Triple RNA-Seq Reveals Synergy in a Human Virus-Fungus Co-infection Model.

Article Triple RNA-Seq Reveals Synergy in a Human VirusFungus Co-infection Model Graphical Abstract Highlights dTriple RNA-seq measures gene expression of co-infected immune cells dGene correlation networks reveal different hub gene sets under co-infection dCo-infection expression includes synergies and interferences between host and pathogens dMolecular basis of viral/fungal pulmonary infection has potential for the clinic Authors Bastian Seelbinder, Julia Wallstabe, Lothar Marischen, ..., Alexander J. Westermann, Sascha Scha ¨uble, Juergen Loeffler Correspondence [email protected]e In Brief Seelbinder et al. demonstrate simultaneous sequencing of RNA isolated from human immune cells infected with Aspergillus fumigatus and cytomegalovirus, two pulmonary pathogens regularly affecting the lungs of immunosuppressed patients. They detect characteristic gene expression patterns for single and co-infections and reveal synergistic virulence strategies between the two pathogens. Seelbinder et al., 2020, Cell Reports 33, 108389 November 17, 2020 ª2020 The Authors. https://doi.org/10.1016/j.celrep.2020.108389 ll Article Triple RNA-Seq Reveals Synergy in a Human Virus-Fungus Co-infection Model Bastian Seelbinder, 1,11 Julia Wallstabe, 2,11 Lothar Marischen, 2,11 Esther Weiss, 2 Sebastian Wurster, 2,3 Lukas Page, 2 Claudia Lo ¨ffler, 2 Lydia Bussemer, 2 Anna-Lena Schmitt, 2 Thomas Wolf, 1 Jo ¨rg Linde, 4 Luka Cicin-Sain, 5,6 Jennifer Becker, 7 Ulrich Kalinke, 7 Jo ¨rg Vogel, 8,9 Gianni Panagiotou, 1,10 Hermann Einsele, 2 Alexander J. Westermann, 8,9,11 Sascha Scha ¨uble, 1,11 and Juergen Loeffler 2,11,12, * 1 Systems Biology and Bioinformatics, Leibniz Institute for Natural Product Research and Infection Biology – Hans Kno ¨ll Institute (HKI), 07745 Jena, Germany 2 University Hospital W€ urzburg, Medical Hospital II, WU ¨4i, 97080 W€ urzburg, Germany 3 The University of Texas MD Anderson Cancer Center, Department of Infectious Diseases, Infection Control and Employee Health, Houston, TX 77030, USA 4 Friedrich-Loeffler-Institut, Federal Research Institute for Animal Health, Institute of Bacterial Infections and Zoonoses, 07743 Jena, Germany 5 Department of Vaccinology and Applied Microbiology, Helmholtz Centre for Infection Research, Hannover-Braunschweig Site, 38124 Braunschweig, Germany 6 Cluster of Excellence RESIST (EXC 2155), Hannover Medical School (MHH) Braunschweig, 38124 Braunschweig, Germany 7 Institute for Experimental Infection Research, TWINCORE–Centre for Experimental and Clinical Infection Research, a joint venture between the Hannover Medical School and the Helmholtz Centre for Infection Research, Cluster of Excellence RESIST (EXC 2155), Hannover Medical School (MHH), 30625 Hannover, Germany 8 Institute of Molecular Infection Biology (IMIB), University of W€ urzburg, 97080 W€ urzburg, Germany 9 Helmholtz Institute for RNA-based Infection Research (HIRI), Helmholtz Centre for Infection Research (HZI), 97080 W€ urzburg, Germany 10 Department of Medicine and State Key Laboratory of Pharmaceutical Biotechnology, University of Hong Kong, Hong Kong S.A.R., China 11 These authors contributed equally 12 Lead Contact *Correspondence: [email protected] https://doi.org/10.1016/j.celrep.2020.108389 SUMMARY High-throughput RNA sequencing (RNA-seq) is routinely applied to study diverse biological processes; however, when performed separately on interacting organisms, systemic noise intrinsic to RNA extraction, library preparation, and sequencing hampers the identification of cross-species interaction nodes. Here, we develop triple RNA-seq to simultaneously detect transcriptomes of monocyte-derived dendritic cells (moDCs) infected with the frequently co-occurring pulmonary pathogens Aspergillus fumigatus and human cytomegalovirus (CMV). Comparing expression patterns after co-infection with those after single infections, our data reveal synergistic effects and mutual interferences between host responses to the two pathogens. For example, CMV attenuates the fungus-mediated activation of pro-inflammatory cytokines through NF-kB (nuclear factor kB) and NFAT (nuclear factor of activated T cells) cascades, while A. fumigatus impairs viral clearance by counteracting viral nucleic acid-induced activation of type I interferon signaling. Together, the analytical power of triple RNA-seq proposes molecular hubs in the differential moDC response to fungal/viral single infection or co-infection that contribute to our understanding of the etiology and, potentially, clearance of post-transplant infections. INTRODUCTION Allogenic stem cell transplantation (alloSCT) has advanced the therapy of hematological malignancies and is potentially curative for a spectrum of nonmalignant hematological disorders (Singh and McGuirk, 2016). The first successful solid organ transplantation (SOT) took place in 1954; today, transplant statistics are steadily increasing, with over 36,500 organ transplants in the United States in 2018 (based on OPTN data as of January 9, 2019; Harrison et al., 1956). Reduced-intensity conditioning regimes, novel therapeutic strategies to combat graft-versushost disease, and tailored supportive care have improved alloSCT and SOT outcomes. However, opportunistic infections are still a clinical challenge and a major source of post-transplant complications (Ullmann et al., 2016). Invasive aspergillosis (IA), predominantly causing pulmonary infections, is responsible for significant post-transplant morbidity, mortality, and incremental cost burdens (Drgona et al., 2014). In addition, human cytomegalovirus (CMV)-associated infections, including CMV pneumonia, remain the most common infectious complications in alloSCT recipients (Camargo and Komanduri, 2017), and CMV viremia is associated Cell Reports 33, 108389, November 17, 2020 ª2020 The Authors. 1 This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). ll OPEN ACCESS with increased early overall mortality after alloSCT (Green et al., 2016). CMV-disease-related mortality has significantly declined with improved prophylactic medication, PCR-based diagnostics, and pre-emptive antiviral treatment, yet indirect CMV effects continue to adversely impact alloSCT outcomes (de la Ca ´mara, 2016;Duarte and Lyon, 2018). Notably, CMV infections pose an independent risk factor for development of IA in alloSCT recipients (Garcia-Vidal et al., 2008;Marr et al., 2002), and invasive mycoses are a frequent cause of mortality in patients surviving CMV disease (de la Ca ´mara, 2016;Nichols et al., 2002). Ample evidence indicates that CMV alters the human immune response to escape host surveillance and establish latent persistence (Cheung et al., 2009;Hahn et al., 1998;Kaminski and Fishman, 2016;Kotenko et al., 2000;Taylor-Wiedeman et al., 1991). Several proteins encoded by CMV broadly modulate the magnitude and quality of host immune cell functions (Miller-Kittrell and Sparer, 2009). For example, CMV-secreted immunosuppressive cytokine homologs inhibit dendritic cell maturation and survival as