Multi-timescale dynamics organize descending pain modulation

  1. Carl Ashworth
  2. Melissa Martenson
  3. Zhigang Shi
  4. Caitlynn C. De Preter
  5. Mary Heinricher
  6. Flavia Mancini

  • Curated by eLife

    eLife logo

    eLife Assessment

    An intriguing set of electrophysiological and analytical results suggests that adult rat neurons in the medulla oblongata crucial for mediating descending pain modulation are embedded in a network that generates a specific pattern of oscillatory activity. These valuable findings may broaden our understanding of how descending pain modulation is achieved. The supporting evidence is incomplete, and the authors need to provide several methodological clarifications and additional data interpretations to strengthen the manuscript.

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

Effective pain therapies increasingly target neural circuits that regulate nociceptive processing; yet, how descending control systems regulate pain across time remains poorly understood. Because pain regulation must coordinate rapid defensive responses with slower fluctuations in physiological state, these neural circuits are likely to operate across multiple timescales. However, whether such dynamics exist in brainstem pain-control circuits remains largely unknown. Here, we investigated this question in populations of rostral ventromedial medullary (RVM) pain-modulating neurons. The RVM contains ON- and OFF-cells that exert descending control over spinal nociceptive transmission, regulating pain sensitivity and behaviors. By integrating neuronal recordings with probabilistic modeling, we show that unstimulated and stimulus-driven conditions give rise to distinct timescales of ON- and OFF-cell dynamics. During noxious stimulation, we find that population responses undergo rapid activation followed by superimposed slow and fast recovery dynamics over tens of seconds. In contrast, the same neurons exhibit quasi-periodic fluctuations in firing activity on the order of minutes in the absence of stimulation. Gaussian-process models show that these slow dynamics are statistically predictable from past activity, indicating structured temporal organization beyond stimulus-evoked responses. Taken together, these results indicate that descending pain-control circuits exhibit structured dynamics spanning rapid pain-related signaling and slower fluctuations associated with ongoing physiological state.

Significance

Pain regulation requires coordination between rapid defensive responses and slower changes in physiological state, yet how these processes are integrated in the brain remains unclear. We show that neurons in a key brainstem pain-control center, the rostral ventromedial medulla, operate across multiple timescales. Using neuronal recordings and computational modeling, we find that these neurons exhibit both fast responses to painful stimuli and slow, structured fluctuations in ongoing activity. These results demonstrate that descending pain control is temporally organized beyond immediate stimulus-evoked responses. This provides a framework for understanding how pain is regulated over time.

Article activity feed

  1. eLife

    eLife Assessment

    An intriguing set of electrophysiological and analytical results suggests that adult rat neurons in the medulla oblongata crucial for mediating descending pain modulation are embedded in a network that generates a specific pattern of oscillatory activity. These valuable findings may broaden our understanding of how descending pain modulation is achieved. The supporting evidence is incomplete, and the authors need to provide several methodological clarifications and additional data interpretations to strengthen the manuscript.

  2. eLife

    **Reviewer #1 (Public review):
    **
    Summary:

    The authors hypothesized that "RVM neurons operate across multiple temporal scales, integrating fast responses associated with reflex-linked control with slower fluctuations reflecting ongoing network or state-dependent modulation". The hypothesis was tested with the established ON/OFF-cell model and probabilistic modeling. The study is conceptually interesting and methodologically sophisticated. The findings build toward the conclusion that pain-control circuits operate across multiple timescales.

    Strengths:

    The use of Bayesian regression and Gaussian process modeling to quantify and characterize recovery dynamics and ongoing oscillatory activity.

    The authors show that slow rhythmic activity appears preferentially in ON- and OFF-cells but not in NEUTRAL-cells, suggesting that the oscillations are related to pain-modulatory circuitry rather than being a generic feature of all recorded neurons.

    The observation that some oscillatory activity is coherent with autonomic measures aligns with broader views of the RVM as a hub integrating nociceptive and homeostatic regulation.

    Some pitfalls are appreciated and discussed by the authors, including the functional significance of slow fluctuations, the influence of anesthetics on global brain-state dynamics, the molecular profiles of the studied ON- and OFF-cells, and the heart rate as a covarying signal of RVM neuronal activity.

    Weaknesses:

    A general weakness is that the work is mostly descriptive and relies on anesthetized preparations. Whether the observed rhythms occur in awake animals and are linked to fluctuations in pain behavior needs to be confirmed in future studies.

    The study measures limited autonomic variables. The causal relationship between "slow fluctuations" and "ongoing physiological state" is unclear and overstated, since the data presented appear correlational.

    ON- and OFF-cells in the RVM are identified by their responses correlated with reflexive activity. The significance of the observed oscillations in spontaneous pain conditions is unclear.

    It is uncertain whether the observed rhythms truly reflect intrinsic RVM organization rather than anesthesia-dependent phenomena; the authors appreciated this pitfall, though.

    The Gaussian process analysis suggests predictability and quasi-periodicity, but predictability alone does not necessarily imply a true biological oscillator.

    Conclusion:

    The results support the authors' hypothesis. The findings provide a compelling conceptual message about the multiscale organization and dynamics of descending pain-control circuits and encourage further studies on the topic.

  3. eLife

    Reviewer #2 (Public review):

    Using electrophysiological recordings in a well-characterized animal model of acute pain, and analytical and modeling methods, the authors show that descending pain-modulatory neurons in the rostral ventromedial medulla (RVM) operate across various timescales. They have both rapid multi-phase responses to noxious stimuli that unfold over tens of seconds, with distinct fast and slow recovery dynamics. Additionally, they generate slow quasi-periodic oscillations with approximately 5-minute periods during ongoing activity. These oscillations are statistically predictable and cell-type specific, demonstrating that descending pain control is organized through structured temporal dynamics that encompass immediate stimulus-evoked responses and slower fluctuations associated with physiological state.

