Auditory mismatch responses are differentially sensitive to changes in muscarinic acetylcholine versus dopamine receptor function

Curation statements for this article:
  • Curated by eLife

    eLife logo

    Evaluation Summary:

    This study adds to the considerable, but often conflicting, work on how neurotransmitter systems contribute to auditory processing dysfunction. The paper details a thorough and careful analysis of an important hypothesis from the point of view of schizophrenia research: do muscarinic and dopaminergic receptors contribute to mismatch negativity effects? The answers could be useful for future treatment allocation in psychosis. The analysis was pre-registered and departures from the planned analysis were well-motivated and clearly described.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

This article has been Reviewed by the following groups

Read the full article See related articles

Abstract

The auditory mismatch negativity (MMN) has been proposed as a biomarker of NMDA receptor (NMDAR) dysfunction in schizophrenia. Such dysfunction may be caused by aberrant interactions of different neuromodulators with NMDARs, which could explain clinical heterogeneity among patients. In two studies (N = 81 each), we used a double-blind placebo-controlled between-subject design to systematically test whether auditory mismatch responses under varying levels of environmental stability are sensitive to diminishing and enhancing cholinergic vs. dopaminergic function. We found a significant drug × mismatch interaction: while the muscarinic acetylcholine receptor antagonist biperiden delayed and topographically shifted mismatch responses, particularly during high stability, this effect could not be detected for amisulpride, a dopamine D2/D3 receptor antagonist. Neither galantamine nor levodopa, which elevate acetylcholine and dopamine levels, respectively, exerted significant effects on MMN. This differential MMN sensitivity to muscarinic versus dopaminergic receptor function may prove useful for developing tests that predict individual treatment responses in schizophrenia.

Article activity feed

  1. Author Response:

    Evaluation Summary:

    This study adds to the considerable, but often conflicting, work on how neurotransmitter systems contribute to auditory processing dysfunction. The paper details a thorough and careful analysis of an important hypothesis from the point of view of schizophrenia research: do muscarinic and dopaminergic receptors contribute to mismatch negativity effects? The answers could be useful for future treatment allocation in psychosis. The analysis was pre-registered and departures from the planned analysis were well-motivated and clearly described.

    Thank you for this positive statement. We would like to make sure that the nature of our pre-registration is fully understood: we did not formally pre-register our study (i.e., there was no independent peer review). Instead, we defined an analysis plan ex ante (i.e., before beginning the data analysis for examining drug effects), and time-stamped and uploaded this plan on our institutional Git repository, prior to the unblinding of the analysing researcher. This a priori analysis plan is publicly available as well as our analysis code, and we report any departures from the analysis plan in our manuscript.

    Reviewer #1 (Public Review):

    The reduced amplitude of the mismatched negativity (MMN) in Schizophrenic patients has been associated with NMDA receptor malfunction. Weber and colleagues adjusted the systemic levels of two neurotransmitters (acetylcholine and dopamine), that are known to modulate NMDA receptor function, and examined the effects on mismatch related ERPs. They examined mismatch related ERPs elicited during a novel passive auditory oddball paradigm where the probability of hearing a particular tone was either constant for at least 100 trials (stable phases) or changed every 25-60 trials (volatile phases). Using impressive statistical testing the authors find that mismatch responses are selectively affected by reduced cholingeric function particularly during stable phases of the paradigm, but not by reduced dopamine function. Interestingly neither enhanced cholingeric or dopamine function affected MM responses at all. While the presented data support the main conclusions mentioned above, there are some claims in the abstract and text that are not supported by the results.

    1. The authors state in the abstract that "biperiden reduced and/or delayed mismatch responses......", while the results (Figure 2) support the statement that biperiden delayed mismatch responses, the claim that biperiden reduced mismatch responses is misleading as on P13 the authors actually report that "mismatch signals were stronger in the biperiden group compared to the placebo group at right central and centro-parietal sensors" around 200ms. This is close both in time and spatially to the traditional temporal and spatial locations of the MMN component. If one were to only read the abstract they would take away the result that the muscarinic acetylcholine receptor antagonist biperiden has an attenuative effect on MMN which is not what the results show.

