Implication of DNA methylation and PAX5 factor in the transcriptional regulation of hTERT, the human telomerase reverse transcriptase gene

RESUME La télomérase est une enzyme dite "d'immortalité" qui permet aux cellules de maintenir la longueur de leurs télomères, ce qui confère une capacité de réplication illimitée aux cellules reproductrices et cancéreuses. A l'inverse, les cellules somatiques normales, qui n'expriment pas la télomérase, ont une capacité de réplication limitée. La sous-unité catalytique de la télomérase, hTERT, est définie comme le facteur limitant l'activité télomérasique. Entre activateurs et répresseurs, le rôle de la méthylation de l'ADN et de l'acétylation des histones, de nombreux modèles ont été suggérés. La découverte de l'implication de CTCF dans la régulation transcriptionnelle de hTERT explique en partie le mécanisme de répression de la télomérase dans la plupart des cellules somatiques et sa réactivation dans les cellules tumorales. Dans les cellules télomérase-positives, l'activité inhibitrice de CTCF est bloquée par un mécanisme dépendent ou non de la méthylation. Dans la plupart des carcinomes, une hyperméthylation de la région 5' de hTERT bloque l'effet inhibiteur de CTCF, alors qu'une petite région hypométhylée permet un faible niveau de transcription du gène. Nous avons démontré que la protéine MBD2 se lie spécifiquement sur la région 5' méthylée de hTERT dans différentes lignées cellulaires et qu'elle est impliquée dans la répression partielle de la transcription de hTERT dans les cellules tumorales méthylées. Par contre, nous avons montré que dans les lymphocytes B normaux et néoplasiques, la régulation de hTERT est indépendante de la méthylation. Dans ces cellules, le facteur PAX5 se lie sur la région 5' de hTERT en aval du site d'initiation de la traduction (ATG). L'expression exogène de PAX5 dans les cellules télomérase-négatives active la transcription de hTERT, alors que la répression de PAX5 dans les cellules lymphomateuses inhibe la transcription du gène. PAX5 est donc directement impliqué dans l'activation de l'expression de hTERT dans les lymphocytes B exprimant la télomérase. Ces résultats révèlent des différences entre les niveaux de méthylation de hTERT dans les cellules de carcinomes et les lymphocytes B exprimant la télomérase. La méthylation de hTERT en tant que biomarqueur de cancer a été évaluée, puis appliquée à la détection de métastases. Nous avons ainsi montré que la méthylation de hTERT est positivement corrélée au diagnostic cytologique dans les liquides céphalorachidiens. Nos résultats conduisent à un modèle de régulation de hTERT, qui aide à comprendre comment la transcription de ce gène est régulée par CTCF, avec un mécanisme lié ou non à la méthylation du gène hTERT. La méthylation de hTERT s'est aussi révélée être un nouveau et prometteur biomarqueur de cancer. SUMMARY Human telomerase is an "immortalizing" enzyme that enables cells to maintain telomere length, allowing unlimited replicative capacity to reproductive and cancer cells. Conversely, normal somatic cells that do not express telomerase have a finite replicative capacity. The catalytic subunit of telomerase, hTERT, is defined as the limiting factor for telomerase activity. Between activators and repressors, and the role of DNA methylation and histone acetylation, an abundance of hTERT regulatory models have been suggested. The discovery of the implication of CTCF in the transcriptional regulation of hTERT in part explained the mechanism of silencing of telomerase in most somatic cells and its reactivation in neoplastic cells. In telomerase-positive cells, the inhibitory activity of CTCF is blocked by methylation-dependent and -independent mechanisms. In most carcinoma cells, hypermethylation of the hTERT 5' region has been shown to block the inhibitory effect of CTCF, while a short hypomethylated region allows a low transcription level of the gene. We have demonstrated that MBD2 protein specifically binds the methylated 5' region of hTERT in different cell lines and is therefore involved in the partial repression of hTERT transcription in methylated tumor cells. In contrast, we have shown that in normal and neoplastic B cells, hTERT regulation is methylation-independent. The PAX5 factor has been shown to bind to the hTERT 5'region downstream of the ATG translational start site. Ectopic expression of PAX5 in telomerase-negative cells or repression of PAX5 expression in B lymphoma cells respectively activated and repressed hTERT transcription. Thus, PAX5 is strongly implicated in hTERT expression activation in telomerase-positive B cells. These results reveal differences between the hTERT methylation patterns in telomerase-positive carcinoma cells and telomerase-positive normal B cells. The potential of hTERT methylation as a cancer biomarker was evaluated and applied to the detection of metastasis. We have shown that hTERT methylation correlates with the cytological diagnosis in cerebrospinal fluids. Our results suggest a model of hTERT gene regulation, which helps us to better understand how hTERT transcription is regulated by CTCF in methylation-dependant and independent mechanisms. Our data also indicate that hTERT methylation is a promising new cancer biomarker.

