ARTICLE Bioproduction of Highly Charged Designer Peptide Surfactants Via a Chemically Cleavable Coiled-Coil Heteroconcatemer Nicholas L. Fletcher,1 Nicolas Paquet,1 Ellyce L. Dickinson,2 Annette F. Dexter1 1

Australian Institute for Bioengineering and Nanotechnology, The University of Queensland, Brisbane, QLD 4072, Australia; telephone: þ61-7-3346 3199; e-mail: [email protected] 2 School of Chemical Engineering, The University of Queensland, Brisbane, QLD 4072, Australia

ABSTRACT: Designer peptides have recently attracted attention as self-assembling fibrils, hydrogelators and green surfactants with the potential for sustainable bioproduction. Carboxylate-rich peptides in particular have shown potential as salt-resistant emulsifiers; however the expression of highly charged peptides of this kind remains a challenge. To achieve expression of a strongly anionic helical surfactant peptide, we paired the peptide with a cationic helical partner in a coiled-coil miniprotein and optimized the polypeptide sequence for net charge, hydropathy and predicted protease resistance (via the Guruprasad instability index). Our design permitted expression of a soluble concatemer that accumulates to high levels (22% of total protein) in E. coli. The concatemer showed high stability to heat and proteases, allowing isolation by simple heat and pH precipitation steps that yield concatemer at 133 mg per gram of dry cell weight and >99% purity. Aspartate-proline sites were included in the concatemer to allow cleavage with heat and acid to give monomeric peptides. We characterized the acid cleavage pathway of the concatemer by coupled liquid chromatography-mass spectrometry and modeled the kinetic pathways involved. The outcome represents the first detailed kinetic characterization of protein cleavage at aspartateproline sites, and reveals unexpected cleavage preferences, such as favored cleavage at the C-termini of peptide helices. Chemical denaturation of the concatemer showed an extremely high thermodynamic stability of 38.9 kcal mol1, with cleavage decreasing the stability of the coiled coil to 32.8 kcal mol1. We determined an interfacial pressure of 29 mN m1 for both intact and cleaved concatemer at the air-water interface, although adsorption was slightly more rapid for the cleaved peptides. The cleaved peptides could be used to prepare heat-stable emulsions with droplet sizes in the nanometer range.

The present address of Nicolas Paquet is Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, QLD 4102, Australia Conflict of interest disclosure: A. F. Dexter is the inventor of patent WO 2011/116412-A1 (cited in the text). Authors contribution: The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Correspondence to: A.F. Dexter Received 3 February 2014; Revision received 2 August 2014; Accepted 18 August 2014 Accepted manuscript online 11 September 2014; Article first published online 10 October 2014 in Wiley Online Library (http://onlinelibrary.wiley.com/doi/10.1002/bit.25446/abstract). DOI 10.1002/bit.25446

242

Biotechnology and Bioengineering, Vol. 112, No. 2, February, 2015

Biotechnol. Bioeng. 2015;112: 242–251. ß 2014 Wiley Periodicals, Inc. KEYWORDS: concatemer; charge pairing; alpha helix; instability index; de novo design, acid cleavage

Introduction Designer peptides are building blocks of interest for the construction of novel sensors, fibers, hydrogels, and other high-value materials (Apostolovic et al., 2010; Hamley 2011). One promising area of research involves the use of lowmolecular weight peptides as responsive surfactants (Dexter et al., 2006). These peptides allow formation of emulsions or foams that can be rapidly collapsed on application of a switch such as a chelating agent or a change in pH (Dexter et al., 2006; Malcolm et al., 2009). Work in our group has also shown that amphipathic peptides with a high carboxylate content can be used to prepare stable emulsions in high salt. Such peptides offer potential for intravenous delivery of hydrophobic drugs, formulation of agrichemical emulsions, and clean-up of oil spills, among other possibilities. While high-value peptides such as pharmaceuticals can be produced economically by solid-phase synthesis, use of peptides for material applications is only viable if they can be produced biologically. This is a non-trivial process, as unfolded peptides expressed in a living cell are susceptible to proteolysis (Hannig and Makrides 1998). Fusion to a carrier protein has been used to improve stability (Young et al., 2012), but has the disadvantage that the target peptide only comprises a small part of the bioexpressed product (typically 3.5–15%). Yields can in some cases be improved by linking several copies of a peptide to form a concatemer (Chen et al., 2008), but this can be difficult if the target peptide is highly charged (Kim et al., 2006). In the present work, we designed a series of artificial proteins, similar in design to those developed by the Dutton ß 2014 Wiley Periodicals, Inc.

and DeGrado laboratories (Koder and Dutton 2006). In our designs, an anionic surfactant peptide is connected to a cationic partner to form a coiled-coil heteroconcatemer. The component peptides, EDP-11 and RDP-4, were sequenceoptimized for helix propensity, molecular hydropathy, and protease resistance (Guruprasad et al., 1990). Constructs with differing ratios (2:1, 2:2, and 3:2) of anionic to cationic partner were tested, and the charge-compensated 2:2 construct Het2-6 was found to express at high levels in E. coli. By making use of its high thermodynamic stability and protease resistance, we were able to develop a chromatography-free downstream process that yielded >99% pure protein at 133 mg per gram of dry cell weight. Expression constructs were designed to allow liberation of the constituent peptides by acid cleavage at aspartate-proline sites (Li et al., 2001; Zhang and Basile 2007). We carried out kinetic analysis of the acid cleavage of Het2-6 as well as determining the thermodynamic stability of both cleaved and intact Het2-6. We have also determined the surface activity of cleaved and intact concatemer and show that the product peptides can be used to prepare thermostable oil-in-water emulsions with droplet sizes in the nanometer range.

