Voltage Imaging in Drosophila Using a Hybrid Chemical-Genetic Rhodamine Voltage Reporter

We combine a chemically-synthesized, voltage-sensitive fluorophore with a genetically encoded, self-labeling enzyme to enable voltage imaging in Drosophila melanogaster. Previously, we showed that a rhodamine voltage reporter (RhoVR) combined with the HaloTag self-labeling enzyme could be used to monitor membrane potential changes from mammalian neurons in culture and brain slice. Here, we apply this hybrid RhoVR-Halo approach in vivo to achieve selective neuron labeling in intact fly brains. We generate a Drosophila UAS-HaloTag reporter line in which the HaloTag enzyme is expressed on the surface of cells. We validate the voltage sensitivity of this new construct in cell culture before driving expression of HaloTag in specific brain neurons in flies. We show that selective labeling of synapses, cells, and brain regions can be achieved with RhoVR-Halo in either larval neuromuscular junction (NMJ) or in whole adult brains. Finally, we validate the voltage sensitivity of RhoVR-Halo in fly tissue via dual-electrode/imaging at the NMJ, show the efficacy of this approach for measuring synaptic excitatory post-synaptic potentials (EPSPs) in muscle cells, and perform voltage imaging of carbachol-evoked depolarization and osmolarity-evoked hyperpolarization in projection neurons and in interoceptive subesophageal zone neurons in fly brain explants following in vivo labeling. We envision the turn-on response to depolarizations, fast response kinetics, and two-photon compatibility of chemical indicators, coupled with the cellular and synaptic specificity of genetically-encoded enzymes, will make RhoVR-Halo a powerful complement to neurobiological imaging in Drosophila.

We combine a chemically-synthesized, voltage-sensitive fluorophore with a genetically encoded, self-labeling enzyme to enable voltage imaging in Drosophila melanogaster.
Previously, we showed that a rhodamine voltage reporter (RhoVR) combined with the HaloTag self-labeling enzyme could be used to monitor membrane potential changes from mammalian neurons in culture and brain slice. Here, we apply this hybrid RhoVR-Halo approach in vivo to achieve selective neuron labeling in intact fly brains. We generate a Drosophila UAS-HaloTag reporter line in which the HaloTag enzyme is expressed on the surface of cells. We validate the voltage sensitivity of this new construct in cell culture before driving expression of HaloTag in specific brain neurons in flies. We show that selective labeling of synapses, cells, and brain regions can be achieved with RhoVR-Halo in either larval neuromuscular junction (NMJ) or in whole adult brains. Finally, we validate the voltage sensitivity of RhoVR-Halo in fly tissue via dualelectrode/imaging at the NMJ, show the efficacy of this approach for measuring synaptic excitatory post-synaptic potentials (EPSPs) in muscle cells, and perform voltage imaging of carbachol-evoked depolarization and osmolarity-evoked hyperpolarization in projection neurons and in interoceptive subesophageal zone neurons in fly brain explants following in vivo labeling. We envision the turn-on response to depolarizations, fast response kinetics, and two-photon compatibility of chemical indicators, coupled with the cellular and synaptic specificity of genetically-encoded enzymes, will make RhoVR-Halo a powerful complement to neurobiological imaging in Drosophila.
-Chemical indicators possess fast response kinetics, turn-on responses to membrane depolarization, and can be compatible with two-photon excitation. -However, selective cell labeling in intact tissues and in vivo remains a challenge for completely synthetic fluorophores. -Here, we show that a chemical-genetic hybrid approach in Drosophila enables cellspecific staining in vivo and voltage imaging in whole-brain explants and at neuromuscular junction synapses.

