The ESX-1 secretion system is a major determinant of Mycobacterium tuberculosis virulence, although the pathogenic mechanisms resulting from ESX-1-mediated transport remain unclear. By global transcriptional profiling of tissues from mice infected with either wild-type or ESX-1 mutant bacilli, we found that host genes controlled by ESX-1 in vivo are predominantly IFN regulated. ESX-1-mediated secretion is required for the production of host type I IFNs during infection in vivo and in macrophages in vitro. The macrophage signaling pathway leading to the production of type I IFN required the host kinase TANK-binding kinase 1 and occurs independently of TLR signaling. Importantly, the induction of type I IFNs during M. tuberculosis infection is a pathogenic mechanism as mice lacking the type I IFNR were more restrictive for bacterial growth in the spleen than wild-type mice, although growth in the lung was unaffected. We propose that the ESX-1 secretion system secretes effectors into the cytosol of infected macrophages, thereby triggering the type I IFN response for the manipulation of host immunity.

Control of inflammation is crucial to prevent damage to the host during infection. Lipoxins and aspirin-triggered lipoxins are crucial modulators of proinflammatory responses; however, their intracellular mechanisms have not been completely elucidated. We previously showed that lipoxin A4 (LXA4) controls migration of dendritic cells (DCs) and production of interleukin (IL)-12 in vivo. In the absence of LXA4 biosynthetic pathways, the resulting uncontrolled inflammation during infection is lethal, despite pathogen clearance. Here we show that lipoxins activate two receptors in DCs, AhR and LXAR, and that this activation triggers expression of suppressor of cytokine signaling (SOCS)-2. SOCS-2-deficient DCs are hyper-responsive to microbial stimuli, as well as refractory to the inhibitory actions of LXA4, but not to IL-10. Upon infection with an intracellular pathogen, SOCS-2-deficient mice had uncontrolled production of proinflammatory cytokines, decreased microbial proliferation, aberrant leukocyte infiltration and elevated mortality. We also show that SOCS-2 is a crucial intracellular mediator of the anti-inflammatory actions of aspirin-induced lipoxins in vivo.

The actin nucleation factors Spire and Cappuccino regulate the onset of ooplasmic streaming in Drosophila 1-5 . Although this streaming event is microtubule-based, actin assembly is required for its timing. It is not understood how the interaction of microtubules and microfilaments is mediated in this context. Here we demonstrate that Cappuccino and Spire have microtubule and microfilament crosslinking activity. The spire locus encodes several distinct protein isoforms (SpireA, SpireC, and SpireD). SpireD was recently shown to nucleate actin, but the activity of the other isoforms has not been addressed. We find that SpireD does not have crosslinking activity, while SpireC is a potent crosslinker. We show that SpireD binds to Cappuccino and inhibits F-actin/microtubule crosslinking, and activated Rho1 abolishes this inhibition, establishing a mechanistic basis for the regulation of Capu and Spire activity. We propose that Rho1, cappuccino and spire are elements of a conserved developmental cassette that is capable of directly mediating crosstalk between microtubules and microfilaments.Cytoskeletal elements must be coordinately regulated for cells to carry out complex functions, such as the cytoplasmic movements required to disperse or localize intracellular components 5 . The formin homology (FH) protein Cappuccino (Capu) and the WASP homology 2 (WH2) domain-containing protein Spire are both required for the proper timing of one such cytoplasmic movement, ooplasmic streaming in Drosophila 1,3 . In wildtype oocytes, vigorous ooplasmic streaming is associated with rapid growth during stages 10b-13, and is never observed prior to this stage. Mutations in capu and spire result in premature ooplasmic streaming, beginning at stage 7/8 and continuing through stage 13. This premature streaming interferes with transport mechanisms required for the localization of early polarity markers, resulting in disruption of dorsal-ventral and anterior-posterior body axes 1,2 . Both the wildtype streaming event and the premature streaming in capu and spire mutants are microtubule-based 1 . Streaming never takes place in oocytes lacking kinesin, and colcemid injection blocks premature streaming in these mutants 5,6 . Recent work suggests that streaming is restrained by the competing effects of dynein and kinesin, and can be initiated by blocking dynein function 7 . Thus, it is somewhat paradoxical that Spire and Capu nucleate actin, but are not known to affect microtubule architecture or dynamics. Interestingly, the premature streaming seen in capu and spire mutant oocytes can be recapitulated by injection of the actindepolymerizing drug cytochalasin D into wildtype oocytes, suggesting that actin assembly may restrict microtubule rearrangements required for ooplasmic streaming 8 . Presumably, microtubule and microfilament dynamics are coordinated in oogenesis by a group of proteins that includes Spire and Capu, as well as one or more upstream signals. However, the signaling events that combine to encode a "switch" from the n...