Genes were cloned into pET-48b(þ) (Novagen, Merck KGaA, Darmstadt, Germany) using NdeI and BamHI restriction sites. Constructs were verified by DNA sequencing. For expression testing of Het2-5, Het2-6, and Het2-7, DNA constructs were transformed into E. coli BL21 (DE3) and grown at 26 C in 50 mL Luria broth (LB) containing 15 mg mL1 kanamycin. Following induction with 1 mM isopropyl-b-D-thiogalactoside (IPTG), samples were taken at 2 and 4 h and overnight. To test Het2-6 expression at larger scale, a single colony was used to inoculate 50 mL starter culture in LB containing 15 mg mL1 kanamycin. This culture was then used to inoculate 2  500 mL shake flask cultures, which were grown at 32 C in the same medium. The cultures were grown to OD600 ¼ 0.6 and induced with 1 mM IPTG. Samples were taken at 0, 2, 5, and 8 h. For preparative shake flask expression, a single colony was used to inoculate 100 mL starter culture in 2  LB with 30 mg mL1 kanamycin. This culture was used to inoculate 750 mL shake flask cultures in the same medium. Cultures were grown to OD660 ¼ 1.0 at 37 C, induced with 1 mM IPTG and harvested at 4 h.

Materials and Methods

Cell pellets were resuspended in 5 mL of H2O per gram of wet cell weight and lysed by three passes through an EmulsiflexC5 valve homogenizer (Avestin, Ottawa, Canada) at 19,000 psi. Lysate was centrifuged at 20,000g for 10 min at 20 C then heated to 90 C for 10 min and centrifuged again to remove heat-precipitated cellular proteins. The supernatant was adjusted to pH 5 and centrifuged (20,000g, 10 min, 20 C). The pellet was taken up in water and the pH adjusted to nine. Magnesium sulfate was added to 1 M and the solution was heated for 1 h at 90 C then centrifuged (20,000g, 10 min, 20 C). The pellet containing nucleic acid and lipopolysaccharide was discarded and the supernatant again adjusted to pH 5 and centrifuged (20,000g, 10 min, 20 C) to recover Het2-6. The pellet was taken up in water and the pH adjusted to two.

Materials Reagents and chemicals were of analytical grade unless otherwise indicated. Water was from an Elga Purelab Classic system (Veolia, Pyrmont, Australia). Tris(hydroxymethyl) aminomethane (Tris) (99%), 2-(N-morpholino)ethanesulfonic acid (MES) (Biotech, performance certified) and guanidinium chloride (G.HCl) (99%) were from SigmaAldrich (St. Louis, MO). Magnesium sulfate (98%) was from Ajax Finechem (Thermo Fisher Scientific, Scoresby, Australia). NuPAGE 4-12% Bis-Tris gels, lithium dodecyl sulfate sample buffer, reducing agent, antioxidant and SeeBlue Plus2 Prestained Standard were from Invitrogen (Carlsbad, CA). Protein and peptide concentrations were determined by quantitative amino acid analysis (Australian Proteome Analysis Facility, Sydney, Australia). Peptides EDP-11 (H2NPG IAELEAE LSAVEAE LEAILAE LD-COOH, FW 2596), RDP-4 (H2N-PG IRALARA IRALARA VRALIRA VRDCOOH, FW 2826), and AM1 (Ac-MKQLADS LHQLARQ VSRLEHA-CONH2, FW 2473) were synthesized and purified by GenScript (Piscataway, NJ). Peptides EP-11 (H2N-PG IAELEAE LSAVEAE LEAILAE L-COOH, FW 2481) and RP-4 (H2N-PG IRALARA IRALARA VRALIRA VR-COOH, FW 2711) were synthesized and purified by Peptide 2.0 (Chantilly, VA). The final purity was >95% in each case. Bovine serum albumin (premium grade) was from Ausgenex (Molendinar, Australia). Group II base oil (Jurong 150) was from Mobil Oil Australia (Melbourne, Australia).

Downstream Processing

Liquid Chromatography-Mass Spectrometry (LCMS) LCMS used a Waters 2695 separations module with a Waters 2489 UV detector (Waters Corporation, Milford, MA). Samples at 1 mg mL1 were applied to a Jupiter 5 mm C4  300 A column (Phenomenex, Torrance, CA) at a flow rate of 0.2 mL min1, using a linear gradient of 10–100% (v/v) buffer B over 45 min. Buffer A was 0.01% (v/v) trifluoroacetic acid (TFA) and buffer B was 0.01% (v/v) TFA in acetonitrile. Eluate was injected directly into a Quattro micro API tandem quadrupole system (Waters Corporation, Milford, MA) in positive ion mode. Peptide species were quantified by integration of UV peak areas using synthetic RDP-4 as a standard. Where two species co-eluted, quantification was by integration of specific ion counts.

Cloning and Expression of Concatemers

Heteroconcatemer Cleavage and Kinetic Modeling

DNA sequences for designed proteins were codon-optimized and synthesized by GENEART (Regensburg, Germany).

Cleavage was carried out by heating Het2-6 (10 mg mL1, pH 2) to 70 C for 1–24 h. Samples were frozen at 80 C until analysis.

