Bioconjugate Chem. 2007, 18, 1619−1624
1619
Design of a Thermocontrollable Protein Complex Yoshihiko Fujita, Hisakage Funabashi, Masayasu Mie, and Eiry Kobatake* Department of Biological Information Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, 4259 Nagatsuta-cho, Midori-ku, Yokohama, Kanagawa 226-8501, Japan. Received April 11, 2007; Revised Manuscript Received June 27, 2007
It is widely recognized that stimuli-responsive nanostructures play a promising role in nanodevices for medical treatments and experimental tools. We have designed and constructed a basic structure which controls the distance between two termini domains through temperature reversibility. Our structure, shaped like a bouquet, is composed of two proteins, R-helix and elastin-like protein (ELP). R-Helices align and bundle the ELP while ELP twists and forms a fiber-like structure at warm temperatures. This ELP conformational change alters the distance between the structure termini at the site opposite the R-helix. We connected enhanced yellow fluorescent protein (EYFP) and enhanced cyan fluorescent protein (ECFP) at the structure’s two termini to evaluate the distance using fluorescence resonance energy transfer (FRET) efficiency. These proteins spontaneously formed a complex which decreased the distance between the two fluorescent proteins located at its termini, at physiologically relevant temperatures. This change was repeated with complete reversibility (n ) 5).
INTRODUCTION External modification of nanomachine function and conformation is essential. In the past, chemically based, simple molecular machines were constructed (1-3). These machines could change structural conformation in response to an external chemical or physical stimulation with complete reversibility. Natural biomotor molecules also have been discovered (4-6). However, few artificial biomaterial-based molecular machines have been constructed due to the unpredictable nature of biomaterials. To create these biomaterial-based machines, one approach was to position a new stimuli-responsive tag on existing biomaterials (7, 8). The characteristics of these biomaterials are constantly controlled by external chemical or physical stimulation. Modification of the existing natural biomolecule, by exploiting its original conformation and function, can create highly functional machines. However, the degree of design freedom is limited, as this approach highly depends on the original biomolecule conformation. An alternate approach is to design biomaterial building blocks that form biomachines using a bottom-up strategy. Many structures composed of nucleic acids (9) or peptides (10, 11) have been constructed using this technique; however, few can be controlled by external stimulation. A temperature-responsive protein complex we designed in this article contains artificial R-helices and ELP. Three artificial R-helices, A, B, and C (12, 13) (we call them helixA, helixB, and helixC, respectively, in this paper), served as self-assembly domains along with ELP, since they are small and assemble into a highly stable heterotrimer coiled coil up to approximately 80 °C. The amino acids sequence (GVGVP)n, which is one of the ELPs derived from the hydrophobic region of elastin (14, 15), shrinks under warm conditions to form fibril structures (16) and aggregations called coacervation. Since its conformational change is well-characterized and completely reversible with temperature change, we adopted (GVGVP)n as the temperatureresponsive domain. A model of the structure at high temperature is proposed as a β-spiral structure composed of three twisted *
[email protected].
(GVGVP)n (17). On the basis of this model, we expected that, when the opposite sites were bundled, the two domains fused to (GVGVP)n individually could be approximated at high temperature. Then, we constructed helixA-(GVGVP)n, helixB(GVGVP)n, and helixC-(GVGVP)n recombinant proteins, which contain the R-helix in order to align the termini of the (GVGVP)n. These proteins were expected to spontaneously form a trimer, aligning the termini by R-helices, bringing the opposite termini together at high temperatures and subsequently releasing them at low temperatures. In this study, we fused ECFP and EYFP to the helixA-(GVGVP)n and helixB-(GVGVP)n, respectively, and were able to control the distance between the ECFP and EYFP (Figure 1). This study demonstrates future possibilities of this structure becoming a thermoresponsive part for complete biomaterial-based machines which can be expressed and assembled in cells.
