Visual Overview
Visual Abstract
Abstract
Zebrafish have gained prominence as a model organism in neuroscience over the past several decades, generating key insight into the development and functioning of the vertebrate brain. However, techniques for whole-brain mapping in adult stage zebrafish are lacking. Here, we describe a pipeline built using open-source tools for whole-brain activity mapping in adult zebrafish. Our pipeline combines advances in histology, microscopy, and machine learning to capture c-fos activity across the entirety of the brain. Following tissue clearing, whole-brain images are captured using light-sheet microscopy and registered to the recently created adult zebrafish brain atlas (AZBA) for automated segmentation. By way of example, we used our pipeline to measure brain activity after zebrafish were subject to the novel tank test, one of the most widely used behaviors in adult zebrafish. c-fos levels peaked 15 min following behavior and several regions, including those containing serotoninergic and dopaminergic neurons, were active during exploration. Finally, we generated a novel tank test functional brain network. This revealed that several regions of the subpallium form a cohesive subnetwork during exploration. Functional interconnections between the subpallium and other regions appear to be mediated primarily by ventral nucleus of the ventral telencephalon (Vv), the olfactory bulb, and the anterior part of the parvocellular preoptic nucleus (PPa). Taken together, our pipeline enables whole-brain activity mapping in adult zebrafish while providing insight into neural basis for the novel tank test.
- brain mapping
- exploratory behavior
- functional brain network
- functional connectome
- light-sheet microscopy
- image registration
Significance Statement
Zebrafish have grown in popularity as a model organism over the past several decades due to their low cost, ease of genetic manipulation, and similarity to other vertebrates like humans and rodents. However, to date, tools for whole-brain mapping in adult stage animals has been lacking. Here, we present an open-source pipeline for whole-brain mapping in adult zebrafish. We demonstrate the use of our pipeline by generating a functional brain network for one of the most widely used behavioral assays in adult zebrafish, the novel tank test. We found that exploration of a novel tank engages the olfactory bulb and a network of subpallial regions that correspond to the mammalian subpallial amygdala and basal ganglia.
Introduction
A fundamental goal of neuroscience is to understand how patterns of brain activity give rise to behavior. Identifying general principles of brain function is facilitated by cross-species comparisons. Over the past several decades, zebrafish have contributed to our understanding of the brain, a trend that promises to continue due to their low cost, ease of genetic manipulation, and sophisticated behavioral repertoire (Grunwald et al., 1988; Brockerhoff et al., 1995; Kenney, 2020; Loring et al., 2020; Burgess and Burton, 2023; Gerlai, 2023). Although several methods have been developed for whole-brain activity mapping in larval zebrafish (Ahrens et al., 2012; Portugues et al., 2014; Randlett et al., 2015; Shainer et al., 2023), equivalent approaches have yet to be developed for adult stage animals.
Adult and larval zebrafish each have distinct advantages and disadvantages in the study of brain–behavior relationships. Whereas larval animals are amenable to high-throughput work and live brain imaging due to their small size and transparency, adults have the advantage of mature neuroanatomy and more extensive behavioral repertoire. This adult behavioral repertoire includes a variety of social behaviors (Gerlai, 2014; Jones and Norton, 2015; Kareklas et al., 2023); short- and long-term associative, nonassociative, and spatial memories (Gerlai, 2020; Kenney, 2020); and different types of exploratory behaviors (Cachat et al., 2010; Toms and Echevarria, 2014; Rajput et al., 2022). Thus, to fully realize the utility of zebrafish as a model organism in neuroscience, methods for whole-brain mapping are also required for adult zebrafish.
Whole-brain activity mapping can yield unexpected insights into brain function that may be lost using more targeted methods. Measuring neural activity across the entire brain also facilitates the use of powerful analytic tools, like network analysis that captures complex interactions that are a hallmark of brain function and improves predictions of brain–behavior relationships (Bullmore and Sporns, 2009; Wheeler et al., 2013; Vetere et al., 2017). However, mapping whole-brain activity presents several technical challenges. One roadblock is that the brain of adult animals is not transparent and thus requires the use of tissue clearing (Richardson et al., 2021). Imaging intact organs presents another technical hurdle due to the increased volume, a challenge met by the recent development of light-sheet microscopy (Hillman et al., 2019). Finally, whole-brain mapping results in large amounts of data that cannot be analyzed via traditional approaches like manual cell counting and segmentation. This last challenge was met by combining advances in machine learning to automate cell detection (Tyson et al., 2021) and image registration (Gholipour et al., 2007) to the digital adult zebrafish brain atlas (AZBA; Kenney et al., 2021). Here, we describe our pipeline of open-source tools to enable whole-brain mapping in adult zebrafish and demonstrate an application by identifying a functional brain network that underlies the novel tank test.
Materials and Methods
Subjects and behavior
Zebrafish
Subjects were 8–10-month-old zebrafish of the TU strain from both sexes. Fish were bred and raised at Wayne State University and within two generations of animals obtained from the Zebrafish International Resource Center (ZIRC, catalog ID: ZL84) at the University of Oregon. Fish were maintained in high-density racks under standard conditions: water temperature of 27.5 ± 0.5°C, salinity of 500 ± 10 µS, and pH of 7.4 ± 0.2. Lighting followed a 14:10 h light/dark cycle, with lights on at 8:00 A.M. Fish were fed twice daily with dry feed (Gemma 300, Skretting) in the morning and brine shrimp (Artemia salina, Brine Shrimp Direct) in the afternoon.
Sex determination was based on secondary sex characteristics such as shape, color, and the presence of pectoral fin tubercles (McMillan et al., 2015). Confirmation was conducted postexperimentation by killing the animals and observing the presence or absence of eggs. All experimental procedures were conducted under the ethical approval of the Wayne State University Institutional Animal Care and Use Committee (Protocol ID: 21-02-3238).
Behavioral stimuli and tissue collection
The novel tank test was used as the behavioral stimulus, using tanks that were distinct from housing tanks. Behavioral tanks were open top five-sided (15 × 15 × 15 cm) and made from frosted acrylic (TAP Plastics). Each tank was filled to a height of 12 cm with 2.5 L of fish facility water and housed within a white corrugated plastic enclosure to minimize external disturbances and diffuse light.
One week before the novel tank test, animals were housed in 2 L tanks divided into two chambers with transparent dividers. Male and female pairs were kept in each chamber to enable identification of individuals without social isolation or tagging. A day prior to the experiment, animals were acclimatized to the behavior room for 1 h before being placed back on the housing racks. On the day of the experiment, animals were removed from the housing rack and allowed to acclimate in the behavioral room for 1 h. After acclimation, animals were individually transferred to a novel tank and allowed to explore the tank for 6 min. After 6 min, fish were removed and placed back in their home tank for a designated periods of time (5, 15, 30, 60, or 120 min) prior to killing. Water was replaced between animals. A subset of animals was killed 1 h after acclimation to the room (home tank control) and another set of animals were killed immediately after removal from the housing racks (rack control).
