Talk:Models of midbrain dopaminergic neurons
We are very grateful for the reviewer’s comments since they really help to improve the quality of the article. Here are our suggestions to the comments:
1. In modeling each type of firing, it would be very helpful to show actual traces from the literature so that readers can understand exactly what firing activity is being modeled. Also, a schematic diagram depicting different components (channels, receptors, calcium, etc) in the model and their interaction should accompany each model. It is difficult to grasp what is being dealt with in the present form.
We have added two schematic diagrams for the background firing mechanism and for bursting evoked by bath application of NMDA in vitro. Since modeling of other firing patterns is partially based on similar assumptions, these diagrams should clarify those sections as well. Also, a number of figures from the literature showing modeled firing activity have been reproduced in the article.
2. The relationship between different models (what are similar and what are the critical differences) needs to be explained more clearly. For example, in the first paragraph of the “Conclusions and open problems” section, it is stated that “the latest results (Canavier and Landry 2006, Kuznetsov et al. 2006) suggest that the oscillatory mechanisms causing repetitive bursting in DA cells in vitro are irrelevant for in vivo bursts.” This sounded very abrupt (and bold) to me and the ground for this statement needs to be articulated. It would be nice to create a table summarizing and comparing all the models.
In in vitro bursting, two types of oscillations are superimposed: the first type is fast spiking, the second type is slow repetitive depolarizing wave that underlies bursts. The paragraph is meant to suggest that no repetitive depolarizing wave, but rather a synaptic input is required to trigger a burst in the DA neuron. We have modified the sentence to read more clearly, indicated that the text following the sentence supports this statement and added a sentence at the beginning of “In vivo bursting” that clarifies the absence of any mechanism solely designated for bursts in our previous article (Kuznetsov et al. 2006). Please also see a table summarizing the relationship between different models.
3. The title should be changed, for example to “Models of dopamine neuron firing”.
We agree that the article is focused on only one aspect of the dopaminergic neuron activity: its firing patterns. However, in our opinion, the scope of Scholarpedia as a computational neuroscience publication put the article in a context where it’s not strictly necessary to mention that.
4. It would be helpful to explain the biological significance of tonic and phasic DA neuron firing and the resulting tonic and phasic DA signals in more plain terms.
We have added a couple of sentences on this matter to the first paragraph of the section “Firing patterns”. However, we believe that this question deserves a separate article and want to minimize the discussion of the question here given the volume limitations.
1. Does DA released from midbrain DA neurons ever act as a neurohormone?
In midbrain, DA is not documented to act as a neurohormone.
2. What are the neuropeptides colocalized and released together with DA?
The most well known neuropeptides colocalized with DA is cholecystokinin and neurotensin. The release of these neuropeptides relies on firing of the neuron, as well as the release of dopamine. However, dopamine and the neuropeptides are released from different vesicles and may be spatially separated from each other.
3. It is stated that “biophysical properties of the DA cells in these groups are similar”. However, Jochen Roeper’s group has shown that this may not necessarily be the case (J. Neurosci., May 2001; 21: 3443 - 3456, J. Neurosci., Feb 2002; 22: 1290 - 1302).
We agree that there are some distinctions between SNc and VTA DA neurons. In particular, VTA DA neurons are shown to fire more irregularly and have a higher inclination to bursting. However, there still a great deal of similarity between the groups of DA cells. In particular, the properties described in our article apply to DA neurons from both the VTA and SNc. We believe that adding “but see Wolfart et al. 2001” to the sentence acknowledges the presence of the differences.
4. It may be more common to say “tonic firing” instead of “background firing”.
Background firing of the DA neuron can be irregular. Therefore, this term is broader than tonic firing. Background firing is linked to the putative role of this pattern: maintaining a background level of DA. Moreover, the models for background firing can reproduce both regular and irregular low-frequency firing patterns without noisy synaptic inputs. This suggests that this broadly defined background firing may be intrinsic for the DA neuron. Please see changes at the very beginning of the section “Firing patterns” that clarify the notion of background firing.