EXPERIMENTAL PROCEDURES Plasmid Construction. The ELP, helixA, helixB, and helixC genes were obtained by elongating partially hybridized two DNA oligonucleotides. About 20 bases at the 3′ end of one oligonucleotide were hybridized with a region at the 5′ end of another oligonucleotide, followed by a polymerase reaction. Enhanced cyan fluorescent protein (ECFP) and enhanced yellow fluorescent protein (EYFP) genes were amplified by PCR, from pECFP-N1 (BD Biosciences, Clontech) and pEYFP-actin (BD Biosciences, Clontech), respectively. All the genes were cloned on pBluescriptII SK (+) (Stratagene) and then sequenced (Beckman; CEQ2000XL). ELP tandem repeats (×18, 36, and 72) were constructed by recursive directional ligation (18) between BamHI and BglII sites. The expression plasmids for the helices (helixA, helixB, and helixC) and fluorescent proteins (ECFP and EYFP) were constructed by ligating the cloned coding DNA fragments with a modified pET28b (Novagen) vector and digested with NcoI and XhoI or BamHI and BglII, respectively. The expression plasmids for helixA-(GVGVP)nECFP,helixB-(GVGVP)n-EYFP,helixC-(GVGVP)n,(GVGVP)nECFP, and (GVGVP)n-EYFP fusion proteins were constructed by inserting each cloned gene fragment into the modified pET28b in the order described.
10.1021/bc070120x CCC: $37.00 © 2007 American Chemical Society Published on Web 08/28/2007
1620 Bioconjugate Chem., Vol. 18, No. 5, 2007
Fujita et al.
Figure 1. Schematic diagram of PCn. Upper illustration indicates a macroscopic view; lower is a microscopic view. The helix parts assemble independently and form a protein complex (PCn). Under cold conditions, the fluorescent proteins at the termini of (GVGVP)n reside away from each other due to the relaxed nature of the (GVGVP)n components. As the temperature increases and the (GVGVP)n components stabilize, the two fluorescent proteins come close together. The distance between them then decreases due to its inverse temperature transition.
Protein Expression and Purification. The expression vectors were transformed into the E. coli strain BL21(DE3). LB medium, supplemented with 20 µg/mL kanamycin, was inoculated with transformed cells from confluent precultures and incubated with shaking at 37 °C. When the OD660 reached between 0.5 and 0.6, expression was induced by adding IPTG, reaching a final concentration of 1 mM. The cells containing pET28b-helixA, pET28b-helixB, pET28b-helixC, pET28bECFP, pET28b-EYFP, pET28b-helixA-(GVGVP)n-ECFP, pET28b-helixB-(GVGVP)n-EYFP, or pET28b-helixC-(GVGVP)n were harvested after induction for 16 h at 16 °C, and centrifuged at 2000 g for 3 min at 4 °C. The cells with pET28b-(GVGVP)nECFP, and pET28b-(GVGVP)n-EYFP were harvested after induction for 6 h at 30 °C. Next, the pellets were lysed by 1×BugBuster (Novagen) including benzonase nuclease (Novagen), and the cell lysates were centrifuged at 16 000 g, for 15 min at 4 °C, to remove insoluble cellular debris. The proteins with (GVGVP)n (n g 36) were purified using the modified method of inverse transition cycling (19) and then dissolved in PBS. Other proteins were purified according to the standard metal affinity chromatography protocol of His-Select Nickel Affinity Gel (Sigma) and dialyzed with PBS using a Slide-ALyzer Dialysis Cassette (Pierce). The purified protein concentrations were determined using the BCA Protein Assay Kit (Pierce) and then stored at -80 °C. The purified proteins are listed in Figure 2.
Figure 2. Constructed proteins. helixA, helixB, and helixC are the artificial helices. The repetitive number of the GVGVP (n) was indicated at the right. Proteins are fused with polyhistidine tag. The amino acids sequences are indicated at the bottom.