Animals were killed by immersion in ice-cold water for 5 min to induce anesthesia and then decapitated using a sharp blade. We found no difference in c-fos cell counts whether fish were immersed in cold water for 1 or 5 min (unpublished observations). Heads were then washed in ice-cold phosphate-buffered saline (PBS) for 60 s to allow for blood drainage and then fixed in 4% paraformaldehyde in PBS at 4°C overnight. Brains were then dissected in ice-cold PBS and subject to iDISCO and in situ HCR.
Histology
Tissue pretreatment
We adapted the iDISCO protocol (Renier et al., 2016) for zebrafish brain tissue staining. Following dissection, brain samples were washed for 30 min, three times, in PBS at room temperature. This was followed by dehydration using a methanol concentration gradient (20, 40, 60, 80, and 100%) for 30 min each at room temperature. Samples were further washed in 100% methanol, chilled on ice, and then incubated in chilled 5% hydrogen peroxide in methanol overnight at 4°C. The next day, the samples were rehydrated through a reverse methanol series (80, 60, 40, and 20%) at room temperature, followed by a 1 h PBS wash, two 1 h PBS-T washes (1× PBS, 0.1% Tween 20), and a 3 h PBS-T wash. Samples were then equilibrated overnight in 5× SSCT (sodium chloride sodium citrate/0.1% Tween-20) buffer.
In situ HCR and tissue clearing
We modified the original HCR method described by Choi et al. (2018) and informed by the work of Kumar et al. (2021). These modifications include the initial acetylation step to reduce background and reduced volumes and decreased volume of HCR amplification buffer to reduce waste and increased concentration of hairpins to increase the signal strength. After dissection, samples were acetylated in 0.25% v/v acetic anhydride solution in ultrapure water for 30 min. Samples were then washed in ultrapure water three times for 5 min and then equilibrated in probe hybridization buffer (30% formamide, 5× SSC, 9 mM citric acid, 0.1% Tween-20, 50 µg/ml heparin, 1× Denhardt's solution, 10% dextran sulfate) for 15 min at room temperature. Samples were then incubated in probe hybridization buffer for 1 h at 37°C with shaking and then incubated with 1 µM of c-fos probes (Molecular Instruments) in hybridization buffer at 37°C with shaking for 48–60 h. Samples were then washed with probe wash buffer (30% formamide, 5× SSCT, 9 mM citric acid, 50 µg/ml heparin) three times at 37°C and then twice with 5× SSCT for 1 h each with shaking. The tissue was then equilibrated in amplification buffer (5× SSC, 0.1% Tween-20, 10% dextran sulfate) at room temperature for 1 h with shaking. Alexa647 labeled hairpins (B1; Molecular Instruments) were prepared by heating to 95°C for 90 s prior to cooling at room temperature in the dark. We diluted 7.5 pmol of each hairpin into 125 µl of amplification buffer for each sample. Samples were incubated for 48–60 h in the dark at room temperature. Finally, tissue was washed in 5× SSCT, five times for 1 h each before being washed overnight in 5× SSCT.
Following the last 5× SSCT wash, samples were dehydrated in a series of methanol–water mixtures (20, 40, 60, 80, 100%) at room temperature for 1 h each and then left in 100% methanol overnight. The next day, samples were incubated at room temperature in a mixture of 66% dichloromethane and 33% methanol for 3 h followed by two 15 min washes in dichloromethane. After removing the dichloromethane, samples were incubated and stored in dibenzyl ether at room temperature for at least 24 h until imaging.
Brain imaging and processing
Cleared samples were imaged on a LaVision BioTec UltraMicroscope II (Miltenyi Biotec) using Imspector software for image acquisition. The microscope setup included a 4.2 megapixel sCMOS camera and a 2× objective lens with a dipping cap with spherical aberration correction. Images were taken at a magnification of 6.4×. Samples were mounted on the sample holder using an ultraviolet cured resin (NOA 61, Norland Products) with a refractive index (1.56) that matched DBE (Extended Data Fig. 2-1). Imaging was conducted from the right laser with a 4 μm step size using dynamic horizontal focus. Both 480 nm autofluorescence and 640 nm signal channels were used. The imaging settings used were the following: 90 and 20% laser power for the 640 and 480 nm lasers, respectively, 200 ms exposure time, 50% sheet width, and a sheet numerical aperture of 0.156. Acquired images were stitched using TeraStitcher (Bria and Iannello, 2012).
Computational analysis
Automated cell detection
We used CellFinder (Tyson et al., 2021) for the automated detection and quantification of c-fos-positive cells. It comprises two steps: cell candidate detection and cell classification. The initial step of cell detection identifies cell-like objects in the image. We optimized parameters to capture as many cell-like objects in our images as possible. Running from the Linux terminal, we used the following command for cell detection:
cellfinder -s path/to/folder/signal/channel/c-fos -b /path/to/folder/background/channel/AF -o path/to/output1 -v 3.990 0.943 0.943 –orientation sal –no-register –no-classification –soma-diameter 5 –threshold 3 –ball-xy-size 2 –ball-z-size 7 –ball-overlap-fraction 0.77 –log-sigma-size 0.1 –save-csv –batch-size 64 –epochs 100
After detecting cell candidates, a customized Python script was used to remove cell candidates that were within 9 μm of one another. This last step was done because the cell detection algorithm tended to double count cells; altering parameters to remove the double counting resulted in unacceptable loss of cell detection in other parts of the brain. We chose the 9 μm empirically by exploring several distances, finding that this distance resulted in the removal of overlapping candidates without unduly affecting the detection of distinct cells in close proximity.
Napari was used for visualization and labeling. We manually annotated 10,597 cells and 7,303 noncells across five brains for training the artificial neural network. CellFinder was trained using the following command:
Cellfinder_train -y path/to/brain1_labels.yml path/to/brain2_labels.yml path/to/brain3_labels.yml path/to/brain4_labels.yml path/to/brain5_labels.yml -o /trained_network –batch-size 64 –epochs 100 –no-save-checkpoints –save-progress
The trained network achieved 96.1% accuracy. Finally, the trained network was applied to all the experimental brains to classify the detected cell candidates into cells and noncells. This was achieved by the following command:
cellfinder -s /path/to/folder/signal/channel/c-fos/ -b /path/to/folder/background/channel/AF/ -o path/to/output -v 3.990 0.943 0.943 –orientation sal –no-register –no-detection –soma-diameter 5 –threshold 3 –ball-xy-size 2 –ball-z-size 7 –ball-overlap-fraction 0.77 –log-sigma-size 0.1 –save-csv –trained-model /trained_network/model.h5
Differentiating puncta and diffuse patterns of c-fos staining
To differentiate between punctate and diffuse patterns of c-fos staining, we developed a convolutional neural network (CNN) built in Python using the TensorFlow library. The architecture of the CNN is outlined in Extended Data Table 4-1. c-fos images from 10 brains were labeled, totaling 2,448 puncta and 1,916 cytoplasmic labels. A training dataset was created by isolating 11 × 11 × 11 pixel cubes centered around each of the labeled cells. The dataset was split 80/20 into a training set and a testing set. The input data was augmented through a series of horizontal and vertical flips, 90° rotations, and 2 pixel horizontal translations to create a total training dataset of 13,706 puncta and 10,724 diffuse labels. No data augmentation was performed on the testing set. The model was trained using an NVIDIA GeForce 3090 GPU for 500 epochs. The batch size was 32, the weight decay rate was 0.0005, and the learning rate was 0.0001. The model achieved an accuracy of 95.3% on the testing set.