Fletcher et al.: Bioproduction of Designer Peptide Surfactants Biotechnology and Bioengineering

243

Pathway modeling was based on interconversion of 18 polypeptide species observed during Het2-6 cleavage, using a basis set of 25 reactions. Individual steps were assumed to occur with first-order kinetics at a fixed pH and temperature. Numerical integration of concentrations was carried out with a step size of 0.1 h. Least-squares fitting was carried out using Microsoft Excel Solver. To assist in fitting the overall kinetic model to minor species, differential weighting was applied for species present at low (M-E-L-R-L-E-L, M-E-L-R-L, L-R-L-E, LE-L, M-E-; 100% weighting), intermediate (M-E-L-R-L-E, E-LR-L-E, L-R-L-E-L-R, M-E-L-R, L-E-L-R, E-L-R, M-E; 50% weighting), or high concentrations (Het2-6, L-R, R, R-, L-E, E, E-; 10% weighting). Preparation of Uninduced E. coli Extract E. coli KC6 cells were grown in 1 L of defined medium (Zawada et al., 2003) in 2 L shake flasks at 37 C, and harvested at OD600 5–6 (6,000g, 30 min). The pellet was washed twice in cold S30 buffer (10 mM Tris acetate, 14 mM Mg2þ acetate, 60 mM Kþ acetate, pH 8.2) and resuspended in 1 mL S30 buffer per gram of wet cells. Cells were lysed by a single pass through an Emulsiflex-C5 valve homogenizer at 17,500 psi. The lysate was cleared by centrifugation (30,000g, 4 C, 1 h). Extract (est. 60 mg mL1 protein) (Goerke and Swartz 2008) was aliquotted, flash frozen, and stored at 80 C. On thawing, protease inhibitor cocktail (Sigma P8465) was added at recommended levels. Extracts were stable to proteolysis over 19 h as assessed by SDS-PAGE and electronic circular dichroism. Electronic Circular Dichroism (ECD) ECD spectra were recorded on a Jasco J-815 spectropolarimeter (Jasco, Easton, MD) in 0.1–1 mm quartz cuvettes. All measurements used a data integration time of 4 s and a band width of 1 nm. Spectra were recorded at a scan speed of 50 nm min1 and data pitch of 0.1 nm over 190–260 nm, averaged over three scans. For single wavelength measurements, samples were equilibrated for 10 min at the selected temperature; and data were recorded over 60 sec with a 5 sec pitch. ECD spectra for cleaved and intact Het2-6 (1.3 mg mL1) were recorded at 20 C in 10 mM Naþ citrate pH 3 or 10 mM Tris.HCl pH 8. For chemical denaturation studies, ellipticity at 222 nm was recorded for cleaved or intact Het2-6 (0.4 mg mL1) in 1.0–7.5 M G.HCl in 10 mM Tris.HCl pH 8, made by mixing separate stocks prepared in 0 or 8 M G.HCl. E. coli extract (est. 1.2 mg mL1 protein) in 1–6 M G.HCl was prepared by mixing separate protein stocks in 10 mM MES pH 7 containing either 0 or 7 M G.HCl. Interfacial Tension (IFT) IFT data were collected on a Dataphysics (Filderstadt, Germany) OCA20 contact angle system. Drops (ca. 25 mL) of cleaved or intact Het2-6, bovine serum albumin or AM1 (1 mg mL1 in 10 mM Tris. HCl pH 8) were formed in air

244

Biotechnology and Bioengineering, Vol. 112, No. 2, February, 2015

using a straight dosing needle in an 8 mL quartz cuvette. Measurements of buffer alone gave IFT 70.2  0.3 mN m1. IFT measurements were collected at 5 sec intervals over 10 min, averaged over three measurements. Equilibrium surface pressure values were determined from the y-intercept of a plot of surface pressure versus (time)1/2 (Dexter 2010). Emulsion Preparation and Characterization Cleaved Het2-6 (2.5 mg mL1) in 10 mM triethanolamine. HCl pH 9 was pre-mixed with 5% (v/v) Jurong 150 base oil for 1 min at 24,000 rpm using an IKA (Selangor, Malaysia) T25 digital Ultra-Turrax mixer fitted with a 24 mm dispersing tool. The resulting coarse emulsion was processed through an Emulsiflex-C5 homogenizer for 3 passes at 16,000 psi. Droplet sizing was carried out using dynamic light scattering (Zetasizer Nano ZS, Malvern Instruments, Worcestershire, UK). The sample was heated for 3 h at 90 C in a heating block and size measurements were repeated after this time.

Results and Discussion De Novo Sequence Design Anionic Peptide Anionic peptide EDP-11 was designed with an emphasis on carboxylate content and helical structure. The peptide sequence was built around a canonical a-helix structure comprising 3.6 amino acid residues per turn, giving a core 18-mer repeat (Fig. 1A). While isolated a-helices are not thermodynamically stable (Scholtz and Baldwin 1992), they can be stabilized by association with a suitable surface. If hydrophobic residues are placed at positions 1, 4, 8, 11, 15, and 18 of the 18-mer shown in Figure 1A (boxed), with hydrophilic residues populating the remaining positions, a facially amphiphilic helix is generated. Incorporating multiple carboxylate residues in a peptide of this kind yields emulsifiers with high salt resistance (unpublished experiments), particularly if these residues, such as helix-favoring glutamates (E), are close to the interface in positions 3, 5, 7, 12, 14, and 16. Amino acids for the 18-mer core sequence were selected based on high helix propensity and chemical stability (Blaber et al., 1993; O’Neil and Degrado 1990). We chose valine (V), leucine (L), and isoleucine (I) for the hydrophobic positions in the helix, emphasizing L due to its higher helix propensity. The remaining hydrophilic sites were populated with alanine (A) and serine (S). To assist product solubility, we calculated grand average of hydropathy (GRAVY) (Kyte and Doolittle 1982) values for the candidate core sequence, adjusting the sequence composition to obtain values less than 0.6. In addition, we optimized sequence patterning with the aim of improving protease resistance. To this end, we employed the Guruprasad instability index (Guruprasad et al., 1990), developed by a statistical analysis of 44 proteins found to have either long or

