A New Highly Specific and Soluble Protease for Precise Removal of N‑Terminal Purification Tags

Abstract

Biotherapeutics production has been significantly enhanced by affinity purification. After purification, however, it is often necessary to remove the affinity purification tag. Thus, we aim for a protease suitable for such a task with properties that include high production yields, good solubility and stability, high cleavage specificity, sufficiently fast turnover, and tolerance of the amino acid identity at the P₁′ position (the C-terminus of the recognition site). Here, we describe the development and characterization of a novel protease named Con1, which is expressed and purified with high solubility and stability. The active site of Con1 harbors a Cys-His-Asp catalytic triad like most of the natural cysteine proteases from viral origins. This validates the optimum enzyme activity under ambient conditions, including physiological pH. Like the Turnip mosaic virus (TuMV) protease, Con1 recognizes the amino acid sequence EAVYHQ (P₆–P₁) and tolerates many different residues at the P₁′ position. The studied 12 amino acids at the P₁′ position represent the different grouping of hydrophobic (A, F, G, I, M, and W), polar uncharged (Q, S, and T), positively charged (R), and negatively charged (D and E) amino acids. Con1 shows faster kinetics than TuMV protease against selected P₁′ substrates. As is typical for this class of viral proteases, Con1 does not cleave substrates with proline at the P₁′ position. We also showed that Con1 efficiently removed the purification tags from representative pharmaceutical/research products such as StefinA, DARPin, and INF2α. Because the active site is close to the C-terminus, we found that Con1 is C-terminal sensitive. The activity is decreased upon trimming the last 7 residues; on the other hand, by cutting 13 residues, the catalytic efficiency has improved with a 2× lower K _m value than that of Con1. Overall, Con1 and its variant have suitable characteristics for biotech applications that will aid the biopharma industry.

Introduction

Affinity purification of recombinant proteins is a powerful tool. − The use of protein purification tags, such as polyhistidine (His-tag), FLAG-tag, glutathione S transferase (GST), or maltose binding protein (MBP), has greatly simplified the purification of recombinantly expressed proteins. − In addition, tags like GST or MBP may play a crucial role in improving the expression and solubility of a protein of interest (POI). − The affinity tag is often placed at the N-terminus of the POI so that it can be enzymatically removed during the purification. , Removal of such affinity tags is often accomplished by incorporating a protease cleavage site between the affinity tag and the POI. Another cleaving method that showed success to release the C-terminus affinity tags is by incorporating a metal-chelating residue that, by complexing with metal ions, hydrolyzes the peptide bonds. Several enzymes have been developed for such tag removal; however, although several of these are suitable for use in the laboratory, they have had limited application at an industrial scale. ,, A cost-effective and robust protease is one that offers high solubility and stability to allow large-scale production at the industrial level, along with suitably rapid catalysis and the precise removal of purification tags.

To date, there is still a gap to be filled by a protease that offers most of the required characteristics. ,− Thrombin is one of the most common serine proteases that shows fast catalysis, but it cleaves the POI nonspecifically. − The split-inteins have been investigated as autoremoving tags, but they often exhibit slow catalysis and/or spontaneous cleavage, which results in a decrease in the final yield of the POI. ,− More recently, caspase-based technologies are being used to remove tags; ,, however, the narrow substrate specificity, low efficacy in cleaving off large protein tags, and off-target cleavage can reduce their applicability. ,,

Most importantly, the majority of the developed enzymes are restricted to specific amino acids at the C-terminus of the recognition site (P₁′), which results in the need for an extra, unwanted amino acid being added to the N-terminus of the POI. For example, viral proteases are attractive because of their high specificity at the N-terminal side of the cleavage position; however, few of these enzymes have been studied in detail. , One of the most widely used viral proteases is the Tobacco Etch Virus nuclear-inclusion-A endopeptidase (TEV protease) with the recognition sequence ENLYFQ−G/S, where (−) represents the site of cleavage. However, TEV preferred residue Gly or Ser at the P₁′ position, which means that after cleavage, a Gly or Ser residue is added to the N-terminus of the POI. For many laboratory applications, the addition of an extra residue to the POI is not problematic. If a recombinant POI is to be a biotherapeutic, however, it must be closely similar to the natural source protein “the reference drug”, for which there is already regulatory approval. Thus, the addition of extra amino acids to the N-terminus is not desirable. Additionally, it is well-known that TEV protease exhibits low solubility and stability that diminishes the purification yield, high concentration storage, and use at the large-scale level. ,,, However, great effort has been made to improve TEV protease solubility by directed evolution, in silico designs, or multiple mutagenesis. ,,−

In addition to TEV protease, the recognition site (P₆–P₁′) of certain viral proteases has been determined such as VRFQ-z for tobacco vein mottling virus (TVMV) protease, VxHQ-z for plum pox potyvirus (PPV) and potato virus Y (PVY) proteases, and VxHQ-z for Turnip mosaic virus (TuMV) protease, where (x) is any amino acid and (z) is a relatively small residue. − Although these proteases are more than 50% identical in their amino acid sequences, they target different recognition sites and have variations in their solubility profiles. With these observations, we hypothesized that it should be possible to create a protease with high substrate specificity, with little restriction on the identity of the P₁′ residue, and with good solubility.