INTRODUCTION
Voltage imaging in intact brains offers the tantalizing promise to watch, in real time, the electrical changes that underlie physiology. Approaches for voltage imaging rely on fluorescent indicators, either chemically synthesized, genetically encoded, or combinations of the two. Chemically-synthesized indicators have a storied past (Davila et al., 1973), enabling visualization of membrane potential changes in a diverse range of species and preparations (Grinvald et al., 1977(Grinvald et al., , 1981Cacciatore et al., 1999;Grinvald and Hildesheim, 2004). One serious drawback of chemically-synthesized indicators is their poor innate ability to localize to specific neurons. On the other hand, geneticallyencoded voltage indicators (GEVIs) circumvent problems of localization to specific neurons. Fresh rounds of innovation improved upon first-generation GEVIs (Storace et al., 2016;Kannan et al., 2019), enhancing membrane trafficking and response kinetics, and introducing alternative approaches to fluorescent protein/voltage-sensing domain fusions, including voltage-sensitive opsins and electrochromic-FRET pairs (Lin and Schnitzer, 2016;. Our group has focused on the development of chemicallysynthesized voltage-sensitive fluorophores that respond to changes in membrane potential via a photoinduced electron transfer (PeT) based mechanism. At hyperpolarizing potentials, the voltage across the membrane accelerates PeT from one side of the molecule to the other, short-circuiting and quenching fluorescence (Liu and Miller, 2020). At depolarized potentials, PeT is slowed, and the quantum yield of the dye increases. This configuration allows fast (Beier et al., 2019), linear, turn-on responses to depolarizations (with corresponding fluorescence decreases for hyperpolarization), good signal to noise, and compatibility with 2P excitation (Kulkarni et al., 2018;Kazemipour et al., 2019). However, attempts to deploy voltagesensitive fluorophores in brain tissues resulted in comprehensive staining of all neuronal membranes, making it difficult to visualize clear boundaries between cells or regions of the brain (Woodford et al., 2015;Kulkarni et al., 2018). Therefore, there is strong interest in developing hybrid systems in which voltagesensitive dyes are directed to cells of interest, either via expression of exogenous enzymes (Ng and Fromherz, 2011;Liu et al., 2017;Sundukova et al., 2019) or via targeting of native ligands (Fiala et al., 2020). Other strategies involve targeting synthetic fluorophores to genetically-encoded voltage-sensitive proteins, whether opsins Liu et al., 2021) or voltage-sensing domains (Deo et al., 2021).
We recently reported a chemical-genetic hybrid, in which a chemically-synthesized rhodamine-based voltage reporter (RhoVR) (Deal et al., 2016) attached to a flexible polyethyleneglycol (PEG) linker terminating with a chloroalkane forms a covalent bond with a cell-expressed HaloTag (Figure 1), enabling voltage imaging from defined neurons, in mouse cortical brain slices . This approach, RhoVR-Halo, takes advantage of the fast kinetics, linear turn-on response, and 2P compatibility of RhoVR-type indicators (Kulkarni et al., 2018;Kazemipour et al., 2019), and pairs it with the ability to target specific cells using traditional genetic methods. Other dyes, like tetramethylrhodamine (TMR)-Halo (Figure 1), can be used to provide a convenient, non-voltage sensitive stain for control experiments in the same genetic background.
The wealth of genetic tools, small brain size for optical imaging, and short generation time make Drosophila melanogaster an attractive model organism (Simpson, 2009;Caygill and Brand, 2016;Simpson and Looger, 2018). Genetically encoded indicators have been previously deployed in Drosophila and fall into two broad classes: (1) fluorescent protein (FP) fusions with voltage-sensing domains and (2) electrochromic FRET indicators (eFRET) that couple voltage-dependent changes in opsin absorbance with FRET to a fluorescent protein. FP-VSD fusions like ArcLight (Jin et al., 2012;Jourjine et al., 2016;Chen et al., 2017) or ASAP Chamberland et al., 2017), have been used in multiple Drosophila contexts and show negative-going responses to membrane depolarizations, use "GFP"-like excitation and emission, and display nonlinear response kinetics across the entire physiological range. Electrochromic-FRET indicators (Zou et al., 2014) like Ace2N-mNeon (Gong et al., 2015) or Varnam (Kannan et al., 2018) (and their chemigenetic relative, Voltron, which replaces the FP with a HaloTag)  have also been used in Drosophila and provide fast, negative-going responses to depolarizations. These types of indicators are not compatible with 2P excitation, likely owing to the complex photocycle involved in opsin-based voltage sensitivity (Maclaurin et al., 2013). Therefore, to expand the RhoVR-Halo methodology beyond vertebrate systems, we developed a stable transgenic UAS reporter line in Drosophila to express HaloTag on the extracellular surface of neurons, enabling the selective staining of defined neuronal populations when crossed with existing GAL4 driver lines. When paired with voltage-sensitive RhoVR-Halo , HaloTag-expressing flies allow cell type-specific labeling in vivo, and voltage imaging in a variety of contexts, including synaptic imaging at the neuromuscular junction (NMJ) and across defined neuronal populations in fly brain explants.

