Forced Ambiguity of the Leucine Codons for Multiple-Site-Specific Incorporation of a Noncanonical Amino Acid

Multiple-site-specific incorporation of a noncanonical amino acid into a recombinant protein would be a very useful technique to generate multiple chemical handles for bioconjugation and multivalent binding sites for the enhanced interaction. Previously combination of a mutant yeast phenylalanyl-tRNA synthetase variant and the yeast phenylalanyl-tRNA containing the AAA anticodon was used to incorporate a noncanonical amino acid into multiple UUU phenylalanine (Phe) codons in a site-specific manner. However, due to the less selective codon recognition of the AAA anticodon, there was significant misincorporation of a noncanonical amino acid into unwanted UUC Phe codons. To enhance codon selectivity, we explored degenerate leucine (Leu) codons instead of Phe degenerate codons. Combined use of the mutant yeast phenylalanyl-tRNA containing the CAA anticodon and the yPheRS_naph variant allowed incorporation of a phenylalanine analog, 2-naphthylalanine, into murine dihydrofolate reductase in response to multiple UUG Leu codons, but not to other Leu codon sites. Despite the moderate UUG codon occupancy by 2-naphthylalaine, these results successfully demonstrated that the concept of forced ambiguity of the genetic code can be achieved for the Leu codons, available for multiple-site-specific incorporation.


Introduction
Site-specific incorporation of a noncanonical amino acid into a protein has been widely used to provide unique physical, chemical, or biological properties to a protein [1][2][3][4][5][6][7][8][9][10]. In most cases, a noncanonical amino acid was introduced into a single site of a target protein. An amber codon was most commonly used as an incorporation site, though other stop codons and four-base codons were also used [3,[11][12][13][14]. In order to expand the utility of site-specific incorporation of a noncanonical amino acid, researchers attempted to achieve site-specific incorporation at multiple sites [15][16][17]. A pre-requisite to achieve noncanonical amino acid incorporation at multiple sites is to develop a new codon that can be reassigned to a noncanonical amino acid (s). Combinations of stop codons and four-base codons have been successfully used to encode two different noncanonical amino acids, resulting in a protein with two different noncanonical amino acids at two programmed sites [15,18]. In this suppression strategy, incorporation of each noncanonical amino acid requires the expression of a corresponding heterologous orthogonal pair in host cells. Due to the limited number of orthogonal pairs available, it is challenging to incorporate noncanonical amino acids into more than two sites with the suppression strategy.
In order to circumvent this limitation, several groups achieved site-specific incorporation of a single noncanonical amino acid into multiple sites [19,20]. We can easily imagine many situations where site-specific incorporation of a single amino acid into multiple sites is required. For instance, conjugation of multiple poly(ethylene glycol) molecules to reactive noncanonical amino acids of a therapeutic protein is expected to be more effective in enhancing the serum half-life than single poly(ethylene glycol) molecule conjugation. Recently, incorporation of pazidophenylalanine into several sites was achieved using engineered E. coli cells of which all amber codons in genomic DNA were mutated into other stop codons and the release factor 1 gene was removed from genomic DNA to enhance amber suppression efficiency [19]. In vivo evolution of aaRS in the E. coli genome further enhanced the incorporation efficiency of Phe analogs into multiple sites of a target protein [21]. Alternatively, the degeneracy of the genetic code was explored for multiple-site-specific noncanonical amino acid incorporation [20]. In nature, several canonical amino acids, such as valine, leucine, isoleucine, methionine, phenylalanine, tyrosine, and tryptophan, are expected to be late addition to the genetic code [22]. Evolutionary analyses revealed that inclusion of methionine and tryptophan required the complete breaking of the codon wobble degeneracy [23]. Very recently it was reported that all tryptophan codons from E. coli genome are efficiently reassigned to a noncanonical amino acid, L-β-(thieno[3,2-b]pyrrolyl)alanine [24]. These are all very good indications that the exploitation of the degeneracy in the genetic code could represent a promising route for the expansion of amino acids available for protein biosynthesis [25]. By breaking the degeneracy of the Phe codons, a phenylalanine analog, 2-naphthylalanine (2Nal), was incorporated into multiple UUU phenylalanine (Phe) codons [20]. In order to reassign a UUU codon to 2Nal, the anticodon of yeast phenylalanyl-tRNA was mutated from CUA to AAA to generate ytRNA Phe AAA . A mutant yeast phenylalanyl-tRNA synthetase (ytRNA_T415G) was co-expressed with ytRNA-Phe AAA in E. coli cells to incorporate 2Nal into a target protein. Since this latter method uses a sense codon, the number of incorporation sites is not restricted [20]. Since UUU codons were not completely occupied by 2Nal, strictly speaking, the degeneracy of the Phe codons was not completely broken, but forced ambiguity of the Phe codons was achieved. Later, a yPheRS variant with a higher specificity toward 2Nal (yPheRS_naph) was selected from high-throughput screening of yPheRS libraries [26]. However, application of this technique has been limited, partly because 2Nal was misincorporated at the unwanted sites (UUC Phe codons) due to the less selective codon recognition of the AAA anticodon of ytRNA Phe AAA [20]. Misincorporation of a noncanonical amino acid at unwanted sites might cause severe perturbation or loss of native properties of a target protein [27,28].
Due to the poor discrimination of UUU codon from UUC codon by the AAA anticodon of ytRNA Phe AAA , we explored degenerate leucine (Leu) codons for noncanonical amino acid incorporation in this study. Several considerations recommend degenerate Leu codons. First, Leu is encoded as six codons: UUA, UUG, CUA, CUG, CUU, and CUC. Discrimination of UUG from CUN (N = A/U/G/C) codons should be highly efficient due to discrimination at the first position in the codon [29]. Second, our existing yeast orthogonal pair should be readily adapted to the incorporation of Phe analogs in response to UUG codons. In practical terms, generalization of the concept of forced ambiguity of the genetic code is limited by the availability of orthogonal pairs. Third, the modified CAA anticodon would more efficiently discriminate UUG from UUA. According to the wobble rules, C in the first position of the anticodon recognizes only G in the third position of the codon [29]. In this study, we first quantitatively evaluated the misincorporation level of 2Nal at unwanted sites (UUC codons) when the UUU Phe codon was reassigned to 2Nal. Then, we showed that an engineered ytR-NA Phe CAA containing a modified CAA anticodon could completely discriminate UUG from the remaining five Leu codons and achieve incorporation of 2Nal at multiple programmed sites in recombinant proteins.