We first found all viral proteases with 50–60% sequence identity to the wild-type sequence of the well-characterized TEV protease. , These proteases were then filtered by their predicted solubility score, and the least soluble one was excluded from the subsequent consensus alignments. Several different consensus sequences were possible, depending on which residues were selected at poorly conserved positions. We computationally assessed the potential solubility of the sequences, and the highest ranking sequence, Con1, was chosen for further experimental study. In the wet laboratory, we found that Con1 has excellent solubility and stability. The P₆ to P₁ substrate binding specificity is the same as that of TuMV protease, but the measured kinetic properties revealed that the maximum rate of Con1 catalysis is faster. We determined that Con1 could readily cleave substrates with each of 12 different residues at the P₁′ position (these residues were chosen to represent different amino acid classes). Importantly, Con1 efficiently cleaved off the purification tags from exemplary test proteins: the human cysteine proteinase inhibitor, StefinA; the human cytokine, Interferon α2; a designed ankyrin repeat protein (DARPin), and green fluorescent protein (GFP).

Materials and Methods

FASTA Search for TEV-Like Proteases

The full-length sequence of the TEV protease (1–242), also known as nuclear inclusion protein a (NIa), was taken from the UniProt database (UniProt code of the polyprotein origin: P04157; Genome position in TEV virus: 2038–2279). We performed a sequence similarity search of this sequence against the UniProt Knowledgebase and UniProtKB/Swiss-Prot isoforms databases using the FASTA suite of programs. We shortlisted the results by excluding variants with ≤50% (which is the desired threshold) and ≥90% (which are mutant versions of TEV protease) identity to the TEV protease sequence.

Intrinsic Solubility Prediction

We used CamSol, a web server, to predict protein solubility based on the amino acid sequence. − We used the sequence-based method to estimate the solubility intrinsic score of each protease at neutral pH. Scores >0 indicate soluble proteins, whereas scores of <0 indicate poorly soluble proteins.

TEV Structure Analysis

To identify the substrate binding site residues, we referred to the available TEV structure (PDB ID: 1LVB). This is a costructure of an inactive TEV protease (C151A) with a peptide substrate (TENLYFQSGT). We used ICM-BrowserPro software to identify the substrate-active site interactions based on geometrical distances between atoms in the receptor and atoms in the substrate. The binding residues of the substrate P₁′ (S), P₁ (Q), P₃ (Y), and P₆ (E) positions which are responsible for the specificity were identified.

Alignment and Consensus

We aligned the full-length sequence of the selected proteases (TEV, TVMV, TuMV_Q, TuMV_J, PPV, OMV, LMVE, and LMV0) using the Clustal Omega tool using the default settings. The alignment was then exported and analyzed by the Jalview 2.11.2.6 software. The alignment was colored by the Clustal scheme with a >75% identity threshold. The software automatically generates the consensus sequence as a percentage (∼40% conservation) of the modal residue per column. A clear consensus backbone with 243 residue lengths comprising 207 conserved residues and 36 nonconserved residues was identified. In addition to the 36 nonconserved residues, we randomly selected another 36 positions from the conserved ones that have less than 75% identity to be varied. These 72 positions were filled on a random basis by choosing the most hydrophilic residues from the equivalent positions of the original proteases (the eight natural proteases in the alignment). These random combinations allowed us to manually build 10 different full-length variants (Con1–Con10) to start with. Finally, the C-terminal residues (235–243) were deleted to improve the predicted solubility since most TEV protease studies have been performed using C-terminal truncated versions. ,,

AlphaFold Structures

Since not all of the selected proteases have available crystal structures, we used ColabFold-based AlphaFold2 to predict the structures of the shortlisted proteases. The protease sequences were used as inputs to generate multiple sequence alignments (MSAs) from UniRef100 and PDB70 databases by the MMseqs2 suite (Many-against-Many sequence searching) and HHsearch tools, along with AlphaFold2 to predict the structure of Con1.

Plasmid Construction

We obtained synthetic genes encoding 7× His-tagged TuMV_J and Con1 (IDT, Belgium). These were cloned by polymerase chain reaction (PCR) to a pMAL expression vector using AQUA cloning. Following transformation into the Escherichia coli strain Top 10, the desired plasmids were identified by DNA sequencing. C-terminal deletions of Con1 were performed using PCR with primers designed to delete the nucleotides corresponding to amino acids 228–234 or 222–234 (Table S1). The synthetic genes encoding CFP-linker-cleavage site-linker-YFP (IDT, Belgium) were cloned by digestion/ligation into the backbone of a pET28 expression vector. The inserts and a vector were digested by BamHI/BsrGI enzymes and ligated by T4 ligase (NEB). We designed constructs encoding the N-terminal His tag separated from the POI by the P₆–P₁ residues of the Con1 cleavage site. The N-terminal residue of the POI contributes to the P₁′ position of the cleavage site. The correct clones were identified by DNA sequencing. See the supporting materials for genes and primer sequences. TEV clone in a pMAL vector (pRK793) was obtained from the laboratory inventory.

Protein Expression

Proteins were expressed by using standard protocols. Briefly, colonies of BL21­(DE3) containing the desired construct were inoculated into 5 mL of LB plus antibiotic and grown overnight at 37 °C. The next day, the overnight cultures were added to LB plus antibiotic media and grown with shaking at 37 °C until the OD₆₀₀ was 0.6. IPTG was then added to a final concentration of 1 mM, the temperature was dropped to 20 °C, and the incubation continued overnight. The following day, cells were harvested by centrifugation at 4000g for 15 min and stored at −80 °C.

Purification of Con1, TuMV_J, and TEV Proteases

Cell pastes (from 100 mL culture) were resuspended in 10 mL buffer A (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, and 5% (v/v) glycerol) supplemented with protease inhibitor cocktail tablets (cOmpleteTM, Roche). The suspensions were kept on ice, and the cells were disrupted by sonication. After sonication, the solution was centrifuged at 15,000g for 30 min at 4 °C to remove insoluble material, and the pellet was discarded. Supernatants were applied to a 5 mL Ni-NTA agarose column pre-equilibrated in buffer A. After sample loading, the resin was washed with buffer A plus 20 mM imidazole. Finally, proteins were eluted by buffer A with an imidazole concentration from 75 to 500 mM. The fractions were analyzed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE), pooled, buffer exchanged into buffer A, and concentrated using 3 kDa MW-CO Amicon (Merck-Millipore) at 4000g for 30 min at 4 °C.

