RNA Sensing and Innate Immunity Constitutes a Barrier for Interspecies Chimerism

Abstract

Interspecies chimera formation with human pluripotent stem cells (PSCs) holds great promise to generate humanized animal models and provide donor organs for transplant. However, the approach is currently limited by low levels of human cells ultimately represented in chimeric embryos. Different strategies have been developed to improve chimerism by genetically editing donor human PSCs. To date, however, it remains unexplored if human chimerism can be enhanced in animals through modifying the host embryos. Leveraging the interspecies PSC competition model, here we discovered retinoic acid-inducible gene I (RIG-I)-like receptor (RLR) signaling, an RNA sensor, in "winner" cells plays an important role in the competitive interactions between co-cultured mouse and human PSCs. We found that genetic inactivation of Ddx58/Ifih1-Mavs-Irf7 axis compromised the "winner" status of mouse PSCs and their ability to outcompete PSCs from evolutionarily distant species during co-culture. Furthermore, by using Mavs-deficient mouse embryos we substantially improved unmodified donor human cell survival. Comparative transcriptome analyses based on species-specific sequences suggest contact-dependent human-to-mouse transfer of RNAs likely plays a part in mediating the cross-species interactions. Taken together, these findings establish a previously unrecognized role of RNA sensing and innate immunity in "winner" cells during cell competition and provides a proof-of-concept for modifying host embryos, rather than donor PSCs, to enhance interspecies chimerism.

Extended Data Fig. 1 |. Comparative RNA-seq analysis between co-cultured and separately cultured mEpiSCs.

a-c, KEGG pathways enriched on days 1 (a), 2 (b), and 3 (c) co-URGs in mEpiSCs. Top 15 (with the lowest P values) KEGG pathways are shown. d-f, Volcano plots showing significantly upregulated (red) and downregulated (blue) genes in co-cultured versus separately cultured mEpiSCs on days 1 (d), 2 (e) and 3 (f). RLR-pathway-related genes are highlighted. g, Volcano plot showing no differentially expressed genes in transwell co-cultured versus separately cultured mEpiSCs. h, KEGG pathway analysis of upregulated genes in mEpiSCs co-cultured with P65^KO hiPSCs (versus separate cultures). The color of each dot represents log₁₀-transformed P value of each term, while the dot size represents the number of co-URGs in each term. Terms that are not significantly enriched are shown in gray. i, Heatmap showing fold changes of RLR-pathway-related genes in mEpiSCs co-cultured with P65^KO hiPSCs (versus separate cultures). Blue and red represent log₂-transformed fold changes < 0 and > 0, respectively. Asterisks indicate statistically differentially expressed.

Extended Data Fig. 2 |. Comparative RNA-seq analysis between co-cultured and separately cultured H9 hES cells.

a, Venn diagram showing the numbers of co-URGs in H9 hES cells. b, KEGG pathway analysis of co-URGs in H9 hES cells. The color of each dot represents log₁₀-transformed P value of each term, while the dot size represents the number of co-URGs in each term. Terms that are not significantly enriched are shown in gray. c, Heatmap showing fold changes of RLR-pathway-related genes in co-cultured versus separately cultured H9 hES cells. Blue and red represent log₂-transformed fold changes < 0 and > 0, respectively. d-f, KEGG pathways enriched on days 1 (d), 2 (e), and 3 (f) co-URGs in H9 hES cells. Top 15 (with the lowest P values) KEGG pathways are shown. g-i, Volcano plots showing significantly upregulated (red) and downregulated (blue) genes in co-cultured versus separately cultured H9 hES cells on days 1 (g), 2 (h) and 3 (i). RLR-pathway-related genes are highlighted.

Extended Data Fig. 3 |. Genetic inactivation of Mavs in mEpiSCs compromises winner cell competitiveness.

