend and continue our approach to exploit and understand diverse chordotonal organs in insects and crustaceans.
I believe the use of OCT to measure organ movements is a significant strength of this paper; however, using ex vivo preparations undermines any conclusions drawn about the system's in vivo mechanics.
Having re-read the manuscript, we failed to explicitly describe our ex vivo preparation of Müller’s organ including key references that detail the largely retained physiological function of Müller’s organ. We have now revised this detail in the method section:
“We used an excised locust ear preparation for all experiments, following a previously described dissection protocol [9]. In short, the tympanum, with Muller’s organ attached was left intact suspended between the cuticular rim. The cuticular rim of the tympanum was fixed into a hole in a preparation dish that allowed Muller’s organ to be submerged with extracellular saline, whilst the outside of the tympanum was dry and could be stimulated with airborne sound. This ex vivo preparation of Muller’s organ retained frequency tuning (Warren & Matheson, 2018), similar electrophysiological function as freshly dissected Muller’s organs (Hill, 1983a, 1983b; Michelsen, 1968: frequency discrimination in the locust ear by means of four groups of receptor cells), and amplitude coding (Warren & Matheson, 2018). Since Müller’s organ is backed by an air-filled trachea in vivo, the addition of saline solution in the ex vivo preparation decreased its displacements ~100 fold due to a dampening effect (Warren et al., 2020).”
And in the last section of the introduction:
“Here, we combined FIB-SEM to resolve the 3D ultrastructure of a scolopidium, OCT and high-speed microscopy to examine sound-evoked motion at both the organ and individual scolopidium levels, and direct mechanical stimulation of the scolopale cap, where the ciliary tip is anchored, whilst simultaneously recording transduction currents. Here, Muller’s organ and the tympanum was excised from the locust for physiological experiments. This ex vivo preparation of Muller’s organ retained frequency tuning, amplitude coding and electrophysiological function. This preparation also permitted the enzymatic isolation of individual scolopidia whilst recording transduction currents (Warren & Matheson, 2018).”
To further clarify physiological differences between the in vivo and ex vivo operation of the tympanum and Müller’s organ, we will perform an additional experiment for the revised manuscript by quantifying the changes in the sound-evoked tonotopic travelling wave of the tympanum using Laser Doppler Vibrometry (LDV). This result will be added to the Supplementary Text.
The choice of Group III scolopidia is also good. Research on the mechanics of locust tympana has shown that travelling waves are formed on the tympanum and waves of different frequencies show highest amplitudes at different positions on the tympanum, and therefore also on different groups of scolopidia within the Müller's organ (Windmill et al, 2005; 2008, and Malkin et al, 2013). The lowest frequency modal waves (F0) observed by Windmill et al 2008 were at about 4.4 kHz, which are slightly higher than the ~3 kHz frequencies studied in this paper but do show large deflections where these group III scolopidia attach at the styliform body (Windmill et al, 2005).
Thank you very much. We accept that the frequencies studied in this manuscript were lower than the lowest modal wave observed by Windmill et al., 2008. Other authors, according to Jacobs et al. 1999, found broad tuning form 3.4-3.74 kHz (Michelson et al., 1971) and 2-3.5 kHz (Halex et al., 1988). We settled on tuning previously measured for Group-III neurons in the same kind of preparation as in this manuscript, which was broadly around 3 kHz (Warren & Matheson, 2018).
This should be mentioned in the paper since the electrophysiology justification to use group III neurons is less convincing, given that Jacobs et al 1999 clearly point out that group III neurons are very variable and some of them are tuned much higher to 10 kHz, and others even higher to 20-30 kHz.
Looking at Fig. 7 from Jacobs et al., 1999, we indeed see that the four Group-III neurons recorded in this study are broadly tuned to 3-4 kHz. Often these tuning curves have threshold dips at higher frequencies at least 20 dB higher. We settled on the most sensitive frequency that we previously measured, and which also overlaps the most sensitive frequencies from several other studies.