Purification of FRET Substrates, His-POI, and Proteases (Con1, ^Δ228–234Con1, and ^Δ222–234Con1)

Cell lysates were prepared as described above. The supernatant was applied to a HisTrap Excel column (Cytiva) equilibrated in buffer A and eluted by a gradient of 0–1 M Imidazole in buffer A. The peak fractions were analyzed by SDS–PAGE, pooled, buffer exchanged, and concentrated using a 3 kDa MW-CO Amicon filter at 4000g for 30 min at 4 °C and then stored at – 80 °C. Förster resonance energy transfer (FRET) substrates and proteases were additionally purified by gel filtration using a Superdex-75 column, equilibrated, and run-in buffer A. After checking the purity on SDS–PAGE, appropriate fractions were pooled, concentrated using a 3 kDa MW-CO Amicon filter at 4000g for 30 min at 4 °C, and then stored at – 80 °C.

Protease Solubility

The solubility of TEV and Con1 proteases was studied by concentrating proteins and monitoring the soluble versus insoluble distribution over time. Five milliliter aliquots of proteins at a low concentration (1.5 mg/mL) were centrifuged at 15,000g for 10 min at 4 °C to pellet any aggregates before the experiment began. Protein concentrations were then measured by the Nanodrop spectrophotometer using the extinction coefficients of 31,970 and 34,950 M^–1 cm^–1 for TEV (M _W = 28560.45 Da) and Con1 (M _W = 27660.22 Da), respectively. The proteins were loaded into 5 mL of 3 kDa MW-CO Amicon and centrifuged at 4000g at 4 °C to concentrate, and samples were taken every 30 min until aggregation was observed or the concentrator volume limit was reached. High concentrations of TEV (15 mg/mL) or Con1 (30 mg/mL) proteases were incubated overnight at room temperature; aliquots were collected over time, and optical density (OD) at 400 nm was measured.

Substrate Specificity

We assessed the specificity of Con1, TEV, and TuMV_J for the cleavage sites EAVYHQ-S and ENLYFQ-G. Protease (1 μM) and FRET substrate (2 μM) were incubated in Buffer A at room temperature (∼22–25 °C). The mixtures were excited at 430 nm, and fluorescence emission intensity at 480 and 520 nm was measured every minute over 2 h. The amount of cleavage was determined from the change in the FRET ratio, calculated by the intensity at 480 nm (donor) divided by the intensity at 520 nm (acceptor) after blank subtraction. Cleavage is a percentage of the end point after saturation. All of the experiments were performed in duplicate. To screen the P₁′ specificity of Con1, FRET substrates with different cleavage sequences: EAVYHQ-X (X = S, D, M, or P). A 0.5 μM Con1 was mixed with 5 μM substrate in buffer A, and the fluorescence was measured over 5 h at room temperature.

pH Dependency of Con1 Activity

We performed the FRET cleavage assay in buffers of different pHs: 0.1 M sodium citrate (pH 4, 5, or 6), 0.1 M Tris-HCl (pH 7, 8, or 9), and 0.1 M sodium borate (pH 10). The Con1 protease (0.5 μM) and the FRET substrate (2.5 μM) were preincubated in the different pH buffers at room temperature (∼22–25 °C) for 10 min. The reaction was initiated by mixing the protease with the substrate, and incubation continued for a further 2 h at room temperature (∼22–25 °C). The end-point fluorescence was measured at 480 and 520 nm, and the amount of cleavage was calculated from the change in FRET. Additional SDS-PAGE experiments were performed when Con1 and His7-(EAVYHQ-M)-DARPin were preincubated in pH 7.0 or 8.0 of 100 mM Tris-HCl, 150 mM NaCl, or phosphate-buffered saline (PBS), pH 7.0. Then, different concentrations of Con1 (0.3, 0.6, 1, 3, or 6 μM) were mixed with the substrate (30 μM) at a certain pH/buffer and incubated for 4 h at 25 °C. The mixtures were then resolved on 10% SDS-PAGE to evaluate the His-tag cleavage.

Temperature Dependency of Con1 Activity

The Con1 protease and the FRET substrate were preincubated in buffer A at different temperatures (4, 15, 25, 37, 50, and 70 °C) for 10 min. The reaction was initiated by mixing Con1 with substrate to final concentrations of 0.5 and 2.5 μM, respectively, and incubation continued for a further 2 h. The end-point fluorescence was measured at 480 and 520 nm, and the amount of cleavage was calculated. Additional SDS-PAGE experiments were performed when Con1 and His7-(EAVYHQ-M)-DARPin were preincubated in 100 mM Tris-HCl, 150 mM NaCl, pH 7.0 at different temperatures (0, 4, 12, 25, 37, or 50 °C). Then, 1 μM Con1 was mixed with the 100 μM substrate and incubated for 4 h at the specified temperature. The mixtures were then resolved on 10% SDS-PAGE to evaluate the His-tag cleavage.

