**3. Results**

### *3.1. In the Presence of RV, iPSCs Maintain Pluripotent Properties and Lineage Identity after Initiation of Di*ff*erentiation*

Specification to one of the three germ lineages is the first critical step in directing di fferentiation to downstream cellular phenotypes. Therefore, directed as well as undirected di fferentiation was induced in RV-infected iPSCs which were subcultured for two to five passages. Passaging of infected iPSCs results in a homogenous level of infection within iPSC cultures without a ffecting the protein expression level of the pluripotency marker OCT4 [13]. During passaging of RV-infected iPSCs, replication occurred at a rather constant rate as assessed by viral titer and E1 protein expression rate [13]. Furthermore, passaging allows for adaptation of RV to iPSCs and excludes any possible effects of the di fferentiation process itself on the otherwise acute infection with RV (Figure 1A). Undirected di fferentiation is spontaneous and was thus induced to assess whether RV, without a specific di fferentiation stimulus, directs a gene expression profile di fferent from the mock-infected population. Directed di fferentiation of RV-infected iPSCs into ecto-, endo-, or mesodermal cells (thereafter referred to as RV-infected) was initiated with the STEMdi ff trilineage di fferentiation kit as an endpoint di fferentiation approach to determine which of the early cell fate decision pathways could be a ffected by RV.

RV establishes a noncytopathic infection of iPSCs with a homogenous distribution of infected cells within the respective colony (Figure 1B), [13]. Di fferentiation into all three embryonic germ layers supported RV replication at a comparable rate (Figure 1C [i,ii]), [13]. As a next step, we generated an expression heatmap of selected marker genes (based on microarray whole transcriptome data) for assessment of pluripotency and lineage identity (Figure 1D). In agreemen<sup>t</sup> with the maintenance of OCT4 (octamer-binding transcription factor 4, also known as POU5F1) expression in RV-infected iPSCs [13], high expression of pluripotency markers CDH1 and OCT4 was noted. Their expression was maintained to some degree in endodermal cells, which is in agreemen<sup>t</sup> with the conditions of the STEMdi ff trilineage di fferentiation kit. The same applies to the expression of the pluripotency marker SOX2 (SRY (Sex Determining Region Y)-Box 2) in ectodermal cells. The expression profile of lineage-specific markers confirmed ectodermal (PAX6 (Paired Box 6), DLK1 (Delta-Like 1 Homolog), and FABP7 (Fatty Acid Binding Protein 7)), mesodermal (HAND1 (Heart and Neural Crest-Derived Transcript 1), CDX2 (Caudal Type Homeobox 2), APLNR (Apelin Receptor)) and endodermal (LEFTY1 (Left-Right Determination Factor 1), EOMES (Eomesodermin), NODAL (Nodal Growth Di fferentiation Factor)) identity after initiation of directed differentiation in mock- and RV-infected iPSCs (Figure 1D). Additionally, some overlap between lineages, especially between mesoderm and endoderm, was noted for RV-infected samples (Figure 1D). Transcriptomic data was confirmed by RT-qPCR of pluripotency marker OCT4 and of selected markers for ectoderm (PAX6), mesoderm (HAND1), and endoderm (NODAL) lineages followed by relative quantification by qbase+ method (Figure 1E). Figure 1D and E indicate that during undirected differentiation, especially mesodermal markers were expressed, which occurred at a comparable level between mock- and RV-infected cells. Among the lineage-specific markers, the expression level of HAND1 was significantly downregulated in mesodermal cells after RV infection. Additionally, RV infection did not alter stemness-related expression signatures as indicated by the transcriptomic activity of the GO gene set telomere maintenance (Figure 1F). Telomere maintenance is active in stem cells, but gets deactivated in differentiated somatic cells [24].

In summary, comparable to the mock-control, RV maintained the pluripotent properties of iPSCs and enabled initiation of differentiation into embryonic germ layer cells as indicated by expression of essential germ layer markers.