    A novel discovery is a ~5-minute quasi-periodic oscillation in ongoing ON- and OFF-cell activity. This oscillation, along with its coherence with heart rate, forms the basis for the claim that descending pain circuits exhibit intrinsic multi-timescale organization. However, it's crucial to demonstrate that this periodicity is independent of external experimental cycles such as methohexital infusion pharmacokinetics, servo-controlled temperature regulation, or slow autonomic feedback loops, all of which operate on similar timescales. For instance, the 300-second period closely matches typical drug infusion cycling and thermoregulatory feedback intervals. Therefore, heart-rate coherence peaks at multiples of this period could equally reflect a shared external driver rather than intrinsic RVM organization. Although the absence of this cyclic structure in Neutral cells argues against this possibility, the authors might want to explicitly discuss this potential confound.

    The findings are important and novel in that they characterize an intriguing structure in the activity of ON and OFF neurons in the RVM. However, in the absence of a causal manipulation causality can only be inferred. That there is no phase-dependence of withdrawal latency argues against a causal role. The author are encouraged to qualify their conclusions (and their title) accordingly.

    Because anesthesia can affect global dynamics, this might affect the oscillations reported. Without awake validation, it remains uncertain whether these rhythms reflect an intrinsic property or an anesthesia-induced regime. Again, the absence of oscillations in Neutral cells argues against this possibility, but it is still possible that ON/OFF cells are embedded in different circuits that are affected differently by anesthesia.

    Analyses of many of the ON-cells had longer training windows (>1 sec) compared to those for the NEUTRAL cells. Could this have reduced the ability to fit and validate periodicity for the latter cell type?

  4. eLife

    Reviewer #3 (Public review):

    Summary:

    In this manuscript by Ashworth and colleagues, the authors investigate the temporal dynamics of the rostral ventromedial medulla (RVM), a key output node in a major descending pain-modulation circuit. Using data from extrasellar single-unit recordings of RVM ON, OFF, and NEUTRAL cells in lightly anesthetized rats, the authors' computational modeling yielded two major findings: (1) heat-evoked ON burst and OFF pause, followed by exponential recovery components in10s of seconds; and (2) ON and OFF cells exhibit periodic fluctuations in ~5-minute cycles that are statistically predictable.

    Strengths:

    The manuscript's concept is innovative, offering the first quantitative analysis of multi-timescale dynamics in physiologically characterized RVM pain-modulating neurons. This advances a field that has mostly depended on qualitative or single-timescale descriptions. The authors use contemporary Gaussian process and probabilistic models to capture statistically predictable slow dynamics. The study is further strengthened by identifying ON-, OFF-, and NEUTRAL-type cells using well-established criteria grounded in decades of RVM research. The combination of rapid reflex-related responses and slower ongoing rhythms supports a dual-timescale framework, providing a more integrated understanding of how these neurons may regulate reflex activity and state-dependent processes.

    Weaknesses:

    Several limitations are noted. Incomplete characterization of light anesthesia during recording sessions, such as methohexital stability and clear criteria for identifying "lightly anesthetized" states. While the NEUTRAL cell control is helpful, it does not fully address concerns about circuit specificity or systemic confounds. The findings are male-dominant, which may limit their generalizability. The synchrony between ON and OFF cells was suggested but not directly tested. The heart rate coherence with ON, OFF, and NEUTRAL cell activity results is intriguing but does not fully clarify how these neurons influence heart rate, particularly within the "lightly anesthetized" model.

  5. eLife

    Author response:

    We are pleased that the reviewers found the study conceptually novel and the analytical framework rigorous. In response we have substantially revised the manuscript to clarify methodological details, temper several interpretations, expand discussion of alternative explanations, and include additional analyses using the existing dataset. We have deliberately revised the manuscript so that our conclusions are limited to those directly supported by the data, namely that physiologically identified RVM pain-modulatory neurons exhibit structured dynamics spanning multiple temporal scales. We do not interpret the slow fluctuations as evidence for a specific intrinsic oscillator or for a causal role in physiological state regulation. We have also expanded the rationale for the lightly anaesthetized preparation, emphasizing that it provides both the recording stability required for prolonged single-unit recordings from sparse neurons in the deep RVM and a controlled physiological setting in which the baseline temporal organization of the circuit can be characterized while minimizing ongoing sensory, motor, and behavioral influences.

    Regarding the rationale for the lightly anaesthetized preparation, these experiments take advantage of the well-validated lightly anaesthetized Sprague-Dawley rat in which much of the foundational data concerning physiology and function of RVM neurons was obtained. This “middle-out” strategy [1] has allowed direct connections between the activity and pharmacology of identified RVM neurons and altered nociceptive behavior. This protocol demonstrably spares the essential links between brainstem pain-modulating neurons and nociceptive transmission pathways. Although the focus here was on ongoing activity, precluding the repeated nociceptive testing needed to link neuronal activity to nociceptive threshold, previous work has demonstrated that ongoing activity of OFF and ON-cells is correlated with nociceptive sensitivity [2] and that alterations in OFF- and ON cell firing in response to pharmacological manipulation and in models of persistent pain states, stress, and sickness have behavioral relevance [3–6,6–24]. Further, conclusions from work in lightly anaesthetized rats have repeatedly been found to be congruent with behavioral observations by other groups in awake rats and mice [19,25–36] and with functional imaging evidence in humans [37–40]. The lightly anaesthetized model has thus established a circuit-level explanatory framework for behavioral findings obtained in several species in multiple laboratories.