    Thank you for this comment. We agree that the description in the abstract might be misleading and have changed our wording there. We now say (in the overall shortened abstract):

    “We found a significant drug x mismatch interaction: while the muscarinic acetylcholine receptor antagonist biperiden delayed and topographically shifted mismatch responses, particularly during high stability, this effect could not be detected for amisulpride, a dopamine D2/D3 receptor antagonist.”

    1. The conclusion that biperiden reduced mismatch responses may be due to the finding that at pre-frontal sensors mismatch responses were significantly smaller in the biperiden group than in the amisulpride (a dopaminergic receptor antagonist) group (P9) around 164ms. However, it is difficult to interpret if this is a meaningful result as amisulpride was found not to significantly alter mismatch responses in any way compared to placebo. It would be more convincing if the significant difference here were between biperiden and placebo groups. Or are we to think of amisulpride as being comparable to a placebo?

    We agree with your previous point and have adjusted our wording in the abstract accordingly (see response to previous comment).

    Furthermore, we have included an additional section in the Discussion in which we address the points you raise:

    "One might wonder whether the early difference between the biperiden and the amisulpride group at pre-frontal sensors is difficult to interpret, given the lack of differences of either drug group compared to placebo. However, given our research question – i.e., whether auditory mismatch signals are differentially susceptible to muscarinic versus dopaminergic receptor status – showing a significant difference between biperiden and amisulpride is critical.

    Clearly, such a differential effect would be even more compelling if biperiden differed significantly from amisulpride and placebo at the same time (and in the same sensor locations). While we do not find this in our main analysis, we do see it for the analysis using the alternative pre-processing pipeline and the trial definition (Figure 2—figure supplement 3) that was also specified a priori in our analysis plan. In this alternative analysis, mismatch responses under biperiden did differ significantly from both placebo and amisulpride."

    We suspect this difference in results between the analysis pipelines might partly be due to the different re-referencing. Compared to the average reference used in the main analysis, the linked mastoid reference in the alternative pre-processing pipeline subtracts the effects at sensors which show positive mismatch signals from those at fronto-central channels (with opposite sign), effectively enhancing the signal at the fronto-central channels (for evidence of this effect see also current Figure 3—figure supplement 1) but weakening it at temporal and pre-frontal sensors.

    We now discuss the question of sensitivity of both our paradigm and processing strategy in the discussion.

    1. The authors use the words mismatch negativity (MMN) and mismatch responses interchangeably however in some cases it is clearly mismatch responses being described and not the classical MMN ERP component. This occurs especially in the Introduction where the authors describe the study and that they plan to focus on the MMN but in the results section, since the initial analysis focuses on all sensors, other mismatch responses are consistently discussed. These differences in wording need to be precisely defined and used consistently in the text.

    We agree that it is important to use precise definitions of the terms and be consistent in their use. The dipole source signal of mismatch detection shows up with different signs across different sensor locations, and “MMN” traditionally refers to the effect in fronto-central channels, where it is a deviant-induced negativity. However, even when we constrain the use of “MMN” to the (difference in) negative deflection at fronto-central channels between 100 and 250ms (or similar) there remains some ambiguity due to the choice of reference. A common choice in MMN research is a linked mastoid reference. Because the mismatch signal shows up at the mastoids with opposite sign to fronto-central channels, this reference maximizes the observed difference at fronto-central channels (see also our Figure 3—figure supplement 1 and our reply to the previous comment) and minimizes it elsewhere, effectively forcing all (drug or other) effects to show up at frontocentral channels. This demonstrates that we typically think of the effects at different sensor locations as (caused by) one and the same (dipole source) signal. In our average referenced data (our main analysis), we observe some effects at fronto-polar sensors, where they are expressed as a modulation of a positive deflection, however, we think of these as being part of what is typically referred to as “MMN” for the above reasons.

    However, to avoid any confusion that this may cause, we have adapted the wording in our manuscript everywhere and mention this distinction in the methods section:

    “To avoid confusion, we will only use the term “MMN” when we talk about effects in the classical time window (100-200ms) and sensor locations (frontocentral sensors) for the MMN, and use “mismatch responses” for all other effects.”

