Removal of N-terminal methionine from recombinant proteins by engineered E. coli methionine aminopeptidase

Production of MetAP and target proteins

The gene encoding engineered MetAP was expressed itself or coexpressed with target protein in E. coli BL21(DE3). The MetAP was soluble and purified to electrophoretical homogeneity (Fig. 2A ▶, lane 2). The yields of MetAP varied with the type of coexpressed target protein (Fig. 2A ▶, lanes 5 and 7), for example, 50 mg purified protein was obtained per 1 L of culture in the absence of target protein. The yield of coexpressed target proteins varied with the property of the protein, for example, the amount of frog ribonucleases (onconase and RC-RNaseL1) is different as shown in Fig. 2B ▶, lanes 6 and 8).

Two residues involved in substrate specificity

Residue Gln 233, adjacent to two Co⁺⁺ ions and two catalytic residues, Glu 204 and Glu 235, in the substrate-binding pocket of E. coli MetAP (Fig. 1 ▶) was replaced by a small residue, for example, Gly, Thr, Asp, or Asn, and coexpressed with the onconase that begins with MQD. The percentage of Met removal was calculated by dividing the amount of proteins initiated with the penultimate residue by the amount of total proteins (i.e., proteins initiated with the penultimate residue plus unprocessed protein initiated with Met) after Edman degradation. The Met preceding Gln was removable by MetAP-Q233G (74% of Met removal) and MetAP-Q233T (88%). Furthermore, a second residue, Met 206, in the pocket (Fig. 1 ▶) was replaced by a small residue, for example, Gly, Thr, or Val. The Met of MQD-Onc was also removable by MetAP-M206G (85% of Met removal), MetAP-M206T (86%), and MetAP-M206V (61%). To make the MetAP more competent, we made a double mutation at residues Met 206 and Gln 233. For the substrate MQD-Onc, the double mutants, MetAP-M206G-Q233G (96%), MetAP-M206T-Q233G (91%), MetAP-M206T-Q233T (93%), and MetAP-M206V-Q233T (90%) variants, designated as MetAP-*GG, MetAP-*TG, MetAP-*TT, and MetAP-*VT, respectively, possess high in vivo catalytic activities (Table 1; data not shown). For other substrates, only MetAP-*TG exerted high catalytic activity for MLD-Onc (86% Met removal) and MID-Onc (22%; Table 1), whereas the MetAP-*GG did not, regardless of the apparent large pocket for substrate binding (data not shown).

A third residue involved in substrate specificity

A third residue, Tyr 168, in the substrate-binding pocket of MetAP (Fig. 1 ▶) was further replaced with a small residue, Gly, Ser, Thr, Val, Asp, or Glu. For the substrate MID-Onc, the efficiency of Met removal by the parental MetAP-*TG was 22%, but it increased markedly to 94% by Y168G mutation and to 68% by Y168T mutation, which were designated as MetAP-GTG and MetAP-TTG, respectively (Table 1). In contrast, for substrates with an acidic penultimate residue, for example, MDD-Onc or MED-Onc, the third Y168T mutation exhibited higher activity than that of the Y168G mutation (Table 1). The result suggests that the hydroxyl group of residue 168 of E. coli MetAP with a short side chain (Thr) may play a crucial role in the recognition of the acidic penultimate residue.

The substrate specificity of engineered MetAPs

The substrate specificities of MetAP variants were determined by using MXD-Onc as a substrate that possesses a different penultimate residue (Table 1). The Met preceding the moderate penultimate residue—that is, Asn (85%), Leu (86%), His (92%), Gln (90%), and Met (100%)—as well as a small and uncharged residue—that is, Gly (95%), Ala (99%), Pro (93%), Ser (99%), Thr (85%), and Val (80%)—were mostly removed by the MetAP-*TG. For proteins with a bulky penultimate residue, Ile and Trp, most of Met was removed by MetAP-GTG, at 94% and 78%, respectively. For protein with acidic penultimate residues, Asp and Glu, the Met was also removable by MetAP-TTG, at 80% and 58%, respectively. However, the Met preceding bulky basic penultimate residues, Lys and Arg, is still not removable by these variants.

Effect of antepenultimate residue on substrate specificity

Using MXD-Onc as a substrate, we demonstrated the importance of the penultimate residue on the substrate specificity of engineered MetAP in vivo (Table 1). However, we also found that the substrate specificity of MetAP was influenced by the antepenultimate residue. With MAX-Onc as a substrate, the removal of terminal Met was not significantly altered by the change of the antepenultimate residue, whereas both terminal Met and penultimate Ala were removable by the engineered MetAPs when the antepenultimate residue was small, that is, Gly or Ala (Table 2). With MLX-Onc or MQX-Onc as a substrate, the terminal Met and penultimate Leu/Gln residues were also removable by engineered MetAPs when the antepenultimate residue was small or moderate, such as Gly, Ala, Ser, Thr, Val, or Asn, but not bulky or charged, such as Asp, Glu, Phe, Trp, Lys, or Arg. Furthermore, both Met and Phe/Trp residues were removable by MetAP-GTG when the antepenultimate residue was Ala but not Asp (Tables 1, 2). With respect to wild-type MetAP, the N-terminal Met and penultimate Ala of MAA- or MAG-initiated onconase and RC-RNase 4 were also removable (Tables 2, 3).