    A further consideration for the present study is that the lightly anaesthetized preparation allows us to examine the underlying temporal organization of the RVM under controlled conditions, without the additional factors that would necessarily come into play in an awake animal. Dynamics would inevitably be influenced by ongoing sensory input, behavioral priorities, arousal and other internal state changes. These factors would make it difficult to distinguish the intrinsic dynamics of the descending pain-modulatory system from the effects of the animal’s constantly changing experience.

    Public Reviews:

    Reviewer #1 (Public review):

    Summary:

    The authors hypothesized that “RVM neurons operate across multiple temporal scales, integrating fast responses associated with reflex-linked control with slower fluctuations reflecting ongoing network or state-dependent modulation”. The hypothesis was tested with the established ON/OFF-cell model and probabilistic modeling. The study is conceptually interesting and methodologically sophisticated. The findings build toward the conclusion that pain-control circuits operate across multiple timescales.

    Strengths:

    The use of Bayesian regression and Gaussian process modeling to quantify and characterize recovery dynamics and ongoing oscillatory activity.

    The authors show that slow rhythmic activity appears preferentially in ON- and OFF-cells but not in NEUTRAL-cells, suggesting that the oscillations are related to pain-modulatory circuitry rather than being a generic feature of all recorded neurons.

    The observation that some oscillatory activity is coherent with autonomic measures aligns with broader views of the RVM as a hub integrating nociceptive and homeostatic regulation.

    Some pitfalls are appreciated and discussed by the authors, including the functional significance of slow fluctuations, the influence of anesthetics on global brain-state dynamics, the molecular profiles of the studied ON- and OFF-cells, and the heart rate as a covarying signal of RVM neuronal activity.

    Weaknesses:

    A general weakness is that the work is mostly descriptive and relies on anesthetized preparations. Whether the observed rhythms occur in awake animals and are linked to fluctuations in pain behavior needs to be confirmed in future studies.

    The study measures limited autonomic variables. The causal relationship between “slow fluctuations” and “ongoing physiological state” is unclear and overstated, since the data presented appear correlational.

    ON- and OFF-cells in the RVM are identified by their responses correlated with reflexive activity. The significance of the observed oscillations in spontaneous pain conditions is unclear.

    It is uncertain whether the observed rhythms truly reflect intrinsic RVM organization rather than anesthesia-dependent phenomena; the authors appreciated this pitfall, though.

    The Gaussian process analysis suggests predictability and quasi-periodicity, but predictability alone does not necessarily imply a true biological oscillator.

    Conclusion:

    The results support the authors’ hypothesis. The findings provide a compelling conceptual message about the multiscale organization and dynamics of descending pain-control circuits and encourage further studies on the topic.

    We thank Reviewer R1 for their thoughtful and balanced assessment of our work. We are grateful for the reviewer’s positive evaluation of the conceptual framework, the analytical methodology, and the conclusion that the results support the hypothesis that RVM pain-modulatory neurons operate across multiple temporal scales. We agree with the reviewer’s central assessment that the present study is primarily descriptive and that several important questions regarding the origin and functional significance of the slow dynamics remain unresolved. We also appreciate the reviewer’s emphasis on clearly distinguishing observations directly supported by the data from their mechanistic interpretation. Although the manuscript already acknowledged that the present findings do not establish the mechanistic origin of the slow dynamics, we agree that this distinction could be made more explicit. We have therefore revised the manuscript to clarify that the approximately 5-minute fluctuations represent structured, quasi-periodic activity whose underlying origin cannot be determined from the present experiments. Throughout the manuscript we now explicitly acknowledge that these dynamics may arise from interactions between RVM circuitry and broader physiological or network processes, including anaesthesia-related state modulation, autonomic regulation, or other slow network influences. We also emphasise that the relationship between RVM activity and heart rate is correlational and does not establish a causal interaction. Finally, we now discuss more explicitly the complementary roles of controlled lightly anaesthetized and awake preparations. The present preparation was chosen to characterize baseline RVM dynamics under controlled sensory and behavioral conditions, whereas future awake studies will be important for determining how these dynamics are expressed and modulated during ongoing behavior, sensory experience, and chronic pain.

    (1) In the abstract, the ”timescales” are vaguely stated as ”rapid activation”, ”fast recovery dynamics,” and ”slow dynamics”. Quantifying these expressions with approximate ranges (milliseconds, seconds, tens of seconds, minutes, etc.) whenever possible would benefit readers.

    We thank the reviewer for this helpful suggestion. We have revised the Abstract to provide approximate timescales for the different phases of neuronal activity, distinguishing the rapid stimulus-evoked response (sub-second), recovery dynamics (seconds to hundreds of seconds), and ongoing quasi-periodic fluctuations (approximately 5 minutes). We believe these revisions improve the clarity of the Abstract and better convey the central findings of the study.

    (2) The abstract states that ”Effective pain therapies increasingly target neural circuits...” The connection to therapy is not clarified in the manuscript. A brief statement about how temporal dynamics might influence neuromodulation, analgesic interventions, or chronic pain could strengthen translational impact.

    We appreciate this suggestion. We have revised both the Abstract and Discussion to better explain the potential translational relevance of our findings. Rather than making a broad statement regarding pain therapies, we now briefly discuss how understanding the temporal organisation of descending pain-modulatory circuits may ultimately inform the design and timing of neuromodulatory interventions. We also emphasise that these implications remain speculative and require future investigation.

    (3) To address inter-animal and inter-neuron variability. Are the effects consistent across animals? Are all neurons oscillatory? Are the reported timescales driven by a subset of cells?