    1. A weakness of the paper would be that the authors offer no prediction in the Introduction about what the expected effects of these specific neurotransmitter modulations would be on mismatch responses.

    Thank you for this suggestion and apologies for this oversight. We have now added a sentence to the Introduction, describing the effects we expected based on previous literature.

    Based on previous literature, one would expect mismatch responses in our paradigm to be sensitive to (1) volatility, with larger mismatch amplitudes during more stable phases (Dzafic et al., 2020; Todd et al., 2014; Weber et al., 2020), and (2) cholinergic manipulations, with galantamine increasing and biperiden reducing mismatch amplitudes (Moran et al., 2013; Schöbi et al., 2021). Furthermore, we expected a differential effect of cholinergic (muscarinic) and dopaminergic receptor status on mismatch responses, as postulated by initial work on MMN-based computational assays (Stephan et al., 2006). Our results suggest that muscarinic receptors play a critical role for the generation of mismatch responses and their dependence on environmental volatility, whereas no such evidence was found for dopamine receptors.”

    1. A nice aspect of this paper is that the authors re-analyzed their data using pre-processing settings identical to those used in comparable research papers examining the effect of cholinergic modulation on MMN. The main findings did not differ following this re-analysis.

    Reviewer #2 (Public Review):

    The authors found that Biperiden (M1 antagonist) delayed and altered the topography of MMN responses, particularly in the stable condition. Amisulpride did not do so, and neither did Galantamine or L-DOPA. The analysis using an ideal Bayesian observer (the HGF) detailed in the Appendix showed that Biperiden reduced the representation of lower-level prediction errors and increased that of higher-level prediction errors (about volatility).

    The methods were rigorous (including obtaining drug plasma levels and detailing alternative preprocessing techniques) and I have no suggestions for improvement from that point of view.

    I only have one main comment that I think could be discussed. I'm not an expert on this but as I understand it, Olanzapine is most selective for M2 receptors rather than M1 (https://www.nature.com/articles/1395486), although Clozapine metabolites do have some M1 selectivity (https://www.pnas.org/content/100/23/13674) - I'm not sure about Clozapine itself. So Biperiden (very M1 selective) might not be the ideal drug to use to explore a treatment allocation paradigm, at least for Olanzapine? I suspect the options are quite limited but it would probably be worth commenting on this.

    Thank you for pointing this out, this is indeed an important point for the discussion.

    First, clarifying the pharmacodynamics of psychopharmacological drugs and their relative affinity to different receptor subtypes is notoriously difficult as this depends on many methodological factors. The seminal paper on the binding profile of olanzapine (which, at the same time, also examined clozapine) is (Bymaster et al., 1996). Using in vitro assays, this study found that both olanzapine and clozapine showed by far the greatest affinity for the M1 receptor (see the Table 5). By contrast, using SPECT data from seven patients with schizophrenia treated with olanzapine, the paper you mentioned (Raedler et al., 2000) estimated the affinity of olanzapine to the M2 receptor as being roughly twice as high as to the M1 receptor. Both studies have methodological pros and cons (as discussed by (Raedler et al., 2000)). From our view, an important limitation by the study of (Raedler et al., 2000) is that they used the ligand [I-123]IQNB which is not selective and "does not allow discrimination between the different subtypes of the muscarinic receptors" (Raedler, Knable, Jones, Urbina, Gorey, et al., 2003). Instead, the M1/M2 comparison by (Raedler et al., 2000) rested on conclusions from a mathematical approximation – under various assumptions and with only 7 data points available. We note that subsequent studies by the same group on muscarinic receptors in schizophrenia (Raedler, Knable, Jones, Urbina, Egan, et al., 2003; Raedler, Knable, Jones, Urbina, Gorey, et al., 2003) no longer used this approach and refrained from making statements about relative selectivity of olanzapine and clozapine with regard to M1/M2 receptors. Furthermore, the results by (Raedler et al., 2000) are potentially confounded by the fact that they were not obtained from healthy controls, but from patients with schizophrenia. This is potentially problematic: if schizophrenia is characterised by an aberration related to M1 receptors (see below), this would affect the interpretability of the results by (Raedler et al., 2000). Overall, the relative affinity of olanzapine and clozapine to M1/M2 receptors remains a matter of debate, but it seems safe to say that both drugs affect both receptors.

