Effect of training-sample size and classification difficulty on the accuracy of genomic predictors.

Introduction: As part of the MicroArray Quality Control (MAQC)-II project, this analysis examines how the choice of univariate feature-selection methods and classification algorithms may influence the performance of genomic predictors under varying degrees of prediction difficulty represented by three clinically relevant endpoints. Methods: We used gene-expression data from 230 breast cancers (grouped into training and independent validation sets), and we examined 40 predictors (five univariate feature-selection methods combined with eight different classifiers) for each of the three endpoints. Their classification performance was estimated on the training set by using two different resampling methods and compared with the accuracy observed in the independent validation set. Results: A ranking of the three classification problems was obtained, and the performance of 120 models was estimated and assessed on an independent validation set. The bootstrapping estimates were closer to the validation performance than were the cross-validation estimates. The required sample size for each endpoint was estimated, and both gene-level and pathway-level analyses were performed on the obtained models. Conclusions: We showed that genomic predictor accuracy is determined largely by an interplay between sample size and classification difficulty. Variations on univariate feature-selection methods and choice of classification algorithm have only a modest impact on predictor performance, and several statistically equally good predictors can be developed for any given classification problem.

ZNRD1 (zinc ribbon domain-containing 1) is a host cellular factor that influences HIV-1 replication and disease progression.

BACKGROUND: Human immunodeficiency virus (HIV) takes advantage of multiple host proteins to support its own replication. The gene ZNRD1 (zinc ribbon domain-containing 1) has been identified as encoding a potential host factor that influenced disease progression in HIV-positive individuals in a genomewide association study and also significantly affected HIV replication in a large-scale in vitro short interfering RNA (siRNA) screen. Genes and polymorphisms identified by large-scale analysis need to be followed up by means of functional assays and resequencing efforts to more precisely map causal genes. METHODS: Genotyping and ZNRD1 gene resequencing for 208 HIV-positive subjects (119 who experienced long-term nonprogression [LTNP] and 89 who experienced normal disease progression) was done by either TaqMan genotyping assays or direct sequencing. Genetic association analysis was performed with the SNPassoc package and Haploview software. siRNA and short hairpin RNA (shRNA) specifically targeting ZNRD1 were used to transiently or stably down-regulate ZNRD1 expression in both lymphoid and nonlymphoid cells. Cells were infected with X4 and R5 HIV strains, and efficiency of infection was assessed by reporter gene assay or p24 assay. RESULTS: Genetic association analysis found a strong statistically significant correlation with the LTNP phenotype (single-nucleotide polymorphism rs1048412; [Formula: see text]), independently of HLA-A10 influence. siRNA-based functional analysis showed that ZNRD1 down-regulation by siRNA or shRNA impaired HIV-1 replication at the transcription level in both lymphoid and nonlymphoid cells. CONCLUSION: Genetic association analysis unequivocally identified ZNRD1 as an independent marker of LTNP to AIDS. Moreover, in vitro experiments pointed to viral transcription as the inhibited step. Thus, our data strongly suggest that ZNRD1 is a host cellular factor that influences HIV-1 replication and disease progression in HIV-positive individuals.

Transepithelial transport of HIV-1 by M cells is receptor-mediated.

Human colon carcinoma Caco-2 cell monolayers undergo conversion into cells that share morphological and functional features of M cells when allowed to interact with B lymphocytes. A lymphotropic (X4) HIV-1 strain crosses M cell monolayers and infects underlying CD4(+) target cells. Transport requires both lactosyl cerebroside and CXCR4 receptors, which are expressed on the apical surface of Caco-2 and M cells. Antibodies specific for each receptor block transport. In contrast, a monotropic (R5) HIV-1 strain is unable to cross M cell monolayers and infect underlying monocytes, despite efficient transport of latex beads. Caco-2 and M cells do not express CCR5, but transfection of these cells with CCR5 cDNA restores transport of R5 virus, which demonstrates that HIV-1 transport across M cells is receptor-mediated. The follicle-associated epithelium covering human gut lymphoid follicles expresses CCR5, but not CXCR4, and lactosyl cerebroside, suggesting that HIV-1 infection may occur through M cells and enterocytes at these sites.