Protein Complex Preparation. A protein complex with various lengths of (GVGVP)n was prepared by incubating a mixture containing equal amounts of helixA-(GVGVP)nECFP, helixB-(GVGVP)n-EYFP, and helixC-(GVGVP)n in PBS for 30 min at 4 °C. Mixtures of fluorescent proteins (ECFP and EYFP) only, without helixA and helixB ((GVGVP)n-ECFP, (GVGVP)n-EYFP, and helixC-(GVGVP)n), were also prepared as above and served as negative controls. The protein
Thermocontrollable Protein Complex
complex consisting of helixA-(GVGVP)n-ECFP, helixB(GVGVP)n-EYFP, and helixC-(GVGVP)n is named PCn (n indicates the number of the repeated GVGVP peptapeptide) in this article. Estimation of FRET Efficiency. The PC72 was prepared in 700 µL of PBS and incubated at 4 °C for 20 min. The fluorescent spectra excitation under cold conditions was measured at 425 nm excitation immediately after transfer to the spectrofluorometer (Jasco, FP-777). The sample was heated with a dryer and immediately measured in the same manner to obtain the fluorescent spectra under warm conditions. The spectra measurements were then normalized by dividing intensity from 450 to 600 nm by that at 475 nm. To estimate fluorescence resonance energy transfer (FRET) efficiency, the fluorescence of the purified protein was determined using the CYTOFLUOR Multi-Well Plate Reader series 4000 (PerSeptive Biosystems). The PCn solutions were prepared directly on a 96-well ELISA plate (Sumilon) in 200 µL of PBS and gently vortexed. After a 20 min incubation at each preset temperature, the fluorescent signal was obtained using 440/40 nm, 485/20 nm, and 530/10 nm filters for excitation, ECFP emission, and EYFP emission, respectively. At each temperature, FRET efficiency was defined as the fluorescent intensity ratio I(530 ( 10 nm)/I(485 ( 20 nm). Sequential measurements of the same sample were taken at low through to high temperatures. To test temperature reversibility, protein complexes were prepared as described above. The fluorescent signals were obtained first at 37 °C, followed by at 4 °C sequentially. This measurement was repeated five times on the same sample. Protein Complex Coacervation. The inverse phase transition temperature was determined by monitoring absorbance to quantify the turbidity of the solution as a function of temperature at 350 nm, using a UV-visible spectrophotometer DU-7500 (Beckman) equipped with a multicell thermal controller. The protein solutions were heated 1 °C every minute, from 25 °C to 55 °C. Turbidity was calculated by subtracting the absorbance at 25 °C from each absorbance reading. The inverse phase transition temperature was defined as the temperature where the liquid absorbance is midway between the two parallel lines, tangent to the slope of the data at the beginning and end of the analysis. Complex Disruption Test. Protein self-assembly was inhibited by blocking the helix region in the components (helixA(GVGVP)n-ECFP, helixB-(GVGVP)n-EYFP, and helixC(GVGVP)n) of the PCn using the respective counterpart helices, i.e., helixA and helixB for helixC-(GVGVP)n, helixB and helixC for helixA-(GVGVP)n-ECFP, and helixC and helixA for helixB-(GVGVP)n-EYFP. Aliquots of components (0.4 µM) with counterparts, whose concentrations were 0.1, 0.2, 0.4, 0.8, or 1.6 µM, were prepared and incubated for 1 h at room temperature in order to block the helix regions. Next, the blocked components were mixed in 200 µL of PBS to obtain the solutions containing 0.1 µM of each blocked component. The final concentrations of the R-helix counterparts in prepared samples were 0.075, 0.15, 0.3, 0.6, and 1.2 µM.
RESULTS AND DISCUSSION Temperature Control of the Distance between Proteins. First, we investigated the change in distance between ECFP and EYFP at various temperatures by measuring the fluorescent spectra of assembled PC72 (Figure 3A), which consists of helixA-(GVGVP)72-ECFP, helixB-(GVGVP)72-EYFP, and helixC-(GVGVP)72 as described in the Experimental Section. At low temperatures (approximately 4 °C), the fluorescent light derived from EYFP was undetected. In contrast, at high temperatures, a remarkable peak occurred around 530 nm, suggesting that FRET from ECFP to EYFP had occurred. These
Bioconjugate Chem., Vol. 18, No. 5, 2007 1621
Figure 3. (A) The normalized fluorescent spectra of PCs (helixA(GVGVP)72-ECFP, helixB-GVGVP)72-EYFP, and helixC-(GVGVP)72) under cold conditions (broken line) and warm conditions (solid line). (B) Complex FRET efficiencies at various lengths of (GVGVP), at 26 °C (open bar) and 37 °C (filled bar). The fluorescence data are the mean ( SD from four separate samples.
results indicate that the two fluorescent proteins came close at the warm temperature, as illustrated in Figure 1. The shapes of the relative fluorescent intensity did not change significantly except for the intensity at 530 nm which was emission maxima of EYFP, at low or high temperatures. That is, the FRET signal can be detected even if the light scattering of coacervation might affect the emission light from the fluorescent proteins. To optimize fluorescent protein approximation by temperature, the FRET efficiencies of PCn having various lengths of (GVGVP)n with or without the helix were tested (Figure 3B). The PCn’s (n ) 9, 18), which have a short (GVGVP)n, caused only a slight increase in FRET efficiency at 37 °C, and reached control levels. This slight increasing in FRET efficiency is due to the inactivation of the ECFP by temperature. The apparent FRET efficiency increased because both ECFP and EYFP indicated the thermal inactivation with heating, but the intensity of ECFP more rapidly decreased over that of EYFP. In contrast, an obvious FRET signal was detected in the PCn (n ) 36, 72) samples. This is not a false FRET signal due to the decrease of ECFP fluorescence light depending on the inactivation of the ECFP by temperature, because the control, which is a mixture
1622 Bioconjugate Chem., Vol. 18, No. 5, 2007
Fujita et al.
