Figures and data
Dimensionality reduction of RABV input tracing to 3 different VTA cell types.
(A) Schematic and timeline of viral injections into mice.
(B) Representative images of the BNST, GPe, and DCN of DAT-Cre, GAD2-Cre, and vGlut2-Cre mice. Green indicates RABV-labeled cells. Scale, 200 μm.
(C) Bar plot showing percent of RABV-labeled cells in each input region for DAT-Cre, GAD2-Cre, and vGlut2-Cre mice.
(D) Heatmap of the contributions of each brain region, or feature, in the data to PCs 1 through 3.
(E) PCA of input labeling from brains from DAT-Cre, GAD2-Cre, and vGlut2-Cre mice. For this and other figures, ellipsoids were centered at the average coordinate of a condition and stretched one standard deviation along the primary and secondary axes.
(F) Box plot comparisons of PC1. One-way ANOVA p = 0.010, pairwise t-tests DAT-Cre vs. GAD2-Cre multiple comparisons adjusted p = 0.0087, DAT-Cre vs. vGluT2-Cre p = 0.46, GAD2-Cre vs. vGluT2-Cre p = 0.19. n=8 for DAT-Cre, n = 13 for GAD2-Cre, and n = 7 for vGluT2-Cre.
(G) Box plot comparisons of PC1. One-way ANOVA p = 0.016, pairwise t-tests DAT-Cre vs. GAD2-Cre multiple comparisons adjusted p = 0.65, DAT-Cre vs. vGluT2-Cre p = 0.016, GAD2-Cre vs. vGluT2-Cre p = 0.051.
(H) Schematic of how projection portraits are constructed. A representative slice of the VTA is considered for experiments from the Allen Mouse Brain Connectivity Atlas with injections into relevant brain regions that provide input to the VTA. Projection density values in this slice are averaged and visualized to obtain the projection portrait in the VTA for that group of regions.
(I) Projection portrait of the VTA based on inputs from regions with a high positive contribution to PC1: BNST, EP, GPe, CeA, and ZI.
(J) Projection portrait of the VTA based on inputs from regions with a large negative contribution to PC1: anterior cortex, NAcMed, VP, PO, septum, LHb, MHb, and LH.
(K) Projection portrait of the VTA based on inputs from regions with a high positive contribution to PC2: LHb, MHb, EP, ZI, DCN, and anterior cortex.
(L) Projection portrait of the VTA based on inputs from regions with a large negative contribution to PC2: NAcMed, NAcLat, NAcCore, DStr, and GPe.
(M) UMAP embedding of input labeling of brains from DAT-Cre, GAD2-Cre, and vGlut2-Cre mice.
(N) Correlogram generated using the average Euclidean distance of data points from each other from 20 UMAP embeddings.
PCA can deconvolve drug-induced effects from experimental variation.
(A) Bar graph representation of inputs from DAT-Cre mice anesthetized with isoflurane and treated with a single injection of saline or fluoxetine. N = 4 for saline, n = 5 for fluoxetine. Two-way ANOVA, p = 0.91.
(B) PCA plot showing PC1 and PC2 of brains from saline or fluoxetine-treated mice.
(C) PCA plot showing PC2 and PC3 of brains from saline or fluoxetine-treated mice.
(D) Box plot of PC1, p = 0.75. n = 4 for saline, n = 5 for fluoxetine.
(E) Box plot of PC2, p = 0.78.
(F) Box plot of PC3, p = 0.81.
(G) Bar graph representation of inputs from DAT-Cre mice anesthetized with isoflurane and treated with a single injection of saline or fluoxetine (controls), or one of the drugs of abuse cocaine, methamphetamine, amphetamine, nicotine, or morphine. n = 9 for control, n = 24 for combined drugs of abuse. Two-way ANOVA, p = 0.0001; unpaired t-tests with Sidak corrections for multiple comparisons: DStr, p = 0.0051, GPe, p = 0.0036.
(H) PCA plot showing PC1 and PC2 of brains from control or drug-treated mice.
(I) PCA plot showing PC2 and PC3 of brains from control or drug-treated mice.
(J) Box plot of PC1, p = 0.78. n = 9 for control, n = 24 for drugs.
(K) Box plot of PC2, p = 0.0009.
