**1. Introduction**

The enveloped, single stranded (positive-sense) RNA virus rubella virus (RV) of the genus *Rubivirus* within the family *Togaviridae* is one of the few viruses that can cause an intrauterine infection. How these viruses are transmitted vertically from the infected mother to the fetus and how they impact human development is only partially resolved. In the case of the very e fficient teratogen

RV, the human-specific symptoms are categorized as congenital rubella syndrome (CRS) with the classical triad of clinical symptoms being sensorineural deafness, congenital heart disease (including cardiovascular and vascular anomalies), and cataracts [1,2]. Heart defects in CRS may comprise ventricular/atrial septal defects, patent ductus arteriosus, and patent foramen ovale. In congenital rubella, ocular (ophthalmic) pathologies include cataract, microphthalmia, glaucoma, and pigmentary retinopathy [1,2]. Furthermore, in tissue samples from three fatal CRS cases RV was detected in cardiac and adventitia (aorta and pulmonary artery) fibroblasts in association with vascular lesions [3]. The risk for the development of congenital defects is especially prevalent during maternal rubella until gestational week 11 and 12 [4–6]. Thus, intrauterine RV infection is only of concern during the first trimester. While congenital malformations are common, premature delivery and stillbirths are not markedly increased after intrauterine RV infection [1].

There are a number of ethical constraints associated with the study of human embryogenesis and congenital malformations, especially as early implantation stages of human embryos are inaccessible [7]. With embryonic stem cells (ESCs) and induced pluripotent stem cells (iPSCs), as the two types of human pluripotent stem cells (PSCs), these novel ex vivo cell culture platforms allow for the analysis of human embryonic germ layer segregation and as well as for developmental toxicity testing [8]. As a cell culture model, they represent a blastocyst-like stage, which can be extended to gastrulation-like stages through their di fferentiation into derivatives of the embryonic germ layers (ectoderm, mesoderm and endoderm). Additionally, their suitability as a developmental model has been demonstrated for cardiac commitment during development [9] as the heart is the first organ to develop and cardiac cell fate decisions occur very early. Furthermore, cultivation of ESCs in combination with suitable 3D matrices or together with trophoblast cells enables the formation of blastoids, gastruloids, and even embryoids (or embryo-like entities) as culture dish models for human embryogenesis [7,10].

PSCs and PSC-based di fferentiation models, especially the mouse (m) ESC test, are already validated for testing of teratogenic and embryotoxic substances such as thalidomide (brand name Contergan ®), [11,12]. However, their potential for the study of infections during pregnancy is just at the beginning of evaluation [13,14]. In line with the limited number of viruses that can cause perinatal infection, iPSCs possess intrinsic mechanisms that restrict virus infections. In addition, compared to di fferentiated somatic cells, iPSCs have a higher expression level of a distinct set of interferon (IFN)-induced genes [14]. This appears to counterbalance the absence of a type I IFN response in iPSCs as an essential component of antiviral innate immunity [15].

Teratogenic RV can be maintained in iPSCs over several passages followed by directed di fferentiation into embryonic germ layer cells [13], highlighting iPSCs as a promising model for the very early mechanisms involved in rubella embryopathy. As a follow-up to this study we aimed at the identification of RV-induced molecular alterations in these cells before and after initiation of directed di fferentiation through transcriptomics. The most profound e ffects associated with RV infection were detected in endodermal cells derived from RV-infected iPSCs. Markers for definitive endoderm were upregulated, which occurred in association with profound epigenetic changes, an upregulation of factors involved in vasculogenesis, and reduced activity of the TGF-β signaling pathway. Additionally, ectodermal cells revealed an altered expression profile of essential transcription factors for eye field development during RV infection. Thus, the study of RV infection on iPSCs and derived lineages provides insights into viral alterations of early developmental pathways and as such into congenital diseases in general.