well as dendritic cell-mediated T-helper (Th) cell activation and Th1 differentiation (Chang et al., 2004;Raftery et al., 2004), which are crucial mechanisms for linking innate and adaptive immunity to fungal pathogens. Vice versa, Aspergillus fumigatus, the most frequent cause of IA, suppresses human T cell activation in response to CMV (Stanzani et al., 2005). However, the molecular events underlying the differential impact of CMV and A. fumigatus on human mononuclear cell functions and the reciprocal immune defense in co-infections are largely unexplored (Grow et al., 2002;Martino et al., 2009;Mikulska et al., 2009;Solak et al., 2013;Upton et al., 2007). High-throughput RNA sequencing (RNA-seq) has greatly advanced our understanding of infections (Saliba et al., 2017; Colgan et al., 2017) and even allows simultaneous studies of host and pathogen transcriptomes (Westermann et al., 2012). These ‘‘dual RNA-seq’’ approaches concurrently isolate host and pathogen RNA, convert it into cDNA libraries for sequencing, and separate the transcriptomes at the computational level by mapping sequencing reads to the respective reference genomes. To date, dual RNA-seq has been applied to diverse infection models to study virulence mechanisms of—and mammalian immune responses to—viral (Juranic Lisnic et al., 2013;Wesolowska-Andersen et al., 2017), bacterial (Westermann et al., 2017), and fungal (Wolf et al., 2018) pathogens as well as eukaryotic parasites (Choi et al., 2014;Pittman et al., 2014). However, multi-organism RNA-seq has not previously been applied to co-infection settings. Here, we advanced the concept of multi-organism RNA-seq by developing triple RNA-seq, which we applied to human monocyte-derived dendritic cells (moDCs) challenged with the two pulmonary pathogens CMV strain TB40 and A. fumigatus. The identified modulations of the host’s immunological state upon single infection and co-infection were independently validated by flow cytometry and multiplex cytokine secretion assays. Our collective findings suggest unique interdependencies of CMV and A. fumigatus during co-infection that add to our molecular understanding of the synergy between CMV infection and the development of invasive mold infections in immunocompromised patients. RESULTS Triple RNA-Seq of Viral/Fungal Co-infections to Simultaneously Study Host and Pathogen Gene Expression Infection assays of human immune cells with either CMV or A. fumigatus were previously established (Paijo et al., 2016; Lother et al., 2014). Here, we built upon these protocols and challenged moDCs with A. fumigatus germ tubes or CMV, either separately (single infection) or in combination (co-infection). Host cell viability (Figure S1A), infection rates (Figures S1B and S1C), and morphology (Figure S1D), analyzed by flow cytometry and fluorescence microscopy, demonstrated sustained infections with the pathogens in both singleand co-infection settings. No prominent changes in fungal morphology or infection rate were observed between the two infection settings (Figures S1C and S1D). In contrast, we observed an increased virus infection rate in the presence of A.fumigatus as compared to CMV single infection (Figure S1B). As a prerequisite for multi-organism RNA-seq analysis, several parameters—including lysis conditions, multiplicities of infections (MOIs), and efficient removal of ribosomal RNA (rRNA)— need to be empirically determined for a given infection system. Here, we first established these parameters for single-infection settings of moDCs with either CMV or A. fumigatus (Figures S2A–S2C; Data S1). Application of dual RNA-seq to the resulting RNA samples led to the detection of the presumed infectionspecific host expression patterns (Figures S2D and S2E), including an induction of pro-inflammatory host marker genes IL1A/B,CCL3, and TLR2 upon A. fumigatus infection (Braedel et al., 2004;Lass-Flo ¨rl et al., 2013;Walsh et al., 2005) and IFNB and CCL2 activation after CMV infection (Loewendorf and Benedict, 2010;McNab et al., 2015). In addition, altered fungal expression levels in the presence of moDCs (Figure S2D), e.g., of the gliotoxin mRNA (gliF)that encodes a mycotoxin known to be produced during host infection, as well as induced expression of immunomodulatory viral mRNAs (Figure S2E), further supported the reliability of our approach. To characterize all three transcriptomes in sequential or simultaneous co-infection settings, we expanded the method toward triple RNA-seq (Figure 1A). MoDCs were either first infected with CMV and, after 4.5 h, additionally with A. fumigatus, or first infected with A. fumigatus followed by CMV challenge 2 h later, or simultaneously infected with both pathogens. Since dual RNA-seq data indicated that major gene expression changes occurred during the first hours after fungal and viral exposure, co-infection samples were harvested after 9 h and subjected to triple RNA-seq. Co-infection data were compared with transcriptome data from single infections (harvested in parallel with co-infections) and with data from uninfected moDCs or A. fumigatus mono-cultures. We extrapolated sequencing depth requirements for triple RNA-seq to sufficiently cover all three transcriptomes. Since dual RNA-seq indicated host transcriptome coverage to be rate limiting (Figures S2A and S2B), we increased the sequencing depth for triple RNA-seq by 5-fold, yielding 15 million non-ribosomal human reads, a threshold above which further increases in sequencing depth have diminishing returns 2Cell Reports 33, 108389, November 17, 2020 Article ll OPEN ACCESS Figure 1. Triple RNA-Seq Outline (A) Triple RNA-seq pipeline. Setup and time frame for controls, single infection, and co-infection are indicated; n = 4. (B) Percentages of sequencing reads that mapped to the annotated reference genome of the three studied organisms. Reads per organism were further assigned to the indicated transcript classes. Afu, Aspergillus fumigatus; moDC, monocyte-derived dendritic cell; CMV, cytomegalovirus; mitoRNA, mitochondrial RNA; miRNA, microRNA; lncRNA, long noncoding RNA; snoRNA, small nucleolar RNA; snRNA, small nuclear RNA; ncRNA, noncoding RNA; miscRNA, miscellaneous RNA. Cell Reports 33, 108389, November 17, 2020 3 Article ll OPEN ACCESS (Ching et al., 2014;Liu et al., 2014). Between 67% and 95% of the quality-filtered reads (Data S1;Data S2) were successfully aligned to the reference genomes. While the vast amount of reads mapped to the human genome, A. fumigatus contributed up to 24% and CMV contributed up to 1.6% of mapped reads in single-infection and co-infection samples (Figure 1B). As expected, the majority of human (55%) and fungal (55%–75%) reads derived from mRNA. Albeit that ribosomal depletion was less efficient for the fungal than for the human transcriptome, rRNA-derived reads did not exceed 25% in the A. fumigatus data subset, allowing for differential gene expression analyses. Other noncoding RNA classes were adequately represented in the two eukaryotic transcriptomes, whereas CMV-annotated genes all encode mRNAs (Figure 1B). Collectively, these