(L) Box plot of PC3, p = 0.024.
(M) Heatmap of the contributions of each brain region to PC1 through PC3 for data shown in panels E-F.
(N) Distribution of the starter cell center of mass along the medial-lateral axis for each experimental condition. Saline n = 4, fluoxetine n = 5, cocaine n = 5, methamphetamine n = 4, amphetamine n = 5, nicotine n = 5, morphine n = 5.
(O) PC1 vs. PC2 for control vs. drug conditions as shown in E, with each point colored according to the medial-lateral coordinate of the center of mass of starter cells for each brain.
(P) PC2 vs. PC3 for control vs. drug conditions as shown in F, with each point colored according to the medial-lateral coordinate of the center of mass of starter cells for each brain.
(Q) Linear regression between the x-coordinate of the center of mass of starter cells for each brain vs. PC1. r2 = 0.770, p = 1.62e-7,
(R) Linear regression between the x-coordinate of the center of mass of starter cells for each brain vs. PC2. r2 = 0.308, p = 0.082.
(S) Linear regression between the x-coordinate of the center of mass of starter cells for each brain vs. PC3, r2 = 0.085, p = 0.638.
Dimensionality reduction analysis of brains from animals treated with a single drug exposure.
(A) PCA plot showing PC1 and PC2 of brains from DAT-Cre and GAD2-Cre mice. DAT-Cre mice were anesthetized with K/X and treated with saline or anesthetized with isoflurane and treated with saline or one of the drugs cocaine, methamphetamine, amphetamine, nicotine, morphine, or fluoxetine. GAD2-Cre mice were anesthetized with isoflurane and treated with saline or cocaine.
(B) Box plot of PC1, One-way ANOVA p < 0.0001, pairwise t-tests DAT-Cre saline vs. DAT-Cre fluoxetine multiple comparisons adjusted p = 0.99, DAT-Cre saline vs. DAT-Cre K/X p = 1.0, DAT-Cre saline vs. DAT-Cre drugs p = 0.23, DAT-Cre saline vs. GAD2-Cre p < 0.0001, DAT-Cre fluoxetine vs. DAT-Cre K/X p = 1.0, DAT-Cre fluoxetine vs. DAT-Cre drugs p = 0.46, DAT-Cre fluoxetine vs. GAD2-Cre p < 0.0001, DAT-Cre K/X vs. DAT-Cre drugs p = 0.31, DAT-Cre K/X vs. GAD2-Cre p < 0.0001, DAT-Cre drugs vs. GAD2-Cre p < 0.0001. n = 4 for DAT-Cre saline and DAT-Cre K/X, n = 5 for DAT-Cre fluoxetine, n = 24 for DAT-Cre drugs, n = 17 for GAD2-Cre.
(C) PCA plot showing PC1 and PC2 of brains from DAT-Cre mice anesthetized with isoflurane and treated with cocaine, methamphetamine, amphetamine, nicotine, morphine, fluoxetine, or saline, or anesthetized with K/X and treated with saline.
(D) PC2 and PC3 of the same group of brains as shown in panel B.
(E) Box plot of PC1, One-way ANOVA p = 0.81.
(F) Box plot of PC2, One-way ANOVA p = 0.0093, pairwise t-tests DAT-Cre saline vs. DAT-Cre fluoxetine multiple comparisons adjusted p = 0.94, DAT-Cre saline vs. DAT-Cre K/X p = 0.87, DAT-Cre saline vs. DAT-Cre drugs p = 0.034, DAT-Cre fluoxetine vs. DAT-Cre K/X p = 0.99, DAT-Cre fluoxetine vs. DAT-Cre drugs p = 0.10, DAT-Cre K/X vs. DAT-Cre drugs p = 0.27.
(G) Box plot of PC3, One-way ANOVA p = 0.086.
(H) Heatmap of the contributions of each brain region to PC1 through PC3 for data shown in panel A.
(I) Heatmap of the contributions of each brain region to PC1 through PC3 for data shown in panels B and C.
(J) Bar plot showing percent of RABV-labeled cells in each input region for the DAT-Cre mice shown in panel B.
(K) Representative images of brain slices from DAT-Cre mice showing, from top to bottom, the NAc, BNST, GPe, and PBN. Scale, 500 μm.
