2National Institute for Nanotechnology, National Research Council, Edmonton, Alberta, Canada
3Faculty of Pharmacy and Pharmaceutical Sciences, University of Alberta, Edmonton, Alberta, Canada
4Department of Chemical & Materials Engineering, University of Alberta, Edmonton, Alberta, Canada
*M.S.B. and A.K.A. contributed equally to this work
The protocol described here is designed as an extension of existing techniques for creating elastin-like polypeptides. It allows for the expression and purification of elastin-like polypeptide (ELP) constructs that are poorly expressed or have very low transition temperatures. DNA concatemerization has been modified to reduce issues caused by methylation sensitivity and inefficient cloning. Linearization of the modified expression vector has been altered to greatly increase cleavage efficiency. The purification regimen is based upon using denaturing metal affinity chromatography to fully solubilize and, if necessary, pre-concentrate the target peptide before purification by inverse temperature cycling (ITC). This protocol has been used to express multiple leucine-containing elastin-like polypeptides, with final yields of 250–660 mg per liter of cells, depending on the specific construct. This was considerably greater than previously reported yields for similar ELPs. Due to the relative hydrophobicity of the tested constructs, even compared with commonly employed ELPs, conventional methods would not have been able to be purify these peptides.
Elastin-like polypeptides (ELPs) are protein-based biopolymer analogs of mammalian elastin (1). The general architecture of an ELP sequence is any number of repeats of the amino acid sequence valine-proline-glycine-X-glycine (VPGXG), where X can be any amino acid except proline (2). In elastin, this guest amino acid is typically valine. What makes this particular sequence of amino acids distinct is that it can reversibly undergo self-assembly in reaction to changes in local environmental factors such as temperature and salinity (1). The ELP amino acid chain typically exists in a random coil conformation in solution, but upon heating, it undergoes a conformational change, adopting a β-spiral conformation (3). These spirals then self-associate to form aggregates. Upon cooling, it is thought that ELPs revert to a random coil conformation and regain solubility.
This efficient and dependable method for the high-yield production and purification of short, marginally soluble, elastin-like polypeptides (ELPs) involves increasing the reliability of DNA concatemerization procedures and utilizing denaturing metal affinity chromatography to ensure these ELPs are fully soluble and recoverable from insoluble cell debris. If the ELPs are poorly expressed, this also serves as a pre-concentration step to maximize the ELP concentration before temperature cycling purification.
The exact environmental conditions under which ELPs change their conformation and solubility depend greatly upon the number of repeats of the VPGXG sequence and the side chain chemistry of the amino acid in the guest amino acid position (2). This effectively means that ELPs can be customized to aggregate under very specific conditions and can be employed in a variety of situations. ELPs are being used in a number of applications, including protein purification, environmental sensing, customizable drug delivery vehicles, and as a method of targeting the delivery of drugs to specific areas (4,–7). In addition to their flexibility, the preliminary biocompatibility of ELPs has been favorable and because they are made up of amino acids, there are few concerns about toxic degradation products (8). Another advantage of using protein-based polymers such as ELPs is that they can be produced recombinantly by inserting a DNA gene encoding an ELP into Escherichia coli, using the bacteria as molecular factories. Making these types of molecules using recombinant methods allows for an unprecedented level of control over the final product compared with traditional polymer synthesis and can allow for the straightforward incorporation of various biofunctional moieties, including targeting, cell penetrating, and enzymatically cleavable sequences.
The purification of recombinantly produced proteins is typically the largest bottleneck in their production, but with ELPs, their reversible solubility can be exploited to partially simplify purification under the correct conditions. This approach is known as inverse temperature cycling (ITC) and was first demonstrated by Meyer and Chilkoti (9). While ELPs and their potential applications are attractive and worth pursuing, there has until now been no published systematic protocol for expressing and purifying marginally soluble, short ELPs, and in our experience, standard protocols are inadequate for this subset of ELPs.
DNA sequences encoding for ELPs are by their very nature highly repetitive and are often rich with G-C base pairs. These two characteristics makes it difficult to work with these types of sequences in the lab and also prohibitively expensive to synthesize large ELP genes commercially. Techniques such as PCR, which depends on the melting and re-annealing of double-stranded DNA, may not function well, if at all, for ELP-encoding DNA sequences. Fortunately, through the clever application of specific restriction enzymes, short commercially sourced ELP genes can be concatemerized together. This process is known as recursive directional ligation (RDL) (10). Here, we describe a protocol featuring modifications to RDL designed to address technical bottlenecks experienced in our lab, including inefficient restriction enzyme cleavage due to DNA methylation, significant numbers of incorrect vector-only clones during attempted ELP sequence concatemerization, and minimal SfiI cleavage when using only one cleavage site in the expression vector modifying sequence. Our modifications greatly increase the overall efficiency of the RDL process while minimizing time- and reagent-intensive steps, such as modified-expression-vector linearization and mass colony screening.
