Second, because E. coli lacks an endogenous sumoylation system, the pathway may be isolated up to the point of the E2 for study. However, these systems are not without shortcomings. E3-independent sumoylation itself occurs at quantifiable levels only for protein concentrations far exceeding physiological levels. While proteins are clearly sumoylated, the physiological relevance of the modified proteins is unclear. For example, Menca and de Lorenzo observed attachment of polySUMO-1 chains to target proteins in E. coli. Because SUMO1 lacks the consensus sequence present on SUMO-2 and SUMO-3, it is not believed to homo-polymerize. However, more recent in vitro studies have shown that SUMO-1 is capable of forming chains through non-consensus lysines, albeit to a far lesser extent compared to SUMO-2 and SUMO-3.
The physiological relevance of such poly-SUMO-1 chains is unclear, and SUMO-1 itself may be more involved in chain termination of SUMO-2 and SUMO-3 rather than formation in vivo. Along similar lines, the physiological Z-VAD-FMK significance of some sumoylation sites observed by Okada et al. using sumo-engineered E. coli is also unclear. Here, we engineered an E3-dependent SUMO-conjugation system in E. coli that employs members of the mammalian PIAS E3 ligase family and, as a result, involves no observable polysumoylation of target proteins. Furthermore, because E. coli lacks organelles and an endogenous sumoylation pathway, our system provides an alternative in vivo context that is insulated from factors such as target localization, downstream interactions. Finally, we show that addition of the E3 increases the efficiency of sumoylation, yielding as much as,5 mg/L of SUMO-modified proteins. This makes possible greater titers of specifically sumoylated target proteins for use in biochemical and structural characterization. Mutation of this lysine residue to arginine abolished Smad4 sumoylation. To verify that K159 is the major site of SUMO attachment in our system, we performed MALDI-TOF mass spectrometry analysis on the SUMO-Smad4 band, which was purified on a Ni-NTA column and separated from unmodified Smad4 by SDS-PAGE. As expected, nearly all of the Smad4 was sumoylated at the consensus K159. An even higher molecular weight band relative to SUMOSmad4 was also produced in our sumo-engineered E. coli. This band might correspond to the attachment of SUMO-1 to a minor site on Smad4 or to the formation of SUMO-1 chains on Smad4.
We favored the former possibility for two reasons. First, low-level expression of the E1 and E2 along with the E3 promoted mono-sumoylation in the case of GFP-PML. Consistent with this result, MS analysis of SUMOSmad4 failed to reveal evidence for the formation of SUMO-1 chains at either K16 or K17 of the already conjugated SUMO-1. Second, a faint sumoylation band was observed for Smad4. Indeed, a known minor site of sumoylation on Smad4 is the non-consensus K113 residue. However, MS analysis did not provide any evidence for SUMO-1 conjugation at this position.