Figure 1. Heteroconcatemer design. A) Helical wheel diagram showing residue positions in the 18-mer peptide core, where the box indicates the hydrophobic face of the peptide, B) Helical wheel diagrams for EDP-11 and RDP-4 cores, C) Tetrameric heteroconcatemer comprising EDP-11 (white) and RDP-4 (grey) helices connected by linkers, D) Heptad-based helical wheel diagram corresponding to C), with EDP-11 (top left; bottom right) and RDP-4 (top right; bottom left).

short in vivo half-lives. By analyzing these sequences for dipeptide composition, it was possible to identify apparently stabilizing or destabilizing residue pairs. The instability index (II) is calculated by averaging the pair values for each dipeptide within a sequence and applying a scaling factor:  X   10 L1 DIWV x y ð1Þ II ¼ i iþ1 i¼1 L

where L is the sequence length, xiyiþ1 is a dipeptide and DIWV is the dipeptide instability weight value. An II value of greater than 40 is taken to predict instability. In our design, we minimized the instability index of test sequences by replacing “destabilizing” dipeptides with others having similar physicochemical properties, obtaining an 18-residue core sequence that we extended to 22 residues for greater stability (Fairman et al., 1995). Addition of end residues consistent with an aspartate-proline (DP) chemical cleavage strategy (Li et al., 2001; Zhang and Basile 2007) yielded peptide EDP-11 (PG IAELEAE LSAVEAE LEAILAE LD, MW 2596; Fig. 1B)

with a predicted charge of -8.1 at pH 7 (Fletcher et al., 2011), II of þ13.9 and GRAVY value of 0.488.

Cationic Peptide The design of a complementary charged peptide for incorporation into a heteroconcatemer used cationic arginine (R) as the main charged residue, driven by the solubilizing properties of this residue (Kyte and Doolittle 1982) and its higher oxidative stability relative to lysine. Using an optimization process similar to that described above, we obtained peptide RDP-4 (PG IRALARA IRALARA VRALIRA VRD, MW 2826; Fig. 1B). RDP-4 has a predicted charge of þ5.9 at pH 7, an II of 6.4 and GRAVY value of 0.412. Linker and Heteroconcatemer To obtain a charge-compensated structure, we designed polypeptides with alternating EDP-11 (E) and RDP-4 (R) sequences connected by a flexible linker (L), PGRGMD, which also incorporated acid-cleavable DP sites. An MD (M)

Fletcher et al.: Bioproduction of Designer Peptide Surfactants Biotechnology and Bioengineering

245

sequence was placed at the start of the polypeptide sequence for initiation of expression and subsequent removal of methionine. Three different heteroconcatemers were designed: trimeric Het2-5 (M-E-L-R-L-E), tetrameric Het2-6 (M-E-L-R-L-E-L-R), and pentameric Het2-7 (M-E-L-R-L-E-L-RL-E). Figure 1D shows EDP-11 and RDP-4 in heptad-based helical wheel projections as expected for the tetrameric coiled coil Het2-6. Values for isoelectric point (pI) (Bjellqvist et al., 1993), predicted molecular charge at pH 7, II and GRAVY were computed for each heteroconcatemer (Table I). pI, II and GRAVY were similar for all three constructs but the molecular charge varied, with Het2-6 showing the lowest charge while still maintaining an average charge per helix of at least one unit, which we consider necessary to maintain solubility in the folded protein. Bacterial Expression We tested expression of the three designed heteroconcatemers in E. coli. Trimeric Het2-5 and pentameric Het2-7 expressed poorly and were not investigated further (not shown). Failure of expression for these two constructs may be a result of excess charge or poor hydrophobic core packing. In contrast, strong expression was seen for the charge-balanced tetramer Het2-6. Figure S1 shows results following IPTG induction in E. coli. There is strong overexpression of a protein with a molecular weight close to 10 kDa, which at 8 h constitutes 22% of total protein. Based on this observation, expression was scaled up and the protein was purified for further characterization. Purification of Het2-6 Heteroconcatemer Figure 2 shows results obtained for a low-cost downstream process avoiding the use of buffer salts, protease inhibitors or chromatography resins. Lanes 2 and 3 show total and clarified cell lysate obtained from freeze-thawed cells in the absence of protease inhibitors. There is a high degree of autolysis of cellular proteins, but the expressed protein is resistant to degradation, comprising 60% of residual protein after initial processing. When clarified lysate was heated to 90 C, the remaining cellular protein aggregated and could be removed by centrifugation, while the expressed protein remained in the supernatant (Lane 4). Adjustment to pH 5, in accordance

Figure 2. SDS-PAGE analysis of Het2-6 fractions across purification steps. Lanes show: 1), 10) molecular weight marker, 2) total lysate, 3) clarified lysate, 4) heat-treated lysate, 5) pH 5 supernatant, 6) pH 5 pellet, 7) MgSO4-treated supernatant, 8) second pH 5 supernatant, 9) second pH 5 pellet.