    We thank the reviewer for raising this important point. We have expanded the Results and Discussion to clarify the degree of variability observed across neurons and animals. In particular, we now emphasise that the slow quasi-periodic dynamics are not uniformly expressed across all neurons, but rather represent a structured population-level phenomenon with variability in predictability and modulation strength between cells, particularly within the OFF-cell population. We also clarify the consistency of the observed timescales across animals and discuss this variability as an important feature of the underlying circuitry rather than evidence for a single homogeneous oscillatory process.

    (4) Discuss what circuit mechanisms generate the oscillations. Are they driven by inputs from the PAG or intrinsic to RVM?

    We agree that the mechanisms underlying the slow temporal dynamics are an important question. We have expanded the Discussion to consider several possible sources of these dynamics, including intrinsic RVM circuitry, descending inputs from higher-order structures, and broader physiological or brain-state fluctuations. We emphasise that the present experiments cannot distinguish between these possibilities and have revised the manuscript to make this limitation more explicit while highlighting it as an important direction for future work.

    Reviewer #2 (Public review):

    Using electrophysiological recordings in a well-characterized animal model of acute pain, and analytical and modeling methods, the authors show that descending pain-modulatory neurons in the rostral ventromedial medulla (RVM) operate across various timescales. They have both rapid multi-phase responses to noxious stimuli that unfold over tens of seconds, with distinct fast and slow recovery dynamics. Additionally, they generate slow quasi-periodic oscillations with approximately 5-minute periods during ongoing activity. These oscillations are statistically predictable and cell-type specific, demonstrating that descending pain control is organized through structured temporal dynamics that encompass immediate stimulus-evoked responses and slower fluctuations associated with physiological state.

    A novel discovery is a 5-minute quasi-periodic oscillation in ongoing ON- and OFF-cell activity. This oscillation, along with its coherence with heart rate, forms the basis for the claim that descending pain circuits exhibit intrinsic multi-timescale organization. However, it’s crucial to demonstrate that this periodicity is independent of external experimental cycles such as methohexital infusion pharmacokinetics, servo-controlled temperature regulation, or slow autonomic feedback loops, all of which operate on similar timescales. For instance, the 300-second period closely matches typical drug infusion cycling and thermoregulatory feedback intervals. Therefore, heart-rate coherence peaks at multiples of this period could equally reflect a shared external driver rather than intrinsic RVM organization. Although the absence of this cyclic structure in Neutral cells argues against this possibility, the authors might want to explicitly discuss this potential confound.

    The findings are important and novel in that they characterize an intriguing structure in the activity of ON and OFF neurons in the RVM. However, in the absence of a causal manipulation causality can only be inferred. That there is no phase-dependence of withdrawal latency argues against a causal role. The author are encouraged to qualify their conclusions (and their title) accordingly. Because anesthesia can affect global dynamics, this might affect the oscillations reported. Without awake validation, it remains uncertain whether these rhythms reflect an intrinsic property or an anesthesia-induced regime. Again, the absence of oscillations in Neutral cells argues against this possibility, but it is still possible that ON/OFF cells are embedded in different circuits that are affected differently by anesthesia.

    We thank Reviewer R2 for their careful and constructive assessment of our work and for recognising the novelty of identifying structured multi-timescale dynamics in physiologically characterised RVM neurons. We particularly appreciate the reviewer’s thoughtful consideration of alternative explanations for the observed low-frequency temporal structure.

    The reviewer raises an important question regarding the extent to which the approximately 5-minute quasi-periodic dynamics reflect processes generated within descending pain-modulatory circuitry versus broader physiological or experimental influences. As discussed in the original manuscript, the present experiments cannot determine the precise mechanistic origin of these dynamics, and we have revised the Discussion to make this distinction more explicit. We now consider possible contributions from autonomic regulation, thermoregulatory processes, anaesthesia-related state modulation, and other slow physiological influences. We also clarify that the NEUTRAL-cell population argues against a uniform global effect acting similarly across all RVM neurons, but cannot exclude systemic influences that preferentially engage ON- and OFF-cell circuitry.

    We have additionally expanded the rationale for the lightly anaesthetized preparation. This preparation was not used solely for technical convenience. Stable single-unit recordings from physiologically identified ON- and OFF-cells are technically challenging because the RVM is a deep brainstem structure and these functional cell classes are relatively sparse; suppression of spontaneous movement therefore permits substantially greater recording stability over the prolonged epochs required here. Importantly, the preparation also provides a controlled physiological setting in which the underlying temporal organization of RVM activity can be examined while reducing the continuously changing sensory, motor, arousal, and behavioral influences that would necessarily contribute to RVM activity in an awake animal. Awake preparations are essential for determining how RVM neurons respond during ongoing behavior and natural sensory experience, but that is a complementary question to the one addressed here: whether physiologically identified RVM neurons exhibit structured temporal dynamics under controlled conditions.

    We have therefore revised the manuscript to present the lightly anaesthetized preparation as both a methodological choice and an important boundary condition on interpretation. We continue to acknowledge that anaesthesia may influence slow network dynamics, and that future awake recordings will be required to determine how the temporal structure identified here is expressed in the behaving animal. We have also revised the title and several sections of the manuscript to ensure that our conclusions consistently reflect the correlational nature of the data and do not imply mechanistic or causal interpretations beyond those directly supported by the experiments.

    (1) Analyses of many of the ON-cells had longer training windows (> 1 sec) compared to those for the NEUTRAL cells. Could this have reduced the ability to fit and validate periodicity for the latter cell type?

