Biomacromolecules 2000, 1, 534-542
534
Reduction-Oxidation Control of β-Sheet Assembly in Genetically Engineered Silk Sandra Szela,† Peter Avtges,† Regina Valluzzi,† Stefan Winkler,† Donna Wilson,† Dan Kirschner,‡ and David L. Kaplan*,† Department of Chemical Engineering and Biotechnology Center, Tufts University, 4 Colby Street, Medford, Massachusetts 02155; and Department of Biology, Boston College, 140 Commonwealth Avenue, Chestnut Hill, Massachusetts 02167 Received May 25, 2000; Revised Manuscript Received September 4, 2000
Genetically engineered spider dragline silk protein was modified to incorporate methionines flanking the β-sheet forming polyalanine regions. The methionines could be selectively chemically oxidized and reduced. This chemical change altered the bulkiness and charge of the sulfhydryl groups, and in turn, the β-sheet forming tendencies of the polyalanine domains and solubility of the protein. The genes encoding these redesigned proteins were constructed, cloned and expressed in Escherichia coli. In the reduced state (βmercaptoethanol) the ∼25 kDa protein behaved similarly to native spider dragline silk, crystallizing into β-sheets based on diffraction analysis and appearing fibrous by TEM. The addition of the methionines into the consensus dragline silk sequence did not disrupt the normal macromolecular assembly behavior of the protein. In the oxidized state (phenacyl bromide) the protein did not form β-sheet crystals and appeared morphologically featureless based on TEM. A reduction in β-strand content was also observed upon oxidation based on FTIR and TEM analysis and confirmed by X-ray diffraction analysis. To further confirm changes in assembly behavior observed for the recombinant protein containing the methionines, a model peptide with the same repeat amino acid sequence was synthesized and characterized. Shifts in molecular weight, observed by MALDI, along with corresponding changes in crystallinity, by electron diffraction, agreed with the changes expected on activation and deactivation of the redox trigger. These results support the use of a redox trigger as a useful feature with which to control the assembly of β-sheet forming proteins. Introduction Spider dragline silk from Nephila claVipes displays mechanical properties that exceed all other known natural fibers and even most synthetic fibers.1 These observations have generated a great deal of interest in understanding the molecular basis for these properties. One direction that has been actively pursued has been the cloning, expression and study of native and synthetic spider silk genes and proteins.2-5 While recombinant proteins have been successfully generated from these genetically engineered systems, a significant limitation in these studies has been the difficulty in maintaining solubility of the purified genetically engineered silk-like proteins. Premature precipitation arises due to crystallization of the protein into β-sheet domains, rendering it insoluble in aqueous and buffered systems. This precipitation process is problematic since once the protein is crystallized it is extremely difficult to resolubilize for subsequent characterization of solution structure and properties, or for the fabrication of materials such as membranes or fibers.6 Thus, a need exists to gain control of the crystallization process so that high concentrations of silk proteins can be maintained in aqueous solution in order to study the mechanism of silk * Corresponding author.
[email protected]. † Tufts University. ‡ Boston College.
Telephone:
(617)
627-3251.
E-mail:
assembly and subsequent formation of fibers and other materials during native or artificial silk processing. Furthermore, this control of solubility will lead to options to emulate the natural silk spinning process in which high concentrations (∼30%, w/v) of protein can be maintained in an aqueous soluble state prior to spinning into a water-insoluble fiber.7-9 Spider dragline silk contains highly repetitive amino acid sequences that form both crystalline and amorphous domains in the fibers.10 On the basis of solid-state NMR analysis, polyalanine domains in the protein form the β-sheet crystals,11,12 while the remaining regions of the protein form less crystalline or amorphous domains. The alanines and other short-side-chain amino acids form antiparallel β-sheets by hydrogen bonding within each sheet. Alanine’s small size may stabilize the β-sheet crystal structure by facilitating interdigitation of small aliphatic side chains, permitting close packing in the intersheet direction.13,14 Spider and silkworm silks are semicrystalline and selectively incorporate specific repetitive motifs into β-sheet crystallites. The solid materials formed by these proteins consist of flexible amorphous domains reinforced by strong stiff crystals (β-sheets). Both the crystalline and amorphous regions contribute to the mechanical properties of spider silk, and the nature of the amorphous regions as well as the amorphous-crystalline interphase may be important contributors to the impressive mechanical properties of silks.