### *3.2. High-Resolution Transcriptomic Maps Reveal Modules of Coregulated Genes Promoted by RV Infection during Endodermal Di*ff*erentiation*

As we found out that RV infection did not affect unspecific differentiation of iPSCs and enabled their lineage-specific differentiation, we wanted to focus on the effect of RV on lineage identity. The self-organizing map (SOM) transcriptome data portrayal provides a high-resolution visualization of the transcriptome landscape of each cell system studied in terms of a quadratic mosaic image and decomposes into clusters of coregulated genes. They are represented as colored spot-like areas where red and blue colors code activated and deactivated gene clusters, respectively. These transcriptomic portraits were then used to evaluate the mutual relatedness between the cell systems by means of a phylogenetic similarity tree (Figure 2A). The tree structure results from the fact that common and different spot patterns in the portraits reflect mutual similarities and differences of the activated cellular programs which enable judging the effect of RV-infection on the different lineages (see the portraits in Figure 2A). For an overview, we generated a spot-summary map in Figure 2B which shows the activated spots observed in any of the samples together with their functional context as extracted by means of gene set enrichment analysis of the genes in each of the spot-clusters of coexpressed genes (see also Supplement Table S1). In total, we identified five relevant spots labeled with capital letters A–E. Each of the spots is characterized by a specific expression profile (Supplement Figures S1 and S2) which, in turn, shows close similarities with the expression profiles of distinct gene sets (shown as 'barcode' plots in Figure 2C).

**Figure 1.** The identity of iPSCs and derived lineages was maintained in the presence of rubella virus (RV). (**A**) Overview of the methods applied to initiate differentiation in mock- and RV-infected iPSCs. (**A**) [**i**] Undirected differentiation in the respective induction medium started at the rim region of iPSC colonies (indicated by white arrows) and extended to their center over time of incubation. (**A**) [**ii**] Additionally, directed differentiation into the primary germ layers ectoderm, mesoderm, and endoderm was induced with the STEMdiff differentiation kit. (**B**) Immunofluorescence analysis with anti-E1 antibody (shown in red) was performed to monitor distribution of RV-positive cells within iPSC colonies. Nuclei are shown in blue. Ph, phase contrast (**C**) To assess RV replication in iPSCs and derived lineages, (**i**) virus progeny, and (**ii**) the amount of genomic viral RNAs was determined by standard plaque assay (n = 11 for passaged iPSCs, otherwise n = 3) and TaqMan-based reverse transcription-quantitative PCR (passaged iPSCs and ectodermal cells n = 3, mesodermal cells n = 7, endodermal cells n = 4), respectively. (**D**) Expression heatmap of selected marker genes of pluripotency and lineage identity (based on microarray whole transcriptome data) in mock- and RV-infected iPSCs and iPSC-derived lineages. Shading indicates overlap in the expression of some of the marker genes between iPSCs and iPSC-derived lineage cells, respectively. (**E**) The expression of indicated target genes in mock- and Wb-12-infected iPSCs and derived cells after initiation of undirected and directed differentiation was determined by real-time quantitative PCR (RT-qPCR) and analyzed by qbase+ software. For normalization, chromosome 1 open reading frame 43 (C1orf43) and hypoxanthine guanine phosphoribosyl transferase 1 (HPRT1) were used. Relative gene expression was calculated as normalized relative quantity (NRQ) and given as means ± SD (n = 3 to 5). As a reference, lineage-specific expression levels were assigned based on literature data and transcriptomics data by Stemcell Technologies for the Trilineage differentiation kit. (**F**) Analysis of stemness-related expression signatures was based on the GO gene set telomere maintenance. An embryonal stem cell signature and a gene set collecting OCT4 targets were both taken from [25].

**Figure 2.** RV supports expression of markers for definite endoderm. (**A**) Similarity tree of the gene expression portraits of the cell systems studied. Both, mock- and RV-infected mesodermal and ectodermal cells share relatively high mutual similarity of their transcriptomes. In contrast, mock- and RV-infected endodermal cells form a separate branch that is closer to the mock- and RV-infected iPSCs. (**B**) Spot summary map (**I**) provides an overview of activated cellular programs and their functional context, which is depicted in more detail in (**II**). (**C**) Barplot representation of the expression profiles of gene sets related to different functions. Their genes were enriched in the spots that were identified in the SOM portraits. They are thus indicated accordingly (see also the overview map in part B of the Figure and Table S2). (**D**) Selected marker genes for definite endoderm are illustrated in the heatmap of the transcriptome of mock- and RV-infected iPSCs and iPSC-derived lineages. (**E**) The expression of marker genes for definite endoderm was determined by RT-qPCR in mock- and Wb-12-infected iPSCs and iPSC-derived lineages by qbase+ software. For normalization, C1orf43 and HPRT1 were used. Relative gene expression was calculated as normalized relative quantity (NRQ) and given as means ± SD (n = 3 to 5). See also Figures S1 and S2.