The protein sequences created from these difficult-to-process DNA sequences can also be troublesome to deal with. Given that ELPs are generally hydrophobic regardless of their conformation, even routine lab procedures such as spin concentration, dialysis, and long-term storage become more complicated. Care must also be taken to not lose ELPs due to non-specific adsorption to sample tubes during routine handling. The ELPs covered here are highly repetitive, typically uncharged, relatively hydrophobic, and can spontaneously precipitate out of solution when stored at high concentrations at temperatures as low as 4°C for long periods of time. While these characteristics are integral to the unique reversible solubilization of ELPs, they also make their purification and downstream sample handling a delicate affair.
When ELPs have been purified successfully, for the most part the guest amino acids in these constructs fall toward the middle of Urry’s ELP hydrophobicity scale (11). While ITC purification has been sufficient for these types of constructs, it is not applicable to all ELPs. Successfully employing an ITC-only purification procedure means the ELP construct and its expression have to meet certain criteria: The ELP must not end up in the insoluble fraction of the cell lysate during the lysis procedure; the ELP must be in its soluble monomer form in the cell lysate; the phase transition needs to be triggered in the cell lysate under reasonable temperature and salinity conditions; and the protein must be expressed at a concentration high enough for a phase transition to be possible under reasonable conditions but must also be low enough to avoid significant depression of the transition temperature as well as the formation of inclusion bodies. Not all ELP purifications meet these criteria. If an ELP is poorly expressed, or if it contains guest amino acids that are significantly more hydrophilic or hydrophobic than those commonly employed in the literature, or if the ELPs are significantly longer or shorter than what is commonly used, the transition temperature of the ELP construct may be too high or too low for an ITC-only approach.
Some work has been carried out on shorter, hydrophobic ELPs using maltose binding protein (MBP) as a solubility and affinity-purification partner as a way to circumvent the above issues (12). This has, however, introduced another complication in that the MBP must then be separated from the hydrophobic ELP, an approach with varying levels of success depending on the ELP construct in question (12). More recently an ITC-only purification approach has been used to produce the ELPs without the interfering MBP, but the yields of purified protein severely limit the applications of such an approach (13). Materials and methods
A complete listing of the materials used in this study and a detailed protocol are provided in the Supplementary Material. ELP gene concatemerization and cloning
A synthetic oligonucleotide encoding for 10 repeats of VPGLG (L10) was purchased from Integrated DNA Technologies (Coralville, IA). Due to its length the oligonucleotide was provided in a pIDT-blue plasmid. This plasmid was transformed into XL10 Gold competent E. coli cells (Agilent Technologies, Santa Clara, CA), and the cells were grown in liquid culture to produce large amounts of the ELP-containing plasmid. The ELP gene was obtained from the plasmid by double digesting it with the EcoRI and HinDIII restriction enzymes (all restriction enzymes were purchased from New England BioLabs, Ipswich, MA) and purified by agarose gel electrophoresis. The gene was ligated into a pUC19 vector (Bio Basic, Ontario, Canada) that was also digested by the same restriction enzymes. Correct gene insertion was confirmed by double digestion with restriction enzymes BglI and NdeI.
Oligomerization of the ELP gene was carried out using a modified recursive directional ligation (RDL) scheme. Vector DNA was linearized using PflMI, purified using gel electrophoresis, digested again with PflMI, and dephosphorylated using Antarctic phosphatase (New England Biolabs). Insert DNA was double-digested with PflMI and BglI and purified using gel electrophoresis. The two ELP-containing genes were ligated together and transformed into XL10 Gold cells. A double digest using PflMI and BglI confirmed successful gene concatemerization. This RDL process was repeated to generate ELP genes with increasing numbers of repeats.
