Figures and data
snRNA-seq reveals a unique adipocyte subpopulation and delineates a brown-to-white transition process in mPRAT during postnatal development
(A) UMAP of all cell types in mPRAT from 1-, 2- and 6-month-old C57BL/6J male mice. A total of 11,708 nuclei were integrated, including 3,342, 3,608 and 4,758 nuclei from 15, 12 and 8 animals for the 3 time points, respectively.
(B) Violin plot of the marker gene expression levels of all the identified cell types in mPRAT. Expression level was log normalized for all the violin plots in the study. Cell types are represented by circles following the same color scheme in corresponding UMAP.
(C) Histogram illustrating the percentage of each cell type in mPRAT relative to the total number of analyzed nuclei.
(D) UMAP of all cell types in iBAT from 1-, 2- and 6-month-old C57BL/6J male mice. A total of 31,079 nuclei were integrated, including 6,468, 8,323 and 16,288 nuclei from the same mice littler as that in mPRAT for the 3 time points, respectively. SMC, smooth muscle cell.
(E) Violin plot of the marker gene expression levels of all the identified cell types in iBAT.
(F) Histogram illustrating the percentage of each cell type in iBAT relative to the total number of analyzed nuclei.
(G) UMAP and cellular composition of the adipocyte subpopulations in mPRAT from 3 time points.
(H) Violin plot of the marker gene expression levels of each mPRAT adipocyte subpopulation.
(I) Ucp1, Cidea, Cyp2e1 and Aldh1a1 expression pattern in the mPRAT adipocyte subpopulations. Scale bar represents log normalized gene expression levels.
(J) Monocle 3 trajectory map of the mPRAT adipocyte subpopulations.
(K) Heatmap of DEGs along the pseudo-time trajectory in (J). Genes were selected by Moran’s I test in Monocle 3 with a threshold larger than 0.2. Representative genes were highlighted. Adipocytes were aligned along the increasing pseudo-timeline and plotted using Z-scaled gene expression. Gene module patterns, including Decreasing, Intermediate (IM) and Increasing were determined using k-means clustering (k=3).
(L) Gene ontology (GO) analysis of DEGs from each module in (K). Color and size of each circle represent the adjusted p-value by Benjamini & Hochberg method and the gene ratio within each module, respectively.
lPRAT has distinct cellular composition compared with eWAT
(A) UMAP of all cell types in lPRAT from 2- and 6-month-old C57BL/6J male mice. A total of 6,048 nuclei were analyzed, including 1,428 and 4,620 nuclei for the 2 time points, respectively.
(B) Violin plot of the marker gene expression levels of all cell types in lPRAT.
(C) Histogram illustrating the percentage of each cell type in lPRAT relative to the total number of analyzed nuclei.
(D, G, J) UMAP and cellular composition of the adipocyte (D), ASPC (G) and macrophage (J) subpopulations in lPRAT.
(E, H, K) Violin plot of the marker gene expression levels of each adipocyte (E), ASPC (H) and macrophage
(K) subpopulation in lPRAT.
(F, I, L) Merged clusters of lPRAT adipocytes (F), ASPCs (I) and macrophages (L) with the corresponding clusters in Sarvari et al(Sarvari et al., 2021). Grey oval shape in (F) highlights the unique adipocyte subpopulation in lPRAT. LSA, lipid-scavenging adipocyte; LGA, lipogenic adipocyte; SLSA, stressed lipid-scavenging adipocyte. RM, regulatory macrophage; NPVM, non-perivascular macrophage; PVM, perivascular macrophage; LAM, lipid-associated macrophage; P-LAM, proliferating LAM; CEM, collagen-expressing macrophage.
puPRAT contains majority of the whitened Bas
(A) Representative immunofluorescence images of tdTomato (red) and UCP1 (green) expression in pvPRAT, puPRAT and iBAT of 2-month-old Ucp1Cre;Ai14 male mice. Whitened adipocytes are marked with Tomato expression, but not UCP1 (white arrows). Lipid droplets were stained with BODIPY (grey). Insets highlight the whitened adipocytes with relatively large lipid droplet size. Scale bar, 50µm.
