Analyzing gene expression data in terms of gene sets: methodological issues

MOTIVATION: Many statistical tests have been proposed in recent years for analyzing gene expression data in terms of gene sets, usually from Gene Ontology. These methods are based on widely different methodological assumptions. Some approaches test differential expression of each gene set against differential expression of the rest of the genes, whereas others test each gene set on its own. Also, some methods are based on a model in which the genes are the sampling units, whereas others treat the subjects as the sampling units. This article aims to clarify the assumptions behind different approaches and to indicate a preferential methodology of gene set testing. RESULTS: We identify some crucial assumptions which are needed by the majority of methods. P-values derived from methods that use a model which takes the genes as the sampling unit are easily misinterpreted, as they are based on a statistical model that does not resemble the biological experiment actually performed. Furthermore, because these models are based on a crucial and unrealistic independence assumption between genes, the P-values derived from such methods can be wildly anti-conservative, as a simulation experiment shows. We also argue that methods that competitively test each gene set against the rest of the genes create an unnecessary rift between single gene testing and gene set testing.     

Dynamic Changes in ANGUSTIFOLIA3 Complex Composition Reveal a Growth Regulatory Mechanism in the Maize Leaf

Most molecular processes during plant development occur with a particular spatio-temporal specificity. Thus far, it has remained technically challenging to capture dynamic protein-protein interactions within a growing organ, where the interplay between cell division and cell expansion is instrumental. Here, we combined high-resolution sampling of the growing maize (Zea mays) leaf with tandem affinity purification followed by mass spectrometry. Our results indicate that the growth-regulating SWI/SNF chromatin remodeling complex associated with ANGUSTIFOLIA3 (AN3) was conserved within growing organs and between dicots and monocots. Moreover, we were able to demonstrate the dynamics of the AN3-interacting proteins within the growing leaf, since copurified GROWTH-REGULATING FACTORs (GRFs) varied throughout the growing leaf. Indeed, GRF1, GRF6, GRF7, GRF12, GRF15, and GRF17 were significantly enriched in the division zone of the growing leaf, while GRF4 and GRF10 levels were comparable between division zone and expansion zone in the growing leaf. These dynamics were also reflected at the mRNA and protein levels, indicating tight developmental regulation of the AN3-associated chromatin remodeling complex. In addition, the phenotypes of maize plants overexpressing miRNA396a-resistant GRF1 support a model proposing that distinct associations of the chromatin remodeling complex with specific GRFs tightly regulate the transition between cell division and cell expansion. Together, our data demonstrate that advancing from static to dynamic protein-protein interaction analysis in a growing organ adds insights in how developmental switches are regulated.     

A kaleidoscopic view of the Arabidopsis core cell cycle interactome

Although protein-protein interaction (PPI) networks have been shown to offer a systems-wide view of cellular processes, only a few plant PPI maps are available. Recently, the core cell cycle of Arabidopsis thaliana has been analyzed by three independent PPI technologies, including yeast two-hybrid systems, bimolecular fluorescence complementation and tandem affinity purification. Here, we merge the three interactomes with literature-curated and computationally predicted interactions, paving the way for a comprehensive picture of the plant core cell cycle machinery. Platform-specific interactions unveil the strengths and weaknesses of each detection method and give insights into the nature of the interactions among cell cycle proteins. Moreover, comparison of the obtained data reveals that a complete interactome can only be obtained when multiple techniques are applied in parallel.     

Location of immunization and interferon-γ are central to induction of salivary gland dysfunction in Ro60 peptide immunized model of Sjögren's syndrome

Introduction

Anti-Ro antibodies can be found in the serum of the majority of patients with Sjögren''s syndrome (SS). Immunization with a 60-kDa Ro peptide has been shown to induce SS-like symptoms in mice. The aim of this study was to investigate factors involved in salivary gland (SG) dysfunction after immunization and to test whether the induction of SS could be improved.

Methods

Ro60 peptide immunization was tested in Balb/c mice, multiple antigenic peptide (MAP)-Ro60 and Pertussis toxin (PTX) were tested in SJL/J mice. In addition, two injection sites were compared in these two strains: the abdominal area and the tailbase. Each group of mice was tested for a loss of SG function, SG lymphocytic infiltration, anti-Ro and anti-La antibody formation, and cytokine production in cultured cells or homogenized SG extracts.

Results

Ro60 peptide immunization in the abdominal area of female Balb/c mice led to impaired SG function, which corresponded with increased Th1 cytokines (IFN-γ and IL-12) systemically and locally in the SG. Moreover, changing the immunization conditions to MAP-Ro60 in the abdominal area, and to lesser extend in the tailbase, also led to impaired SG function in SJL/J mice. As was seen in the Balb/c mice, increased IFN-γ in the SG draining lymph nodes accompanied the SG dysfunction. However, no correlation was observed with anti-MAP-Ro60 antibody titers, and there was no additional effect on disease onset or severity.

Conclusions

Effective induction of salivary gland dysfunction after Ro60 peptide immunization depended on the site of injection. Disease induction was not affected by changing the immunization conditions. However, of interest is that the mechanism of action of Ro60 peptide immunization appears to involve an increase in Th1 cytokines, resulting in the induction of SG dysfunction.     

