Dendritic delay lines shape the computation of sound location in neurons of the gerbil medial superior olive

  1. Jared Casarez
  2. Rebecca I. Voglewede
  3. Bradley D. Winters
  4. Ken Ledford
  5. Nace L. Golding

  • Curated by eLife

    eLife logo

    eLife Assessment

    This is a fundamental study that clarifies the cellular mechanism of sound localization in the horizontal plane. The analysis of medial superior olivary neurons provides experimental and computational evidence for a new mechanism in which a range of asymmetric dendritic delays permits individual MSO neurons to represent the full range of biologically relevant ITDs. Using elegant 2-photon guided simultaneous recordings from distal dendrites and soma, along with compartmental modeling on anatomically reconstructed neurons, the authors provide compelling evidence that this mechanism contributes to microsecond-level tuning.

Reviewed by

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

Log in to save this article

Abstract

In mammals, neurons of the medial superior olive encode information about sounds arising from discrete spatial locations along the horizon. This tuning requires that an internal delay in the brain must offset acoustic disparities to ensure coincident arrival of excitatory inputs driven from the two ears. The source of this optimal internal delay, originally assumed to arise from axonal delay lines, is currently controversial in mammals. Here we use 2-photon guided paired dendritic and somatic recordings together with compartmental modeling of 40 complete MSO neuron morphologies to demonstrate that the dendrites themselves serve as a significant source of internal delay. We show that most MSO neurons exhibit morphological asymmetries that impose different EPSP delays across dendrites and shifts in optimal interaural time differences. Dendrite-based delays in the mammalian MSO are heterogeneous within each isofrequency laminae and provide a stable, structural mechanism to help tune individual neurons to sounds from different azimuthal locations.

Article activity feed

  1. eLife

    eLife Assessment

    This is a fundamental study that clarifies the cellular mechanism of sound localization in the horizontal plane. The analysis of medial superior olivary neurons provides experimental and computational evidence for a new mechanism in which a range of asymmetric dendritic delays permits individual MSO neurons to represent the full range of biologically relevant ITDs. Using elegant 2-photon guided simultaneous recordings from distal dendrites and soma, along with compartmental modeling on anatomically reconstructed neurons, the authors provide compelling evidence that this mechanism contributes to microsecond-level tuning.

  2. eLife

    Reviewer #1 (Public review):

    Overview:

    This study examines cellular computations in the dendrites of neurons in the medial superior olive (MSO) required for computing sound location based on interaural time differences (ITD). This field had, for many decades, depended on the so-called Jeffress model, which stated that an array of binaural coincidence detector neurons fire only when a given sound lateralization is balanced by a given difference in presynaptic axonal conduction time. The apparent absence of such calibrated axonal delay lines has left the field with little mechanistic handle for the strong ITD computations in MSO. This study suggests that dendritic delay along the dendrites of the bipolar MSO neurons makes a significant contribution to a calibrated delay line.

    Strengths:

    The authors used a combination of in vitro patch-clamp recordings, morphological analysis of a large dataset, and computational modelling to gain experimental access to dendritic computations. A technical tour-de-force set of distal dendritic patch-clamp recordings allowed an evaluation of this otherwise inaccessible parameter, and detailed modeling based on large datasets revealed the functional consequences. The use of this broad methodological toolbox enabled a detailed study of dendritic integration in MSO neurons and revealed a prominent role for graded variation in dendrite structure in shaping the coincidence detection in MSO neurons. In addition, the modeled effects of synaptic inhibition were quite striking and shaped our understanding of ITD coding in the MSO.

    Weaknesses:

    The paper's organization does not set up the reader very well for the major point to be made about exactly how dendritic asymmetry could bias ITD curves. This point only arises later in the paper after discussion of uncorrelated physiological measures that merely hint that what is important is "larger morphological and electrotonic structure". The paper could also benefit from a more complete description of the methodology. As an example, bridge balance goes unmentioned, and series resistance is hardly mentioned, even though both could distort the measurements of simulated EPSP amplitudes made through tiny electrodes used for dendrite recording.