Implication of DNA methylation, CTCF and BORIS factors in the transcriptional regulation of the human telomerase catalytic subunit hTERT

RESUME La télomérase confère une durée de vie illimitée et est réactivée dans la plupart des cellules tumorales. Sa sous-unité catalytique hTERT est définie comme le facteur limitant pour son activation. De l'identification de facteurs liant la région régulatrice d'hTERT, au rôle de la méthylation de l'ADN et de la modification des histones, de nombreux modèles de régulation ont été suggérés. Cependant, aucun de ces modèles n'a pu expliquer l'inactivation de la télomérase dans la plupart des cellules somatiques et sa réactivation dans la majorité des cellules tumorales. De plus, les observations contradictoires entre le faible niveau d'expression d'ARN messager d'hTERT dans les cellules télomérase-positives et la très forte activité transcriptionnelle du promoteur d'hTERT en transfection restent incomprises. Dans cette étude, nous avons montré que la région proximale du gène hTERT (exon 1 et 2) était impliquée dans la répression de l'activité de son promoteur. Nous avons identifié le facteur CTCF comme étant un inhibiteur du promoteur d'hTERT, en se liant au niveau de son premier exon. La méthylation de l'exon 1 du gène hTERT, couramment observée dans les tumeurs mais pas dans les cellules normales, empêcherait la liaison de CTCF. L'étude du profil de méthylation du promoteur d'hTERT indique qu'une partie du promoteur reste déméthylée et qu'elle semble suffisante pour permettre une faible activité transcriptionnelle du gène hTERT. Ainsi, la méthylation particulière des régions régulatrices d'hTERT inhibe la liaison de CTCF tout en permettant une faible transcription du gène. Cependant, dans certaines cellules tumorales, le promoteur et la région proximale du gène hTERT ne sont pas méthylés. Dans les lignées cellulaires tumorales de tesitcules et d'ovaires, l'inhibition de CTCF est contrée par son paralogue BORIS, qui se lie aussi au niveau de l'exon 1 d'hTERT, mais permet ainsi l'activation du promoteur. L'étude de l'expression du gène BORIS montre qu'il est exclusivement exprimé dans les tissus normaux de testicules et d'ovaires jeunes, ainsi qu'à différents niveaux dans la plupart des tumeurs. Sa transcription est sous le contrôle de deux promoteurs. Le promoteur proximal est régulé par méthylation et un transcrit alternatif majoritaire, délété de l'exon 6, est trouvé lorsque ce promoteur est actif. Tous ces résultats conduisent à un modèle de régulation du gène hTERT qui tient compte du profil épigénétique du gène et qui permet d'expliquer le faible taux de transcription observé in vivo. De plus, l'expression de BORIS dans les cancers et son implication dans l'activation du gène hTERT pourrait permettre de comprendre les phénomènes de dérégulation épigénétique et d'immortalisation qui ont lieu durant la tumorigenèse. SUMMARY Telomerase confers an unlimited lifespan, and is reactivated in most tumor cells. The catalytic subunit of telomerase, hTERT, is defined as the limiting factor for telomerase activity. Between activators and repressors that bind to the hTERT 5' regulatory region, and the role of CpG methylation and histone acetylation, an abundance of regulatory models have been suggested. None of these models can explain the silence of telomerase in most somatic cells and its reactivation in tumor cells. Moreover, the contradictory observations of the low level of hTERT mRNA in telomerase-positive cells and the high transcriptional activity of the hTERT promoter in transfection experiments remain unresolved. In this study, we demonstrated that the proximal exonic region of the hTERT gene (exon 1 and 2) is involved in the inhibition of its promoter. We identified the protein CTCF as the inhibitor of the hTERT promoter, through its binding to the first exon. The methylation of the first exon region, which is often observed in cancer cells but not in noimal cells, represses CTCF binding. Study of hTERT promoter methylation shows a partial demethylation sufficient to activate the transcription of the hTERT gene. Therefore, we demonstrated that the particular methylation profile of the hTERT regulatory sequences inhibits the binding of CTCF, while it allows a low transcription of the gene. Nevertheless, in some tumor cells, the promoter and the proximal exonic region of hTERT are unmethylated. In testicular and ovarian cancer cell lines, CTCF inhibition is counteracted by its BORIS paralogue that also binds the hTERT first exon but allows the promoter activation. The study of BORIS gene regulation showed that this factor is exclusively expressed in normal tissue of testis and ovary of young woman, as well as in almost all tumors with different levels. Two promoters were found to induce its transcription. The proximal promoter was regulated by methylation. Moreover, a major alternative transcript, deleted of the exon 6, is detected when this promoter is active. All these results lead to a model for hTERT regulation that takes into account the epigenetic profile of the gene and provides an explanation for the low transcriptional level observed in vivo. BORIS expression in cancers and its implication in hTERT activation might also permit the understanding of epigenetic deregulation and immortalization phenomena that occur during tumorigenesis.

