Smoothed Quantile Regression with Large-Scale Inference.

Quantile regression is a powerful tool for learning the relationship between a response variable and a multivariate predictor while exploring heterogeneous effects. This paper focuses on statistical inference for quantile regression in the "increasing dimension" regime. We provide a comprehensive analysis of a convolution smoothed approach that achieves adequate approximation to computation and inference for quantile regression. This method, which we refer to as conquer, turns the non-differentiable check function into a twice-differentiable, convex and locally strongly convex surrogate, which admits fast and scalable gradient-based algorithms to perform optimization, and multiplier bootstrap for statistical inference. Theoretically, we establish explicit non-asymptotic bounds on estimation and Bahadur-Kiefer linearization errors, from which we show that the asymptotic normality of the conquer estimator holds under a weaker requirement on dimensionality than needed for conventional quantile regression. The validity of multiplier bootstrap is also provided. Numerical studies confirm conquer as a practical and reliable approach to large-scale inference for quantile regression. Software implementing the methodology is available in the R package conquer.

The expanding significance of keratin intermediate filaments in normal and diseased epithelia.

Intermediate filaments are assembled from a diverse group of evolutionary conserved proteins and are specified in a tissue-dependent, cell type-dependent, and context-dependent fashion in the body. Genetic mutations in intermediate filament proteins account for a large number of diseases, ranging from skin fragility conditions to cardiomyopathies and premature aging. Keratins, the epithelial-specific intermediate filaments, are now recognized as multi-faceted effectors in their native context. In this review, we emphasize the recent progress made in defining the role of keratins towards the regulation of cytoarchitecture, cell growth and proliferation, apoptosis, and cell motility during embryonic development, in normal adult tissues, and in select diseases such as cancer.

Type I keratin 17 protein is phosphorylated on serine 44 by p90 ribosomal protein S6 kinase 1 (RSK1) in a growth- and stress-dependent fashion.

Keratin 17 (K17) is a type I intermediate filament protein that is constitutively expressed in ectoderm-derived epithelial appendages and robustly induced in epidermis following injury, during inflammation, and in chronic diseases such as psoriasis and cancer. Mutations within K17 are responsible for two rare diseases related to ectodermal dysplasias. Studies in K17-null mice uncovered several roles for K17, including structural support, resistance to TNFα-induced apoptosis, regulation of protein synthesis, and modulation of cytokine expression. Yet, little is known about the regulation of K17 protein via post-translational modification. Here, we report that serine 44 in the N-terminal head domain of K17 (K17-Ser(44)) is phosphorylated in response to extracellular stimuli (serum, EGF, and the phorbol ester 12-O-tetradecanoylphorbol-13-acetate) that alter skin keratinocyte growth, and to cellular stresses (sorbitol-induced hyperosmotic shock, UV irradiation, and hydrogen peroxide-induced oxidative stress). It also occurs in basaloid skin tumors in situ. Upon its stimulation in skin keratinocytes, K17-Ser(44) phosphorylation is induced rapidly but stays on transiently. The majority of the phosphorylated K17-Ser(44) pool is polymer-bound and is not obviously related to a change in filament organization. The amino acid sequence surrounding K17-Ser(44) matches the consensus for the AGC family of basophilic kinases. We show that p90 RSK1, an AGC kinase involved in the regulation of cell survival and proliferation, phosphorylates K17-Ser(44) in skin keratinocytes. These findings confirm and expand the tight link that has emerged between K17 up-regulation and growth and stress responses in the skin epithelium.

IFN-gamma-dependent regulatory circuits in immune inflammation highlighted in diabetes.