For the RDL-generated ELP genes to be inserted into a pET-25b(+) expression vector (EMD Millipore, Ontario, Canada), a short modifying sequence containing two SfiI cleavage sites along with the N- and C-terminal sequences of the ELP was first added. As with the L10 gene above, the synthetic oligonucleotide was purchased commercially (Integrated DNA Technologies), supplied in a vector, transformed into XL10 E. coli, and the sequence of interest was isolated using the EcoRI and NdeI restriction enzymes. The expression vector was similarly linearized and then ligated with the isolated modifying sequence. A post-ligation digest with BamHI before transformation was used to greatly reduce the number of vector-only clones. Vector modification was confirmed by single digests with either BamHI or SfiI.
To prepare for insertion of the ELP gene into the modified expression vector, the vector was digested with SfiI, purified by gel electrophoresis, digested again, and then dephosphorylated. The insert was prepared in the same manner as the inserts for RDL. The two were ligated together, and a post-ligation SfiI digest was performed before transformation to reduce the number of incorrect colonies observed after cloning. Restriction enzyme digestions and DNA sequencing confirmed the final plasmid sequences. Expression and purification
ELP expression vectors were transformed into OneTouch BL21 (DE3) E. coli (Invitrogen, Carlsbad, CA) cells, and 1 L cultures of these cells were grown in Terrific Broth (TB) (Thermo Fisher Scientific, Waltham, MA) supplemented with 100 µg/mL ampicillin (Thermo Fisher Scientific) and 10 mM L-proline (Sigma-Aldrich, St. Louis, MO). Expression was induced for 24 h using 2 mM isopropyl β-D-1-thiogalactopyranoside (Thermo Fisher Scientific).
Purification was achieved by first performing denaturing metal-affinity chromatography and eluting the ELPs using an imidazole step gradient. Buffered 8 M urea (Thermo Fisher Scientific) was used to lyse the cells and ensure the ELPs were fully soluble and free from any inclusion bodies. The urea was removed during the extensive column washing. Eluents were screened for the presence of ELPs by gel electrophoresis or by heating them to room temperature or 37°C, depending on the construct, and observing which samples turned reversibly cloudy. Samples confirmed to contain ELPs were combined and then subjected to one round of ITC for final purification. The temperature used to cause ELP aggregation varied depending on the construct. The purification was confirmed with denaturing PAGE (SDS-PAGE) and the ELP concentrations were measured by sample absorbance at 280 nm. Results and discussion
Many of the design considerations for new ELP sequences have been discussed in detail previously (10). Briefly, the design of the initial DNA sequences should take into account E. coli codon bias; that is, the frequency of the employed triplet codons should reflect the naturally occurring frequency of their corresponding tRNAs. The ends of the ELP DNA sequences should be designed such that oligomerization of the DNA eliminates the PflMI and BglI cleavage sites without introducing another cut site for a restriction enzyme employed downstream. While these restriction enzymes are used here to facilitate ELP gene oligomerization, they are not the only two that could be used. In addition to the published guidelines, if individual oligonucleotides are being used, rather than genes already contained in plasmids, the oligonucleotides should be ordered in a semi-purified state. If the oligonucleotides are ordered with only standard desalting as the purification method, there is a high probability that an undesirable heterogeneous product will be obtained because of the highly repeated sequence.
The basic workflow of RDL has been based on the work of Meyer and Chilkoti (10). Modifications have been introduced to both maximize restriction enzyme digest efficiency by minimizing interference by methylation and significantly decrease the likelihood of incorrect background colonies by maximizing vector linearization and purity. This drastically reduces the number of colonies that must be screened in order to find the correct RDL product. We found that methylation was less of a concern when using XL10-Gold E. coli as opposed to the XL1-Blue strain. Before switching host cell lines, restriction enzyme digests were usually only 30%–40% efficient. This drastically reduced the yield of linearized vector and correctly digested inserts necessary for RDL. The number of background colonies was decreased by gel purifying the digested vectors, repeating the linearization procedure, and then dephosphorylating the vector. These additional steps help to reduce the number of background colonies resulting from both uncut vector DNA and re-closed empty vectors. Without this additional effort, the number of clones containing only the vector completely eclipsed the number of correctly ligated clones. These precautions reduced the proportion of incorrect clones from ~95% to <5%. Figure 1 shows the results of an RDL colony screening procedure and clearly demonstrates the complete lack of empty vector clones. The time required to screen clones to find a correct ligation is greatly reduced by performing all of the steps above to reduce both uncut vector and religated DNA. Useful controls for the RDL cloning procedure consisting of DNA ligation reactions where the ELP insert and/or ligase enzyme are replaced by water can be used to assess the amount of empty vector that makes it through the ligation procedure. The sequence of the L10 gene used as an example in the detailed protocol is shown in Supplementary Figure S1.