with the expected isoelectric point of Het2-6, led to protein precipitation (Lane 5) along with removal of some nucleic acid (Table II). The protein could be redissolved at pH 9 (Lane 6) and had an estimated purity of >99%. While the initial purification steps yielded the overexpressed protein free of other cellular proteins, we found that large amounts of nucleic acid were still present (Table II). Nucleic acids appeared to associate with the expressed protein, leading to co precipitation at pH 5 (Table II). We removed a large fraction of the remaining nucleic acid by heating the expressed protein with 1 M MgSO4 at 90 C, followed by a second precipitation at pH 5. The final preparation contained nucleic acid at less than 5% (w/w) and was soluble at 50 mg mL1 protein at pH 2. Mass spectrometry gave a molecular weight of 12,878.3 (Fig. S2), matching the expected value for Het2-6 (12,877.9). The low apparent molecular weight of the protein on gel electrophoresis (7–10 kDa) appears to be an artifact of incomplete denaturation, as electronic circular dichroism Table II. Quantification of Het2-6 purification. Sample

Table I. Computed properties of designed concatemers. Concatemer a

pI Chargeb Instability Indexc GRAVYd

Het2-5

Het2-6

Het2-7

4.3 11.1 9.5 0.183

4.9 5.1 8.2 0.153

4.5 13.1 8.7 0.146

a

Following Bjellqvist et al. (1993). Charge per concatemer at pH 7, following Fletcher et al. (2011). c Following Guruprasad et al. (1990). d Following Kyte & Doolittle (1982). b

246

Biotechnology and Bioengineering, Vol. 112, No. 2, February, 2015

Clarified lysate Heat-treated lysate Supernatant at pH 5 Redissolved Het2-6 at pH 9 MgSO4-treated supernatant Second supernatant at pH 5 Redissolved protein at pH 2

Het2-6 (mg)a

Yieldb

Nucleic acid (mg)c

274 182 0 166 147 7 133

– 100% – 91% 81% – 73%

246 209 48 153 146 138 6

All values given per gram of dry cell weight. Het2-6 yield estimated from ellipticity at 222 nm. b Yield relative to initial heat precipitation step. c Nucleic acid concentration estimated by absorption at 260 nm. a

(ECD) showed that heating Het2-6 in loading buffer, as done for SDS-PAGE sample preparation, gives only a 6% loss of helical secondary structure (not shown). As Het2-6 does not possess aromatic residues suitable for UV spectroscopic analysis, we used ECD to monitor yields during purification. Measurements of raw ellipticity in the initial lysate include contributions from other proteins with helical secondary structure, but after heat treatment, Het2-6 is essentially the only protein present. We obtained consistent estimates of protein concentration using either ellipticity at 222 nm, SDS-PAGE or quantitative amino acid analysis. Quantification of Het2-6 by amino acid analysis showed that approximately 73% of the expressed protein was recovered from the lysate. We obtained 133 mg of Het2-6 per gram of dry cell weight, equivalent to 150 mg per liter of culture, which compares favorably to charge-paired constructs reported elsewhere (60–107 mg recovered peptide per liter of culture) (Kim et al., 2006; Lee et al., 1998). Notably, this volumetric yield of Het2-6 was obtained using low celldensity culture in shake flasks. It is anticipated that much higher yields could be obtained in high cell-density fermenters, as would be utilized for large-scale production (Lee 1996; Shiloach and Fass 2005). Acid Hydrolysis of Heteroconcatemer Within Het2-6 (M-E-L-R-L-E-L-R), each peptide is flanked by DP dipeptides susceptible to cleavage at acid pH with mild heating. Complete cleavage of all seven DP sites in Het2-6 will produce two copies each of EDP-11 and RDP-4, one MD peptide and three copies of the linker. We carried out cleavage at 70 C and pH 2, where the solubility of intact Het2-6 exceeds 50 mg mL1. Cleavage was monitored over 24 h using liquid chromatography-mass spectrometry (LCMS) to fractionate and assign the cleaved products (Fig. 3). We identified 18 different oligopeptide species including intact Het2-6, monomeric helical peptides and partly-cleaved intermediates. On heating of Het2-6, dimer and trimer combinations of EDP-11 and RDP-4 are observed in the first several hours. From 4 h, these species decrease in concentration, in parallel with the emergence of monomer helix species with or without linkers. Intact Het2-6 is no longer detected after 10 h, and at 24 h cleavage is largely complete, with 93% of species present at this time point corresponding to monomeric peptides with or without linker. In addition to EDP-11 and RDP-4, at longer incubation times, we found monomeric peptides in which the C-terminal aspartate has been lost in a secondary cleavage event. We designated the resulting products as EP-11 (E-) and RP-4 (R-). While cleavage C-terminal to aspartate, especially at aspartate-proline, is preferred under acidic conditions (Fig. S3A), cleavage N-terminal to aspartate is also possible, and can account for loss of the final aspartate of EDP-11 and RDP-4 (Fig. S3B; Li et al. (2001); Zhang and Basile (2007)). Initial cleavage of the Het2-6 backbone appears to occur Cterminal to aspartate, as we do not observe peptides having an N-terminal aspartate, and C-terminally truncated species

Figure 3.

Liquid chromatography elution profiles showing intact Het2-6 and products of cleavage after 6 and 24 h at 70 C. Peak positions for authentic standard peptides are indicated.

arise only late in the cleavage process. Incubation of synthetic EDP-11 and RDP-4 at pH 2 and 70 C also gives a rate of aspartate loss similar to that seen during Het2-6 cleavage (not shown). All species observed using LCMS were assigned to cleavage intermediates as described in the Supporting Information. Examination of intermediates showed that several species that would be expected to arise if cleavage occurred equally at all DP positions were either not present or present in only trace amounts, greatly simplifying kinetic analysis. Based on the observed and non-observed intermediates, we have proposed a minimal scheme for cleavage of Het2-6 to products (Fig. 4). We used this scheme to model the concentrations of cleavage intermediates with time and obtain values for individual rate constants, assuming sequential first-order reactions. Data for experimentally determined and model concentrations are given in Figure 5. Limitations in the data, in particular difficulties in obtaining integrated concentrations for several low-concentration intermediates, mean that there is uncertainty in the values of individual rate constants. Nevertheless, the overall fit

Fletcher et al.: Bioproduction of Designer Peptide Surfactants Biotechnology and Bioengineering

247

Proposed cleavage pathway for Het2-6. M is the N-terminal MD dipeptide, E is EDP-11, L is linker R is RDP-4, E is EP-11 and R is RP-4. Line thicknesses are in approximate proportion to the magnitude of the rate constants.