10.1021/bm0055697 CCC: $19.00 © 2000 American Chemical Society Published on Web 10/04/2000
Triggered Assembly of Silk Proteins
Figure 1. Methionine redox trigger to control β-sheet assembly.
We have previously cloned and expressed a partial cDNA clone of N. claVipes dragline silk, and a series of synthetic dragline silk-like genes, in Escherichia coli.3,5 However, even with successful cloning and expression, the limited solubility of the recombinant proteins suggests that alternative strategies are needed to overcome this problem. Dado and Gellman15 have shown that methionine can be introduced into proteins to act as a redox “trigger”, and this trigger can be used to study protein folding. They reported a reversible methionine side chain modification (oxidation to the more bulky and hydrophilic sulfoxide form) to control the secondary structure of an 18-residue R-helical forming peptide.15 We have redesigned our synthetic dragline silk-like encoding genes16,17 to incorporate codons for methionine flanking a region encoding five alanine residues. Data documenting a number of structural features of these proteins was previously reported.17 Details on the biochemical characteristics as well as additional morphological features for these proteins are reported here employing the redox state of the methionyl side chains to regulate the assembly of the protein. In the reduced state, the methionines (hydrophobic, relatively small) were expected to permit normal polyalanine hydrophobic domain interactions, leading to β-sheet formation, while in the oxidized state (hydrophilic, relatively large) the process would be inhibited due to a combination of increased bulkiness and hydrophilicity at the sulfoxide side chain (Figure 1). A ∼25 kDa protein was expressed and characterized in the reduced and oxidized states by SDSPAGE, TEM, FTIR, electron diffraction, and X-ray diffraction. A peptide analogue of this protein was synthesized for further studies, including MALDI and reversibility of the system. The results show that the redox state of the methionine “triggers” can be used to control the solution and solid-state behavior of β-sheet forming proteins like silks. The results also suggest that it may be possible to extrapolate this design to other β-sheet systems, such as β-amyloid and prion proteins. Materials and Methods Materials. Lysozyme, kanamycin, β-mercaptoethanol, dithiothrietol, and phenacyl bromide were purchased from Sigma. Coomassie brilliant blue R-250 was purchased from Bio-Rad. All other reagents were from major commercial suppliers and were reagent grade or better. Strain and Expression Plasmids. E. coli strain BLR(DE3) (F- ompT hsdSB (rB-mB-) gal dcm ∆(srl-recA)306::
Biomacromolecules, Vol. 1, No. 4, 2000 535
Tn10 (DE3)) (Novagen), was used as a host for cloning and expression. The silk gene construct was cloned into the BamH1 site of the expression vector pET29a (Novagen). One oligonucleotide monomer building block was 102 nucleotides long and encoded a protein of 34 amino acids (2774 Da) [SGRGGLGGQGAGMAAAAAMGGAGQGGYGGLGSQG]. The polyalanine β-sheet forming domain is in italics, and the added methionines are in bold. The expression system, pET29a(+), 5371bp, contains an N-terminal S‚Tag/ thrombin configuration for production of an S‚Tag fusion protein. The S‚Tag was used to quantify the unpurified fusion protein and for affinity purification. Biotinylated thrombin was used to cleave the S‚Tag according to the manufacturer’s protocol. VectorsspUC-link and pCR-link. A synthetic adapter was ligated into the XbaI site of pUC18 forming the ampicillin resistant shuttle vector pUC-link, following the protocol of Prince et al.3 A second vector, with chloramphenicol resistance was also constructed and termed pCRlink. The vector pCR script (Stratagene) was digested with BamH1 and dephosphorylated using CIP. The linker oligonucleotides were cut with BamH1 and ligated into pCR. Silk Gene Construction. The synthetic silk oligonucleotide monomer was cloned into the NheI/SpeI sites of either pUC-link or pCR-link. The plasmid was purified from an overnight cell culture, double digested with NheI/SpeI, dephosphorylated