Should We Abandon the t-Test in the Analysis of Gene Expression Microarray Data: A Comparison of Variance Modeling Strategies

High-throughput post-genomic studies are now routinely and promisingly investigated in biological and biomedical research. The main statistical approach to select genes differentially expressed between two groups is to apply a t-test, which is subject of criticism in the literature. Numerous alternatives have been developed based on different and innovative variance modeling strategies. However, a critical issue is that selecting a different test usually leads to a different gene list. In this context and given the current tendency to apply the t-test, identifying the most efficient approach in practice remains crucial. To provide elements to answer, we conduct a comparison of eight tests representative of variance modeling strategies in gene expression data: Welch''s t-test, ANOVA [1], Wilcoxon''s test, SAM [2], RVM [3], limma [4], VarMixt [5] and SMVar [6]. Our comparison process relies on four steps (gene list analysis, simulations, spike-in data and re-sampling) to formulate comprehensive and robust conclusions about test performance, in terms of statistical power, false-positive rate, execution time and ease of use. Our results raise concerns about the ability of some methods to control the expected number of false positives at a desirable level. Besides, two tests (limma and VarMixt) show significant improvement compared to the t-test, in particular to deal with small sample sizes. In addition limma presents several practical advantages, so we advocate its application to analyze gene expression data.     

Loss of autocrine endothelial-derived VEGF significantly reduces hemangiosarcoma development in conditional p53-deficient mice

Malignant transformation of the endothelium is rare, and hemangiosarcomas comprise only 1% of all sarcomas. For this reason and due to the lack of appropriate mouse models, the genetic mechanisms of malignant endothelial transformation are poorly understood. Here, we describe a hemangiosarcoma mouse model generated by deleting p53 specifically in the endothelial and hematopoietic lineages. This strategy led to a high incidence of hemangiosarcoma, with an average latency of 25 weeks. To study the in vivo roles of autocrine or endothelial cell autonomous VEGF signaling in the initiation and/or progression of hemangiosarcomas, we genetically deleted autocrine endothelial sources of VEGF in this mouse model. We found that loss of even a single conditional VEGF allele results in substantial rescue from endothelial cell transformation. These findings highlight the important role of threshold levels of autocrine VEGF signaling in endothelial malignancies and suggest a new approach for hemangiosarcoma treatment using targeted autocrine VEGF inhibition.     

Diversity of the cassiicolin gene in Corynespora cassiicola and relation with the pathogenicity in Hevea brasiliensis

Corynespora cassiicola is an important plant pathogenic Ascomycete causing the damaging Corynespora Leaf Fall (CLF) disease in rubber tree (Hevea brasiliensis). A small secreted glycoprotein named cassiicolin was previously described as an important effector of C. cassiicola. In this study, the diversity of the cassiicolin-encoding gene was analysed in C. cassiicola isolates sampled from various hosts and geographical origins. A cassiicolin gene was detected in 47 % of the isolates, encoding up to six distinct protein isoforms. In three isolates, two gene variants encoding cassiicolin isoforms Cas2 and Cas6 were found in the same isolate. A phylogenetic tree based on four combined loci and elucidating the diversity of the whole collection was strongly structured by the toxin class, as defined by the cassiicolin isoform. The isolates carrying the Cas1 gene (toxin class Cas1), all grouped in the same highly supported clade, were found the most aggressive on two rubber tree cultivars. Some isolates in which no Cas gene was detected could nevertheless generate moderate symptoms, suggesting the existence of other yet uncharacterized effectors. This study provides a useful base for future studies of C. cassiicola population biology and epidemiological surveys in various host plants.     

Genome-Wide Investigation of DNA Methylation Marks Associated with FV Leiden Mutation

In order to investigate whether DNA methylation marks could contribute to the incomplete penetrance of the FV Leiden mutation, a major genetic risk factor for venous thrombosis (VT), we measured genome-wide DNA methylation levels in peripheral blood samples of 98 VT patients carrying the mutation and 251 VT patients without the mutation using the dedicated Illumina HumanMethylation450 array. The genome-wide analysis of 388,120 CpG probes identified three sites mapping to the SLC19A2 locus whose DNA methylation levels differed significantly (p<3 10⁻⁸) between carriers and non-carriers. The three sites replicated (p<2 10⁻⁷) in an independent sample of 214 individuals from five large families ascertained on VT and FV Leiden mutation among which 53 were carriers and 161 were non-carriers of the mutation. In both studies, these three CpG sites were also associated (2.33 10⁻¹¹<p<3.02 10⁻⁴) with biomarkers of the Protein C pathway known to be influenced by the FV Leiden mutation. A comprehensive linkage disequilibrium (LD) analysis of the whole locus revealed that the original associations were due to LD between the FV Leiden mutation and a block of single nucleotide polymorphisms (SNP) located in SLC19A2. After adjusting for this block of SNPs, the FV Leiden mutation was no longer associated with any CpG site (p>0.05). In conclusion, our work clearly illustrates some promises and pitfalls of DNA methylation investigations on peripheral blood DNA in large epidemiological cohorts. DNA methylation levels at SLC19A2 are influenced by SNPs in LD with FV Leiden, but these DNA methylation marks do not explain the incomplete penetrance of the FV Leiden mutation.     

