Background Fulfilling the promise of human genetics in elucidating disease requires identifying causal variants and genes underlying genetic association signals. Molecular quantitative trait locus (molQTL) analyses, e.g. expression QTL (eQTL) and splicing QTL (sQTL), link genetic variants to intermediate molecular phenotypes, but pinpointing causal variants and their regulatory effects remains challenging. Here, we integrate sQTL analysis with deep-learning-based splicing effect annotation to identify causal genetic variants and elucidate their functional mechanisms affecting human phenotypes. Results Using a single-cell GWAS method (scHi-HOST) on 96 lymphoblastoid cell lines (LCLs) with and without influenza A virus (IAV) infection, we discovered ~ 43,000 sQTLs associated with 217 genes after IAV infection. Integrating sQTLs with AI splice prediction, we uncovered 76 likely causal variants that affect cis-acting molecular splicing components (5’ donor, 3’ acceptor), supported by further computational analysis. Among these, we experimentally validated a causal sQTL signal affecting poly (ADP-ribose) polymerase 2 (PARP2). The causal variant, rs2297616, alters the 5’ splice donor site in the second intron of PARP2 , resulting in two protein isoforms differing by 13 amino acids. The derived A allele was associated with the longer protein isoform and increased IAV levels in LCLs. CRISPR editing validated the causal effect of this variant on both protein length and IAV infection. Lastly, these 76 putative causal sQTLs were further linked to over a hundred GWAS traits, including many variants associated with autoimmune diseases. Conclusions Our work provides a catalog of causal sQTL with direct splicing impacts, providing causal mechanistic insights from genotype to disease susceptibility.

Humans display sexual dimorphism across many traits, but little is known about underlying genetic mechanisms and impacts on disease. We utilized single-cell RNA-seq of 480 lymphoblastoid cell lines to identify 1200 genes with significantly sex-biased expression. While reproducibility was highest among LCL datasets, 71% were found to be sex-biased in at least one GTEx tissue, with a core dataset of 21 genes displaying sex-biased expression across all datasets and tissues examined. While 7.7% of sex-biased genes can be directly explained by differences in the number of sex chromosomes, most sex-biased genes (79%) are targets of transcription factors that display sex-biased expression. FOSL1, ZNF730, ZFX, and ZNF726 appear to make the largest contribution to this based on machine learning and linear modeling approaches, and all four of these transcription factors are regulated by the number of X chromosomes. Further, by testing the difference in slopes of conditionally independent expression quantitative trait loci (eQTL) identified in each sex separately, we identified 2,390 sex-biased eQTL (sb-eQTL) across the genome. While evidence of replication in an independent dataset was modest, permutation analysis demonstrated that sb-eQTL identified using real sex was more likely to have concordant direction of effect. These sb-eQTL are enriched in over 100 GWAS phenotypes, including many loci associated with female-biased autoimmune diseases such as multiple sclerosis. Our results demonstrate widespread genetic impacts on sexual dimorphism and identify possible mechanisms and clinical targets for sex differences in diverse diseases.

Figure S2. Scatter-plots showing a linear relationship between expression of housekeeping genes.

Figure S6. Knockdown of MAPK7 does not increase proliferation rate in primary tumor or lung metastasis.

Figure S5. Knockdown of MAPK7, CDC7, CDC2L5 and CSNKID induces Mesenchymal-Epithelial Transition (MET).

Supplemental Methods, Tables S1-2, Table Legends and Figure Legends Table S1.List of qRT-PCR primers used for epithelial, mesenchymal, inducer and housekeeping genes. Table S2. RNA-ISH quantitation of expression scores used in Figure 5.

Figure S3. Illustration of modeling and correcting for dependence of each marker through the expression level of housekeeping genes.

Figure S1. Summary of kinome screen.

Figure S4. Heat map analysis of the kinome screen generated by the Wilcoxon method.

During pandemics, individuals exhibit differences in risk and clinical outcomes. Here, we developed single-cell high-throughput human in vitro susceptibility testing (scHi-HOST), a method for rapidly identifying genetic variants that confer resistance and susceptibility. We applied this method to influenza A virus (IAV), the cause of four pandemics since the start of the 20th century. scHi-HOST leverages single-cell RNA sequencing (scRNA-seq) to simultaneously assign genetic identity to cells in mixed infections of cell lines of European, African, and Asian origin, reveal associated genetic variants for viral burden, and identify expression quantitative trait loci. Integration of scHi-HOST with human challenge and experimental validation demonstrated that a missense variant in endoplasmic reticulum aminopeptidase 1 (ERAP1; rs27895) increased IAV burden in cells and human volunteers. rs27895 exhibits population differentiation, likely contributing to greater permissivity of cells from African populations to IAV. scHi-HOST is a broadly applicable method and resource for decoding infectious-disease genetics.