using CIP, and ligated into the vector with T4-Ligase. The reaction mixture was used to transform competent E. coli strain BLR(DE3) using the method of Hanahan.18 Transformants were identified by incubation on LB-plates containing the appropriate antibiotic (chloramphenicol for pCR-link and ampicillin for pUC-link). Recombinant plasmids were screened for insertion of the silk gene monomer by digestion with BamH1 and subsequent electrophoresis on 2% agarose gels. Inserts cloned in the correct orientation could be excised from the plasmid by double digestion with NheI/SpeI.3,16 The larger gene construct was prepared by shuttling the oligonucleotides between the two vectors as previously described,16 until the desired size gene was obtained. Cultivation and Purification. E. coli strain BLR(DE3) cells were grown under aerobic conditions at 37 °C in LuriaBertani media (LB) (Fisher). Glucose was added to a final concentration of 5 g/L at time of induction. Protein synthesis was induced with isopropyl-β-D-thiogalactoside (IPTG) at a final concentration of 1 mM. The cultures were harvested and centrifuged at 4 °C, 4000g for 10 min. The pellet was resuspended in a volume 3 times its weight in lysis buffer (20 mM Tris-HCl, 0.15 M NaCl, 0.1% Triton X-100) containing 200 µg/mL lysozyme and Mini Complete protease inhibitor (Boehringer Mannheim) and subsequently put through a French pressure cell press at 10 000 psi. The lysate was centrifuged at low speed (4 °C, 4000g for 5 min) to remove unlysed cells and the supernatant was then centrifuged at 4 °C, 39 000g for 20 min to remove remaining cell debris. The supernatant containing the unpurified protein was stored at -70 °C until further processing. The recombinant silk protein was purified by S‚Tag purification according to the manufacturer’s recommenda-
536
Biomacromolecules, Vol. 1, No. 4, 2000
tions (S‚Tag Purification Kit Protocol, Novagen). Purified proteins were dialyzed extensively using dialysis cassettes (MWCO ) 2000) (Pierce) or Snake Skin dialysis tubing (MWCO ) 10 000) (Pierce) against water to remove salts and then lyophilized. Peptide Synthesis. Peptide synthesis was carried out by the Tufts Protein Core Facility on an Applied Biosystems Peptide-Synthesizer. Peptides were subsequently purified using RP-HPLC. The core repeat (E)5SGRGGLGGQAGM(A)5MGGAGQYGGLGSQG(E)5 was prepared, with solubilizing blocks, (E)5, at each terminus. Redox Reactions. For the S‚Tag purified proteins methionine reduction was carried out with β-mercaptoethanol (BME) at 37 °C for 52 h while maintaining a non-oxygen atmosphere.19 For a 0.1-0.5 mg/mL solution of protein, phenacyl bromide in ethanol was used to oxidize the protein at room temperature in the dark.20 The final concentration of oxidant was 20× the stoichiometric amount of protein, based on one phenacyl bromide molecule reacting with each methionine sulfhydryl group. Additional redox reactions were evaluated using the model peptide, including various concentrations of N-chlorosuccinimide (NCS) and N-bromosuccinimide (NBS) in 0.1 M Tris buffer pH 8 for oxidation,21 reaction with 1 M DMSO in 0.5 N HCl for oxidation,22 and 0.3 M DMS in 6 N HCl for reduction.22 All reactions were carried out for 3 h at room temperature except for the reduction, which was incubated overnight at 37 °C. Protein Characterization. Total protein was measured by the bicinchoninic assay (BCA) (Pierce) against a bovine serum albumin standard (Sigma). To determine expression levels from the crude lysate, the S‚Tag Rapid Assay (Novagen) was performed according to manufacturer’s recommendations. SDS-PAGE was performed according to the standard Laemmli protocol using 4-12% Bis-Tris gels (Novex) with Coomassie brilliant blue G-250 (Biorad). All samples for SDS-PAGE were normalized to an OD600 ) 1 with 4X SDS loading buffer (Novex). For reduced gels, 0.05 M dithiothreitol or 0.1 M β-mercaptoethanol was used, unless otherwise noted. Western Blots. The S‚Tag fusion proteins were detected on blots using an S-protein alkaline phosphatase conjugate. Standards containing the S‚Tag peptide were used with molecular weights of 15, 25, 35, 50, 75, and 150 