### **2. Materials and Methods**

### *2.1. Cell Lines and Cultivation*

Vero (green monkey kidney epithelial cell line, ATCC CCL-81) and A549 (human lung carcinoma epithelial cells, ATCC, LGC Standards GmbH, Wesel, Germany) were cultured in Dulbecco's modified Eagle's medium (DMEM; Thermo Fisher Scientific, Darmstadt, Germany) with high glucose, GlutaMAX, 10% fetal calf serum (FCS) and 100 U/mL penicillin–streptomycin. If not otherwise indicated, the vector-free human episomal A18945 iPS cell line (alias TMOi001-A), (Thermo Fisher Scientific) was maintained in mTeSR™1 medium (StemCell Technologies, Cologne, Germany) with 10 μg/mL gentamycin on Matrigel™ (BD Biosciences, dispensed in DMEM/F-12)-coated culture plates with daily medium change. They were passaged enzymatically at a ratio of 1:6 to 1:10 every 3 to 5 days with collagenase type IV (Thermo Fisher Scientific) in DMEM-F12 with the addition of 10 μM Y-27632 ROCK inhibitor.

### *2.2. Directed and Undirected Di*ff*erentiation of iPSCs*

Directed differentiation was performed as an endpoint differentiation assay through the STEMdiffTM trilineage differentiation kit (StemCell Technologies). The differentiation protocol was performed according to the manufacturer's instructions and required cultivation of A18945 iPSCs in mTeSR™1 medium. Single cells, as obtained after treatment with Accutase (Merck/Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany), were plated on Matrigel. Every 24 h medium change of the respective STEMdiffTM trilineage differentiation medium for ectoderm, mesoderm, and endoderm was performed. Samples were collected after 5 days (mesoderm and endoderm) and 7 days (ectoderm) of cultivation. Undirected differentiation was initiated 24 h after collagenase-passaging of iPSC cultures at a ratio of 1:4 through application of undirected differentiation medium (DMEM-F12, 1x MEM-NEAA, 0.2 mM L-glutamine, 20% FBS, 0.11 mM β-mercaptoethanol, and 100 U/mL penicillin) followed by further cultivation for 5 days.

### *2.3. Embryoid Body Formation*

EB formation as based on a previous publication [16] and (http://www.biolamina.com/media. ashx/instructions-bl010.pdf) was carried out in suspension culture and single cell suspensions were obtained after Accutase (Sigma-Aldrich) treatment. A total of 1 × 10<sup>6</sup> cells was seeded in 200 μL of EB culture medium (DMEM-F12, 20% KnockOut™ Serum Replacement (Thermo Fisher Scientific), 1× MEM-NEAA, 0.2 mM L-glutamine, 0.11 mM β-mercaptoethanol and 1 mg/mL Gentamicin) medium into one well of a nontreated conical 96-well plate and centrifuged at 600× *g* for 5 min. After cultivation for 2 days the EBs were transferred according to the protocol to a low attachment flat-bottom six-well plate and medium was changed every third day.

### *2.4. Virus Infection and Interferon Assays*

The supernatant of infected Vero cells was collected and cleared from cellular debris by centrifugation at 350× *g* for 10 min at 4 ◦C and filtration through a 0.45 μm syringe filter. Thereafter ultracentrifugation with a 20% sucrose cushion (*w*/*v* in PBS) was performed for 2 h at 25,000 rpm and 4 ◦C. The obtained pellets were resuspended in mTeSR1. Viral titers were determined by standard plaque assay. As described previously [13], iPSC cultures with a 40–50% confluency were acutely infected with 7.5 × 10<sup>5</sup> plaque forming units (PFU) of RV per well of a 24-well plate. This corresponds approximately to an MOI of 20. The applied MOI can only be estimated as iPSCs were passaged enzymatically in clumps. The inoculum was replaced with fresh mTesR1 medium after 2 h of incubation [13]. After 4 to 5 days of cultivation, RV-infected iPSCs were passaged.

For exogenous (or paracrine) IFN treatment human recombinant IFN lambda 1 (IL-29, #300-02L) and 2 (IL28A, #300-02K), were purchased from Peprotech (Hamburg, Germany), and 3 (IL-28B, #CS26) from Novoprotein (Novoprotein, PELOBIOTECH GmbH, Planegg/Martinsried, Germany). The Accuri C6 flow cytometer (BD Bioscience, Heidelberg, Germany) was used for IFN measurement by the LEGENDplex human type 1/2/3 IFN panel (BioLegend, San Diego, CA, USA). The double-stranded (ds) RNA analogue polyinosinic-polycytidylic acid (poly I:C; Santa Cruz Biotechnology, Heidelberg, Germany) was added either directly to the cell culture or transfected at a concentration of 1 μg using Lipofectamine 2000 (Thermo Fisher Scientific) as transfection reagent.