data confirmed the high technical quality of the triple RNA-seq data, with relative proportions of host-topathogen read ratios and assignment to major RNA classes matching the predictions extrapolated from the dual RNA-seq pilot (Data S1). Transcriptomes during Co-infection Differ Globally from Single Infections Although principal-component analysis (PCA) of moDC transcriptomes revealed no time-point-specific segregation (Figure 2A), distinct clusters were obtained for single-infection conditions with A. fumigatus or CMV and co-infection. Similarly, A. fumigatus samples failed to cluster according to infection time point yet globally differed between fungal mono-culture and (co-) infection samples (Figure 2B). Finally, CMV transcriptome profiles formed clusters for single infection and co-infection despite the compact size of the CMV genome (166 genes; Figure 2C). In the following, due to the absence of time-point-specific segregation, we disregard temporal information and analyzed infection samples based on etiology (i.e., uninfected, single viral or single fungal infection, and co-infection). The triple RNA-seq approach potentially reduces systemic noise that may impede cross-species gene expression correlations when separately sampling host and pathogen transcriptomes (Westermann et al., 2017). To globally investigate interspecies co-expression, we calculated node betweenness and degree, two metrics to quantify co-expression network complexity and identify hub genes (Figure 3;Data S3). Focusing on innate immunity-associated effects, we restricted the network analysis to human genes contained within InnateDB (Breuer et al., 2013). The first two networks depicted in Figure 3 show host-pathogen gene-gene correlations present in the intersection of the respective singleand co-infection networks, whereas the remaining networks refer to correlations in the set differences between infection settings. Co-infection networks shared only few cross-species correlations with single-infection networks, whereas co-infection exclusive correlations exhibited a high degree of connectivity and were mostly based on positive correlations. Very few correlations were specific to single A. fumigatus infection, but fungus-host connectivity increased during co-infection, suggesting the fungus to adapt to—and maybe benefit from—the presence of CMV. In contrast, the single CMV infection network consisted of many unique, mostly positive, inter-species correlations. This implies a specialized CMV response to human target cells largely ignoring the presence of the fungus. Cross-species expression correlation analysis can pinpoint host factors with potential as future biomarkers or drug targets. CXCL11, for example, showed considerable network importance specifically during co-infection without any obvious correlations in single-infection settings (Data S3). Similarly, TNF showed high node degree and betweenness, specifically during co-infection, but occupied a hub position also in the intersection correlation network of fungal singleand co-infection (Data S3). This hints toward a generally important role for tumor necrosis factor (TNF) in the response to A. fumigatus, regardless of the additional presence of CMV. RELA, in contrast, possesses central network importance in the intersection of CMV singleand co-infection (Data S3), suggesting a particular relevance of this factor in the immune response against CMV. Differential Gene Expression Profiles in the Three Interacting Organisms Relative to uninfected moDCs, there was an increased number of differentially expressed human genes upon co-infection as compared to either single infection (Figure S3A). Moreover, we identified specific moDC gene sets with distinct expression patterns between, but low variance within, different infection etiologies (Figure 4A). For example, TNF,IL1A,andCXCL8 were specifically upregulated upon the sensing of A. fumigatus as previously reported (Balloy et al., 2008;Caffrey-Carr et al., 2017;Caffrey et al., 2015;Cortez et al., 2006;Mehrad et al., 1999;Roilides et al., 1998). On the other hand, induction of IFNB (encoding a first-line defense type I interferon [IFN] to CMV infection; Marshall and Geballe, 2009), CXCL10,andCXCL11 (associated with immune clearance in CMV viremia; Cheeran et al., 2003;Knoblach et al., 2011;Murayama et al., 2012) was specific to CMV infection. The majority of A. fumigatus genes differentially expressed in the presence of moDCs compared with the fungal mono-culture were shared between single infection and co-infection (Figure S3B). For example, the cat1 mRNA that encodes a fungal catalase to break down host-derived hydrogen peroxide was equally highly expressed by A.fumigatus during singleand co-infection. This suggests that moDCs were the primary driver of fungal transcriptional reprogramming. This notwithstanding, a greater total number of fungal genes were regulated during single infection than during co-infection, thereby defining an A. fumigatus gene set whose regulation might be dispensable for infection in the presence of CMV. CMV expression analysis, on the other hand, revealed largely overlapping sets of differentially expressed viral genes during single infection and co-infection (Figure S3B). The Host Response to Co-infection Suggests Synergy among A. fumigatus and CMV Closer inspection of moDC expression data identified a subset of key immune pathways whose activity differed markedly between infection settings, including Toll-like receptor (TLR) signaling, nucleic acid sensing, and C-type lectin receptor signaling (Figure S4). In co-infected moDCs, expression levels of AIM2 and CCL5, factors involved in cytosolic DNA sensing (Figure S5) and with known roles in the defense against both fungal and viral 4Cell Reports 33, 108389, November 17, 2020 Article ll OPEN ACCESS A B C Figure 2. PCA Identifies Co-clustering Transcriptomes (A–C) PCA for all three organisms after median-by-ratio normalization. Ellipses depict 95% confidence intervals for conditions. PCA for genes associated with (A) Homo sapiens, (B) Aspergillus fumigatus, or (C) cytomegalovirus (CMV) reference genome. Left: PCAs for all conditions. Right: (A) infection conditions grouped independent of infection time point, (B) data for single infection and co-infection grouped and compared to single-culture A. fumigatus, and (C) data for single infection versus co-infection independent of time point. Abbreviations are as explained in the Figure 1 legend. Cell Reports 33, 108389, November 17, 2020 5 Article ll OPEN ACCESS Figure 3. Interspecies Correlation Networks Pinpoint Synergies during Co-infection Networks indicate significant correlations (edges: p < 0.05; |rho| R0.95; red indicates positive correlation; and blue indicates negative correlation) of gene expression values (nodes: green represents Homo sapiens, yellow indicates Aspergillus fumigatus, and red indicates cytomegalovirus [CMV]) between (legend continued on next page) 6Cell Reports 33, 108389, November 17, 2020 Article ll OPEN ACCESS infections (Huang and Levitz, 2000;Man