Weaknesses:
Specifically, it is understandable that the authors decided to use excised ears for the light microscopy, where Müller's organ would not be accessible in situ. However, it is very likely that excision will change the system's mechanics, especially since any tension or support to Müller's organ will be ablated.
We completely understand this criticism. We have now added descriptions in the methodology and introduction (as detailed previously). In short, the tympanum was left intact suspended on the cuticle. Müller’s organ retains all (measured) physiological properties: frequency tuning, amplitude coding and electrophysiological function. To further investigate whether this excised preparation is a representative of the in vivo conditions, we plan to measure tympanal mechanics, such as the travelling wave, as part of the revisions.
OCT enables in vivo measurements in fully undissected systems (Mhatre et al, Biorxiv, 2021) or in systems with minimal dissection where the mechanics have not been compromised (Vavakou et al, 2021). The choice to entirely dissect out the membrane is difficult to understand here.
The pioneering OCT works by Mhatre et al, Biorxiv, 2021 and Vavakou et al, 2021 set the new standard of in vivo measurements in the field. We also totally agree with Reviewer#1’s view that OCT is best performed on in vivo Müller’s organ and we tried OCT imaging of Müller’s organ for several months in vivo. Although the OCT penetrates the tympanum the OCT beam does not penetrate the tracheal air sac that surrounds Müller’s organ and therefore OCT cannot be used in vivo. Please also see previous comment with regards to the intact physiological operation of Muller’s organ in the ex vivo preparation.
My main concern with this paper, however, is the use of light microscopy very close to the Nyquist limit to study scolopidial motion, and the fact that the OCT data contradict and do not match the light microscopy data. The light microscopy data is collected at ~8 kHz, and hence the Nyquist limit is ~4 kHz. It is possible to measure frequencies reliably this close to the limit, but the amplitude of motion is quite likely to be underestimated, given that the technique only provides 2 sample points per cycle at 4 kHz and approximately 2.66 sample points at 3 kHz. At that temporal resolution, the samples are much more likely to miss the peak of the wave than not, and therefore, amplitudes will be mis-estimated. A much more reasonable sample rate for amplitude estimation is generally about 10 samples per cycle. I do not believe the data from the microscopy is reliable for what the authors wish to use them for.
We understand your concern that the study of sound-evoked motion of the scolopidium using light microscopy was done near the Nyquist limit (with our average sampling rate at 8.6 ± 0.3 kHz and the Nyquist limit at 4.3 kHz). We also agree with your comment that amplitude of the motion could be underestimated at frequencies closer to the limit. However, we find that this systematic error does not change the key observation from our direct light microscopy observation that axial stretch of the scolopidium occurs around 3 kHz.
To address this concern, we plan to study the scolopidial motion within Group 1 auditory neurons, which are tuned to lower frequencies (0.5-1.5 kHz). This new set of data will allow us to obtain more data points per cycle (up to ~8.6 data points at 1 kHz). We will consider adding this result into the revised Fig. 4 or its extended data.
Regarding increasing the sampling rate, we did try to achieve higher sampling rate (> 10 kHz), however, there is a technical limitation of our camera and a trade-off between other key parameters, such as the size of the region of interest (ROI) and magnification. To increase the sampling rate, we will have to reduce the magnification or the ROI and in turn lose the spatial resolution required for quantification of the scolopidial motion or the ROI does not cover the whole scolopidial motion. The sampling rate at 8.6 ± 0.3 kHz was the best we could achieve.
Using the light microscopy data, the authors claim that the strains experienced by the group III scolopidia at 3 kHz are greater along the AP axis than the ML axis (Figure 4). However, this is contradicted by the OCT data, which show very low strain along the AP axis (black traces) at and around 3 kHz (Figure 3c and extended data Figure 2f) and show some movement along the ML axis (red traces, same figures). The phase at low amplitudes of motion cannot be considered very reliable either, and hence phase variations at these frequencies in the OCT cannot be considered reliable indicators of AP motion; hence, I'm unclear whether the vector difference in the OCT is a reliable indicator of movement.