Brain registration
Image registration was performed using ANTs (Avants et al., 2009). For the nonlinear diffeomorphic step, four parameters were optimized: cross-correlation, gradient step, update field variance in voxel space, and total field variance in voxel space to achieve the best alignment. We chose the parameters based on a combination of qualitative and quantitative measures. Using mutual information (MI; using the measureImageSimilarity function in ANTs with 32 bins) and a binarized segmentation mask, we quantified how well we were able to register the AZBA autofluorescence image to our autofluorescence template (Extended Data Fig. 2-2A). We found that most parameters did not make a large difference in MI except total field variance where a value of 0 was superior to 0.5. We got a similar result when we measured the average MI for AZBA registered to individual brains for the 15 min c-fos group (Extended Data Fig. 2-2B). Thus, we qualitatively determined the best parameters by examining broad alignment of regions like the optic tectum, mamillary bodies, cerebellum, and anterior commissure. These parameters were cross-correlation of 3, gradient step of 0.3, update field variance of 4, and a total field variance of 0. Using these optimized parameters, brain registration was carried out in two steps: first, an average brain template was created, and second, AZBA was registered to this average template.
Before registration, images were downsampled to 4 μm isotropic using brainreg from the BrainGlobe suite of tools (Tyson et al., 2021):
brainreg /path/to/raw/data /path/to/output/directory -v 3.990 0.943 0.943 –orientation sal –atlas azba_zfish_4µm –debug
The average template was generated using 10 autofluorescence images. Initially, nine autofluorescence images were individually brought into the space a single image (template) using the following ANTs command:
antsRegistration –dimensionality 3 –float 1 -o [${AF_sample_1_for_avg_},${ AF_sample_1_for_avg-warped}] –interpolation WelchWindowedSinc -u 0 -r [${AF_template.nii},${AF_sample_1.nii},1] -t Rigid[0.1] -m MI[${AF_template.nii},${AF_sample_1.nii},1,32,Regular,0.25] -c [200 x 200 x 200 x 0,1e-8,10] –shrink-factors 12x8x4x2 –smoothing-sigmas 4x3x2x1vox -t Affine[0.1] -m MI[${AF_template.nii},${AF_sample_1.nii}, 1,32,Regular,0.25] -c [200 x 200 x 200 x 0,1e-8,10] –shrink-factors 12x8x4x2 –smoothing-sigmas 4x3x2x1vox -t SyN[0.3,4,0] -m CC[${AF_template.nii},${AF_sample_1.nii}, 1,3] -c [200 x 200 x 200 x 200, 1e-6,10] –shrink-factors 12x8x4x2 –smoothing-sigmas 4x3x2x1vox –verbose 1
These outputs were then used to create an average image using the “AverageImages” command in ANTs. Next, the autofluorescence image from AZBA was registered to the average template using the following command:
antsRegistration –dimensionality 3 –float 1 -o [${AZBA_to_avg_temp_},${AZBA_to_avg_temp-warped}] –interpolation WelchWindowedSinc -u 0 -r [${avg_template.nii.gz},${AZBA/20180628_AF_average.nii.gz},1] -t Rigid[0.1] -m MI[${avg_template.nii.gz},${AZBA/20180628_AF_average.nii.gz},1,32,Regular,0.25] -c [200 x 200 x 200 x 0,1e-8,10] –shrink-factors 12x8x4x2 –smoothing-sigmas 4x3x2x1vox -t Affine[0.1] -m MI[${avg_template.nii.gz},${AZBA/20180628_AF_average.nii.gz}, 1,32,Regular,0.25] -c [200 x 200 x 200 x 0,1e-8,10] –shrink-factors 12x8x4x2–smoothing-sigmas 4x3x2x1vox -t SyN[0.3,4,0] -m CC[${avg_template.nii.gz},${AZBA/20180628_AF_average.nii.gz}, 1,3] -c [200 x 200 x 200 x 200, 1e-6,10] –shrink-factors 12x8x4x2 –smoothing-sigmas 4x3x2x1vox –verbose 1
To bring the segmentation from AZBA into the space of the template, we used the following command:
antsApplyTransforms -d 3 –float -n NearestNeighbor -i /AZBA/2021-08-22_AZBA_segmentation.nii.gz -r avg_template.nii.gz -o AZBA_to_avg_temp_transformed.nii.gz -t AZBA_to_avg_temp_1Warp.nii.gz -t AZBA_to_avg_temp_0GenericAffine.mat
The newly generated average template image was used as a reference image and was registered onto individual autofluorescence images:
antsRegistration –dimensionality 3 –float 1 -o [${AF_sample_},${AF_sample-warped}] –interpolation WelchWindowedSinc -u 0 -r [${AF_sample.nii},${avg_template.nii.gz },1] -t Rigid[0.1] -m MI[${AF_sample.nii},${ avg_template.nii.gz },1,32,Regular,0.25] -c [200 x 200 x 200 x 0,1e-8,10] –shrink-factors 12x8x4x2 –smoothing-sigmas 4x3x2x1vox -t Affine[0.1] -m MI[${AF_sample.nii},${ avg_template.nii.gz },1,32,Regular,0.25] -c [200 x 200 x 200 x 0,1e-8,10] –shrink-factors 12x8x4x –smoothing-sigmas 4x3x2x1vox -t SyN[0.3,4,0] -m CC[${AF_sample.nii},${ avg_template.nii.gz },1,3] -c [200 x 200 x 200 x 200, 1e-6,10] –shrink-factors 12x8x4x2–smoothing-sigmas 4x3x2x1vox –verbose 1
Finally, segmentation of individual brains was done using the same command as above but applied to the segmentation file as the floating image.
C-fos cell counts and network analysis
R (version 4.1.1; R Core Team, 2016) was used for network analysis and to integrate the output from CellFinder with the brain segmentation using the RNifti package (Clayden et al., 2021) to read in the segmentation files. The number c-fos-positive cells in each brain were summed excluding white matter and clear labeled regions yielding 143 gray matter regions for analysis. The olfactory bulbs of two brains in the 15 min group did not register properly, so these regions from these two brains were removed from the analysis.