5. When burst firing is first introduced, the definition of “burst” should be provided.
Burst firing is first introduced in the first paragraph of the section ”Firing patterns”. We define burst firing there as characterized by much higher firing rates attained for short time intervals. Although this definition doesn’t provide the particular values of interspike interval at which the start and the end of a burst are registered, we believe that it gives the general idea how a burst is identified for the DA neuron.
6. Explain the pharmacological actions of barium and cadmium. Furthermore, the original reference should be provided here and other places whenever experimental observations are described.
In different neurons, barium is shown to blocks a number of potassium currents. The anomalous rectifier current is one of those, but this current doesn’t influence the firing pattern of the DA neuron much. Another current blocked by barium is shown to be the human ether-a-go-go potassium current (I HERG), which is identified in the DA neuron and suggested to influence the firing pattern. Barium has been shown to block several calcium-activated voltage-dependent potassium channels. The BK-type of such channels has been identified in DA neurons. Therefore, there are many ways for barium to influence the firing pattern of the neuron, and we can add here only that barium acts presumably by blocking K currents.
Cadmium is a nonselective blocker of calcium channels. However, based on this influence, it’s hard to justify the data showing that cadmium promotes bursting and moderately increases the firing frequency.
Even though the mechanisms of influence of barium and cadmium on DA neuron firing are not clear, these experiments help to make a clear conclusion that the firing pattern and rate can be manipulated differentially.
The paragraph stating the results on cadmium and barium applications had no citation, and we agree that this and other such instances must be rectified.
7. There are also other minor editing issues. These may be addressed later.
I feel that this review article has improved significantly from the original version. I only have a couple of comments.
1) Ca-dependence of various forms of burst is still not clearly described in terms of extracellular vs. intracellular Ca, as commented by the second reviewer. For example, it is stated that “Apamin-induced bursting is critically dependent on calcium, but in a way different from bursting in vivo.” How exactly are they different? It may be helpful to incorporate this information in the table, as differential “Ca-dependence” appears to be the important factor discriminating different types of burst. In the current table, “Ca balance” only in terms of location (soma only or distributed) is listed. Is “Ca balance” equivalent to intracellular “Ca-K mechanism”?
We regard this comment of the reviewer as allowing us to say a little bit more about Ca dependence of apamin-induced bursting, despite the volume limitations. We agree that the difference was not obvious form the previous text. Please see changes in the section on apamin-induced bursting. However, we believe we shouldn’t include this into the table because it summarizes modeling studies, reflecting the main focus of our article.
Ca balance includes calcium currents, pumps and buffering. The Ca-K mechanism includes the SK potassium current, the L-type Ca current and relies on Ca balance. So, these two notions are not the same, although the two corresponding columns are almost identical. The only difference is shown for the article by Li et al., where the SK current and the L-type Ca current are spatially separated to be in different compartments.
2) It is stated that “In contrast to other types of neurons, burst firing of DA cells cannot be elicited in vitro by somatic current injections.” Is this due to the difference in the density of Na channels or in their inactivation properties, thus making dopamine neurons more vulnerable to depolarization block?
This is one of open questions. Please see the new first paragraph in the concluding section.
3) I agree with the second reviewer that the statement “The neuron synaptically as well as dendritically releases dopamine along with colocalized neuropeptides” should be accompanied by a reference. I do not think this is generally shared knowledge. Also, is dopamine ever released as a “neurohormone”? It may be released as a “hormone” from the adrenal gland.
We decided to skip mentioning release of neuropeptides together with dopamine. Although the machinery for the release has been shown to be present in the dendrites as well as in the axon of the DA cell, there is no direct demonstration of simultaneous release of the neuropeptides and dopamine form dendrites of the DA neuron. Regarding dendritic release of dopamine, please see the added citation.