### *2.5. Calcein Live Cell Staining*

For live cell staining, EBs were incubated with mTeSR1 plus calcein FM (Sigma-Aldrich) at 1 μM. After an incubation period for 30 min, EBs were washed twice with PBS and analyzed on an inverted fluorescence microscope.

### *2.6. RNA Isolation*

Total RNA was extracted from mock- and RV-infected cells by Trizol reagen<sup>t</sup> (Thermo Fisher Scientific). The purification was performed with the Direct-zol RNA kit (Zymo Research, Freiburg, Germany) according to manufacturer's instructions. The integrity of the RNA samples was confirmed through analysis on a fragment analyzer (Advanced Analytical). Only samples with a RIN (RNA integrity number as a means of quality assessment) equal to 7 or greater were subjected to further analysis.

### *2.7. Microarray Gene Expression Analysis and SOM Portrayal*

Isolated RNA was processed and hybridized to Illumina HT-12 v4 Expression BeadChips (Illumina, San Diego, CA, USA) and measured on the Illumina HiScan. Raw intensity data of 47,323 gene probes was extracted by Illumina GenomeStudio and subsequently background corrected, transformed into log10-scale, quantile normalized, and centralized to obtain gene expression estimates. Two independent samples per condition and cell type were processed.

Expression data were then further processed using self-organizing map (SOM) machine learning. The method distributes the gene-centered expression values among 2500 microclusters called meta-genes, which were arranged in a two-dimensional 50 × 50 lattice and colored in maroon-to-blue for high-to-low meta-gene expression values. These mosaic images visualize the transcriptome patterns of each individual sample and therefore can be understood as their molecular portraits exhibiting clusters of coexpressed genes in the samples studied [17]. Mean portraits over replicates were calculated by averaging the meta-gene landscapes of replicated samples while difference portraits between different cell types were obtained by subtracting the respective metagene values to highlight differentially expressed genes. Clusters of coexpressed genes were identified by selecting so-called 'spot-areas' in the SOM portraits using overexpression criteria as described previously [17]. For functional interpretation of the expression-modules, we applied gene set enrichment analysis using the gene set Z-score (GSZ), [17]. Enrichment of functional gene sets in the spot cluster was calculated by applying Fisher's exact test. We considered gene sets related to biological processes (BP) of the gene ontology (GO) classification, standard literature sets [17,18], and literature sets curated by our group. Downstream analysis methods were described previously [17,19] and are implemented in the R-package 'oposSOM' used for analysis [20].

Pathway activity was analyzed based on pathway topologies and gene expression data using the pathway signal flow method as implemented in oposSOM [21].

### *2.8. Quantitative Real-Time PCR Analysis of Viral and Cellular RNA*

For determination of the mRNA expression level of selected cellular genes, 1.2 μg of total RNA were reverse transcribed with Oligo(dT)18 primer and AMV reverse transcriptase (Promega, Mannheim, Germany) at 42 ◦C for 1 h. This was followed by an incubation step at 70 ◦C for 10 min. The carousel-based LightCycler 2.0 (Roche, Mannheim, Germany) was used for quantitative real-time PCR (qRT-PCR) experiments. These experiments included a 1:5 dilution of the respective cDNA samples together with 1 μg BSA and the *GoTaq*® *qPCR master mix* (Promega). Supplement Table S1 lists oligonucleotides and probes targeting viral p90 gene that were used for quantification of viral RNA as described [22]. Two different approaches for relative expression analysis were pursued. For direct comparison of one sample type after mock- and RV-infection, comparative delta delta Ct (ΔΔCt) was used. For comparison of gene expression levels among different cell types within a large data

set, a modified version of the comparative delta delta Ct (ΔΔCt) method was used. The normalized relative quantity (NRQ) values were derived from qbase+ software (Biogazelle, Zulte, Belgium) which are based on the mean expression values of all samples and replicates within a given data set [23].