Figure 4.

provides interpretable results. Initial cleavage occurs primarily at the C-termini of peptide helices (M-E/L-R-L-E-L-R, M-EL-R/L-E-L-R and M-E-L-R-L-E/L-R), a preference which may be a result of helix dipole effects. Cleavage rates are similar at these three sites, with first-order rate constants of 0.100– 0.120 h1. Some cleavage occurs by reaction at two additional sites (M-E-L-R-L/E-L-R and M-E-L-R-L-E-L/R), with rate constants of 0.050 and 0.016 h1 respectively. Second-round cleavage can also partly be accounted for by cleavage at helix C-termini, as seen from reactions at M-E-L-R/L (0.498 h1), M-E-L-R/L-E-L (0.142 h1), L-R/L-E-L-R (0.171 h1), and L-R-L-E/L-R (0.090 h1). However, it becomes necessary to account for cleavage at L-E-L/R (0.154 h1), as well as the loss of N-terminal MD at M/E-L-R-L-E (0.324 h1) and M/E-L-R (0.103 h1). Subsequent steps give rise to lower molecular weight intermediates, including monomer helices with short groups attached at the helix N-terminus (L-R, L-E, M-E). For L-E and L-R, it is possible to model loss of the linker group

248

Biotechnology and Bioengineering, Vol. 112, No. 2, February, 2015

at 0.144 h1 and 0.058 h1, with a subsequent slower loss of C-terminal aspartate. For M-E, it is necessary to model an additional pathway by which the C-terminal aspartate is lost first (forming M-E-), followed by loss of MD to give E-. To our knowledge, this work represents the first attempt to quantitatively analyze the acid cleavage of a pure protein containing distinct aspartate sites. Although the local sequence at each site is similar, we found that particular sites were preferentially cleaved, apparently due to secondary structure effects. This was particularly striking for initial cleavage of intact Het2-6, where similar cleavage rates were observed at each of three sites located at the C-terminus of an a-helix, while cleavage at the four sites N-terminal to an a-helix was either slower or undetected. It appears that this is a result of helix dipole effects, possibly mediated by an effect on the pKa of the aspartate residues. At the same time, after initial cleavage, it is probable that at least some cleavage reactions occur in peptides liberated from the coiled coil,

Product Characterization Work-up of Het2-6 cleavage products at alkaline pH showed decreased solubility when cleavage was allowed to proceed for longer than 9 h at 70 C, correlating with the accumulation of end-truncated EP-11 and RP-4 peptides which possess lower solubility than full length EDP-11/RDP-4 (not shown). To minimize accumulation of EP-11/RP-4, we terminated heating at 9 h when carrying out preparative cleavage of Het2-6.

Secondary Structure Both intact and cleaved Het2-6 fold into strongly helical structures at acid and alkaline pH as judged by ECD (Supporting Information), consistent with the coiled-coil design. Thermodynamic Stability To characterize the thermodynamic stability of intact and cleaved Het2-6, we carried out chemical denaturation with comparison to a reference sample of E. coli proteins. Figure 6 shows the effects of guanidinium chloride (G.HCl) on the secondary structure of intact and cleaved Het2-6 and E. coli total protein. The secondary structure of E. coli cellular proteins is already partly lost at 1 M G.HCl at 20 C, with denaturation complete by 6 M G.HCl. In contrast, Het2-6 does not unfold fully at either 20 or 50 C with 8 M G.HCl (not shown); it was thus necessary to carry out chemical denaturation studies of intact and cleaved Het2-6 at 90 C. In

Figure 5. Time course of Het2-6 cleavage. Separate panels show species present at A) high, B) intermediate and C) low concentrations. Symbols indicate species concentrations determined by LCMS, while solid lines indicate data from the model fit.

where local unfolding of a-helices and/or the effects of local chemical context begin to be important. For example, we find that linker cleavage from L-E (0.144 h1) occurs much faster than for L-R (0.058 h1), implying an effect of local charge. We also found that the most rapid cleavage occurred for intermediates generated by a prior N-terminal cleavage (E-L-R/L-E, 1.93 h1; E/L-R-L-E, 2.08 h1; L-E/L, 1.13 h1). No reason for this difference is apparent, but it is possible that local unraveling of uncapped helix ends may play a role.

Figure 6. ECD analysis of chemical denaturation at pH 8 for intact Het2-6, cleaved Het2-6 and E. coli soluble protein. Intact (*) and cleaved () Het2-6 were monitored at 90 C and 222 nm, while E. coli soluble protein (&) was monitored at 20 C and 230 nm. Data are plotted as fraction folded (Ff) as described in the supporting information. Fits show a three-state unfolding transition for intact Het2-6, disassembly of a tetramer coiled coil for cleaved Het2-6 and non-linear regression for E. coli soluble protein.