Generation of HaloTag Constructs for Expression in Flies
Although HaloTag and other self-labeling enzymes have been successfully expressed in transgenic flies, the reported lines localize HaloTag intracelluarly (Kohl et al., 2014;Sutcliffe et al., 2017;Meissner et al., 2018;Abdelfattah et al., 2019). Our first task was to generate a HaloTag that expressed on the extracellular face of membranes. Previous chemical-genetic hybrids deployed in mammalian cells used a transmembrane domain from the platelet-derived growth factor receptor (PDGFR) to localize HaloTag to the cell surface and a secretion signal peptide from immunoglobulin K (IgK) to enhance export of the construct . To adapt HaloTag-directed chemical-genetic hybrids for voltage imaging in Drosophila, we selected CD4 as a transmembrane anchor, on account of its good membrane association in Drosophila neurons (Han et al., 2011), fusing it to the C-terminus of the HaloTag. We sub-cloned this construct into different vectors for expression in mammalian (pcDNA3.1) and insect cells (pJFRC7) (Pfeiffer et al., 2010).
HaloTag-CD4 shows good expression on cell surfaces. In mammalian cells, CD4 localizes to the cell surface by anti-CD4 immunocytochemistry (Supplementary Figure 1). Inclusion of the self-labeling enzyme, HaloTag, affords the opportunity to confirm not only localization, but activity of the expressed enzyme by delivering HaloTag substrates. HEK cells expressing HaloTag-CD4 and treated with RhoVR-Halo (100 nM) show good membrane localization (Figure 2a), while cells that do not express HaloTag-CD4 show approximately 3.5-fold lower fluorescence levels (Figures 2b,c). RhoVR-Halo survives fixation: following live-cell imaging, cells can be fixed and retain their RhoVR-Halo staining, which serves as a useful counterstain to the anti-CD4 immunocytochemistry (Supplementary Figure 1). Live cells labeled with RhoVR-Halo and subsequently fixed, permeabilized with detergent and assayed for CD4 via immunochemistry reveal the majority of CD4 is found intracellularly. Confocal imaging shows that RhoVR-Halo localizes to cell membranes ( Supplementary Figures 2A-C). HaloTag-mediated labeling works with other dyes, too: HEK cells expressing HaloTag-CD4 and labeled with TMR-Halo show approximately 15-fold greater fluorescence than cells that do not express HaloTag-CD4 (Supplementary Figure 2).
In S2 cells, an immortalized Drosophila cell line, we also observe cell surface localization of HaloTag-CD4, as visualized by anti-CD4 immunocytochemistry (Supplementary Figure 3). S2 cells show similar HaloTag-CD4 dependent staining with TMR-Halo (100 nM, Figure 3 and Supplementary Figures 3A,B) with a 20-fold enhancement in fluorescence intensity in HaloTag-CD4 expressing cells compared to non-expressing cells (Figure 3c). TMR-Halo staining in S2 cells is also retained post-fixation (Supplementary Figure 3).
HaloTag remains functional when expressed on the cell surface of Drosophila neurons, enabling a range of brain regions and neurons to be labeled with small molecules. We delivered TMR-Halo (1 µM) to live flies via application of a solution of TMR-Halo in artificial hemolymph (AHL) to flies with their cuticle removed (Harris et al., 2015) (see Supplementary Material for dissection details). We then imaged via confocal microscopy to establish the extent of labeling (Figure 6). In GH146-GAL4 > HaloTag flies (PN labeling) treated with TMR-Halo, we observe strong fluorescence localized to the antennal lobe   be used to label single cells. VT011155-GAL4 > HaloTag-CD4 fly brains drive expression in single interoceptive subesophageal zone neurons (ISNs) (Jourjine et al., 2016), and treatment with TMR-Halo results in bright fluorescence localized to these neurons (Figure 6c). Similar staining profiles can be achieved with the voltage-sensitive RhoVR-Halo, which clearly labels PNs of the antennal lobe (Figure 6d, GH146-GAL4). High magnification examination of labeled projection neurons reveals membrane-localized staining (Figures 6e,g, red) alongside Hoechst 33342 nuclear staining (Figures 6f,h, blue). RhoVR-Halo can also label smaller sub-sets of neurons cells; treatment of Nan-GAL4 > HaloTag-CD4 brains with RhoVR-Halo results in labeling of ISNs (Figure 6i).