In our opinion, DA can be called a neurohormone since, along with it’s hormonal functions, it is expressed by neurons and can be released as a result of neuronal electrical activity. Please see Stuart R. Snider, Carl Miller, A. L. N. Prasad, Vernice Jackson and Stanley Fahn, Is dopamine a neurohormone of the adrenal medulla? Naunyn-Schmiedeberg's Archives of Pharmacology Volume 297, Number 1 / March, 1977
Although many issues were resolved by the revision a few remain. Several citations were inserted by me into the text, but need to be added to the reference list.
1) The following sentence must be supported by a citation. Also, it should be made clear whether it is only dopamine that is released somato-dendritically or if peptides are also released somatodendritically.
The neuron synaptically as well as dendritically releases dopamine along with colocalized neuropeptides.
There is no direct demonstration of simultaneous release of the neuropeptides and dopamine form dendrites of the DA neuron, although the machinery for the release has been shown to be present in the dendrites as well as in the axon. Therefore, we decided to avoid mentioning the neuropeptide release. Please see the text for the citation on the dendritic DA release.
2) Regarding the following sentence, calcium entry in the dendrites is supported by several studies, but where is the evidence for SK channel localization in both the soma and dendrites?
The mechanism is present in both soma and dendrites.
The dendritic localization of the SK current in the DA neuron has not been directly documented, to our knowledge. Therefore, we agree that it’s better to say that oscillations have been shown to be present in dendrites as well as in the soma, and this suggests that the SK current is distributed along the dendrites. Please see corrections in the text.
3) "Injection of a calcium buffering agent (calcium chelator), which decreases the concentration of free calcium, has been reported to abolish baseline as well as depolarization-induced bursting." Earlier in the article it is stated that bursting cannot be induced by depolarization, so this sentence is very confusing and needs to be changed or the statement that high rates cannot be evoked in vitro should possibly be tempered by the observation that they can be evoked by depolarizing current in vivo.
Regarding the idea that burst firing cannot be elicited by the injection of a depolarizing current, on page 2882 of Grace and Bunney 1984, it is stated that in vivo, "burst firing can be initiated by a depolarizing current injection" and an example of a high frequency burst is given in Fig 8B." This just emphasizes the contradictory nature of the literature. The same paper states in the caption of figure 12 that "In the presence of EGTA, the long-term depolarization does not alter its firing pattern. Instead of initiating a burst-firing pattern, the DA cell retains its pacemaker-like firing characteristics and merely fires at a faster rate in response to increasing levels of membrane depolarization." I do not see where EGTA was reported to abolish spontaneous bursting. However, two recent papers (Ji and Shepard 2006 and Waroux et al 2005) confirm that SK channel blockers increase burst firing in vivo, consistent with the actions of SK blockers in vitro, which is also consistent with Grace and Bunney's observation, cited in the article, that calcium applied extracellularly by microiontophoresis decreased burst firing by two thirds, since such a manipulation presumably elevated calcium influx and therefore SK channel activation. See the comments of Reviewer 3 on SK blockers. These references were inserted into the text.
We agree that depolarization-induced bursting in vivo should be introduced earlier, and we added a phrase on that to the second paragraph in Burst Firing.
Regarding the influence of EGTA, we echo the reviewer comment on the contradictory nature of the literature. The abstract of Grace and Bunney 1984 states “Increases in burst firing could also be elicited by intracellular Ca injection and could be prevented by intracellular injection of EGTA…” However, only suppression of depolarization-induced bursting has been supported in the main text (p. 2883). Therefore, we corrected our sentence to mention only this kind of bursting. We thank the reviewer for inserting the sentence on the influence of the SK current blockers. We added the cited articles to the reference list. However, we would say that induction of bursting by the SK blockers in vivo similar, but not completely consistent with their action in vitro. We do find the calcium dependence of the apamin-induced bursting in vitro different from the in vivo bursting. In our opinion, blocking SK current may play a permissive role allowing different kinds of bursting, as proposed by Johnson and Wu (2004).