(L) Projection portrait of the VTA based on inputs from regions with a large negative contribution to PC1, based on data from panel E: NAcCore, NAcLat, DStr, GPe.
(M) Projection portrait based on inputs from regions with a high positive contribution to PC1, based on data from panel E: LHb, DR, VP, PO, EAM, LH.
(N) Projection portrait based on inputs from regions with a high positive contribution to PC2, based on data from panel E: EAM, PBN, DR, EP, LH, GPe, ZI.
(O) Projection portrait based on inputs from regions with a large negative contribution to PC2, based on data from panel E: NAcMed, NAcCore, DStr, Septum, MHb.
(P) UMAP embedding of data from the brains of DAT-Cre mice shown in panels B and F.
(Q) Correlogram showing average Euclidean distance of data points over 20 UMAP embeddings of the data presented in panel J. Experiment conditions from bottom to top and left to right are labelled 1-10 and are ordered as follows: GAD2-Cre cocaine, GAD2-Cre saline, saline, fluoxetine, cocaine, nicotine, amphetamine, meth, K/X, and morphine.
Dimensionality reduction analysis of brains from animals anesthetized with isoflurane and treated with either cocaine or saline, or anesthetized with ketamine/xylazine and treated with saline.
(A) PCA plot of input data from VTADA cells from mice anesthetized either with isoflurane or K/X and given saline. Box plot comparisons of PC1 and PC2 are shown to the right for this plot and D, G, J, and M. PC1, p = 0.086, PC2, p = 0.037, unpaired t-tests. N = 4 for both.
(B) UMAP plot of input data from VTADA cells from mice anesthetized either with isoflurane or K/X and given saline.
(C) Bar plot showing percent of RABV-labeled cells in each input region from mice anesthetized either with isoflurane or K/X and given saline.
(D) PCA plot of RABV input mapping experiments to VTADA®NAcLat cells in mice anesthetized with isoflurane and given saline or cocaine, or anesthetized with K/X and given saline. PC1, One-way ANOVA p = 0.0003, pairwise t-tests saline vs. cocaine multiple comparisons adjusted p = 0.011, saline vs. K/X p = 0.071, K/X vs. cocaine p = 0.0002. PC2, One-way ANOVA p = 0.16. n=4 for saline and K/X, n = 5 for cocaine.
(E) UMAP plot of input mapping experiments to VTADA®NAcLat cells.
(F) Bar plot showing percent of RABV-labeled cells in each input region for cTRIO experiments from VTADA®NAcLat cells.
(G) PCA plot of input mapping experiments to VTADA®Amygdala cells. PC1, One-way ANOVA p = 0.48. PC2, One-way ANOVA p = 0.30. n=4 for saline, n = 5 for K/X and cocaine.
(H) UMAP plot of input mapping experiments to VTADA®Amygdala cells.
(I) Bar plot showing percent of RABV-labeled cells in each input region for cTRIO experiments from VTADA®Amygdala cells.
(J) PCA plot of input mapping experiments to VTADA®NAcMed cells. PC1, One-way ANOVA p = 0.64. PC2, One-way ANOVA p = 0.27. n=4 for saline and K/X, n = 5 for cocaine.
(K) UMAP plot of input mapping experiments to VTADA®NAcMed cells.
(L) Bar plot showing percent of RABV-labeled cells in each input region for cTRIO experiments from VTADA®NAcMed cells.
(M) PCA plot of input mapping experiments to VTADA®mPFC cells. PC1, One-way ANOVA p = 0.67. PC2, One-way ANOVA p = 0.97. n=4 for saline and cocaine, n = 5 for K/X.
(N) UMAP plot of input mapping experiments to VTADA®mPFC cells.
(O) Bar plot showing percent of RABV-labeled cells in each input region for cTRIO experiments from VTADA®mPFC cells.
(P-T) Heatmap of the contributions of each brain region to PC1 to PC3 for data shown in (P) panel A, (Q) panel D, (R) panel G, (S) panel J, (T) panel M.
(U) Projection portrait of regions with high PC1 contributions in all VTADA cells: VP, PO, EAM, PVH, LH, PBN.
(V) Projection portrait of regions with high PC2 contributions in VTADA cells: NAcLat, DStr, GPe, PVH.