Enzyme Kinetics

Kinetic parameters for the Con1 vs TuMV_J were determined by incubating 0.4 μM enzyme with the substrates at a range of concentrations (1–24 μM) in buffer A at room temperature (∼22–25 °C). Different FRET substrates EAVYHQ-X (X = Ala, Ser, Met, or Asp) between the FRET donor (CFP) and acceptor (YFP) proteins were used to assess the influence of the identity of the P₁′ position. Kinetics of Con1 at different pHs was determined when Con1 and substrate fractions were preincubated at different pHs of 0.1 M citrate, pH 6.0; 0.1 M Tris-HCl, pH 7.0; 0.1 M Tris-HCl, pH 8.0 or 0.1 M Tris-HCl, pH 9.0 at room temperature (∼22–25 °C). The reaction mixtures containing 0.4 μM Con1 incubated with an increasing concentration of the FRET substrate incorporating the EAVYHQ-M site (0.3125, 0.625, 1.25, 2.5, 5, 7.5, 10, and 12.5 μM) at different pHs. The kinetic parameters for Con1, ^Δ228–234Con1, and ^Δ222–234Con1 (0.4 μM) against FRET substrate incorporating EAVYHQ-M site (0.625–25 μM) were determined in 50 mM Tris-HCl, pH 7.0, 200 mM NaCl buffer. For the data analysis, the data were collected and analyzed as previously described. The full-scale range (FSR) was calculated for each wavelength by subtracting the minimum emission Em^min (the intensity at 480 nm at time zero and the end-point intensity at 520 nm) from the maximum emission (Em^max; which is at the end-point intensity at 480 nm and the time zero intensity at 520 nm), as shown in the following equations:

$${\mathrm{FSR}}_{480\mathrm{nm}}={{\mathrm{Em}}_{480\mathrm{nm}}}^{\max }-{{\mathrm{Em}}_{480\mathrm{nm}}}^{\min }$$

$${\mathrm{FSR}}_{520\mathrm{nm}}={{\mathrm{Em}}_{520\mathrm{nm}}}^{\max }-{{\mathrm{Em}}_{520\mathrm{nm}}}^{\min }$$

Then, the emission values were normalized by subtracting the emission at time zero, as shown in the following equations:

$$\mathrm{\Delta }{\mathrm{Em}}_{480\mathrm{nm}}={\mathrm{Em}}_{480\mathrm{nm}}-{{\mathrm{Em}}_{480\mathrm{nm}}}^{t=0}$$

$$\mathrm{\Delta }{\mathrm{Em}}_{520\mathrm{nm}}=-1({\mathrm{Em}}_{520\mathrm{nm}}-{{\mathrm{Em}}_{520\mathrm{nm}}}^{t=0})$$

Then, the cleavage percentages (C ^%) were calculated by dividing the normalized values by the FSR values, as shown in the following equations:

$${C_{480\mathrm{nm}}}^{\%}=100(\mathrm{\Delta }{\mathrm{Em}}_{480\mathrm{nm}}/{\mathrm{FSR}}_{480\mathrm{nm}})$$

$${C_{520\mathrm{nm}}}^{\%}=100(\mathrm{\Delta }{\mathrm{Em}}_{520\mathrm{nm}}/{\mathrm{FSR}}_{520\mathrm{nm}})$$

The cleaved product concentration [P] was calculated as a percentage of the total substrate concentration at time zero by the following equation

$$[\mathrm{P}]=(C^{\%}\times {[\mathrm{S}]}_{t=0})/100$$

After that, the product concentrations (on the Y-axis) were plotted vs time (on the X-axis), and the curves were fitted by the exponential plateau equation on GraphPad Prism software:

$$[\mathrm{P}]=[{\mathrm{P}}_{\max }]-([{\mathrm{P}}_{\max }]-[{\mathrm{P}}_{\min }]){\mathrm{e}}^{-kt}$$

where [P_max] is the maximum product concentration, [P_min] is the minimum product concentration, k is the rate constant, and t is the time. From the calculated rate constants, the reaction rates (V) were calculated by the following equation:

$$V=k\left[{\mathrm{P}}_{\max }\right]$$

Finally, the reaction rates (Y-axis) were plotted vs substrate concentration (X-axis), and the curves were fitted by the Michaelis–Menten equation using GraphPad Prism software:

$$V=(V_{\max }[\mathrm{S}])/(K_{\mathrm{m}}+[\mathrm{S}])$$

where V is the reaction rate, V _max is the maximum velocity, [S] is the substrate concentration; and K _m is the Michaelis constant. The values of V _max and K _m were obtained from the nonlinear regression, and then k _cat was calculated by the following equation:

$$k_{\mathrm{cat}}=V_{\max }/[\mathrm{E}]$$

where k _cat is the catalytic turnover number and [E] is the total enzyme concentration.

His-Tag Removal from Different Substrates

We used Con1 to cleave an N-terminal His-tag from different POIs. We mixed Con1 (0.5 μM) with the His7-tagged POI (50 μM). Tested POI include His7-(EAVYHQ-M)-StefinA, His7-(EAVYHQ-M)-Interferonα2, His7-(EAVYHQ-G)-DARPin and His7-(EAVYHQ-M)-GFP. The mixtures were incubated at room temperature (∼22–25 °C) overnight with gentle shaking. The cleavage was assessed by separating proteins on SDS-PAGE. The P₁′ tolerance was examined by incubating 0.5 μM of Con1 or ^Δ222–234Con1 with 50 μM of different His7-(EAVYHQ-X)-DARPin substrates with 12 different N-terminals and incubating overnight at room temperature. The cleavage was assessed by separating proteins on SDS-PAGE.

