In SILAC, normalization was performed using the original mixture of the cells at a 1:1 ratio and reaching 100% labeling efficiency for both cell populations. The limitation of selecting a threshold of expression to consider proteins to be differentially expressed requires a followup validation analysis for key data. Verification of changes by two independent analytical methods, and using independent in vitro Cinoxacin strategies and clinical material provided confidence that the experimental design permitted significant changes in abundance to be validated. The limited correlation between transcript and protein expression at their steady state was similar to the 0.28 previously reported in pancreatic cells. This could be attributed to the wider range of ratios of expression measured by gene arrays while the majority of the SILAC ratios were in the low range. SILAC ratios were more limited due to the internal labelling and the characteristic 1:1 mixture of the protein extracts analyzed. The weak correlation between the gene array and SILAC ratios highlighted the relevance of quantitative proteomic approaches to estimate the expression of proteins of interest, in concordance with previous reports. There were missing data between both techniques because not all the coding Gomisin-D products of the genes measured by the early version of the Affymetrix oligonucleotide array were detected by SILAC. Similarly, genes coding for the 831 proteins identified by SILAC duplicates were not included among the probes contained in the commercial U133 oligonucleotide array. Availability of both transcript and protein expression levels could also be utilized to uncover potential regulatory mechanisms modifying translation or protein degradation. Immunoblotting validation was closely correlated to the SILAC results, and also served to validate candidates identified in oligonucleotide arrays. Cul3 was selected from the top over-expressed candidates in T24T not previously characterized in bladder cancer for which we had available reagents for further studies. Cul3 was differentially expressed in T24T using three different methodologies: SILAC, gene arrays and immunoblotting. Cul3 is one of the four members of the cullin protein family. It belongs to the core component of multiple ubiquitin-protein ligase complexes that mediate the ubiquitination and subsequent proteasomal degradation of their target proteins. Cul3 acts as a scaffolding protein in a heterodimeric complex playing a central role in the specificity of polyubiquitinization of these proteins, positioning the substrate and the ubiquitin-conjugating enzyme. Although the full list of targets whose ubiquitination and degradation is mediated by Cul3 remains unknown, cancer-related proteins reported include cyclin E, or Rho, among others. In concordance with the interaction network shown in Figure S3, it could be proposed that Cul3 would be involved in the proteasomal degradation of adhesion associated cytoskeletal proteins such as filamin A, ezrin, caveolin1 or moesin. Indeed, the expression of these proteins increased upon Cul3 silencing, observations highlighting the impact of Cul3 expression not only on the aggressive phenotype of T24T shown by functional assays, but also modifying the expression of other proteins identified by SILAC. It remains 
to be characterized whether Cul3 might be directly involved in the proteasomal degradation of cytoskeleton proteins, potentially regulating the migration and invasive aggressiveness properties of T24T cells.