Mapping of the position of genes into the SOM enables us to deduce their expression profile in the cell systems studied according to the gene's location in or near the spots and the respective spot profiles (Figure 2B and Supplement Figure S1). Importantly, key genes of ectoderm and mesoderm development are found in spots D and E, respectively, and are confirmed to be upregulated in the ectoderm and mesoderm cells. On the other hand, genes related to gastrulation and heart-tube development were enriched in spot B and found to be activated in RV-infected endoderm cells. This occurred together with genes that are activated upon interferon response (see next subsection). Stemness key genes locate in spot C together with part of the developmental genes of the endoderm (Supplement Figure S1). Di fference portraits clearly indicate that spot B associates with RV infection (Figure 2B), which contains the genes NODAL, CER1, SOX17, and GATA4, reflecting their activation upon RV infection in endodermal cells. In addition, we are able to show that mesoderm and ectoderm cells share similar expression patterns characterized by upregulated spots D and E and downregulated stemness genes (spot C), which, in contrast, are upregulated in endoderm cells. This is in agreemen<sup>t</sup> with the expression level of pluripotency genes in the endodermal lineage which was higher than in the remaining two embryonal germ layers (Figure 1D,E). As a consequence, mesoderm and ectoderm cells occupy neighboring positions at one end of the similarity tree, while endoderm cells are found at the opposite end. These findings highlight two important aspects: (I) there is a close similarity of transcriptional patterns among ecto-, mesoderm, and iPSC samples without and with RV infection; and (II) the endodermal lineage is an exception, where RV infection induces a notable shift away from the corresponding mock sample.

To further elucidate this specific e ffect of RV, we analyzed the transcriptomic data for markers for definitive endoderm (CXCR4 (C-X-C Motif Chemokine Receptor 4), MIXL1, SOX17, FOXA2 (Forkhead Box A2), EOMES, GATA6, CER1 (Cerberus, DAN family BMP Antagonist), and LEFTY, [26]. The expression heatmap shown in Figure 2D indicates an increased expression level of CER1 and SOX17 after RV infection. Figure 2E shows qPCR analysis followed by relative quantification by qbase+ method and highlights that the definite endoderm markers CXCR4, MIXL1, SOX17, GATA4/6, and CER1 were significantly higher expressed in RV-infected endodermal cells as compared to their mock-infected counterparts. While RV infection did not induce these markers above a cut-o ff of a two-fold increase during mesodermal and ectodermal di fferentiation, EOMES, GATA4, and CER1 were specifically induced by RV during undirected di fferentiation.

In conclusion, similarity analysis supports the hypothesis that ectodermal and mesodermal lineage identity was maintained after infection with RV, while endodermal cells derived from RV-infected iPSCs were enriched in markers for definitive endoderm.

### *3.3. RV Infection Activates IFN Type III Response Pathways on iPSCs and Derived Lineages*

Di fference portraits (Figure 3A) indicated that RV-infection specifically upregulated genes in spot B, which were associated with "IFN response" characteristics. This is supported by the profiles of gene expression signatures of viral infections such as by influenza virus and pneumonia that is accompanied by interferon activation. These genes were consistently upregulated in RV-infected samples with the largest observed e ffect in endodermal cells (Supplement Figure S3), [27–29]. Spot B highlighted in the SOM landscapes contained genes involved in IFN and viral response mechanisms (Figure 3A, a list of genes is given in Figure 3B). This is further emphasized by the expression heatmap shown in Figure 3C. Whereas genes involved in IFN-sensing, including the type III IFN receptor IFNLR1 (IFN lambda receptor 1), were not altered in their expression level, the IFN-signaling components STAT1 (signal transducer and activator of transcription 1) and IRF9 (interferon regulatory factor 9) appeared to be slightly upregulated at their mRNA expression level after infection with RV, especially in endodermal cells. The highest level of upregulation was found for IFN-stimulated genes (ISGs), notably for MX1 (MX dynamin like GTPase 1), IFITM2 (interferon induced transmembrane protein 2), and ISG15 (interferon-stimulated gene 15). Mapping of these genes into SOM space further underlines these findings: The IFN-signaling genes and the ISGs accumulate in and around spots B