a, Schematic of mouse Mavs gene targeting with two sgRNAs and homozygous 1.6-kb deletion (233-bp out-of-frame deletion in CDS) in mEpiSC clone C2. Bold, PAM sequence. b, Western blot confirmed the lack of MAVS protein expression in several independent Mavs^KO mEpiSC clones. Tubulin was used as a loading control. c, Representative brightfield and immunofluorescence images showing long-term cultured Mavs^KO mEpiSCs maintained stable colony morphology and expressed core (OCT4, red; SOX2, green) pluripotency markers. Blue, DAPI. Scale bars, 100 μm. d, Growth dynamics of human-mouse 3-day-co-culture assay. Left, growth curves of H9 hES cells in co-culture with WT mEpiSCs (red), co-culture with Mavs^KO mEpiSCs (green) and separate culture (blue). Right, growth curves of WT mEpiSCs in co-culture (red) with H9 hES cells and separate culture (blue), Mavs^KO mEpiSCs in co-culture (green) with H9 hES cells and separate culture (orange). n = 4, biological replicates. Data are mean ± s.e.m. P values determined by one-way ANOVA with Dunnett’s multiple comparison. e, Representative fluorescence images of day-3 co-cultured and separately cultured WT mEpiSCs (green), Mavs^KO mEpiSCs (green) and H9 hES cells (red) in MEF-coated micropattern wells. White, DAPI. Scale bar, 100 μm. f, Dot-plot showing numbers of live H9 hES cells in separate culture (n = 6), co-culture with WT mEpiSCs (n = 5) and co-culture with Mavs^KO mEpiSCs (n = 5) on day 3 in micropatterned wells. n, biological replicates.

Extended Data Fig. 4 |. Genetic inactivation of Ddx58, Ifih1 and Irf7 in mEpiSCs compromise winner cell viability during human-mouse PSC co-culture.

a, Upper, schematic of mouse Ddx58 gene targeting with two sgRNAs and homozygous 1.5-kb deletion (139-bp out-of-frame deletion in CDS) in Ddx58^KO clone. Middle, schematic of mouse Ifih1 gene targeting with one sgRNA and homozygous 64-bp out-of-frame deletion in Ifih1^KO clone. Lower, schematic of mouse Irf7 gene targeting with two sgRNAs and homozygous 655-bp deletion (305-bp out-of-frame deletion in CDS) in Irf7^KO clone. Bold, PAM sequence. b, Representative brightfield and immunofluorescence images showing long-term cultured Ddx58^KO, Ifih1^KO and Irf7^KO mEpiSCs maintained stable colony morphology and expressed core (OCT4, red; SOX2, green) pluripotency markers. Blue, DAPI. Scale bars, 100 μm. c, Growth curves of co-cultured and separately cultured WT, Ddx58^KO, Ifih1^KO and Irf7^KO mEpiSCs. n = 6 (WT), n = 3 (Ddx58^KO), n = 3 (Ifih1^KO) and n = 3 (Irf7^KO) independent culture experiments. Data are mean ± s.e.m. P values determined by unpaired two-tailed t-test.

Extended Data Fig. 5 |. Conserved role of Mavs during competitive interactions of mEpiSCs with primed PSCs from other species.

a, Left, growth curves of rhesus ES cells (ORMES23) in co-culture with WT mEpiSCs (red), co-culture with Mavs^KO mEpiSCs (green) and separate culture (blue). Right, growth curves of WT mEpiSCs in co-culture (red) with rhesus ES cells (ORMES23) and separate culture (blue), Mavs^KO mEpiSCs in co-culture (green) with rhesus ES cells (ORMES23) and separate culture (orange). n =3, biological replicates. All data are mean ± s.e.m. P values determined by one-way ANOVA with Dunnett’s multiple comparison. b, Representative fluorescence images of day-4 co-cultured and separately cultured WT mEpiSCs (red), Mavs^KO mEpiSCs (red) and rhesus ES cells (ORMES23) (green). Scale bar, 200 μm. c, Representative fluorescence images of day-4 co-cultured and separately cultured WT mEpiSCs (red), Mavs^KO mEpiSCs (red) and bovine ES cells (green). Scale bar, 200 μm. d, RT-qPCR analysis of relative expression levels of RLR-pathway-related genes in days 1, 2 and 3 co-cultured (with bovine ES cells) Mavs^KO mEpiSCs compared to separately cultured Mavs^KO mEpiSCs. n = 3, biological replicates. Data are mean ± s.e.m. P values determined by unpaired two-tailed t-test. e, Representative fluorescence images of E8.5–9.5 Mavs^KO mouse embryos showing eGFP signal after blastocyst injection of eGFP-labeled hiPSCs and embryo transfer. Scale bars, 50 μm. f, Representative immunofluorescence images showing contribution of eGFP-labeled hiPSCs in E8.5–9.5 mouse embryos. Embryo sections were stained with antibody against eGFP and DAPI. Scale bars, 50 μm.