We used the same live-animal staining procedure to optimize the loading of RhoVR-Halo (Supplementary Figure 5). We find that 2 µM RhoVR-Halo provides good staining in the antenna lobes of GH146-GAL4 > HaloTag-CD4 crosses (Supplementary Figure 5). Fluorescence from RhoVR-Halo is localized to the periphery of cell bodies, again supporting the extracellular expression of HaloTag-CD4 (Figures 6e-h). Compared to regions of the brain that do not express HaloTag-CD4, RhoVR-Halo fluorescence is approximately three times higher (Supplementary Figures 5B,C). We find homozygous FIGURE 5 | HaloTag-CD4 expression in transgenic Drosophila. (a) nSyb-GAL4, HaloTag-CD4 brains express CD4 pan-neuronally. Maximum z-projection of a confocal fluorescence microscopy stack of brain explant from either (a-e) nSyb-GAL4 > HaloTag-CD4 or (f-l) GH146-GAL4 > HaloTag-CD4, fixed and stained for an extracellular epitope of the CD4 protein (OKT4, green) and counterstained for nuclei with Hoechst 33342 (16 µm or 10 µg/mL, magenta). Scale bar is 50 µm for whole-brain images (a-c,f-h) and 5 µm for zoomed-in regions (d,e,i-l). Insets on panels (a,f) show schematized brains with an approximate location of the staining for reference.  flies for GH146-GAL4 > HaloTag-CD4 have slightly higher fluorescence levels compared to levels of heterozygous flies, when stained with the same concentration of RhoVR-Halo (Supplementary Figures 5H-J). However, because the difference in fluorescence intensity in homozygous flies was not significantly larger than heterozygotes, we used heterozygous flies for subsequent experiments.

Functional Imaging
We established the voltage sensitivity of RhoVR-Halo in fly tissue expressing HaloTag-CD4 using two different approaches. First, we performed electrophysiology using dual two-electrode voltage-clamp combined with fluorescence imaging at the larval Drosophila neuromuscular junction (NMJ). We used the motor neuron driver OK6-GAL4 to drive pre-synaptic expression of HaloTag-CD4 (Figures 7a-f) or the muscle driver G14-GAL4 to express HaloTag-CD4 in the post-synaptic muscle (Figures 7gl). In live 3rd instar larval NMJ preparations, RhoVR-Halo (2 µM) clearly stains pre-synaptic neuronal compartments when HaloTag-CD4 expression is targeted in motor neurons (red, Figure 7d), co-localizing with the neuronal plasma membrane marker horseradish peroxidase (HRP, gray, Figures 7e,f). In a complementary fashion, when HaloTag-CD4 is expressed in post-synaptic muscle cells, RhoVR-Halo fluorescence (red, Figure 7j) accumulates at NMJs outside of the neuronal membrane outlined by HRP (gray, Figures 7k,l). RhoVR-Halo readily detects excitatory post-synaptic potentials (EPSPs) in muscle cells, confirmed by simultaneous optical imaging and sharp electrode recordings (Figures 7m-o). Importantly, we next used two-electrode voltage-clamp recordings in a semi-dissected larval preparation with muscle HaloTag-CD4 expression (G14-GAL4 > HaloTag-CD4, Figure 7p). This approach demonstrated that depolarizing potentials result in an increase in RhoVR-Halo signal (m6, Figures 7q-s) with an overall voltage sensitivity of approximately 12% F/F per 100 mV (± 0.2%, n = 8), in reasonably close agreement to the value determined in HEK293T cells (14%, Figure 4). Analysis of electrode (Figure 7r) and optical recordings (Figure 7s) show good correspondence. In contrast, no change in fluorescence signals was observed in an adjacent unclamped/unstimulated muscle cell (m7, Figure 7q, gray).
