The transcriptome of endosymbiotic Bradyrhizobium japonicum bacteroids was assessed, using RNA extracted from determinate soybean root nodules. Results were compared with the transcript profiles of B. japonicum cells grown in either aerobic or microaerobic culture. Microoxia is a known trigger for the induction of symbiotically relevant genes. In fact, one third of the genes induced in bacteroids at day 21 after inoculation are congruent with those up-regulated in culture by a decreased oxygen concentration. The other induced genes, however, may be regulated by cues other than oxygen limitation. Both groups of genes provide a rich source for the possible discovery of novel functions related to symbiosis. Samples taken at different timepoints in nodule development have led to the distinction of genes expressed early and late in bacteroids. The experimental approach applied here is also useful for B. japonicum mutant analyses. As an example, we compared the transcriptome of wild-type bacteroids with that of bacteroids formed by a mutant defective in the RNA polymerase transcription factor sigma54. This led to a collection of hitherto unrecognized B. japonicum genes potentially transcribed in planta in a sigma54-dependent manner.

Pollen, the male gametophyte of flowering plants, represents an ideal biological system to study developmental processes, such as cell polarity, tip growth, and morphogenesis. Upon hydration, the metabolically quiescent pollen rapidly switches to an active state, exhibiting extremely fast growth. This rapid switch requires relevant proteins to be stored in the mature pollen, where they have to retain functionality in a desiccated environment. Using a shotgun proteomics approach, we unambiguously identified ∼3500 proteins in Arabidopsis pollen, including 537 proteins that were not identified in genetic or transcriptomic studies. To generate this comprehensive reference data set, which extends the previously reported pollen proteome by a factor of 13, we developed a novel deterministic peptide classification scheme for protein inference. This generally applicable approach considers the gene model–protein sequence–protein accession relationships. It allowed us to classify and eliminate ambiguities inherently associated with any shotgun proteomics data set, to report a conservative list of protein identifications, and to seamlessly integrate data from previous transcriptomics studies. Manual validation of proteins unambiguously identified by a single, information-rich peptide enabled us to significantly reduce the false discovery rate, while keeping valuable identifications of shorter and lower abundant proteins. Bioinformatic analyses revealed a higher stability of pollen proteins compared to those of other tissues and implied a protein family of previously unknown function in vesicle trafficking. Interestingly, the pollen proteome is most similar to that of seeds, indicating physiological similarities between these developmentally distinct tissues.

Rhizobia are able to infect legume roots, elicit root nodules, and live therein as endosymbiotic, nitrogen-fixing bacteroids. Host recognition and specificity are the results of early programming events in bacteria and plants, in which important signal molecules play key roles. Here, we introduce a new aspect of this symbiosis: the adaptive response to hosts. This refers to late events in bacteroids in which specific genes are transcribed and translated that help the endosymbionts to meet the disparate environmental requirements imposed by the hosts in which they live. The host-adaptation concept was elaborated with Bradyrhizobium japonicum and three different legumes (soybean, cowpea, and siratro). Transcriptomes and proteomes in root-nodule bacteroids were analyzed and compared, and genes and proteins were identified which are specifically induced in only one of the three hosts. We focused on those determinants that were congruent in the two data sets of host-specific transcripts and proteins: seven for soybean, five for siratro, and two for cowpea. One gene cluster for a predicted ABC-type transporter, differentially expressed in siratro, was deleted in B. japonicum. The respective mutant had a symbiotic defect on siratro rather than on soybean or cowpea. This result demonstrates the value of the applied approach and corroborates the host-specific adaptation concept.

We present a concept of interactive learning and probabilistic retrieval of user-specific cover types in a content-based remote sensing image archive. A cover type is incrementally defined via user-provided positive and negative examples. From these examples, we infer probabilities of the Bayesian network that link the user interests to a pre-extracted content index. Due to the stochastic nature of the cover type definitions, the database system not only retrieves images according to the estimated coverage but also according to the accuracy of that estimation given the current state of learning. For the latter, we introduce the concept of separability. We expand on the steps of Bayesian inference to compute the application-free content index using a family of data models, and on the description of the stochastic link using hyperparameters. In particular, we focus on the interactive nature of our approach, which provides instantaneous feedback to the user in the form of an immediate update of the posterior map, and a very fast, approximate search in the archive. A java-based demonstrator using the presented concept of content-based access to a test archive of Landsat TM, X-SAR, and aerial images are available over the Internet [http://www.vision.ee.ethz.ch/~rsia/ClickBayes].

We present Gibbs-Markov random field (GMRF) models as a powerful and robust descriptor of spatial information in typical remote-sensing image data. This class of stochastic image models provides an intuitive description of the image data using parameters of an energy function. For the selection among several nested models and the fit of the model, we proceed in two steps of Bayesian inference. This procedure yields the most plausible model and its most likely parameters, which together describe the image content in an optimal way. Its additional application at multiple scales of the image enables us to capture all structures being present in complex remote-sensing images. The calculation of the evidences of various models applied to the resulting quasicontinuous image pyramid automatically detects such structures. We present examples for both synthetic aperture radar (SAR) and optical data.