Investigating Different Molar Ratios and Incubation Times

Either increasing Con1 concentrations (0.1, 0.2, 1, 2, and 10 μM) were mixed with 100 μM of His₇-(EAVYHQ-X)-DARPin, where X = Gly, Asp, Met, or Ile. The reactions were incubated at 25 °C for either 4 or 17 h or increasing substrate concentrations of His₇-(EAVYHQ-X)-DARPin, where X = Asp or Met (100, 200, 300, 400, 500, or 600 μM) were mixed with 1 μM of Con1. The reactions were incubated at 25 °C for 17 h. Finally, 1 μM Con1 was mixed with 100 μM His₇-(EAVYHQ-X)-DARPin, where X = Gly, Asp, Met, or Ile. The reactions were incubated at 25 °C, and fractions were collected at time points (0, 0.5, 1, 2, 4, and 24 h). The cleavage was assessed by separating proteins on SDS-PAGE.

Results and Discussion

Identification of TEV-Like Proteases and Consensus Design

We used a FASTA search to identify TEV-like proteases in the UniProt Knowledgebase and UniProtKB/Swiss-Prot isoforms databases. From the list of 50 proteases identified, we excluded those with more than 90% or less than 50% identity to the TEV protease. Eight proteases meet these criteria, ranging in sequence identity to TEV protease from 52.3 to 60.6% (Table ). All were plant viral proteases of the type nuclear inclusion protein a (NIa) with high substrate specificity; however, most of them are little studied. In Table , we specify the mostly identified cleavage site, if known, with the predominant glutamine (Q) residue at the P₁ position of all of the sequences and relatively small residues (A, G, or S) at the P₁′ position of the cleavage site (data taken from the MERPOS database; Figure S1).

1. Proteases with ≥50% and ≤90% Identity to TEV Protease.

UniProt ID	Viral origin	Abbreviation	Genome position	Identity (%)	Cleavage site
P04517	Tobacco etch virus	TEV	2038–2279	reference	ENLYFQ-G
Q85197	Potato virus A	PVA	2032–2274	60.6	EAVQFQ-S
Q02597	Turnip mosaic virus (strain Quebec)	TuMV_Q	2116–2358	54.8	TVYHQ-A
P89509	Turnip mosaic virus (strain Japanese)	TuMV_J	2117–2359	54.8	EAVYHQ-S
Q01681	Plum pox potyvirus (strain El amar)	PPV	426–668	53.7	TVYHQ-A
P20234	Ornithogalum mosaic virus	OMV	123–365	53.3	–
P09814	Tobacco vein mottling virus	TVMV	2002–2242	52.3	ETVRFQ-S
P89876	Lettuce mosaic virus (strain E)	LMVE	2215–2457	52.9	–
P31999	Lettuce mosaic virus (strain 0/French)	LMV0	2215–2457	52.5	–

^aPercent identity of the viral protease to the TEV protease, (−) Not identified.

We assessed the relative solubility of the shortlisted proteases using CamSol analysis of their sequences. − TuMV-J was predicted to be the most soluble and PVA the least (Table S1). Therefore, the PVA protease was excluded from the shortlist. The remaining eight proteases were aligned using Clustal Omega (Figure A); , from this alignment, we generated a consensus sequence. The entire consensus sequence is 243 amino acids. In the sequence, 207 residues have more than 40% conservation among the eight aligned sequences; however, 36 positions of the conserved sequence were selected for variation in the design along with the 36 nonconserved ones. All of the 72 positions were chosen and varied based on the degree of hydrophilicity of the original residues in the parent sequences (the eight natural proteases in the alignment). We randomly created ten sequences (Con1–Con10) with different combinations at 72 positions and predicted their solubility using CamSol (Figure B and Table S2).

1.

Design of consensus protease. (A) Alignment of the full-length amino acid sequences of the indicated proteases, using Clustal Omega, with a 75% identity threshold colored by the default Clustal scheme (light blue for hydrophobic, red for + ve charge, magenta for – ve charge, green for polar, pink for cysteine, orange for glycine, yellow for proline, cyan for aromatic, and colorless for nonconserved residues). The consensus sequence was automatically defined by the Jalview viewer. The asterisk annotations indicate the TEV residues that interact with the substrate residues (P₁, P₃, P₆, and P₁′). The hash mark annotations indicate the catalytic triad (H46, D81, and C151) based on the structural analysis of PDB 1LVB (Figure S3). (B) Intrinsic solubility prediction of the full-length (1–243) and C-terminal truncated (1–234; Δ^235–243) versions of the designed library, along with TEV and TuMV_J proteases as references. The bluer, the more soluble. (C) Ribbon representation of TEV protease structure (pink) complexed with uncleaved substrate (orange; PDB ID: 1LVB). (D) AlphaFold2 model of the full-length Con1 protease (1–243 aa) predicted by ColabFold. (E) Close-up view of the active site of Con1 (light blue) superimposed onto TEV protease (pink); the C-terminus of Con1 is indicated in orange.

We also used CamSol to evaluate the predicted effect of C-terminal (235–243) deletion on the solubility of the designs. The deletion was predicted to enhance solubility (Figure B). This result is consistent with a form of TEV with the C-terminal tail deleted often being used to enhance solubility (Figure C–E). ,, We selected the top hit, Con1, for the experimental studies.

Con1 Protease Solubility

Con1 (pRS11, Addgene #227090) or TuMV_J with the C-terminal amino acids (235–243) deleted and TEV (pRK793, Addgene #8827) were expressed in soluble form and purified as described (Figure S2).

