Network-based analysis of transcriptome dynamics during activation in two human T-cell subpopulations identifies key regulators, and reveals that PLAU plays a critical role in both human and murine regulatory T-cell function.
We construct a Treg-specific correlation network from a high time-resolution transcriptome of human Tregs versus CD4+ T effector cells measured during their very early activation process.We propose a queen bee-surrounding principle to predict key candidate genes from the simplified undirected correlation network rather than an advanced directed transcription regulatory network. These potential key genes would have not been easily identified by a differential expression analysis.We show that the plasminogen activator urokinase (PLAU) is critical for suppressor function of both human and murine Tregs.We further demonstrate that PLAU is particularly important for memory Tregs and that PLAU mediates Treg suppressor function via STAT5 and ERK signaling pathways.
Human FOXP3+CD25+CD4+ regulatory T cells (Tregs) are essential to the maintenance of immune homeostasis. Several genes are known to be important for murine Tregs, but for human Tregs the genes and underlying molecular networks controlling the suppressor function still largely remain unclear. Here, we describe a strategy to identify the key genes directly from an undirected correlation network which we reconstruct from a very high time-resolution (HTR) transcriptome during the activation of human Tregs/CD4+ T-effector cells. We show that a predicted top-ranked new key gene PLAU (the plasminogen activator urokinase) is important for the suppressor function of both human and murine Tregs. Further analysis unveils that PLAU is particularly important for memory Tregs and that PLAU mediates Treg suppressor function via STAT5 and ERK signaling pathways. Our study demonstrates the potential for identifying novel key genes for complex dynamic biological processes using a network strategy based on HTR data, and reveals a critical role for PLAU in Treg suppressor function.
Network-based analysis of transcriptome dynamics during activation in two human T-cell subpopulations identifies key regulators, and reveals that PLAU plays a critical role in both human and murine regulatory T-cell function.
high time-resolution time series; human CD4 regulatory T cell; infer key genes from undirected gene networks; Plau knockout mice; Treg development and suppressor function
The need for the timely collection of diagnostic biosamples during symptomatic episodes represents a major obstacle to large-scale studies on acute respiratory infection (ARI) epidemiology. This may be circumvented by having the participants collect their own nasal swabs. We compared self- and staff-collected swabs in terms of swabbing quality and detection of viral respiratory pathogens.
We conducted a prospective study among employees of our institution during the ARI season 2010/2011 (December-March). Weekly emails were sent to the participants (n = 84), reminding them to come to the study center in case of new symptoms. The participants self-collected an anterior nasal swab from one nostril, and trained study personnel collected one from the other nostril. The participants self-collected another two swabs (one from each nostril) on a subsequent day. Human β-actin DNA concentration was determined in the swabs as a quality control. Viral respiratory pathogens were detected by multiplex RT-PCR (Seeplex RV15 kit, Seegene, Eschborn, Germany). Of 84 participants, 56 (67%) reported at least one ARI episode, 18 participants two, and one participant three. Self-swabbing was highly accepted by the participants. The amount of β-actin DNA per swab was higher in the self- than in the staff-collected swabs (p = 0.008). β-actin concentration was lower in the self-swabs collected on day 1 than in those collected on a subsequent day (p<0.0001). A respiratory viral pathogen was detected in 31% (23/75) of staff- and in 35% (26/75) of self-collected swabs (p = 0.36). With both approaches, the most frequently identified pathogens were human rhinoviruses A/B/C (12/75 swabs, 16%) and human coronavirus OC43 (4/75 swabs, 5%). There was almost perfect agreement between self- and staff-collected swabs in terms of pathogen detection (agreement = 93%, kappa = 0.85, p<0.0001).
Nasal self-swabbing for identification of viral ARI pathogens proved to be equivalent to staff-swabbing in this population in terms of acceptance and pathogen detection.
The mouse represents an important model system to study the host response to influenza A infections and to evaluate new prevention or treatment strategies. We and others reported that the susceptibility to influenza A virus infections strongly varies among different inbred mouse strains. In particular, DBA/2J mice are highly susceptible to several influenza A subtypes, including human isolates and exhibit severe symptoms after infection with clinical isolates.
Upon intra-muscular immunization with live H1N1 influenza A virus (mouse-adapted PR8M, and 2009 pandemic human HA04), DBA/2J mice mounted virus-specific IgG responses and were protected against a subsequent lethal challenge. The immune response and rescue from death after immunization in DBA/2J was similar to those observed for C57BL/6J mice.