Additive functions in boolean models of gene regulatory network modules.

Gene-on-gene regulations are key components of every living organism. Dynamical abstract models of genetic regulatory networks help explain the genome's evolvability and robustness. These properties can be attributed to the structural topology of the graph formed by genes, as vertices, and regulatory interactions, as edges. Moreover, the actual gene interaction of each gene is believed to play a key role in the stability of the structure. With advances in biology, some effort was deployed to develop update functions in Boolean models that include recent knowledge. We combine real-life gene interaction networks with novel update functions in a Boolean model. We use two sub-networks of biological organisms, the yeast cell-cycle and the mouse embryonic stem cell, as topological support for our system. On these structures, we substitute the original random update functions by a novel threshold-based dynamic function in which the promoting and repressing effect of each interaction is considered. We use a third real-life regulatory network, along with its inferred Boolean update functions to validate the proposed update function. Results of this validation hint to increased biological plausibility of the threshold-based function. To investigate the dynamical behavior of this new model, we visualized the phase transition between order and chaos into the critical regime using Derrida plots. We complement the qualitative nature of Derrida plots with an alternative measure, the criticality distance, that also allows to discriminate between regimes in a quantitative way. Simulation on both real-life genetic regulatory networks show that there exists a set of parameters that allows the systems to operate in the critical region. This new model includes experimentally derived biological information and recent discoveries, which makes it potentially useful to guide experimental research. The update function confers additional realism to the model, while reducing the complexity and solution space, thus making it easier to investigate.

Stochastic models and numerical algorithms for a class of regulatory gene networks.

Regulatory gene networks contain generic modules, like those involving feedback loops, which are essential for the regulation of many biological functions (Guido et al. in Nature 439:856-860, 2006). We consider a class of self-regulated genes which are the building blocks of many regulatory gene networks, and study the steady-state distribution of the associated Gillespie algorithm by providing efficient numerical algorithms. We also study a regulatory gene network of interest in gene therapy, using mean-field models with time delays. Convergence of the related time-nonhomogeneous Markov chain is established for a class of linear catalytic networks with feedback loops.

Binge eating in binge eating disorder: a break-down of emotion regulatory process?

Current explanatory models for binge eating in binge eating disorder (BED) mostly rely onmodels for bulimianervosa (BN), although research indicates different antecedents for binge eating in BED. This studyinvestigates antecedents and maintaining factors in terms of positive mood, negative mood and tension in asample of 22 women with BED using ecological momentary assessment over a 1-week. Values for negativemood were higher and those for positive mood lower during binge days compared with non-binge days.During binge days, negative mood and tension both strongly and significantly increased and positive moodstrongly and significantly decreased at the first binge episode, followed by a slight though significant, andlonger lasting decrease (negative mood, tension) or increase (positive mood) during a 4-h observation periodfollowing binge eating. Binge eating in BED seems to be triggered by an immediate breakdown of emotionregulation. There are no indications of an accumulation of negative mood triggering binge eating followed byimmediate reinforcing mechanisms in terms of substantial and stable improvement of mood as observed inBN. These differences implicate a further specification of etiological models and could serve as a basis fordeveloping new treatment approaches for BED.

Dynamical modeling and analysis of large cellular regulatory networks.

