Measuring protein-protein interactions generally means choosing between looking at a single protein pair in an individual living cell or assaying multiple interactions in cell extracts. But in Angewandte Chemie, Gandor et al. describe tantalizing steps toward combining the best of both approaches. The authors envision a chip in which chemically patterned arrays dictate the position of bait proteins within cells growing on the chip's surface. In essence, the chip provides spatial addressing, and the cell acts as the reaction vessel. The array begins with micropatterned DNA oligonucleotide spots on a glass slide, which capture complementary streptavidin-conjugate oligonucleotides that in turn recruit a biotinylated antibody. It is this microstructure that positions an artificial receptor protein expressed in the cells grown on the chip. The authors dub the receptor bait-PARC (bait-presenting artificial receptor construct); each bait-PARC consists of an extracellular domain displaying an epitope for recognition by the antibody on the microstructure bound to the slide, spacer sequences to make the epitope accessible, a transmembrane domain, and an intracellular fluorescent protein fused with the bait protein. Binding between the arrayed antibody and the artificial receptor epitope should rearrange the bait-PARCs into defined spots. If the cell also contains fluorescently labeled prey, bait-prey colocalization can be detected. Such an elaborately engineered system might sound a bit far-fetched, but Gandor et al. present evidence that bait-PARCs can work as envisioned. First, they demonstrate that patterned microstructures can position the bait-PARCs within cells by showing that two differently labeled bait-PARCs clustered in a way that mimics the checkerboard pattern established by oligonucleotide spotted on the chip. Next, to show that the system can identify protein-protein interactions, the authors tracked the association of protein kinase A (PKA) catalytic subunit prey with PKA regulatory subunit bait in response to changes in the intracellular concentration of cAMP. Then, to demonstrate side-by-side analysis of two protein-protein interactions, they compared the binding behavior of bait- PARCs containing either regulatory subunit I-α or II-β. Their studies revealed substantial cell-to-cell variability, highlighting a key benefit of the intracellular protein interaction array: the ability to detect temporal cross-correlations that become blurred when measurements are derived from many cells studied in aggregate.
S. Gandor et al. 2013. A protein-interaction array inside a living cell. Angew Chem Int Ed Engl. 52:4790-4.Consider the Alternatives
Not long ago, asking which genes are expressed in a particular cell type seemed insurmountably complicated. Today, however, that question is overly simplistic; the real stumper is identifying which proteoforms of which genes are present. Many of these protein variants derive from alternative splicing, and while analysis of cellular RNA can uncover alternatively spliced transcripts, this does not necessarily mean that the variant mRNAs are translated. On the other hand, identification of cellular proteins by mass spectrometry is limited to peptides already present in a database. In Molecular and Cellular Proteomics, Sheynkman et al. describe how they have combined customized RNA and proteomic data to tackle this problem. Pointing out that it is possible to sequence a complete transcriptome in a few days using RNA-Seq, they advocate empirically determining the noncanonical splice products in the cell type to be analyzed proteomically. Their publication describes a bioinformatics workflow for creating a splice junction database and then applying it to the discovery of novel proteoforms. Using Jurkat cells, the authors collected RNA-Seq data and then ran it through splice-junction discovery programs. Over 144,000 unannotated splice junctions were detected. They set a threshold of six supporting reads per putative junction, based on the expected abundance needed for detection of the corresponding protein. Just under 25,000 junction sequences made the cut. From these, amino acid sequences were predicted and combined with the UniProt database. The authors looked for peptide matches with mass spectrometry of Jurkat lysate using this enhanced database. To guard against artifactual hits, the method requires a better peptide score than usual, corresponding to a 1% false discovery rate threshold. The process yielded 57 novel junction peptides. The authors argue that parallel analysis of RNA and protein is preferable to expanding existing databases with in silico–predicted peptides or transcriptomic data from unrelated cell types, since these approaches increase false positives, database redundancy, and data analysis times. By contrast, this sample-specific method for preparing high-quality, customized splice junction databases makes discovery of novel proteoforms efficient and highly reliable.
G.M. Sheynkman et al.. Discovery and mass spectrometric analysis of novel splice-junction peptides using RNA-Seq. Mol Cell Proteomics. [Epub ahead of print, April 29, 2013].