Misincorporation of 2Nal at unwanted UUC Phe codons
In order to evaluate the misincorporation level of 2Nal at UUC codons, we used a green fluorescent protein variant with Phe codons encoded by only the UUC codon (GFP6) by mutating all UUU Phe codons to UUC codons [26]. It was previously reported that 2Nal incorporation into multiple sites of GFP led to a significant loss of fluorescence due to the structural perturbation of GFP [26]. Since GFP6 does not have any 2Nal incorporation sites (UUU codon), little or no change in fluorescence intensity was expected, assuming 2Nal is not incorporated into UUC codons. However, in the presence of 3 mM 2Nal and 5 μM Phe, the mean fluorescence intensity of cells expressing GFP6 as well as ytRNA Phe AAA /yPheRS_naph decreased almost 10-fold compared to that of cells in the absence of 2Nal (Fig 1A and 1B). In the presence of 3 mM 2Nal and either 2.5 or 5.0 μM Phe, the occupancy of UUU and UUC codons by 2Nal was evaluated using liquid chromatography-tandem mass spectrometric (LC-MS/MS) analysis of tryptic digests of a mDHFR variant expressed in E. coli cells co-expressing ytRNA Phe AAA / yPheRS_naph (Fig 1C and 1D). At 5 μM Phe and 3 mM 2Nal, about 50% of the UUU codon was occupied by 2Nal, but 6% of the UUC codon was also occupied by 2Nal. Therefore, the reduction in cellular fluorescence of GFP6 expressed in the presence of 3 mM 2Nal (Fig 1B) could be attributed to misincorporation of 2Nal at multiple UUC codon sites.
We reasoned that misincorporation of 2Nal at the UUC codon resulted from the recognition of UUC codons by the AAA anticodon of ytRNA Phe AAA . According to Crick's wobble rule proposed in 1966 [29], the base A in the first position of the anticodon can recognize only the base U in the third position of the codon. Therefore, the UUC codon should not be recognized by the AAA anticodon. The discrepancy between the experimental results and Crick's wobble rule may be explained by the expanded wobble rule proposed by Lim and Curran in 2001 [30]. The expanded wobble rule is based on new experimental findings [30][31][32][33][34][35][36][37] and stereochemical modeling [38][39][40][41] of codon-anticodon interactions. According to the expanded wobble rule (Fig 2A), A in the first position of the anticodon can recognize all four bases in the third position in the codon. The base A in the first position of the anticodon favors bases in the order U > C > G > A (Fig 2B), consistent with the codon-biased incorporation of 2Nal observed in this work.

