The role of PDF neurons in setting the preferred temperature before dawn in Drosophila

Essential revisions:

There are several technical issues that dampen our enthusiasm for this study as currently presented. Yet, addressing these concerns would significantly increase the confidence with which several central results could be interpreted and these concerns are addressable.

One major technical concern is the genetic drivers used to differentially express transgenes within the DN2s and s-LNvs for the GRASP and Kir2.1 experiments. Ideally these experiments would employ completely independent GAL4 and LexA drivers, one expressing strongly in the DN2s and the other in the LNvs. Of course, life is never this easy. The GAL4 driver used here for the DN2s (Clk9M-Gal4) is also, unfortunately, expressed in the s-LNvs, thus requiring the need for Pdf-GAL80 to turn-off GAL4/UAS-mediated expression in the s-LNvs. As designed, the experiments in the study essentially require Pdf-GAL80 silencing of UAS expression to be perfectly complete. Unfortunately, there is no evidence that this is true. This leads to serious concerns about the GRASP signal seen this study and the effects of Kir2.1 expression driven by Clk9M-Gal4/Pdf-GAL80, both of which may simply reflect persistent GAL4 expression in the s-LNvs.

We recently published the immunostaining data of the Clk9M-Gal4, Pdf-GAL80>UAS-GFP flies and showed that there were no GFP signals in sLNvs (Figure 2A in Goda et al., The Journal of Neuroscience, 2016). We conclude that Pdf-GAL80 can silence the Clk9M-Gal4 expression in sLNvs and added this information in the Materials and methods.

DN2/s-LNv GRASP: The authors describe GRASP located in the dorsal protocerebrum that is with a synaptic connection between the s-LNvs and the DN2s. Maximal putative GRASP signal between s-LNvs and DN2s corresponded exactly to the previously established time of maximal spread of the s-LNv dorsal projection. Given the possibility of some GAL4 driven expression in the s-LNvs (i.e., incomplete silencing by Pdf-GAL80), we are worried that the authors are simply visualizing GFP reconstitution in the axoplasm of s-LNvs. How sure are they that Pdf-GAL80 has completely silenced UAS driven spGFP1-10 expression in the s-LNvs? One way to reassure the reader (and the authors) that this isn't the case, is to show a complete lack of GRASP signal outside of the dorsal protocerebrum. It would be very reassuring, for example, to see no GFP fluorescence in the s-LNv cell bodies or the first 1/2 of their dorsal projections. We did not see this possibility addressed clearly in the Results section or the figures.

We updated the GRASP image in Figure 2A to includes the area and depth from the sLNvs cell bodies to the dorsal projections. The GRASP signals were clearly visible in the dorsal protocerebrum, but not in the s-LNv cell bodies or the first half of their dorsal projections. We also confirm it using PDF antibody to label LNvs in Figure 2—figure supplement 1.

P2X2: Though the data presented for a physiological connection between s-LNvs and DN2s does support the presence of such a connection, critical negative controls for this experiment are not shown. Though the authors state that preparations in which P2X2 was not driven in the s-LNvs, displayed no responses to ATP, the details of this control were not described and data are not shown. It is critical to show the ATP responses of Clk9M-Gal4/UAS-GCaMP/LexAop-P2X2 flies, using the same LexAop-P2X2 element used in the experimental fly. Unfortunately, leaky P2X2 expression is always a possibility and must be controlled for. We think the authors should show these control experiments to fully control for this important observation.

As the reviewer suggested, we performed the additional negative control in the P2X2 experiment using Clk9M-Gal4::UAS-GCaMP x LexAop-P2X2 flies (Figure 3—figure supplement 1). The fluorescence in sLNvs and DN2s did not increase upon the ATP application. On the other hand, we also performed the same experiment in Figure 3AB as a positive control and reached the same conclusion. Thus, the data indicates that the responses in Figure 3AB were specific due to P2X2 activation in sLNvs by ATP.

Kir2.1 inhibition of DN2s: Isn't it possible, given the concern about incomplete silencing of UAS expression by Pdf-GAL80, that this is simply phenocopying Pdf-GAL4/UAS-Kir2.1 because there is enough Kir2.1 expression in the s-LNvs to result in inhibition? This is especially worrisome in this case, as it appears that there are two copies of the Clk9M-Gal4 present in these experimental flies.

