Genome-wide expression changes induced by bisphenol A, F and S in human stem cell derived hepatocyte-like cells.

The debate about possible adverse effects of bisphenol A (BPA) has been ongoing for decades. Bisphenol F (BPF) and S (BPS) have been suggested as "safer" alternatives. In the present study we used hepatocyte-like cells (HLCs) derived from the human embryonic stem cell lines Man12 and H9 to compare the three bisphenol derivatives. Stem cell-derived progenitors were produced using an established system and were exposed to BPA, BPF and BPS for 8 days during their transition to HLCs. Subsequently, we examined cell viability, inhibition of cytochrome P450 (CYP) activity, and genome-wide RNA profiles. Sub-cytotoxic, inhibitory concentrations (IC50) of CYP3A were 20, 9.5 and 25 µM for BPA, BPF and BPS in Man12 derived HLCs, respectively. The corresponding concentrations for H9-derived HLCs were 19, 29 and 31 µM. These IC50 concentrations were used to study global expression changes in this in vitro study and are higher than unconjugated BPA in serum of the general population. A large overlap of up- as well as downregulated genes induced by the three bisphenol derivatives was seen. This is at least 28-fold higher compared to randomly expected gene expression changes. Moreover, highly significant correlations of expression changes induced by the three bisphenol derivatives were obtained in pairwise comparisons. Dysregulated genes were associated with reduced metabolic function, cellular differentiation, embryonic development, cell survival and apoptosis. In conclusion, no major differences in cytochrome inhibitory activities of BPA, BPF and BPS were observed and gene expression changes showed a high degree of similarity.

Statistical analysis of MCC-IMS data for two group comparisons-an exemplary study on two devices.

The Multi-capillary-column-Ion-mobility-spectrometry (MCC-IMS) technology for measuring breath gas can be used for distinguishing between healthy and diseased subjects or between different types of diseases. The statistical methods for classifying the corresponding breath samples typically neglects potential confounding clinical and technical variables, reducing both accuracy and generalizability of the results. Especially measuring samples on different technical devices can heavily influence the results. We conducted a controlled breath gas study including 49 healthy volunteers to evaluate the effect of the variables sex, smoking habits and technical device. Every person was measured twice, once before and once after consuming a glass of orange juice. The two measurements were obtained on two different devices. The evaluation of the MCC-IMS data regarding metabolite detection was performed once using the software VisualNow, which requires manual interaction, and once using the fully automated algorithm SGLTR-DBSCAN. We present statistical solutions, peak alignment and scaling, to adjust for the different devices. For the other potential confounders sex and smoking, in our study no significant influence was identified.

Subtype-specific prognostic impact of different immune signatures in node-negative breast cancer.

BACKGROUND: The role of different subtypes of immune cells is still a matter of debate. METHODS: We compared the prognostic relevance for metastasis-free survival (MFS) of a B-cell signature (BS), a T-cell signature (TS), and an immune checkpoint signature (CPS) in node-negative breast cancer (BC) using mRNA expression. Microarray-based gene-expression data were analyzed in six previously published cohorts of node-negative breast cancer patients not treated with adjuvant therapy (n = 824). The prognostic relevance of the individual immune markers was assessed using univariate analysis. The amount of independent prognostic information provided by each immune signature was then compared using a likelihood ratio statistic in the whole cohort as well as in different molecular subtypes. RESULTS: Univariate Cox regression in the whole cohort revealed prognostic significance of CD4 (HR 0.66, CI 0.50-0.87, p = 0.004), CXCL13 (HR 0.86, CI 0.81-0.92, p < 0.001), CD20 (HR 0.76, CI 0.64-0.89, p = 0.001), IgκC (HR 0.81, CI 0.75-0.88, p < 0.001), and CTLA-4 (HR 0.67, CI 0.46-0.97, p = 0.032). Multivariate analyses of the immune signatures showed that both TS (p < 0.001) and BS (p < 0.001) showed a significant prognostic information in the whole cohort. After accounting for clinical-pathological variables, TS (p < 0.001), BS (p < 0.05), and CPS (p < 0.05) had an independent effect for MFS. In subgroup analyses, the prognostic effect of immune cells was most pronounced in HER2+ BC: BS as well as TS showed a strong association with MFS when included first in the model (p < 0.001). CONCLUSION: Immune signatures provide subtype-specific additional prognostic information over clinical-pathological variables in node-negative breast cancer.