This is our fault for not clearly explaining the orientation of the light microscopy measurement, which then leads to the reviewer’s concern about contradiction between OCT and light microscopy. Our OCT measurements was done along the Antero-Posterior (AP) and Mesio-Lateral axes (ML), while the axial stretch of the scolopidium occurs along the Dorso-Ventral (DV) axis. We recognise that the anatomical references in this manuscript can be confusing, and we tried to show the orientation of the scolopidium relative to Müller’s organ in Fig. 3b. To further clarify the orientation of our observations, we will add anatomical references in Fig. 4a and Fig. 5a. in the revised manuscript.
As stated in our result section (Line 165-167)
“Notably, we could not resolve the Group-III scolopidia along the ventro-dorsal axis—which runs parallel to the dendrite—as the OCT beam was obstructed by either the cuticle or the elevated process”
We did try to perform OCT measurement along the VD axis, but we could not resolve the scolopidial region along the scolopidial or ciliary axes because the OCT beam could not go through the thick cuticle at the edge of the tympanic membrane and the elevated process. For this reason, it is impossible for us to find an agreement or rule out any contradiction between the OCT and light microscopy since they are measuring motion along different axes. We plan to address this accessibility issue in a separate work using OCT measurements in combination with mirrors.
The OCT data are significantly more reliable as they are acquired at an appropriate sampling rate of 90 kHz. The authors do not mention what microphone they use to monitor or calibrate their sound field and phase measurements in OCT, but I presume this was done since it is the norm.
We use a condenser microphone (MK301, Microtech) and measuring amplifier (type 2610, Brüle & Kjær) for calibration. The calibration microphone was also calibrated beforehand using a sound calibrator type 4231 from B&K.
Thus, the OCT data show that the movement within the Müller's organ is complex, probably traces an ellipse at some frequencies as observed in bushcrickets (Vavkou et al, 2021) and also thought to be the case in tree crickets based on the known attachment points of the tympanal organ (Mhatre et al, 2021). The OCT data shows relatively low AP motion at frequencies near 3 kHz, and higher ML motion, which contradicts the less reliable light microscopy data. Given that the locust membrane shows peaks in motion at ~4.5 kHz, ~11 kHz, and also at ~20 kHz (Windmill et al, 2008), I am surprised that the authors limited their OCT experiments and analyses to 5 kHz.
We found that immediately above 5 kHz the displacements reduced to undetectable magnitudes. We accept that there may be other modes of vibration at higher frequencies >10 kHz (based on Jacobs et al., 1999) that we could have detected with OCT. However, we focused our analysis on Group-III neurons at the best frequency and frequencies that we could cross-compere between our high-speed imaging system and OCT.
In summary for this section, I am not convinced of the conclusion drawn by the authors that group III scolopidia receive significantly higher stimulation along the AP axis in their native configuration, if indeed they were studied in the appropriate force regime (altered due to excision).
Again, we accept our faults for not clearly displaying the anatomical references of the scolopidial and ciliary axes in Fig. 4 and Fig. 5. We also did not clearly describe in detail that our ex vivo preparation largely retains its physiological properties. We will address the errors of our measurement near Nyquist and provide additional information from Group 1 scolopidia where we could achieve higher data points per cycle.
In the scolopidial patch clamp data, the authors study transduction currents in response to steady state stimulation along the AP axis and the ML axis. The responses to steady state and periodic forces may well be different, and the authors do not offer us a way to clearly relate the two and therefore, to interpret the data.
We will revise the Fig. 5a to clarify that the push-pull were done along the Dorso-Ventral (DV) axis and the push-pull were done along the Antero-Posterior (AP) axis. We do agree that steady-state and periodic forces may well be very different. However, valuable insight can be gained from mechanical systems when displaced outside of their normal physiological frequency (e.g. the transformative work on vertebrate hair bundle mechanics, Howard & Hudspeth, 1988). For the same reason, we believe artificial stimulation of the scolopidium gives us new and crucial information to understand scolopidial mechanics. Our main finding that stretch is the dominant stimulus should still, or at least provide strong support, that stretch is the dominant stimulus in periodical motion.
