SUMO fusions and SUMO-specific protease for efficient expression and purification of proteins

Among the ubiquitin-like proteins tested as N-terminal fusion partners of His6-GFP in E. coli, ubiquitin, SUMO and Urm1 gave the best enhancement of expression. GFP fused to these three partners accumulates to up to 60% of total E. coli protein, remains soluble and is properly folded as judged by a fluorescence assay.

Fusing His6-SUMO to the N terminus of the difficult-to-express matrix metalloproteinase MMP13 (residues 20–274 and the catalytic domain 104–274) changes its behavior in E. coli. His6-MMP13 is detected in lysates but is always insoluble, whereas the His6-SUMO-MMP13 fusions are mostly soluble, showing that SUMO acts as a chaperone-like solubility enhancer for MMP13.

The expression-enhancing effect of SUMO in E. coli depends on SUMO being N-terminal to its partner. The reverse configuration, GFP-SUMO, does not show enhanced expression, consistent with the proposal that the tightly folded, fast-folding SUMO domain primes folding of the C-terminally fused protein.

The catalytic domain of yeast SUMO protease 1 (Ulp1 residues 403–621, called ULP) cleaves His6-SUMO-X-GFP fusions at the SUMO-GFP junction whichever of the 19 amino acids other than methionine is placed in the position after the SUMO Gly-Gly motif, except proline, which blocks cleavage. ULP therefore releases target proteins with almost any desired N-terminal residue, because it recognizes the SUMO structure and not the sequence of the partner protein.

Cleavage of the SUMO junction by ULP is not equally fast for all N-terminal residues. With a limiting amount of enzyme, His6-SUMO-MMP13 (20–274), which has Leu after SUMO, is only about 60% cleaved while the Tyr-bearing His6-SUMO-MMP13 (104–274) is completely cleaved, and His6-SUMO-LXR183 is not completely cleaved because three consecutive glutamic acid residues follow its cleavage site.

ULP cleaves the SUMO-M-GFP substrate in buffers from pH 6.0 to 10.0, with optimal cleavage between pH 8.0 and 9.0. It cleaves 100% of the substrate at temperatures as low as 4 °C and is most efficient between 22 °C and 37 °C, and doubling the enzyme amount at 4 °C gives results similar to those at 22 °C and 37 °C.

ULP activity on SUMO-M-GFP is unaffected by 20 mM DTT or beta-mercaptoethanol, 300 mM imidazole, 20 mM reduced glutathione, 20 mM maltose, 20% glycerol, 20% ethylene glycol, 20% sucrose, 10% ethanol or 150 mM NaCl. Higher NaCl reduces cleavage to 60% at 500 mM and 30% at 1 M, 1 M and 2 M urea leave 100% and 95% of cleavage while 3 M urea reduces it to 5%, and Gu-HCl reduces it to 60% at 500 mM and 0% at 1 M.

ULP cleaves SUMO fusions of diverse structure without producing unexpected hydrolysis products. His6-SUMO fusions of the liver X receptor ligand-binding domain, beta-galactosidase, the MAPKAPK2 serine/threonine kinase and maltose-binding protein are completely cleaved in standard buffer or PBS, and no nonspecific cleavage is seen even when ULP is deliberately raised to a 1:1 enzyme-to-substrate ratio.

ULP cleaves SUMO fusions efficiently at very low enzyme-to-substrate ratios. Substrates from 4 to 200 kDa are consistently cleaved at a typical 1:5000 enzyme-to-substrate ratio, and the enzyme remains efficient at 1:10,000.

A SUMO-GUS fusion is cleaved inefficiently by ULP in standard buffer, but cleavage is complete when 3 M urea is added to the reaction, consistent with denaturant loosening the SUMO-GUS structure and exposing the cleavage site to ULP.

ULP remains active in 2 M urea, so SUMO fusions recovered from insoluble aggregates can be solubilized in 8 M urea, diluted to 2 M urea and cleaved with ULP, followed by dialysis and Ni-NTA subtraction of His6-SUMO. Attaching SUMO increases production of proteins that are prone to inclusion bodies and facilitates their refolding.

A three-step capture, cleavage and subtraction scheme purifies SUMO-tagged proteins. After His6-SUMO fusions are captured on Ni-NTA and cleaved with ULP, passing the digest over Ni-NTA again retains the cleaved His6-SUMO and the C-terminally His-tagged ULP and removes metal-binding contaminants, which leaves essentially pure MMP13 in the flow-through.

SUMO can be combined with other affinity tags, and MBP-SUMO and GST-SUMO fusions have been used. In a double-tagged MBP-His6-SUMO fusion, capture on amylose resin followed by subtraction of the tag on Ni-NTA omits the desalting or dialysis step.

A serine/threonine kinase expressed as a GST fusion in E. coli is mostly insoluble, whereas the SUMO fusion solubilizes 95% of the protein and it stays soluble after cleavage with ULP. MMP13, insoluble when expressed unfused, likewise remains soluble after SUMO is cleaved off, consistent with SUMO helping the partner protein fold correctly.

Over 60 proteins have been expressed as SUMO fusions in E. coli in one laboratory, with SUMO fusion generally giving a 10–50-fold increase in the quantity and quality of expressed protein.

In the pET24 His6-SUMO vector, a Bsa I type II restriction site placed directly downstream of the SUMO coding sequence produces a four-base sticky end that is part of the SUMO 3′ sequence. Only these four bases are needed for correct ligation, so the downstream sequence of the insert, and therefore the N-terminal residue after SUMO, is specified freely by the user.

Recombinant ULP (Ulp1 residues 403–621 with a C-terminal His tag, about 31 kDa by SDS-PAGE) is easy to produce in E. coli Rosetta(DE3)pLysS. Purification on Ni-NTA from 125 ml of culture at OD600 of 1.2 yields ULP that is 99% pure, giving 10 ml of stock at 425 units/µl in 50% glycerol, where one unit cleaves 15 µg of SUMO-M-GFP at 25 °C in 1 h in 20 mM Tris-HCl pH 8.0 with 5 mM 2-mercaptoethanol. Concentrated stocks stored for 6 months show no detectable loss of activity.