A His6-SUMO-eXact tag for producing human prepro-urocortin 2 in Escherichia coli for raising monoclonal antibodies.

This is a first report of recombinant production of human prepro-Urocortin 2 in Escherichia coli by N-terminal fusion with a triple His6-SUMO-eXact tag and its subsequent use as an antigen for the production and screening of very high affinity monoclonal antibodies. The rationale for this combinatorial construct is that the His tag allows first step protein purification of insoluble and soluble proteins, the SUMO tag enhances protein expression level and solubility, while the eXact tag facilitates anion-triggered on-column cleavage of the triple tag to recover pure native proteins in a simple two-step protein purification procedure. Compared with an eXact fusion alone, the presence of the SUMO moiety enhanced overall expression levels by 4 to 10 fold but not the solubility of the highly basic prepro-Urocortin 2. Insoluble SUMO-eXact-preproUCN2 was purified in milligram quantities by denaturing IMAC and solubilized in native phosphate buffer by on-column refolding or step-wise dialysis. Only a small fraction of this solubilized protein was able to bind onto the eXact™ affinity column and cleaved by NaF treatment. To test whether binding and cleavage failure was due to improperly refolded SUMO-eXact-preproUCN2 or to the presence of N- and C-terminal sequences flanking the eXact moiety, we created a SUMO-eXact-thioredoxin construct which was overexpressed mainly in the soluble form. This protein bound to and was cleaved efficiently on the eXact™ column to yield native thioredoxin. Solubilized SUMO-eXact-preproUCN2 was used successfully to generate two high affinity mouse monoclonal antibodies (KD~10⁻¹° and 10⁻¹¹ M) specific to the pro-region of Urocortin 2.

An SRLLR motif downstream of the scissile bond enhances enterokinase cleavage efficiency.

In a previous paper, we reported more efficient enterokinase cleavage at a C-terminal non-target LKGDR(201) site compared with an internally sited canonical recognition site, DDDDK(156). When this non-target site was placed internally to replace DDDDK(156) between the thioredoxin moiety and mouse NT-proCNP(1-50), this site was poorly processed leading us to conclude that efficient processing at LKGDR(201) in the first instance was due to its accessibility at the C-terminus of the fusion protein. Subsequently, we reasoned that treatment of thioredoxin-fused NT-proCNP(1-81) would allow us to retrieve full-length NT-proCNP(1-81) without undue processing at the LKGDR(201) site since this non-target site would now be located internally about 36 residues away from the C-terminus and hence not be hydrolyzed efficiently. Surprisingly, ESI-MS data showed that the LKGDR site in thioredoxin-fused human NT-proCNP(1-81) was still very efficiently cleaved and revealed a new but slow hydrolysis site with the sequence RVDTK/SRAAW to yield a peptide consistent with NT-proCNP(58-81). The evidence obtained from these experiments led us to postulate that efficient cleavage at the non-target LKGDR(201) site was not merely influenced by steric constraints but also by the sequence context downstream of the scissile bond. Hence, we constructed variants of thioredoxin-mouse NT-proCNP(1-50) where SRLLR residues (i.e. those immediately downstream from the LKGDR(201) site in NT-proCNP(1-50)) were systematically added one at a time downstream of the internal DDDDK(156) site. To evaluate the relative effects of site accessibility and downstream sequence context on the efficiency of enterokinase cleavage, we have also replaced the native LKGDR(201) sequence with DDDDK(201). Our results showed that incremental addition of SRLLR residues led to a steady increase in the rate of hydrolysis at DDDDK(156). Further variants comprising DDDDK(156)SS, DDDDK(156)SD and DDDDK(156)RR showed that the minimal critical determinants for enhanced enterokinase cleavage are serine in the P1' position followed by a serine or a basic residue, lysine or arginine, in the P2' position. Our data provided conclusive evidence that the influence of downstream sequences on recombinant light chain enterokinase activity was greater than accessibility of the target site at the terminus region of the protein. We further showed that the catalytic efficiency of the native holoenzyme was influenced primarily by residues on the N-terminal side of the scissile bond while being neutral to residues on the C-terminal side. Finally, we found that cleavage of all nine fusion proteins reflects accurate hydrolysis at the DDDDK(156) and DDDDK(201) sites when recombinant light chain enterokinase was used while non-specific processing at secondary sites were observed when these fusion proteins were treated with the native holoenzyme.