et al., 2016;Tyner et al., 2005), matched their cumulative induction after single infections (Figure 4A). However, a number of immune-related moDC genes did not display such additive expression patterns. For instance, the expression of nuclear factor kB (NF-kB)-dependent genes was mainly driven by A. fumigatus through TNF signaling (Figure 4B). Co-infection with CMV, however, reduced the expression of those genes as compared to fungal single infection. Similar expression patterns were detected for IL10 and IL1B. Vice versa, RIG-I and ZBP1 signaling (previously associated with type I IFN responses; Onomoto et al., 2010;Yang et al., 2020), as well as expression of IFNB,CXCL10, and TLR3, displayed strong activation upon CMV single infection, which was counteracted by the additional presence of A. fumigatus (Figure 4B). The gene for apoptosis-associated speck-like protein (ASC), which is critical for host cell survival upon viral infection (Kumar et al., 2013), was highly expressed in uninfected and CMV-infected moDCs. In contrast, expression of ASC was downregulated in the presence of A. fumigatus, thereby mirroring the expression pattern of Dectin-1, encoding an important fungal infection sensor (Taylor et al., 2007). This points at a disparate role for ASC in response to viral and fungal infections, suggesting an inhibiting effect upon fungal infection that is dominant over the induction upon viral challenge. This hypothesis was further supported by the fact that IL1B, despite its functional connection to ASC (Martinon and Tschopp, 2007), was not coexpressed with ASC. Finally, cGAS and STING, both encoding receptors for foreign nucleic acids, and STING also functioning as an IFN-stimulating factor associated with the viral glycoprotein US9 (Choi et al., 2018), were induced upon single CMV infection but downregulated upon co-infection (Figure 4B). Taken together, these expression data support the existence of two distinct host response patterns to A. fumigatus or CMV infection (Figure 4C). Expression of IL1B, IL10, and NF-kB-associated genes peaked in A. fumigatus-infected moDCs, whereas cGAS, STING, and RIG-I signaling and expression of IFNB, CXCL10,TLR3,and ZBP1 showed maximal expression upon CMV single infection. Relative to their induction upon the respective single infections, expression levels for all those host genes dropped in co-infected cells, suggesting mutual interfering effects between the two host responses with possibly synergistic effects for the two pathogens. Independent Validation of Expression Changes Quantitative real-time PCR measurement of genes from the three organisms that were differentially expressed in the triple RNA-seq data supported the sequencing-derived expression changes (Figure 5A; Figure S5). For instance, differential expression of human ZBP1, fungal cat1, and the mRNA for the viral envelope glycoprotein UL4 in single infection and co-infection settings could be confirmed in this way (Figure 5A). We next traced the expression of selected host factors, which were called as differentially expressed in the RNA-seq analysis, on the protein level. For instance, as shown by flow cytometry of antibody-stained moDCs, expression of the surface marker CCR7 was highest after co-infection, which was in line with mRNA levels in the triple RNA-seq dataset (Figure S6). Additionally, altered secretion levels of key cytokines produced from genes that showed pathogen-specific expression patterns were confirmed by multiplex ELISA (Figure 5B; Figure S7). For example, the secreted levels of interleukin (IL)-1band IFN-b differed between fungal and viral infections (Figure 5B), echoing the differential expression of their cognate mRNAs in the sequencing data. Altogether, these results underpin the sequencing data, confirm some of the key transcriptomic changes in the infected host cells to extend to the protein level, and provide further support of the notion that pulmonary pathogens differentially affect their host at a global level during co-infection (Reese et al., 2016). DISCUSSION Cross-kingdom interactions in polymicrobial infections are increasingly recognized as crucial virulence determinants that shape the outcome of life-threatening infectious diseases (Arvanitis and Mylonakis, 2015;Bergeron et al., 2017). In addition to direct physical interaction and inter-kingdom signaling, altered immunopathology is considered to foster co-infections (Arvanitis and Mylonakis, 2015). Specifically, viral pathogens associated with long-term persistence such as CMV have evolved an armamentarium of counter-strategies to shape-shift the host environment, allowing for immune surveillance evasion and establishment of latent infection (Freeman, 2009;Picarda and Benedict, 2018). Although evidence is limited, previous studies reported broad alterations of host immunity elicited by CMV in immunocompromised patients, entailing the predisposition to subsequent opportunistic fungal diseases (Yong et al., 2018). To improve our understanding of this co-occurrence, we here set out to dissect the molecular interplay of A.fumigatus, CMV, and their shared host cells during co-infection. Establishment of the Triple RNA-Seq Approach To study the complex interplay of different pathogen classes with their human host and among each other, we implemented the previously proposed concept of multi-organism triple RNAseq (Westermann et al., 2017). We selected moDCs—being at the border of innate and adaptive immunity—as a host model to establish this technology for a variety of both biological and technical aspects. Myeloid precursors form a reservoir for latent CMV infection and their differentiation into dendritic cells can trigger virus reactivation (Hahn et al., 1998;Jarvis and Nelson, 2002), while moDCs represent a well-studied surrogate model in the context of fungal infection (Hsieh et al., 2017;Mezger et al., 2008;Morton et al., 2011). Additionally, moDCs can be generated in large quantities by well-standardized protocols for the upfront optimization of infection and RNA processing organisms. The asterisk indicates that correlations between CMV and A. fumigatus were removed. Node betweenness and degree are measures of network complexity. Higher degree indicates increased (local) connectivity. Higher betweenness indicates density. Corresponding boxplots show network-wise median (center line), confidence interval (boxes), and quantile (25% and 75%) values. # Genes and # Correlations display numbers of genes and correlation types, respectively. Cell Reports 33, 108389, November 17, 2020 7 Article ll OPEN ACCESS Figure 4. Detailed Differential Gene Expression Analysis of Monocyte-Derived Dendritic Cells (A) Genes selected for high variance across all conditions and low variance within individual conditions regardless of infection time point. Left: differentially expressed genes (DEGs) with distinct expression patterns across experimental conditions (no stimulation, single infection, and co-infection). Right: examples of innate immune-relevant genes allowing for clear separation of infection types (inferred from innateDB; https://www.innatedb.com/). (B and C) Selection of host factors involved in monocyte-derived dendritic cell response to cytomegalovirus (CMV), Aspergillus fumigatus infection, or co-infection is depicted based on the clustering of similar expression profiles (B) and in topological context (C). Topological analyses are based on Kyoto Encyclopedia of Genes and Genomes reference signaling maps (Figure S4). Abbreviations are as explained in the Figure 1 legend. 