Incorporation of 2Nal into UUG Leu codons
According to the expanded wobble rules, the base C in the first position of the anticodon will recognize only the base G in the third position of the codon (Fig 2A). Therefore, we hypothesized that ytRNA Phe CAA (containing the modified CAA anticodon) would selectively recognize UUG codons (Figs 2C and 3). Endogenous E. coli leucyl-tRNAs recognize all six Leu codons. One E. coli leucyl-tRNA containing the UAA anticodon (E. coli tRNA Leu UAA ) recognizes UUA codons via Watson-Crick base pairing, but not UUG codons (Fig 3). In order to test this hypothesis of forced ambiguity of the Leu codons, we mutated the AAA anticodon of ytRNA Phe AAA [20] into CAA anticodon by PCR mutagenesis using pREP4_ytRNA Phe AAA as a template. Then, mDHFR was expressed in MP [pQE16_mDHFR2_lacI_yPheRS_naph/pRE-P4_ytRNA Phe CAA ] cells. mDHFR contains twenty Leu codons, of which six are UUG, two are UUA and twelve are CUN (N = A/T/G/C). The expression level of mDHFR was 2.9 mg/L.
Occupancy of each Leu codon by various amino acids was determined by LC-MS analysis of tryptic digests of mDHFR expressed with and without 2Nal. We focused on four peptides. Peptide 1 (residues 165-180; L CUU L CUC PEYPGVL CUC SEVQEEK) contains three Leu residues, encoded as CUU and CUC codons. Peptide 2 (residues 54-61; QNL CUG VIMGR) contains a Leu residue, encoded as a CUG codon. Peptide 3 (residues 62-70; L CUU IEQPEL UUG ASK) contains two Leu residues, encoded as CUU and UUG codons. Peptide 4 (residues 99-105; SL UUG DDAL UUA R) contains two Leu residues, encoded as UUG and UUA codons. 2Nal was not detected at any CUN codon in Peptide 1 and 2 (Fig 4A-4D). However, 50% of the UUG codons in Peptide 3 and 4 were occupied by 2Nal (Fig 4E-4H). In order to determine UUA codon occupancy by 2Nal, Peptide 4 UUA was tested. Peptide 4 UUA is the same as Peptide 4 except both Leu residues are encoded as UUA codons. Since the Peptide 4 UUA variant containing 2Nal was not detected, we concluded that 2Nal incorporation is highly specific to the UUG codon.
However, this strategy achieved only moderate UUG codon occupancy (~50%) by 2Nal ( Fig  4E-4H), because UUG codon was partly occupied by Leu. Since E. coli leucyl-tRNA containing CAA anticodon (E. coli tRNA Leu CAA ) can also recognize UUG codons (Fig 3), 2Nal-charged ytRNA Phe CAA should competes against Leu-charged E. coli tNRA Leu CAA . Considering that the efficiency of amber codon suppression has been greatly improved in the past decade since its development, the occupancy of UUG codon by 2Nal is expected to increase in the future. For instance, an E. coli mutant deficient of E. coli tRNA Leu CAA was previously reported [42]. If this mutant is further developed as an expression host for 2Nal incorporation, UUG codon occupancy by Leu would be eliminated, resulting in high fidelity incorporation of 2Nal at UUG codons. Furthermore, even partial incorporation of a noncanonical amino acid into programmed multiple sites would be useful for many applications, in particular, engineering protein-based biomaterials.