O-acetylation of cryptococcal capsular glucuronoxylomannan is essential for interference with neutrophil migration

The capsular polysaccharide glucuronoxylomannan (GXM) of Cryptococcus neoformans has been shown to interfere with neutrophil migration. Although several receptors have been implied to mediate this process, the structural perspectives are unknown. Here, we assess the contribution of 6-O-acetylation and xylose substitution of the (1-->3)-alpha-d-mannan backbone of GXM, the variable structural features of GXM, to the interference with neutrophil migration. We compare chemically deacetylated GXM and acetyl- or xylose-deficient GXM from genetically modified strains with wild-type GXM in their ability to inhibit the different phases of neutrophil migration. Additionally, we verify the effects of de-O-acetylation on neutrophil migration in vivo. De-O-acetylation caused a dramatic reduction of the inhibitory capacity of GXM in the in vitro assays for neutrophil chemokinesis, rolling on E-selectin and firm adhesion to endothelium. Genetic removal of xylose only marginally reduced the ability of GXM to reduce firm adhesion. In vivo, chemical deacetylation of GXM significantly reduced its ability to interfere with neutrophil recruitment in a model of myocardial ischemia (65% reduction vs a nonsignificant reduction in tissue myeloperoxidase, respectively). Our findings indicate that 6-O-acetylated mannose of GXM is a crucial motive for the inhibition of neutrophil recruitment.     

A goodness-of-fit test for multinomial logistic regression

This article presents a score test to check the fit of a logistic regression model with two or more outcome categories. The null hypothesis that the model fits well is tested against the alternative that residuals of samples close to each other in covariate space tend to deviate from the model in the same direction. We propose a test statistic that is a sum of squared smoothed residuals, and show that it can be interpreted as a score test in a random effects model. By specifying the distance metric in covariate space, users can choose the alternative against which the test is directed, making it either an omnibus goodness-of-fit test or a test for lack of fit of specific model variables or outcome categories.     

Effects of Lactococcus lactis on Composition of Intestinal Microbiota: Role of Nisin

This study examined the ability of (i) pure nisin, (ii) nisin-producing Lactococcus lactis strain CHCC5826, and (iii) the non-nisin-producing L. lactis strain CHCH2862 to affect the composition of the intestinal microbiota of human flora-associated rats. The presence of both the nisin-producing and the non-nisin-producing L. lactis strains significantly increased the number of Bifidobacterium cells in fecal samples during the first 8 days but decreased the number of enterococci/streptococci in duodenum, ileum, cecum, and colon samples as detected by selective cultivation. No significant changes in the rat fecal microbiota were observed after dosage with nisin. Pearson cluster analysis of denaturing gradient gel electrophoresis profiles of the 16S rRNA genes present in the fecal microbial population revealed that the microbiota of animals dosed with either of the two L. lactis strains were different from that of control animals dosed with saline. However, profiles of the microbiota from animals dosed with nisin did not differ from the controls. The concentrations of nisin estimated by competitive enzyme-linked immunosorbent assay (ELISA) were approximately 10-fold higher in the small intestine and 200-fold higher in feces than the corresponding concentrations estimated by a biological assay. This indicates that nisin was degraded or inactivated in the gastrointestinal tract, since fragments of this bacteriocin are detected by ELISA while an intact molecule is needed to retain biological activity.     

Genome-wide microRNA expression analysis of clear cell renal cell carcinoma by next generation deep sequencing

MicroRNAs (miRNAs), non-coding RNAs regulating gene expression, are frequently aberrantly expressed in human cancers. Next-generation deep sequencing technology enables genome-wide expression profiling of known miRNAs and discovery of novel miRNAs at unprecedented quantitative and qualitative accuracy. Deep sequencing was performed on 11 fresh frozen clear cell renal cell carcinoma (ccRCC) and adjacent non-tumoral renal cortex (NRC) pairs, 11 additional frozen ccRCC tissues, and 2 ccRCC cell lines (n?=?35). The 22 ccRCCs patients belonged to 3 prognostic sub-groups, i.e. those without disease recurrence, with recurrence and with metastatic disease at diagnosis. Thirty-two consecutive samples (16 ccRCC/NRC pairs) were used for stem-loop PCR validation. Novel miRNAs were predicted using 2 distinct bioinformatic pipelines. In total, 463 known miRNAs (expression frequency 1-150,000/million) were identified. We found that 100 miRNA were significantly differentially expressed between ccRCC and NRC. Differential expression of 5 miRNAs was confirmed by stem-loop PCR in the 32 ccRCC/NRC samples. With respect to RCC subgroups, 5 miRNAs discriminated between non-recurrent versus recurrent and metastatic disease, whereas 12 uniquely distinguished non-recurrent versus metastatic disease. Blocking overexpressed miR-210 or miR-27a in cell line SKCR-7 by transfecting specific antagomirs did not result in significant changes in proliferation or apoptosis. Twenty-three previously unknown miRNAs were predicted in silico. Quantitative genome-wide miRNA profiling accurately separated ccRCC from (benign) NRC. Individual differentially expressed miRNAs may potentially serve as diagnostic or prognostic markers or future therapeutic targets in ccRCC. The biological relevance of candidate novel miRNAs is unknown at present.     

The inheritance procedure: multiple testing of tree-structured hypotheses

Hypotheses tests in bioinformatics can often be set in a tree structure in a very natural way, e.g. when tests are performed at probe, gene, and chromosome level. Exploiting this graph structure in a multiple testing procedure may result in a gain in power or increased interpretability of the results.We present the inheritance procedure, a method of familywise error control for hypotheses structured in a tree. The method starts testing at the top of the tree, following up on those branches in which it finds significant results, and following up on leaf nodes in the neighborhood of those leaves. The method is a uniform improvement over a recently proposed method by Meinshausen. The inheritance procedure has been implemented in the globaltest package which is available on www.bioconductor.org.