Since the identification of sickle cell trait as a heritable form of resistance to malaria, candidate gene studies, linkage analysis paired with sequencing, and genome-wide association (GWA) studies have revealed many examples of genetic resistance and susceptibility to infectious diseases. GWA studies enabled the identification of many common variants associated with small shifts in susceptibility to infectious diseases. This is exemplified by multiple loci associated with leprosy, malaria, HIV, tuberculosis, and coronavirus disease 2019 (COVID-19), which illuminate genetic architecture and implicate pathways underlying pathophysiology. Despite these successes, most of the heritability of infectious diseases remains to be explained. As the field advances, current limitations may be overcome by applying methodological innovations such as cellular GWA studies and phenome-wide association (PheWA) studies as well as by improving methodological rigor with more precise case definitions, deeper phenotyping, increased cohort diversity, and functional validation of candidate loci in the laboratory or human challenge studies.

Human genetic diversity can have profound effects on health outcomes upon exposure to infectious agents. For infections with Chlamydia trachomatis (C. trachomatis), the wide range of genital and ocular disease manifestations are likely influenced by human genetic differences that regulate interactions between C. trachomatis and host cells. We leveraged this diversity in cellular responses to demonstrate the importance of variation at the Toll-like receptor 1 (TLR1), TLR6, and TLR10 locus to cytokine production in response to C. trachomatis. We determined that a single-nucleotide polymorphism (SNP) (rs1057807), located in a region that forms a loop with the TLR6 promoter, is associated with increased expression of TLR1, TLR6, and TLR10 and secreted levels of ten C. trachomatis-induced cytokines. Production of these C. trachomatis-induced cytokines is primarily dependent on MyD88 and TLR6 based on experiments using inhibitors, blocking antibodies, RNAi, and protein overexpression. Population genetic analyses further demonstrated that the mean IL-6 response of cells from two European populations were higher than the mean response of cells from three African populations and that this difference was partially attributable to variation in rs1057807 allele frequency. In contrast, a SNP associated with a different proinflammatory cytokine (rs2869462 associated with the chemokine CXCL10) exhibited an opposite response, underscoring the complexity of how different genetic variants contribute to an individual's immune response. This multidisciplinary study has identified a long-range chromatin interaction and genetic variation that regulates TLR6 to broaden our understanding of how human genetic variation affects the C. trachomatis-induced immune response.

Chlamydia trachomatis is the leading cause of sexually transmitted bacterial infections and a major threat to women’s reproductive health in particular. This obligate intracellular pathogen resides and replicates within a cellular compartment termed an inclusion, where it is sheltered by unknown mechanisms from gamma-interferon (IFNγ)-induced cell-autonomous host immunity. Through a genetic screen, we uncovered the Chlamydia inclusion membrane protein gamma resistance determinant (GarD) as a bacterial factor protecting inclusions from cell-autonomous immunity. In IFNγ-primed human cells, inclusions formed by garD loss-of-function mutants become decorated with linear ubiquitin and are eliminated. Leveraging cellular genome-wide association data, we identified the ubiquitin E3 ligase RNF213 as a candidate anti-Chlamydia protein. We demonstrate that IFNγ-inducible RNF213 facilitates the ubiquitylation and destruction of GarD-deficient inclusions. Furthermore, we show that GarD operates as a cis-acting stealth factor barring RNF213 from targeting inclusions, thus functionally defining GarD as an RNF213 antagonist essential for chlamydial growth during IFNγ-stimulated immunity.

While recent breakthroughs have enabled intense study of bacterial DNA modifications, limitations in current work have potentiated a surprisingly untested narrative that DNA methylation is a common mechanism of the bacterial response to environmental conditions. Essentially, whether epigenetic regulation of bacterial transcription is a common, generalizable phenomenon is a critical unanswered question that we address here.

Embryos (28 hpf) were hybridized to foxd1 and foxg1a DIG-labelled antisense RNA probes. Arrows point at normal expression of nasal (A-A”) and temporal (B-B”) expression of foxg1a and foxd1, respectively. Dorsal views anterior left. Scale bar: 50 μm. (PDF)

(A-F) Ultrastructure analysis of the cornea (arrow) from sox10t3/+ or sox10+/+ sibling embryos (A-B, n = 2 embryos) and sox10t3/t3 homozygous mutants (C-F, n = 4). Low magnification overviews (A, C and E) and respective magnified views (B, D and F) are shown. Stippled line demarcates the lens. sox10t3/t3 homozygous mutants showed vacuolated thick layer of cells (C-D) beneath the corneal stroma (St). Thinning of the same layer with oedema (asterisk) was also observed (E-F). Scale bar: 1 μm (B, D); (G-I) Morpholino knockdown of sox10 caused abnormal AS with reduced number of the corneal endothelium. (G-H’) At 5 dpf the corneal endothelium (grey) forms a monolayer of cells over the lens, which is surrounded by a ring of annular ligament cells (stippled magenta circle) in en face views (G-H). (G’-H’) Corresponding transverse section views along the nasotemporal axis of a control and sox10-KD embryo shown in G-H. (G’-H’) Stippled double-head arc arrows (magenta) show the extent of the corneal endothelium. Note in agreement with the electron micrographs the endothelial cells appear thicker and more irregular than the control endothelial cells in G’. Orientation: nasal (N), temporal (T), dorsal (D) and ventral (V); Scale bar: 50 μm (I) Control embryos at 5 dpf have 28.6 ± 3.6 corneal endothelial cells (n = 25), whereas sox10-KD embryos show 12.8 ± 6.7 cells (n = 28). ***Welch Two sample t-test p-value = 1.3 x10-13 (t = 10.7, df = 42.1). (PDF)