    We thank the reviewer for raising this important point. The difference between ON/OFFand NEUTRAL-cell analyses reflects the available recording durations rather than differences in the Gaussian process fitting procedure. All cell classes were fitted using the same GP model and training strategy; however, some NEUTRAL-cell recordings were shorter (960 s versus 1500 s for ON- and OFF-cells), resulting in correspondingly shorter training segments. Whilst the minimum frequency recoverable from a 960 s training segment is 0.00104 Hz, meaning that the 0.0033 Hz frequency observed in ON- and OFF-cells would be recoverable if present in a 960 s recording. We agree that shorter recordings could, in principle, reduce the ability to estimate slow periodic structure. We have therefore clarified this point in the Methods and Discussion. Importantly, the absence of predictable low-frequency dynamics in NEUTRAL-cells is supported not only by GP prediction performance but also by the independent power spectral analysis, the low latent GP variance, and the near-flat phase-normalised reconstructions, suggesting that the difference between cell classes is not solely attributable to recording duration. To address this concern, we will revise the manuscript to repeat the GP analysis after truncating the ON- and OFF-cell recordings to match the duration of the NEUTRAL-cell recordings. We will also include a 960-second-long simulated NEUTRAL-cell recording with periodic structure, to demonstrate that this would be located by our method if present.

    Reviewer #3 (Public review):

    Summary:

    In this manuscript by Ashworth and colleagues, the authors investigate the temporal dynamics of the rostral ventromedial medulla (RVM), a key output node in a major descending pain-modulation circuit. Using data from extrasellar single-unit recordings of RVM ON, OFF, and NEUTRAL cells in lightly anesthetized rats, the authors’ computational modeling yielded two major findings: (1) heat-evoked ON burst and OFF pause, followed by exponential recovery components in10s of seconds; and (2) ON and OFF cells exhibit periodic fluctuations in 5-minute cycles that are statistically predictable.

    Strengths:

    The manuscript’s concept is innovative, offering the first quantitative analysis of multitimescale dynamics in physiologically characterized RVM pain-modulating neurons. This advances a field that has mostly depended on qualitative or single-timescale descriptions. The authors use contemporary Gaussian process and probabilistic models to capture statistically predictable slow dynamics. The study is further strengthened by identifying ON-, OFF-, and NEUTRAL-type cells using well-established criteria grounded in decades of RVM research. The combination of rapid reflex-related responses and slower ongoing rhythms supports a dual-timescale framework, providing a more integrated understanding of how these neurons may regulate reflex activity and state-dependent processes.

    Weaknesses:

    Several limitations are noted. Incomplete characterization of light anesthesia during recording sessions, such as methohexital stability and clear criteria for identifying “lightly anesthetized” states. While the NEUTRAL cell control is helpful, it does not fully address concerns about circuit specificity or systemic confounds. The findings are male-dominant, which may limit their generalizability. The synchrony between ON and OFF cells was suggested but not directly tested. The heart rate coherence with ON, OFF, and NEUTRAL cell activity results is intriguing but does not fully clarify how these neurons influence heart rate, particularly within the “lightly anesthetized” model.

    We thank Reviewer R3 for their thoughtful and constructive assessment of our work. We appreciate the reviewer’s emphasis on providing additional methodological detail and placing the findings within the context and limitations of the experimental preparation. In response, we have substantially expanded the Methods to provide a more complete description of the lightly anaesthetized preparation, the methohexital infusion protocol, physiological monitoring, and the rationale for the ongoing recording paradigm.

    We have also clarified why this preparation was appropriate for the question addressed here. In addition to enabling stable long-duration single-unit recordings from sparse, physiologically identified neurons in the deep RVM, the lightly anesthetized preparation provides a controlled physiological setting in which baseline temporal dynamics can be characterized while minimizing ongoing sensory, motor, and behavioral influences. We nevertheless acknowledge that anaesthesia may alter slow brain-state dynamics, and we now make this limitation more explicit throughout the manuscript. We have also revised the Discussion to more clearly acknowledge the predominantly male sample, the interpretation of the NEUTRAL-cell population as a comparison group rather than a definitive control for systemic effects, the limitations of inferring synchrony from pseudo-population data, and the correlational nature of the heart-rate coherence analysis.

    (1) How does the lightly anesthetized preparation affect evoked and oscillation activity modeling? Given that cell activities can be highly influenced by the state of sedation and the pharmacology of methohexital, detailing how light anesthesia was achieved and determined can help interpret the limitations of the current model. For example, did the methohexital rate adjustments occur during the ongoing activity period used for GP modeling? What specific criteria defined “lightly anesthetized” beyond stable paw withdrawal latency, such as stable respiratory rate, EMG (reflex vigor?), and core temperature? Given RVM activity coupled to autonomic/thermoregulatory circuits, data on these variables should be reported, or their absence should be acknowledged.

    We thank the reviewer for this important comment. We have substantially expanded the Methods and Discussion to describe both the rationale for the lightly anaesthetized preparation and the criteria used to maintain it.

    The preparation offers both technical and conceptual advantages for the present question. Technically, the RVM is a deep brainstem structure and physiologically identified ON- and OFF-cells are relatively sparse. Prolonged extracellular recordings therefore depend on maintaining stable electrode–neuron contact, which is readily disrupted by spontaneous movement. Light methohexital anaesthesia suppresses spontaneous movement while preserving nocifensive withdrawal responses and the canonical physiological response patterns used to identify ON-, OFF-, and NEUTRAL-cells. Conceptually, the aim of the present study was to characterize the baseline temporal organization of identified RVM neurons rather than to determine which sensory, cognitive, or behavioral events drive their activity in an awake animal. An awake preparation would necessarily introduce continuously changing sensory input, motor activity, arousal, behavioral priorities, and other internal-state variables, all of which are known to influence RVM activity. These are important influences in their own right, but for the present question they would make it more difficult to distinguish underlying temporal structure from activity driven by ongoing experience. We therefore view controlled lightly anaesthetized and awake preparations as complementary: the former is useful for identifying foundational circuit dynamics under controlled conditions, whereas the latter will be essential for determining how those dynamics are modified and expressed during natural behavior.