The dynamical analysis of large biological regulatory networks requires the development of scalable methods for mathematical modeling. Following the approach initially introduced by Thomas, we formalize the interactions between the components of a network in terms of discrete variables, functions, and parameters. Model simulations result in directed graphs, called state transition graphs. We are particularly interested in reachability properties and asymptotic behaviors, which correspond to terminal strongly connected components (or "attractors") in the state transition graph. A well-known problem is the exponential increase of the size of state transition graphs with the number of network components, in particular when using the biologically realistic asynchronous updating assumption. To address this problem, we have developed several complementary methods enabling the analysis of the behavior of large and complex logical models: (i) the definition of transition priority classes to simplify the dynamics; (ii) a model reduction method preserving essential dynamical properties, (iii) a novel algorithm to compact state transition graphs and directly generate compressed representations, emphasizing relevant transient and asymptotic dynamical properties. The power of an approach combining these different methods is demonstrated by applying them to a recent multilevel logical model for the network controlling CD4+ T helper cell response to antigen presentation and to a dozen cytokines. This model accounts for the differentiation of canonical Th1 and Th2 lymphocytes, as well as of inflammatory Th17 and regulatory T cells, along with many hybrid subtypes. All these methods have been implemented into the software GINsim, which enables the definition, the analysis, and the simulation of logical regulatory graphs.

Hard-wired heterogeneity in blood stem cells revealed using a dynamic regulatory network model.

MOTIVATION: Combinatorial interactions of transcription factors with cis-regulatory elements control the dynamic progression through successive cellular states and thus underpin all metazoan development. The construction of network models of cis-regulatory elements, therefore, has the potential to generate fundamental insights into cellular fate and differentiation. Haematopoiesis has long served as a model system to study mammalian differentiation, yet modelling based on experimentally informed cis-regulatory interactions has so far been restricted to pairs of interacting factors. Here, we have generated a Boolean network model based on detailed cis-regulatory functional data connecting 11 haematopoietic stem/progenitor cell (HSPC) regulator genes. RESULTS: Despite its apparent simplicity, the model exhibits surprisingly complex behaviour that we charted using strongly connected components and shortest-path analysis in its Boolean state space. This analysis of our model predicts that HSPCs display heterogeneous expression patterns and possess many intermediate states that can act as 'stepping stones' for the HSPC to achieve a final differentiated state. Importantly, an external perturbation or 'trigger' is required to exit the stem cell state, with distinct triggers characterizing maturation into the various different lineages. By focusing on intermediate states occurring during erythrocyte differentiation, from our model we predicted a novel negative regulation of Fli1 by Gata1, which we confirmed experimentally thus validating our model. In conclusion, we demonstrate that an advanced mammalian regulatory network model based on experimentally validated cis-regulatory interactions has allowed us to make novel, experimentally testable hypotheses about transcriptional mechanisms that control differentiation of mammalian stem cells. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Synchronous versus asynchronous modeling of gene regulatory networks.

MOTIVATION: In silico modeling of gene regulatory networks has gained some momentum recently due to increased interest in analyzing the dynamics of biological systems. This has been further facilitated by the increasing availability of experimental data on gene-gene, protein-protein and gene-protein interactions. The two dynamical properties that are often experimentally testable are perturbations and stable steady states. Although a lot of work has been done on the identification of steady states, not much work has been reported on in silico modeling of cellular differentiation processes. RESULTS: In this manuscript, we provide algorithms based on reduced ordered binary decision diagrams (ROBDDs) for Boolean modeling of gene regulatory networks. Algorithms for synchronous and asynchronous transition models have been proposed and their corresponding computational properties have been analyzed. These algorithms allow users to compute cyclic attractors of large networks that are currently not feasible using existing software. Hereby we provide a framework to analyze the effect of multiple gene perturbation protocols, and their effect on cell differentiation processes. These algorithms were validated on the T-helper model showing the correct steady state identification and Th1-Th2 cellular differentiation process. AVAILABILITY: The software binaries for Windows and Linux platforms can be downloaded from http://si2.epfl.ch/~garg/genysis.html.

The immunological synapse and actin assembly: a regulatory role for PKC theta.

Engagement of the T cell receptor leads to the accumulation of filamentous actin, which is necessary for the formation of the immunological synapse and subsequent T cell activation. In the December issue of Molecular Cell, Sasahara et al. provide new insights into the link between the T cell receptor and actin assembly in the immunological synapse, and reveal a critical regulatory role for PKC theta in this process.