4. "This type of bursting is calcium-independent: bursts persist in Ca2+-free solution, in contrast to in vivo burst firing. " With respect to the above sentence, where is the citation for abolition of in vivo bursting in Ca free solution? I'm not sure how such an experiment could be conducted.
Certainly, Ca-free solution refers to in vitro experiments on NMDA-induced bursting. We have changed the sentence since it could be interpreted incorrectly and added a reference.
5 "However, the latest results (Canavier and Landry 2006, Kuznetsov et al. 2006) question the relevance of the oscillatory mechanisms causing the slow depolarizing wave that underlie repetitive bursting in DA cells to in vivo bursts."
The above sentence is not a consensus, and I strongly disagree that the mechanisms underlying the slow depolarizing wave do not contribute to burst generation in Canavier and Landry 2006. See also the comments of reviewer 3. Here are some examples of how the "slow wave" dynamics affect bursting in that paper. In Fig 4 A3 and B3 The repetitive bursting activity is generated by the interaction of the sodium pump and the NMDA current in A3, and only modified by the change in B3 if modeling the NMDA conductance as with synaptic dynamics rather than as a constant. The accompanying text states clearly "both intrinsic and synaptic dynamics contribute to the timing of bursts". Again "Figure 5 shows how synaptic and intrinsic dynamics interact". Although the paper does state that "most bursts are triggered by a barrage of synaptic input", the burst does ride on a depolarizing wave and is followed by a postburst inhibition mediated by intrinsic processes. Also, when the level of glutamatergic input is excessive, repetitive bursting is exhibited by the model (Fig 10). The existence of mechanisms to generate repetetive bursting is consistent with a large body of data showing that blocking GABA A synapses or SK channels increases bursting in vivo, presumably due to mechanisms that are masked by these channels in vivo.
We should agree that we misinterpreted the results by Canavier and Landry 2006. Please see the changes to this section, which also take into account the comments of the third reviewer.
6) I echo reviewer 3's comment that it should be clearly stated that the mechanism of burst repolarization in vivo is controversial.
Please see the changes in the text and our answer to the third reviewer.
End of new comments
We thank the reviewer for very useful comments and answer the comments below in italic
The comments are arranged by section. The main comments are to include citations to support each observation, and to acknowledge that many of the experimental results are in conflict with each other such that there is little consensus on many issues. The other main comment involves a paper that was published since the article was written.
Background firing 1) Need a citation to support that the SOP persists after TTX (Wilson and Callaway 2000 and Ping and Shepard 1996 are appropriate). 2) Need a citation that nifedipine blocks the SOP to support the role of the L-type current- Harris et al is one of them, but I would remove the sentence "These two currents constitute a calcium-potassium mechanism for pacemaking in this neuron" as it seems redundant. 3) You also need a citation (Ping and Shepard 1996) for the role of SK. The citations are probably all in the paper somewhere but they need to be arranged so that it is clear that each statement is supported.
We agree that each statement must be supported, but we think that if several sentences in a row come from the same source, we may cite it only once for brevity. We agree that the first paragraph in “Background firing” does need an additional citation.
We don’t think that the sentence about the calcium-potassium mechanism is redundant. First, it introduces this term, which is widely used in the literature. Second, the text following the sentence says that the mechanism is distributed and has different natural frequency in the soma and dendrites. This is in distinction to the SOP (introduced by the previous sentence) because “a mechanism for the SOP” would imply conditions on proper synchronization of compartments and therefore would apply to the whole neuron, whereas the “calcium-potassium mechanism” may function in each compartment independently.
