Inflammatory skin disease in transgenic mice that express high levels of interleukin 1 alpha in basal epidermis.

Resting epidermal keratinocytes contain large amounts of interleukin 1 (IL-1), but the function of this cytokine in the skin remains unclear. To further define the role of IL-1 in cutaneous biology, we have generated two lines of transgenic mice (TgIL-1.1 and TgIL-1.2) which overexpress IL-1 alpha in basal keratinocytes. There was high-level tissue-specific expression of transgene mRNA and protein and large quantities of IL-1 alpha were liberated into the circulation from epidermis in both lines. TgIL-1.1 mice, which had the highest level of transgene expression, developed a spontaneous skin disease characterized by hair loss, scaling, and focal inflammatory skin lesions. Histologically, nonlesional skin of these animals was characterized by hyperkeratosis and a dermal mononuclear cell infiltrate of macrophage/monocyte lineage. Inflammatory lesions were marked by a mixed cellular infiltrate, acanthosis, and, in some cases, parakeratosis. These findings confirm the concept of IL-1 as a primary cytokine, release of which is able to initiate and localize an inflammatory reaction. Furthermore, these mice provide the first definitive evidence that inflammatory mediators can be released from the epidermis to enter the systemic circulation and thereby influence, in a paracrine or endocrine fashion, a wide variety of other cell types.     

Efficient denoising algorithms for large experimental datasets and their applications in Fourier transform ion cyclotron resonance mass spectrometry

Modern scientific research produces datasets of increasing size and complexity that require dedicated numerical methods to be processed. In many cases, the analysis of spectroscopic data involves the denoising of raw data before any further processing. Current efficient denoising algorithms require the singular value decomposition of a matrix with a size that scales up as the square of the data length, preventing their use on very large datasets. Taking advantage of recent progress on random projection and probabilistic algorithms, we developed a simple and efficient method for the denoising of very large datasets. Based on the QR decomposition of a matrix randomly sampled from the data, this approach allows a gain of nearly three orders of magnitude in processing time compared with classical singular value decomposition denoising. This procedure, called urQRd (uncoiled random QR denoising), strongly reduces the computer memory footprint and allows the denoising algorithm to be applied to virtually unlimited data size. The efficiency of these numerical tools is demonstrated on experimental data from high-resolution broadband Fourier transform ion cyclotron resonance mass spectrometry, which has applications in proteomics and metabolomics. We show that robust denoising is achieved in 2D spectra whose interpretation is severely impaired by scintillation noise. These denoising procedures can be adapted to many other data analysis domains where the size and/or the processing time are crucial.Big data are becoming predominant in modern science, and found in many scientific domains: astrophysics (1), large-scale physics experiments (2), biology (3, 4), or even natural text analysis (5). This introduces a new challenge for data handling and analysis, as it requires special processing approaches, which avoid accessing the whole data at once (6), and make use of adapted algorithms easily parallelized. Such constraints may be difficult to fulfill, even for simple procedures such as noise reduction.Every measurement is corrupted by unwanted noise, which is the combination of the effect of random fluctuations in the sample and the apparatus, but can also originate from external events like environmental noise. Denoising methods focus mainly on removing or reducing as much as possible an additive Gaussian wide sense stationary noise.For stationary signals the optimal linear denoising filter in the mean-square error sense is the Wiener filter. However, it suffers from the requirement of a precise estimate of the signal and noise auto- and cross-correlation functions. Many advanced denoising methods have been developed using linear algebra, which usually requires considerable processing power. One of the main alternative approaches relies on a multiresolution analysis, which sets noise apart from signal components more efficiently than classical orthogonal basis methods. In this respect, wavelets associated with soft thresholding have been considered for denoising purposes (7).Harmonic signals can be modeled as the sum of damped sinusoids. They are typically found in spectroscopies like NMR, Fourier transform mass spectrometry (FT-MS), FTIR but also in seismology, astrophysics, genetics, or financial analysis. They are easily analyzed by Fourier transformation if regularly sampled. For such specific signals, one class of denoising methods is based on modeling a sum of a fixed number of exponentials as devised by Prony (8). This was recently revisited and improved by Beylkin and Monzon (9, 10).There are also statistical methods related to the Karhunen–Loève transform which use adaptive basis instead of a priori basis. Relying on the autoregressive model (AR) (11, 12), a Hankel matrix is built and its rank is then reduced to the number of expected frequencies. Rank reduction by the singular value decomposition (SVD) (13) of this matrix is known as Cadzow’s method (14), also known as singular spectrum analysis (SSA) (15). The advantage is that no assumption about the noise or signal power is required and the number of frequencies is the only parameter.However, the benefits of denoising are balanced by several drawbacks. If the assumed number of frequencies is incorrect, the denoised signal is polluted with spurious artifacts that are indistinguishable from the real signal. Additionally, the SVD decomposition is slow and scales in operations. Alternative rapid SVD algorithms can be used, such as the Lanczos bidiagonalization method (16, 17), the truncated SVD (18), or random projections (19), as was recently applied in seismology (20). However, these algorithms do not solve the artifacts issue.Capitalizing on recent progress in algebra on random projection and probabilistic algorithms (21–24), we present here an efficient approach to denoising which can be easily applied to the large datasets found in Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometry experiments, and more generally, to any big data analysis. The main driving idea is to avoid explicit computation of data-derived quantities, but rather estimate the needed values, based on a partial sampling of the data. Extending from previous ideas (19), the denoising algorithm is based on a subsampling of the data-associated matrix. Here, rather than truncating the rank by removing some of the components of the SVD decomposition, we compute a randomized low-rank approximation of the Hankel matrix (24) that retains preferentially more signal than noise.We show that this leads to a substantial improvement of the processing in terms of speed, with little compromise on the quality, allowing gains of 2–3 orders of magnitude in processing time and in memory size. Applications of this approach are demonstrated on the large datasets obtained in FT-ICR mass spectrometry.     

