Rsp5-mediated ubiquitination of the yeast proteome. (A) Assay development. To optimize ubiquitination conditions using protein microarrays, known substrates of Rsp5 (CTD and Ydl203c) and proteins not ubiquitinated by Rsp5 in vitro (Yer036c and GST alone) were robotically printed on slides and incubated in ubiquitination reactions containing Rsp5 and FITC-labeled ubiquitin. The fluorescent signal demonstrates CTD and Ydl203c ubiquitination in the presence of ATP (right panel), while negative control proteins are not ubiquitinated (left panel). Blue color represents ubiquitination (detected with FITC-Ub). GFP was used as a positive control, as it has the same excitation wavelength as FITC. The colors associated with the protein microarray spots indicate of the intensity of the signal (with light blue<bright blue<white). (B) Image of a scanned ubiquitinated protein microarray with an enlargement of one grid. All proteins are printed in duplicate and arrows indicate proteins that were identified as substrates after quantitative data analysis. Alexa dyes are spotted as controls in the left-hand corners of each grid. (C) Reproducibility. Two protein microarrays were ubiquitinated in separate experiments and the same grid from each array is shown. Arrows point to ubiquitinated proteins that were identified as substrates. Most spots producing significantly higher signal than background can be seen on both arrays, suggesting high reproducibility between slides.

Validation of substrate ubiquitination in vitro and in vivo. (A, B) In vitro ubiquitination: (A) 15 proteins identified as ‘high-confidence' Rsp5 substrates using protein microarrays were expressed (fused to GST) in yeast, purified and incubated in ubiquitination reactions containing Rsp5. (B) Six randomly selected proteins that were not identified as Rsp5 substrates in the protein microarray experiments were used as negative controls. All 15 of the ‘high-confidence' Rsp5 substrates and none of the negative control proteins were visibly ubiquitinated in the Western blots with anti-GST antibodies (arrows indicate the original size of the protein in the absence of ubiquitination (i.e. without ATP)). (C) In vivo ubiquitination: example of three Rsp5 substrates from the protein microarray exhibiting ubiquitination in vivo. The three proteins (HA tagged) were expressed in RSP5 (WT) or rsp5-1 mutant yeast cells. Following a shift to the non-permissive temperature (37°C), proteins were immunoprecipitated with anti-HA antibodies and immunoblotted with anti-ubiquitin antibodies. Note ubiquitination in the RSP5-WT but not the rsp5-1 cells.

Sequence logos for substrates of Rsp5. PPxY motifs, together with six residues upstream and downstream of the motif from proteins identified as substrates of Rsp5, were aligned and used to generate sequence logos (Crooks et al, 2004). In each logo, stacks of letters indicate the relative frequency of certain amino acids at each position in the sequence. The overall height provides a guide to the level of sequence conservation associated at that position.

Phage display logos. Peptides identified as substrates of Rsp5, using the phage display system, were aligned and are graphically displayed as logos, as shown in Figure 3 above.

Conservation of Rsp5 substrates across fungal species. BLAST analyses were used to identify potential orthologues and homologues of the top 49 most highly ubiquitinated Rsp5 substrates, together with the presence of a (L/P)PxY motif in 17 species of ascomycetes. Putative orthologues were defined as those demonstrating reciprocal best BLAST matches, and best scoring homologues were defined as those matches with the highest BLAST bit scores, which were considered homologous through manual inspection of sequence alignments (see Materials and methods). The cladogram at the top of the figure shows the phylogenetic relationships of the 18 fungal species (including S. cerevisiae) considered.

Network diagram describing Rsp5 functional pathways. Proteins identified from this study, together with previously published studies, were used to generate a protein–protein interaction network describing functional pathways associated with Rsp5. Two studies have recently attempted to comprehensively define the physical interactome in budding yeast through systematic affinity tagging, purification and mass spectrometry (Gavin et al, 2006; Krogan et al, 2006). A more recent study that combines these data sets to produce a single high-quality integrated data set (Collins et al, 2007a, 2007b) was used to identify proteins with physical interactions to: 42 (of 150) proteins identified in the current study as being a substrate of Rsp5; 40 (of 155) proteins identified in this study as binding partners of Rsp5; 24 (of 59) proteins identified in other small-scale experiments as being substrates of Rsp5; 38 (of 110) proteins identified in a previous high-throughput screen of Rsp5 binding partners (Hesselberth et al, 2006) and 39 (of 79) proteins, whose genes have been found to either positively or negatively genetically interact with Rsp5. Genetic interactions were obtained from an rsp5-DamP hypomorphic allele (Schuldiner et al, 2005; Collins et al, 2007b) derived from a recently generated E-MAP (epistatic miniarray profile), using a normalized score of >1.5 for positive interactions and <−1.5 for negative interactions. Physical interactions were obtained using a purification enrichment (PE) score of >3.2 (Collins et al, 2007a, 2007b). In total, 1258 interactions (light blue lines) between 694 proteins were identified. Colors and sizes of nodes are described in the inset key.