The removal of N-terminal Met from the penultimate Ala/Leu/Gln residue by MetAP-*TG was reduced when the antepenultimate residue was Pro. In contrast, most of the newly exposed Leu/Gln residue preceding Pro was removed by MetAP-GTG, although only half of the N-terminal Met was removed (Table 2). This result suggests that the Pro residue at the antepenultimate site may exert a steric hindrance on the catalytic activity of the MetAP enzyme.

Applications in other recombinant proteins

To investigate the possible application of engineered MetAPs in the production of recombinant proteins, we used several soluble or insoluble proteins as substrates. The Met of insoluble MQD-initiated bullfrog ribonucleases (RC-RNase 3, RC-RNase 4, RC-RNase 6, and RC-RNase L1-N2D) was removed by MetAP-*TG, and both Met and Gln of insoluble MQN-initiated bullfrog ribonucleases (RC-RNase 2 and RC-RNase L1) were removed by the enzyme (Table 3). The terminal Met of MQD-initiated glutathione S-transferase (GST) of Schistosoma japonicum, either soluble or insoluble, was also removable by the MetAP-*TG. Both Met and Leu of the soluble MLA-initiated WW domain of CA150 protein, a transcription elongation regulator, were also removable (Table 3). To increase the competence of the engineered MetAP, we cloned the MetAP-*TG gene in the vector pET29b, which possesses kanamycin-resistance, and we cloned the gene of target protein, for example, onconase, the WW domain of CA150, or GST, in a separate expression vector pET22b, which possesses ampicillin resistance and is coexpressed in E. coli BL21(DE3). We found that the Met of these MQD- or MLD-initiated proteins was also removable, but with a slightly lower efficiency (10%–15% less; Table 3). This reduction may be due to lower copy number of the plasmid and the RNA transcript encoding the engineered MetAP.

From the eukaryotic curated (NP) protein sequence entries of the NCBI RefSeq database (dated January 7, 2003) using the SignalP program, we found that 37.4% of 5298 secretory proteins possesses a small residue, such as Gly, Ala, Cys, Pro, Ser, Thr, or Val, at N termini and 58.5% of 17,978 nonsecretory proteins possesses a small penultimate residue. The N-terminal Met of these recombinant proteins may be processed by wild-type MetAP. On the basis of the results we obtained by the coexpressed MetAPs, up to 84.5% of secretory proteins and 90.2% of nonsecretory proteins should be produced in authentic form.

Properties of recombinant ribonucleases

In vitro catalytic activities of MetAP variants

For oligopeptides ranging from tetra- to dodecapeptides with Ala at the penultimate site, the wild-type MetAP and MetAP-*TG possessed high catalytic activity, whereas the MetAP-*GG, MetAP-*TT, and MetAP-GTG variants did not (Table 4). For the oligopeptide with Gln at the penultimate residue, wild-type and most engineered MetAPs did not exhibit significant catalytic activities, whereas the MetAP-GTG exerted 43% catalytic activity for the dodeca-peptide MQDWLTFQKKHI in comparison with that for MADWLTFQKKHI, which is identical to the N-terminal α-helix of rM-Onc. The low in vitro catalytic activity of the purified MetAP for MQD-initiated oligopeptides suggests that some factor(s) may be required for the in vivo removal of Met from a bulky penultimate residue.