Fletcher et al.: Bioproduction of Designer Peptide Surfactants Biotechnology and Bioengineering

249

the absence of G.HCl, the mean residue ellipticity at 222 nm at 90 C for Het2-6 is approximately 23% lower than at 20 C. While this suggests a degree of unfolding, addition of 1-4 M G.HCl does not further affect the helix content, while higher concentrations of G.HCl induce denaturation in a two-phase process. We found that cleaved Het2-6 precipitated at 90 C in the absence of G.HCl; however addition of 1 M G.HCl restored solubility and unfolding occurred progressively above 2 M G.HCl. The G.HCl concentration required to achieve 50% unfolding was 7.0 M for intact Het2-6 at 90 C, 4.7 M for cleaved Het2-6 at 90 C and 1.7 M for E. coli cellular extract at 20 C, highlighting the extreme thermodynamic stability of intact and cleaved Het2-6 relative to cellular proteins. Quantitative study of protein unfolding commonly assumes a two-state model (Scholtz et al., 2009). We attempted to model intact Het2-6 denaturation in this manner, but could not obtain a good fit. We thus used a threestate model adapted from Morjana et al. (1993), as described in the Supporting Information, where unfolding is taken to proceed through an intermediate (I) between native (N) and fully unfolded (U) states. K N!I

K I!U

N Ð I Ð U

ð2Þ

Using least squares fitting to this model we determined free energy of unfolding values of 7.7 and 31.2 kcal kcal mol1 for the N!I and I!U transitions respectively, giving a summed value of 38.9 kcal mol1 or 9.7 kcal mol1 helix1. To analyze the stability of cleaved Het2-6 using ECD data, we applied a model originally developed by Fairman et al. (1995), as described in the Supporting Information, assuming that cleaved protein retains a tetramer configuration. Using least squares fitting to this model we determined a free energy of unfolding of 32.8 kcal mol1 or 8.2 kcal mol1 helix1 for cleaved Het2-6, which is slightly lower than that obtained for intact Het2-6. The relatively small change afforded by the cleavage of linker regions suggests that Het2-6 stability derives more from the hydrophobic association of amphipathic helices, than the fact that the helices have been constrained into a single polypeptide chain. The per helix stabilities of 9.7 kcal mol1 for intact and 8.2 kcal mol1 for cleaved Het2-6 are both high for a tetramer of 23–24 residue helices. Maquette tetramers based on 27-residue a-helices had DG ¼ 6.4 kcal mol1 helix1 (Gibney et al., 1997), while Fairman et al. (1995) reported tetramer coiled coils in which the stability ranged from 3.9 kcal mol1 (21 residues) to 7.6 kcal mol1 (28 residues) to 12.9 kcal mol1 (35 residues) per helix. All reference values were obtained at room temperature. Surfactant Function To investigate the surface activity of the expressed peptides at fluid interfaces we carried out air-water interfacial tension measurements on cleaved and intact Het2-6, with the previously designed peptide AM1 (Dexter et al., 2006) and

250

Biotechnology and Bioengineering, Vol. 112, No. 2, February, 2015

Figure 7. Interfacial activity of Het2-6 (solid), cleaved Het2-6 (long dash), bovine serum albumin (dotted) and reference peptide AM1 (short dash).

native bovine serum albumin (BSA) as reference points (Fig. 7). BSA is slow to adsorb, with an equilibrium surface pressure of only 16 mN m1. Intact and cleaved Het2-6 both adsorb more readily, with an initial increase in surface pressure over 15–30 sec, followed by a steady rise to equilibrium surface pressures of 29 mN m1. AM1, which is largely unstructured in solution (Dexter and Middelberg 2007), adsorbed rapidly and gave an equilibrium surface pressure of 25.2 mN m1. As a final test of cleaved Het2-6 functionality, we prepared an oil-in-water emulsion. We determined a droplet size of 200 nm and polydispersity index of 0.27 for the emulsion as initially prepared, containing 5% (v/v) of an industrial lubricating oil. Heating to 90 C for 3 h gave only a slight increase in droplet size to 210 nm, while the polydispersity index increased to 0.35. The resulting emulsion is a candidate for a renewable, low-toxicity industrial coolant that could be separated into oil and water phases once spent (WO 2011/116412-A1), allowing potential reuse of the oil phase.

Conclusions Bioproduction of designer peptides is key to providing a lowcost and renewable source of peptides for large scale applications. High molecular charge offers functional benefits in some surfactant applications; but expression of highly charged peptides remains a challenge. In the present work, we have shown that pairing of an anionic a-helical peptide with a cationic partner in a coiled-coil miniprotein leads to high expression in E. coli. In designing Het2-6, we optimized the parameters of pI, GRAVY, and II to enhance function and expression. The expressed protein displays extremely high proteolytic and thermal stability, allowing purification to be accomplished without chromatography.