kDa. Membranes were developed with TBST (10 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% Tween 20) and 1% gelatin and then incubated with S-protein alkaline phosphatase conjugate according to the manufacturers recommendations (Novagen). Amino Acid Composition Analysis. Amino acid analysis was carried out at the Tufts University Medical School Core Protein Facility using a Waters Pico-Tag system. Proteins were hydrolyzed in 6 M HCl at 150 °C for 85 min followed by derivatization and HPLC separation and quantitation. MALDI-TOF. A Perkin-Elmer Voyager mass spectrometer (Protein Core Facility, Tufts Medical School) was used for molecular weight determinations. A 1-10 pmol sample of the protein or peptide was mixed with a 104-fold molar excess of the matrix in an aqueous 30% acetonitrile solution
Szela et al. Table 1. Characteristics of the Proteins (after Removal of the S-Tag) with the Methioine Triggers Expressed and Characterized
a
characteristic
tetramer
octamer
no. of kilobases (kb) no. of amino acids mol wt (kDa) expression level (mg/L)a
504 164 14.6 27-41
912 304 25.7 25-95
Crude cell lysate by S‚Tag assay.
containing 0.1% trifluoroacetic acid. Sinapinic acid was used as the matrix with the recombinant protein and 2,5-dihydroxybenzoic acid for the peptide. Fourier Transform Infrared Spectroscopy. FTIR studies were performed using a Bruker Equinox 55 FTIR spectrometer with an IR microscope attachment. Samples were prepared by dropping a small amount of 7 mg/mL solution onto ZnSe crystals and examining the dried drop in the FTIR microscope in transmission mode. Separate spectra were obtained for the dried solution and for the precipitates in samples where large precipitates were observed upon drying. Additional analyses were performed on saturated solutions of the reduced and oxidized proteins. Circular Dichroism. Most circular dichroism spectra were recorded on a JASCO J710 spectropolarimeter between 190 and 240 nm. Protein solutions, of 0.1-0.5 mg/mL, were prepared in Milli-Q (Millipore) ultrapure water. The solutions were passed through a 0.22 µm filter. Calibration was with (+)-10-camphorsulfonic acid (CSA), using measurements of ∆290.5 value of 2.36 M-1 cm-1 and a ∆192.5 value of -4.9 M-1 cm-1. Spectra deconvolution was performed using the Yang algorithm supplied with the instrument.23 X-ray Diffraction. X-ray diffraction experiments were performed using nickel-filtered and double-mirror focused Cu KR radiation from an Elliot GX-20 rotating anode operated at 35 kV by 25 mA. The samples were placed in siliconized thin walled capillaries (Charles Supper Co., Natick, MA). A Franks camera was used with a 200 µm spot and a specimen to film distance of 74.2 mA calibrated with calcite. Kodak DEF film was used with an exposure time of 50 h. Transmission Electron Microscopy and Diffraction. Transmission electron microscopy was performed on a JEOL 2000 Mark II transmission electron microscope (TEM) (University of Massachusetts, Amherst, MA), operated at 200 kV using Brightfield imaging and electron diffraction, imaged on Kodak DEF-5 X-ray film. A Gatan cryogenic sample holder at -160 °C was used. Results Expression. Two synthetic silk genes were constructed to encode proteins of ∼14.6 kDa (four repeats of the monomer sequence) and ∼25.7 kDa (eight repeats of the monomer sequence) (Tables 1 and 2). After expression and purification of the octamer the protein appeared as a single band at the expected molecular weight (Figure 2). Figure 2 shows the uninduced and induced cell extracts and the final purified protein by SDS-PAGE. There was no difference
Biomacromolecules, Vol. 1, No. 4, 2000 537
Triggered Assembly of Silk Proteins Table 2. Amino Acid Composition Analysis of the Predominant Amino Acids in the Tetramer and Octamer tetramer (%) amino acid glycine alanine serine arginine methionine glutamate + glutamineb
octamer (%)
expected
actual
expected
actual
38.6 17.6 6.3 5.1 5.1 8.7
28.0a
41.3 18.4 5.9 4.1 5.3 8.4
39.2 17.7 6.8 5.6 4.9 8.4
18.3 8.2 4.6 4.5 6.7
a Result based on peak response more than 10% outside of calibration range. b Amino acid analysis cannot distinguish between Glu and Gln.