DBA/2J mice represent a suitable mouse model to evaluate virulence and pathogenicity as well as immunization regimes against existing and newly emerging human influenza strains without the need for prior adaptation of the virus to the mouse.
Influenza A virus; Mouse; DBA/2J; Immunization
There is strong but mostly circumstantial evidence that genetic factors modulate the severity of influenza infection in humans. Using genetically diverse but fully inbred strains of mice it has been shown that host sequence variants have a strong influence on the severity of influenza A disease progression. In particular, C57BL/6J, the most widely used mouse strain in biomedical research, is comparatively resistant. In contrast, DBA/2J is highly susceptible.
To map regions of the genome responsible for differences in influenza susceptibility, we infected a family of 53 BXD-type lines derived from a cross between C57BL/6J and DBA/2J strains with influenza A virus (PR8, H1N1). We monitored body weight, survival, and mean time to death for 13 days after infection. Qivr5 (quantitative trait for influenza virus resistance on chromosome 5) was the largest and most significant QTL for weight loss. The effect of Qivr5 was detectable on day 2 post infection, but was most pronounced on days 5 and 6. Survival rate mapped to Qivr5, but additionally revealed a second significant locus on chromosome 19 (Qivr19). Analysis of mean time to death affirmed both Qivr5 and Qivr19. In addition, we observed several regions of the genome with suggestive linkage. There are potentially complex combinatorial interactions of the parental alleles among loci. Analysis of multiple gene expression data sets and sequence variants in these strains highlights about 30 strong candidate genes across all loci that may control influenza A susceptibility and resistance.
We have mapped influenza susceptibility loci to chromosomes 2, 5, 16, 17, and 19. Body weight and survival loci have a time-dependent profile that presumably reflects the temporal dynamic of the response to infection. We highlight candidate genes in the respective intervals and review their possible biological function during infection.
An infection represents a highly dynamic process involving complex biological responses of the host at many levels. To describe such processes at a global level, we recorded gene expression changes in mouse lungs after a non-lethal infection with influenza A virus over a period of 60 days. Global analysis of the large data set identified distinct phases of the host response. The increase in interferon genes and up-regulation of a defined NK-specific gene set revealed the initiation of the early innate immune response phase. Subsequently, infiltration and activation of T and B cells could be observed by an augmentation of T and B cell specific signature gene expression. The changes in B cell gene expression and preceding chemokine subsets were associated with the formation of bronchus-associated lymphoid tissue. In addition, we compared the gene expression profiles from wild type mice with Rag2 mutant mice. This analysis readily demonstrated that the deficiency in the T and B cell responses in Rag2 mutants could be detected by changes in the global gene expression patterns of the whole lung. In conclusion, our comprehensive gene expression study describes for the first time the entire host response and its kinetics to an acute influenza A infection at the transcriptome level.
Regulatory T cells (Tregs) play an essential role in the control of the immune response. Treg cells represent important targets for therapeutic interventions of the immune system. Therefore, it will be very important to understand in more detail which genes are specifically activated in Treg cells versus T helper (Th) cells, and which gene regulatory circuits may be involved in specifying and maintaining Treg cell homeostasis.
We isolated Treg and Th cells from a genetically diverse family of 31 BXD type recombinant inbred strains and the fully inbred parental strains of this family--C57BL/6J and DBA/2J. Subsequently genome-wide gene expression studies were performed from the isolated Treg and Th cells. A comparative analysis of the transcriptomes of these cell populations allowed us to identify many novel differentially expressed genes. Analysis of cis- and trans-expression Quantitative Trait Loci (eQTLs) highlighted common and unique regulatory mechanisms that are active in the two cell types. Trans-eQTL regions were found for the Treg functional genes Nrp1, Stat3 and Ikzf4. Analyses of the respective QTL intervals suggested several candidate genes that may be involved in regulating these genes in Treg cells. Similarly, possible candidate genes were found which may regulate the expression of F2rl1, Ctla4, Klrb1f. In addition, we identified a focused group of candidate genes that may be important for the maintenance of self-tolerance and the prevention of allergy.