A protease with high solubility is a key goal of this study. To experimentally assess protein solubility and because TEV protease is well-studied and characterized, ,, we compared the solubility of Con1 to that of a version of TEV with the S219 V mutation to prevent the self-cleavage and polyarginine (R5) C-terminal to improve the solubility. TEV was expressed downstream to the maltose binding protein (MBP) tag to aid the soluble expression, giving a final yield of ∼10 mg of pure protein per liter of culture. Con1 was expressed without either MBP or the extra R5 tail, with a final yield of ∼25 mg of pure protein per liter of culture. Previous studies reported that engineered versions of TEV protease with multiple mutations yielded between 10 and 50 mg/L soluble protein. , We further assessed the solubility and stability by tracking the highest stable concentration of Con1 or TEV protease that could be achieved by concentrating until the solubility threshold beyond which was aggregation initiated. TEV started to aggregate at above 15 mg/mL, whereas Con1 did not aggregate up to 30 mg/mL, the highest concentration possible in this experiment (Figure A). After filtering and bringing TEV to 15 mg/mL and Con1 to 30 mg/mL, aliquots were incubated at room temperature, and aggregation was determined by measuring the absorbance at 400 nm. As shown in Figure B, TEV protease aggregated significantly, while Con1 is quite stable.

2.

Solubility profile of Con1 and TEV proteases. (A) TEV (red) or Con1 (blue) protease concentration over a 3 K MW-CO Amicon concentrator. Each concentration round was carried out for 30 min, aliquots were collected every round, and the concentrations were measured by Nanodrop spectrophotometer in duplicate readings using the extinction coefficient of 31,970 and 34,950 M^–1 cm^–1 for TEV (28560.45 Da) and Con1 (27660.22 Da), respectively. (B) Optical density tracking of 15 mg/mL TEV and 30 mg/mL Con1 at 400 nm was measured by Nanodrop in duplicate readings.

Con1 Is Specific to EAVYHQ-S but Not ENLYFQ-G

The consensus sites for substrate recognition and cleavage for TEV protease are ENLYFQ-G (Figure A). ,, The substrate binding site of Con1 is similar to that of TuMV_J. Therefore, the target sequence for TuMV_J and Con1 is EAVYHQ-S (P₆–P₁′), where the cut is between P₁ (Q) and P₁′ (S) (Figure B). − , We assessed the specificity of TEV, TuMV_J, and Con1 using a FRET-based assay in which different potential substrate recognition sequences were incorporated between cyan fluorescent protein (CFP, donor) and yellow fluorescent protein (YFP, acceptor; Figures C and S4 and Notes S1.4 and S1.5) , (see Materials and Methods for more details). Con1 and TuMV_J cleave EAVYHQ-S but not ENLYFQ-G (Figures D,E and S5), which is consistent with the reported TuMV_J specificity. ,, The key residues that change the specificity between TEV and TuMV_J/Con1 are (TEV → TuMV_J/Con1): T30 → N30, S170 → A170, N176 → I176, H214 → L215 and K220 → Q221. Previous studies suggested that the consensus sequence for TuMV_J cleavage site is xxVxHQ-z, where (x) is any amino acid and (z) is small ones (e.g., G, S, A, or T). ,,,, Thus, the substrate specificity arises from the P₁ (Q), P₂ (H), and P₄ (V).

3.

Substrate specificity of Con1 protease. (A, B) Substrate sequence preferences for cleavage by TEV protease; n = 19 (A) or TuMV_J protease; n = 7 (B). The amino acid preferences at each position (P₆–P₅–P₄–P₃–P₂–P₁–P₁′) are indicated as logo plots (WebLogo online application). (C) Schematic illustration describing the FRET-based assay. The substrates contain cyan fluorescent protein (CFP; donor) and yellow fluorescent protein (YFP; acceptor) separated by a potential cleavage site. Before cleavage, by exciting the substrate at 430 nm, the emission of YFP is high at 520 nm, while CFP emission at 480 nm is low. If the substrate is cleaved by the protease, the YFP emission decreases and CFP emission increases. (D, E) Screening Con1, TuMV_J, or TEV proteases against FRET substrates with different cleavage sites, ENLYFQ-G (D) or EAVYHQ-S (E). All of the results are the mean of 2 replicates ± standard deviation. Ordinary one-way ANOVA was used: ****P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05; ns, P > 0.05 with Tukey’s test correction. The results are representative of two independent experiments.

Con1 and TuMV_J Cleavage of EAVYHQ-X Substrates

We used the FRET-based assay to determine the kinetic parameters of Con1 and TuMV_J against substrates with different residues at the P₁′ position. The substrates have Ala, Met, Asp, Ser, or Pro at the P₁′ position. As expected, neither enzyme cleaves a substrate with proline at the P₁′ position (Figure S6), which is a common behavior for the Potyvirus genus cysteine proteases. On the other hand, both Con1 and TuMV_J cleave all of the other substrates, with Con1 exhibiting a higher catalytic rate than TuMV_J against them (Figures , S7, and S8). The kinetic parameters of Con1 are indicated in Table , but the low catalytic activity of TuMV_J precluded accurate calculations of the kinetic parameters by this method.

4.

Kinetic studies of Con1 and TuMV_J. (A, B) Rate versus substrate concentration plots for Con1 (A) and TuMV_J (B) against the different FRET substrates (P₁′ = Ser, Ala, Asp, or Met). The results are the mean of 2 replicates ± standard deviation. The curves are fitted by the Michaelis–Menten nonlinear regression using GraphPad Prism software. The curve colors are indicated in part (B). The results are representative of two independent experiments.

2. Summary of the Kinetic Parameters of Con1 .

Substrate P1′	Km (μM)	kcat (s–1)	kcat/Km (s–1 M–1)
Ala	13 ± 3	0.012 ± 0.002	924 ± 45
Asp	23 ± 4	0.011 ± 0.004	454 ± 70
Met	23 ± 7	0.011 ± 0.003	477 ± 3
Ser	11 ± 2	0.013 ± 0.002	1248 ± 66

^aThe results are mean ± standard deviation of 2 independent experiments with 2 technical replicates each.

