Skip to main content
. 2019 Oct 18;8:e43775. doi: 10.7554/eLife.43775

Figure 1. Whole brain activity mapping reveals anti-correlated hypothalamic regions.

(a) Top: The protocols used to quantify feeding behavior in larval zebrafish. At 7 or 8 dpf, larvae were either food-deprived for 2 hr, or fed with excess paramecia for this duration. After 2 hr (2–4 hr in the case of behavioral imaging), they were subject to a quick wash, followed either by: 1) addition of excess fluorescently-labeled paramecia (left), 2) high-resolution behavioral imaging (right; see Johnson et al., 2019, and Materials and methods). Gut fluorescence is both cumulative and diminished by digestion (Jordi et al., 2015) and so lags the dynamics of hunting behavior. Bottom left: Gut fluorescence measurements of food-deprived (red) or fed (blue) fish as a function of duration of feeding labeled paramecia. Groups of fed or food-deprived larvae were fixed at the indicated time points (fed: n = 7/18/19/17/17 fish, food-deprived: n = 8/23/20/14/15 fish). Food-deprived fish had significantly higher gut fluorescence than fed fish overall (p = 7.5859×10−10, Two-way ANOVA, asterisk indicates corrected p-values<0.05. Bottom right: The probability of performing a hunting-related swim bout across fed and food-deprived fish groups in 3 min time bins over 45 min. Error bars represent 90% confidence intervals. For all bins except those indicated with triangles, the null hypothesis that initial feeding condition has no effect on hunting-bout probability is rejected (p<0.00001, Fisher’s Exact Test comparing binomial probability distributions per bin). Fed: n = 85655 bouts from 73 fish; Food-deprived: n = 75357 bouts from 57 fish. Since the rate of food intake and hunting behavior was highest in the first 15 min (voracious feeding phase, gray boxes), we chose this time point for subsequent MAP-mapping experiments. (b) Brain-wide activity mapping of food-deprived (Dep.) fish exposed to food for 15 min, with subtraction of activity in continuously fed (Fed) fish. Data from nine experiments were combined to generate this difference map based on anti-pERK staining fluorescence. Relative activation from feeding after food deprivation yields activated regions including the telencephalon (Tel), Arborization field 7 (AF7), cerebellum (CB), hindbrain (HB), Vagal ganglion (VG) and lateral lobe of the intermediate hypothalamus (LH). Reduced activity was observed in the caudal hypothalamus (cH) and some areas of the telencephalon. Scale bar = 100 μm. Also see Video 1. (c) ROI-specific pixel intensity analysis of LH and cH regions in nine independent MAP-mapping experiments (20–30 fish per treatment per experiment). The cH or LH ROI intensities of each individual fish was normalized to the mean cH or LH ROI intensity of all fed fish. Food-deprived fish consistently displayed higher LH and lower cH pERK fluorescence after the onset of feeding (p = 0.0019 for both cH and LH, one-tailed Wilcoxon signed-rank test). (d) Z-projection of same MAP-map as described in (b) in planes revealing the hypothalamus (right panel), where lateral regions (e.g. lateral hypothalamus, LH) display strong relative activation and medial regions (e.g. caudal hypothalamus, cH) display reduced activity in when food-deprived animals were fed for 15 min. The map is overlaid onto a stack for the transgenic line Tg(VMAT:GFP) (left panel) to localize the cH region. (e) Six examples of independent component analysis (ICA) maps. Voxels for each recovered independent component (IC) are shown as maximum projections, with intensity proportional to the z-score of the loadings of the ICA signal. These ICs, along with others (22/30) highlight LH and cH regions of opposite loadings, suggesting they may be included in a network that displays anti-correlated activity patterns between the cH and LH. A subset of these ICs (e.g. #14 and #24) only showed partial anti-correlation between the cH and the LH. All ICs are shown in Figure 1—figure supplement 3. Positive (+) loading and Negative (-) loadings (z-score values of IC signals) are reflected in green and magenta, respectively. (f) Confocal micrographs of anti-pERK antibody stained brains from animals that were continuously fed (panel (i), left), food-deprived for 2 hr (panel (i), center) and fed for 5 min after food deprivation (panel (i), right). cH (ii) and LH (iii) insets are shown at higher magnification on the bottom and right side respectively. The lateral hypothalamus is shown with subdivisions lateral lateral hypothalamus (lLH) and medial lateral hypothalamus (mLH). (i) scale bar: 50 μm; (ii) and (iii) scale bar: 20 μm. Fish are mounted ventral side up. (g) Quantification of cH and LH activities by normalized anti-pERK fluorescence intensity averaging. The normalized anti-pERK staining intensity for each region (ROI) was obtained by dividing the anti-pERK fluorescence from each fish (in all experimental groups) by the average anti-pERK fluorescence for the same ROI of continuously fed fish. Quantitative analysis performed on fish in six independent conditions (n = 13/11/9/9/13/12). Normalized anti-pERK fluorescence intensity (cH/mLH/lLH): Fed vs Dep. (p = 0.016/0.12/0.11), Dep. vs Dep. + 5 min food (p = 3.1×10−4/9.9 × 10−5/0.020), Fed vs Dep. + 5 min food (p = 0.0097/8.5 × 10−4/0.11). Asterisks denote p<0.05, one-tailed Wilcoxon rank-sum test. (h) The active cell count metric (bottom panels) was determined as described in Figure 1—figure supplement 4 by a thresholding protocol to isolate and count individual pERK-positive cells within a z-stack. This approach could be reliably performed for areas of sparse active cells (e.g. mLH and lLH) but not where individually labeled pERK-positive neurons are not well separated (such as the cH). Active cell count (mLH/lLH): Fed vs Dep. (p = 0.001/0.0038), Dep. vs Dep. + 5 min food (p = 9.7×10−5/1.3 × 10−5), Fed vs Dep. + 5 min food (p = 0.0038/0.048). Asterisks denote p<0.05, one-tailed Wilcoxon rank-sum test. (i) Schematic of inferred cH and LH activity in relation to feeding behavior. Note that, based on data in Figure 2, the LH active cell count appears to decline more rapidly than the rise in cH activity (based on cH average fluorescence intensity). Data plotted in Figure 1 are provided in Figure 1—source data 1.