8Cell Reports 33, 108389, November 17, 2020 Article ll OPEN ACCESS interaction, relevant to competition in airways. Med. Mycol. 57 (Suppl 2), S228–S232. Seelbinder, B., Wolf, T., Priebe, S., McNamara, S., Gerber, S., Guthke, R., and Linde, J. (2019). GEO2RNAseq: An easy-to-use R pipeline for complete preprocessing of RNA-seq data. bioRxiv. https://doi.org/10.1101/771063. Shankar, J., Cerqueira, G.C., Wortman, J.R., Clemons, K.V., and Stevens, D.A. (2018). RNA-Seq Profile Reveals Th-1 and Th-17-Type of Immune Responses in Mice Infected Systemically with Aspergillus fumigatus. Mycopathologia 183, 645–658. Singh, A.K., and McGuirk, J.P. (2016). Allogeneic Stem Cell Transplantation: A Historical and Scientific Overview. Cancer Res. 76, 6445–6451. Solak, Y., Biyik, Z., Cizmecioglu, A., Genc, N., Ozbek, O., Gaipov, A., and Yeksan, M. (2013). Cytomegalovirus and Aspergillus spp. coinfection in organ transplantation: a case report and review of the literature. CEN Case Rep. 2, 59–67. Stanzani, M., Orciuolo, E., Lewis, R., Kontoyiannis, D.P., Martins, S.L., St John, L.S., and Komanduri, K.V. (2005). Aspergillus fumigatus suppresses the human cellular immune response via gliotoxin-mediated apoptosis of monocytes. Blood 105, 2258–2265. Taylor, P.R., Tsoni, S.V., Willment, J.A., Dennehy, K.M., Rosas, M., Findon, H., Haynes, K., Steele, C., Botto, M., Gordon, S., and Brown, G.D. (2007). Dectin-1 is required for beta-glucan recognition and control of fungal infection. Nat. Immunol. 8, 31–38. Taylor-Wiedeman, J., Sissons, J.G., Borysiewicz, L.K., and Sinclair, J.H. (1991). Monocytes are a major site of persistence of human cytomegalovirus in peripheral blood mononuclear cells. J. Gen. Virol. 72, 2059–2064. Tyner, J.W., Uchida, O., Kajiwara, N., Kim, E.Y., Patel, A.C., O’Sullivan, M.P., Walter, M.J., Schwendener, R.A., Cook, D.N., Danoff, T.M., and Holtzman, M.J. (2005). CCL5-CCR5 interaction provides antiapoptotic signals for macrophage survival during viral infection. Nat. Med. 11, 1180–1187. Ullmann, A.J., Schmidt-Hieber, M., Bertz, H., Heinz, W.J., Kiehl, M., Kr€ uger, W., Mousset, S., Neuburger, S., Neumann, S., Penack, O., et al.; Infectious Diseases Working Party of the German Society for Hematology and Medical Oncology (AGIHO/DGHO) and the DAG-KBT (German Working Group for Blood and Marrow Transplantation) (2016). Infectious diseases in allogeneic haematopoietic stem cell transplantation: prevention and prophylaxis strategy guidelines 2016. Ann. Hematol. 95, 1435–1455. Upton, A., Kirby, K.A., Carpenter, P., Boeckh, M., and Marr, K.A. (2007). Invasive aspergillosis following hematopoietic cell transplantation: outcomes and prognostic factors associated with mortality. Clin. Infect. Dis. 44, 531–540. Vanderbeke, L., Spriet, I., Breynaert, C., Rijnders, B.J.A., Verweij, P.E., and Wauters, J. (2018). Invasive pulmonary aspergillosis complicating severe influenza: epidemiology, diagnosis and treatment. Curr. Opin. Infect. Dis. 31, 471–480. Walsh, T.J., Roilides, E., Cortez, K., Kottilil, S., Bailey, J., and Lyman, C.A. (2005). Control, immunoregulation, and expression of innate pulmonary host defenses against Aspergillus fumigatus. Med. Mycol. 43 (Suppl 1), S165–S172. Wesolowska-Andersen, A., Everman, J.L., Davidson, R., Rios, C., Herrin, R., Eng, C., Janssen, W.J., Liu, A.H., Oh, S.S., Kumar, R., et al. (2017). Dual RNA-seq reveals viral infections in asthmatic children without respiratory illness which are associated with changes in the airway transcriptome. Genome Biol. 18,12. Westermann, A.J., and Vogel, J. (2018). Host-Pathogen Transcriptomics by Dual RNA-Seq. Methods Mol. Biol. 1737, 59–75. Westermann, A.J., Gorski, S.A., and Vogel, J. (2012). Dual RNA-seq of pathogen and host. Nat. Rev. Microbiol. 10, 618–630. Westermann, A.J., Barquist, L., and Vogel, J. (2017). Resolving host-pathogen interactions by dual RNA-seq. PLoS Pathog. 13, e1006033. Wiertz, E.J., Jones, T.R., Sun, L., Bogyo, M., Geuze, H.J., and Ploegh, H.L. (1996). The human cytomegalovirus US11 gene product dislocates MHC class I heavy chains from the endoplasmic reticulum to the cytosol. Cell 84, 769–779. Wolf, T., Ka ¨mmer, P., Brunke, S., and Linde, J. (2018). Two’s company: studying interspecies relationships with dual RNA-seq. Curr. Opin. Microbiol. 42, 7–12. Yang, D., Liang, Y., Zhao, S., Ding, Y., Zhuang, Q., Shi, Q., Ai, T., Wu, S.Q., and Han, J. (2020). ZBP1 mediates interferon-induced necroptosis. Cell. Mol. Immunol. 17, 356–368. Ye, J., Coulouris, G., Zaretskaya, I., Cutcutache, I., Rozen, S., and Madden, T.L. (2012). Primer-BLAST: a tool to design target-specific primers for polymerase chain reaction. BMC Bioinformatics 13, 134. Yong, M.K., Slavin, M.A., and Kontoyiannis, D.P. (2018). Invasive fungal disease and cytomegalovirus infection: is there an association? Curr. Opin. Infect. Dis. 31, 481–489. Cell Reports 33, 108389, November 17, 2020 15 Article ll OPEN ACCESS STAR+METHODS KEY RESOURCES TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies anti-human CD40-PE-Vio770 (recombinant human IgG1, REAfinity) Miltenyi Biotec, Bergisch Gladbach, Germany Cat#130-110-948, Clone: monoclonal REA733; RRID:AB_2658003 anti-human CD80-APC (recombinant human IgG1, REAfinity) Miltenyi Biotec, Bergisch Gladbach, Germany Cat#130-117-719, Clone: monoclonal REA661; RRID:AB_2751414 anti-human CD209-VioBlue(recombinant human IgG1, REAfinity) Miltenyi Biotec, Bergisch Gladbach, Germany Cat#130-110-456, Clone: monoclonal REA617; RRID:AB_2656252 anti-human CCR7-VioBlue(recombinant human IgG1, REAfinity) Miltenyi Biotec, Bergisch Gladbach, Germany Cat#130-117-353, Clone: monoclonal REA546; RRID:AB_2733933 anti-human TLR2-PE-Vio770 (recombinant human IgG1, REAfinity) Miltenyi Biotec, Bergisch Gladbach, Germany Cat#130-099-022, Clone: monoclonal REA109; RRID:AB_2656975 anti-human TLR3-APC (mouse IgG1k) Miltenyi Biotec, Bergisch Gladbach, Germany Cat#130-096-885, Clone: monoclonal TLR3.7; RRID:AB_2660004 Bacterial and Virus Strains human CMV strain TB40/E-mNeonGreen Helmholtz Centre for Infection Research, Braunschweig, Germany Kasmapour et al., 2017 https://www.helmholtz-hzi.de human CMV strain TB40/SE University Freiburg, Medical Center, Institute of Virology, Freiburg, Germany; Sampaio et al., 2017 https://www.uniklinik-freiburg.de/de.html Biological Samples Human peripheral venous blood from healthy adult donors for generation of monocyte-derived dendritic cells Paijo et al., 2016;Sallusto and Lanzavecchia, 1994 N/A Chemicals, Peptides, and Recombinant Proteins RNAprotectCell Reagent QIAGEN, Hilden, Germany Cat#76526 Viobility 405/520 Fixable Dye Miltenyi Biotec, Bergisch