Figure 3. Western blots of (A) SDS-PAGE and (B) SDS-PAGE with β-mercaptoethanol. Lane 1 ) markers; lanes 2-6 ) 0, 1, 1.5, 2, and 2.5 h after induction, respectively.
Figure 2. SDS-PAGE of cell extract during induction (lanes 2-6) and purified protein (octamer) (lane 7). Lane 1 ) markers; lanes 2-7 ) 0, 0.5, 1, 1.5, 2, and 2.5 h after induction, respectively; lane 8 ) purified protein.
Figure 4. Western blots of (A) SDS-PAGE and (B) oxidized protein on SDS-PAGE. For A: lane 1 ) markers, lanes 2 and 5 ) 0 h after induction, lanes 3 and 6 ) 1 h after induction, and lanes 4 and 7 ) 2.5 h postinduction. Lanes 2-4 are at 37 °C and lanes 5-7 at 70 °C. For B: lane 1 ) markers, lanes 2, 5, and 8 are 0 h after induction, lanes 3, 6, and 9 are 1 h after induction, and lanes 4, 7, and 10 are 2.5 h after induction. Lanes 2-4 are room-temperature incubations, lanes 5-7 are 37 °C, and lanes 8-10 are 70 °C.
observed in uninduced and induced cell extracts suggesting that the octamer is expressed at relatively low levels. The octamer was the focus for characterization of the redox trigger due to its higher molecular weight. On the basis of S‚Tag purification of the soluble and insoluble fractions, the tetramer product appeared exclusively in the soluble fraction of the cell extract. The yield of tetramer, determined per liter of cells from the S‚Tag assay at an OD600 of 1.5, ranged from 27 to 41 mg/L. The octamer was exclusively found in the soluble fraction of the cell extract. Expression levels of the octamer, based on the S‚Tag assay, varied from 25 to 95 mg/L (Table 1). Amino acid composition analysis confirmed the correct composition for the purified recombinant protein (Table 2). Solubility. The protein in reduced and oxidized forms was characterized for solubility in aqueous solution free of salts by measuring UV absorbance at 280 nm. A calibration curve was generated using the corresponding silk peptide. The oxidized protein was soluble at ∼0.48 mg/mL with no visible precipitate. After a series of dilutions the reduced form of the protein was still insoluble. Interpolation of absorbance readings for the diluted reduced protein samples indicated a concentration of ∼0.02 mg/mL, although precipitate was still present even at this dilution.
Western Blots. Although cell extract samples run by SDS-PAGE did not show overexpression of the recombinant silk protein, Western blots using an S‚Tag assay were able to detect the protein both in whole cell lysates and in the purified (uncleaved S‚Tag) form (Figure 3). As expected the uninduced protein lane and purified, S‚Tag cut lane (not shown), contained no signal (e.g., no S‚Tag protein). The induced protein lane showed a doublet at ∼30 kDa (protein plus fusion) for the octamer, suggestive of two forms in the cell extract, perhaps reduced and oxidized. The theoretical difference in molecular weight between the fully oxidized and the fully reduced form should be 272 Da. The blot suggests that although the protein is expressed in the reducing environment of the cell,24 it was not fully reduced. Other Western blot experiments indicate that in some instances either band could be eliminated by heavily reducing or oxidizing the sample extracts prior to loading the gel. Processing. To investigate whether the affinity purification could be optimized by oxidizing the protein and disrupting the β-sheet structure, the protein was oxidized prior to the S‚Tag purification step or prior to dialysis (Figure 4). For other analyses, oxidation was performed after dialysis. A maximum of 1 mg (based on the maximum binding capacity of S‚Tag β-galactosidase) of protein could be purified per batch run; however, the average binding of the octamer was
538
Biomacromolecules, Vol. 1, No. 4, 2000
Szela et al.