Eliminated misincorporation of 2Nal into unwanted Leu codons
Next, we evaluated misincorporation of 2Nal at unwanted Leu codons using a GFP variant. We already showed that misincorporation of 2Nal at unwanted UUC Phe codons in GFP6 led to a 10-fold reduction in fluorescence of cells, even though there are no UUU codons in GFP6 ( Fig  1A and 1B). Similar to GFP6, we used GFP3 containing twenty-three Leu codons, of which there were no UUG, four UUA, and nineteen CUN (N = A/T/G/C) codons. In order to investigate the effect of misincorporation of 2Nal at unwanted sites (Leu codons other than UUG), GFP3 was expressed in the E. coli strain MP [pQE9_GFP3_lacI_yPheRS_naph/pREP4_ytRNA Phe CAA ]. The fluorescence intensities of cells expressing GFP3 without 2Nal or with 2Nal were compared ( Fig  5A and 5B). There was no detectable difference in the fluorescence of cells prepared under these two conditions, implying the absence of significant misincorporation of 2Nal at Leu codons other than UUG.
Codon occupancy depending on the anticodon of ytRNA Phe As the first base in the anticodon of ytRNA Phe changed from A to C, 2Nal assignment was accordingly changed from UUU to UUG. As an extension, we compared the UUU or UAG codon occupancy according to the anticodon of ytRNA Phe . First, UUU codon occupancy by tRNA Phe containing AAA (ytRNA Phe AAA ) was obtained by N-terminal sequencing of purified GFP6 variant containing a UUU Phe codon at position 2 (GFP6 (2UUU)). DHF [pQE9_GF-P6_lacI_yPheRS_naph/pREP4_ytRNA Phe AAA ] cells were induced to express GFP6 (2UUU) in MM18_FW medium supplemented with 50 μM Trp and 3 mM 2Nal, 80% and 20% of the  L CUC PEYPGVL CUC SEVQEEK) contains three Leu residues encoded as CUU and CUC codons. Peptide 2 (residues 54-61; QNL CUG VIMGR) contains a Leu residue encoded as CUG codon. Peptide 3 (residues 62-70; L CUU IEQPEL UUG ASK) contains two Leu residues encoded as CUU and UUG codons. Peptide 4 (residues 99-105; SL UUG DDAL UUA R) contains two Leu residues encoded as UUG and UUA codons. Peptide 4 UUA is the same as Peptide 4 except both Leu residues are encoded as UUA codon. Peptide 1; 2; 3; 4; 4 UUA variants containing Leu and 2Nal were designated 1L and 1Z; 2L and 2Z; 3L and 3Z; 4L and 4Z; 4 UUA L and 4 UUA Z, respectively. These peptides were separated by LC and detected by MS. Unmodified mDHFR was synthesized in the absence of 2Nal in a Phe/Leu auxotrophic expression host (A, C, E, and UUU codon at the 2 nd position of GFP6 (UUU) were decoded as 2Nal and Phe, respectively, but Trp was not detected at this position (Table 1).
With an appropriate tRNA, the selective yPheRS_naph variant can be used for residue-and single-site-specific incorporation of 2Nal into proteins. In order to realize residue-specific incorporation of 2Nal, DHF [pQE9_GFP6 (2UUU)_lacI_yPheRS_naph/pREP4_ytRNA Phe -GAA ] expression hosts were induced to express GFP6 (2UUU) in minimal medium supplemented with 25 mg/mL 18 amino acids without Phe and Trp, 50 μM Phe, 50 μM Trp, and 3 mM 2Nal. N-terminal sequencing of the purified GFP6 (2UUU) showed that 92% of position 2 was occupied by 2Nal (Table 1), slightly higher than the 80% occupancy achieved by multiplesite-specific incorporation. The enhanced 2Nal incorporation may be a consequence of the known 12-fold higher aminoacylation rate for ytRNA Phe GAA by yPheRS as compared to ytR-NA Phe AAA [43]. Site-specific incorporation of 2Nal into mDHFR_38Am, the mDHFR variant containing an amber codon at the 38 th position, was achieved by AFWK [pQE16_mDH-FR_38Am_yPheRS_naph/pREP4_ytRNA Phe CUA_UG ] in minimal medium supplemented with 25 μg/mL 17 amino acids (MM17_FWK), 50 μM Phe, 50 μM Trp, 50 μM Lys, and 3 mM 2Nal. The pQE16_mDHFR_38Am_yPheRS_naph plasmid was generated by replacing a yPheR-S_T415A gene in pQE16_mDHFR_38Am_yPheRS_T415A plasmid [44] with a yPheRS_naph  Table 1). Neither Trp nor Phe was detected, confirming the high selectivity of yPheRS_naph toward 2Nal.