Gladbach, Germany Cat#130-109-814 iTaq Universal SYBRGreen Supermix Bio-Rad Laboratories GmbH, Feldkirchen, Germany Cat#1725124 Recombinant human IL-4, premium grade Miltenyi Biotec, Bergisch Gladbach, Germany Cat#130-093-922 Recombinant human GM-CSF,premium grade Miltenyi Biotec, Bergisch Gladbach, Germany Cat#130-093-866 Critical Commercial Assays RiboPure-Yeast Kit Thermo Fisher Scientific, Waltham, MA, USA Cat#AM1926 DNase I, RNase free (1.000 U) Thermo Fisher Scientific, Waltham, MA, USA Cat#EN0521 M-MLV reverse transcriptase Invitrogen, Carlsbad, CA, USA Cat#28025013 cDNA First Strand Synthesis Kit Thermo Fisher Scientific, Waltham, MA, USA Cat#K1612 Ribo-Zero Gold rRNA removal kit (human, mouse, rat) (24 reactions) Illumina, San Diego, CA, USA Cat#MRZG12324 ProcartaPlex 16-plex Immunoassay (IFN-a, IFN-b, IFN-g, IL-1a,IL-1b, IL-2, IL-6, IL-8, IL-10, IL-12p70, IL-17A, IL-23, CXCL10, CXCL11, CCL5, TNF-a) Thermo Fisher Scientific, Waltham, MA, USA Cat#PPX-16, customized (Continued on next page) e1 Cell Reports 33, 108389, November 17, 2020 Article ll OPEN ACCESS Continued REAGENT or RESOURCE SOURCE IDENTIFIER Deposited Data Triple RNA-seq, A. fumigatus, CMV and H. sapiens https://www.ncbi.nlm.nih.gov/geo/ GSE134344 Dual RNA-seq A. fumigatus and H. sapiens https://www.ncbi.nlm.nih.gov/geo/ GSE135450 Dual RNA-seq CMV and H. sapiens https://www.ncbi.nlm.nih.gov/geo/ GSE136217 Experimental Models: Organisms/Strains Aspergillus fumigatus (wildtype) ATCC ATCC: 46645 Aspergillus fumigatus dTomato ATCC; Lother et al., 2014 ATCC: 46645 Aspergillus fumigatus GFP ATCC; Lother et al., 2014 ATCC: 46645 Oligonucleotides See Table S1 for primer sequences (ALAS1, IL15, CCR7, ZBP1, VEGFA, IFNG, TLR3, CD40, GliT, GliZ, Ilv3, Cat1, SrbA, UL4, UL8, UL20, US16, UL9, US9) This paper N/A Software and Algorithms CellQuest Pro Software (Becton & Dickinson) https://www.bd.com/en-us Becton & Dickinson, Franklin Lakes, NJ, USA FACSDiva Software https://www.bd.com/en-us Becton & Dickinson, Franklin Lakes, NJ, USA FlowJo Software, v10 https://www.flowjo.com; https://www.bd.com/en-us Treestar/ Becton & Dickinson, Ashland, OR, USA FCS Express Software, v7 https://denovosoftware.com De Novo Software, Pasedena, CA, USA NIS Elements Imaging software, v5.02.00 https://www.microscope. healthcare.nikon.com Nikon Instruments, Amsterdam, Netherlands Rhttps://cran.r-project.org/ 3.5.1 featureCounts https://cran.r-project.org/ Rsubread 1.28.0 DESeq https://bioconductor.org 1.30.0 DESeq2 https://bioconductor.org 1.18.1 limma voom https://bioconductor.org 3.34.6 edgeR https://bioconductor.org 3.20.7 geo2RNaseq https://singularity-hub.org/accounts/ login/?next=/collections/4387; https://bitbucket.org/Xentrics/ geo2rnaseq/src/master/ 0.9.12 FastQC https://www.bioinformatics. babraham.ac.uk/projects/fastqc/ 0.11.8 Trimmomatic http://www.usadellab.org/cms/? page=trimmomatic 0.36 HiSat2 https://daehwankimlab.github.io/ hisat2/ 2.1.0 SAMtools https://www.htslib.org/ 1.7 MultiQC https://multiqc.info/ 1.5 Jupyter https://jupyter.org/ 4.4.0 Tidyverse https://cran.r-project.org/ 1.2.1 Magrittr https://cran.r-project.org/ 1.5.0 Reshape2 https://cran.r-project.org/ 1.4.4 ggplot2 https://cran.r-project.org/ 3.1.1 Ggpubr https://cran.r-project.org/ 0.2.5 Ggsci https://cran.r-project.org/ 2.9 Dplyr https://cran.r-project.org/ 0.7.6 Tidyr https://cran.r-project.org/ 1.3.1 Stringr https://cran.r-project.org/ 1.3.0 (Continued on next page) Cell Reports 33, 108389, November 17, 2020 e2 Article ll OPEN ACCESS RESOURCE AVAILABILITY Lead Contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Juergen Loeffler ([email protected]). Materials Availability All primary material generated in this study will be made available upon request following publication. A completed Materials Transfer Agreement might be necessary, especially if there is potential for commercial application. Data and Code Availability All primary sequencing data and processed data described in this manuscript have been deposited in the NCBI Gene Expression Omnibus under the accession numbers GEO: GSE134344, GSE135450 and GSE136217. Code for preprocessing RNA sequencing data and analysis is deposited and available at https://github.com/SchSascha/ manuscript_tripleRNAseq. EXPERIMENTAL MODEL AND SUBJECT DETAILS Healthy blood donors were exclusively in the age between 18-59 years and otherwise excluded from providing blood samples. Both sexes were included in the study equally and at random; the sex of the blood donors was anonymised and represents the normal distribution within this population. Ethics Statement The processing of human peripheral venous blood from healthy adult donors was approved by the Ethical Committee of the University Hospital W€ urzburg (#302/12). METHOD DETAILS Primary cell isolation and differentiation To generate moDCs, monocytes were isolated from leukoreduction system chambers containing blood from healthy volunteers and standard density-gradient centrifugation followed by positive selection using magnetic-activated cell sorting (CD14 MicroBeads, human, Miltenyi Biotec). Monocytes were cultured in CellGenix GMP dendritic cell medium (serum-free, CellGenix) supplemented with 120 mg gentamicin (Refobacin, Merck) in 24-well plates with 1 310 6 cells/ml. Cells were differentiated for 6 days by addition of 1,000 U/ml granulocyte macrophage-colony stimulating factor (Miltenyi Biotec) and 1,000 U/ml interleukin (IL)-4 (Miltenyi Biotec) (Mezger et al., 2008;Paijo et al., 2016;Sallusto and Lanzavecchia, 1994). Continued REAGENT or RESOURCE SOURCE IDENTIFIER AnnotationDbi https://bioconductor.org 1.44.0 org.Hs.eg.db https://bioconductor.org 3.10.0 setRank https://cran.r-project.org/ 1.1.0 Pathview https://bioconductor.org 1.26.0 Pheatmap https://cran.r-project.org/ 1.0.12 igraph https://cran.r-project.org/ 1.2.5 WriteXLS https://cran.r-project.org/ 5.0.0 corrplot https://cran.r-project.org/ 0.84 RcolorBrewer https://cran.r-project.org/ 1.1-2 enrichR https://cran.r-project.org/ 2.1.0 cowplot https://cran.r-project.org/ 1.1.0 Other H. sapiens reference genome https://www.ncbi.nlm.nih.gov/ GRCH 38 v89 A. fumigatus Af293 reference genome http://www.aspergillusgenome.org/ s03-m05-r09 CMV reference genome https://www.ncbi.nlm.nih.gov/ EF999921.1 e3 Cell Reports 33, 108389, November 17, 2020 Article ll OPEN ACCESS Pulmonary pathogens Reporter strain human CMV TB40/E-mNeonGreen and CMV TB40/SE (non-fluorescent) were generated according to published protocols (Kasmapour et al., 2017;Paijo et al., 2016;Sampaio et al., 2017). A. fumigatus (ATCC 46645) germ tubes were prepared overnight in RPMI (Invitrogen) and used for dual or triple RNA-seq. For flow cytometry and microscopy (see below), dTomatoor GFP-expressing A. fumigatus germ tubes were used. moDC infection assays Primary human cells were infected with CMV at a multiplicity of infection (MOI) 3 and with subsequent centrifugal enhancement at 300 xgfor 30 min. A. fumigatus germ tubes, which represent an invasive immunogenic morphotype of the fungus, were added at MOI 0.5. For co-infections, pathogens were added simultaneously