Extended Data Fig. 6 |. Presence of human transcripts in co-cultured mEpiSCs revealed by bulk RNA sequencing.

a, Statistics of bulk RNA-seq reads in mEpiSCs that could map to mouse genome. b, Statistics of bulk RNA-seq reads in H9 hES cells that could map to human genome. For a and b, Sep, Co-d1, Co-d2, Co-d3, Sep-T and Co-T represent separate culture, co-culture on days 1, 2, 3, separate culture in transwell and co-culture in transwell, respectively. c, Mapping rates of human genome in mEpiSCs co-cultured with P65^KO hiPSCs determined by bulk RNA-seq. Asterisks indicate statistically significant differences: (***) P < 0.001; Fisher’s exact test. d, Schematic of human and mouse PSC co-culture in transwells and the following RNA-seq experimental setup. e, Normalized eGFP expression in mEpiSCs determined by bulk RNA-seq. Asterisks indicate statistically significant differences: (***) P < 0.001; Fisher’s exact test.

Extended Data Fig. 7 |. Presence of RNAs encoding human repeat elements in co-cultured mEpiSCs revealed by bulk RNA sequencing.

a, Normalized abundance of human repeat elements in H9 hES cells (left) or mEpiSCs (right) aligned to the human genome (assembly HG38). Shown are repeat elements expressed in separately cultured H9 hES cells (Normalized Expression >20) and not detected in separately cultured mEpiSCs (Normalized Expression <20). Only reads uniquely assignable to a single repeat element type are shown. Heat maps were sorted on expression from high to low in H9 hES cells. b, Normalized abundance of near-consensus human repeat elements in H9 hES cells (left) or mEpiSCs (right). Shown are least divergent repeat elements (full-length relative to consensus, <10 mismatches or indels) that were expressed in separately cultured H9 hES cells (Normalized Expression >20) and were not detected in separately cultured mEpiSCs (Normalized Expression <20). Heat maps were sorted on expression from high to low in H9 hES cells. c-f, Expression of human repeat element classes (c), retroelement families (d), Alu (e) and evolutionarily young LINE1 (f) elements in H9 hES cells under different culture conditions. Sep, Co-d1, Co-d2, Co-d3, Sep-T and Co-T represent separate culture, co-culture on days 1, 2, 3, separate culture in transwell and co-culture in transwell, respectively.

Extended Data Fig. 8 |. Presence of human transcripts in co-cultured mEpiSCs revealed by single cell RNA sequencing.

a, Violin plots showing the number of genes (left), UMIs (middle) and the percentage of mitochondria genes (right) per cell. b, UMAP visualization showing eGFP⁺ (red) and eGFP⁻ (gray) mEpiSCs in separate culture (left) and co-culture (right). c, Proportion of eGFP⁺ mEpiSCs determined by scRNA-seq. Asterisks indicate statistically significant differences: (***) P < 0.001; Fisher’s exact test. Cross-species doublets served as a negative control. d, Cumulative distribution showing proportion of eGFP⁺ mEpiSCs with different thresholds. Asterisks indicate statistically significant differences: (**) P < 0.01; (***) P < 0.001; Fisher’s exact test; (N.S.) not significant. e, Scatterplot showing global expression correlation of human genes detected in mEpiSCs and H9 hES cells determined by bulk RNA-seq and scRNA-seq. X-axis represents Pearson Correlation Coefficients whereas Y-axis represents P values. Triangles represent bulk RNA-seq datasets (indicated in the plot) or single-cell datasets. All datasets are summarized and shown as density plots along each axis.

Fig. 1 |. RLR signaling activation in mEpiSCs co-cultured with H9 hES cells.

a, Schematic of human and mouse PSC co-culture and RNA-seq experimental setup. b, Left, Venn diagram showing the numbers of co-URGs in mEpiSCs. Right, KEGG pathway analysis of co-URGs in mEpiSCs. The color of each dot represents log₁₀-transformed P value of each term, while the dot size represents the number of co-URGs in each term. Terms that are not significantly enriched are shown in gray. c, Heatmap showing fold changes of RLR-pathway-related genes in co-cultured versus separately cultured mEpiSCs. Blue and red represent log₂-transformed fold changes < 0 and > 0, respectively. Asterisks indicate statistically differentially expressed. d, RT-qPCR analysis of relative expression levels of RLR-pathway-related genes in co-cultured WT mEpiSCs compared to separately cultured WT mEpiSCs. n = 3, biological replicates. Data are mean ± s.e.m. P values determined by unpaired two-tailed t-test. e, Native gel electrophoresis detecting IRF3 dimerization in WT mEpiSCs transfected with poly (I:C), co-cultured and separately cultured WT mEpiSCs, co-cultured and separately cultured Mavs^KO mEpiSCs. f, RT-qPCR analysis of relative expression levels of RLR-pathway-related genes in co-cultured Mavs^KO mEpiSCs compared to separately cultured Mavs^KO mEpiSCs. n = 3, biological replicates. Data are mean ± s.e.m. P values determined by unpaired two-tailed t-test.