Figure 1—source data 1. Source data for plots displayed in Figure 1a, c, g and h.

Figure 1.

Figure 1—figure supplement 1. Anatomical characterization of intermediate hypothalamus expression of appetite related peptides.

Figure 1—figure supplement 1.

(a) Expression patterns of a number of feeding-related peptides in the zebrafish hypothalamus, based on antibody-staining or transgenic labels (see Materials and methods). HCRT = hypocretin (orexin), CART = cocaine and amphetamine related transcript MCH = melanin concentrating hormone, TH = tyrosine hydroxylase (labels dopaminergic and/or noradrenergic neurons), MSH = alpha melanocyte stimulating hormone, AgRP = Agouti related peptide, NPY = neuropeptide Y, VMAT = vesicular monoamine transporter (labels dopaminergic (DA) and serotonergic neurons (5-HT)). Note that MCH and HCRT staining is absent from the zebrafish LH. Though not apparent from the schematic, HCRT is located more dorsally. The preoptic area, which contains oxytocinergic as well as other peptidergic neurons, is located more dorsally and not reflected in this schematic. (b) Schematic view from the ventral perspective summarizing zebrafish hypothalamic peptide expression. GABA (dark blue) and glutamatergic (blue) neurons are found in the zebrafish LH (see Figure 1—figure supplement 2) and also throughout the medial regions of the hypothalamus. PVO = paraventricular organ, which also contains DA and 5-HT neurons. A number of peptidergic neurons are located within the anterior and posterior pituitary/hypophysis (aPit and pPit). Color code corresponds to images in (a). A = anterior, R = right.
Figure 1—figure supplement 2. Characterization of neuronal transmitter types in the zebrafish lateral hypothalamus.

Figure 1—figure supplement 2.