Variation of expression across the strains allowed us to find many novel gene-interaction networks in both T cell subsets. In addition, these two data sets enabled us to identify many differentially expressed genes and to nominate candidate genes that may have important functions for the maintenance of self-tolerance and the prevention of allergy.
The immune response to viral infection is a temporal process, represented by a dynamic and complex network of gene and protein interactions. Here, we present a reverse engineering strategy aimed at capturing the temporal evolution of the underlying Gene Regulatory Networks (GRN). The proposed approach will be an enabling step towards comprehending the dynamic behavior of gene regulation circuitry and mapping the network structure transitions in response to pathogen stimuli.
We applied the Time Varying Dynamic Bayesian Network (TV-DBN) method for reconstructing the gene regulatory interactions based on time series gene expression data for the mouse C57BL/6J inbred strain after infection with influenza A H1N1 (PR8) virus. Initially, 3500 differentially expressed genes were clustered with the use of k-means algorithm. Next, the successive in time GRNs were built over the expression profiles of cluster centroids. Finally, the identified GRNs were examined with several topological metrics and available protein-protein and protein-DNA interaction data, transcription factor and KEGG pathway data.
Our results elucidate the potential of TV-DBN approach in providing valuable insights into the temporal rewiring of the lung transcriptome in response to H1N1 virus.
Gene Regulatory Network; Time Varying Dynamic Bayesian Network; Immune System; Influenza A
During a meeting of the SYSGENET working group ‘Bioinformatics’, currently available software tools and databases for systems genetics in mice were reviewed and the needs for future developments discussed. The group evaluated interoperability and performed initial feasibility studies. To aid future compatibility of software and exchange of already developed software modules, a strong recommendation was made by the group to integrate HAPPY and R/qtl analysis toolboxes, GeneNetwork and XGAP database platforms, and TIQS and xQTL processing platforms. R should be used as the principal computer language for QTL data analysis in all platforms and a ‘cloud’ should be used for software dissemination to the community. Furthermore, the working group recommended that all data models and software source code should be made visible in public repositories to allow a coordinated effort on the use of common data structures and file formats.
QTL mapping; database; mouse; systems genetics
The lung is critical in surveillance and initial defense against pathogens. In humans, as in mice, individual genetic differences strongly modulate pulmonary responses to infectious agents, severity of lung disease, and potential allergic reactions. In a first step towards understanding genetic predisposition and pulmonary molecular networks that underlie individual differences in disease vulnerability, we performed a global analysis of normative lung gene expression levels in inbred mouse strains and a large family of BXD strains that are widely used for systems genetics. Our goal is to provide a key community resource on the genetics of the normative lung transcriptome that can serve as a foundation for experimental analysis and allow predicting genetic predisposition and response to pathogens, allergens, and xenobiotics.
Steady-state polyA+ mRNA levels were assayed across a diverse and fully genotyped panel of 57 isogenic strains using the Affymetrix M430 2.0 array. Correlations of expression levels between genes were determined. Global expression QTL (eQTL) analysis and network covariance analysis was performed using tools and resources in GeneNetwork http://www.genenetwork.org.
Expression values were highly variable across strains and in many cases exhibited a high heri-tability factor. Several genes which showed a restricted expression to lung tissue were identified. Using correlations between gene expression values across all strains, we defined and extended memberships of several important molecular networks in the lung. Furthermore, we were able to extract signatures of immune cell subpopulations and characterize co-variation and shared genetic modulation. Known QTL regions for respiratory infection susceptibility were investigated and several cis-eQTL genes were identified. Numerous cis- and trans-regulated transcripts and chromosomal intervals with strong regulatory activity were mapped. The Cyp1a1 P450 transcript had a strong trans-acting eQTL (LOD 11.8) on Chr 12 at 36 ± 1 Mb. This interval contains the transcription factor Ahr that has a critical mis-sense allele in the DBA/2J haplotype and evidently modulates transcriptional activation by AhR.
Large-scale gene expression analyses in genetic reference populations revealed lung-specific and immune-cell gene expression profiles and suggested specific gene regulatory interactions.
