2Department of Chemical and Biomolecular Engineering, Clemson University, Clemson, South Carolina
Typically, chromatography is the most costly and time-consuming step in protein purification. As a result, alternative methods have been sought for bioseparations, including the use of stimulus-responsive tags that can reversibly precipitate out of solution in response to the appropriate stimulus. While effective, stimulus-responsive tags tend to require temperature changes or relatively harsh buffer conditions to induce precipitation. Here we describe a synthetic peptide, based on the natural repeat-in-toxin (RTX) domain that undergoes gentler calcium-responsive, reversible precipitation. When coupled to the maltose binding protein (MBP), our calcium-responsive tag efficiently purified the fusion protein. Furthermore, when the MBP was appended to green fluorescent protein (GFP), β-lactamase, or a thermostable alcohol dehydrogenase (AdhD), these constructs could also be purified by calcium-induced precipitation. Finally, protease cleavage of the precipitating tag enables the recovery of pure and active target protein by cycling precipitation before and after cleavage.
Non-chromatographic purification techniques are of significant interest since chromatography is typically the most expensive step in protein purification (1). Alternative approaches often rely on targeted precipitation of the protein of interest. One approach is metal chelate affinity precipitation, where thermoresponsive copolymers can be used to specifically precipitate out poly-histidine tagged recombinant proteins (2, 3). Another purely protein-based approach is the use of thermoresponsive elastin-like peptides (ELPs) that consist of tandem repeats of the sequence VPGXG and precipitate with small temperature increases(4, 5). ELPs undergo an inverse phase transition and aggregation, which is thought to be driven by the exposure of hydrophobic patches in the peptides upon heating (6). As part of a purification system, ELPs have been coupled to intein domains that have been genetically engineered into minimal self-cleaving units (7). When coupled, the ELP-intein system allows for a simple two-stage purification scheme. In the first step, precipitation of the ELP is triggered and the fusion protein is purified. Then, the intein is induced to cleave off the target protein and the ELP is again precipitated, leaving behind pure target protein in solution (8). While effective for many purification applications, the necessary heating of samples or the alternative use of high salt concentrations (9) can be problematic in many situations. Another protein-based non-chromatographic purification scheme developed by Ding et al. relies on calcium-dependent precipitation of an annexin B1 tag (10). As with ELPs, a self-cleaving intein is also incorporated in the fusion protein to remove the tag following purification.
A new calcium-responsive tag based on a consensus sequence found in the natural repeat-in-toxin (RTX) domain is presented. This calcium-responsive tag works under gentler reaction conditions than existing approaches and can be removed through protease cleavage, resulting in a pure, active target protein.
Our interest in alternatives to chromatography for purifying proteins is a product of discoveries made while exploring repeat scaffolds for protein engineering applications. Repeat scaffolds are of interest to protein engineers as their repetitive, predictable secondary structures make them ideal for studying folding and engineering novel functions (11, 12). There are examples of repeat scaffolds being engineered for biomolecular recognition, most notably the ankyrin repeats (13). In order to improve the engineerability of these scaffolds, efforts have been made at consensus design. Consensus design seeks to identify the core repeating peptide unit. Once this sequence is identified, multiple repeats can be concatenated as necessary for the desired application. Consensus design approaches have been successfully used for a number of repeat scaffolds, including ankyrin repeats, tetratricopeptide repeats, and armadillo repeats (14-17). The ability to alter the size of a scaffold is of particular use when engineeringbinding, as the interface size can be tuned to the particular target.
In an effort to explore novel scaffolds for protein engineering, we have sought to identify a repeat scaffold that is also stimulus-responsive. To this end, we investigated the calcium-responsive repeat-in-toxin (RTX) domain. The RTX domain is found in proteins secreted through the bacterial type 1 secretion system (18). The domain consists of repeats of the consensus amino acid sequence GGXGXDXUX, where X is variable and U is a hydrophobic amino acid. One of the most well characterized RTX domains is the block V RTX domain from the adenylate cyclase toxin (CyaA) of B. pertussis. The domain is intrinsically disordered in the absence of calcium and forms a β roll structure (Figure 1A) in the presence of calcium (19). Of note, the block V RTX domain retains its reversible calcium-responsiveness even when expressed separately from the larger protein (20, 21). Previous attempts have been made to use RTX domains in protein engineering, including incorporation into mesh networks, design of synthetic RTX peptides, and generation of hydrogel-forming RTX domains (22-25).