The less-studied TuMV_J kinetics showed a K _m of ∼1 mM for TuMV to cleave EPTVYHQ-TL, , which is about 100× higher than what we determined for Con1 (Table ). Substrates with different residues at the P₁′ position are cleaved at similar rates by Con1, and differences are primarily in K _m. The reported P₁′ residues for TuMV_J suggest only small amino acids such as Gly, Ser, Ala, or Thr; ,,, however, for Con1, we observed cleavage of Asp and, most importantly, Met, which is the natural start codon in protein translation.

Dependency of Con1 Activity on pH and Temperature

To ascertain the conditions in which Con1 could be used in practice, we characterized Con1 activity at different pHs, buffers, and temperatures. For the pH studies, Con1 protease and substrate were preincubated in buffers of the desired pHs including sodium citrate (pH 4.0–6.0), Tris-HCl (pH 7.0–9.0), or sodium borate (pH 10.0), and then the reaction was initiated by mixing the substrate with the enzyme. The optimal pH for Con1 activity was determined to be around pH 7.5 (Figures A and S9A), independent of whether Tris-HCl or PBS buffer was used (Figure S9B), which is slightly lower than that previously determined (pH 8.2–8.3) for TuMV protease. , The pH dependency profile of Con1 aligns with the presence of the two ionizable residues, H46 and D81, of the catalytic triad in the active site that contribute to the proton transfer from C151 and stabilize the transition state during the proteolysis.

5.

Characterization of the pH and temperature dependency of Con1 activity. (A) Determining the optimal pH. Con1 and FRET substrate (EAVYHQ-S) were mixed to initiate the reaction after 10 min equilibration at the desired pH. The amount of FRET substrate cleaved after 2 h incubation at room temperature at a given pH was determined. Results are expressed as a percentage of the cleavage that occurred at the optimal pH (pH 7.5). (B) Kinetic summary of Con1 at the different pH. The curves are fitted by the Michaelis–Menten nonlinear regression using GraphPad Prism. (C) Determining the optimal temperature. Con1 and substrate were equilibrated in buffer A at the specified temperature for 10 min, and the reaction was initiated by mixing them. The amount of FRET substrate cleaved after 2 h incubation was determined. Results are expressed as a percentage of cleavage that occurred at the optimal temperature (24 °C). All of the results are the mean of 2 replicates ± standard deviation. The results are representative of two independent experiments.

We further assessed the kinetics of Con1 at different pHs (6.0, 7.0, 8.0, and 9.0) against the FRET substrate with the Met at the P₁′ position. The best kinetic parameters were observed at pH 7.0 with a lower K _m and a 2× higher cleavage efficiency than at pH 8.0 (Figures B and S10). Con1 was very slow at pH 6.0 and 9.0, at which the kinetic parameters could not be calculated.

To investigate the effect of temperature on the proteolysis, Con1 and substrate were preincubated for 10 min at various temperatures (4, 15, 25, 37, 50, or 70 °C) in buffer A. Then, the proteolytic reaction was initiated by mixing the substrate with the enzyme. The results indicate that Con1 activity increased by raising the incubation temperature to 37 °C beyond which the activity started to decline, and the optimum activity was determined at a temperature of 24 °C; however, there is no significant variation observed in the activity at 15, 25, or 37 °C when testing the activity by the FRET substrate (Figure B). On the other hand, His-tag cleavage significantly increased by the temperature up to 37 °C as observed on the SDS-PAGE (Figure S9C). Because the FRET substrate consists of two fluorescent proteins with the cleavage site in between, this explains the possible slower rate to cleave large tags such as GST or MBP, while the small tags are cleaved faster. TuMV protease was previously identified as a cold-adapted protease due to the low-temperature optimum activity, which is related to structural stability and flexibility.

Con1 Protease Cleaves His Tag Independent of the Identity of the P₁′ Residue

A key goal of this study was to develop a protease that could be used to remove N-terminal affinity purification tags from a POI after affinity purification. Therefore, we designed different His-tagged POIs with P₆–P₁ of the Con1 recognition site (EAVYHQ) directly upstream of the POI. The first amino acid of the POI contributes to the P₁′ residue. These POIs are the human cysteine proteinase inhibitor, StefinA, the human cytokine, Interferon α2, a designed ankyrin repeat protein (DARPin), and green fluorescent protein (GFP) (Notes S1.10–S1.13 and Figure S11). After affinity purification using Ni-NTA, the His-tagged POI was incubated overnight with Con1. The removal of the N-terminal purification tag was then assessed by separating the proteins using SDS-PAGE. As seen in Figure A, the tags are all completely removed, and no undesired additional cleavage was observed.

6.

SDS-PAGE analysis of different His-tagged proteins of interest before (−) and after (+) incubation with protease. (A) Evaluation of His-tag removal by Con1 from different substrates as indicated. (B) Screening the P₁′ tolerance of Con1 against His-tagged DARPin proteins with different N-terminal tags indicated. Complete removal of the purification tag is evident by the decrease in the M _W of the protein. The reaction mixtures contained 0.5 μM Con1 and 50 μM substrate final concentrations and were incubated overnight at room temperature.

To screen the P₁′ tolerance of Con1, we mutated the natural N-terminal amino acid of the DARPin (G) to 11 different residues covering different grouping of amino acids: hydrophobic amino acids (A, F, I, M, and W), polar uncharged amino acids (Q, S, and T), positively charged amino acids (R), and negatively charged amino acids (D and E; Figure S12). Con1 efficiently cleaved the His tag from all of the studied substrates with different N-termini (acting as a P₁′ position). This result indicates that Con1 can cleave off N-terminal tags, independent of the P₁′ position (i.e., the N-terminal amino acid of the POI) (Figure B).