An Animal Model of Type A Cystinuria Due to Spontaneous Mutation in 129S2/SvPasCrl Mice

Cystinuria is an autosomal recessive disease caused by the mutation of either SLC3A1 gene encoding for rBAT (type A cystinuria) or SLC7A9 gene encoding for b^0,+AT (type B cystinuria). Here, we evidenced in a commonly used congenic 129S2/SvPasCrl mouse substrain a dramatically high frequency of kidney stones that were similar to those of patients with cystinuria. Most of 129S2/SvPasCrl exhibited pathognomonic cystine crystals in urine and an aminoaciduria profile similar to that of patients with cystinuria. In addition, we observed a heterogeneous inflammatory infiltrate and cystine tubular casts in the kidney of cystinuric mice. As compared to another classical mouse strain, C57BL/6J mice, 129S2/SvPasCrl mice had an increased mortality associated with bilateral obstructive hydronephrosis. In 129S2/SvPasCrl mice, the heavy subunit rBAT of the tetrameric transporter of dibasic amino acids was absent in proximal tubules and we identified a single pathogenic mutation in a highly conserved region of the Slc3a1 gene. This novel mouse model mimicking human disease would allow us further pathophysiological studies and may be useful to analyse the crystal/tissue interactions in cystinuria.     

Cdc42 controls the dilation of the exocytotic fusion pore by regulating membrane tension

Membrane fusion underlies multiple processes, including exocytosis of hormones and neurotransmitters. Membrane fusion starts with the formation of a narrow fusion pore. Radial expansion of this pore completes the process and allows fast release of secretory compounds, but this step remains poorly understood. Here we show that inhibiting the expression of the small GTPase Cdc42 or preventing its activation with a dominant negative Cdc42 construct in human neuroendocrine cells impaired the release process by compromising fusion pore enlargement. Consequently the mode of vesicle exocytosis was shifted from full-collapse fusion to kiss-and-run. Remarkably, Cdc42-knockdown cells showed reduced membrane tension, and the artificial increase of membrane tension restored fusion pore enlargement. Moreover, inhibiting the motor protein myosin II by blebbistatin decreased membrane tension, as well as fusion pore dilation. We conclude that membrane tension is the driving force for fusion pore dilation and that Cdc42 is a key regulator of this force.     

Immunohistochemical Evaluation of Global DNA Methylation: Comparison with in Vitro Radiolabeled Methyl Incorporation Assay

The in vitro radiolabeled methyl incorporation assay, a commonly used technique to evaluate global methylation of DNA, has some disadvantages and limitations. The purpose of the present study was to compare the results of global DNA methylation evaluated by radiolabeled methyl incorporation (CPM/μg of DNA) with immunohistochemical staining of the same tissue sections with a monoclonal antibody developed against 5-methylcytosine (5-mc). We used archival specimens of squamous cell cancer (SCC) of the human lung with a matched uninvolved specimen (n = 18 pairs) and 18 lung specimens from subjects without lung cancer (noncancer specimens) to make this comparison. The immunostaining for 5-mc was reported as a percentage of cells positive for staining as well as a weighted average of the intensity score. The results suggested that both radiolabeled methyl incorporation assay and immunostaining for 5-mc can be used to demonstrate hypomethylation of DNA in SCC tissues compared to matched uninvolved tissues. An advantage of immunostaining, however, is its ability to demonstrate hypomethylation of SCC compared to adjacent bronchial mucosa on the same archival specimen, obviating the need to use sections from both SCC and matched uninvolved tissues. Only by using the immunostaining technique were we able to document a statistically significant difference in DNA methylation between SCC and noncancer tissues. We conclude that the immunostaining technique has advantages over the radiolabeled methyl incorporation assay and may be best suited for evaluation of global DNA methylation when the methylation status of cancer cannot be normalized by methyl incorporation of normal tissues or when the number of samples available for evaluation is small.