Construction of expression plasmids

The rpr (AF332139) gene encoding for the target protein, onconase, from R. pipiens was tagged with the NdeI and BamHI sites at the 5′ and 3′ ends, respectively, and inserted into expression vector pET22b (Novagen; Huang et al. 1998^; Liao et al. 2000^{, 2003}). The other target proteins were also cloned by the same method including RC-RNase L1 (AF288642.2), RC-RNase 2 (AF242553), RC-RNase 3 (AF242554), RC-RNase 4 (AF242555), RC-RNase6 (AF242556) from Rana catesbeiana, the WW domain of a transcription elongation regulator 1, CA150 (nt 1295–1717 of NM_006706), and glutathione S-transferase (M14654). The Escherichia coli methionine aminopeptidase gene (accession no. M 15106) or its mutant was fused with the T7 promoter by three steps of PCR and inserted into the same pET22b vector containing a target protein, as described earlier, through SacI and HindIII. The first PCR was performed using oligonucleotide 1 (5′-CGCG GAGCTCGATCCCGCGAAATTAATACG-3′) and oligonucleo-tide 2 (5′-CTTGATTGAGATAGCCATTATCTCCTTCTTAAAG TTAAACAAAATTATTTCTAGAGG-3′) as the 5′ and 3′ primers, respectively, and pET22b as the template. The second PCR was performed using the earlier-mentioned PCR product as the 5′ megaprimer and oligonucleotide 3 (5′-CCGGAAGCTTTTATTC GTCGTGCGAGATTATCG-3′) as the 3′ primer and E. coli MetAP gene as the template (Ben-Bassat et al. 1987; Huang et al. 1998). The third PCR was to amplify the second PCR product using oligonucleotide 1 and oligonucleotide 3 as primers. Therefore, the genes encoding target protein and E. coli MetAP have their own T7 RNA polymerase promoter, but their encoded mRNAs may be located in the same RNA molecule for cotranslation because only one transcription terminator exists downstream of the MetAP gene. Alternatively, the MetAP gene alone may be directly subcloned into vector pET29b through the SacI and HindIII sites. Residues Tyr 168, Met 206, and Gln 233 in the putative substrate-binding site of E. coli MetAP were mutated by site-directed mutagenesis on the basis of the crystal structure (Fig. 1 ▶, PDB code 3MAT) and amino acid sequence alignment of MetAPs (Ben-Bassat et al. 1987; Lowther et al. 1999; Walker and Bradshaw 1999). For the expression of secretory ribonucleases, the ribonuclease gene was cloned and expressed as described previously (Huang et al. 1998).

Expression and purification of MetAP and target proteins

The MetAP was expressed alone or coexpressed with a target protein, that is, onconase and RC-RNaseL1, in E. coli BL21 (DE3) in the presence of 0.5 mM isopropyl-β-D-thiogalactopyranoside at 37°C. After sonication, the MetAP in soluble fraction was dialyzed against buffer A (20 mM HEPES at pH 8.0, 50 mM KCl) and purified to homogeneity by phosphocellulose (Whatman P-11) and DEAE-cellulose (Whatman DE52) column chromatographies with a yield of 50 mg from 1 L of culture.

Most of the coexpressed target protein, for example, onconase or RC-RNase L1, existed in insoluble cell lysate. These proteins were denatured, renatured, and purified as described previously (Liao et al. 2003). The purified ribonucleases were stored in 20 mM sodium phosphate (pH 7.0) for Pyr formation at the N terminus. Alternatively, the ribonuclease secreted into culture medium through the pelB system was concentrated and purified to electrophoretical homogeneity with a yield of 1~5 mg per 1 L of culture (Huang et al. 1998).

Analysis of N termini of recombinant proteins

Proteins in the cell pellet or in the soluble fraction were separated by 13.3% SDS-PAGE and contact-transferred to ProBlott membrane (Applied Biosystems) in transfer buffer (10 mM Tris-HCl at pH 7.5, 50 mM NaCl, 2 mM EDTA, 0.5 mM 2-mercaptoethanol) overnight. The N-terminal five residues of the transferred protein were determined by Edman degradation. The percentage of Met removal was calculated by dividing the amount of protein initiated with the penultimate residue by the amount of total proteins (i.e., proteins initiated with the penultimate residue and unprocessed protein initiated with Met) after the first degradation cycle. The same calculation was made from five consecutive degradation cycles, for example, MQNWL for onconase. The efficiency of Met removal is reported as the mean values of at least three cycles. The mass spectrum analysis was carried out as described previously (Liao et al. 2003).

In vitro MetAP activity assay

The oligopeptide substrates used in this experiment were synthesized on the 432A Synergy peptide synthesizer (Applied Biosystems). The enzymatic activity of purified MetAP was assayed as described by Ben-Bassat et al. (1987). Briefly, the diluted MetAP enzyme in 10 μL was added to 90 μL of the substrate solution containing 4 mM oligopeptide, 0.1 M potassium phosphate buffer (pH 7.5), and 0.2 mM CoCl₂, incubated for 10 min at 37°C and stopped in boiling water for 2 min. After addition of 0.9 mL of color development mixture containing 0.2 mg of L-amino acid oxidase, 0.03 mg of horseradish peroxidase, and 0.2 mg of o-dianisidine in 0.1 M Tris-HCl (pH 7.4), the tubes were incubated for 10 min at 37°C and the optical density at 440 nm was recorded. One unit of activity is defined as 1 μmole of amino acids produced per min under the assay conditions and the pure Met was used as standard. The N-terminal residues of MetAP-treated oligopeptide were further verified by Edman degradation.

Ribonuclease activity assay

Ribonuclease activity was analyzed by zymogram assay on RNA-casting PAGE (Liao and Wang 1994). In addition, the ribonuclease activity of each purified ribonuclease was determined by the release of acid-soluble nucleotides from bakers’ yeast total RNA after ribonuclease digestion. One unit of enzyme activity was defined as the amount of enzyme producing one A₂₆₀ acid-soluble material for 15 min at 37°C (Huang et al. 1998).