(B-D) Quantification of the percentage of Tomato+/UCP1- (B), Tomato+ (C) and UCP1+ (D) cells in (A). n=5 mice for each tissue, represented by a dot in the graph. 3 tissue slices were quantified for each mouse for all analysis in the study. Bars represent mean±SD for all analysis in the study. ** p < 0.01; **** p<0.0001 by one-way ANOVA.
(E) Representative hematoxylin and eosin (HE) images of pvPRAT, puPRAT and iBAT of 1-, 2- and 6-month-old C57BL/6J male mice. Scale bar, 50µm. V, blood vessel; RH, renal hilum.
(F) Quantification of the lipid droplet size in (E). A total of 3,082-4,857 adipocytes from 4-6 mice were quantified for each tissue. Boxplot shows the area distribution of all lipid droplets where the whiskers show the 10-90 percentile. *p < 0.05; ****p < 0.0001 by Two-way ANOVA.
(G) Representative immunofluorescence images of tdTomato (red), UCP1 (green) and CYP2E1/ALDH1A1 (magenta) expression in puPRAT of 2-month-old Ucp1Cre;Ai14 male mice. tdTomato+/UCP1-/CYP2E1+ and tdTomato+/UCP1-/ALDH1A1+ cells represent the whitened adipocytes that became mPRAT-ad3 and 4 adipocytes (white arrows). Scale bar, 40µm.
(H-I) Quantification of the percentage of tdTomato+/UCP1-/CYP2E1+ (H) and tdTomato+/UCP1-/ALDH1A1+ cells (I) in (G). n=3 mice for each tissue, represented by a dot in the graph. Bars represent mean±SD. ***, p < 0.001; **** p<0.0001 by unpaired t-test.
Cold exposure inhibits BA whitening and restores UCP1 expression in UCP1- adipocytes of mPRAT
(A) Illustration of the cold exposure experimental design. RT, room temperature; CE, cold exposure.
(B) qPCR analysis of Ucp1 mRNA level in mPRAT, lPRAT and iBAT of 2-month-old C57BL/6J male mice in (A). n=8 mice for each condition, represented by a dot in the graph. **, p < 0.01; ****, p<0.0001 by one-way ANOVA.
(C) Western blot and quantification of UCP1 protein level in mPRAT and iBAT of 2-month-old C57BL/6J male mice in (A). n=4 mice for each condition, represented by a dot in the graph. ∗p < 0.05 by unpaired t-test.
(D) Representative HE images of pvPRAT, puPRAT, lPRAT and iBAT of 2-month-old C57BL/6J male mice in (A). Insets highlight the changes in lipid droplet size. Scale bar, 50µm.
(E-F) Representative immunofluorescence images of tdTomato (red) and UCP1 (green) expression in pvPRAT and puPRAT of 2-month-old Ucp1CreERT2;Ai14 male mice kept under room temperature (E) or cold exposure
(F) condition. Tamoxifen was injected at 1-month-old to trace the UCP1-expressing cells. tdTomato+/UCP1- cells represent the population that is whitened during the 1-to 2-month-old tracing period (white arrows). Tomato-/UCP1+ cells represent the recruited BAs (yellow arrows). Scale bar, 40µm.
(G-H) Quantification of the percentage of whitened adipocytes (tdTomato+/UCP1-) (G) and recruited BAs (tdTomato-/UCP1+) (H) in (E-F). n=3 mice for each tissue, represented by a dot in the graph. **, p < 0.01; ***, p<0.001; ****, p<0.0001 by two-way ANOVA.