4) A citation of Richards et al 1997 is needed to support " burst firing of DA cells cannot be elicited in vitro by somatic current injections" and that the firing rate cannot be elevated to bursting rates. One would not expect to induce rhythmic bursting by current injection alone in any neuron in which the negative slope conductance region required for bursting is provided by a synaptic rather than an intrinsic current. The concept of rates attained during bursts needs to be made more precise, because by definition frequencies as low as 12.5 Hz (80 ms) can initiate a burst and as low as 6.2 Hz (160 ms ISI) can comprise part of a burst, and Richards et al 1997 showed sustained frequencies of 10 Hz and transient frequencies of 20 Hz. Also,the sentence "Thus, the mechanism responsible for burst firing in DA neurons is very different from the usual formulation, in which synaptic currents bring the cell closer to or farther from the spike threshold." needs to be rewritten. Many bursters do not require synaptic input to drive a burst, but instead rely on intrinsic currents for the burst mechanism (see neuron R15 in Aplysia for example). Synaptic input is required only for network bursters (the archtypical half center oscillator in Izhikevich et al 2007) or for bursters in which the negative slope conductance region underlying bursting is provided by the NMDA receptor mediated current. Even in these cases, intrinsic mechanisms such as post-inhibitory rebound in the case of network bursting and the outward current that provides the negative feedback in the case of NMDA-mediated bursting are important components of bursting. Therefore I think that it is inaccurate to say that in the "usual formulation" synaptic currents bring the cell closer to or farther from the threshold. I suggest rewriting the first paragraph as follows:
"Rhythmic bursting cannot be evoked by current injection alone. This is also true for a phasic burst because the maximum frequencies attained during bursting in vivo cannot be attained by current injection alone. Instead, the neuron enters depolarization block without exceeding transient frequencies of 20 Hz and steady firing frequencies of 10 Hz (Richards et al 1997)."
Stating that the mechanism for burst firing is different in this neuron, we had no intention to differentiate synaptic from intrinsic currents that may underlie bursting, as it may erroneously follow form our sentence. Whatever synaptic or intrinsic (or applied) currents bring a usual neuron to the spike threshold, it will respond with a spike train of an elevated frequency. This means that, in such a neuron, what is needed for bursting is a subthreshold oscillatory depolarizing wave, which is converted to a burst by producing repetitive spiking on top of the wave. This is the case with the R15 Aplysia neuron. However, in the DA neuron, such a wave is not enough, or it must have some specific additional properties, because repetitive high-frequency spiking is very fragile. To rectify the sentence stating this, we have changed it to "… synaptic or intrinsic currents slowly bring it closer to or further from the spike threshold".
We don’t think that there is a significant inconsistency in the definition of the burst. As the instantaneous frequency (reciprocal of the interspike interval) rises slightly above the 10 Hz limit of the sustained firing, a burst is registered. By this definition, once the beginning of the burst is registered, spikes are counted to belong to the burst even if they are separated by interspike intervals above 100 ms (below 10 Hz). However, since the average firing rate of the neuron in vivo is usually about 3-4 Hz, the chosen interspike interval works well for localization of a burst.
"Models of firing activity of the DA cell" 5) For completeness, you might want to mention Penney MS and Britton NF Modelling natural burst firing in nigral dopamine neurons. J. theor. Biol. (2002) 219, 207-233. They suggested that the calcium-mediated inactivation of the delayed rectifier causes the decrease in amplitude and increase in ISI during a burst. Although there is no experimental support for this hypothesis that I am aware of, the BK current likely mediates spike repolarization as well, and calcium-mediated inactivation (in addition to the more well-known activation) has been shown for BK channels in hippocampus (Shao et al. 1999, J. Physiol. 521.1:135-146). I do not feel strongly about his suggestion.
We have not included the article by Penny and Britton because, in our opinion, the relevance of their results to properties of the DA neuron is questionable as also noted by the reviewer. Many of their modeling assumptions are not relevant, or even contradictory to the facts known about the neuron.