Dynamic simulation of regulatory networks using SQUAD.

BACKGROUND: The ambition of most molecular biologists is the understanding of the intricate network of molecular interactions that control biological systems. As scientists uncover the components and the connectivity of these networks, it becomes possible to study their dynamical behavior as a whole and discover what is the specific role of each of their components. Since the behavior of a network is by no means intuitive, it becomes necessary to use computational models to understand its behavior and to be able to make predictions about it. Unfortunately, most current computational models describe small networks due to the scarcity of kinetic data available. To overcome this problem, we previously published a methodology to convert a signaling network into a dynamical system, even in the total absence of kinetic information. In this paper we present a software implementation of such methodology. RESULTS: We developed SQUAD, a software for the dynamic simulation of signaling networks using the standardized qualitative dynamical systems approach. SQUAD converts the network into a discrete dynamical system, and it uses a binary decision diagram algorithm to identify all the steady states of the system. Then, the software creates a continuous dynamical system and localizes its steady states which are located near the steady states of the discrete system. The software permits to make simulations on the continuous system, allowing for the modification of several parameters. Importantly, SQUAD includes a framework for perturbing networks in a manner similar to what is performed in experimental laboratory protocols, for example by activating receptors or knocking out molecular components. Using this software we have been able to successfully reproduce the behavior of the regulatory network implicated in T-helper cell differentiation. CONCLUSION: The simulation of regulatory networks aims at predicting the behavior of a whole system when subject to stimuli, such as drugs, or determine the role of specific components within the network. The predictions can then be used to interpret and/or drive laboratory experiments. SQUAD provides a user-friendly graphical interface, accessible to both computational and experimental biologists for the fast qualitative simulation of large regulatory networks for which kinetic data is not necessarily available.

Strength and limitations of regulatory T Cells for immunotherapy in transplantation.

In many experimental models, CD4+CD25+Foxp3+ regulatory T cells (nTreg) have been identifi ed as key players in promoting peripheral transplantation (Tx) tolerance. We have been focusing on therapies based on antigen-specifi c nTreg that can control effector T cells (Teff) and prevent allograft rejection. The use of nTreg in immunotherapeutic protocols for solid organ Tx is however limited by their overall low numbers as well as the low precursor frequency of alloantigen cross-reactive nTreg expected to be found in a normal individual. Moreover, although we previously described robust protocols to generate and expand antigen-specifi c nTreg in vitro, the process requires careful selection of highly pure nTreg and cumbersome ex-vivo manipulations, rendering this strategy not easily applicable in clinical solid organ Tx. In this study, we aimed to expand Treg directly in vivo and determine their suppressive function, effi cacy and stability in promoting donor-specifi c tolerance in a stringent murine Tx model. Our data suggest that IL-2-based therapies lead to a signifi cant increase of Treg in vivo. The expanded Treg suppressed Teff proliferation (albeit slightly less effi ciently than nTreg isolated from control mice) and allowed prolonged graft survival of major MHC-mismatched skin grafts in wild-type non-lymphopenic recipients. The expanded Treg alone were however not suffi cient to induce tolerance in stringent experimental conditions. Rapamycin reduced the frequency of Teff but did not impede expansion of Treg. Pro-infl ammatory stimuli hindered the expansion of Treg and resulted in an increase in the frequency of CD4+IFN-γ+ and CD4+IL17+ T cells. We propose that IL-2-based treatments would be an effi cient method for expanding functional Treg in vivo without affecting other immune cell populations, thereby favorably shifting the pool of alloreactive T cells towards regulation in response to an allograft. However, we also highlight some potential limitations of Treg expansion such as concomitant infl ammatory events.

Modeling stochasticity and robustness in gene regulatory networks.

MOTIVATION: Understanding gene regulation in biological processes and modeling the robustness of underlying regulatory networks is an important problem that is currently being addressed by computational systems biologists. Lately, there has been a renewed interest in Boolean modeling techniques for gene regulatory networks (GRNs). However, due to their deterministic nature, it is often difficult to identify whether these modeling approaches are robust to the addition of stochastic noise that is widespread in gene regulatory processes. Stochasticity in Boolean models of GRNs has been addressed relatively sparingly in the past, mainly by flipping the expression of genes between different expression levels with a predefined probability. This stochasticity in nodes (SIN) model leads to over representation of noise in GRNs and hence non-correspondence with biological observations. RESULTS: In this article, we introduce the stochasticity in functions (SIF) model for simulating stochasticity in Boolean models of GRNs. By providing biological motivation behind the use of the SIF model and applying it to the T-helper and T-cell activation networks, we show that the SIF model provides more biologically robust results than the existing SIN model of stochasticity in GRNs. AVAILABILITY: Algorithms are made available under our Boolean modeling toolbox, GenYsis. The software binaries can be downloaded from http://si2.epfl.ch/ approximately garg/genysis.html.