  3. eLife

    Reviewer #2 (Public review):

    Medial superior olivary neurons are sensitive to interaural time differences in the microsecond range, and many cellular mechanisms have been advanced to explain this temporal sensitivity. This study provides experimental and computational evidence for a new mechanism in which a range of asymmetric dendritic delays permits individual MSO neurons to represent the full range of biologically relevant ITDs. Using elegant 2-photon guided simultaneous recordings from distal dendrite and soma, along with compartmental modeling on anatomically reconstructed neurons, the authors provide compelling evidence that this mechanism contributes to microsecond-level tuning. The experimental design, analyses, and narrative are all well-crafted. It's a beautiful study. As outlined below, I have two general questions about interpretations drawn from the experimental data and modeling.

    (1) Both excitatory and inhibitory synapses on MSO neurons display significant short-term depression (Couchman et al., 2010). Given the amount of attenuation at the soma, the role that the distal inputs would play after stimulus onset has not been tested. Were simulated EPSC pulse trains with endogenous short-term plasticity kinetics injected into distal dendrites? If not, were EPSP and IPSP trains with endogenous short-term plasticity kinetics studied in the model? The fundamental question is how much distal synapses contribute to somatic spike initiation as a function of synaptic pulse number.

    (2) The model provides a credible line of evidence that synaptic inputs from distal and tertiary compartments can generate reliable increases in the time of arrival at the soma. It would be relatively simple to sequentially prune dendritic compartments to address how the time difference at which the maximal firing rate scales with tertiary or distal compartments. Similarly, one could eliminate the primary dendrites to determine whether or not they play a functional role. I would expect these chores to be largely confirmatory, but since EPSP delay and amplitude are convolved, it would increase confidence in the interpretation.

    (3) Two technical questions. The age range is fairly broad, and it is not clear at which ages the experimental recordings were obtained, especially for the key experimental graphs that show correlations between delay (Figure 1d) or tau (Figure 2e) and distance. In addition, age could be added to Supplementary Figure 1, and the data could be ordered from youngest to oldest. Second, the Methods section indicates that brain slices were gradually cooled to 25 {degree sign}C, but should specify whether or not the recordings were obtained at this temperature.

  4. eLife

    Reviewer #3 (Public review):

    Summary:

    The study addresses how mammalian medial superior olive (MSO) neurons generate the internal delays required for interaural time difference (ITD) coding and sound localization. The authors demonstrate that dendritic morphology, particularly asymmetry between lateral and medial dendritic arbors, contributes to differential EPSP propagation delays and thereby shifts the optimal ITD of individual MSO neurons, using two-photon-guided paired dendritic and somatic recordings with compartmental modeling. This is a strong and potentially impactful manuscript. The work provides compelling evidence that dendritic morphology contributes to coincidence detection and ITD tuning in MSO neurons.

    Strengths:

    A major strength of the study is its technically rigorous combination of experimental electrophysiology, detailed neuronal reconstructions, and computational modeling. The use of paired dendritic and somatic recordings provides direct physiological insight into EPSP propagation, while the modeling approach allows the authors to test how cell-specific morphology influences coincidence detection. The analysis of multiple reconstructed MSO neurons further supports that dendritic asymmetry generates differential EPSP propagation delays that contribute to ITD tuning. This is a novel and potentially important mechanism that may complement classical axonal delay-line models. The study is strong in its anatomical and electrophysiological approach.

    Weaknesses:

    No major weakness. However, some aspects of the methods and interpretation would benefit from clarification. First, the assumptions used in the compartmental models should be more explicitly described, including the distribution of glutamatergic synaptic inputs and synaptic conductance parameters. It would be useful to clarify whether excitatory inputs were assumed to be homogeneously distributed along primary and higher-order dendritic branches or assigned based on known MSO input organization. Anatomical validation using VGluT staining together with dendritic labeling could strengthen the physiological relevance of the modeled input patterns. Second, the morphological analysis is informative, but additional measures of dendritic complexity could further support the conclusions. In addition to path length and membrane surface area, analyses of primary neurite number, branch points, and terminal arbors, using Sholl profiles or fractal dimension, could provide a more comprehensive assessment of lateral-medial dendritic asymmetry.