and D, respectively, while IFN-sensing genes are located outside these spot regions (Figure 3B). The increase in the expression level of selected marker genes of the IFN pathway in the presence of RV was confirmed by RT-qPCR (Figure 3D). Compared to RV-infected iPSCs and ecto- and mesodermal cells, the expression level of IRF9 and STAT1 and selected ISGs (IFITM1/2, IFIT1, and ISG15) was significantly higher in RV-infected endodermal cells. The highest increase in mRNA expression after RV infection was noted for the ISGs IFIT1 and ISG15. Therefore, we determined whether this gene expression pattern was indeed associated with IFN generation during RV infection through quantification of type I (α and β), type II ( γ), and type III (λ1 and λ2/3) IFNs by the LEGENDplex assay from the supernatants of RV-infected cells (Figure 3E). In iPSCs as well as iPSC-derived lineage cells, RV infection induced secretion of type III IFNs, namely IFN λ2/3 (Figure 3E). As a positive control, the synthetic dsRNA analog poly I:C was used, which was either transfected into iPSCs or added directly to the supernatant. Either application of poly I:C did not lead to secretion of any type I, II, or III interferons (Figure 3E). The activation of type III IFNs by RV was also confirmed at the mRNA level by RT-qPCR (Figure 3F). Compared to the mock-infected control, RV induced a significant increase in the mRNA expression of IFN λ2/3 in endodermal cells (Figure 3F). Thereafter, we addressed the discrepancy between IFN λ2/3 protein (Figure 3E) and mRNA level (Figure 3F). Gene set analysis revealed that mRNAs associated with the KH type-splicing regulatory protein (KHSRP) were specifically enriched in endodermal cells (Figure 3G). KHSRP is involved in post-transcriptional regulation of mRNA expression, including IFN λ3 [30]. This could explain the discrepancy between IFN λ2/3 protein and mRNA expression level.

To address the influence of the type III IFNs secreted during RV to the cell culture supernatant on the gene expression landscape of iPSCs, type III IFNs were added exogenously for two weeks of cultivation during daily medium change of iPSCs. The zoom-in similarity tree shown in Figure 3H highlights the relatedness between the expression portraits of mock- and RV-infected iPSCs as well as iPSCs after application of type III IFNs. The gene expression profile of passaged RV-infected cells shifted away from iPSCs after exogenous IFN type III application, but closer to the mock-infected cells, suggesting an adaptation of RV to iPSCs.

In conclusion, in iPSCs and iPSC-derived embryonic lineages, RV infection induced a type III IFN response together with activation of ISGs, notably MX1 and ISG15. This activation appeared to be specifically profound in endodermal cells.

### *3.4. RV Infection Is Associated with Chromatin Remodeling*

Alterations of gene expression patterns during development are governed by epigenetic mechanisms in cooperation with regulation via transcription factor networks [32,33]. Particularly, we found that gene signatures of epigenetic impact, such as targets of the polycomb repressive complex 2 (PRC2), of H3K27me3, and of bivalently (H3K4me3 and H3K27me3) marked gene promoters, have an almost antagonistic expression profile as compared to the stemness signatures (Figure 4A, in comparison to Figure 1F).