DAI/ZBP1 recruits RIP1 and RIP3 through RIP homotypic interaction motifs to activate NF-kappaB.

Detection of viral nucleic acids is central to antiviral immunity. Recently, DAI/ZBP1 (DNA-dependent activator of IRFs/Z-DNA binding protein 1) was identified as a cytoplasmic DNA sensor and shown to activate the interferon regulatory factor (IRF) and nuclear factor-kappa B (NF-kappaB) transcription factors, leading to type-I interferon production. DAI-induced IRF activation depends on TANK-binding kinase 1 (TBK1), whereas signalling pathways and molecular components involved in NF-kappaB activation remain elusive. Here, we report the identification of two receptor-interacting protein (RIP) homotypic interaction motifs (RHIMs) in the DAI protein sequence, and show that these domains relay DAI-induced NF-kappaB signals through the recruitment of the RHIM-containing kinases RIP1 and RIP3. We show that knockdown of not only RIP1, but also RIP3 affects DAI-induced NF-kappaB activation. Importantly, RIP recruitment to DAI is inhibited by the RHIM-containing murine cytomegalovirus (MCMV) protein M45. These findings delineate the DAI signalling pathway to NF-kappaB and suggest a possible new immune modulation strategy of the MCMV.

Characterization of cytoplasmic antiviral signaling pathways

Summary Multicellular organisms have evolved the immune system to protect from pathogen such as viruses, bacteria, fungi or parasites. Detection of invading pathogens by the host innate immune system is crucial for mounting protective responses and depends on the recognition of microbial components by specific receptors. The results presented in this manuscript focus on the signaling pathways involved in the detection of viral infection by the sensing of viral nucleic acids. First, we describe a new regulatory mechanism controlling RNA-sensing antiviral pathways. Our results indicate that TRIF and Cardif, the crucial adaptor proteins for endosomal and cytoplasmic RNA detection signaling pathway, are processed and inactivated by caspases. The second aspect investigated here involves a signaling pathway triggered upon cytosolic DNA sensing. The interferon inducible protein DAI was recently described as a DNA sensor able to induce the activation of IRFs and NF-κΒ transcription factors leading to type I interferon production. Here we identify two RIP homotypic interaction motifs (RHIMs) in DAI and demonstrate that they mediate the recruitment of RIP1 and RIP3 and the subsequent NF-κΒ activation. Moreover, we observed that the mouse cytomegalovirus RHIM- containing protein M45 has the potential to block this signaling cascade by interfering with the formation of the DAI-RIP1/3 signaling complex. Finally, we report the generation and the initial characterization of NLRX1-deficient mice. NLRX1 is a member of the NOD-like receptor family localized to the mitochondria. The function of NLRX1 is still controversial: one study proposed that NLRX1 acts as an inhibitor of the RIG-like receptor (RLR) antiviral pathway by binding the adaptor protein Cardif, whereas another report implicated NLRX1 in the generation of reactive oxygen species (ROS) and the amplification of NF-κΒ and JNK triggered by TNF-α, poly(I:C) or Shigella infection. Collectively, our results indicate that NLRX1-deficiency does not affect RLR signaling nor TNF-α induced responses. Proteomics analysis identified UQCRC2, a subunit of the complex III of the mitochondrial respiratory chain, as a NLRX1 binding partner. This observation might reveal a possible functional link between NLRX1 and mitochondrial respiration and/or ROS generation. Résumé Au cours de l'évolution, les organismes multicellulaires ont développé le système immunitaire afin de se protéger contre les pathogènes. Une étape cruciale pour le déclenchement des réponses protectrices est la reconnaissance par les cellules du système immunitaire de molécules propres aux microbes grâce à des récepteurs spécifiques. Les résultats présentés dans cette thèse décrivent des nouveaux aspects concernant les voies de signalisation impliquées dans la détection des virus. Le premier projet décrit un mécanisme de régulation des voies activées par la détection d'ARN virale. Nos résultats montrent que TRIF et Cardif, des protéines adaptatrices des voies déclenchées par la reconnaissance de ces acides nucléiques au niveau des endosomes et du cytoplasme, sont clivés et inactivés par les caspases. Le projet suivant de notre recherche concerne une voie de signalisation activée par la détection d'ADN au niveau du cytoplasme. La protéine DAI a été récemment décrite comme un senseur pour cet ADN capable d'activer les facteurs de transcription IRF et NF-κΒ et d'induire ainsi la production des interférons de type I. Ici on démontre que DAI interagit avec RIP1 et RIP3 par le biais de domaines appelés RHIM et que ce complexe est responsable de l'activation de NF-κΒ. On a aussi identifié une protéine du cytomégalovirus de la souris, M45, qui contient ce même domaine et on a pu démontrer qu'elle a la capacité d'interférer avec la formation du complexe entre DAI et RIP1/RIP3 bloquant ainsi l'activation de NF-κΒ. Enfin on décrit ici la génération de souris déficientes pour le gène qui code pour la protéine NLRX1. Cette protéine fait partie de la famille des récepteurs NOD et est localisée dans la mitochondrie. Une étude a suggéré que NLRX1 agit comme un inhibiteur des voies antivirales activées par les récepteurs du type RIG-I (RLR) en interagissant avec la protéine adaptatrice Cardif. Une autre étude propose par contre que NLRX1 participe à la production des dérivés réactifs de l'oxygène et contribue ainsi à augmenter l'activation de NF- κΒ et JNK induite par le TNF-α ou le poly(I:C). Nos résultats montrent que l'absence de NLRX1 ne modifie ni la voie de signalisation RLR ni les réponses induites par le TNF-α. Des analyses ultérieures ont permis d'identifier comme partenaire d'interaction de NLRX1 la protéine UQCRC2, une des sous-unités qui composent le complexe III de la chaîne respiratoire mitochondriale. Cette observation pourrait indiquer un lien fonctionnel entre NLRX1 et la respiration mitochondriale ou la production des dérivés réactifs de l'oxygène au niveau de cette organelle.