Network analysis was performed using the igraph (version 2.0.2) package (Csardi and Nepusz, 2006). The network was generated by treating the correlation matrix (Fig. 6) as an adjacency matrix. For thresholding, we chose the network density using efficiency cost optimization to maximize the quality function (Fallani et al., 2017):
For identifying node roles, we first calculated community structure using the Louvain algorithm (Blondel et al., 2008), and then the within module degree z-score:
The small worldness parameter was calculated as described in Humphries and Gurney (2008):
Statistical analysis
Statistical analysis was done using R. Data were analyzed using 2 × 2 ANOVAs as indicated in the results. For the overall time course c-fos data, Dunnett's t tests were used to compare all other groups to the home tank control group (time = 0). False discovery rate (FDR; Benjamini and Hochberg, 1995) corrected paired t tests at each time point were used for diffuse versus punctate data. For comparing c-fos counts across regions between the home tank and 15 min time point, independent samples t tests were used with p values were corrected using the FDR.
Results
Overview of strategy
We begin by giving an overview of our strategy for whole-brain activity mapping (Fig. 1) before describing the results of each step in more detail. Following behavior, animals are killed and head fixed in 4% paraformaldehyde overnight. After careful dissection, brains are rendered optically transparent using iDISCO (Renier et al., 2016), which we modified to make compatible with in situ hybridization chain reaction (HCR) for the detection of c-fos mRNA (Choi et al., 2018; Kramer et al., 2018; Kumar et al., 2021). Imaging intact cleared brain tissue was done using light-sheet microscopy. To automatically identify c-fos-positive cells in the brain, we used the open-source CellFinder package (Tyson et al., 2021) that is part of the BrainGlobe suite of Python-based software tools (Claudi et al., 2020). Finally, to automatically parcellate the brain into individual regions, we used advanced normalization tools (ANTs; Avants et al., 2009) to register autofluorescence images to AZBA (Kenney et al., 2021). The final output of our pipeline is a list of c-fos-positive cell counts for each brain region and each animal. This enables the use of a variety of downstream analytic tools; one example we demonstrate here is functional network analysis.
Overview of pipeline for mapping neural activity in adult zebrafish. Following behavior, zebrafish are killed and brains carefully removed. In situ HCR is then used to label c-fos. Brains are then cleared using iDISCO and imaged via light-sheet microscopy. Cells are then detected using CellFinder and brains are registered to AZBA. Regional c-fos counts are then used to generate brain networks for further analysis. A bench protocol for these methods are available on Dryad; see the code availability statement for a link.
Automated cell detection
After in situ HCR, tissue was cleared using iDISCO, which allowed us to use light-sheet microscopy to capture whole-brain images in both the c-fos (Fig. 2A, top) and autofluorescence channels (Fig. 2A, bottom). Detection of c-fos-positive cells was done using CellFinder (Tyson et al., 2021), an artificial neural net-based supervised machine learning algorithm. The first step in the cell detection process uses image filtering to detect cell shaped objects in the c-fos image. We found parameters that captured c-fos-positive cells throughout the entire brain (described in the Materials and Methods section), including areas with cells of different sizes and densities like the telencephalon (Fig. 2B) and cerebellum (Fig. 2C). Because the cell detection algorithm generated many overlapping cells, we used a custom-written Python script to remove cell candidates that were within 9 μm of one another. We then trained a CellFinder artificial neural network by manually labeling 10,597 cells and 7,303 noncells across five brains. Noncells were unambiguously identified by the presence of a signal in both the c-fos and autofluorescence channels, suggesting the presence of background bleeding into the c-fos channel. Cells only appeared in the c-fos channel. The resulting network achieved over 95% accuracy where the cells and noncells were clearly differentiated across different brain regions (Fig. 2B,C).
Staining for c-fos and identifying c-fos-positive cells. A, Adult zebrafish brain stained for c-fos (top) and the corresponding autofluorescence image (bottom). Scale bar, 0.5 mm. To see how brains are mounted for imaging, see Extended Data Figure 2-1. B, C, Zoomed in sections of the brain corresponding to red squares in part A showing c-fos staining and autofluorescence with labeling of cells (yellow arrows) and noncells (pink triangles). Scale bars, 0.1 mm. D, Examples of c-fos staining characterized as punctate or diffuse. Scale bar, 10 μm.
Figure 2-1
Mounting of the zebrafish brain for light sheet imaging. A) Ultraviolet cured resin is first used to form a hemisphere to raise the brain above the platform. B) Zebrafish brain mounted on top of the resin hemisphere. The brain is illuminated with ultraviolet light to make it visible. C) Sample holder where the platform is mounted before being placed in the imaging chamber. D) Sample holder in the imaging chamber of the Ultramicroscope II. Download Figure 2-1, TIF file.
Figure 2-2
Download Figure 2-1, TIF file.
During imaging, we noticed that we had sufficient resolution to differentiate distinct patterns of c-fos staining: punctate and diffuse, which likely represent nuclear and cytoplasmic staining, respectively (Fig. 2D). This localization of c-fos may be an indication of how long ago the cell was active as the mRNA is first transcribed in the nucleus before being shuttled to the cytoplasm for translation. To capture these distinct patterns, we created and trained an artificial neural net on 2,448 examples of punctate and 1,916 examples of diffuse staining. This network achieved >95% accuracy.
Registration to the adult zebrafish brain atlas
The adult zebrafish brain contains over 200 regions, making manual segmentation implausible. To automate parcellation of brains into individual regions, we used ANTs (Avants et al., 2009) to register brains to AZBA making use of common autofluorescence images in the atlas and present study. Initially, we attempted to register the autofluorescence image in AZBA directly to individual autofluorescence images, but the results were inconsistent (data not shown). We had more success using an average template that was generated by registering together 10 autofluorescence images from present study (Fig. 3A, top). The autofluorescence image from AZBA was then successfully registered to this template brain (Fig. 3A, segmentation overlay on bottom). A handful of small anomalies arose from this registration process that we manually fixed using ITK-SNAP (Yushkevich et al., 2019). These arose in parts of the image that tend to be highly variable between individuals, such as where mounting occurs at the ventral hypothalamus and the dorsal sac that extends from the diencephalon. To segment individual brains, we used the transforms from registering the template average autofluorescence brain to individual images (Fig. 3B). Using inverse transformations from the registration process, we also brought c-fos images into the space of AZBA (Fig. 3C). Finally, to demonstrate the quality of our registration, we juxtaposed segmentation from individual and average autofluorescence brains from the present study with the autofluorescence and nuclear stain (TO-PRO) in AZBA (Fig. 3D). We included the AZBA nuclear stain because it was the primary image used for segmentation due to its similarity to the cresyl violet stain of the original atlas (Wulliman et al., 1996), and it more clearly delineates regions than the autofluorescence images. Registration was able to successfully segment white matter tracts (e.g., the anterior commissure; Fig. 3D, second row); large areas like the cerebellum, optic tectum, and the periventricular gray zone (Fig. 3D, bottom row); and numerous telencephalic regions and midbrain nuclei (Fig. 3D, first, second, and third rows).