    Second, we would like to explain that we think of biperiden as a model of a (potential) impairment, rather than a treatment. A series of studies have provided compelling evidence for a role of muscarinic (M1) receptor dysfunction in the pathophysiology of schizophrenia. In particular, there is compelling evidence for a subgroup of patients with markedly decreased M1 availability in the prefrontal cortex ((E. Scarr et al., 2009); see also (Gibbons et al., 2013) and (Elizabeth Scarr et al., 2018)). Moreover, multiple studies have found antipsychotic effects of xanomeline, an M1/M4 agonist (Bodick et al., 1997; Shekhar et al., 2008).

    Against this background, clozapine and olanzapine may seem counterintuitive as treatment options since they antagonize muscarinic receptors. However, the muscarinic system is complex, and the mechanisms by which muscarinic receptors are involved in the therapeutic effects of clozapine and olanzapine are far from being understood. One interesting observation is that both clozapine and olanzapine have been found to elevate extracellular acetylcholine concentrations in cortical regions (Ichikawa et al., 2002; Shirazi-Southall et al., 2002), potentially by blocking muscarinic autoreceptors (Johnson et al., 2005), although this is debated (Tzavara et al., 2006). There is clinical evidence that clozapine or its metabolites may exert their pro-cognitive effects by increasing the release of actetylcholine (Weiner et al., 2004), and preclinical evidence that clozapine is able to normalize M1 receptor availability in cortex (Malkoff et al., 2008).

    Irrespective of the exact mechanism by which clozapine and olanzapine exert their antipsychotic effects, their much higher affinity to muscarinic cholinergic receptors compared to dopaminergic receptors sets them apart from other antipsychotics. If a functional readout of the relative contribution of cholinergic versus dopaminergic deficits could be obtained in individual patients, this might be predictive of whether this patient would profit from clozapine, olanzapine, or, in the future, potential new treatments targeting the muscarinic system specifically.

    Given the above considerations, we have amended the relevant paragraph in the discussion to state this rationale more clearly.

    Notably, there is compelling evidence for a subgroup of patients with markedly decreased M1 availability in the prefrontal cortex ((E. Scarr et al., 2009); see also (Gibbons et al., 2013) and (Elizabeth Scarr et al., 2018)). This is consistent with the possibility that a key pathophysiological dimension of the heterogeneity of schizophrenia derives from a differential impairment of cholinergic versus dopaminergic modulation of NMDAR function (Stephan et al., 2006, 2009). Distinguishing these potential subtypes of schizophrenia could be highly relevant for treatment selection, as some of the most effective neuroleptic drugs (e.g., clozapine, olanzapine) differ from other atypical antipsychotics (e.g., amisulpride) in their binding affinity to muscarinic cholinergic receptors. The exact mechanisms by which muscarinic receptors are involved in the therapeutic effects of clozapine and olanzapine are still under debate and include, for example, elevation of extracellular levels of acetylcholine in cortex (Ichikawa et al., 2002; Shirazi-Southall et al., 2002; Weiner et al., 2004), possibly via blocking presynaptic muscarinic autoreceptors (see (Johnson et al., 2005; Tzavara et al., 2006) for conflicting data), and normalization of M1 receptor availability in cortex (Malkoff et al., 2008). Irrespective of the exact mechanism by which clozapine and olanzapine exert their antipsychotic effects, their much higher affinity to muscarinic cholinergic receptors compared to dopaminergic receptors sets them apart from other antipsychotics. If a functional readout of the relative contribution of cholinergic versus dopaminergic deficits could be obtained in individual patients, this might be predictive of whether this patient would profit from clozapine, olanzapine, or, in the future, potential new treatments targeting the muscarinic system specifically. Indeed, muscarinic receptors have become an important target of drug development for schizophrenia (Yohn & Conn, 2018).