Conclusions
In this study, we developed a strategy to incorporate a noncanonical amino acid into multiple sites in a site-specific manner based on forced ambiguity of the Leu codons. Misincorporation of 2Nal at unwanted sites resulting from the less selective codon recognition of the AAA anticodon of ytRNA Phe was overcome by use of the more codon-selective ytRNA Phe CAA . The CAA anticodon of ytRNA Phe CAA completely discriminates UUG codon from the other five Leu codons. When both yPheRS_naph and ytRNA Phe CAA were overexpressed in E. coli expression hosts, 50% of UUG codon sites were occupied by 2Nal, but no other Leu codon sites were occupied by 2Nal. Combined use of yPheRS_naph and the codon-selective ytRNA Phe CAA realized multiple-site-specific incorporation of 2Nal into proteins. Furthermore, for the first time, these results successfully demonstrated that the concept of forced ambiguity of the genetic code is not limited to degenerate Phe codons, but can be generalized to other degenerate codons. Although the incorporation level of 2Nal at UUG codons was moderate (about 50%) due to UUG codon recognition by endogenous E. coli tRNA Leu CAA , we are working to improve the The second position in the amino acid sequence of GFP6. b UUU and UAG codon occupancy was determined by N-terminal protein sequencing and MALDI-TOF MS analysis, respectively. c Not detected.  level of incorporation of 2Nal at programmed UUG sites using an E. coli mutant deficient of E. coli tRNA Leu CAA . Since the technique and strategy described here are very general, they would be applicable to incorporation of another noncanonical amino acid into a target protein when an appropriate orthogonal pair of the noncanonical amino acid is available.

Materials
Restriction enzymes were purchased from New England Biolabs (Beverly, MA, USA). Quikchange mutagenesis kits were purchased from Stratagene (La Jolla, CA, USA). Plasmid pREP4 and nickel-nitrilotriacetic acid affinity columns and were purchased from Qiagen (Valencia, CA, USA). 2Nal was obtained from Chem-Impex (Wood Dale, IL, USA). DNA primers were obtained from Operon Technologies (Huntsville, AL, USA) and Integrated DNA Technologies (Coralville, IA, USA). The reagents were purchased from commercial suppliers and used without further purification unless otherwise indicated.

Preparation of E. coli expression hosts
A Phe/Trp double auxotrophic strain (AFW) and Phe/Trp/Lys triple auxotrophic strain (AFWK) were previously reported [44]. The Phe/Leu double auxotrophic strain, MPC390 (leuB6(Am), PheA18::Tn10), was supplied from the E. coli Genetic Stock Center (CGSC) at Yale University. A Phe auxotrophic derivative of DH10B (Stratagene) E. coli strain was generated by chemical mutagenesis [45] and replica plating. DH10B E. coli mutants were subjected to replica plating on minimal medium agar plates containing either all twenty amino acids or nineteen amino acids without Phe. An E. coli mutant that could not grow on minimal medium agar plates without Phe was selected as a Phe auxotrophic derivative of DH10B, designated DHF.

Construction of Plasmids and Expression Hosts for Incorporation of 2Nal at Phe Codons
Construction of several GFP variants used in this study was previously reported [26]. An EGFP variant (GFP3) [46] has excitation maximum at 488 nm suitable for FACS analysis. A GFP variant (GFP6) was constructed by replacing all UUC Phe codons and one CUG Leu codon (at position 64) with UUU codons. A GFP variant (GFP3_WC) contains 12 Phe residues encoded as only UUC codons. In order to determine Phe codon occupancy by various amino acids, the AGA (Arg) codon in the second position of GFP6 was replaced by UUU or UUC Phe codon by PCR mutagenesis to generate GFP6(2UUU) or GFP6(2UUC), respectively [26]. Then, various E. coli host cells expressing either mDHFR or GFP variant were prepared using plasmids commercially available (Qiagen) or constructed previously [26]. Both pQE16_mDHFR_yPheRS (T415G) and pQE16_mDHFR_yPheRS naph were co-transformed with pREP4_ytRNA