Unconventional myosin proteins of the MyTH-FERM superclass are involved in intrafilopodial trafficking, are thought to be mediators of membrane-cytoskeleton interactions, and are linked to several forms of deafness in mammals. Here we show that the Drosophila myosin XV homolog, Sisyphus, is expressed at high levels in leading edge cells and their cellular protrusions during the morphogenetic process of dorsal closure. Sisyphus is required for the correct alignment of cells on opposing sides of the fusing epithelial sheets, as well as for adhesion of the cells during the final zippering/fusion phase. We have identified several putative Sisyphus cargos, including DE-cadherin (also known as Shotgun) and the microtubule-linked proteins Katanin-60, EB1, Milton and aPKC. These cargos bind to the Sisyphus FERM domain, and their binding is in some cases mutually exclusive. Our data suggest a mechanism for Sisyphus in which it maintains a balance between actin and microtubule cytoskeleton components, thereby contributing to cytoskeletal cross-talk necessary for regulating filopodial dynamics during dorsal closure.

Predictive modeling is increasingly being employed to assist human decision-makers. One purported advantage of replacing or augmenting human judgment with computer models in high stakes settings-such as sentencing, hiring, policing, college admissions, and parole decisionsis the perceived "neutrality" of computers. It is argued that because computer models do not hold personal prejudice, the predictions they produce will be equally free from prejudice. There is growing recognition that employing algorithms does not remove the potential for bias, and can even amplify it if the training data were generated by a process that is itself biased. In this paper, we provide a probabilistic notion of algorithmic bias. We propose a method to eliminate bias from predictive models by removing all information regarding protected variables from the data to which the models will ultimately be trained. Unlike previous work in this area, our framework is general enough to accommodate data on any measurement scale. Motivated by models currently in use in the criminal justice system that inform decisions on pre-trial release and parole, we apply our proposed method to a dataset on the criminal histories of individuals at the time of sentencing to produce "race-neutral" predictions of re-arrest. In the process, we demonstrate that a common approach to creating "race-neutral" models-omitting race as a covariate-still results in racially disparate predictions. We then demonstrate that the application of our proposed method to these data removes racial disparities from predictions with minimal impact on predictive accuracy.

Many modern applications collect highly imbalanced categorical data, with some categories relatively rare. Bayesian hierarchical models combat data sparsity by borrowing information, while also quantifying uncertainty. However, posterior computation presents a fundamental barrier to routine use; a single class of algorithms does not work well in all settings and practitioners waste time trying different types of MCMC approaches. This article was motivated by an application to quantitative advertising in which we encountered extremely poor computational performance for common data augmentation MCMC algorithms but obtained excellent performance for adaptive Metropolis. To obtain a deeper understanding of this behavior, we give strong theory results on computational complexity in an infinitely imbalanced asymptotic regime. Our results show computational complexity of Metropolis is logarithmic in sample size, while data augmentation is polynomial in sample size. The root cause of poor performance of data augmentation is a discrepancy between the rates at which the target density and MCMC step sizes concentrate. In general, MCMC algorithms that have a similar discrepancy will fail in large samples -a result with substantial practical impact.

Contingency table analysis routinely relies on log-linear models, with latent structure analysis providing a common alternative. Latent structure models lead to a reduced rank tensor factorization of the probability mass function for multivariate categorical data, while log-linear models achieve dimensionality reduction through sparsity. Little is known about the relationship between these notions of dimensionality reduction in the two paradigms. We derive several results relating the support of a log-linear model to nonnegative ranks of the associated probability tensor. Motivated by these findings, we propose a new collapsed Tucker class of tensor decompositions, which bridge existing PARAFAC and Tucker decompositions, providing a more flexible framework for parsimoniously characterizing multivariate categorical data. Taking a Bayesian approach to inference, we illustrate empirical advantages of the new decompositions.

Protein tyrosine phosphatases (PTPs) are critical cell-signaling molecules. Inhibitors that are selective for individual PTPs would be valuable tools for dissecting complicated phosphorylation networks. However, the common architecture of PTP active sites impedes the discovery of such compounds. To achieve target selectivity, we have redesigned a PTP/inhibitor interface. Site-directed mutagenesis of a prototypical phosphatase, PTP1B, was used to generate "inhibitor-sensitized" PTPs. The PTP1B mutants were targeted by modifying a broad specificity PTP inhibitor with chemical groups that are sterically incompatible with wild-type PTP active sites. From a small panel of putative inhibitors, compounds that selectively inhibit Ile219Ala PTP1B over the wild-type enzyme were identified. Importantly, the corresponding mutation also conferred novel inhibitor sensitivity to T-cell PTP, suggesting that a readily identifiable point mutation can be used to generate a variety of inhibitor-sensitive PTPs.