Additional experiments were performed to evaluate the effects of changing the enzyme to substrate molar ratio (1:1000, 1:600, 1:500, 1:400, 1:300, 1:200, 1:100, 1:50, and 1:10) either by fixing substrate concentration (100 μM) against increasing enzyme concentration (0.1, 0.2, 1, 2, or 10 μM) or fixing enzyme concentration (1 μM) against increasing substrate concentration (100, 200, 300, 400, 500, and 600 μM) versus the incubation time (4 and 17 h) to achieve 100% cleavage of a tag from selected P₁’ residues (Gly, Asp, Met, or Ile). We found that the His tag was fully cleaved from the Gly and Asp substrates after 2 h of incubation at a 1:100 molar ratio; a longer incubation at the same ratio resulted in the cleavage of the tag from the Met and Ile substrates (Figure S13).

Effect of Con1 C-Terminal on the Activity

Con1 is 234 amino acids long. The predicted AlphaFold2 model of Con1 indicates that the C-terminal residues (222–234) have low confidence (Figure S14A). This motivated us to study the effect of C-terminal deletion on the activity of the Con1. We prepared two protease versions by deleting 228–234 (^Δ228–234Con1) or 222–234 (^Δ222–234Con1) residues (Figure S14B). Kinetic studies of Con1, ^Δ228–234Con1, and ^Δ222–234Con1 against FRET substrate with the EAVYHQ-M cleavage site show better cleavage by ^Δ222–234Con1 than by the original Con1. Cleavage by ^Δ228–234Con1 is worse than cleavage by ^Δ222–234Con1 (Figures A and S15). The superior catalytic activity of ^Δ222–234Con1 derives from a K _m value that is about half that of the original Con1. Both have similar k _cat values (Table ).

7.

Kinetic studies of C-terminal truncated versions of Con1. (A) Kinetics summary of Con1, ^Δ228–234Con1, and ^Δ222–234Con1 against the EAVYHQ-M FRET substrate. The results are the mean of 2 replicates ± standard deviation. The curves are fitted by the Michaelis–Menten nonlinear regression using GraphPad Prism software. The results are representative of two independent experiments. (B) Screening the P₁′ tolerance of ^Δ222–234Con1 against His-tagged DARPins proteins with different N-terminal tags as indicated. Complete removal of the purification tag is evident by the decrease in M _W of the protein. The reaction mixture contained 0.5 μM Con1 and 50 μM substrate final concentrations and was incubated overnight at room temperature.

3. Summary of the Kinetic Parameters of Truncated Versions of Con1.

Variant	Km (μM)	kcat (s–1)	kcat/Km (s–1 M–1)
Con1	24.9	0.014	597.2
Δ228–234Con1	>25	NA	NA
Δ222–234Con1	13.65	0.011	814.6

We speculate that because the C-terminus is close to the active site, the unstructured tail may interfere in some fashion with substrate binding. Most importantly, ^Δ222–234Con1 exhibits the same P₁′ tolerance as Con1. We screened ^Δ222–234Con1 activity against the 12 different DARPins with different N-termini (Figure S12). ^Δ222–234Con1 efficiently cleaved the His tag from all of the substrates (Figure B).

Conclusions

For many applications, such as the production of biotherapeutics, it is necessary to remove the affinity tags after purification, especially without leaving extra residue in the final product. An ideal protease for that purpose should be soluble, simple in purification and handling, highly specific, fast in catalysis, and, most important, P₁′ tolerant. Many proteases have been employed in the technology of affinity tag removals; however, only a limited number fulfill most of the required characteristics. Some viral proteases were exploited for this purpose; however, their limited activity against certain residues at the P₁′ position and their low solubility limit their industrial applicability. Inspired by the need for a protease that exhibits all of the desirable characteristics, we sought to create a protease that has activity against substrates with different residues at the P₁′ position and which displays high solubility. Here, we have described the properties of a new enzyme, Con1 protease, whose high solubility, stability, and catalysis make it well-suited for large-scale industrial applications. Con1 is optimally active under ambient conditions without the need for extra loads. It has been tested to work perfectly in different buffer systems and different temperatures. The high substrate specificity of Con1 (via the substrate P₆, P₅, P₄, P₃, P₂, P₁ positions), combined with its tolerance for different amino acids at the substrate P₁′ position, enables the precise and efficient removal of affinity purification tags from a wide variety of proteins.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.5c01764.

• Con1 synthetic gene; predicted solubility profiles of the selected proteases; conserved cleavage sites of proteases; screening substrate specificity of Con1, TuMV_J, and TEV (PDF)

†.

Institute of Genetics and Cancer, Western General Hospital, Crew Road, University of Edinburgh, Edinburgh EH4 2XU, United Kingdom

R.S. and L.R. led and designed the study. R.S. performed the computational and experimental work, analyzed data, generated figures, and wrote the first draft of the manuscript. L.R. conceptualized the study, analyzed data, and edited the manuscript.

This work was funded by an Industrial Biotechnology Innovation Centre (IBioIC) Innovator Award and the BBSRC Impact Acceleration Account (IAA) fund (BBSRC IAA PIII117) in partnership with Fujifilm Diosynth Biotechnology UK (FDBK). The University of Edinburgh is a member of the Centre of Excellence in Bioprocessing, along with Fujifilm Diosynth Biotechnology (FDB), the University of York, and the University of Manchester.

The authors declare the following competing financial interest(s): A UK Patent Application No. 2411203.9, in the name of The University Court of the University of Edinburgh, covering aspects of this work, has been filed.