In vivo bursting: 6) Need a citation for Firing rates of >= 20 Hz- I suggest one of the Schultz papers. 7) "episodes of elevated firing rate don't follow any apparent repetitive pattern" In order to illustrate how bursting in dopamine neurons is different from that in thalamic relay neurons, for example, the following sentence or something similar might convey the point better: "Although rhythmic bursting may occasionally be observed in vivo (Freeman, Meltzer and Bunney 1985), in general the firing pattern has great variability in the proportion of spikes fired in phasic bursts and episodes of irregular single spiking (Grace and Bunney 1984, Hyland et al 2001)." Also, you may want to note that Hyland et al 2001 dispute the Grace and Bunney contention that there is a stereotypical pattern of amplitudes and ISIs within the burst.
We agree with these comments. Please see our corrections introduced to this paragraph.
8) With respect to the role of AMPA, the recent paper by Blythe et al suggests a critical role for AMPA. Furthermore, the Meltzer et al 1997 paper cited in support of the dependency of bursting on NMDA but not AMPA states "there was no difference in the ability of NMDA AMPA and 1S,3R-ACPD to increase firing rate and burst-firing of A9 neurons". This citation cannot be used to support a privileged role for NMDA, but rather to bolster the opposing viewpoint that they are both important.
We cite Meltzer et al. 1997 to introduce the discussion about the significance of different types of glutamatergic synapses. This is a review article, and it includes the reference supporting the role of AMPA and NMDA. This article states “The effects of exogenously applied glutamate appear to be preferentially mediated by NMDA versus non-NMDA receptors.” Then, the article tries to reconcile this result with the result you cite. The authors suggest that, in vivo, glutamate causes bursting by acting primarily on NMDA receptors, but (artificial) application of a selective AMPA agonist can do the job as well. We suggest changing the way we cite the article to be “see e.g. Meltzer et al. 1997 for discussion”.
Blythe et al. 2007 speaks for the importance of AMPA, but this study is done in in vitro, so we think we should not add their result to this section.
9) With respect to the critical dependence on calcium, the one thing that everyone in the field agrees on is that blocking the calcium-activated SK potassium current increases burst firing, and it is difficult to reconcile this fact with the critical dependence of a burst on calcium. I think that this statement should be toned down a little since these results have not been replicated or explained.
The fact that blocking the SK current induces burst firing in vitro does not contradict to the calcium dependence of bursting. This manipulation blocks only one calcium-dependent channel, whereas other membrane mechanisms, such as calcium currents, have been shown to be necessary for apamin-induced bursting. So, we think that this is reasonable to say that this kind of bursting depends on calcium and explain how exactly the calcium dependence is displayed in experiments, as we do in further sentences.
In vitro bursting induced by apamin 10) I suggest "plateau potential oscillations" instead of square-wave oscillations (see Canavier et al 2007 posted on J Neurophysiology Online in August. 11) This type of bursting is critically dependent on an inward calcium current, but has been proven NOT to be dependent on calcium accumulation in the cytosol (see the same paper Canavier et al 2007). Please clarify this point, it would be easy to misinterpret, especially since the implication is that in vivo bursting and apamin induced bursting are similar because they depend on accumulation of calcium in the cytosol, which is not true because apamin induced bursting is clearly not calcium-dependent (The hypothesis presented for apamin-induced bursting Amini et al. 1999 has been disproved) .
We agree that calcium dependence of the in vivo and apamin-induced bursts is different. Please see changes in the text.
NMDA-induced bursting 12) What is meant by "rebound bursting " in the Komendantov et al 2004 model?
Rebound bursting is not reported in Komendantov et al. 2004, but only in Li et al. 1996. We have clarified this fact.
13. See 11 above. The Amini et al 1999 model of apamin-induced bursting has been replaced because the mechanism has been proven not to be dependent upon the accumulation of calcium in the cytosol. The new hypothesis is that a slow voltage-activated potassium current is responsible. I think the old, incorrect mechanism should be replaced with an illustration of the new hypothesized mechanism.