Figure 4. The FRET and turbidity profiles of the PC36 or PC72 with (filled diamond) or without (open square) helixA and helixB regions. (A) FRET profiles under the same conditions as (C). (B) FRET profiles under the same conditions as (D). The fluorescence data are the mean ( SD from four separate samples. (C) Turbidimetric profiles of the PBS solutions containing 0.1 µM PC36 with or without helixA and helixB regions. (D) Similar data as (C) but in a PC72 solution.
of ECFP and EYFP, did not cause an obvious increase of the signal. The effect of thermal inactivation of fluorescent proteins on the FRET signal may exist, but only slightly. In addition, FRET efficiency of the PC72 was higher than PC36. This result suggests that long (GVGVP)ns twist and form more stable fiber structures at 37 °C than shorter sequences. This coincides with research on the relationship between coacervation transition temperature and length of (GVGVP)n (20). Comparison of FRET with Turbidity Profiles. When coacervation occurs, FRET increases derived from intercomplex proximity must be taken into account. To determine whether the increase of FRET is due to the intercomplex proximity by coacervation or intracomplex proximity due to a protein complex conformational change, we compared FRET and the turbidity of solutions containing 0.1 µM of helixA-(GVGVP)n-ECFP, helixB-(GVGVP)n-EYFP, and helixC-(GVGVP)n (PCn, n ) 36 or 72) with solutions containing 0.1 µM of (GVGVP)nECFP, (GVGVP)n-EYFP, and helixC-(GVGVP)n (nonhelixPCn, n ) 36 or 72). The former protein group contains the R-helix region that self-assembles, while the (GVGVP)n-ECFP and (GVGVP)n-EYFP in the latter group do not contain these helices. Improvements in FRET efficiency have been detected in PCn solutions, while no changes have been detected in non-helixPCn (Figure 4A,B). Although differences in FRET efficiency have been observed in solutions with or without helixA and helixB, only small differences in the turbidimetric profiles have been detected (Figure 4C,D). The FRET derived from intercomplex proximity by the coacervation was expected to be dependent on the turbidity regardless of the helices. Thus, these results suggest that increases in FRET efficiency were derived from intracomplex proximity, since increased FRET efficiency was not detected in the samples without an R-helix that formed coacervates. The fact that FRET efficiency of PC72 with helixA and helixB increased throughout the higher temperature (>35 °C), despite the turbidity, increased a little supports the hypothesis. Presumably, the FRET efficiency is mainly dependent on the stability of the microscopic structure formed by (GVGVP)n. It is notable that the temperature at which the FRET
signal rises shifts to lower when the longer (GVGVP)n is used. This indicates that the distance between ECFP and EYFP can be controlled at various temperatures by changing the length of (GVGVP). The small difference between the turbidity profiles of PC72 with or without helixA and helixB was speculated as follows. ELP molecules including GVGVP need three steps before coacervation (21). First, ELP changes from random structure into a β-turn-like structure. Second, ELPs associate with each other and form a microscopic structure which includes just a few ELP molecules. Third, the microscopic structures assemble into macroscopic aggregation. They suggest that the process from step 1 to step 2 is stoichiometric; that is, this process is dependent on the concentration of ELP. In the case of a protein complex with helixA and helixB, the microscopic concentration of (GVGVP)n was high because a helix put one (GVGVP)n close to other (GVGVP)n’s conjugated with counterpart helices. This proximity effect promoted the process from step 1 to step 2 of (GVGVP)72. Consequently, the temperature of the coacervation decreased, and intensity increased more rapidly. On the other hand, the microscopic structure of (GVGVP)36 is less stable than that of (GVGVP)72 because the free energy difference between the monomeric state and the associated state is dependent on the length of ELP (longer ELP indicates larger free energy difference). Thus, even if the microscopic concentration increases by proximity effect, (GVGVP)36 could form fewer microscopic structures than (GVGVP)72, and we could not detect the small change as turbidity. Complex Disruption by Counterparts. To confirm whether the FRET signal disappeared when the complex was inhibited, the R-helix region in the PC72 was blocked by monomer R-helice binding. FRET efficiencies of the mixed solutions containing the components of PC72 and R-helices decreased with increasing concentrations of R-helices at 37 °C (Figure 5). FRET efficiency attenuation at low counterpart concentrations (0-0.1 µM) was greater than high concentrations (>0.2 µM). This nonlinear inhibition can be explained through interactions between the counterpart helices. Counterpart helices’ binding to the helix regions of the PCn components was nearly complete
Thermocontrollable Protein Complex
Figure 5. FRET efficiencies of the PC72 containing various concentrations of the counterpart R-helices. FRET efficiencies at 37 °C (filled diamond) and 4 °C (open square) as indicated, where each component of the PC72 was 0.1 µM.