(I-J) Representative immunofluorescence images of tdTomato (red) and UCP1 (green) expression in pvPRAT and puPRAT of 2-month-old PdgfraCreERT2;Ai14 male mice kept under room temperature (I) or cold exposure
(J) condition. Tamoxifen was injected at 1-month-old to trace the Pdgfra-expressing cells. Scale bar, 40µm.
(K) Quantification of the percentage of ASPC-derived BAs (tdTomato+/UCP1+) in (I-J). n=3 mice for each tissue, represented by a dot in the graph. **, p < 0.01; ***, p < 0.001 by two-way ANOVA.
(L) Illustration of the differential cellular contributions to the cold-recruited adipocytes in puPRAT. Cold exposure activated ASPC-derived BA adipogenesis and browning of whitened adipocytes, while preventing BA whitening. N, nucleus; M, mitochondria; LD, lipid droplet.
mPRAT-ad2 is the major cold-recruitable adipocytes in mPRAT
(A-B) Representative immunofluorescence images of tdTomato (red) and UCP1 (green) expression in pvPRAT and puPRAT of 7-month-old Ucp1CreERT2;Ai14 male mice. Tamoxifen was injected at 6 months old to trace the UCP1-expressing cells. tdTomato+/UCP1- cells represent the population that is whitened during the 6- to 7-month-old tracing period (white arrows). tdTomato-/UCP1+ cells represent the recruited BAs (yellow arrows). Scale bar, 40µm. RT, room temperature; CE, cold exposure.
(C) Quantification of the percentage of whitened BAs (tdTomato+/UCP1-) in (A-B). n=3 mice for each tissue, represented by a dot in the graph. *, p < 0.05; ****, p<0.0001 by two-way ANOVA.
(D-E) Integrated (D) and separated (E) UMAP of all cell types in mPRAT of 6-month-old C57BL/6J male mice kept under room temperature or cold exposure condition. A total of 5,636 nuclei were integrated, including 3,618 and 2,018 nuclei from 8 and 8 animals for the 2 conditions, respectively.
(F) Violin plot of the marker gene expression levels of each mPRAT adipocyte subpopulation under room temperature or cold exposure condition.
(G) DEGs of each adipocyte subpopulation under room temperature or cold exposure condition. The top 8 genes with the highest fold change (FC) were labelled.
(H-J) Gene regulatory network (GRN) and functional pathway enrichment comparison between room temperature and cold exposure conditions within each of the mPRAT-ad1, mPRAT-ad2 and mPRAT-ad3 subpopulations. Bar graphs illustrate the most enriched functional pathways according to the GO, KEGG, Reactome and WikiPathways databases.
mPRAT-ad2 adipocytes have different transcriptomes from iWAT BeAs
(A-B) Integrated (A) and separated (B) UMAP of the adipocytes in mPRAT and iWAT of 6-month-old C57BL/6J male mice kept under room temperature condition.
(C-D) Integrated (C) and separated (D) UMAP of the adipocytes in mPRAT and iWAT of 6-month-old C57BL/6J male mice kept under cold exposure condition. The grey-shaded oval was determined by the distribution of iWAT-ad1 population in (D).
(E) Heatmap illustrating the transcriptome of the BAs, BeAs and WAs of mPRAT, iBAT and iWAT under cold exposure condition. Overall similarity comparison was illustrated by hierarchical clustering.
(F-G) Cell-cell communication analysis by Cellchat in each adipocyte population of mPRAT, iBAT and iWAT under room temperature or cold exposure condition. The interaction number was labelled on corresponding linked lines in (F). The communication numbers and strength were shown as bar graphs in (G).
Tissue weight, quality control and additional analysis on the snRNA-seq data obtained from mPRAT and iBAT of 1-, 2-and 6-month-old C57BL/6J male mice, related to Figure. 1
(A) mPRAT and iBAT weight at 1-, 2- and 6-month-old of age. n=4-7 mice per time point, represented by a dot in the graph. ns, non-significant; **p < 0.01; ***p<0.001 by one-way ANOVA.