**Figure 3.** RV infection activates an IFN response in iPSCs and derived lineages. (**A**) Expression portraits of the embryonal germ layers before and after RV infection and the respective difference portraits reveal characteristic spot patterns, where spots C, D, and E are specific for endoderm, ectoderm, and mesoderm cells, respectively. Spot B appears after RV infection mainly in endodermal cells. (**B**) IFN response genes with signaling and stimulated functions in the IFN response pathways accumulate in spots B and C and were upregulated predominantly in iPS and endodermal cells after RV infection as also indicated by the IFN and viral response gene signature profiles. (**C**) Expression heatmap of selected marker genes involved in IFN-sensing and -signaling in mock- and RV-infected iPSCs and iPSC-derived lineages. Interferon-stimulated genes (ISGs) that were identified by the SOM analysis shown in (**A**) are included. (**D**) The mRNA expression level of the IFN-signaling components IRF9 and STAT1 and indicated ISGs was verified by RT-qPCR analysis. Data are given as means ± SD (n = 3 and n = 5 for RV-infected mesoderm, IRF9 and STAT1). (**E**) The IFN profile for RV-infected iPSCs and derived lineages was determined by the LEGENDplex IFN panel for undiluted supernatants collected after five (iPSCs and mesodermal cells) and seven days (endodermal and ectodermal) of cultivation. (**F**) The mRNA expression level of type III IFNs was verified by qPCR analysis and given as means ± SD (n = 3). (**G**) Mean expression of a gene set (gene set Z-score, GSZ) that is controlled by KSRP, which appears to keep inflammatory gene expression within defined limits [31]. (**H**) Similarity tree of the gene expression portraits of mock- and RV-infected iPSCs in comparison to iPSCs after cultivation in the presence of exogenous type III IFNs. (**D**,**F**) For normalization of qRT-PCR data in the 2−ΔΔCt method, the HPRT1 gene was used. See also Figure S3.

**Figure 4.** RV infection is accompanied by an altered expression of components of the SWI/SNF and NURF chromatin remodeling complexes. (**A**) Mean GSZ expression signatures of stemness-related transcriptional programs which act via epigenetic programming (PRC2 targets, repressive and bivalent chromatin marks). (**B**) Expression heatmap of chromatin modifying enzymes in the cell systems studied. The gene expression data of methyltransferases (MTs) and demethylases (DMs) of DNA cytosines, histone lysine and arginine side chains were assigned to the expression spot-cluster A–E according to their expression profiles. This suggests their involvement in the regulation of chromatin structure as writers and erasers of methylation marks at histone lysine and arginine side chains and affecting DNA methylation. Notably, a very strong variability was observed for KDM6a (alias UTX), a constituent of the SWI/SNF ATP-dependent chromatin remodeling machinery. (**C**) Expression of KDM6a directly relates to stemness programs (Figure 1F) and inversely relates to programs repressing stemness functions (part A of the Figure). See also Figure S4.

Next, we focused on chromatin remodeling which is essential for lineage segregation [9,34]. The analysis of transcription factor networks that act in regions of euchromatin with transcriptionally active genes and regulatory elements revealed that genes in repressed and bivalent states of endoderm progenitors become more rigorously deactivated in RV-infected endoderm cells than genes from these states of mesoderm progenitors in RV-infected mesoderm cells (Supplement Figure S4A). This suggests that RV infection specifically dedi fferentiates endoderm cells by suppressing developmental suppressors. We then studied enzymes a ffecting methylation of DNA and of arginine and lysine side chains of histones such as H3K4, H3K9, H3K27, and H3K36, with potential impact on chromatin structure. The profiles of the DNA methylation maintenance methyltransferase DNMT1, of PRMT6, a methyltransferase of histone arginine side chains, and of JMJD1c, a H3K9 demethylase, and KDM5b demethylating H3K4me3 correlate with the 'stemness' spot cluster C upregulated in iPS and endodermal cells.

Notably, the gene encoding the H3K4 demethylase KDM6a (alias UTX) was markedly upregulated in RV-infected iPSCs and, especially, endoderm-derived cells (Figure 4B). KDM6a is a constituent of the SWI/SNF ATP-dependent chromatin remodeling machinery. Figure 4C highlights that in comparison to iPSCs and ecto- and mesodermal cells, the upregulation of KDM6a was highest in endodermal cells. The alterations of its expression sugges<sup>t</sup> its role in chromatin remodeling after RV infection described above. This motivated us to estimate the expression patterns of other genes encoding components of the SWI/SNF and of the NURF chromatin remodeling complexes [9] by mapping them into the SOM (Supplement Figure S4B). We found that, indeed, Smarcc2, Smarcd3, and Btpf were all upregulated in endoderm-derived cells after RV infection, which further supports the assumption that SWI/SNF and possibly also NURF contribute to chromatin remodeling during RV infection. In conclusion, expression changes of di fferent sets of genes involved in epigenetic regulation and of constituents of the ATP-dependent chromatin remodeling complexes such as KDM6a-UTX were detected in association with RV-infection, especially during endodermal di fferentiation.