We would like to point out that only one copy of the Clk9M-Gal4 was used in the Kir2.1 inhibition of DN2s (Figure 3). Because some of the genotypes of fly lines were not clearly written in the previous manuscript, we have revised them.

As mentioned above, we showed that Clk9M-Gal4; Pdf-GAL80 does not drive sLNvs and that sLNvs-DN2s GRASP signals were not visible in the sLNvs’ somas. The data suggest that Pdf-GAL80 is sufficient to block the expression of Clk9M-Gal4 in sLNvs. Therefore, we conclude that DN2 inhibition causes the phenotype of the lower temperature preference of Clk9M-Gal4/UAS-Kir2.1; Pdf-GAL80/+ flies.

AC/s-LNv connection. This would be a major finding with big implications for how temperature input makes its way into the circadian system. It is very nice to see that the GRASP signal between these two neuron types persists even when the more specific AC driver (NP0002-Gal4) is used. GRASP alone is not compelling evidence that the AC neurons form synapses on or modulatory connections with the s-LNvs. The two neurons may simply rest in close apposition (the dorsal protocerebrum is a busy place) or, if there is a connection, the PDF neurons could be modulating the AC neurons. Without evidence for a physiological connection between AC neurons and s-LNvs it is impossible to interpret the GRASP results. We are curious why the authors did not try to confirm the presence of a connection with P2X2, as this technique is used to make the case for the s-LNv to DN2 connection earlier in the paper. Existing sensors are quite likely to be sensitive enough to detect even an inhibitory or modulatory connection between the AC neurons and the s-LNvs.

We have tried to obtain evidence for a physiological connection between ACs and sLNvs. We focused on both the synaptic locus of sLNvs-ACs and the somas of sLNvs using a normal upright microscope. While GCaMP3.0 was expressed in sLNvs, P2X2 was expressed in ACs. We activated ACs by ATP or alternatively by warm temperature. However, we have not successfully detected any GCaMP fluorescence change in sLNvs.

The most challenging aspect was to find an optimal depth or location at the synaptic locus of sLNvs-ACs because there was no marker at the sLNvs-ACs synaptic locus. The sLNVs-ACs contacts were fewer and weaker than that of sLNvs-DN2s based on experiments relying on GRASP signals. Therefore, the G- CaMP3.0 fluorescence changes in the synaptic regions of sLNvs-ACs might be difficult to capture. We are currently setting up confocal microscopy with GaAsP which should allow us to capture weaker signals located at the deeper brain regions.

Given that serotonin signaling is important for the ACs-sLNvs communication (Figures 4E and F), driving the Epac-cAMP sensor in sLNvs might be an option for detecting a functional connectivity within the sLNvs-ACs synaptic locus.

Furthermore, to obtain a robust fluorescence change in both terminal regions and somas of sLNvs, other reporters such as Arclight or G-CaMP6.0 might be useful. Thus, further studies will be necessary for investigating the detailed properties of the functional connection between sLNvs and ACs.

Due to the lack of physiological evidence, the functional basis between ACs and sLNvs is still not clear. It is possible that information could be modified in both direction of the AC-sLNv connection and we do not exclude the possibility that sLNvs could modulate ACs' activity. That said, it is known that ACs are serotonergic and do not have dendrite projections in the dorsal protocerebrum. Given that 5HT1B-RNAI in sLNvs caused abnormal temperature preference before dawn, we prefer a model in which the information flows from ACs to sLNvs at this point.

In summary this is a promising and interesting study that would be significantly strengthened by an increase in technical rigor. Further, we recommend they add citation of a recent publication Barber et al., 2016, Genes & Dev, that supports their use of the pre-synaptic P2X2/post-synaptic GCaMP assay for functional connectivity.

We added the reference. Thank you very much for this suggestion.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

In particular, please see the concerns of reviewer 2, which needs to be well-addressed.

We have revised and tried not to overstate the explanation of the physiological connection between ACs and sLNvs.

Also, are the experiments that you suggest (re reviewer 1's) being done/done. It will not hurt to have them in place too.

Because we have been setting up the new system using different UAS-reporters and/or different microscopes, it may take more time to successfully detect the physiological signals between ACs and sLNvs. Therefore, we would like to pursue this question in the next report.