(B-C) Quality metrices showing UMI counts, number of detected genes and percentage of sequencing reads from mitochondria of the mPRAT (B) and iBAT (C) datasets.
(D, F) UMAP of integrated mPRAT (D) and iBAT (F) datasets. Same landscapes as in Fig. 1A for mPRAT and Fig. 1D for iBAT were used. Cells were colored based on different time points to indicate the overall unbiased integration by Seurat v4.
(E, G, H, L) Heat map of the top 20 marker gene expression levels of all cell types in mPRAT (E), iBAT (G), mPRAT adipocytes (H) and iBAT adipocytes (L).
(I) UMAP and cellular composition of the adipocyte subpopulations in iBAT of 1-, 2- and 6-month-old C57BL/6J male mice.
(J) Violin plot of the marker gene expression levels of each iBAT adipocyte subpopulation.
(K) Ucp1, Cidea, Cyp2e1 and Aldh1a1 expression pattern in the iBAT adipocyte subpopulations.
(M-N) Cell identity comparison by gene module analysis between iBAT-ad4 and the P4 subpopulation identified in Sun et al10 (M), and the mPRAT and iBAT adipocyte subpopulations (N). Top 50 marker genes from each reference cell type was selected as a module to calculate the module score of each target cell. Statistical differences in gene module scores were determined by one-way ANOVA and the corresponding p values were presented in the graph.
(O) Monocle 3 trajectory map of the iBAT adipocyte subpopulations.
Analysis on the ASPC population of mPRAT and iBAT and quality control of the lPRAT datasets, related to Figure. 1 and 2
(A, E) UMAP of the ASPC subpopulations in mPRAT (A) and iBAT (E) of 1-, 2- and 6-month-old C57BL/6J male mice.
(B, F) Violin plot of the marker gene expression levels of each ASPC subpopulation in mPRAT (B) and iBAT (F).
(C, G) Heat map of the top 20 marker gene expression levels of each ASPC subpopulation in mPRAT (C) and iBAT (G).
(D, H) Histogram illustrating the percentage of each ASPC subpopulation in mPRAT (D) and iBAT (H) relative to the total ASPC number.
(I) Merged clusters of mPRAT and iBAT ASPC populations.
(J) Violin plot of the marker gene expression levels of each ASPC subpopulation in merged mPRAT and iBAT ASPC populations.
(K) Histogram illustrating the percentage of each ASPC subpopulation in merged mPRAT and iBAT ASPC populations.
(L) Quality metrices showing UMI counts, number of detected genes and percentage of sequencing reads from mitochondria of the lPRAT datasets.
Difference in adipocyte composition between pvPRAT and puPRAT, related to Figure. 3
(A) Whole tissue immunostaining of UCP1 in PRAT of 6-week-old and 8-month-old C57BL/6J male mice using the ImmuView method. Scale bar, 1000µm. A, anterior; P, posterior; D, dorsal; V, ventral; L, lateral; M, medial.
(B) Illustration photos of the definition of pvPRAT and puPRAT.
(C-D) Ucp1CreERT2 labelled over 90% of the brown adipocytes in pvPRAT and puPRAT of 1- and 6-month-old mice. Tissues were collected one day after the last tamoxifen injection and all Tomato+ cells were stained by UCP1 antibody. n=3 mice, represented by a dot in the graph. Scale bar, 40µm.
(E) puPRAT had higher proportion of mPRAT-ad2 cells than pvPRAT. The Tomato-&CYP2E1-&BODIPY+ mPRAT-ad2 cells were highlighted by asterisks. n=3 mice, represented by a dot in the graph. Scale bar, 40µm.
mPRAT adipocytes are entirely derived from Pdgfra-expressing cells, related to Figure 4
pvPRAT and puPRAT of the PdgfraCre;Ai14 mice showed 100% overlap between tdTomato signal and PLIN1 staining. A total of 832 and 628 adipocytes in pvPRAT and puPRAT of two animals were examined. Scale bar, 40µm.