Our original objective was to design consensus RTX domains. Specifically, we identified the frequency of amino acids at each position of the nine amino acid repeat unit from a set of RTX-containing proteins (Figure 1B). This led to identification of the consensus sequence GGAGNDTLY. We then sought to create a library of consensus RTX constructs consisting of 5, 9, 13, or 17 repeats of the consensus unit. Upon purification of a number of these constructs, we observed that many of them precipitated in the presence of calcium. Therefore, we decided to explore the possibility of using these consensus β roll tags (BRTs) as a tools for bioseparation, similar to the ELP system. Here, we report the use of BRTs to purify recombinant proteins. We first purified a maltose binding protein (MBP)-BRT17 fusion as a proof of principle. Then, this MBP-BRT17 construct was fused to green fluorescent protein (GFP), which was used as a reporter during initial purification experiments. We have also fused β-lactamase and a thermostable alcohol dehydrogenase (AdhD) to demonstrate the feasibility of purifying enzymatic proteins. Finally, a specific protease site was engineered downstream of the tag to show that target proteins can be fully purified by protease cleavage while retaining their activity. Materials and methods Cloning
All cloning enzymes were purchased from New England Biolabs (Ipswich, MA). All oligonucleotides were synthesized by Integrated DNA Technologies (IDT) (Coralville, IA) and all sequences are available in Supplementary Table S1. Four differently sized MBP-BRT fusions were prepared consisting of 5, 9, 13, or 17 repeats of the consensus RTX sequence (named BRT5, BRT9, BRT13, and BRT17). In order to generate the DNA fragment for BRT9, three oligonucleotides were synthesized: cons_β_1, cons_β_2, and cons_β_3. One ng each of these oligonucleotides was mixed along with the primers cons1_AvaI_F and cons9_BseRI_HindIII_R. PCR was performed and a clean product was obtained and gel extracted. This fragment was digested with AvaI and HindIII and cloned into the similarly digested pMAL_c4E vector to generate pMAL_BRT9.
To generate the BRT5 construct, pMAL_BRT9 was used as a template for PCR with the primers cons1_AvaI_F and cons5_BseRI_HindIII_R. This product was digested with AvaI and HindIII and cloned into the pMAL_c4E vector producing pMAL_BRT5.
BRT13 was produced by concatenation of four additional repeats to BRT9. Concatenations were achieved using a recursive ligation technique we developed, similar to those previously described (26, 27). This four repeat insert was amplified using primers cons1_BtsCI_F and cons4_BseRI_HindIII_R. The product was digested with. BtsCI and HindIII and then cloned into pMAL_BRT9 cut with BseRI and HindIII to yield pMAL_BRT13. BRT17 was produced analogously to BRT13, except that the reverse primer cons8_BseRI_HindIII_R was used instead of cons4_BseRI_HindIII_R.
The emGFP gene was amplified from the Invitrogen pRSET/emGFP vector using primers GFP_BseRI_F and GFP_HindIII_R. The β-lactamase gene was amplified from the pMAL_c4E vector using primers βlac_BseRI_F and βlac_HindIII_R. The AdhD gene was amplified out of pWUR85 using primers AdhD_BserI_F and AdhD_HindIII_R (28). All three of these inserts were digested with BseRI and HindIII and cloned into similarly digested pMAL_BRT17 to yield pMAL_BRT17_GFP, pMAL_BRT17_βlac and pMAL_ BRT17_AdhD.
The native enterokinase site in the pMAL_c4E vector, which sits between MBP and BRT17, was knocked out in the pMAL_BRT17_βlac and pMAL_BRT17_AdhD plasmids. Two rounds of site-directed mutagenesis were required to change the native recognition site, DDDDK, to DDGEQ, which was shown to be resistant to cleavage. A novel enterokinase recognition site was also engineered downstream of BRT17 in these constructs to allow for purification of the untagged protein of interest. Full plasmid maps of all cloned constructs are available in Supplementary Figure S1. Expression and purification
For expression and cloning, Life Technologies (Grand Island, NY) Omnimax T1 E. coli cells were used. One L cultures of TB supplemented with 100 µg/mL ampicillin and 0.2% glucose were inoculated with 10 mL of overnight culture. Cultures were grown at 37°C with shaking at 225 RPM to an approximate OD600 of 0.5 and induced with 0.3 mM IPTG. Cells harboring pMAL_BRT17 and pMAL_BRT17_βlac were allowed to express for an additional two hours and then harvested. Cultures transformed with pMAL_BRT17_GFP were transferred to a shaker at 25°C and allowed to express for an additional 16 h and then harvested as no fluorescence was observed when expressed at 37°C. Cultures transformed with pMAL_BRT17_AdhD were allowed to express at 37°C for an additional 16 h as previously reported (28). Cells were harvested after expression and resuspended in 1/20 culture volume of 50 mM tris-HCl, pH 7.4 for precipitation purification. For amylose resin purification, cells were resuspended in 1/20 culture volume of MBP column buffer (20 mM tris-HCl, 200 mM NaCl, 1 mM EDTA, pH 7.4). In both cases, cells were subsequently lysed viasonication using 15 s pulses for a total of 150 s. Lysate was then clarified by centrifugation at 15,000 g for 30 min at 4°C. For amylose resin purification, clarified lysate was diluted with five volumes of column buffer and purified as previously described (21). All subsequent steps were performed at room temperature.
For precipitation purification, clarified lysate was added to a concentrated calcium stock according to the data presented in Figure 2. For example, for precipitation of MBP-BRT17 lysate in 100 mM CaCl2, 950 µL of clarified lysate was added to 50 µL of 2 M CaCl2 solution. The sample was promptly mixed by gentle pipetting, allowed to sit at room temperature for 2 min and then centrifuged at 16,000 g in a microcentrifuge for 2 min. The supernatant was carefully removed and the pellet was resuspended in the same tris buffer by gentle pipetting. The turbid solution was centrifuged and washed for four additional cycles. For the final step, the pellet was resuspended in tris buffer with a concentration of EGTA equivalent to the original calcium concentration. Gentle pipetting was sufficient to cause the sample to redissolve as confirmed by observation and the lack of a precipitate upon subsequent centrifugation.