We characterized acid-catalyzed cleavage of the heteroconcatemer to monomer peptides, with kinetic analysis of the cleavage pathways. Surprisingly, we found that the cleavage of chemically similar sites occurred preferentially at the helix C-termini, an effect of secondary structure on chemical reactivity that does not appear to have been previously described. We then assessed the secondary structure and thermodynamic stability of the cleaved peptides, and found that the stability of the coiled coil remained remarkably high. The cleaved peptides adsorbed at the air-water interface more rapidly than BSA but slower than the unstructured peptide AM1, suggesting the stability of the coiled-coil structure may retard interfacial adsorption. As a final test of functionality, the cleaved peptides were used to prepare a thermostable oilin-water emulsion with droplet sizes in the nanometer range. References Apostolovic B, Danial M, Klok HA. 2010. Coiled coils: Attractive protein folding motifs for the fabrication of self-assembled, responsive and bioactive materials. Chem Soc Rev 39:3541–3575. Bjellqvist B, Hughes GJ, Pasquali C, Paquet N, Ravier F, Sanchez JC, Frutiger S, Hochstrasser D. 1993. The focusing positions of polypeptides in immobilized ph gradients can be predicted from their amino-acidsequences. Electrophoresis 14:1023–1031. Blaber M, Zhang XJ, Matthews BW. 1993. Structural basis of amino-acid alpha-helix propensity. Science 260:1637–1640. Chen YH, Zhao LX, Shen G, Cui LJ, Ren WW, Zhang H, Qian HM, Tang KX. 2008. Expression and analysis of thymosin alpha1 concatemer in escherichia coli. Biotechnol Appl Biochem 49:51–56. Dexter AF. 2010. Interfacial and emulsifying properties of designed betastrand peptides. Langmuir 26:17997–18007. Dexter AF, Malcolm AS, Middelberg APJ. 2006. Reversible active switching of the mechanical properties of a peptide film at a fluid-fluid interface. Nat Mater 5:502–506. Dexter AF, Middelberg APJ. 2007. Switchable peptide surfactants with designed metal binding capacity. J Phys Chem C 111:10484–10492. Fairman R, Chao HG, Mueller L, Lavoie TB, Shen LY, Novotny J, Matsueda GR. 1995. Characterization of a new four-chain coiled-coil: Influence of chain length on stability. Protein Sci 4:1457–1469. Fletcher NL, Lockett CV, Dexter AF. 2011. A ph-responsive coiled-coil peptide hydrogel. Soft Matter 7:10210–10218. Gibney BR, Johansson JS, Rabanal F, Skalicky JJ, Wand AJ, Dutton PL. 1997. Global topology & stability and local structure & dynamics in a synthetic spin-labeled four-helix bundle protein. Biochemistry 36:2798–2806. Goerke AR, Swartz JR. 2008. Development of cell-free protein synthesis platforms for disulfide bonded proteins. Biotechnol Bioeng 99:351– 367. Guruprasad K, Reddy BVB, Pandit MW. 1990. Correlation between stability of a protein and its dipeptide composition - a novel-approach for predicting invivo stability of a protein from its primary sequence. Protein Eng 4:155–161.

Hamley IW. 2011. Self-assembly of amphiphilic peptides. Soft Matter 7:4122–4138. Hannig G, Makrides SC. 1998. Strategies for optimizing heterologous protein expression in escherichia coli. Trends Biotechnol 16:54–60. Kim HK, Chun DS, Kim JS, Yun CH, Lee JH, Hong SK, Kang DK. 2006. Expression of the cationic antimicrobial peptide lactoferricin fused with the anionic peptide in escherichia coli. Appl Microbiol Biotechnol 72:330–338. Koder RL, Dutton PL. 2006. Intelligent design: The de novo engineering of proteins with specified functions. Dalton Trans 3045–3051. Kyte J, Doolittle RF. 1982. A simple method for displaying the hydropathic character of a protein. J Mol Biol 157:105–132. Lee JH, Minn I, Park CB, Kim SC. 1998. Acidic peptide-mediated expression of the antimicrobial peptide buforin ii as tandem repeats in escherichia coli. Protein Expr Purif 12:53–60. Lee SY. 1996. High cell-density culture of escherichia coli. Trends Biotechnol 14:98–105. Li AQ, Sowder RC, Henderson LE, Moore SP, Garfinkel DJ, Fisher RJ. 2001. Chemical cleavage at aspartyl residues for protein identification. Anal Chem 73:5395–5402. Malcolm AS, Dexter AF, Katakdhond JA, Karakashev SI, Nguyen AV, Middelberg APJ. 2009. Tuneable control of interfacial rheology and emulsion coalescence. Chemphyschem 10:778–781. Morjana NA, McKeone BJ, Gilbert HF. 1993. Guanidine-hydrochloride stabilization of a partially unfolded intermediate during the reversible denaturation of protein disulfide isomerase. Proc Natl Acad Sci USA 90:2107–2111. O’Neil KT, Degrado WF. 1990. A thermodynamic scale for the helix-forming tendencies of the commonly occurring amino acids. Science 250:646– 651. Scholtz JM, Baldwin RL. 1992. The mechanism of alpha-helix formation by peptides. Annu Rev Biophys Biomol Struct 21:95–118. Scholtz JM, Grimsley GR, Pace CN. 2009. Solvent denaturation of proteins and interpretations of the m value. In: Johnson ML, Holt JM, Ackers GK, editors. Methods in enzymology Biothermodynamics. San Diego: Elsevier Academic Press Inc. p 549-565. Shiloach J, Fass R. 2005. Growing e-coli to high cell density - A historical perspective on method development. Biotechnol Adv 23:345–357. Young CL, Britton ZT, Robinson AS. 2012. Recombinant protein expression and purification: A comprehensive review of affinity tags and microbial applications. Biotechnol. J 7:620–634. Zawada J, Richter B, Huang E, Lodes E, Shah A, Swartz JR. 2003. Highdensity, defined media culture for the production of escherichia coli cell extracts. In: Saha BC, editor. Fermentation biotechnology. Washington, DC: American Chemical Society. p 142-156. Zhang SF, Basile F. 2007. Site-specific pyrolysis-induced cleavage at aspartic acid residue in peptides and proteins. J Proteome Res 6: 1700–1704.

Supporting Information Additional Supporting Information may be found in the online version of this article at the publisher’s web-site.

Fletcher et al.: Bioproduction of Designer Peptide Surfactants Biotechnology and Bioengineering

251

Bioproduction of highly charged designer peptide surfactants via a chemically cleavable coiled-coil heteroconcatemer.

Designer peptides have recently attracted attention as self-assembling fibrils, hydrogelators and green surfactants with the potential for sustainable...
1MB Sizes 0 Downloads 5 Views