ERly Detection

Given both the relevance of apoptosis to cancer research and the significance of Ca²⁺ release from the endoplasmic reticulum (ER) to programmed cell death, more information about Ca²⁺ dynamics would permit a fuller understanding of apoptotic signaling cascades and the action of anti-apoptotic proteins such as the Bcl-2 family members. Current methods for imaging Ca²⁺ concentrations use either small-molecule Ca²⁺ indicators or expression of ER-targeted aequorin. The latter method is limited in part by the low signal of this luminescent reporter, while the former suffers from a lack of a straightforward means to selectively measure ER-specific Ca²⁺. As an alternative to these approaches, Palmer et al. investigated whether modified cameleon sensors might allow direct imaging of intra-ER Ca²⁺ in individual cells. Cameleons are proteins designed for fluorescent detection of Ca²⁺; they consist of a fusion of cyan fluorescent protein (CFP), calmodulin, a calcium-dependent calmodulin-binding peptide, and yellow fluorescent protein (YFP). Upon binding Ca²⁺, calmodulin interacts with the adjacent peptide, resulting in a conformational change that induces fluorescence resonance energy transfer (FRET). Because existing cameleons have unfavorable Ca²⁺ Kd values for measuring intra-ER Ca²⁺ levels and since endogenous calmodulin can interfere with their function, the authors reengineered the protein and identified a mutant that overcomes these two deficiencies. With the addition of ER targeting and retention signals, the modified cameleon (named D1ER) outperformed the best-existing ER-targeted cameleon, displaying a dramatic increase in sensitivity. Furthermore, the FRET partners do not spectrally overlap with the intracellular calcium indicator Fura-2, allowing simultaneous imaging of ER and cytosolic Ca²⁺. Together, these reagents allow tracking of Ca²⁺ oscillations, and pleasingly, the ER and cytosolic patterns are inversely related. Having validated the tool, the authors go on to demonstrate its utility for teasing out the effects of Bcl-2 on calcium signaling and for analyzing the action of a Bcl-2 inhibitor. Thus, by allowing straightforward detection of Ca²⁺ in the ER, the new cameleon offers exciting new possibilities for understanding the role of antiapoptotic factors in tumorigenesis.

Proceedings of the National Academy of Sciences of the USA 101(50):17404–17409 (2004) ND

Protein ELIZA

ELIZA, a computer program developed at MIT in the mid-1960s, used natural language communication to mimic the responses of a psychoanalyst. Nearly 40 years later, therapists show no sign of being replaced by machines, but the burgeoning amount of biomedical literature has made computerized parsing of natural language a high-priority goal. In this case, the object is to automatically examine the scientific literature and use natural language techniques to derive information regarding the function and relationships of proteins. A first step to achieve this involves the assignment of a controlled vocabulary term [e.g., a gene ontology (GO) identification] to each gene product. For example, if a publication implicates gene A in nucleocytoplasmic transport of poly(A) RNA, then the GO process ID “mRNA-nucleus export” could be assigned, and functional information about the product of gene A could become available to researchers even if they haven't come across the original publication. Not surprisingly, this worthy goal is anything but straightforward to implement. Koike et al. describe a method for automatically assigning GO process IDs using natural language processing. Their strategy uses PubMed abstracts as input and analyzes sentence structure to assign GO IDs. Because specific GO IDs are not habitually used by scientists in writing abstracts, the process relies upon semiautomatic augmentation of functional terms, for example, by looking for terms that tend to appear together with bona fide GO terminology. Gene product-function relationships are extracted based on an analysis of sentence structure and the syntactic relationship between the gene name and the function term. This strategy is more precise than simply scoring co-occurrence of gene name and function. Although the authors emphasize the challenges still posed by the sheer complexity of natural language, their method extracted an impressive set of gene product-function relationships, and of these, some 80% were not already available in widely used databases. The authors provide these data in a searchable web database that may be imperfect like ELIZA, but no less intriguing.

Bioinformatics doi:10.1093/bioinformatics/bti084 2004 Oct 27; [Epub ahead of print]

Dysfunctional Relationship

In an ideal world, the subcomponents of a recombinant fusion protein would work in harmony; the protein of interest would go about its usual cellular duties, and the tag would quietly report their whereabouts to interested observers. Nothing, of course, is that easy, and often the tag can interfere with the localization or activity of its fusion partner. For example, green fluorescent protein (GFP)-tagged yeast -tubulin cannot functionally replace the native protein, probably as a result of steric hindrance by the 25-kDa tag. A much smaller 6 amino acid tag has been described for fluorescent labeling—this strategy employs a tetracysteine motif that can bind an exogenously introduced arsenic-containing fluorescent compound. The fluorescein version, termed FlAsH, has proven adept for in vivo labeling of metazoan cells, but to date, no method has been available for budding yeast, owing to the problem of obtaining a sufficient intracellular concentration of the tetracysteine binding moiety. Andresen et al. describe a novel staining procedure for using the FlAsH system in Saccharomyces cerevisiae. Moreover, the authors describe a series of experiments to clarify the relevant considerations in selecting a fluorescent tagging strategy. For instance, a comparison of proteins with single, double, or triple tetracysteine motifs revealed that while the multimers did not improve fluorescence intensity, having more than one FlAsH-binding peptide did decrease photobleaching. On the other hand, protein function can diminish as the tag size increases. For instance, when the essential factor -tubulin is tagged with three tetracysteine motifs, it, like the GFP-tagged version, results in nonviable haploid cells. Moreover, including just one tetracysteine motif in the fusion protein produces a strain with a growth defect. Therefore, even a 10 amino acid extension to -tubulin (the motif plus a 4 amino acid linker) can begin to compromise protein function, showing the sensitivity of some factors to tagging strategies. Although the FlAsH system is not as bright in yeast as GFP, nonspecific signal is minimal, making this the method of choice for pairing up picky proteins in functional fusions.

Molecular Biology of the Cell in Press, published online ahead of print October 6, 2004 10.1091/mbc.E04-06-0454ND