As a second confirmation of voltage sensitivity in fly tissues, we developed a stereotyped stimulation protocol for imaging in fly brain explants. We generated flies that express both HaloTag and the voltage-sensitive fluorescent protein, ArcLight, in PNs (GH146 GAL4, HaloTag/CyO; ArcLight/TM2) for use as an internal positive control. The use of RhoVR-Halo, with excitation and emission profiles in the green/orange region of the visible spectrum, allows for the simultaneous deployment of GFP-based indicators (Deal et al., 2016, like ArcLight (Jin et al., 2012). Drosophila antennal lobe projection neurons receive input from the olfactory receptor neurons (ORNs) in the antennae (Kazama and Wilson, 2008). As these projection neurons primarily receive cholinergic input from the ORNs (Restifo and White, 1990), and stimulated with 100 µM carbachol (red, n = 7 brains), stained with voltage-insensitive TMR-Halo (100 nM) and stimulated with 100 µM carbachol (gray, n = 7 brains), or stained with voltage-sensitive RhoVR-Halo (2 µM) and treated with vehicle control (blue, n = 6 brains). (d) Plots of individual F/F responses from RhoVR-Halo to carbachol stimuli (gray) and the average across all responses (red, SEM in light red). Traces of responses were aligned by peak response time and display 50 s before peak response and 150 s after peak response (gray). (e) Plots of average F/F traces for Drosophila brains stained with voltage-sensitive RhoVR-Halo (2 µM) and stimulated with 100 µM carbachol (red, n = 7 brains). ArcLight responses are recorded simultaneously (green, n = 7 brains). RhoVR traces are replicated from panel (c) for comparison with ArcLight. (f) Plots of average F/F traces for Drosophila brains stained with voltage-sensitive RhoVR-Halo (2 µM) and then treated with a vehicle control (red, n = 6 brains). ArcLight responses are recorded simultaneously (green, n = 6 brains). RhoVR traces are replicated from panel (c) for comparison with ArcLight. (g) Plots of average F/F traces for Drosophila brains stained with voltage-insensitive TMR-Halo (100 nM) and stimulated with 100 µM carbachol (blue, n = 6 brains). ArcLight responses are recorded simultaneously (green, n = 6 brains). TMR traces are replicated from panel (c) for comparison with ArcLight. For all plots, data are mean ± SEM for the indicated number of samples. Drosophila brain explants were stimulated three times for 30 s with either 100 µM carbachol or vehicle. Stimulus (delivery of carbachol or vehicle) is depicted by small black bars immediately below the traces).
we hypothesized that PNs could be readily stimulated with carbachol (CCH), a non-hydrolyzable acetylcholine mimic. We treated ArcLight/HaloTag-CD4 expressing fly brain explants with carbachol (100 µM) and observed robust fluorescence decreases timed to carbachol treatment, indicating a depolarizing membrane potential response to this neurotransmitter analog (Supplementary Figures 6A-D).
Using this robust stimulation protocol in fly brain explants, we next performed two-color voltage imaging with RhoVR-Halo and ArcLight. As before, we loaded RhoVR-Halo (2 µM) in live flies, removed the brains, and imaged the brain explants using epifluorescence microscopy. Excitation provided alternately with blue (475 nm) or green (542 nm) light to excite ArcLight or RhoVR-Halo, respectively, revealed robust fluorescence responses to carbachol (100 µM) treatment (Figure 8). RhoVR-Halo fluorescence increases with carbachol stimulation (Figures 8a-d), corresponding to membrane voltage depolarization and the turn-on response of RhoVR-type indicators (Deal et al., 2016. In contrast, ArcLight fluorescence decreases with carbachol stimulation (Figure 8e), showing a fluorescence decrease in response to depolarization, consistent with the turn-off response to depolarization for ArcLight indicators (Jin et al., 2012). Importantly, neither RhoVR-Halo nor ArcLight responds to a control experiment that omits carbachol from the perfusion solution (Figures 8c,f). Finally, the chemical-genetic hybrid approach of RhoVR-Halo enables additional controls to be carried out using the same transgenic flies. When HaloTag/ArcLight expressing flies are treated with TMR-Halo and then stimulated with carbachol, there is no response from the voltage-insensitive TMR-Halo (Figures 8c,g), but ArcLight still responds (Figure 8g). Using a "functionally dead" rhodamine dye in this experiment allows for control experiments to be run in the same transgenic animals as the experiments. Similar experiments with inactive mutants of genetically-encoded indicators/actuators (like ArcLight or GCaMP) would require the generation of separate transgenic animals with the inactivating mutation.