Proteolysis of influenza virus hemagglutinin by host cell proteases is essential for viral infectivity, but the proteases responsible are not well defined. Recently, we showed that engineered expression of the type II transmembrane serine proteases (TTSPs) TMPRSS2 and TMPRSS4 allows hemagglutinin (HA) cleavage. Here we analyzed whether TMPRSS2 and TMPRSS4 are expressed in influenza virus target cells and support viral spread in the absence of exogenously added protease (trypsin). We found that transient expression of TMPRSS2 and TMPRSS4 resulted in HA cleavage and trypsin-independent viral spread. Endogenous expression of TMPRSS2 and TMPRSS4 in cell lines correlated with the ability to support the spread of influenza virus in the absence of trypsin, indicating that these proteases might activate influenza virus in naturally permissive cells. Indeed, RNA interference (RNAi)-mediated knockdown of both TMPRSS2 and TMPRSS4 in Caco-2 cells, which released fully infectious virus without trypsin treatment, markedly reduced the spread of influenza virus, demonstrating that these proteases were responsible for efficient proteolytic activation of HA in this cell line. Finally, TMPRSS2 was found to be coexpressed with the major receptor determinant of human influenza viruses, 2,6-linked sialic acids, in human alveolar epithelium, indicating that viral target cells in the human respiratory tract express TMPRSS2. Collectively, our results point toward an important role for TMPRSS2 and possibly TMPRSS4 in influenza virus replication and highlight the former protease as a potential therapeutic target.
Host serine proteases are essential for the influenza virus life cycle because the viral haemagglutinin is synthesized as a precursor which requires proteolytic maturation. Therefore, we studied the activity and expression of serine proteases in lungs from mice infected with influenza and evaluated the effect of serine protease inhibitors on virus replication both in cell culture and in infected mice.
Two different inbred mouse strains were investigated: DBA/2J as a highly susceptible and C57Bl/6J as a more resistant strain to influenza virus infection. The serine proteases from lung homogenates of mice exhibited pH optima of 10.00. Using the substrate Bz-Val-Gly-Arg-p-nitroanilide or in zymograms, the intensities of proteolysis increased in homogenates from both mouse strains with time post infection (p.i.) with the mouse-adapted influenza virus A/Puerto Rico/8/34 (H1N1; PR8). In zymograms at day 7 p.i., proteolytic bands were stronger and numerous in lung homogenates from DBA/2J than C57Bl/6J mice. Real-time PCR results confirmed differential expression of several lung proteases before and after infecting mice with the H1N1 virus. The most strongly up-regulated proteases were Gzma, Tmprss4, Elane, Ctrl, Gzmc and Gzmb. Pretreatment of mouse and human lung cell lines with the serine protease inhibitors AEBSF or pAB or a cocktail of both prior to infection with the H1N1 or the A/Seal/Massachusetts/1/80 (H7N7; SC35M) virus resulted in a decrease in virus replication. Pretreatment of C57Bl/6J mice with either AEBSF or a cocktail of AEBSF and pAB prior to infection with the H1N1 virus significantly reduced weight loss and led to a faster recovery of treated versus untreated mice while pAB alone exerted a very poor effect. After infection with the H7N7 virus, the most significant reduction of weight loss was obtained upon pretreatment with either the protease inhibitor cocktail or pAB. Furthermore, pretreatment of C57BL/6J mice with AEBSF prior to infection resulted in a significant reduction in the levels of both the H1N1 and H7N7 nucleoproteins in mice lungs and also a significant reduction in the levels of the HA transcript in the lungs of the H1N1- but not the H7N7-infected mice.
Multiple serine protease activities might be implicated in mediating influenza infection. Blocking influenza A virus infection in cultured lung epithelia and in mice by the used serine protease inhibitors may provide an alternative approach for treatment of influenza infection.
Quantitative trait locus (QTL) mapping identifies genomic regions that likely contain genes regulating a quantitative trait. However, QTL regions may encompass tens to hundreds of genes. To find the most promising candidate genes that regulate the trait, the biologist typically collects information from multiple resources about the genes in the QTL interval. This process is very laborious and time consuming.
QTLminer is a bioinformatics tool that automatically performs QTL region analysis. It is available in GeneNetwork and it integrates information such as gene annotation, gene expression and sequence polymorphisms for all the genes within a given genomic interval.
QTLminer substantially speeds up discovery of the most promising candidate genes within a QTL region.
The analysis of expression quantitative trait loci (eQTL) is a potentially powerful way to detect transcriptional regulatory relationships at the genomic scale. However, eQTL data sets often go underexploited because legacy QTL methods are used to map the relationship between the expression trait and genotype. Often these methods are inappropriate for complex traits such as gene expression, particularly in the case of epistasis.