  5. eLife

    Author response:

    We thank the reviewers for their enthusiasm for the work as well as for their thoughtful and constructive comments, which will lead to many improvements in the manuscript. We will address their concerns/suggestions in the following ways:

    Reviewer 1

    (1) We will revise text to help the reader more intuitively understand how dendritic asymmetry can translate into alterations in receptive field location, as well as provide a better description of the cited portions of the Methods section.

    Reviewer 2

    (1) The simulations in the current version of the manuscript modeled a transient response via a single synaptic conductance in part because one can better visualize the interplay between synaptic inputs and voltage-gated ion channels across both time and dendritic space. However, we agree that it is also important to show how our results are impacted during ongoing trains of synaptic activity exhibiting short-term depression as documented in the literature. We will add an additional figure showing simulations employing realistic statistical patterns of presynaptic excitatory and inhibitory inputs with appropriate short-term plasticity characteristics. These simulations are already complete and show that the increased complexity minimally alters the location of modeled ITD curves of the cell population over a wide range of frequencies (250 Hz – 2 kHz).

    (2) The reviewer’s suggestion of sequentially pruning the different orders of dendritic branches is an excellent one. However, removal of dendrites also alters overall whole cell resistance and capacitance as well as the cable properties of the remaining dendrites. It is thus impossible to disentangle the branch-specific effects of synapse location from changing intrinsic electrical properties. However, the reviewer has inspired us to address their suggestion in a slightly different way: we will add (via a new figure) simulations that take place in the same dendritic arbor, but with inputs restricted to progressively lower orders of dendritic branches. Thus, the relative contributions of synapses onto higher order dendritic branches can be visualized without fundamentally changing the electrotonic structure of the simulated neurons across the different conditions. These simulations will be performed under the “in vivo-like” conditions described in the previous point. We think they will effectively address the essence of the reviewer’s suggestion.

    (3) We will add more specific information about animal ages in relevant figures, including Supplementary Figure 1. We will also indicate that all physiological recordings were performed near physiological temperature (35°C), which was unintentionally omitted.

    Reviewer 3

    (1) We will add more detail about the anatomical assumptions regarding spatial input patterns vs. higher order dendrites. We do not think that VGluT staining with dendritic labeling will be a productive experiment, since the thin sections that provide high quality labeling conditions also preclude following single dendrites for long distances. The distal portions, which are of particular interest, are most difficult to follow because of their smaller diameter and more extensive branching out of the plane of thin sections. Further, the work of Callan and colleagues (2021) has addressed axonal input patterns as well as dendritic coverage, documenting that single axon inputs follow dendrites for variable distances, and typically provide multiple synaptic contacts. This work also highlights the many challenges and large effort involved in documenting synaptic innervation patterns in single cells at the light microscopic level. Thus, we do not think we can improve upon existing anatomical descriptions without excessively expanding the scope of an already long study, which will have 9 figures after revision.

    (2) We have analyzed many other measures of dendritic complexity but for reasons of clarity and focus included the two measures that appeared most intuitive and impactful (length and surface area). We agree that access to other measures would be useful even if some are less intuitive, and thus we will provide a more comprehensive analysis of dendritic structure in a supplementary figure.

    References:

    Callan, A. R., Heß, M., Felmy, F., & Leibold, C. (2021). Arrangement of Excitatory Synaptic Inputs on Dendrites of the Medial Superior Olive. The Journal of neuroscience, 41(2), 269–283. https://doi.org/10.1523/JNEUROSCI.1055-20.2020

  6. Version published to 10.64898/2026.01.28.702328 on bioRxiv