Prognostic significance of interferon regulating factor 4 (IRF4) in node-negative breast cancer.

PURPOSE: The transcription factor IRF4 regulates immunoglobulin class switch recombination as well as plasma cell differentiation. We examined the prognostic significance of IRF4 expression in node-negative breast cancer (BC). METHODS: IRF4 expression was evaluated by immunostaining in a cohort of 197 node-negative BC patients not treated in adjuvant setting, referred to as Mainz cohort. The prognostic significance of immunohistochemically determined IRF4 expression for metastasis-free survival (MFS) was examined by Kaplan-Meier survival analysis as well as univariate and multivariate Cox analysis adjusted for age, pT stage, histological grade, ER, and HER2 status. For verification of immunohistochemical results, IRF4 mRNA expression was evaluated using microarray-based gene expression profiling in four previously published cohorts (Mainz, Rotterdam, Transbig, Yu) consisting of 824 node-negative breast cancer patients in total, who were not treated with adjuvant therapy. The prognostic significance of IRF4 mRNA expression on metastasis-free survival (MFS) was examined by univariate and multivariate Cox analysis in the Mainz cohort and by a meta-analysis of all node-negative BC patients and different molecular subtypes. IRF4 mRNA levels were compared to immunohistochemically determined IRF4 expression in 140 patients of the Mainz cohort using Spearman correlation. RESULTS: Immunohistochemically determined high IRF4 expression was associated with higher MFS in univariate Cox regression (HR 0.178, 95% CI 0.070-0.453, p < 0.001). IRF4 maintained its significance independently of established clinical factors for MFS (HR 0.088, 95% CI 0.033-0.232, p < 0.001). Immunohistochemically, determined IRF4 correlated moderately with IRF4 mRNA expression (ρ = 0.589). Higher expression of IRF4 was associated with better MFS in a meta-analysis of the total cohort (HR 0.438, 95% CI 0.307-0.623, p < 0.001). Prognostic significance was more pronounced in the HER2+ molecular subtype (HR 0.215, 95% CI 0.090-0.515, p = 0.001) as compared to the luminal A (HR 0.549, 95% CI 0.248-1.215, p = 0.139), luminal B (HR 0.444, 95% CI 0.215-0.916, p = 0.028), and basal-like subtypes (HR 0.487, 95% CI 0.269-0.883, p = 0.018). Further, IRF4 expression showed independent prognostic significance in a multivariate analysis of the Mainz cohort (HR 0.236, 95% CI 0.105-0.527, p < 0.001). CONCLUSIONS: IRF4 had independent prognostic significance in node-negative BC. Higher expression of IRF4 was associated with improved outcome. The prognostic impact differed between diverse molecular subtypes and was most pronounced in HER2+ breast cancer.

In vitro - in vivo correlation of gene expression alterations induced by liver carcinogens.