Registration of brain images to AZBA. A, Average image of 10 brains from the present study registered together (top) with segmentation overlay from AZBA (bottom). B, Segmentation from AZBA applied to an individual zebrafish brain. C, An individual c-fos brain brought into the space of AZBA with and without segmentation. D, Overlays of segmentation from AZBA in autofluorescence images from an individual brain, the average brain template from the present study, AZBA, and the nuclear TO-PRO stain in AZBA. A quantitative assessment of registration quality can be found in Extended Data Figure 3-1. The numbers on the left indicate distance from anterior most portion of the brain in mm. Scale bars, 0.5 mm.
Figure 3-1
Quantitative assessment of image registration. A) Mutual information of the autofluorescence image from AZBA with the average template in the present study. B) Normalized mutual information of the autofluorescence image from AZBA registered to individual brains from the 15-minute cfos group. Error bars represent standard error of the mean; n=13. Download Figure 3-1, TIF file.
Time course for c-fos expression
To effectively map whole-brain activity, we need to know at what point after behavior c-fos expression peaks. We exposed fish of both sexes to a commonly used behavioral task, the novel tank test, and killed animals 5, 15, 30, 60, or 120 min following the behavior (Fig. 4A). We also had two control groups: (1) fish that were killed immediately after removal from the housing racks and (2) fish that were brought into the behavioral room and killed an hour later, mimicking the habituation to the room we use for fish that were exposed to the novel tank (i.e., time = 0). From the c-fos average images at each time point, we observed a large increase in staining between 5 and 30 min (Fig. 4B). We also quantified the number of c-fos-positive cells using CellFinder and used a 2 × 2 (sex × time) ANOVA for analysis (Fig. 4C). We found a large effect of time (p < 0.001, η2 = 0.54), a trend toward a small effect of sex (p = 0.07, η2 = 0.059) and no interaction (p = 0.46). The small, but nonsignificant, effect of sex was in the direction of females having slightly higher c-fos counts than males. Using a Dunnett's t test, we compared all groups with the home tank (HT) control group and found a large increase in c-fos cell density at 15 min (p = 0.00067, d = 2.07). There were no differences at any other time point (p's > 0.05; Fig. 4C).
Time course for c-fos expression following exploration of a novel tank. A, Experimental design for capturing the time course for c-fos expression. B, c-fos stained brains from each time point were brought into the space of AZBA and averaged and displayed in the coronal plane. The numbers on the left of image are the distance (in mm) from the anterior most portion of the brain. Scale bar, 0.5 mm. C, c-fos cell density from animals taken off the rack, that remained in their home tank (HT), or at different times after exploration of the novel tank (5, 15, 30, 60, or 120 min). *p < 0.05 compared with the HT group using Dunnett's t test. Only comparisons with an asterisk were significant. Error bars are standard error of the mean. D, Number of c-fos cells classified as punctate or diffuse at each time point. See Extended Data Table 4-1 for the structure of the artificial neural net used for classification. *p < 0.05 difference between the number of puncta versus diffuse stained cells within time point using FDR-corrected t tests. Error bars are standard error of the mean. Sample sizes were as follows: rack: female, n = 4, male, n = 4; HT: female, n = 5, male, n = 4; 5 min: female, n = 6, male, n = 5; 15 min: female, n = 7, male, n = 6; 30 min: female, n = 5, male, n = 5; 60 min: female, n = 4, male, n = 4; 120 min: female, n = 5, male, n = 3.
Table 4-1
Architecture of CNN for detecting cytoplasmic and nuclear cfos stained cells. Download Table 4-1, DOCX file.
We also examined how the proportion of punctate and diffusely stained cells changed across time (Fig. 4D). A cell type × time ANOVA found a large main effect of time (p < 0.001, η2 = 0.40) and no overall effect of cell type (p = 0.95). There was also a large interaction between cell type and time (p = 0.0082, η2 = 0.13). FDR-corrected paired t tests at each time point found that there were more punctate than diffuse stained cells at 5 min (p = 0.048). This trend switched to more diffuse than punctate stained cells at 15 and 30 min, although the differences at these time points were not statistically significant (p's = 0.16 and 0.22, respectively).
Regions and cell types active during the novel tank test
Comparing c-fos counts between brains of the home tank and 15 min groups identified 46 regions that had elevated c-fos following exposure to the novel tank (Extended Data Table 5-2 with a subset labeled in Fig. 5A). In the telencephalon, regions with elevated activity included parts of both the pallium [lateral (Dl) and central zones (Dc), the anterior part of the nucleus of the olfactory tract (nLOT-a), and the medial division of the bed nucleus of the stria terminalis (BSTm)] and subpallium [lateral (Vl) and central nuclei (Vc)]. Several thalamic nuclei were also upregulated, such as the anterior (A) and intermediate (I) thalamic nuclei suggesting strong engagement of sensory systems during exploration. The hypothalamus, which is central to endocrine regulation, had increased c-fos in the caudal (Hc) and dorsal (Hd) zones as well as the lateral hypothalamus. Finally, several regions of the cerebellum also had high levels of c-fos during exploration of the novel tank, such as the granular and molecular layers of the cerebellar corpus (CCe-g/m) and the granular eminence (EG).
Regions and cell type active during the novel tank test. A, Average c-fos staining from home tank and brains removed 15 min after the novel tank test with a segmentation overlay. Regions indicated are significantly increased compared with the home tank group (FDR corrected p < 0.05). Regional abbreviations can be found in Extended Data Table 5-1 and full statistical results in Extended Data Table 5-2. B–D, Overlap between c-fos expression and neurotransmitter-related stains in AZBA. Regional overlap for (B) 5-HT, (C) TH, and (D) ChAT. Scale bars, 0.5 mm. Numbers on the bottom or left are distance from anterior most portion of the brain in millimeter.
Table 5-1
List of abbreviations for grey matter regions. Download Table 5-1, DOCX file.
Table 5-2
Comparison of cfos counts in hometank control and animals euthanized 15 minutes after the novel tank test. Download Table 5-2, CSV file.