V3 net charge: additional tool in HIV-1 tropism prediction.

Genotype-based algorithms are valuable tools for the identification of patients eligible for CCR5 inhibitors administration in clinical practice. Among the available methods, geno2pheno[coreceptor] (G2P) is the most used online tool for tropism prediction. This study was conceived to assess if the combination of G2P prediction with V3 peptide net charge (NC) value could improve the accuracy of tropism prediction. A total of 172 V3 bulk sequences from 143 patients were analyzed by G2P and NC values. A phenotypic assay was performed by cloning the complete env gene and tropism determination was assessed on U87_CCR5(+)/CXCR4(+) cells. Sequences were stratified according to the agreement between NC values and G2P results. Of sequences predicted as X4 by G2P, 61% showed NC values higher than 5; similarly, 76% of sequences predicted as R5 by G2P had NC values below 4. Sequences with NC values between 4 and 5 were associated with different G2P predictions: 65% of samples were predicted as R5-tropic and 35% of sequences as X4-tropic. Sequences identified as X4 by NC value had at least one positive residue at positions known to be involved in tropism prediction and positive residues in position 32. These data supported the hypothesis that NC values between 4 and 5 could be associated with the presence of dual/mixed-tropic (DM) variants. The phenotypic assay performed on a subset of sequences confirmed the tropism prediction for concordant sequences and showed that NC values between 4 and 5 are associated with DM tropism. These results suggest that the combination of G2P and NC could increase the accuracy of tropism prediction. A more reliable identification of X4 variants would be useful for better selecting candidates for Maraviroc (MVC) administration, but also as a predictive marker in coreceptor switching, strongly associated with the phase of infection.