Table 3. FTIR Interpretations of the Reduced and Oxidized Proteins, along with a Comparison to Native Spider Dragline Silk sample (cm-1)
amide I expected amide I (cm-1) observed native spider dragline silk recombinant protein (reduced) recombinant protein (oxidized) a
β strand
silk I or random coil
other conformationsa
1625-1630 1628 31 63 not observed
1640-1650 1642 15 not observed 67
1655-1675 55 37 33
E.g., β turns, 3-fold extended helix.
0.37 mg, based on the BCA total protein assay. The low binding efficiency may be due to the conformation of the protein or the inaccessibility of the S‚Tag to the S-protein of the resin. It is also possible that up to 1 mg binds or interacts with the S-protein, but is eluted in the subsequent wash steps and is therefore lost during purification. In both instances (Figure 4, parts a and b), lanes without the expressed S-Tag do not contain the protein and regardless of incubation temperature, purified octamer as well as dimer bands are evident at 1 and 2.5 h postinduction. Conformational Studies in Solution. Fully oxidized and reduced samples of the octamer were characterized by CD.19,25 Approximately 0.15 mg/mL protein (determined from the BCA assay) in ultrapure water was characterized. The amount of R-helix and β-turn was negligible. Since any precipitates present were expected to be rich in β-sheet structure (as observed using the β-sheet FTIR absorption band at ∼1628 cm-1), the concentration of β-strand rich conformation in the supernatant solution was expected to underrepresent the proportion of β-conformation present in the protein. When samples were filtered to remove precipitates the protein primarily consisted of β-strand and random coil. The reduced form of the protein contained a higher percentage of β-sheet than the oxidized protein (46% vs 30%) and a lower content of random coil. In cases where the solutions were not filtered the β-sheet content of the oxidized and reduced proteins was not significantly different. This result, when coupled with the solubility and diffraction data supports the finding that the reduced form of the protein is less soluble due to a higher β-sheet content. These insoluble precipitates would not be detected by CD since only soluble forms of the protein would be detected and analyzed. The Yang algorithm was used in determination of secondary structure; however, it is important to recognize that databases for characterization of secondary structures of proteins by CD are based primarily on globular proteins. Thus, interpretations of secondary structure of fibrous proteins, such as those under study here, can only be taken as qualitative indications of trends. FTIR Analysis. The solutions of the proteins dried on ZnS crystals were analyzed by FTIR. In protein adsorption studies, strong conformational changes are generally associated with adsorption to highly hydrophobic surfaces from aqueous solution, while hydrophilic and mildly hydrophobic surfaces, such as ZnS, preserve the solution state conformation. FTIR spectra of dried films included both the dried solution and small particulates that may have been suspended in solution. The results (Table 3) support the significant difference between the reduced and oxidized samples, with a high proportion of β-strand (1625-1635 cm-1) and random
Figure 5. FTIR analysis of the saturated solutions of oxidized and reduced protein.
coil (1640-1650 cm-1) in the reduced and oxidized forms, respectively.17 Additional analysis of the saturated protein solutions (Figure 5) showed strong absorptions in the amide I region for both the oxidized and reduced forms of the silk. The reduced form showed absorption peaks at 1624, 1630, 1641, 1648, 1664, 1672, and 1678 cm-1 with the main intensities in the β-structural region (1624, 1630, 1672, and 1678 cm-1) and smaller contributions from random coil and turns and bends, 1641 and 1648 cm-1 and 1664 cm-1, respectively (Figure 5). The oxidized form of the silk displayed absorption peaks at 1624, 1632, 1639, 1645, 1657, 1665, 1674, and 1686 cm-1; however, the relative intensities were lower in the β-structural region and higher for random coil and helical absorption frequencies than the reduced form. Differences in spectral features compared to the dried films are not unexpected since the proteins retain more conformational freedom in the solution state. Morphology (TEM) and Diffraction Analysis (ED, XRD). Films formed from the reduced octamer at an airwater interface (see refs 26 and 27 for details on film preparation) had a banded appearance imaged by TEM. The bands alternated between light and dark with a 200 nm width (Figure 6, parts a-c). It has not been determined whether these bands are due to changes in film thickness or changes in crystallinity. The darker bands contained needlelike crystallites, approximately 200 nm long and