Time perception in children treated for a cerebellar medulloblastoma

The aim of the present study was to investigate temporal abilities in children treated by surgery for a malignant tumor in the cerebellum. Children with a diagnosed medulloblastoma and age-paired control children were given a temporal discrimination task (bisection task) and a temporal reproduction task with two duration ranges, one shorter than 1 s and the other longer than 4 s. The motor and cognitive capacities of these children were also assessed by a battery of age-adapted neuropsychological tests. The results did not show any significant difference in performance between the children with or without cerebellar lesions in the temporal discrimination task. It was only in the temporal reproduction task that the children with cerebellar lesions reproduced longer and more variable durations than the other children, but only for the short stimulus durations (≤1 s). In addition, a hierarchical regression analysis revealed that the best predictor of variance in temporal performance was a significantly lower processing speed in children with cerebellar lesions in comparison to their controls. These results indicated that the major cause of deficits in temporal judgments in children with cerebellar lesions was due to their inability to reproduce accurately short temporal intervals in association with low processing speed, rather than to a specific deficit in the perception of time.     

Effects of social contexts on overweight and normal-weight children's food intake

This study investigates the effects of peer influence on the food intake of overweight and normal-weight children. A mixed factorial design was employed, with children's weight status (overweight vs. normal-weight) as a between-subjects factor, and social context (alone vs. group) as a within-subjects factor. A total of 32 children (n=17 overweight and n=15 normal-weight) between the ages of 6-10 years participated in this study. Findings from the random regression model indicated that overweight children ate more when with others than when alone, while in contrast normal-weight ate more with others than they did when alone. Therefore, social context differentially impacts the eating behavior of overweight and normal-weight children. This study underscores differences in responses to the social environment between overweight and non-overweight youths, and suggests that social involvement may be an important tool in treatment and prevention programs for overweight and obesity.     

Arterial manifestations of Beh?et's disease. 12 cases

Out of 196 patients with Beh?et's disease, 12 (10 men and 2 women, mean age 34 +/- 7 years) had non-coronary arterial lesions. Beh?et's disease was complete in 4 patients. The arterial lesions had appeared 8.6 +/- 8 years on average (20 years at most) after the first sign of the disease. Three patients showed evidence of stenosis or occlusion involving one or several arteries. Eight patients had both stenotic and aneurysmal lesions. One patient had an arteriovenous fistula. Another developed a false aneurysm at the site of introduction of a femoral catheter. Yet another patient developed an anastomotic aneurysm one year after implantation of an abdominal aortic graft. In 2 cases histology showed fragmentation of the media associated with vasculitis of the vasa vasorum. Two patients with pulmonary aneurysm died of massive haemoptysis. In 2 patients combined corticosteroid and cyclophosphamide therapy failed to prevent the development of aneurysmal lesions. Phlebitis was associated with arterial involvement in 7 patients. Comparison between patients with or without arterial lesions showed no significant difference in time of onset of Beh?et's disease, sex, main clinical features and presence of HLA B5. Aneurysmal lesions respond poorly to medical treatment, and surgery is mandatory. Since recurrence at the site of anastomosis is possible, prolonged monitoring is required.     