  2. Evaluation Summary:

    This study adds to the considerable, but often conflicting, work on how neurotransmitter systems contribute to auditory processing dysfunction. The paper details a thorough and careful analysis of an important hypothesis from the point of view of schizophrenia research: do muscarinic and dopaminergic receptors contribute to mismatch negativity effects? The answers could be useful for future treatment allocation in psychosis. The analysis was pre-registered and departures from the planned analysis were well-motivated and clearly described.

    (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. The reviewers remained anonymous to the authors.)

  3. Reviewer #1 (Public Review):

    The reduced amplitude of the mismatched negativity (MMN) in Schizophrenic patients has been associated with NMDA receptor malfunction. Weber and colleagues adjusted the systemic levels of two neurotransmitters (acetylcholine and dopamine), that are known to modulate NMDA receptor function, and examined the effects on mismatch related ERPs. They examined mismatch related ERPs elicited during a novel passive auditory oddball paradigm where the probability of hearing a particular tone was either constant for at least 100 trials (stable phases) or changed every 25-60 trials (volatile phases). Using impressive statistical testing the authors find that mismatch responses are selectively affected by reduced cholingeric function particularly during stable phases of the paradigm, but not by reduced dopamine function. Interestingly neither enhanced cholingeric or dopamine function affected MM responses at all. While the presented data support the main conclusions mentioned above, there are some claims in the abstract and text that are not supported by the results.

    1. The authors state in the abstract that "biperiden reduced and/or delayed mismatch responses......", while the results (Figure 2) support the statement that biperiden delayed mismatch responses, the claim that biperiden reduced mismatch responses is misleading as on P13 the authors actually report that "mismatch signals were stronger in the biperiden group compared to the placebo group at right central and centro-parietal sensors" around 200ms. This is close both in time and spatially to the traditional temporal and spatial locations of the MMN component. If one were to only read the abstract they would take away the result that the muscarinic acetylcholine receptor antagonist biperiden has an attenuative effect on MMN which is not what the results show.

    2. The conclusion that biperiden reduced mismatch responses may be due to the finding that at pre-frontal sensors mismatch responses were significantly smaller in the biperiden group than in the amisulpride (a dopaminergic receptor antagonist) group (P9) around 164ms. However, it is difficult to interpret if this is a meaningful result as amisulpride was found not to significantly alter mismatch responses in any way compared to placebo. It would be more convincing if the significant difference here were between biperiden and placebo groups. Or are we to think of amisulpride as being comparable to a placebo?

    3. The authors use the words mismatch negativity (MMN) and mismatch responses interchangeably however in some cases it is clearly mismatch responses being described and not the classical MMN ERP component. This occurs especially in the Introduction where the authors describe the study and that they plan to focus on the MMN but in the results section, since the initial analysis focuses on all sensors, other mismatch responses are consistently discussed. These differences in wording need to be precisely defined and used consistently in the text.

    4. A weakness of the paper would be that the authors offer no prediction in the Introduction about what the expected effects of these specific neurotransmitter modulations would be on mismatch responses.

    5. A nice aspect of this paper is that the authors re-analyzed their data using pre-processing settings identical to those used in comparable research papers examining the effect of cholinergic modulation on MMN. The main findings did not differ following this re-analysis.

  4. Reviewer #2 (Public Review):

    The authors found that Biperiden (M1 antagonist) delayed and altered the topography of MMN responses, particularly in the stable condition. Amisulpride did not do so, and neither did Galantamine or L-DOPA. The analysis using an ideal Bayesian observer (the HGF) detailed in the Appendix showed that Biperiden reduced the representation of lower-level prediction errors and increased that of higher-level prediction errors (about volatility).

    The methods were rigorous (including obtaining drug plasma levels and detailing alternative preprocessing techniques) and I have no suggestions for improvement from that point of view.

    I only have one main comment that I think could be discussed. I'm not an expert on this but as I understand it, Olanzapine is most selective for M2 receptors rather than M1 (https://www.nature.com/articles/1395486), although Clozapine metabolites do have some M1 selectivity (https://www.pnas.org/content/100/23/13674) - I'm not sure about Clozapine itself. So Biperiden (very M1 selective) might not be the ideal drug to use to explore a treatment allocation paradigm, at least for Olanzapine? I suspect the options are quite limited but it would probably be worth commenting on this.