New technologies used in biology are generating huge quantities of data; up to two petabytes of overall data are to be expected by the end of the decade. Modern biology also has to deal with multitude of information representations that are driven by the continuous evolution of biological knowledge. Conceptual modeling allows layered descriptions whose abstractions are rather stable while their concrete descriptions may evolve: formal ontologies versus domain ontologies, metamodels versus models, etc. In this paper we introduce an incremental integration methodology that is based on the Open Biomedical Ontology (OBO) umbrella. The proof of concept for our methodology is carried out using the Gene Ontology's (GO) case study in which GO extensions are viewed as manifestations of the domain knowledge evolution. Based on this conceptual foundation, we present a metamodeling architecture that encompasses metamodels and models of GO, as well as those of GO extensions. Our architecture can be used for ensuring semantical quality of biological knowledge integration.