In addition, both stimulation types, along the AP axis and the ML, elicit clear transduction responses. Stimulation along the AP axis might be slightly higher, but there is over 40% variation around the mean in one case (pull: 26.22 {plus minus} 10.99 pA) and close to 80% variation in the other (push: 10.96 {plus minus} 8.59 pA). These data are indeed from a very high displacement range (2000 nm), which is very high compared to the native displacement levels, which are in the 1-10 nm range.
In this experiment, we wished to establish the upper limits (and plateau region) of displacement-transduction current response. However, even at 2000 nm we still did not see a plateau. Therefore, we believe that the strain on the scolopidium is still in the operating range even though our displacement is not. This discrepancy can be explained because the base of the scolopidium is not fixed. Therefore, the displacement imposed in our experiment is not equivalent to the strain on the cilium but a combination of pulling and stretching along the length of the dendrite. The force, however, remains along that particular axis, supporting our main finding.
Another important consideration is that the cilium is surrounded by the scolopale wall. It is assumed that the scolopale wall is far stiffer than the ciliary and will therefore limit the amount of ciliary strain.
The factor change from sample to sample is not reported and is small even overall. The statistical analyses of these data are not clearly reported, and I don't see the results of the overall ANOVA in the results section.
We reported the statistical analyses in the Fig. 5 Source Data. We will now add tables displaying these statistics in the supplementary text of the revised manuscript.
I also find the dip in the reported transduction currents between 10 and 100 nm quite odd (Figure 5 j-m) and would like to know what the authors' interpretation of this behaviour is. It seems to me that those currents increase continuously linearly after ~50-100 nm and that the data below that range are in the noise. Thus, the transduction currents observed at the relevant displacement range (1-10 nm) may not actually be reliable. How were these small displacements achieved, and how closely were the actual levels monitored? Is it possible to reliably deliver 1-10 nm displacements using a micromanipulator?
One interpretation is that the cilium has both sensitive and insensitive mechanically gated ion channels. A finding that is also supported by Effertz et al., 2012. We will add a sentence in the discussion highlighting this interpretation. We will also provide our calibration of displacement vs voltage delivered to the piezo in the Supplementary Text.
What is clear, despite the difficulty in interpreting this data, is that both AP and ML stimulation evoke transduction currents, and their relative differences are small. Additionally, in Müller's organ itself, in the excised organ, the scolopidia are stimulated along both axes. Thus, in my opinion, it is not possible to say that axial stretch along the cilium is 'the key mechanical input that activates mechano-electrical transduction'.
We confirm that the scolopidia are displaced along both. We also note that displacements of the scolopidium limited to the up-down axis will also produce a strain on the scolopidium along the push-pull axis. However, we tried to disentangle this complex motion by limiting the displacements to one axis during recordings of the transduction current. We found that displacement along the scolopidial axis generated the largest transduction currents. Even though there is large variation our statistical analysis confirmed a significant difference as stated in the result section (Line 283 – 286)
“Additionally, the transduction current evoked by pull from the resting position was larger than displacement upward, 12.17 ± 5.37 pA (N = 11, n = 11) (Tukey's procedure, p = 1.75e-03, t = -3.83) or downward 7.28 ± 9.76 pA (N = 11, n = 11) (Tukey's procedure, p = 5.10e-06, t = -4.53).”
The reason for large variation is that the discrete depolarisations (random depolarisations of unknown function and a common feature of chordotonal neurons so far recorded) have a similar magnitude to the transduction current produced by the step displacements. We will highlight these discrete depolarisations in Figure 4d and mention them in the results.
Reviewer #2 (Public review):
Summary of strengths and weaknesses:
Using several techniques-FIB-SEM, OCT, high-speed light microscopy, and electrophysiology-Chaiyasitdhi et al. provide evidence that chordotonal receptors in the locust ear (Müller's organ) sense the stretch of the scolapale cell, primarily of its cilium. Careful measurements certainly show cell stretch, albeit with some inconsistencies regarding best frequencies and amplitudes.