We demonstrate diverse roles of IFN-gamma in the induction and regulation of immune-mediated inflammation using a transfer model of autoimmune diabetes. The diabetogenic CD4(+)BDC2.5 (BDC) T cell clone upon transfer into NOD.scid mice induced destruction of islets of Langerhans leading to diabetes. Administration of a neutralizing Ab to IFN-gamma (H22) resulted in long-term protection (LTP) from diabetes, with inflammation but persistence of a significant, albeit decreased, number of beta cells. BDC T cells were a mixture of cells expressing high, intermediate, and low levels of the TCR. Clonotype(low) BDC T cells were required for LTP. Furthermore, islet-infiltrating leukocytes in the LTP mice contained Foxp3(+)CD4 T cells. Islet inflammation in both diabetic and LTP mice was characterized by heavy infiltration of macrophages. Gene expression profiles indicated that macrophages in diabetic mice were M1 type, while LTP mice contained M2 differentiated. The LTP was abolished if mice were treated with either Ab-depleting CD4 T cells or a neutralizing Ab to CTLA-4, in this case, only at a late stage. Neutralization of IL-10, TGF-beta, glucocorticoid-induced TNF receptor (GITR), or CD25 had no effect. Transfer of only clonotype(high)-expressing BDC T cells induced diabetes; in contrast, H22 Abs did not inhibit diabetes. While clonotype(high) T cells induced diabetes even when IFN-gamma was neutralized, paradoxically there was reduced inflammation and no diabetes if host myeloid cells lacked IFN-gamma receptor. Hence, using monoclonal CD4 T cells, IFN-gamma can have a wide diversity of roles, depending on the setting of the immune process.

Matrix metalloproteinases expressed by astrocytes mediate extracellular amyloid-beta peptide catabolism.

It has been postulated that the development of amyloid plaques in Alzheimer's disease (AD) may result from an imbalance between the generation and clearance of the amyloid-beta peptide (Abeta). Although familial AD appears to be caused by Abeta overproduction, sporadic AD (the most prevalent form) may result from impairment in clearance. Recent evidence suggests that several proteases may contribute to the degradation of Abeta. Furthermore, astrocytes have recently been implicated as a potential cellular mediator of Abeta degradation. In this study, we examined the possibility that matrix metalloproteinases (MMPs), proteases known to be expressed and secreted by astrocytes, could play a role in extracellular Abeta degradation. We found that astrocytes surrounding amyloid plaques showed enhanced expression of MMP-2 and MMP-9 in aged amyloid precursor protein (APP)/presenilin 1 mice. Moreover, astrocyte-conditioned medium (ACM) degraded Abeta, lowering levels and producing several fragments after incubation with synthetic human Abeta(1-40) and Abeta(1-42). This activity was attenuated with specific inhibitors of MMP-2 and -9, as well as in ACM derived from mmp-2 or -9 knock-out (KO) mice. In vivo, significant increases in the steady-state levels of Abeta were found in the brains of mmp-2 and -9 KO mice compared with wild-type controls. Furthermore, pharmacological inhibition of the MMPs with N-[(2R)-2-(hydroxamidocarbonylmethyl)-4-methylpentanoyl]-L-tryptophan methylamide (GM 6001) increased brain interstitial fluid Abeta levels and elimination of half-life in APPsw mice. These results suggest that MMP-2 and -9 may contribute to extracellular brain Abeta clearance by promoting Abeta catabolism.

Location, allocation, relocation: isolating adult tissue stem cells in three dimensions.

The literature on isolation of adult tissue stem cells is vast and disparate. To better organize the field, we redefine 'isolation', re-expressing it as the sum of three component vectors: location, allocation and relocation. Location is the isolation of stem cells in situ by anatomical features. Allocation is physical isolation by cell sorting. Relocation is isolation of the functional properties of a stem cell to regenerate its normal progeny when relocated to a new environment. Techniques for the allocation and relocation of stem cells from certain tissues (e.g. hematopoietic) are currently better defined than their location, whereas those of other tissues (e.g. mammary glands) have had recent advances along all three vectors. Yet another group (e.g. gastric glands), have stem cells with well characterized location, emerging techniques for allocation but still rudimentary techniques for relocation.