or subsequently (Figure 1A). After harvesting, cells were centrifuged and pellets collected in HBSS (Invitrogen) or RNAprotect Cell Reagent (QIAGEN), and cell-free supernatants were stored at 80C. For flow cytometry and microscopy, (see below) moDCs were infected with CMV for 24 h before adding germ tubes, because detection of NeonGreen fluorescence in infected cells requires sufficient replication of the virus. Flow cytometry analysis CMV and A.fumigatus infection rates were determined by flow cytometry, measuring fluorescent signals of moDCs positive for mNeonGreen (CMV) or GFP (A. fumigatus), respectively (Figures S1B and S1C). Viability of moDCs was determined by staining with Viobility 405/520 Fixable Dye (Miltenyi Biotec), defining negative populations as viable. Additionally, cells were stained with anti-CD40-PE-Vio770 (REA733, Miltenyi Biotec), anti-CD80-APC (REA661, Miltenyi Biotec), anti-CD209-VioBlue, (REA617, Miltenyi Biotec), anti-CCR7-VioBlue (REA546, Miltenyi Biotec), anti-TLR2-PE-Vio770 (REA109, Miltenyi Biotec) or anti-TLR3-APC (TLR3.7, Miltenyi Biotec). Mean fluorescence intensities of these surface markers were measured and isotypes subtracted. Data were acquired on a FACSCalibur using CellQuest Software (Becton & Dickinson) or FACSCanto II using FACSDiva software (Becton & Dickinson). Data analysis was done with FlowJo (Treestar/Becton & Dickinson, version 10) or FCS Express 7 (De Novo Software). Microscopy Morphology and fluorescence signals for CMV (mNeonGreen) or A.fumigatus (dTomato) were analyzed in 48-well cell culture plates using a Nikon Eclipse Ti microscope (Nikon) with an Okolab incubator set at 37C. Images were obtained at 20-fold magnification and processed using NIS Elements Imaging software (Nikon, version 5.02.00). Establishment of optimized RNA-seq conditions for single infections As a prerequisite for multi-organism RNA-seq analysis in our infection model, lysis conditions should be sufficiently harsh to disrupt cellular membranes of all interacting organisms, but sufficiently mild to maintain high RNA integrity. Additionally, multiplicities of infection (MOIs) must be adjusted to ensure homogeneous coverage of individual transcriptomes in the resulting sequencing data. That is, the relative proportions of transcriptomes in isolated RNA samples should match the ratio of the respective genome sizes. Therefore, we first evaluated yield and quality of RNA isolated from moDCs infected with either A. fumigatus or CMV at different time points and MOIs, and observed high RNA integrity (RIN > 7) for all conditions (Figures S2A and S2B). While highly abundant in any organism (> 90% of total cellular RNA), ribosomal RNA (rRNA) provides little informative value about cellular physiology. Therefore, ribosomal transcripts are typically depleted from sequencing libraries, either by active rRNA pull-out (e.g., Ribo-Zero technology) or enrichment of polyadenylated transcripts. In principle, both options are suitable for our infection model, since mRNAs of all three interacting organisms are polyadenylated. We found that both approaches efficiently depleted human ribosomal reads (Figure S2C). However, to retain potentially interesting non-polyadenylated transcripts (e.g., microRNAs [miRNAs], small nucle(ol)ar RNAs [snRNAs, snoRNAs] and polyAlong noncoding RNAs [lncRNAs]; Figure S2C), we employed Ribo-Zero technology for further RNA-seq experiments. After rRNA removal from single-infection samples, cDNA libraries were prepared and sequenced to shallow depth (5-8 million reads/library) for initial quality assessment. Obtained sequencing reads aligned to their parental reference genome with little cross-mapping observed (Figures S2A and S2B). In fact, the vast majority of cross-mapped reads derived from mitochondrial genes present in both Aspergillus and human cells, and all of these reads were removed from further analyses. As expected, the fungal-tohuman read ratio in Aspergillus-infected samples increased with MOI. In line with the high proportions of fungal reads (20% of total mapped reads), human exon coverage was the rate-limiting factor (Figure S2A), thus favoring the low-dose infections (MOI 0.5) for further experiments. While the vast majority of reads in CMV-infected moDCs mapped to the human genome, viral read proportion and exon coverage increased time-dependently, indicative of viral replication (Figure S2B). Even with the low sequencing depth used in this pilot experiment, the induction of marker genes for human dendritic cell activation and pathogenicity-related fungal genes was detected (Figure S2D). For example, upregulation of IL-1, CCL3, and TLR2 indicated activation of well-described pro-inflammatory cascades in moDCs infected with Aspergillus (Braedel et al., 2004;Lass-Flo ¨rl et al., 2013;Walsh et al., 2005). Fungal cells also showed elevated expression of genes for toxic molecules such as the gliotoxin GliF (Latge ´, 1999). Similarly, expression profiles during virus infection reflected expected patterns (Figure S2E) as CMV-infected moDCs upregulated IFN-gand CCL2 (Loewendorf & Benedict, 2010;McNab et al., 2015), whereas viral gene expression was generally induced over Cell Reports 33, 108389, November 17, 2020 e4 Article ll OPEN ACCESS time (Figure S2E displays viral expression at two later stages relative to an early infection stage) indicating active proliferation (Dunn et al., 2003;Moutaftsi et al., 2002;Wiertz et al., 1996). RNA extraction, rRNA depletion, cDNA library preparation, and triple RNA-seq For extraction of human, fungal and viral RNA, RiboPure RNA Purification Kit yeast (Thermo Fisher Scientific) was used according to the manufacturer’s instructions. Total RNA was treated with 0.13 U/ml DNase I (Thermo Fisher Scientific) for 30 min at 37C to remove contaminating genomic DNA. The integrity of DNase-treated RNA was assessed on a bioanalyzer (Agilent). All samples had RNA integrity numbers (RIN) R7.0. Where explicitly indicated (as polyA + ), total RNA was directly converted into first-strand cDNA using an oligo(dT) 25 primer, fragmented (four 30 s ultrasound pulses), and processed as outlined below. Otherwise, ribosomal transcripts were actively removed using Ribo-Zero Gold rRNA removal kits (human, mouse, rat) (Illumina) following manufacturer’s instructions for 500 ng DNase-treated RNA as input for rRNA depletion. The cDNA libraries for Illumina sequencing were generated by Vertis Biotechnologie AG, Freising-Weihenstephan, Germany, with rRNA-free RNA sheared via ultrasound sonication (four 30 s pulses, 4C) to generate 200to 400-nucleotide fragments, on average. Fragments < 20 nucleotides were removed using Agencourt RNAClean XP kits (Beckman Coulter Genomics), and Illumina TruSeq adapters were ligated to the 30ends of remaining fragments. First-strand cDNA synthesis was performed using M-MLV reverse transcriptase (NEB) with 30adapters used as primer target sites. First-strand cDNA was purified, and 50Illumina TruSeq sequencing adapters were ligated to 30ends of antisense cDNA. Resulting cDNA was PCR-amplified to about 10 to 20 