Cranial neural crest (NC) contributes to the developing vertebrate eye. By multidimensional, quantitative imaging, we traced the origin of the ocular NC cells to two distinct NC populations that differ in the maintenance of sox10 expression, Wnt signalling, origin, route, mode and destination of migration. The first NC population migrates to the proximal and the second NC cell group populates the distal (anterior) part of the eye. By analysing zebrafish pax6a/b compound mutants presenting anterior segment dysgenesis, we demonstrate that Pax6a/b guide the two NC populations to distinct proximodistal locations. We further provide evidence that the lens whose formation is pax6a/b-dependent and lens-derived TGFβ signals contribute to the building of the anterior segment. Taken together, our results reveal multiple roles of Pax6a/b in the control of NC cells during development of the anterior segment.

(A-C’) 3D rendered images of a Tg(sox10:mem-tdEosFP) embryo are shown from three different angles: dorsal views (A-A’), lateral views (B-B’) and ventral views (C-C’). Anterior, left. Two selected time points (A-C: 15 hpf; A’-C’: 18 hpf) are shown. The grey spheroid represents the optic cup (OC). One side of the eye (rectangular region) in the dorsal view (A-A’) is highlighted in lateral (B-B’) and ventral views (C-C’). (A-C) At 15 hpf, 1°NC cells start covering the optic cup from its posterior end and its proximal side (arrowheads, A). Only a small dorsal portion of the eye is covered by 1°NC cells (arrowheads in B, arrow in C), while the ventral half of the eye is not covered with 1°NC cells (C). The arrow in C points at 1°NC cells on the dorsal side of the eye. (A’-C’) By 18 hpf, 1°NC cells have reached the rostral end of the optic cup (arrowhead in A’), with the proximal side of the eye covered roughly halfway (arrows in B’ and arrowheads in C’). Importantly, 1°NC cells exclusively migrate over the proximal side of the eye (arrows, B’) and do not migrate over the lens side (distal side; arrowheads, B’). Note that 2°NC cells in clusters (asterisks in A’) have not yet reached the eye at 18 hpf. (D-F) Ultrastructure of 1°NC cells (magenta). (D) 1°NC cells are located on the proximal side of the optic cup (blue; transverse view at 18 hpf). The regions indicated by squares are shown in E and F. Black honey comb pattern is from the grid. (E) 1°NC cells form a mono-layered sheet. (F) A 1°NC cell (coloured in magenta) show a large cell-cell contact plane with neighbouring 1°NC cells. OSE: optic surface epithelium; Br: brain. (PDF)

Wildtype embryos at 18 hpf were examined for the indicated combinations of diencephalic/mesencephalic marker genes by fluorescence double in situ hybridisation. Anterior top, dorsal views. The boundary between diencephalon (d) and mesencephalon (m) is indicated by a stippled horizontal line. (A-B) In dorsal views, both sox10 and sox9b appear co-expressed with pax6b that is expressed in the posterior diencephalon. However, co-localization analysis showed neither sox10- nor sox9b-positive cells co-express pax6b (A’ and B’). The posterior end of pax6b expression marks the boundary between diencephalon and mesencephalon (stippled line). (C, C’) sox9b-positive cells (magenta) in the diencephalon co-express the epiphysis marker gene, otx5 (green). Scale Bar: (A-C, A’-C’) 50 μm. (PDF)

Susceptibility to infectious diseases is determined by a complex interaction between host and pathogen. For infections with the obligate intracellular bacterium Chlamydia trachomatis, variation in immune activation and disease presentation are regulated by both host genetic diversity and pathogen immune evasion. Previously, we discovered a single nucleotide polymorphism (rs2869462) associated with absolute abundance of CXCL10, a pro-inflammatory T-cell chemokine. Here, we report that levels of CXCL10 change during C. trachomatis infection of cultured cells in a manner dependent on both host and pathogen. Linear modeling of cellular traits associated with CXCL10 levels identified a strong, negative correlation with bacterial burden, suggesting that C. trachomatis actively suppresses CXCL10. We identified the pathogen-encoded factor responsible for this suppression as the chlamydial protease- or proteasome-like activity factor, CPAF. Further, we applied our modeling approach to other host cytokines in response to C. trachomatis and found evidence that RANTES, another T-cell chemoattractant, is actively suppressed by Chlamydia. However, this observed suppression of RANTES is not mediated by CPAF. Overall, our results demonstrate that CPAF suppresses CXCL10 to evade the host cytokine response and that modeling of cellular infection parameters can reveal previously unrecognized facets of host–pathogen interactions.