Here we compare legacy QTL mapping methods with several modern multi-locus methods and evaluate their ability to produce eQTL that agree with independent external data in a systematic way. We found that the modern multi-locus methods (Random Forests, sparse partial least squares, lasso, and elastic net) clearly outperformed the legacy QTL methods (Haley-Knott regression and composite interval mapping) in terms of biological relevance of the mapped eQTL. In particular, we found that our new approach, based on Random Forests, showed superior performance among the multi-locus methods.
Benchmarks based on the recapitulation of experimental findings provide valuable insight when selecting the appropriate eQTL mapping method. Our battery of tests suggests that Random Forests map eQTL that are more likely to be validated by independent data, when compared to competing multi-locus and legacy eQTL mapping methods.
The importance of the adaptive immune response for secondary influenza infections and protection from a lethal challenge after vaccination has been well documented. However, some controversy still exists concerning the specific involvement of B and T cells during a primary infection. Here, we have followed the survival, weight loss, viral load and lung pathology in Rag2-/- knock-out mice after infection with influenza A virus (H1N1). Infected wild type mice initially lost weight early after infection but then cleared the virus and recovered. Rag2-/- mice, however, showed similar weight loss kinetics in the early stages after infection but weight loss continued post infection and culminated in death. In contrast to wild type mice, Rag2-/- mice were not able to clear the virus, despite an increased inflammatory response. Furthermore, they did not recruit virus-specific lymphocytes into the lung in the later stages after infection and exhibited sustained pulmonary lesions.
The first scientific meeting of the newly established European SYSGENET network took place at the Helmholtz Centre for Infection Research (HZI) in Braunschweig, April 7-9, 2010. About 50 researchers working in the field of systems genetics using mouse genetic reference populations (GRP) participated in the meeting and exchanged their results, phenotyping approaches, and data analysis tools for studying systems genetics. In addition, the future of GRP resources and phenotyping in Europe was discussed.
The recent explosion of biological data and the concomitant proliferation of distributed databases make it challenging for biologists and bioinformaticians to discover the best data resources for their needs, and the most efficient way to access and use them. Despite a rapid acceleration in uptake of syntactic and semantic standards for interoperability, it is still difficult for users to find which databases support the standards and interfaces that they need. To solve these problems, several groups are developing registries of databases that capture key metadata describing the biological scope, utility, accessibility, ease-of-use and existence of web services allowing interoperability between resources. Here, we describe some of these initiatives including a novel formalism, the Database Description Framework, for describing database operations and functionality and encouraging good database practise. We expect such approaches will result in improved discovery, uptake and utilization of data resources.
Database URL: http://www.casimir.org.uk/casimir_ddf
The laboratory mouse has become the organism of choice for discovering gene function and unravelling pathogenetic mechanisms of human diseases through the application of various functional genomic approaches. The resulting deluge of data has led to the deployment of numerous online resources and the concomitant need for formalized experimental descriptions, data standardization, database interoperability and integration, a need that has yet to be met. We present here the Mouse Resource Browser (MRB), a database of mouse databases that indexes 217 publicly available mouse resources under 22 categories and uses a standardised database description framework (the CASIMIR DDF) to provide information on their controlled vocabularies (ontologies and minimum information standards), and technical information on programmatic access and data availability. Focusing on interoperability and integration, MRB offers automatic generation of downloadable and re-distributable SOAP application-programming interfaces for resources that provide direct database access. MRB aims to provide useful information to both bench scientists, who can easily navigate and find all mouse related resources in one place, and bioinformaticians, who will be provided with interoperable resources containing data which can be mined and integrated.
Database URL: http://bioit.fleming.gr/mrb
XGAP, a software platform for the integration and analysis of genotype and phenotype data.
We present an extensible software model for the genotype and phenotype community, XGAP. Readers can download a standard XGAP (http://www.xgap.org) or auto-generate a custom version using MOLGENIS with programming interfaces to R-software and web-services or user interfaces for biologists. XGAP has simple load formats for any type of genotype, epigenotype, transcript, protein, metabolite or other phenotype data. Current functionality includes tools ranging from eQTL analysis in mouse to genome-wide association studies in humans.
The integration of information present in many disparate biological databases represents a major challenge in biomedical research. To define the problems and needs, and to explore strategies for database integration in mouse functional genomics, we consulted the biologist user community and implemented solutions to two user-defined use-cases.
