SummaryHuman cytomegalovirus (HCMV) is an important pathogen with multiple immune evasion strategies, including virally facilitated degradation of host antiviral restriction factors. Here, we describe a multiplexed approach to discover proteins with innate immune function on the basis of active degradation by the proteasome or lysosome during early-phase HCMV infection. Using three orthogonal proteomic/transcriptomic screens to quantify protein degradation, with high confidence we identified 35 proteins enriched in antiviral restriction factors. A final screen employed a comprehensive panel of viral mutants to predict viral genes that target >250 human proteins. This approach revealed that helicase-like transcription factor (HLTF), a DNA helicase important in DNA repair, potently inhibits early viral gene expression but is rapidly degraded during infection. The functionally unknown HCMV protein UL145 facilitates HLTF degradation by recruiting the Cullin4 E3 ligase complex. Our approach and data will enable further identifications of innate pathways targeted by HCMV and other viruses.

NKG2D plays a major role in controlling immune responses through the regulation of natural killer (NK) cells, αβ and γδ T-cell function. This activating receptor recognizes eight distinct ligands (the MHC Class I polypeptide-related sequences (MIC) A andB, and UL16-binding proteins (ULBP)1–6) induced by cellular stress to promote recognition cells perturbed by malignant transformation or microbial infection. Studies into human cytomegalovirus (HCMV) have aided both the identification and characterization of NKG2D ligands (NKG2DLs). HCMV immediate early (IE) gene up regulates NKGDLs, and we now describe the differential activation of ULBP2 and MICA/B by IE1 and IE2 respectively. Despite activation by IE functions, HCMV effectively suppressed cell surface expression of NKGDLs through both the early and late phases of infection. The immune evasion functions UL16, UL142, and microRNA(miR)-UL112 are known to target NKG2DLs. While infection with a UL16 deletion mutant caused the expected increase in MICB and ULBP2 cell surface expression, deletion of UL142 did not have a similar impact on its target, MICA. We therefore performed a systematic screen of the viral genome to search of addition functions that targeted MICA. US18 and US20 were identified as novel NK cell evasion functions capable of acting independently to promote MICA degradation by lysosomal degradation. The most dramatic effect on MICA expression was achieved when US18 and US20 acted in concert. US18 and US20 are the first members of the US12 gene family to have been assigned a function. The US12 family has 10 members encoded sequentially through US12–US21; a genetic arrangement, which is suggestive of an ‘accordion’ expansion of an ancestral gene in response to a selective pressure. This expansion must have be an ancient event as the whole family is conserved across simian cytomegaloviruses from old world monkeys. The evolutionary benefit bestowed by the combinatorial effect of US18 and US20 on MICA may have contributed to sustaining the US12 gene family.

Human cytomegalovirus (HCMV) UL141 induces protection against natural killer cell-mediated cytolysis by downregulating cell surface expression of CD155 (nectin-like molecule 5; poliovirus receptor), a ligand for the activating receptor DNAM-1 (CD226). However, DNAM-1 is also recognized to bind a second ligand, CD112 (nectin-2). We now show that HCMV targets CD112 for proteasome-mediated degradation by 48 h post-infection, thus removing both activating ligands for DNAM-1 from the cell surface during productive infection. Significantly, cell surface expression of both CD112 and CD155 was restored when UL141 was deleted from the HCMV genome. While gpUL141 alone is sufficient to mediate retention of CD155 in the endoplasmic reticulum, UL141 requires assistance from additional HCMV-encoded functions to suppress expression of CD112.

Human cytomegalovirus (HCMV) US2, US3, US6 and US11 act in concert to prevent immune recognition of virally infected cells by CD8+ T-lymphocytes through downregulation of MHC class I molecules (MHC-I). Here we show that US2 function goes far beyond MHC-I degradation. A systematic proteomic study using Plasma Membrane Profiling revealed US2 was unique in downregulating additional cellular targets, including: five distinct integrin α-chains, CD112, the interleukin-12 receptor, PTPRJ and thrombomodulin. US2 recruited the cellular E3 ligase TRC8 to direct the proteasomal degradation of all its targets, reminiscent of its degradation of MHC-I. Whereas integrin α-chains were selectively degraded, their integrin β1 binding partner accumulated in the ER. Consequently integrin signaling, cell adhesion and migration were strongly suppressed. US2 was necessary and sufficient for degradation of the majority of its substrates, but remarkably, the HCMV NK cell evasion function UL141 requisitioned US2 to enhance downregulation of the NK cell ligand CD112. UL141 retained CD112 in the ER from where US2 promoted its TRC8-dependent retrotranslocation and degradation. These findings redefine US2 as a multifunctional degradation hub which, through recruitment of the cellular E3 ligase TRC8, modulates diverse immune pathways involved in antigen presentation, NK cell activation, migration and coagulation; and highlight US2’s impact on HCMV pathogenesis.

BackgroundAtrial fibrillation (AF) is the most common sustained heart arrhythmia. However, as many cases are asymptomatic, a large proportion of patients remain undiagnosed until serious complications arise. Efficient, cost-effective detection of the undiagnosed may be supported by risk-prediction models relating patient factors to AF risk. However, there exists a need for an implementable risk model that is contemporaneous and informed by routinely collected patient data, reflecting the real-world pathology of AF.MethodsThis study sought to develop and evaluate novel and conventional statistical and machine learning models for risk-predication of AF. This was a retrospective, cohort study of adults (aged ≥30 years) without a history of AF, listed on the Clinical Practice Research Datalink, from January 2006 to December 2016. Models evaluated included published risk models (Framingham, ARIC, CHARGE-AF), machine learning models, which evaluated baseline and time-updated information (neural network, LASSO, random forests, support vector machines), and Cox regression.ResultsAnalysis of 2,994,837 individuals (3.2% AF) identified time-varying neural networks as the optimal model achieving an AUROC of 0.827 vs. 0.725, with number needed to screen of 9 vs. 13 patients at 75% sensitivity, when compared with the best existing model CHARGE-AF. The optimal model confirmed known baseline risk factors (age, previous cardiovascular disease, antihypertensive medication usage) and identified additional time-varying predictors (proximity of cardiovascular events, body mass index (both levels and changes), pulse pressure, and the frequency of blood pressure measurements).ConclusionThe optimal time-varying machine learning model exhibited greater predictive performance than existing AF risk models and reflected known and new patient risk factors for AF.