Additional analysis on mPRAT under room temperature and cold exposure conditions, related to Figure. 5
(A) Representative HE images illustrating the changes in lipid droplet size in pvPRAT, puPRAT, iBAT and iWAT of 6-month-old C57BL/6J male mice under room temperature and cold exposure conditions. Scale bar, 50µm, 5µm. RT, room temperature; CE, cold exposure.
(B) Quality metrices showing UMI counts, number of detected genes and percentage of sequencing reads from mitochondria of the mPRAT cold exposure dataset.
(C) UMAP of all cell types in mPRAT of 6-month-old C57BL/6J male mice under room temperature or cold exposure condition. A total of 7,387 nuclei were integrated, including 4,731 and 2,656 nuclei from 8 and 8 animals for the 2 conditions, respectively.
(D) Cellular composition of each cell type in (C) relative to the total number of analyzed nuclei.
(E) Heat map of the top 20 marker gene expression levels of all cell types of mPRAT under room temperature and cold exposure conditions.
(F) Cell identity comparison between mPRAT adipocyte subpopulations identified under the room temperature condition of Fig. 5E and Fig. 1G.
(G) Representative images of Tunel, Plin1 and DAPI staining of pvPRAT and puPRAT under room temperature or cold exposure condition. n=3 mice for each condition.
(H) UMAP and cellular composition of the ASPC subpopulations of mPRAT under room temperature and cold exposure conditions.
(I) Violin plot of the marker gene expression levels of each ASPC subpopulation of mPRAT under room temperature and cold exposure conditions.
snRNA-seq data analysis of iBAT under room temperature and cold exposure conditions, related to Figure. 6
(A) Quality metrices showing UMI counts, number of detected genes and percentage of sequencing reads from mitochondria of the iBAT cold exposure datasets of 6-month-old C57BL/6J male mice. CE, cold exposure.
(B) UMAP of all cell types in iBAT under room temperature and cold exposure conditions. A total of 27,717 nuclei were integrated, including 16,293 and 11,424 nuclei from 8 and 8 animals for the 2 conditions, respectively. RT, room temperature.
(C) Violin plot of the marker gene expression levels of all cell types in (B).
(D) Cellular composition of each cell type in (B) relative to the total number of analyzed nuclei. (E, H) UMAP and cellular composition of the adipocyte (E) and ASPC (H) subpopulations in (B).
(F, I) Violin plot of the marker gene expression levels of the adipocyte (F) and ASPC (I) subpopulations in (B).
(G) Cell identity comparison between iBAT adipocyte subpopulations identified in the room temperature condition of (E) and Supplementary Fig. 1I.
snRNA-seq data analysis of iWAT under room temperature and cold exposure conditions, related to Figure. 6
(A) Quality metrices showing UMI counts, number of detected genes and percentage of sequencing reads from mitochondria of the iWAT room temperature and cold exposure datasets of 6-month-old C57BL/6J male mice.
(B) UMAP of all cell types in iWAT under room temperature and cold exposure conditions. A total of 11,486 nuclei were integrated, including 7,342 and 4,144 nuclei from 6 and 6 animals for the 2 conditions, respectively.
(C) Violin plot of the marker gene expression levels of all cell types in (B).
(D) Heat map of the top 20 marker gene expression levels of all cell types of iWAT under room temperature and cold exposure conditions.
(E) Cellular composition of each cell type in (B) relative to the total number of analyzed nuclei.
(F, G, I) UMAP and cellular composition of the adipocyte (F, G) and ASPC (I) subpopulations in (B).
(H, J) Violin plot of the marker gene expression levels of the adipocyte (H) and ASPC (J) subpopulations in (B).
(K) Label transferred annotation of mPRAT-CE adipocytes from the 6-month-old mPRAT-RT dataset in Fig. 1G.