Fluorescence in situ hybridization mapping of translocations and deletions involving the short arm of human chromosome 12 in malignant hematologic diseases

Translocations and deletions of the short arm of chromosome 12 [t(12p) and del(12p)] are common recurring abnormalities in a broad spectrum of hematologic malignant diseases. We studied 20 patients and one cell line whose cells contained 12p13 translocations and/or 12p deletions using fluorescence in situ hybridization (FISH) with phage, plasmid, and cosmid probes that we previously mapped and ordered on 12p12-13. FISH analysis showed that the 12p13 translocation breakpoints were clustered between two cosmids, D12S133 and D12S142, in 11 of 12 patients and in one cell line. FISH analysis of 11 patients with deletions demonstrated that the deletions were interstitial rather than terminal and that the distal part of 12p12, including the GDI-D4 gene and D12S54 marker, was deleted in all 11 patients. Moreover, FISH analysis showed that cells from 3 of these patients contained both a del(12p) and a 12p13 translocation and that the affected regions of these rearrangements appeared to overlap. We identified three yeast artificial chromosome (YAC) clones that span all the 12p13 translocation breakpoints mapped between D12S133 and D12S142. They have inserts of human DNA between 1.39 and 1.67 Mb. Because the region between D12S133 and D12S142 also represents the telomeric border of the smallest commonly deleted region of 12p, we also studied patients with a del(12p) using these YACs. The smallest YAC, 964c10, was deleted in 8 of 9 patients studied. In the other patient, the YAC labeled the del(12p) chromosome more weakly than the normal chromosome 12, suggesting that a part of the YAC was deleted. Thus, most 12p13 translocation breakpoints were clustered within the sequences contained in the 1.39 Mb YAC and this YAC appears to include the telomeric border of the smallest commonly deleted region. Whether the same gene is involved in both the translocations and deletions is presently unknown.     

HPA-1a phenotype-genotype discrepancy reveals a naturally occurring Arg93Gln substitution in the platelet beta 3 integrin that disrupts the HPA-1a epitope

A single nucleotide polymorphism (SNP) at position 196 in the beta 3 integrin causes a Leu33Pro substitution in the mature protein. Alloimmunization against the beta 3Leu33 form (human platelet antigen [HPA]-1a, Pl(A1), Zw(a)) in patients who are beta 3Pro33 homozygous (HPA-1b1b, Pl(A2A2), Zw(bb)) causes neonatal alloimmune thrombocytopenia, posttransfusion purpura, or refractoriness to platelet transfusion. Studies with recombinant proteins have demonstrated that amino acids 1 to 66 and 288 to 490 of the beta 3 integrin contribute to HPA-1a epitope formation. In determining the HPA-1a status of more than 6000 donors, we identified a donor with an HPA-1a(weak) phenotype and an HPA-1a1b genotype. The platelets from this donor had normal levels of surface alpha IIb beta 3 but reacted only weakly with monoclonal and polyclonal anti-HPA-1a by whole blood enzyme-linked immunosorbent assay (ELISA), flow cytometry, and sandwich ELISA. We reasoned that an alteration in the primary nucleotide sequence of the beta 3Leu33 allele of this donor was disrupting the HPA-1a epitope. In agreement with this hypothesis, sequencing platelet RNA-derived alpha IIb and beta 3 cDNA identified a novel G/A SNP at position 376 of the beta 3 integrin that encodes for an Arg93Gln replacement in the beta 3Leu33 allele. Coexpression of the beta 3Leu33Gln93 encoding cDNA in Chinese hamster ovary cells with human alpha IIb cDNA showed that the surface-expressed alpha IIb beta 3 reacted normally with beta 3 integrin-specific monoclonal antibodies but only weakly with monoclonal anti-HPA-1a. Our results show that an Arg93Gln mutation in the beta 3Leu33 encoding allele disrupts the HPA-1a epitope, suggesting that Arg93 contributes to the formation of the HPA-1a B-cell epitope.