Although cultivated hepatocytes are widely used in the studies of drug metabolism, their application in toxicogenomics is considered as problematic, because previous studies have reported only little overlap between chemically induced gene expression alterations in liver in vivo and in cultivated hepatocytes. Here, we identified 22 genes that were altered in livers of rats after oral administration of the liver carcinogens aflatoxin B1 (AB1), 2-nitrofluorene (2-NF), methapyrilene (MP) or piperonyl-butoxide (PBO). The functions of the 22 genes have been classified into two groups. Genes related to stress response, DNA repair or metabolism and genes associated with cell proliferation, respectively. Next, rat hepatocyte sandwich cultures were exposed to AB1, 2-NF, MP or PBO for 24h and expression of the above mentioned genes was determined by RT-qPCR. Significant correlations between the degree of gene expression alterations in vivo and in vitro were obtained for the stress, DNA repair and metabolism associated genes at concentrations covering a range from cytotoxic concentrations to non-toxic/in vivo relevant concentrations. In contrast to the stress associated genes, no significant in vivo/in vitro correlation was obtained for the genes associated with cell proliferation. To understand the reason of this discrepancy, we compared replacement proliferation in vivo and in vitro. While hepatocytes in vivo, killed after administration of hepatotoxic compounds, are rapidly replaced by proliferating surviving cells, in vitro no replacement proliferation as evidenced by BrdU incorporation was observed after washing out hepatotoxic concentrations of MP. In conclusion, there is a good correlation between gene expression alterations induced by liver carcinogens in vivo and in cultivated hepatocytes. However, it should be considered that cultivated primary hepatocytes do not show replacement proliferation explaining the in vivo/in vitro discrepancy concerning proliferation associated genes.

Image analysis for cDNA microarrays.

OBJECTIVES: We characterize typical problems encountered in microarray image analysis and present algorithmic approaches dealing with background estimation, spot identification and intensity extraction. Validation of the quality of resulting measurements is discussed. METHODS: We describe sources for errors in microarray images and present algorithms that have been specifically developed to deal with such experimental imperfections. RESULTS: For the image analysis of hybridization experiments, discriminating spot regions from a background is the most critical step. Spot shape detection algorithms, intensity histogram methods and hybrid approaches have been proposed. The correctness of final intensity estimates is difficult to verify. Nevertheless, the application of sophisticated algorithms provides a significant reduction of the possible information loss. CONCLUSIONS: The initial analysis step for array hybridization experiments is the estimation of expression intensities. The quality of this process is crucial for the validity of interpretations from subsequent analysis steps.

Clustering algorithms and other exploratory methods for microarray data analysis.

OBJECTIVES: We introduce methods for the exploratory analysis of microarray data, especially focusing on cluster algorithms. Benefits and problems are discussed. METHODS: We describe application and suitability of unsupervised learning methods for the classification of gene expression data. Cluster algorithms are treated in more detail, including assessment of cluster quality. RESULTS: When dealing with microarray data, most cluster algorithms must be applied with caution. As long as the structure of the true generating models of such data is not fully understood, the use of simple algorithms seems to be more appropriate than the application of complex black-box algorithms. New methods explicitly targeted to the analysis of microarray data are increasingly being developed in order to increase the amount of useful information extracted from the experiments. CONCLUSIONS: Unsupervised methods can be a helpful tool for the analysis of microarray data, but a critical choice of the algorithm and a careful interpretation of the results are required in order to avoid false conclusions.

Predicting protein structure classes from function predictions.

MOTIVATION: We introduce a new approach to using the information contained in sequence-to-function prediction data in order to recognize protein template classes, a critical step in predicting protein structure. The data on which our method is based comprise probabilities of functional categories; for given query sequences these probabilities are obtained by a neural net that has previously been trained on a variety of functionally important features. On a training set of sequences we assess the relevance of individual functional categories for identifying a given structural family. Using a combination of the most relevant categories, the likelihood of a query sequence to belong to a specific family can be estimated. RESULTS: The performance of the method is evaluated using cross-validation. For a fixed structural family and for every sequence, a score is calculated that measures the evidence for family membership. Even for structural families of small size, family members receive significantly higher scores. For some examples, we show that the relevant functional features identified by this method are biologically meaningful. The proposed approach can be used to improve existing sequence-to-structure prediction methods. AVAILABILITY: Matlab code is available on request from the authors. The data are available at http://www.mpisb.mpg.de/~sommer/Fun2Struc/