High-throughput screening requires large amounts of active protein. Recombinant expression of USP18 in a bacterial system as well as in insect cells has been reported, however with very low yields [5
]. Until now, there has been no expression system available for production of sufficient amounts of recombinant USP18. Therefore, we aimed to establish a high-yield and easy-to-apply expression system for catalytically active USP18.
Expression trials using murine cDNA for USP18 cloned into pET15b or a pGEMEX vector were performed in E. coli
Rosetta(DE3). However, no expression of His6
-tagged USP18 could be observed in Western blots (Table ). We reasoned that some rare codons in the cDNA of the USP18 clone might obstruct expression and therefore switched to an expression construct with codons optimized for expression in E. coli
. In addition, we introduced a SUMO-tag at the N-terminus of USP18 as such a tag was reported to enhance expression levels of this protein in the baculovirus expression system [12
]. Sequence and ligation independent cloning (SLIC) was performed to generate the His6
-SUMO-USP18 construct in the pACE vector backbone [20
] (Figure A). Subsequently, E. coli
BL21(DE3) as well as E. coli
BL21(DE3)pLysS were used as host strains for test expressions. In contrast to the clone derived from mouse cDNA, expression of the SUMO-USP18 fusion protein could be detected in both E. coli
strains on Western blot with an anti-His6
-tag specific antibody. However, expression levels of the fusion protein were too low to be detected on Coomassie-stained SDS-PAGE gel (Table ).
Constructs tested for expression of different USP18 fusion proteins
Figure 1 (A) Generation of SUMO-USP18 and SUMO-TFAAA-USP18 expression vectors using sequence and ligation independent cloning (SLIC). The target vector pACE was linearized using primers NdeI-pACE-rev and XhoI-pACE-for. Recognition sites for the restriction enzymes (more ...)
Overproduction of soluble recombinant protein in E. coli
can be limited by the deprivation of host cell chaperones that are required for correct folding of the respective protein. Co-overexpression of E. coli
chaperones was reported previously to enhance solubility and yield of recombinant proteins [22
]. Recently, also successful expression of a fusion of the chaperone Trigger Factor with the protein of interest was reported using a cold shock expression system in E. coli
] (Takara, pCold TF plasmid). This system provides each translated protein its own chaperone. As the chaperone Trigger Factor (TF) is the first chaperone newly translated proteins encounter [24
] we fused this chaperone to the N-terminus of USP18. TF interacts with the bacterial ribosome and incorporates nascent polypeptide chains that emerge from the ribosomal exit tunnel. In this way, it provides a protective environment that facilitates folding [24
]. To assure interaction of USP18 with TF, which forms a large hydrophobic cradle, we introduced a long flexible linker consisting of six GSS repeats between USP18 and the chaperone (Figure B, C). Moreover, the long linker ensures that the folded USP18 is accessible for substrates and not sterically blocked by TF.
TF binds to the ribosome via the motif 43-GFRxGxxP-50 [26
]. Although TF binds with low affinity to the ribosome [28
], overexpression of TF might become a serious problem for protein synthesis in the expression host. In order to reduce binding of the TF-USP18 fusion protein to the ribosome and facilitate dissociation, residues G43, F44 and R45 of TF were exchanged to alanine (TFAAA
). These residues have been shown previously to be critical for TF-ribosome interaction [27
The resulting fusion protein consists of an N-terminal His6-tag, SUMO, Trigger FactorAAA, and USP18 (= TFAAA-USP18). TFAAA-USP18 in the pACE vector backbone was tested for expression in E. coli BL21 (DE3) and E. coli BL21 (DE3) pLysS. In contrast to SUMO-USP18, insertion of TF increased expression levels so that the fusion protein could be detected on Coomassie-stained SDS-PAGE gel (Table ). However, it did not represent the major fraction compared to endogenous bacterial proteins.
Therefore, we changed the vector backbone from pACE to pSUMO. This boosted expression of the fusion protein which now represented the major band on SDS gel when expression was performed at 37°C (Table and Figure A). However, these expression conditions resulted in poor solubility of the protein as demonstrated by Western Blot with a His6
-tag specific antibody (Figure B). Almost all recombinant protein was detected in the pellet fraction whereas only a weak band was detected in the soluble fraction. Lowering temperature is often reported to increase yield and solubility of expressed proteins [29
]. Test expressions at 25°C had no observable effect and resulted in insoluble protein (not shown). Decreasing further the expression temperature to 15°C yielded soluble TFAAA
-USP18 (Figure C). However, the drop in temperature caused also a severe decrease in protein expression (Table ).