    This preparation has also been extensively used to establish the canonical relationship between ON-/OFF-cell activity and nocifensive responses, pharmacological modulation of the RVM, and top-down control from structures including the hypothalamus and amygdala, with many of these functional relationships subsequently confirmed in awake behavioral experiments. We have added this context to the revised manuscript.

    With respect to physiological monitoring, core temperature was continuously monitored and maintained at 36–37 °C, heart rate was monitored by EKG, and EMG was recorded to monitor withdrawal responses. Light anesthesia was defined functionally by preservation of a stable nocifensive withdrawal response in the absence of spontaneous movement. Respiratory variables were not recorded, and we now acknowledge this explicitly as a limitation. We have also clarified in Methods that the Methohexital rates were not adjusted during the recording windows used for the gaussian process analysis.

    We therefore agree that the findings must be interpreted within the context of the lightly anaesthetized preparation, but we do not view awake recordings as a direct substitute for the present experiment. Rather, awake studies provide the important next step of determining how the structured dynamics identified under controlled conditions are modulated by sensory experience, behavioral state, and ongoing cognition.

    (2) It is unclear how the absence of slow oscillations in NEUTRAL cells can be used as an internal control for anesthesia and systemic drift. It is unlikely that NEUTRAL cells are identified in every single-cell recording session for them to be used as a consistent internal control. Also, as the authors suggested that the shared modulatory inputs to ON/OFF cells explain the coordinating mechanism for ON/OFF rhythmicity, the lack of rhythmicity or coherence in majority of the NEUTRAL cells may indicate that they do not receive the same modulatory inputs as ON/OFF cells. Would this make NEUTRAL cells insensitive to systemic changes throughout the recording sessions? Do rhythmic vs. non-rhythmic cells differ in location within the RVM?

    We appreciate the reviewer’s important distinction. We agree that NEUTRAL-cells should not be considered a definitive internal control for anaesthesia or systemic physiological drift. NEUTRAL-, ON-, and OFF-cells were not necessarily recorded simultaneously within the same session, and the functional classes may differ in the systemic or modulatory inputs they receive. Our intended inference is therefore narrower: the absence of comparable low-frequency temporal structure in most NEUTRAL-cells argues against a uniform global process that imposes the same temporal pattern on all RVM neurons. It does not exclude anaesthesia-related, autonomic, thermoregulatory, or other systemic processes that preferentially influence ON- and OFF-cell circuitry. We have revised the manuscript throughout to make this distinction explicit and now refer to NEUTRAL-cells as an informative comparison population rather than as a definitive control for systemic influences.

    Indeed, as the reviewer suggests, differential sensitivity to common modulatory inputs could itself contribute to the distinction between ON/OFF- and NEUTRAL-cell dynamics. This interpretation is also compatible with the observation that a subset of NEUTRAL-cells shows low-frequency coherence with heart rate despite lacking the structured approximately 5-minute temporal dynamics observed in the ON/OFF populations.

    We additionally examined the reconstructed recording locations and found no obvious anatomical segregation between neurons showing stronger versus weaker low-frequency structure within the sampled RVM region. We now state this in the revised manuscript. We appreciate the reviewer’s point that there may be locational differences between RVM rhythmic and non-rhythmic cells, which should be addressed in future work; however, determining this would require a substantially larger sample size, for example with multichannel probe recording, for a valid analysis.

    (3) It is important to acknowledge that findings are effectively male-only (77 M and 6 F). Although a recent publication demonstrated that RVM ON and OFF cell activities do not differ substantially on an individual level between male and female rats, sex differences in RVM population dynamics remain unexplored. The current finding may not be generalizable to females.

    We thank the reviewer for highlighting this important limitation. We now explicitly acknowledge in the Discussion that the present dataset is predominantly male (77 males, 6 females) and therefore does not permit meaningful assessment of sex differences in population dynamics. Although previous studies suggest that individual ON- and OFF-cell responses are broadly comparable between sexes, the generalisability of the present findings to female animals remains unknown and should be addressed in future work.

    (4) It was suggested that strong synchrony exists within each functional population (e.g., ON and OFF cells). However, phase-relationship or coherence analyses were lacking. Since there were recoding sessions with > 2 cells/animals, were there enough recordings that contain simultaneous ON/OFF pairs to allow for these analyses?

    We thank the reviewer for this helpful suggestion. We agree that direct analyses of synchrony between simultaneously recorded neurons would provide valuable additional information. However, the number of simultaneous recordings containing identifiable ON/OFF-cell pairs was insufficient to support a robust phase or coherence analysis. We have therefore revised the Discussion to avoid implying that synchrony has been directly demonstrated and instead describe the results as evidence for consistent low-frequency temporal structure across recordings. We also identify direct analysis of synchrony in larger simultaneously recorded neuronal populations as an important direction for future work.

    (5) It was intriguing that the ON-cell population’s ongoing activity shows a predictive structure, while the OFF-cell population does not (Figure 5). However, this interesting asymmetry in ongoing activity between two cell classes was not adequately explained in the discussion. For example, since shared modulatory inputs were proposed as the coordinating mechanism for ON/OFF rhythmicity, how may this difference in ON and OFF rhythm predictivity occur?