### *3.5. RV Infection Impairs Aggregation of iPSCs into Embryoid Bodies*

The progression of embryogenesis does not only involve the activation of developmental pathways, but also requires cell–cell interactions based on adhesive forces [35]. The relevance of these observations for RV-infected iPSCs was emphasized by gene ontology analysis regarding focal adhesion and regulation of cell adhesion (Figure 5A). Transcriptomic analysis revealed that THY-1 (also known as cluster of di fferentiation (CD) 90) was among the targets a ffected by RV (Figure 5B). The relevance of THY-1 (CD90) for cellular adhesion capacity was highlighted for CD90 negative carcinoma, which compared to their CD90 positive counterparts lack the ability to form spheres [36]. Accordingly, we have addressed whether RV alters the spontaneous aggregation capacity of iPSCs into 3D aggregates called embryoid bodies (EBs). EB formation relies on cell–cell adhesive interactions [35]. Compared to the mock control, EBs generated from RV-infected iPSCs were reduced in diameter and of irregular shape (Figure 5C). Furthermore, during cultivation they lost stability and small-sized debris was generated (Figure 5C). In contrast to RV-infected iPSCs, the mock-infected controls generated viable EBs as indicated by staining with calcein performed after two weeks of cultivation (Figure 5D). In conclusion, RV infection impaired the adhesion capacity of iPSCs as shown by their reduced ability to assemble into EBs. This suggests an impaired cell–cell interaction capacity during lineage segregation.

**Figure 5.** RV impairs the cellular adhesion capacity of iPSCs. (**A**) Gene signatures related to focal adhesion in mock- and RV-infected iPSCs. (**B**) Expression heatmap of selected marker genes involved in cellular adhesion. (**C**) Assessment of 3D stability through embryoid body (EB) formation. Shown are images before and after cultivation in suspension. (**D**) To verify viability, EBs were stained with calcein after cultivation for two weeks.

### *3.6. RV Infection Specifically A*ff*ects Developmental Pathways during Endodermal Di*ff*erentiation*

For assessment of the effect of RV on global cellular signaling networks, we focused on two important signaling pathways, namely transforming growth factor β (TGF-β) and Wnt/β-catenin (Wnt), (Figure 6A,B, respectively). The TGF-β signaling pathway is involved in cell growth and differentiation during embryogenesis [37]. The Wnt signaling pathway regulates the interaction between cellular pathways involved in primary germ layer formation and is required for mesodermal differentiation from pluripotent stem cells [38]. A more detailed view of the TGF-β and Wnt signaling pathways is provided in Supplementary Figures S5 and S6, respectively. As expected, the highest TGF-β signaling pathway activity was observed in endodermal cells, while the Wnt signaling pathway was most active in mesodermal cells (Supplement Figures S5 and S6). Thus, these two lineages were depicted in Figure 6A,B, respectively, to highlight the effect of RV on their activity in comparison to the respective controls. Within the TGF-β signaling pathway, mock-infected endodermal cells show high cell cycle activity induced by CDKN2B and its downstream interaction partners, which became deactivated during RV infection (Figure 6A and Supplement Figure S5). Additionally, RV infection in endodermal cells was specifically accompanied by a strong activation of NODAL, an essential component of the TGF-β signaling pathway (Figure 6A and Supplement Figure S5). During RV infection, the transcriptional activity of Wnt signaling pathway was reduced in mesodermal cells (Figure 6B), whereas for ectodermal and endodermal cells, almost no alteration in its activity was detected (Supplement Figure S6).

**Figure 6.** RV exerts lineage-specific effects on developmental signaling pathways and alters expression of transcription and growth factors. Pathway signal flow (PSF) activity plot of (**A**) the TGF-β signaling pathway in endodermal cells (highlighted are genes with higher (NODAL, ACVR2A, Myc) and lower (RBL1 and E2F4) activity after RV infection) and (**B**) of the Wnt signaling pathway in mesodermal cells (highlighted are genes with lower (the CSNK1E/AXIN1E and the LEF1/CCND1 axis) activity after RV infection). The calculation of the activity of the nodes was based on the PSF algorithm using the respective gene expression values and the wirings between the nodes [21]. (**C**) Selected genes within pathways that were specifically affected by RV infection are illustrated in the heatmap of the transcriptome of mock- and RV-infected iPSCs and iPSC-derived lineages. (**D**) For qRT-PCR expression analysis, the 2−ΔΔCt method based on normalization to HPRT1 gene was used. Values are given as means ± SD (n = 4 for Wb-12-infected ectoderm, n = 2 for HPV77-infected ectoderm, otherwise n = 3). See also Figures S5–S7.

As congenital rubella leads to defects in heart and eye development, we analyzed the impact of RV infection on the underlying molecular pathways. For members of the gene annotation embryonic

heart tube development and the gene set heart morphogenesis only a slight effect of RV infection was noted (Supplement Figure S7), suggesting the involvement of other factors. In mesodermal cells, expression of HAND1, which is involved in embryonic heart tube development, was reduced after RV infection as compared to the mock controls (Figures 6C and 1E). The ectoderm gives rise to components of the eye. At the molecular level, RV infection of ectodermal cells impaired the gene set eye development (Figure 6C), (Supplement Figure S7). Specifically, SIX3 and SIX6, as key transcription factors for mammalian eye development [39,40], were reduced in their expression level (Figure 6C). Among others, SIX3, together with RAX, initiates transcription of genes required for lens placode formation [39].

To further assess developmental pathways with relevance for the teratogenic outcome of RV infection, we determined the mRNA expression of RAX and SIX3 (as important factors for eye development) besides FGF17 (Fibroblast Growth Factor 17) and SOX17 (as contributing factors for endodermal differentiation and cardiovascular development) by RT-qPCR in ectodermal and endodermal cells, respectively (Figure 6D). Here, the vaccine strain HPV77 was used in addition to Wb-12 strain. Attenuated vaccine strains such as HPV77 are not teratogenic as revealed after immunization of unknowingly pregnan<sup>t</sup> women [4]. Thus, any alteration at the molecular level that is present during wild-type Wb-12, but not HPV77 infection, emphasizes its possible contribution to congenital rubella. In comparison to the mock control, a similar reduction in the expression of RAX and SIX3 was detected in ectodermal cells after infection with both RV strains. However, a different picture emerged for FGF17 and SOX17. Figure 6D [i] highlights an increase in the expression of FGF17 and SOX17, which was significant for FGF17 compared to the mock control, but only for endodermal cells derived from Wb-12-infected iPSCs, not for endodermal cells derived from HPV77-infected iPSCs. Moreover, the increase in the expression of the definitive endoderm markers CER and GATA6 after infection with Wb-12 as shown in Figure 2E was not detected after infection with HPV77 (Figure 6D [ii]). However, both RV strains induced a significant increase of the IFN-signaling component IRF9 (Figure 6D [ii]). The endoderm plays an essential role in the crosstalk between the lineages and contributes to the epithelial lining of many organs, including the vascular network. Accordingly, the gene set vasculogenesis, but not angiogenesis, was affected by RV infection (Supplement Figure S7). This emphasizes our notion on the correlation between the impact of RV infection on endodermal cells and congenital rubella.

In summary, specific signatures including the TGF-β signaling pathway were affected by RV infection, but in a lineage-specific manner. In ectodermal cells, RV infection significantly reduced expression of SIX3 as key transcription factors for eye field development. Only for the clinical isolate Wb-12, but not for the vaccine strain HPV77, was an impact on the growth factor FGF17 and the endodermal transcription factor SOX17 noted.

Figure 7 summarizes the findings of this study in correlation to the main CRS symptoms. The noncytolytic course of infection of RV during directed differentiation is in agreemen<sup>t</sup> with its persistence in multiple organs and tissues during congenital rubella. We have not identified any indication at the molecular level that could contribute to the defects in ear development during congenital rubella. However, sensorineural deafness is often a late-onset symptom and could be associated with pathological alterations in the brain of the infected infants [3].

**Figure 7.** Graphical summary of the identified molecular alterations induced by RV during directed differentiation of iPSCs into the three embryonic germ layer cells. The data was set in a possible relation to the prevailing symptoms of congenital rubella embryopathy. Especially endodermal cells were characterized by profound alterations in their gene expression landscape, including the expression of markers for definitive endoderm and epigenetic factors. This could impair the crosstalk between endodermal and mesodermal cells during differentiation.