(W) Projection portrait of regions with low PC1 contributions in VTADA®NAcLat cells: cortex, NAcMed, NAcLat, NAcCore, septum, PVH, MHb.
(X) Projection portrait of regions with high PC2 contributions in VTADA®NAcLat cells: VP, PO, EAM, LHb, LH.
Gene ontology (GO) analysis of top 50 genes within the Allen Gene Expression Atlas correlated with differences in RABV input labeling obtained from experimental vs. control mice.
(A-B) GO from an average of the addictive drug conditions vs. saline-treated controls, all anesthetized using isoflurane.
(C-D) GO from K/X-anesthetized vs. isoflurane-anesthetized saline-treated mice.
Panels A and C show molecular function-related gene classes whose expression was positively correlated with RABV labeling differences, and panels B and D show GO analyses that reflect biological process-related gene classes whose expression was negatively correlated with RABV labeling differences.
Relationship between RABV labeling ratio in addictive drug/saline-treated groups and mean basal gene expression of gene subgroups, subtypes of ion channels, and neurotransmitter receptors.
Linear regressions are shown for all five examined drugs against different gene classes expressed by brain region within the Allen Gene Expression Atlas, including (A) ion channel-related genes, (B) neurotransmitter-related genes, (C) synapse-related genes, (D) Ca2+ channels, (E) Cl- channels, (F) K+ channels, (G) Na+ channels, (H) glutamate receptors, (I) acetylcholine receptors, (J) GABA receptors, (K) glycine receptors.
Inhibition of Cacna1e expression in the NAcLat reduces RABV-labeled inputs from the NAcLat onto VTADA cells.
(A) Schematic of viral injections performed.
(B) Schematic of connectivity in relevant brain regions and explanation of the normalized input metric.
(C) Representative image of fluorescence in starter cells in the VTA. Scale, 1 mm.
(D) Representative image of fluorescence in the NAcLat after either administration of no gRNA (left) or gRNA targeting Cacna1e (right). Scale, 1 mm.
(E) qPCR results showing slight inhibition of Cacna1e in the NAcLat.
(F) Difference in RABV-labeled inputs after CRISPRi-mediated knockdown of Cacna1e compared to controls15, and no gRNA.
Brains from animals treated with a single exposure to a drug (Figure 3A), but with the brain identities scrambled.
(A) PCA plot showing the data with scrambled identities along PC1 and PC2.
(B) The same scrambled data along PC1 and PC3.
(C) UMAP of the scrambled data.
(D) Correlogram of 20 UMAP embeddings of the scrambled data.
Dimensionality reduction analysis of brains from animals anesthetized with isoflurane and treated with either cocaine or saline, or anesthetized with ketamine/xylazine and treated with saline, focusing on PC3.
(A) PCA plot of input data from VTADA cells from mice anesthetized either with isoflurane or K/X and given saline, showing PC2 and PC3.
(B) Box plot comparisons of PC3, p = 0.99, unpaired t-tests. N = 4 for both.
(C) PCA plot of RABV input mapping experiments to VTADA®NAcLat cells in mice anesthetized with isoflurane and given saline or cocaine, or anesthetized with K/X and given saline, showing PC2 and PC3.
(D) Box plot comparisons of PC3, One-way ANOVA p = 0.12. n=4 for saline and K/X, n = 5 for cocaine.
(E) PCA plot of input mapping experiments to VTADA®Amygdala cells, showing PC2 and PC3.
(F) Box plot comparisons of PC3, PC1, One-way ANOVA p = 0.014, pairwise t-tests saline vs. K/X multiple comparisons adjusted p = 0.011, saline vs. cocaine p = 0.24, K/X vs. cocaine p = 0.16. n=4 for saline, n = 5 for K/X and cocaine.
(G) PCA plot of input mapping experiments to VTADA®NAcMed cells, showing PC2 and PC3.
(H) Box plot comparisons of PC3,, One-way ANOVA p = 0.64. PC2, One-way ANOVA p = 0.16. n=4 for saline and K/X, n = 5 for cocaine.
(I) PCA plot of input mapping experiments to VTADA®mPFC cells, showing PC2 and PC3.
(J) Box plot comparisons of PC3, One-way ANOVA p = 0.67. PC2, One-way ANOVA p = 0.083. n=4 for saline and cocaine, n = 5 for K/X.