    We appreciate the reviewer drawing attention to this interesting observation. We have expanded the Discussion to consider possible explanations for the greater predictability observed in ON-cells relative to OFF-cells. Although both populations exhibited similar dominant timescales, ON-cells displayed larger latent GP variance and more consistent predictive performance, whereas OFF-cells exhibited greater heterogeneity across recordings. We now discuss several possible explanations for this asymmetry, including differences in intrinsic cellular properties, network coupling, or modulation amplitude, while emphasising that the present data do not allow these possibilities to be distinguished.

    (6) The relationship between RVM activity oscillations and cardiac rhythms appears to be covariate but may not support the ”physiologically meaningful” claim with the current analysis. Additional discussion could help clarify the findings of a) how the RVM oscillation period of 300s relates to the heart rate peak/oscillation period of 600s (Figure 6d) and b) how NEUTRAL cells show heart rate coherence but lack rhythmicity.

    We thank the reviewer for this thoughtful comment. We have revised the Discussion to more carefully interpret the heart-rate coherence analysis. In particular, we now emphasise that the observed coherence demonstrates shared low-frequency temporal structure but does not establish a causal relationship between RVM activity and cardiac dynamics. We also discuss the relationship between the approximately 300-s RVM timescale and the broader low-frequency components observed in the heart-rate spectrum, noting the limited frequency resolution available at these timescales. Finally, we expand our discussion of the NEUTRAL-cell results, clarifying that significant coherence in NEUTRAL-cells despite the absence of comparable structured low-frequency firing dynamics is consistent with shared physiological influences acting on multiple cell classes without implying that the slow temporal structure originates within NEUTRAL-cells.

    References

    (1) Noble, D. The Music of Life: Biology beyond the Genome (Oxford University Press, 2006).

    (2) Heinricher, M. M., Barbaro, N. M. & Fields, H. L. Putative Nociceptive Modulating Neurons in the Rostral Ventromedial Medulla of the Rat: Firing of On- and Off-Cells Is Related to Nociceptive Responsiveness. Somatosensory & motor research 6, 427–39 (1989).

    (3) Barbaro, N. M., Heinricher, M. M. & Fields, H. L. Putative Nociceptive Modulatory Neurons in the Rostral Ventromedial Medulla of the Rat Display Highly Correlated Firing Patterns. Somatosensory & Motor Research 6, 413–425 (1989).

    (4) Heinricher, M. M., Haws, C. M. & Fields, H. L. Evidence for GABA-mediated control of putative nociceptive modulating neurons in the rostral ventromedial medulla: Iontophoresis of bicuculline eliminates the off-cell pause. Somatosensory & Motor Research 8, 215–225 (1991).

    (5) Heinricher, M. M. & Kaplan, H. J. GABA-mediated inhibition in rostral ventromedial medulla: Role in nociceptive modulation in the lightly anesthetized rat. Pain 47, 105–113 (1991).

    (6) Heinricher, M. M. & Tortorici, V. Interference with GABA transmission in the rostral ventromedial medulla: Disinhibition of off-cells as a central mechanism in nociceptive modulation. Neuroscience 63, 533–546 (1994).

    (7) Heinricher, M. M., McGaraughty, S. & Grandy, D. K. Circuitry Underlying AntiOpioid Actions of Orphanin FQ in the Rostral Ventromedial Medulla. Journal of Neurophysiology 78, 3351– 3358 (1997).

    (8) Heinricher, M. M., McGaraughty, S. & Farr, D. A. The role of excitatory amino acid transmission within the rostral ventromedial medulla in the antinociceptive actions of systemically administered morphine. Pain 81, 57–65 (1999).

    (9) Heinricher, M. M., McGaraughty, S. & Tortorici, V. Circuitry Underlying Antiopioid Actions of Cholecystokinin Within the Rostral Ventromedial Medulla. Journal of Neurophysiology 85, 280–286 (2001).

    (10) Heinricher, M. M., Schouten, J. C. & Jobst, E. E. Activation of brainstem N-methyl-daspartate receptors is required for the analgesic actions of morphine given systemically. Pain 92, 129–138 (2001).

    (11) McGaraughty, S. & Heinricher, M. M. Microinjection of morphine into various amygdaloid nuclei differentially affects nociceptive responsiveness and RVM neuronal activity. Pain 96, 153–162 (2002).

    (12) Heinricher, M. M. & Neubert, M. J. Neural Basis for the Hyperalgesic Action of Cholecystokinin in the Rostral Ventromedial Medulla. Journal of Neurophysiology 92, 1982–1989 (2004).

    (13) Heinricher, M. M., Martenson, M. E. & Neubert, M. J. Prostaglandin E2 in the midbrain periaqueductal gray produces hyperalgesia and activates pain-modulating circuitry in the rostral ventromedial medulla. Pain 110, 419–426 (2004).

    (14) Heinricher, M. M., Neubert, M. J., Martenson, M. E. & Gonc¸alves, L. Prostaglandin E2 in the medial preoptic area produces hyperalgesia and activates pain-modulating circuitry in the rostral ventromedial medulla. Neuroscience 128, 389–398 (2004).

    (15) Kincaid, W., Neubert, M. J., Xu, M., Kim, C. J. & Heinricher, M. M. Role for Medullary Pain Facilitating Neurons in Secondary Thermal Hyperalgesia. Journal of Neurophysiology 95, 33–41 (2006).

    (16) Ortiz, J., Heinricher, M. & Selden, N. Noradrenergic agonist administration into the central nucleus of the amygdala increases the tail-flick latency in lightly anesthetized rats. Neuroscience 148, 737–743 (2007).