Construction of Plasmids and Expression Hosts for Incorporation of 2Nal at Leu Codons
The AAA anticodon of ytRNA Phe AAA was mutated to CAA by PCR mutagenesis using pRE-P4_ytRNA Phe _AAA as a template to yield pREP4_ytRNA Phe CAA . The expression cassette of mDHFR was excised from pQE16 (Qiagen) by digestion with AatII and NheI and inserted into pQE9_GFP6_lacI_yPheRS_naph [26] between the AatII and NheI sites to generate pQE16_ mDHFR_lacI_yPheRS_naph. In order to increase the number of Leu residues encoded as UUG, UUC and UUU Phe codons in position 38 and 95 of mDHFR were changed to UUG by PCR mutagenesis reactions using pQE16_mDHFR_lacI_yPheRS_naph as a template to generate pQE16_mDHFR2_lacI_yPheRS_naph. PCR mutagenesis reaction was performed to mutate UUG to UUA at position 100 of mDHFR2 to yield pQE16_mDHFR2 (100UUA)_lacI_yPheRS_naph.  The cells were resuspended in fresh M9 minimal medium supplemented with 18 amino acids (25 μg/mL), and the indicated concentrations of Phe, Trp, and 2Nal. The GFP expression was induced by addition of 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG). After induced at 30°C for 4 hours, cells were harvested and either kept at -80°C or subjected to fluorescence measurement according to the procedures described earlier [26]. Whole cell lysates were analyzed by SDS-PAGE. Due to slow growth of DHF and MPC390 expression hosts co-transformed with pQE plasmid variants and pREP4 plasmid variants, transformants were grown in 2xYT medium to prepare glycerol stocks first. Then glycerol stocks were inoculated into minimal medium supplemented with 20 amino acids (at 25 mg/L) and incubated overnight at 37°C. The remaining steps were similar to those for AF and AFWK expression hosts.

Flow cytometric analysis
When OD 600 of DHF or MP900 cells expressing a GFP variant reached 0.6, the cells were washed twice with 0.9% NaCl solution. Then, the cells were resuspended with 20 mL of minimal medium supplemented with an appropriate amount of amino acids. The expression of a GFP variant was induced with 1 mM IPTG. After 3 hrs, 1 mL of the culture was collected, and washed twice with 0.5 mL of PBS (pH 7.4). 100 μL of cells were diluted with 3 mL of distilled water. Fluorescence intensities of the cells were analyzed by a MoFlo cell sorter. At least 20,000 events were collected in each measurement. Data were analyzed with Summit software (DakoCytomation).

Quantitative Analysis of Codon Occupancy
Quantitative analysis of codon occupancy was performed by either N-terminal protein sequencing or LC-MS analysis of tryptic digests. The GFP6 (2UUU) and GFP6 (UUC) variants were expressed in minimal medium and purified by Ni-NTA affinity chromatography according to the manufacturer's protocol (Qiagen) under denaturing conditions. The purified GFP variants were subjected to N-terminal protein sequencing using a 492 cLC Procise 1 protein micro-sequencer (Applied Biosystems, Foster City, CA). Occupancy of Phe codons in mDHFR and Leu codons in GFP was determined by LC-MS analysis. mDHFR expressed in minimal medium were subjected to purification via Ni-NTA affinity chromatography according to the manufacturer's protocol (Qiagen) under denaturing conditions. After purification, expression levels of GFP and mDHFR were determined by UV absorbance at 280 nm using a calculated extinction coefficient of 20,010 cm -1 M -1 and 24,750 cm -1 M -1 , respectively. The purified proteins were concentrated by ultrafiltration (Millipore). 10 μL of the concentrate was diluted into 90 μL of 75 mM (NH 4 ) 2 CO 3 solution and then 1 μL of modified trypsin (Promega, 0.2 μg/μL) was added. Reaction was carried out for 2-4 hrs at 37°C and quenched by addition of 13 μL of 5% trifluoroacetic acid (TFA) solution. The solution was then directly subjected to LC-MS analysis conducted on a LCT Premier XE MICROMASS MS system (MS Technologies, Montgomery Village, MD) with Acquity UPLCTM system (Waters, Milford, MA). Tryptic digests were separated by Acquity BEH300 C18 column (1.7 μm, 300 Å, 2.1 x 50mm) using a gradient of 5-95% of solvent B (90% of acetonitrile/10% of 0.1% formic acid solution) and solvent A (2% of acetonitrile/98% of 0.1% formic acid solution) in 10 min. The column eluent was transferred to the electrospray source and mass spectra were recorded. MALDI-TOF MS analysis of tryptic digests of mDHFR was performed as described previously [20].