To evaluate the ability of RhoVR-Halo to report on physiological stimuli, we probed the response of RhoVR-Halo in ISNs, cells that respond dynamically to changes in osmolarity. Previous studies demonstrated that increases in osmolarity (240-440 mOsm) evoke hyperpolarizing responses in ISNs (Jourjine et al., 2016). Consistent with this, we find that ISNs expressing HaloTag-CD4 (Nanchung-GAL4) and labeled with RhoVR-Halo hyperpolarize upon an increase in osmolarity, as indicated by decreases in RhoVR fluorescence (Figures 9a-c).
In fly brains labeled with voltage-insensitive TMR-Halo, we observe no change in fluorescence (Figures 9a-c). In contrast, flies expressing ArcLight in ISNs show fluorescence increases in response to increased osmolarity (Supplementary Figure 7). Two-color voltage imaging alongside ArcLight in flies that express both HaloTag-CD4 and ArcLight in ISNs (Nanchung-GAL4, UAS-HaloTag-CD4/CyO; UAS-ArcLight/TM2) reveals osmolarity-induced decreases in RhoVR fluorescence coupled with increases in ArcLight fluorescence (Figures 9d-f), while control experiments at constant osmolarity show no responses in either ArcLight or RhoVR fluorescence (Figure 9g). Heterozygous flies expressing HaloTag in ISNs and labeled with RhoVR-Halo also respond to changes in osmolarity (Supplementary Figure 8). Taken together, these data establish the utility of RhoVR-Halo for monitoring sensory-induced changes to membrane potential.

DISCUSSION
In summary, we show that RhoVR-Halo indicators can be used for direct visualization of membrane potential changes in synapses and brains of flies. We show, for the first time, that RhoVR-Halo dyes can label specific neurons in vivo and that voltage changes can be visualized using epifluorescence microscopy at synapses in the NMJ and whole-brain explants. The hybrid chemical-genetic strategy employed here features a turn-on response to membrane depolarization and affords the opportunity to "plug-and-play" different fluorescent dyes to enable imaging in different colors (Ortiz et al., 2020) or to run critical control experiments using a non-voltage-sensitive fluorophore in the same genetic background (Figures 8g-i). We envision that RhoVR-Halos, with their high two-photon (2P) cross-section (93 GM at 840 nm, Supplementary Figure 9), can be combined with high-speed 2P imaging methods to provide fast voltage imaging in the brain.
Despite these advances, several drawbacks are associated with our implementation of the methodology at present. First, in the imaging data presented here, we do not take full advantage of the response kinetics of PeT-based indicators like RhoVR, which should have nanosecond responses times based on the mechanism of voltage sensing (Beier et al., 2019;Lazzari-Dean et al., 2019;Milosevic et al., 2020). Secondly, we do not take full advantage of the high 2P excitation cross-section of RhoVR dyes. Especially notable is the substantial cross-section at ∼1,030-1,040 13 nm (Supplementary Figure 9), which allows for the use of high-powered 2P illumination in emerging fast 2P methods (Kazemipour et al., 2019;Wu et al., 2020). Third, in fly brains, RhoVR-Halo voltage-sensitive indicators are not as bright as their fluorophore-only counterparts. This is likely a result of combinations of (a) lower intrinsic quantum yield for RhoVR-Halo compared to TMR-Halo (since the presence of a molecular wire quenches the fluorescence of the dye) and (b) lower solubility for the rather greasy RhoVR-Halo indicators compared to the smaller, more compact TMR-Halo dyes. The former can be addressed by using published methods to generate brighter fluorophores. The latter can be addressed by the use of new chemistries to attach HaloTag ligands, freeing up other sites for solubilizing groups to address the challenge of delivering fluorophores to intact brain preparations. Even with these limitations, we envision that chemical-genetic hybrids like RhoVR-Halo will be an important complement to the expanding set of methods for visualizing membrane potential changes in living systems.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

FUNDING
Research reported in this publication was supported in part by the National Institutes of Health S10 program under award number 1S10RR026866-01. We acknowledge support from NIH (R01NS098088, EM and R01NS111414, DD) and NSF (NeuroNex Innovation Award 1707350, EM and KS). MK was supported, in part, by a training grant from the NIH (T32GM007232). BB was supported, in part, by a training grant from the NIH (T32GM066698). MD acknowledges support from the NIH BRAIN Initiative (U24NS109107).