    (17) Xu, M., Kim, C. J., Neubert, M. J. & Heinricher, M. M. NMDA receptor-mediated activation of medullary pro-nociceptive neurons is required for secondary thermal hyperalgesia. PAIN 127, 253 (2007).

    (18) Ortiz, J. P., Close, L. N., Heinricher, M. M. & Selden, N. R. α2-Noradrenergic antagonist administration into the central nucleus of the amygdala blocks stress-induced hypoalgesia in awake behaving rats. Neuroscience 157, 223–228 (2008).

    (19) Edelmayer, R. M. et al. Medullary pain facilitating neurons mediate allodynia in headache-related pain. Annals of Neurology 65, 184–193 (2009).

    (20) Martenson, M. E., Cetas, J. S. & Heinricher, M. M. A possible neural basis for stress-induced hyperalgesia. Pain 142, 236–244 (2009).

    (21) Heinricher, M. M., Maire, J. J., Lee, D., Nalwalk, J. W. & Hough, L. B. Physiological Basis for Inhibition of Morphine and Improgan Antinociception by CC12, a P450 Epoxygenase Inhibitor. Journal of Neurophysiology 104, 3222–3230 (2010).

    (22) Heinricher, M. M., Martenson, M. E., Nalwalk, J. W. & Hough, L. B. Neural basis for improgan antinociception. Neuroscience 169, 1414–1420 (2010).

    (23) McGaraughty, S., Farr, D. A. & Heinricher, M. M. Lesions of the periaqueductal gray disrupt input to the rostral ventromedial medulla following microinjections of morphine into the medial or basolateral nuclei of the amygdala. Brain Research 1009, 223–227 (2004).

    (24) Rogness, V. M. et al. Descending Facilitation of Nociceptive Transmission From the Rostral Ventromedial Medulla Contributes to Hyperalgesia in Mice with Sickle Cell Disease. Neuroscience 526, 1–12 (2023).

    (25) Smith, D. J. et al. Dose-Dependent Pain-Facilitatory and -Inhibitory Actions of Neurotensin Are Revealed by SR 48692, a Nonpeptide Neurotensin Antagonist: Influence on the Antinociceptive Effect of Morphine1,2. The Journal of Pharmacology and Experimental Therapeutics 282, 899–908 (1997).

    (26) Hurley, R. W. & Hammond, D. L. The Analgesic Effects of Supraspinal µ and δ Opioid Receptor Agonists Are Potentiated during Persistent Inflammation. Journal of Neuroscience 20, 1249–1259 (2000).

    (27) Kovelowski, C. J. et al. Supraspinal cholecystokinin may drive tonic descending facilitation mechanisms to maintain neuropathic pain in the rat. Pain 87, 265–273 (2000).

    (28) Porreca, F. et al. Inhibition of Neuropathic Pain by Selective Ablation of Brainstem Medullary Cells Expressing the µ-Opioid Receptor. Journal of Neuroscience 21, 5281–5288 (2001).

    (29) Zhang, Y. et al. Identifying local and descending inputs for primary sensory neurons. Journal of Clinical Investigation 125, 3782–3795 (2015).

    (30) Franc¸ois, A. et al. A Brainstem-Spinal Cord Inhibitory Circuit for Mechanical Pain Modulation by GABA and Enkephalins. Neuron 93, 822–839.e6 (2017). URL https://www.ncbi.nlm. nih.gov/pmc/articles/PMC7354674/.

    (31) Kim, J.-H. et al. Yin-and-yang bifurcation of opioidergic circuits for descending analgesia at the midbrain of the mouse. Proceedings of the National Academy of Sciences 115, 11078–11083 (2018).

    (32) Nguyen, E. et al. Medullary kappa-opioid receptor neurons inhibit pain and itch through a descending circuit. Brain 145, 2586–2601 (2022).

    (33) Jiao, Y. et al. Molecular identification of bulbospinal ON neurons by GPER, which drives pain and morphine tolerance. The Journal of Clinical Investigation 133, e154588 (2023).

    (34) Nguyen, E., Grajales-Reyes, J. G., Gereau, R. W. & Ross, S. E. Cell type-specific dissection of sensory pathways involved in descending modulation. Trends in Neurosciences 46, 539–550 (2023).

    (35) Fatt, M. P. et al. Morphine-responsive neurons that regulate mechanical antinociception. Science 385, eado6593 (2024).

    (36) Wang, Q. et al. Deconstruction of a spino-brain–spinal cord circuit that drives chronic pain. Nature 1–10 (2026).

    (37) Brooks, J. C., Davies, W.-E. & Pickering, A. E. Resolving the Brainstem Contributions to Attentional Analgesia. The Journal of Neuroscience 37, 2279–2291 (2017).

    (38) Mills, E. P. et al. Brainstem Pain-Control Circuitry Connectivity in Chronic Neuropathic Pain. Journal of Neuroscience 38, 465–473 (2018).

    (39) Mills, E. P., Keay, K. A. & Henderson, L. A. Brainstem Pain-Modulation Circuitry and Its Plasticity in Neuropathic Pain: Insights From Human Brain Imaging Investigations. Frontiers in Pain Research (Lausanne, Switzerland) 2, 705345 (2021).

    (40) Oliva, V., Hartley-Davies, R., Moran, R., Pickering, A. E. & Brooks, J. C. Simultaneous brain, brainstem, and spinal cord pharmacological-fMRI reveals involvement of an endogenous opioid network in attentional analgesia. eLife 11, e71877 (2022).

  6. Version published to 10.64898/2025.12.15.693905 on bioRxiv