Characterization and binding affinities of SmLANP: a new Schistosoma mansoni member of the ANP32 family of regulatory proteins.

Members of the leucine-rich repeat protein family are involved in diverse functions including protein phosphatase 2-inhibition, cell cycle regulation, gene regulation and signalling pathways. A novel Schistosoma mansoni gene, called SmLANP, presenting homology to various genes coding for proteins that belong to the super family of leucine-rich repeat proteins, was characterized here. SmLANP was 1184bp in length as determined from cDNA and genomic sequences and encoded a 296 amino acid open reading frame that spanning from 6 to 894bp. The predicted amino acid sequence had a calculated molecular weight of 32kDa. Analysis of the predicted sequence indicated the presence of 3 leucine-rich domains (LRR) located in the N-terminal region and an aspartic acid rich region in the C-terminal end. SmLANP transcript is expressed in all stages of the S. mansoni life cycle analyzed, exhibiting the highest expression level in males. The SmLANP protein was expressed in a GST expression system and antibodies raised in mice against the recombinant protein. By immunolocalization assay, using adult worms, it was shown that the protein is mainly present in the cell nucleus through the whole body and strongly expressed along the tegument cell body nuclei of adult worms. As members of this family are usually involved in protein-protein interaction, a yeast two hybrid assay was conducted to identify putative binding partners for SmLANP. Thirty-six possible partners were identified, and a protein ATP synthase subunit alpha was confirmed by pull down assays, as a binding partner of the SmLANP protein.

Efficient generation of human alloantigen-specific CD4+ regulatory T cells from naive precursors by CD40-activated B cells.

CD4(+)CD25(+)Foxp3(+) regulatory T cells (Treg) play an important role in the induction and maintenance of immune tolerance. Although adoptive transfer of bulk populations of Treg can prevent or treat T cell-mediated inflammatory diseases and transplant allograft rejection in animal models, optimal Treg immunotherapy in humans would ideally use antigen-specific rather than polyclonal Treg for greater specificity of regulation and avoidance of general suppression. However, no robust approaches have been reported for the generation of human antigen-specific Treg at a practical scale for clinical use. Here, we report a simple and cost-effective novel method to rapidly induce and expand large numbers of functional human alloantigen-specific Treg from antigenically naive precursors in vitro using allogeneic nontransformed B cells as stimulators. By this approach naive CD4(+)CD25(-) T cells could be expanded 8-fold into alloantigen-specific Treg after 3 weeks of culture without any exogenous cytokines. The induced alloantigen-specific Treg were CD45RO(+)CCR7(-) memory cells, and had a CD4(high), CD25(+), Foxp3(+), and CD62L (L-selectin)(+) phenotype. Although these CD4(high)CD25(+)Foxp3(+) alloantigen-specific Treg had no cytotoxic capacity, their suppressive function was cell-cell contact dependent and partially relied on cytotoxic T lymphocyte antigen-4 expression. This approach may accelerate the clinical application of Treg-based immunotherapy in transplantation and autoimmune diseases.

The MicroArray Quality Control (MAQC)-II study of common practices for the development and validation of microarray-based predictive models.

Gene expression data from microarrays are being applied to predict preclinical and clinical endpoints, but the reliability of these predictions has not been established. In the MAQC-II project, 36 independent teams analyzed six microarray data sets to generate predictive models for classifying a sample with respect to one of 13 endpoints indicative of lung or liver toxicity in rodents, or of breast cancer, multiple myeloma or neuroblastoma in humans. In total, >30,000 models were built using many combinations of analytical methods. The teams generated predictive models without knowing the biological meaning of some of the endpoints and, to mimic clinical reality, tested the models on data that had not been used for training. We found that model performance depended largely on the endpoint and team proficiency and that different approaches generated models of similar performance. The conclusions and recommendations from MAQC-II should be useful for regulatory agencies, study committees and independent investigators that evaluate methods for global gene expression analysis.