Thank you very much for acknowledging the strength of our study. Regarding the inconsistencies between best frequencies and amplitude, we believe that this concern largely arises from our faults for not clearly displaying the anatomical references of the scolopidial and ciliary axes in Fig. 4 and Fig. 5. As previously addressed in our response to Reviewer#1, we will add the anatomical references and revised the text to clarify the orientation of our measurements.
The weakest argument concerns the electrophysiological recordings, because the authors do not show directly that the stimulus stretches the cells. If this latter point can be clarified, then our confidence that ciliary stretch is the proximal stimulus for mechanotransduction will be increased.
We agree that the displacement is not solely stretching the scolopidium. However, the force is still constrained and acting along the push-pull axis. Due to this reason, we overestimate the displacement required to open the MET channels but stand by our conclusion that stretch is the dominant stimulus. For future work, we wish to devise a technique to mechanically clamp the base of the scolopidium and measure the more physiological relevant current-strain relationship.
This conclusion will not come as a surprise for workers in the field, as the chordotonal organ is known as a stretch-receptor organ (e.g., Wikipedia). But it is a useful contribution to the field and allows the authors to suggest transduction mechanisms whereby ciliary stretch is transduced into channel opening.
One of the goals of this manuscript is to highlight the lack of direct evidence for stretch-sensitivity of chordotonal organs, as this is assumed from their structure. More importantly the acceptance of chordotonal organs, as being stretch sensitive does not address the mechanism of how organs work. For instance, one candidate for the MET channel, NompC, is shown to be sensitive to compression (Wang et al., 2021). We find that a preconceived concept of “stretch-sensitive” mechanism, without an appreciation of scolopidium mechanics, cannot explain how NompC can be opened in chordotonal organs.
P. .E. Howse wrote in his work on ‘The Fine Structure and Functional Organisation of Chordotonal Organs’ in 1968 (Symp. Zool. Soc. Lon.) No. 23
“There is, however, a common tendency to refer to chordotonal organs in which scolopidia are contained in a connective tissue strand as “stretch receptor”. This is unfortunate in two senses, for firstly the implied function may not have been proved and secondly even if the organ responds to stretch the scolopidia may not.” then he proceeded to cite a pioneering work in the chordotonal organs of the hermit crab by R.C. Taylor (Comp. Biochem. Physiol. 1966) showing that the scolopidia may experience flexing when the connective strand are stretched.
This work represents the first efforts to investigate the problematic assumption of stretch-sensitivity of scolopidia since it was first highlighted 57 years ago.
Reviewer #3 (Public review):
Summary:
The paper 'A stretching mechanism evokes mechano-electrical transduction in auditory chordotonal neurons' by Chaiyasitdhi et al. presents a study that aims to address the mechanical model for scolopidia in Schistocerca gregaria Müller's organ, the basic mechanosensory units in insect chordotonal organs. The authors combine high-resolution ultrastructural analysis (FIB-SEM), sound-evoked motion tracking (OCT and high-speed light microscopy), and electrophysiological recordings of transduction currents during direct mechanical stimulation of individual scolopidia. They conclude that axial stretching along the ciliary axis is an adequate mechanical stimulus for activating mechanotransduction channels.
Strengths/Highlights:
(1) The 3D FIB-SEM reconstruction provides high resolution of scolopidial architecture, including the newly described "scolopale lid" and the full extent of the cilium.
(2) High-speed microscopy clearly demonstrates axial stretch as the dominant motion component in the auditory receptors, which confirms a long-standing question of what the actual motion of a stretch receptor is upon auditory stimulation.
(3) Patch-clamp recordings directly link mechanical stretch to transduction currents, a major advance over previous indirect models.
Weaknesses/Limitations:
(1) The text is conceptually unclear or written in an unclear manner in some places, for example, when using the proposed model to explain the sensitivity of Nanchung-Inactive in the discussion.