AZBA contains several stains that can be used to identify different cell types across brain regions such as 5-hydroxytryptamine (5-HT), tyrosine hydroxylase (TH), and choline acetyltransferase (ChAT; Kenney et al., 2021). To determine if exposure to a novel tank results in the activation of regions containing these neuronal cell types, we looked for overlap between the stains in AZBA and elevated c-fos (Fig. 5B–D). For regions expressing 5-HT (Fig. 5B), we saw an increase in c-fos in the paraventricular organ (PVO), intermediate nucleus (IN), and caudal zone of the periventricular hypothalamus (Hc). For TH, we saw overlap in the ventromedial thalamic nucleus (VM), the posterior part of the parvocellular preoptic nucleus (PPp), paracommissural nucleus (PCN), and Hc (Fig. 5C), although only the elevation of c-fos in PCN and Hc was statistically significant (Extended Data Table 5-2). Finally, for ChAT, we saw overlap in the paraventricular gray zone of the optic tectum (PGZ; Fig. 5D), however, this elevation did not reach significance. One reason we may see differences in c-fos average images that did not reach statistical significance is the lack of power; with our current number of observations (n's = 9–13), we are only able to detect very large effect sizes (cohen's D > ∼1). Although we can see an overlap at the regional level, our findings are only tentative because the c-fos and antibody-stained images come from separate brains, so we cannot make claims at the cellular level. Nonetheless, this demonstrates how our approach can be used to generate hypotheses about roles different neurotransmitters may play in an underlying a behavior.
Brain network analysis
We used functional network analysis to gain insight into the organization of brain activity that underlies exploration of a novel tank (Wheeler et al., 2013; Vetere et al., 2017; Pinho et al., 2023). Using c-fos counts from the 15 min time point, we computed the correlated activity between all 143 gray matter regions across animals (Fig. 6). To filter the correlation matrix to generate a network, we used efficiency cost optimization where the network density is chosen such that it balances the inclusion of edges to increase global and local efficiency against the putative cost of including additional connections (Fallani et al., 2017). We found a density of 2.5% maximized the efficiency cost optimization quality function (Fig. 7A). This resulted in a network with 256 edges and an average degree of 3.6, which is consistent with functional brain networks from other species using different imaging modalities (Fallani et al., 2017). This network also exhibited small world properties: its average shortest path length between nodes was 5.3, similar to the average path length from equivalently dense random networks (3.9). Our network also had much higher clustering than random networks (0.42 vs 0.025). This yielded a small world coefficient >1 (12.6) indicating the expected small world property (Humphries and Gurney, 2008). We also computed degree and eigenvector centrality for each node to uncover brain regions that may play outsized roles in the network, uncovering several regions that were in the top 15 for both measures (Fig. 7C). These included several telencephalic regions, particularly from the subpallium, such as the ventral nucleus of the ventral telencephalon (Vv), the dorsal and ventral zones of the ventral telencephalon (Vd-dd and Vd-vd), and the dorsal most zone of the ventral telencephalon (Vdd). A handful of regions outside the telencephalon were also high in centrality, such as the intermediate thalamic nucleus (I), the anterior part of the parvocellular preoptic nucleus (PPa), the mesencephalic trigeminal nucleus (Vmn), and two regions in the olfactory bulb [external (ECL) and internal (ICL) cellular layers].
Correlation matrix of c-fos activity across the zebrafish brain. Entries in the matrix are Pearson’s correlations between brain regions across animals killed 15 min after the novel tank test. Regions are organized based on common ontological levels. Regional abbreviations and ontological levels can be found in Extended Data Table 5-1.
Analysis of the functional brain network active during the novel tank test. A, Efficiency–cost optimization for different network densities. J, quality function (see Materials and Methods); El, local efficiency; EG, global efficiency. B, Network filtered at a density of 2.5%. Connections between nodes represent suprathreshold correlations from Figure 6. Color of connections represents the strength (darker means higher absolute value) and direction (red, positive; blue, negative) of the correlation. Node colors correspond to communities. Regions not in the giant component are not shown. C, Degree and eigenvector centrality for the top 15 brain regions. Gray bars are those regions that are in the top 15 for both degree and eigenvector centrality. D, Identification of the role that each node plays in the network based on within module degree z-score and participation coefficient.
To identify the community structure of the network, we used the Louvain algorithm (Blondel et al., 2008), which identified nine clusters (Fig. 7B). Using this community structure, we categorized the roles that different nodes play in interconnecting different parts of the network (Guimerà and Amaral, 2005): provincial hubs (highly connected within its community, but not between communities), connector hubs (highly connected both within and between communities), peripheral nodes (low connectivity within and between communities), and nonhub connectors (low connectivity within a community, but high between communities). The Vv, Vmn, and PPa, which were identified as important based on centrality measures, are connector hubs (or nearly so for the PPa). The PPa and Vv interconnect the module dominated by regions of the ventral telencephalon with other parts of the preoptic area (SC and PPp), thalamus (VM, CP, and ZL), and hypothalamus [ATN (anterior tuberal nucleus), Hv (ventral zone of the periventricular hypothalamus), Hc, and Hd]. Interestingly, Vmn has a lot of inverse connections with parts of the telencephalon and provides a link to a cluster dominated by pretectal regions like the DAO (dorsal accessory optic nucleus), PSm (magnocellular superficial pretectal nucleus), CPN (central pretectal nucleus), PO (posterior pretectal nucleus), and PPd (dorsal part of the periventricular pretectal nucleus). Finally, two olfactory bulb regions [the internal (ICL) and external (ECL) layers] were high in centrality and identified as hubs. Both the ICL and ECL had most of their connections with the subpallium. The ICL also had several negative connections with the pretectal cluster described above, and the ECL connected to the octavolateralis neurons. Thus, our network analysis points to the PPa, ventral telencephalon, olfactory bulbs, and Vmn as likely playing an important role in regulating behavior during exploration of a novel tank.
Discussion
In the present study, we introduce a pipeline for performing whole-brain activity mapping in adult zebrafish. Our pipeline combines several recently developed tools: a digital brain atlas for adult zebrafish (Kenney et al., 2021), registration using ANTs (Avants et al., 2011), machine learning for automated cell detection (Tyson et al., 2021), tissue clearing (Renier et al., 2014), light-sheet microscopy (Reynaud et al., 2015), and in situ HCR (Choi et al., 2018) for detecting c-fos. All computational tools are open source and free to use. Furthermore, to aid in the implementation of this pipeline, we have included a bench protocol. The primary stumbling blocks for implementing this pipeline are likely to be access to a light-sheet microscope for whole-brain imaging, and sufficient computational power for training and applying the registration and CellFinder machine learning algorithms. The former issue is partly mitigated by the increased availability of light-sheet microscopes, particularly in core facilities. Access to computational resources can be addressed by using tools like Google Colaboratory (Bisong, 2019) or high performance computing facilities available at many institutions.