Fig. 2 |. Compromised cell competitiveness and viability of Mavs-deficient mEpiSCs.

a, Left, growth curves of H9 hES cells in co-culture with WT mEpiSCs (red), co-culture with Mavs^KO mEpiSCs (green) and separate culture (blue). Right, growth curves of WT mEpiSCs in co-culture (red) with H9 hES cells and separate culture (blue), Mavs^KO mEpiSCs in co-culture (green) with H9 hES cells and separate culture (orange). n = 6, biological replicates. b, Representative fluorescence images of day-3 co-cultured and separately cultured WT mEpiSCs (red), Mavs^KO mEpiSCs (red) and H9 hES cells (green). Scale bar, 200 μm. c, Growth curves of co-cultured WT, Mavs^KO, Ddx58^KO, Ifih1^KO and Irf7^KO mEpiSCs normalized to corresponding separate cultures. n = 6 (WT), n = 6 (Mavs^KO), n = 3 (Ddx58^KO), n = 3 (Ifih1^KO) and n = 3 (Irf7^KO) independent culture experiments. d, Left, growth curves of bovine ES cells in co-culture with WT mEpiSCs (red), co-culture with Mavs^KO mEpiSCs (green) and separate culture (blue). Right, growth curves of WT mEpiSCs in co-culture (red) with bovine ES cells and separate culture (blue), Mavs^KO mEpiSCs in co-culture (green) with bovine ES cells and separate culture (orange). n =3, biological replicates. All data are mean ± s.e.m. P values determined by one-way ANOVA with Dunnett’s multiple comparison.

Fig. 3 |. Genetic inactivation of Mavs in mouse embryos improves donor human cell survival during chimera formation.

a, Schematic showing the generation of ex vivo and in vivo human-mouse chimera using Mavs-deficient mouse blastocysts. b, Representative brightfield and fluorescence merged images of WT and Mavs^KO mouse embryos cultured for 3 days (d3) and 5 days (d5) after blastocyst injection with hiPSCs. Scale bar, 100 μm. c, Line graphs showing the percentages of eGFP+ mouse embryos at indicated time points during ex vivo culture after injecting hiPSCs into WT and Mavs^KO blastocysts. n = 3 (WT) and n = 3 (Mavs^KO) independent injection experiments. Data are mean ± s.e.m. P values determined by unpaired two-tailed t-test. d, Dot plot showing the percentages of eGFP+ E8.5–9.5 mouse embryos derived from injecting hiPSCs into WT and MavsKO blastocysts. Each blue dot represents one embryo transfer experiment. n = 4 (Mavs^+/+), n = 4 (Mavs^+/−) and n = 7 (Mavs^−/−), independent experiments. e, Representative immunofluorescence images showing contribution of eGFP-labelled hiPSCs in E9.5 mouse embryos. Embryo sections were stained with antibody against eGFP and DAPI. Scale bars, 500 μm (whole embryo) and 50 μm (insets). f, Genomic PCR analysis of E8.5–9.5 mouse embryos derived from injecting hiPSCs into WT and Mavs^KO blastocysts. TPA25-Alu denotes a human-specific primer; PTGER2 was used as a loading control. HFF, HFF-hiPS cells. NTC, non-template control. This experiment was repeated independently three times with similar results.

Fig. 4 |. Presence of human transcripts in co-cultured mEpiSCs revealed by RNA sequencing.

a, Computational pipeline to identify species-specific bulk/single-cell RNA-seq reads. See Methods for details. b, Mapping rates of human genome in mEpiSCs determined by bulk RNA-seq. Asterisks indicate statistically significant differences: (***) P < 0.001; Fisher’s exact test. c, Heatmap showing global expression profiles of human genes expressed in H9 hES cells and detected in mEpiSCs. Genes are sorted by expression levels from high to low in H9 hES cells. Log₂-transformed expression levels were used to draw this plot. For b and c, Sep, Co-d1, Co-d2, Co-d3, Sep-T and Co-T represent separate culture, co-culture on days 1, 2, 3, separate culture in transwell and co-culture in transwell, respectively. d, Clustering of separately cultured and co-cultured mEpiSCs by scRNA-seq analysis. Each point represents one single cell, colored according to cluster. e, Percentage of separately cultured and co-cultured mEpiSCs in each cluster. f, Mapping rates of human genome in mEpiSCs determined by scRNA-seq. Asterisks indicate statistically significant differences: (***) P < 0.001; Wilcoxon test. Cross-species doublets served as a negative control. g, UMAP visualization showing mapping rates of human genome in separately cultured (left) and co-cultured (right) mEpiSCs.