We will rephrase and make clearer the context of our findings for Nanchung-Inactive mechanism of MET in the introduction and the discussion. We will also refine and simplify unclear text overall.
(2) The proposed mechanistic models (direct-stretch, stretch-compression, stretch-deformation, stretch-tilt) are compelling but remain speculative without direct molecular or biophysical validation. For example, examining whether the organ is pre-stretched and identifying the mechanical components of cells (tissues), such as the extracellular matrix and cytoskeleton, would help establish the mechanical model and strengthen the conclusion.
We agree with the speculative nature of our four proposed hypotheses. We have, however, narrowed down from at least ten previous hypotheses (Field and Matheson, 1998). These hypotheses will enable us, and hopefully the field, to test them and more rapidly advance our understanding of how scolopidia work. We will add a section in the discussion as to the best way to experimentally test these four hypotheses (e.g pushing directly onto the cap should elicit sensitive responses for the cap-compression hypothesis).
(3) To some extent, the weaknesses of the paper are part of its strengths and vice versa. For example, the direct push/pull and up/down stimulations are a great experimental advance to approach an answer to the question of how the underlying cellular components are deformed and how the underlying ion channels are forced. However, as the authors clearly state, neither of their stimulations can limit all forces to only one direction, and both orthogonal forces evoke responses in the neurons. The question of which of the two orthogonal forces 'causes' the response cannot be answered with these experiments and has not been answered by this manuscript. But the study has brought the field a considerable step closer to answering the question. The answer, however, might be that both longitudinal ('stretch') and perpendicular ('compression') forces act together to open the ion channels and that both dendritic extension via stretch and bending can provide forces for ion channel gating.
Thank you very much for your acknowledgement of our experimental advances. We agree that this study cannot identify and localise the forces on the cilium as it is enclosed in the scolopidial unit. As previously explained, we plan to address this question in our next work by improving and expanding our experimental techniques, including modelling, to study the scolopidial mechanics based on our experiments using patch-clamp recording in combination with individual and direct manipulation the scolopidium.
The current paper has identified major components (longitudinal stretch components) for the neurons they analysed, but these will surely have been chosen according to their accessibility, and as such, the variety of mechanical responses in Müller's organ might be greater. In light of these considerations, the authors might acknowledge such uncertainties more clearly in their paper.
Our high-speed and OCT imaging confirms complex multi-dimensional displacements (and presumably forces) acting on the scolopidium. We agree that our mechanical stimulation cannot recapitulate such complex motions. But for future work we wish to extend our mechanical stimulation to three axis and also to pivot on the axis of the scolopidial cap.
The paper is an impressive methodological progress and breakthrough, but it simply does not "demonstrate that axial stretch along the cilium is the adequate stimulus or the key mechanical input that activates mechano-electrical transduction" as the authors write at the start of their discussion.
We rephrase to clarity that stretching along the “scolopidial axis”, not “along the ciliary axis” is the adequate stimulus. We cannot yet verify how this translates to forces acting on the cilium, hence the four speculative hypotheses. We will re-write the discussion to make clear that we are only interpretating the forces and displacements at the level of the cilium.
They do show that axial stretch dominates for the neurons they looked at, which is important information. The same applies to the end of the discussion: The authors write, "This relative motion within the organ then drives an axial stretch of the scolopidium, which in turn evokes the mechano-electrical transduction current." Reading the manuscript, the certainty and display of confidence are not substantiated by the data provided. But they are also not necessary. The study has paved the road to answer these questions. Instead, the authors are encouraged to make suggestions on how the remaining uncertainties could be removed (and what experiments or model might be used).
We will moderate our conclusion in the discussion, but we are confident that we have experimental repeats, and the statistical test, to support our conclusion that stretching of the scolopidium provides that largest transduction current responses (although not at the level of the cilium). As mentioned previously, we will include a section in the discussion for the best way to test the hypotheses arising from this work.