C-fos to capture whole-brain activity
We captured neural activity using in situ HCR to detect c-fos mRNA. We chose this approach for several reasons: (1) there are a paucity of antibodies for detecting c-fos protein in zebrafish, none of which are known to work in whole-mount tissue-cleared samples; (2) in situ HCR probes are small (∼150 bp), which easily penetrates chunks of intact tissue like the adult zebrafish brain; and (3) c-fos is one of the most widely used markers of neural activity due to low background, high signal-to-noise, and good temporal resolution that arises from its autoinhibition of transcription (Lucibello et al., 1989; Chung, 2015). Furthermore, c-fos has been successfully used to capture neural activity in both adult and larval zebrafish (Baraban et al., 2005; Lau et al., 2011; Ruhl et al., 2017; Shainer et al., 2023). The findings in the present study further support our rationale: we saw even penetration of c-fos staining throughout the brain (Figs. 2C, 4B) and the levels of background c-fos staining were low, with an ∼3.5-fold increase in c-fos density 15 min following behavior compared with animals removed directly from their housing racks (Fig. 4C). The increase in c-fos was also tightly coupled to the behavior, peaking 15 min after exposure to the novel tank before decreasing to baseline by 60 min. Interestingly, if we look at only cells that have more puncta like staining, we see the increase begins as soon as 5 min after the behavior (Fig. 4D). We believe that these puncta represent the initial burst of transcription in the nucleus. The shift from more puncta-like to more diffuse-like staining over 10–15 min is consistent with the time it would take for mRNA to be transcribed and shuttled from the nucleus to the cytoplasm (Oeffinger and Zenklusen, 2012). However, additional experiments are necessary to more definitively draw this conclusion. The time to maximal c-fos we observed is faster than is seen in rodents, where it is often found to peak at 30 min poststimulation (Ding et al., 1994; Kovács, 1998; Guzowski et al., 2001; Zangenehpour and Chaudhuri, 2002). The reason for this time difference between zebrafish and rodents is unclear. Nonetheless, it emphasizes the importance of performing time course analysis when establishing new methods for brain mapping in different species.
Other markers of neural activity have gained traction in recent years in zebrafish, such as the phosphorylated forms of ribosomal protein S6 (pS6) and extracellular regulated kinase 1/2 (pERK1/2). Our data suggests that c-fos as an activity marker compares favorably with these options. For pS6, the signal-to-noise ratio is comparable with what we see for c-fos, with an ∼2–4-fold increase over baseline both in vivo in zebrafish (Butler et al., 2018; Scaia et al., 2022; Parada et al., 2024) and in vitro neuronal cell culture (Kenney et al., 2015). However, the time course of elevated pS6 is notably slower, taking an hour or more to peak (Kenney et al., 2015; Parada et al., 2024) compared with 15 min for c-fos (Fig. 4C). In contrast, pERK1/2 activity peaks quickly, within 2–5 min, but the signal-to-noise ratio is ∼0.5–1, lower than c-fos (Randlett et al., 2015; Venincasa et al., 2021). This low signal-to-noise ratio likely arises from higher background levels of pERK due to the wide variety of cellular processes that it regulates (Cargnello and Roux, 2011). Thus, the best choice of stain depends on the behavioral paradigm. Large, rapid responses to brief behavioral stimuli are best captured by pERK. However, more subtle responses may be missed due to the low signal-to-noise ratio. S6 phosphorylation excels at capturing long-lasting steady-state neural activity, as suggested by Maruska et al. (2020), and would be ideal for behaviors lasting 30 min or more. c-fos represents a middle ground that is ideal for capturing neural activity from behaviors lasting on the order of 5–10 min, like the novel tank test used in the present study.
Registration to AZBA to identify cell types
We were able to successfully register our brains to AZBA using ANTs (Avants et al., 2009). To do so, we first used ANTs to make an average template from our images by registering nine brains to a single brain before averaging them together. The autofluorescence image in ABZA was then registered to this average template, yielding excellent results (Fig. 3). We chose this method because we found that registering the autofluorescence image from AZBA to individual brains gave inconsistent results. This is likely because the autofluorescence image in AZBA is also an average of many brains (Kenney et al., 2021). We chose ANTs because the nonlinear symmetric diffeomorphic image registration it employs has been found to be one of the best algorithms for 3D image registration (Klein et al., 2009; Murphy et al., 2011). The tool is also well documented and straightforward to use. Finally, ANTs have recently grown in popularity for image registration in larval zebrafish (Marquart et al., 2017; Shainer et al., 2023), which provided a starting point for identifying the best parameters for registration in our samples.
Following registration to AZBA, we were able to identify regions, and potential neuronal cell types, whose activity were increased following exposure to a novel tank (Fig. 5). We found that several regions containing high levels of 5-HT were active during behavior, such as the PVO, IN, and Hc. Consistent with this, several papers have implicated 5-HT as contributing to exploration of a novel tank using pharmacological approaches (Wong et al., 2010; Maximino et al., 2013; Nowicki et al., 2014; Beigloo et al., 2024). Similarly, there was overlap in c-fos activity in several regions that express tyrosine hydroxylase (VM, PPp, PCN, and Hc), implicating these populations of dopaminergic neurons in novel tank behavior (Kacprzak et al., 2017; Nabinger et al., 2023). Of the c-fos-positive cells that overlap with TH, our network analysis suggests that the PPp may be of particular importance in regulating exploratory behavior because it has a direct connection to the PPa region, which ranks high in both eigenvector and degree centrality (Fig. 7C), and connects to the thalamic VM region, another area high in TH expression. This suggests that the PPp and VM may act in concert to mediate the effects of the dopaminergic system on exploration. However, one important caveat to these interpretations is that we are comparing averaged c-fos images to averaged neurotransmitter-related stains in AZBA, and thus we cannot definitively identify the specific cell types that are active. This would require costaining of brains with both c-fos and cell-type markers.
Novel tank functional brain network
Using our whole-brain mapping data, we generated an adult zebrafish novel tank test functional brain network. The novel tank test is one of the most widely used behavioral tests in adult zebrafish, commonly used to study exploratory and anxiety-related behaviors (Spence et al., 2006; Blaser et al., 2010; Luca and Gerlai, 2012; Kalueff et al., 2013; Rajput et al., 2022). Our functional network analysis identifies several key regions that are engaged during exploration of a novel tank (Figs. 6, 7). In particular, the medial portion of the ventral telencephalon stands out, where several subregions (the Vv, Vd-dd, Vdd, Vc, Vd-vd, and Vp) rank highly on at least one measure of centrality (Fig. 7C). These regions are also highly interconnected, a fact that is clear from both the correlation matrix (Fig. 6) and the community they form in the network (Fig. 7B, dark green). Based on molecular markers, these regions of the subpallium are thought to correspond to the mammalian subpallial amygdala (i.e., the central and medial amygdala) and basal ganglia (Porter and Mueller, 2020; Mueller, 2022). In mammals, these brain regions have been found to be important for a wide range of behaviors, from defensive, anxiety-related, and social behaviors to motor control (Grillner and Robertson, 2016; Fadok et al., 2018; Raam and Hong, 2021). Our findings that the ventral telencephalon appears to be strongly engaged during the novel tank test makes sense given that novelty and exploration would be expected to engage circuits involved in decision-making, emotional regulation, and muscle coordination.