(a) Glutamatergic and GABAergic neuron distribution in the hypothalamus. Tg(VGlut2a:dsRed) and Tg(GAD1b:GFP) transgenic fish were dissected, imaged and registered onto a common reference hypothalamus. All fish in this figure were food-deprived for 2 hr and fixed for analysis after 15 min of feeding. (b) Glutamatergic cells, labeled by Tg(VGlut2a:dsRed), overlap with active (pERK-positive) neurons in both the lLH and outer rim of the mLH. (i) Z-projection of hypothalamus. (ii) Higher magnification images of LH (iii-iv) Inset showing overlap of lLH and outer rim of mLH with glutamatergic cells. (c) GABAergic cells, labeled by Tg(Gad1b:GFP), overlap with active neurons in the inner rim of the mLH but not the lLH. (i) Z-projection of hypothalamus. (ii) Higher magnification images of LH showing a subset of z-planes. (iii-iv) Inset showing overlap of inner rim of mLH with GABAergic cells. White arrows point to examples of overlapping cells. All fish were mounted ventral side up. Scale bar (i and ii) = 50 μm. Inset (iii and iv) scale bar = 20 μm.
Figure 1—figure supplement 3. All 30 independent components extracted from ICA analysis.

Figure 1—figure supplement 3.

This method separates pERK signals into statistically independent components based on their correlated and anti-correlated activities, thus identifying putative functional connectivity (both positive or negative relationships) between different brain regions (Randlett et al., 2015; see Materials and methods). Fish included in this analysis were either food-deprived (2 hr), food-deprived and then fed for 15 min prior to harvest, or continuously fed (n = 300 fish total). (a-c) From this analysis, multiple independent component networks (ICs) were identified in which at least part of the LH displayed an inverse activity relationship (i.e. opposite loadings) with the cH (22/30). (d) 4/30 ICs had correlated LH and cH activity. However, in these cases lateral loci displayed some anti-correlated activity with medial loci (especially IC #15 and 29). (e) There were 4/30 ICs that displayed asymmetrical or noisy activity patterns that rendered them unclassifiable.
Figure 1—figure supplement 4. Automated quantification of pERK-positive (active) cells.

Figure 1—figure supplement 4.

(a) Method by which pERK-positive (‘active’) cell count were determined in a high-throughput manner. Brain z-stacks obtained from confocal microscopy are registered with a selected reference brain within the same dataset, using the tERK channel, though in experiments where tERK staining was not performed, unregistered images were used (for which individual ROIs have to be defined for each image). A series of processing steps were uniformly applied to segment pERK-positive cells, which were selected using a manually optimized threshold across the entire dataset. Cell counts were obtained using the Analyze Particles algorithm within the Fiji software.
Figure 1—figure supplement 5. Food deprivation-induced activity in caudal hypothalamus monoaminergic neurons.

Figure 1—figure supplement 5.

(a) Dopaminergic neurons are labeled in Tg(TH2:GCaMP5) fish. These animals were food-deprived for 2 hr and then co-stained with anti-5-HT (to label serotonergic neurons) and anti-pERK antibodies in order to quantify food deprivation-induced activity in both cell types. Each row shows a different z-plane, moving from ventral to dorsal. (i) There is minimal overlap between Tg(TH2:GCaMP5)-positive cells (magenta) and 5-HT labeling (green). There is higher overlap of anti-pERK staining (magenta) with (iii) 5-HT-positive cells (green) as compared to (ii) Tg(TH2:GCaMP5)-positive cells (green). White arrows point to examples of overlapping cells. White boxes indicate region shown in insets. Scale bar = 20 μm. Full z-stacks for (ii) pERK overlap with anti-5-HT staining (Video 2) and (iii) TH2:GCaMP5 expression (Video 3) are also provided. (b) Quantification of overlap between pERK-positive cells and anti-5-HT staining or Tg(TH2:GCaMP5) expression. Other cH cell types, including histaminergic neurons (Chen et al., 2016) are not labeled. Fish one corresponds to the fish shown in (a).