ng/ml using high-fidelity DNA polymerase. TruSeq barcode sequences were included in 50and 30TruSeq sequencing adapters. The cDNA libraries were purified using Agencourt AMPure XP kits (Beckman Coulter Genomics) and analyzed by capillary electrophoresis (Shimadzu MultiNA microchip). For sequencing, cDNA libraries were pooled in approximately equimolar amounts. Pools were size-fractionated to 200-600 bp using differential cleanup with Agencourt AMPure kits (Beckman Coulter Genomics). Aliquots of cDNA pools were analyzed by capillary electrophoresis (Shimadzu MultiNA microchip). Sequencing was performed on a NextSeq 500 platform (Illumina) at Vertis Biotechnologie AG, Freising-Weihenstephan, Germany (single-end mode, 75 cycles). Quantitative reverse transcription PCR-based validation of differential gene expression RNA of untreated moDCs or moDCs after infection with CMV, A. fumigatus or both was reverse transcribed into cDNA using Thermo Fisher cDNA First Strand Synthesis Kits. Primers (Sigma-Aldrich) were designed using Primer-Blast (Ye et al., 2012) and The Aspergillus Genome Database (Cerqueira et al., 2014), avoiding the occurrence of target sequences in the other two organisms (Table S1). Quantitative reverse transcription (qRT)-PCR was conducted using SYBRGreen Master Mix from (BioRad) in a Step One System (Applied Biosystems). Primer specificity was confirmed by agarose (Roth) gel electrophoresis (Serva) of PCR amplicons using ethidium bromide (Thermo Fisher). Gel images were documented in a Multi-Image Light Cabinet (Alpha Innotech). Multiplex cytokine secretion assays Cell culture supernatants were analyzed by multiplex cytokine secretion assays according to the manufacturer’s instructions using 16-plex ProcartaPlex Immunoassays (Thermo Fisher Scientific) including IFN-a, IFN-b, IFN-g, IL-1a, IL-1b, IL-2, IL-6, IL-8, IL-10, IL-12p70, IL-17A, IL-23, CXCL10, CXCL11, CCL5, and TNF-a. Cytokines with concentration levels above measurement range were set to 1.05 times the maximum. Concentration levels below measurement range or negative concentrations were set to 0. Significant changes in the concentration of cytokines were determined using pairwise two-sided Wilcoxon rank-sum tests. P values for each test were adjusted for multiple testing using FDR. We rejected the null hypothesis for FDR < 0.1. QUANTIFICATION AND STATISTICAL ANALYSIS RNA-seq data processing Preprocessing of raw reads including quality control and gene abundance estimation was done with GEO2RNaseq pipeline version 0.9.12 in Rversion 3.5.1 (Seelbinder et al., 2019). Quality analysis was done with FastQC version 0.11.8 before and after trimming. Read-quality trimming was done with Trimmomatic version 0.36. Adaptor sequences were removed, window size trimming performed (15 nucleotides, average Q < 25) including 50and 30per-base trimming for Q < 3 and removal of sequences shorter than 30 nucleotides. The reference genome in FASTA format was created by combining references Homo sapiens (GRCH 38 v89), A. fumigatus Af293 (s03-m05-r09) and human herpesvirus 5 strain TB30/E clone TB40-BAC4 (EF999921.1). Reference annotation was created by extracting and combining exon features from corresponding annotation files. The reference genome was indexed with exon information using HiSat2 version 2.1.0. Paired-end read alignment used HiSat2 on the created reference genome. Only concordantly aligned pairs of reads were used. Mapping statistics per organism were calculated using the ‘‘calc_triple_mapping_ stats’’ function of GEO2RNaseq. SAMtools version 1.7 with the ‘‘flagstat’’ subcommand was used to deduce alignment quality. Gene abundance estimation was done with featureCounts (Rpackage Rsubread version 1.28.0) in paired-end mode with default parameters. MultiQC version 1.5 was used to summarize the output of FastQC, Trimmomatic, HiSat, featureCounts and SAMtools (Data S1). In addition to the count matrix with gene abundance for all three species, species-specific gene count matrices were extracted e5 Cell Reports 33, 108389, November 17, 2020 Article ll OPEN ACCESS from the complete count matrix. For nonstatistical analyses, count matrices were normalized using median-by-ratio normalization (MRN) as described before (Anders and Huber, 2010). Correlation analysis, principal component analysis, and clustering were performed per species based on MRN gene-abundance data. Clustering between samples used the complete linkage (farthest neighbor) method. Raw files are accessible under the Gene Expression Omnibus accession number GSE134344, GSE135450 and GSE136217. Differential gene expression analysis Differential gene expression was analyzed by GEO2RNaseq per species. Pairwise tests were performed between control, singleinfection, and co-infection groups with and without consideration of infection time. Four statistical tools (DESeq 1.30.0, DESeq2 1.18.1, limma voom 3.34.6 and edgeR 3.20.7) were used, and p values were corrected for multiple testing using the false-discovery rate method q = FDR(p) for each tool. In addition, mean MRN, transcripts per kilobase million (TPKM) and reads per kilobase million (RPKM) values were computed per test per group including corresponding log 2 fold-changes. Gene expression differences were considered significant if they were reported significant by all four tools (q < 0.01 and |log 2 MRN| R1 for H. sapiens; q < 0.05 for A. fumigatus and CMV). To compare RNA-seq and qRT-PCR derived relative differences, expression values were scaled by the maximum value per replicate. To keep statistical analysis comparable, significance was assessed based on pairwise t tests. Resulting p values were corrected for multiple testing using FDR. Tests were performed individually for RNA-seq and qRT-PCR per gene (Figure 5;Figure S5). Interspecies and intraspecies gene expression analyses Abundances of genes with nonzero coverage from all three species and all samples were MRN normalized. Spearman’s correlations between gene abundances were calculated per treatment group (moDC, moDC + A. fumigatus [Afu], moDC + CMV, moDC + CMV + Afu) for infection time 0 h. Only significant correlations with p < 0.01 and absolute correlation R30% were used for further analysis. Rpackage iGraph version 1.2.4.1 was used to create, compare, analyze (node degree and betweenness) and plot significant correlations as networks. Node degree centrality describes the number of edges incident to a node. Node betweenness centrality describes the number of shortest paths through a node for all pairs of nodes. For cross-species analysis, correlations between genes of the same species were ignored. For additional immune system relevant gene correlation analysis, genes from H. sapiens were retained only if they were present in the curated database InnateDB (https://www.innatedb.com/). moDC gene set analysis with distinct expression patterns between, but low variance within different infection etiologies yielded 160 genes that were further filtered for immune system-relevant genes (Figure 4). Cell Reports 33, 108389, November 17, 2020 e6 Article ll OPEN ACCESS