Summary of approaches used to predict genes encoding small proteins.
A. Homology-based searches using intergenic DNA sequences as input.
B. Searches for ribosome binding sites using intergenic DNA sequences as input.
C. Homology-based searching using protein sequences predicted from ribosome binding sites as input.
aa = amino acids

Immunoblot analysis of small proteins predicted on the basis of DNA homology.
Whole cell extracts of MG1655 cells grown to (E) exponential and (S) stationary phase in LB medium were analyzed as in Fig. 1. Again the star (*) indicates the band corresponding the fusion protein. The caveats of the marker lane and exposure times are as for Fig. 1.

Immunoblot analysis of small proteins predicted on the basis of the presence of a RBS and protein homology.
Whole cell extracts of MG1655 cells grown to (E) exponential and (S) stationary phase in LB medium were analyzed as in Fig. 1. Again the star (*) indicates the band corresponding the fusion protein. The caveats of the marker lane and exposure times are as for Fig. 1.

Immunoblot analysis of previously-annotated short ORFs tagged with SPA. Immunoblot analysis using anti-FLAG, alkaline phosphatase-conjugated antibodies was carried out with whole cell extracts harvested from MG1655 cultures.
A. Cells were grown to (E) exponential and (S) stationary phase in LB medium. Immunoblots shown on the top row are of highly expressed small proteins, whereas immunoblots on the bottom row are of proteins with lower expression. MG1655 control lanes were run for each blot and representative lanes are shown. For those proteins with lower expression (bottom row) an additional cross-reacting band of approximate 26 kDa can also be observed for MG1655 grown to stationary phase.
B. Cells were grown to (E) exponential and (S) stationary phase in LB and M63 medium containing 0.2% glucose.
C. Cells were grown to (T0) exponential phase in LB medium and then treated with (−) water or (+) 1% α-methylglucoside. In all cases, a fraction equivalent to the cells in OD₆₀₀ = 0.057 was loaded in each lane. The star (*) indicates the band corresponding the fusion protein. The positions of the markers for one blot are shown. This only provides the approximate sizes of the proteins because there was slight variation in the running of gels. Exposure times were optimized for each panel for visualization here, therefore the signal intensity shown does not indicate relative abundance between proteins. Given the need for longer exposure times, some background bands were detected for the immunoblots in the second row of (A).

Alignments for short ORFs identified on the basis of DNA homology.
Gene sequences identified in the DNA-as-input search were translated and the predicted protein was used to search for homologs using tblastn. Alignments were generated using ClustalW (Fig. 1http://align.genome.jp). “*” indicates the residues are identical in all sequences and “:” and “.” respectively indicate that conserved and semi-conserved substitutions as defined by ClustalW. Swiss-Prot organism codes are from EcoGene (www.ecogene.org/modules.php?name=NEWT).

Immunoblot analysis of small proteins predicted on the basis of potential RBSs with high information content.
A. Small proteins predicted in an initial search for genes with canonical RBSs.
B. Small proteins predicted using an information theory-based RBS search for open reading frames. For both (A) and (B), whole cell extracts of MG1655 cells grown to (E) exponential and (S) stationary phase in LB medium were analyzed as in Fig. 1. Again the star (*) indicates the band corresponding the fusion protein. The caveats of the marker lane and exposure times are as for Fig. 1. Given the need for longer exposure times, some background bands were detected for the immunoblots in second row of (B).

Subcellular fractionation of small proteins predicted to have transmembrane segments.
Whole cell lysates were generated from the SPA-tag strains and fractionated into cytoplasmic supernatant and membrane pellet fractions. Immunoblot analysis of the lysates and fractions was carried out as in Fig. 1. The star (*) indicates the band corresponding the fusion protein.