We have included Canavier et al. (2007). Please see the changes in this section of the article.
Conclusions and open problems
14) I agree that Kuznetsov et al 2006 questions the relevance of the oscillatory mechanisms underlying the rhythmic bursts produced in vitro to bursting in vivo, but it was not the intent of Canavier and Landry (2006) to question the relevance of those mechanism, just to suggest that a combination of intrinsic and synaptic factors influence the firing pattern, so I don;t think that the Canavier and Landry citation should be used in this context.
We agree that the relevance of mechanisms underlying rhythmic bursting to in vivo bursts is not directly discussed in Canavier and Landry 2006. However, there are several facts in this article that question the relevance of those mechanisms. First, the article discusses the interaction between NMDA, AMPA and GABA synaptic currents and the SK potassium current as shaping firing patterns in the model. The sodium mechanism responsible for NMDA-induced bursting in vitro is mentioned only briefly in the case of simulated apamin application. By contrast, if the SK current is not blocked by apamin, calcium processes influence the firing pattern more, in concert with the in vivo data.
Second, for a higher maximal NMDA conductance, the number of bursts and the number of spikes per burst are reported to decrease in response to doubling AMPA or blocking SK, a dependence opposite to the rest of the data and to that expected from experiments. Although, this is not written explicitly, one can conclude, e.g. form Fig. 10 in the paper, that the higher maximal NMDA conductance evokes oscillations that underlie bursts by the sodium mechanism suggested for NMDA application in vitro. The sodium mechanism is obviously responsible for the up and down states displayed in the timeseries in this case.
Third, the article is focused on the difference between a regular and stochastic glutamatergic synaptic input to the neuron. It shows a much better replication of experimentally observed firing patterns if the synaptic input is stochastic and obeys a realistic statistics. This difference cannot be mediated by a different response of the sodium oscillatory mechanism because sodium concentration is a slow variable and it averages the fluctuations coming from the synaptic inputs.
Summarizing the above observations, we come to the conclusion that the mechanisms for repetitive firing stay inactive while the model produce a firing pattern that resembles in vivo bursting in appearance.
We thank the reviewer for the comments and suggestions and answer all points below in italic.
This is an interesting and fair review of what is known about bursting in these neurons. There have already been several relevant comments by the two first reviewers and I will therefore only focus on a few topics:
1. I disagree with the idea that the properties of DA neurons are homogenous throughout the nigra and VTA (see comment 3 of referee 1). The medial DA neurons in the VTA are clearly different (see Roeper and also Margolis et al. J Physiol 2006)
We agree that the properties of the DA neuron are inhomogeneous, and we had no intention to state the opposite. Nevertheless, the reviewer will probably agree that there are similarities in the biophysical properties of different subpopulations of DA neurons. Please see the changes introduced in order to convey this point better.
2. I would emphasize Referee 2's comment 9, i.e; SK blockade enhances bursting in all conditions, including in vivo (Waroux et al. Eur J Neurosci 2005).
We agree that this is an important result and have added the reference and a corresponding sentence into the section on in vivo bursting.
3. The sentence which states that oscillatory mechanisms causing bursting in vitro are irrelevant for in vivo is much too strong and basically kills the whole review. I also think it may be wrong because we have little mechanistic information in vivo.
We have removed the statement mentioned above and rewritten the whole paragraph also taking into account the next comment of the reviewer. Please see changes in the text.
4. In the conclusion, I would clearly state that the mechanisms underlying repolarization of the bursts are controversial (active pump phenomenon, ERG-type K current, inactivation of an inward Ca current) and need more investigation.
We believe that it’s better to say that the contribution of each mechanism for the oscillations underlying bursts to in vivo bursting remains controversial. The oscillatory mechanisms themselves seem to be consistent and well supported. Please see the changes in the section “Conclusions and open problems”.