We organised workshops, meetings and used a questionnaire to identify the needs of biologist database users in mouse functional genomics. As a result, two use-cases were developed that can be used to drive future designs or extensions of mouse databases. Here, we present the use-cases and describe some initial computational solutions for them. The application for the gene-centric use-case, "MUSIG-Gen" starts from a list of gene names and collects a wide range of data types from several distributed databases in a "shopping cart"-like manner. The iterative user-driven approach is a response to strongly articulated requests from users, especially those without computational biology backgrounds. The application for the phenotype-centric use-case, "MUSIG-Phen", is based on a similar concept and starting from phenotype descriptions retrieves information for associated genes.
The use-cases created, and their prototype software implementations should help to better define biologists' needs for database integration and may serve as a starting point for future bioinformatics solutions aimed at end-user biologists.
Following the technological advances that have enabled genome-wide analysis in most model organisms over the last decade, there has been unprecedented growth in genomic and post-genomic science with concomitant generation of an exponentially increasing volume of data and material resources. As a result, numerous repositories have been created to store and archive data, organisms and material, which are of substantial value to the whole community. Sustained access, facilitating re-use of these resources, is essential, not only for validation, but for re-analysis, testing of new hypotheses and developing new technologies/platforms. A common challenge for most data resources and biological repositories today is finding financial support for maintenance and development to best serve the scientific community. In this study we examine the problems that currently confront the data and resource infrastructure underlying the biomedical sciences. We discuss the financial sustainability issues and potential business models that could be adopted by biological resources and consider long term preservation issues within the context of mouse functional genomics efforts in Europe.
The genetic make-up of the host has a major influence on its response to combat pathogens. For influenza A virus, several single gene mutations have been described which contribute to survival, the immune response and clearance of the pathogen by the host organism. Here, we have studied the influence of the genetic background to influenza A H1N1 (PR8) and H7N7 (SC35M) viruses. The seven inbred laboratory strains of mice analyzed exhibited different weight loss kinetics and survival rates after infection with PR8. Two strains in particular, DBA/2J and A/J, showed very high susceptibility to viral infections compared to all other strains. The LD50 to the influenza virus PR8 in DBA/2J mice was more than 1000-fold lower than in C57BL/6J mice. High susceptibility in DBA/2J mice was also observed after infection with influenza strain SC35M. In addition, infected DBA/2J mice showed a higher viral load in their lungs, elevated expression of cytokines and chemokines, and a more severe and extended lung pathology compared to infected C57BL/6J mice. These findings indicate a major contribution of the genetic background of the host to influenza A virus infections. The overall response in highly susceptible DBA/2J mice resembled the pathology described for infections with the highly virulent influenza H1N1-1918 and newly emerged H5N1 viruses.
Retinoic acid is a signaling molecule involved in the regulation of growth and morphogenesis during development. There are three types of nuclear receptors for all-trans retinoic acid in mammals, RARα, RARβ, and RARγ, which transduce the retinoic acid signal by inducing or repressing the transcription of target genes (Leid, M., P. Kastner, and P. Chambon. 1992. Trends Biochem. Sci. 17:427–433). While RARα, RARβ, and RARγ are expressed in distinct but overlapping patterns in the developing mouse limb, their exact role in limb development remains unclear. To better understand the role of retinoic acid receptors in mammalian limb development, we have ectopically expressed a modified RARα with constitutive activity (Balkan, W., G.K. Klintworth, C.B. Bock, and E. Linney. 1992. Dev. Biol. 151:622–625) in the limbs of transgenic mice. Overexpression of the transgene was associated with marked pre- and postaxial limb defects, particularly in the hind limb, where expression of the transgene was consistently seen across the whole anteroposterior axis. The defects displayed in these mice recapitulate, to a large degree, many of the congenital limb malformations observed in the fetuses of dams administered high doses of retinoic acid (Kochhar, D.M. 1973. Teratology. 7:289–295). Further analysis of these transgenic animals showed that the defect in skeletogenesis resided at the level of chondrogenesis. Comparison of the expression of the transgene relative to that of endogenous RARα revealed that downregulation of RARα is important in allowing the chondrogenic phenotype to be expressed. These results demonstrate a specific function for RARα in limb development and the regulation of chondroblast differentiation.