Figure 2 Expression of TFAAA-USP18 in pSUMO vector backbone under different conditions(A) TFAAA-USP18 was expressed in E. coli BL21(DE3)pLysS at 37°C. Expression was verified by analyzing protein content directly after lysis on SDS-PAGE followed by Coomassie (more ...)
To achieve again high expression levels combined with high solubility of TFAAA-USP18 we changed to the stringent expression host strains E. coli Tuner(DE3) and E. coli Tuner(DE3)pLysS. Tuner strains are deficient in lactose permease (lacY) and thus allow uniform uptake of IPTG via diffusion. Whereas E. coli Tuner(DE3)pLysS only showed a weak expression of TFAAA-USP18, strong expression of soluble fusion protein was observed when the E. coli Tuner(DE3) strain was grown at 15°C (Table and Figure D). Therefore, these conditions were applied for large scale expression and purification. 2 liter expression cultures typically yielded 24 g of wet weight pellet. For purification of TFAAA-USP18, different IMAC columns were tested of which a cobalt IMAC column provided the best results. Using this column, pure TFAAA-USP18 was eluted allowing one-step purification without need of further purification steps (Figure ). A minor band running at lower molecular weight was observed when the sample was boiled only for a short time or without fresh DTT added. This band most likely represents protein containing an intramolecular disulfide bond formed during boiling. Typical yield was 10 mg pure protein out of 8 g wet weight pellet.
One-step purification of TFAAA-USP18. TFAAA-USP18 was bound to a Co-IMAC column and eluted with imidazole. Purity of the eluted fusion protein was visualized by SDS-PAGE with subsequent Coomassie staining.
Once expression and purification was established we checked whether the large scale preparations represent also catalytically active enzyme. Therefore, we tested isopeptidase activity of TFAAA-USP18 towards ISG15 modified cellular proteins (Figure A). High levels of ISGylated cellular protein were obtained using USP18 deficient mouse embryonic fibroblasts (MEFs) stimulated with IFN β. MEF cell lysates were incubated with and without TFAAA-USP18 and changes in ISGylation levels were monitored by Western blot with an ISG15-specific antibody. Incubation with TFAAA-USP18 drastically decreased the amount of ISGylated proteins, simultaneously the amount of free ISG15 increased demonstrating the ability of the TFAAA-USP18 to recognize and cleave ISG15 from cellular target proteins. To further evaluate enzymatic specificity, TFAAA-USP18 as well as a TFAAA-USP18 variant, where the catalytic cysteine is exchanged to alanine (TFAAA-USP18-C61A), were incubated with the suicide inhibitors ubiquitin vinyl sulfone (Ub-VS) and ISG15 vinyl sulfone (ISG15-VS), respectively. These suicide inhibitors form a covalent adduct upon reaction with the active site cysteine of ubiquitin-specific proteases. The reaction can be visualized as a shift in molecular mass on a Coomassie-stained gel. For TFAAA-USP18, covalent complex formation was detected with ISG15-VS whereas mutation of the catalytic cysteine to alanine resulted in complete loss of the interaction. Neither TFAAA-USP18 nor TFAAA-USP18-C61A showed cross-reactivity towards Ub-VS (Figure B). In summary, these experiments show that TFAAA-USP18 is catalytically active and underline its specificity towards ISG15.
Figure 4 Enzymatic activity of TFAAA-USP18 (A) Cell lysates of USP18 knockout mouse embryonic fibroblasts were stimulated with IFN β resulting in elevated ISGylation or left untreated. Cell lysates were incubated with and without TFAAA-USP18 for the indicated (more ...)
Screening for potential USP18 inhibitors requires a method that allows quantification of USP18 activity and is compatible with standard detection instruments. Therefore, we established assay conditions for TFAAA-USP18-mediated ISG15-AMC cleavage. Different amounts of the fusion protein were incubated with ISG15-AMC and cleavage was measured over a period of 30 minutes. The measured rate of ISG15-AMC cleavage was constant for more than 20 minutes and the rate increased linearly with enzyme concentration (Figure C). At the highest concentration of TFAAA-USP18, a slight decrease in the rate was observed after 25 minutes that is most likely due to a limitation of substrate and not caused by a decrease in enzyme activity. TF itself has no isopeptidase catalytic activity and does not interfere with the assay. These results demonstrate that TFAAA-USP18 is very well suited for kinetic analysis and the assay presented here can be easily adapted for high-throughput screening for specific inhibitors of USP18.