In examining how the regions of the ventral telencephalon interact with the rest of the brain, a few interesting trends emerge. Notably, the interaction of ventral telencephalic regions with many other communities is anticorrelated (Fig. 7B, gold and light purple). This suggests the presence of strong inhibitory connections between the medial ventral telencephalon and other parts of the brain. Consistent with this interpretation, the ventral telencephalon has been found to contain a substantial number of inhibitory GABAergic neurons (Porter and Mueller, 2020). Our network analysis suggests that these inhibitory connections are most likely present between the ventral telencephalon and the Vmn (mesencephalic nucleus of the trigeminal nerve), ENd (entopeduncular nucleus in the lateral portion of the ventral telencephalon), and from the BSTm (bed nucleus of the stria terminalis, medial portion in the dorsal telencephalon) to R (rostrolateral nucleus in the thalamus). However, given that our findings are correlational in nature, techniques like tract tracing and direct manipulation would be needed to determine if these interactions are due to direct connections or are mediated by intermediate regions.
Our network analysis also identified the PPa as a region of high importance. The PPa was high in both eigenvector and degree centrality (Fig. 7C) and was ranked highly for participation coefficient and above average for within module z-score (Fig. 7D). In examining its place in the network (Fig. 7B), the PPa interconnects with several regions of the ventral telencephalon and, working in concert with the PPp, mediates their interactions with parts of the network that contain thalamic and hypothalamic regions (Fig. 7B, dark purple cluster). To our knowledge, the correspondence between the PPa and PPp in teleosts and tetrapods has not been determined. Based on the expression of neuropeptides, like oxytocin and arginine vasopressin, parts of the PPp are thought to be homologous to the supraoptic nucleus in mammals (Herget et al., 2014). In larval zebrafish, the preoptic area has been implicated in behaviors such as navigation, thermoregulation, and stress reactivity (Corradi et al., 2022; Palieri et al., 2024). However, the preoptic area in larval zebrafish cannot be differentiated into subregions like the PPa and PPp due to a lack of cytoarchitectural boundaries (Herget et al., 2014). This makes it unclear as to what specific regions in the adult would subsume the functions identified in larval animals. Future work should determine the role that these different subregions might play in different aspects of exploration and anxiety-like behavior in adult zebrafish.
Unexpectedly, two olfactory bulb regions (ECL and ICL) were high in centrality (Fig. 7C) and identified as connector hubs (Fig. 7D). Exposure to novelty elicits a combination of exploratory and anxiety-like behaviors and fish may be engaging their olfactory system as they assess the new environment. The correlated activity between the bulbs and many other brain regions likely arises through their extensive projections to the dorsal and ventral telencephalon, habenula, and posterior tuberculum (Miyasaka et al., 2014, 2009). Neural activity in the olfactory bulbs is complex. Activity is not only elicited by incoming sensory information but sculpted by centrifugal projections from other brain regions, like the posterior zone of the dorsal telencephalon (Kermen et al., 2020), and neuromodulators like dopamine, serotonin, acetylcholine, and neuropeptide Y (Byrd and Brunjes, 1995; Clemente et al., 2004; Edwards et al., 2007; Schweitzer and Driever, 2009; Bundschuh et al., 2012; Kawai et al., 2012; Kenney et al., 2021). The importance of the olfactory bulbs in regulating activity across the brain makes sense given that reduction in their function has been associated with depression in humans (Croy and Hummel, 2016) and olfactory bulbectomy in rodents has been used to model mood disorders for over 40 years (Leonard and Tuite, 1981). Indeed, removal of the olfactory bulbs causes many structural changes throughout the brain (Song and Leonard, 2005). Taken together with the present data, it may be that the olfactory bulbs in zebrafish are also instrumental to the functioning of the brain during the exploration of a novel environment.
The only prior work examining the neural basis of exploratory behavior in fish is in larval stage animals where the regions underlying spontaneous locomotion were identified (Dunn et al., 2016). The most prominent region associated with spontaneous turn patterning in larval fish was an anterior portion of the hindbrain. Interestingly, in our novel tank functional brain network, we found several regions in a similar part of the brain with high centrality, the trigeminal nuclei (Vsm and Vmn); however, the exact relationship between these areas in the adult and larval brain is unclear. In larval fish, the preoptic area was also found to be weakly correlated with turning (Dunn et al., 2016), echoing our findings that the PPa appears to be a key region regulating exploratory behaviors. This overlap in the present work in adults and prior work in larval fish suggests that similar neural mechanisms may be at play at these two distinct life stages.
Caveats and limitations
An important caveat of the present work is that the functional brain network (commonly referred to as a “functional connectome” in human studies) reflects statistical, not physical, relationships between brain areas. Regional covariation may be due to a physical connection or mediation by one or more intermediate regions. This approach to understanding patterns of brain activity is a mainstay in human functional neuroimaging (Bullmore and Bassett, 2011) and is increasingly used in rodents (Wheeler et al., 2013; Ben-Ami Bartal et al., 2021; Verpeut et al., 2023) where it is able to predict regional involvement in behavior (Vetere et al., 2017). In zebrafish, functional connectivity approaches have been used to measure activity in a predefined set of regions in adults, providing insight into the patterns of brain activity important for social behaviors (Akinrinade et al., 2023; Pinho et al., 2023). Our approach builds on this prior work by enabling the capture of neural activity across the entirety of the brain.
One limitation of the present work is our use of iDISCO for tissue clearing. We chose iDISCO because it results in rapid and robust clearing that has been found to be compatible with multiple staining methods, like in situ HCR (Kramer et al., 2018; Kumar et al., 2021). Furthermore, AZBA was created using iDISCO (Kenney et al., 2021), and we wanted to ensure our autofluorescence images used for registration matched as closely as possible to those in AZBA. However, iDISCO requires imaging in an organic solvent, dibenzyl ether, and thus is not compatible with the objectives of many microscopes. Here, we used the UltraMicroscope II, which was specially designed for use with such solvents (Dodt et al., 2015). The use of other light-sheet microscopes may require the building of chambers to prevent the objectives from coming into contact with the solvents, as proposed when iDISCO was first described (Renier et al., 2014). Nonetheless, with the decreasing cost and wider availability of light-sheet microscopes in core facilities, our approach is likely within reach of many scientists.
Summary
The present study provides an open-source framework for performing whole-brain mapping in adult zebrafish. This work also yielded the first description of brain activity that underlies the novel tank test, suggesting the ventral telencephalon may play an important role in one of the most widely used behavioral tasks in adult zebrafish. Taken together, we anticipate that our pipeline will help generate insights into the principles of brain function by enhancing the utility of adult zebrafish as a model organism.
Data Availability
Data, code, and a bench protocol are available at Dryad: https://doi.org/10.5061/dryad.k3j9kd5js
Footnotes
The authors declare no competing financial interests.
We thank Jacob Hudock and Dinh Luong for excellent care of the zebrafish and facility maintenance.
National Institutes of Health (R35GM142566) and Richard Barber Interdisciplinary Research Program.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license, which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed.