**4. Discussion**

Knowledge on developmental signaling networks is an essential prerequisite to understand congenital abnormalities, either caused by pathogenic, hereditary or environmental risk factors. Models for developmental toxicity testing range from iPSCs to iPSC-derived EBs and three-dimensional organoids. They have different properties regarding high-throughput screening capacity and relevance for in vivo developmental processes [41]. Their proper assessment requires compounds or pathogens with well-known symptoms arising from embryotoxic or teratogenic alterations during embryonal development. Here, we used RV to correlate clinical observations for congenital rubella syndrome with its impact on the differentiation capacity of iPSCs. Although iPSC-based cell culture models reflect only transient stages during human embryogenesis, they allow us to recapitulate essential developmental pathways that are otherwise inaccessible [42].

Among human pathogens, RV is rather exceptional in its ability to replicate noncytopathically in iPSCs, which in general represent a rather restrictive environment to most viral infections [43]. The protection of human development from a pathogenic insult involves several mechanisms, including transcriptional silencing of viruses in pluripotent stem cells [44] and an intrinsic high expression level of IFN-induced genes [14]. This includes interferon-induced transmembrane protein 1 (IFITM1) and its capacity to restrict the potentially harmful reactivation of human endogenous retroviruses [43]. Otherwise, the antiviral innate immune response in iPSCs is rather refractive [15]. The constitutive overexpression of an active IRF7 as a master regulator of the type I IFN system revealed the harmful effects an activated type I IFN response would have on the expression of pluripotency and lineage specific genes, especially of endodermal cells [45]. In contrast to the engineered type I IFN response in iPSCs through overexpression of IRF7, no morphological changes were noted after infection of iPSCs with RV [13]. However, in agreemen<sup>t</sup> with the study on the effect of type I IFNS on differentiation capacity of iPSCs [45], the impact of RV on directed differentiation was most profound during endodermal differentiation. The differences in the signaling cascades of type I and III IFNs [46] might explain the milder effects noted after RV-associated type III IFN activation as compared to the severe effects of an engineered type I IFN response [45]. Our data complements a recent study on the impact of Influenza A virus (IAV) on the pluripotency and proteome of hiPSCs [47]. Whereas, in contrast to RV,

IAV reduces the pluripotency of iPSCs, both virus infections induce ISG15 and IFN λ1 [47], highlighting this observation as an innate immune mechanism that is already developed in iPSCs. Further studies need to address whether the impact of RV infection on endodermal di fferentiation is correlated with the activation of the type III IFN signaling pathway and how this a ffects the course of infection of RV in iPSCs. In ectodermal cells, RV infection was associated with the downregulation of SIX3, an essential transcription factor for early eye development [48]. Together with SIX6, SIX3 suppresses Wnt signaling, which could contribute to the slight activation of this essential developmental signaling pathway in ectodermal cells derived from RV-infected iPSCs [40]. Their functional importance during retinal development and eye field specification was recently shown by the use of iPSC-derived retinal organoids [39]. Our study complements a previous study on the gene expression profile of fetal (HUVEC originating from umbilical cord veins) and adult (HSaVEC derived from the saphenous vein) endothelial cells which revealed a specific enrichment of 18 downregulated genes within the GO terms "sensory organ development", "eye development", and "ear development" [49].

Among the embryonic germ layers, especially di fferentiation to definite endoderm appeared to be a ffected by RV infection. In addition to its role in formation of organs of the digestive tract, the interaction of endodermal cells with precardiac mesoderm drives specification and di fferentiation of cardiac myocytes and cells of endocardial endothelium [50]. This is supported by studies on the contribution of signals from endodermal cells and the interactive crosstalk between the endoderm and mesoderm to di fferentiation of ESCs to a cardiomyogenic lineage [9]. RV infection does not only target the endoderm, but also signals that facilitate this interactive crosstalk. This includes Cerberus as a bone morphogenetic protein (BMP) antagonist [51]. The secretion of Cerberus from endodermal cells initiates di fferentiation of the neighboring tissue, namely the overlying cardiac mesoderm [51,52]. Furthermore, the analysis of endoderm-depleted frog and avian embryos revealed that the endoderm contributes to vasculogenesis and vascular tube formation [53]. Thus, as summarized in Figure 7, the molecular events identified in RV-infected endodermal cells could contribute to cardiovascular defects during congenital rubella [2].

Besides the mere expression level of essential components of developmental pathways, post-translational histone modifications are involved in the regulation of gene expression during development. The balance between H3K4me as an active and H3K27 as an inactive state histone modification directs the switch between active and inactive pathways during di fferentiation [54]. The activity of the KDM6A (UTX) demethylase was especially upregulated in endodermal cells during RV infection. KDM6A demethylase activity was reported to counteract DNA damage response and cell death induction in di fferentiating ESCs [55], which could also apply to RV-infected endodermal cells.

RV infection was associated with an upregulation of definitive endoderm-enriched transcription factors, including GATA4, EOMES, and SOX17 [56]. In a context- and dose-dependent manner, the transcription factor EOMES directs cardiac development as well as endoderm specification [57]. Whereas SOX17-null mice revealed a downregulation of several genes involved in heart development [58], the ectopic overexpression of SOX17 during hematopoiesis impaired survival of early hematopoietic precursors due to induction of apoptosis [59]. This indicates that normal embryonal development, especially cardiac specification, requires fine-tuned expression of several factors [60], which appears to be a ffected by RV infection.

The characterization of teratogens such as RV on iPSC-based models is an essential requirement to emphasize their suitability for the assessment of embryotoxicants and to identify relevant parameters to increase their predictive power. Congenital heart malformations are not only caused by pathogens such as RV, they are the most common among human developmental defects identified for human births. iPSC-based models enable valuable insights into human development and processes that might disturb its normal progression, which will broaden our diagnostic and treatment options for congenital defects. Further studies are needed to correlate the identified transcriptional changes with functional consequences for pathways directing embryonal development.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2073-4409/8/8/870/s1, Table S1: Related to description of quantitative real-time PCR analysis, Table S2: Related to Figure 2C. Spot cluster characteristics, Figure S1: Mean expression profiles of 'spot'-clusters of genes which were denoted with capital letters A–E., Figure S2: Gene set enrichment for pathway analysis of mock- and RV-infected iPSCs and iPSC-derived lineages., Figure S3: Characterization of the IFN-response gene signature in RV-infected iPSCs and iPSC-derived lineages., Figure S4: Characterization of gene expression signatures related to epigenetic regulation., Figure S5: Pathway signal flow (PSF) activity plot of the TGF-beta signaling pathway in ecto-, meso-, and endodermal cells derived from mock- and RV-infected iPSCs., Figure S6: Pathway signal flow (PSF) activity plot of the Wnt signaling pathway in ecto-, meso-, and endodermal cells derived from mock- and RV-infected iPSCs., Figure S7: Gene set expression signatures of developmental programs in mock- and RV-infected iPSCs and iPSC-derived lineages.

**Author Contributions:** C.C. supervision of the study; funding acquisition, C.C., H.B. wrote the first draft of the manuscript, all authors read and revised the final manuscript, N.C.B., J.B., D.H. performed experiments, E.W. and H.B. performed mathematical modelling and processing and analysis of transcriptomic data. M.L.S., S.B. and U.G.L. provided resources.

**Funding:** This work was supported by DFG gran<sup>t</sup> CL 459/3-1 to C.C. We acknowledge support from the German Research Foundation (DFG) and Leipzig University within the program of Open Access Publishing.

**Acknowledgments:** For provision of Wb-12 strain the authors want to thank B. Weißbrich (University of Wuerzburg, Germany). We want to thank Knut Krohn and Petra Süptitz from the core unit DNA technologies, IZKF Leipzig, Medical Faculty of the University of Leipzig, Leipzig, Germany for RNA quality assessment and technical support of the microarray experiments. The authors also want to thank Sandra Bergs for technical support.

**Conflicts of Interest:** The authors declare no conflict of interest.