Inputs to brains of mice anesthetized with isoflurane and receiving saline or cocaine, or K/X-saline (Figure 4), but with the brain identities scrambled.
(A) PCA plot of scrambled data from isoflurane-anesthetized and K/X-anesthetized mice.
(B) UMAP plot of scrambled data from isoflurane-anesthetized and K/X-anesthetized mice.
(C) PCA plot of scrambled data from input mapping experiments to VTADA®NAcLat cells.
(D) UMAP plot of scrambled data from input mapping experiments to VTADA®NAcLat cells.
(E) PCA plot of scrambled data from input mapping experiments to VTADA®Amygdala cells.
(F) UMAP plot of scrambled data from input mapping experiments to VTADA®Amygdala cells.
(G) PCA plot of scrambled data from input mapping experiments to VTADA®NAcMed cells.
(H) UMAP plot of scrambled data from input mapping experiments to VTADA®NAcMed cells.
(I) PCA plot of scrambled data from input mapping experiments to VTADA®mPFC cells.
(J) UMAP plot of scrambled data from input mapping experiments to VTADA®mPFC cells.
Gene ontology analysis of top 50 expressed genes correlated to differences in RABV input labeling obtained from drug-treated vs. control mice.
GO from mice treated with (A) amphetamine, (B) cocaine, (C) methamphetamine, (D) morphine, and (E) nicotine-treated relative to saline-injected controls are shown. Each GO analysis reflects molecular function-related gene classes whose expression was negatively correlated with RABV labeling differences.
Gene ontology analysis of top 50 expressed genes correlated to differences in RABV input labeling obtained from drug-treated vs. control mice.
GO analyses from mice treated with (A) amphetamine, (B) cocaine, (C) methamphetamine, (D) morphine, and (E) nicotine-treated relative to saline-injected controls are shown. Each GO analysis reflects biological process-related gene classes whose expression was positively correlated with RABV labeling differences.
Gene ontology analysis of the top 50 genes correlated with differences in RABV input labeling obtained from fluoxetine vs. saline-treated controls.
Panel (A) shows molecular function-related gene classes whose expression was positively correlated with RABV labeling differences, and panel (B) shows GO analyses that reflect biological process-related gene classes whose expression was negatively correlated with RABV labeling differences.
Relationship between RABV labeling ratio in addictive drug/saline-treated groups and mean basal gene expression of gene subgroups.
Linear regressions are shown for all five examined drugs against all genes and different gene classes expressed by brain region within the Allen Gene Expression Atlas, including (A) all genes, (B) substance dependence-related genes, (C) endocytosis- and exocytosis-related genes, (D) mitochondrial genes, (E) sugar transporter genes, (F) tumor-associated genes, (G) ligand-gated ion channel genes, (H) voltage-gated ion channel genes, (I) other ion channel genes.
Relationship between RABV labeling ratio in K/X-saline vs. isoflurane-saline groups and mean basal gene expression of gene subgroups.
Linear regressions are shown for K/X-saline vs. isoflurane-saline groups against different gene classes, including (A) ion channel-related genes, (B) synapse-related genes, (C) neurotransmitter-related genes, (D) DEGs related to substance abuse, (E) voltage-gated ion channels, (F) ligand-gated ion channels, (G) other ion channels, (H) K+ channels, (I) Ca2+ channels, (J) Cl- channels, (K) Na+ channels, (L) glutamate receptors, (M) glycine receptors, (N) acetylcholine receptors, (O) GABA receptor genes.
Relationship between RABV ratio labeling in fluoxetine vs. saline-treated groups, both anesthetized with isoflurane, and mean basal gene expression of gene subgroups.
Linear regressions are shown for fluoxetine/saline-treated groups against different gene classes, including (A) ion channel-related genes, (B) synapse-related genes, (C) neurotransmitter-related genes, (D) DEGs related to substance abuse, (E) voltage-gated ion channels, (F) ligand-gated ion channels, (G) other ion channels, (H) K+ channels, (I) Ca2+ channels, (J) Cl- channels, (K) Na+ channels, (L) glutamate receptors, (M) glycine receptors, (N) acetylcholine receptors, (O) GABA receptor genes.