Figure 6. Reversibility of 0.1 µM of PC72. FRET efficiencies of PC72 were sequentially measured at 4 °C (open bar) and 37 °C (filled bar) five times.
at low concentrations. In contrast, when there were many counterpart helices, they bound with other monomeric helices and formed stable complexes which were not capable of blocking the components. Thus the exceeded helices could not block the components efficiently. This inhibition indicates that domain proximity at the PCn termini requires helices that permit individual protein bundling. Reversibility Test. To control PCn termini domain function, conformational reversibility is necessary. Therefore, we repeatedly measured FRET efficiencies, alternating between 4 °C and 37 °C as illustrated in Figure 6. FRET efficiency increased at 37 °C and decreased at 4 °C in all cycles. Even after four cooling and warming cycles, FRET efficiency consistently decreased at 4 °C. This result suggests that changing temperature could control the distance between the domains positioned at the PC72 termini in a reversible manner. An upward trend in the FRET efficiency is explained by two possibilities. The first is the difference between thermal inactivation rates of ECFP and EYFP described above. The second is the incorrect homocomplex of helices. Some protein complexes might contain only ECFP or EYFP, such as a complex containing helixA(GVGVP)n-ECFP, helixA-(GVGVP)n-ECFP, and helixC(GVGVP)n, at the start condition, because the same helix can form a homodimer or trimer coiled coil (12, 13). During thermal cycling, these incorrect complexes were substituted to assemble correct heterotrimer complexes. Thus, the upward trend of the FRET efficiency should have been observed.
CONCLUSION Our designed recombinant proteins, helix-(GVGVP)n, spontaneously formed a trimeric complex. The FRET signal from EYFP indicated that its temperature-induced conformational change promoted ECFP and EYFP proximity. Moreover, this conformational change was reversible. This complex, although composed of amino acids, behaved much like a machine. Thus, this complex has the potential to function as basic components
Bioconjugate Chem., Vol. 18, No. 5, 2007 1623
in the construction of future biomaterial-based machines. In this study, fluorescent proteins modeled the functional domain. Alternate proteins could be used in their place, which are activated by multimerizaton or complementation. Thus, we have the ability to control multimeric proteins as typified by DNA binding proteins or split enzymes by altering the external temperature. A problem of the undesirable coacervation between complexes must be solved to use it in liquid because the steric hindrance is expected. This problem may be overcome by modifying the (GVGVP)n or fixing complexes on a solid structure, such as a cell membrane. Although a simple heterotrimer coiled coil was used in this paper in order to distinguish between ECFP and EYFP, and indicate whether proximity occurs or not, we can replace them with not only multimeric helices but also transmembrane helices (22). We also have to consider undesirable effects on cells when this thermal control is applied to cells. It is impossible to completely remove the effects on cells in principle as long as the thermal control is used; the effects, however, will reduce when the activitycontrollable temperature is set around 37 °C by an appropriate length of (GVGVP)n. Therefore, this approach has an advantage in the situation where the effects on cells are well-known or simplicity, speed, or reversibility is required. Our adjustable protein complexes can be expressed and formed in cells, since they are composed of amino acids. Once the plasmids coding this protein complex were introduced into cells, the additional harmful injection is unnecessary every time we would like to activate the interest function. Therefore, this proposed method may control cellular function without causing additional damage.
LITERATURE CITED (1) Badjic, J. D., Balzani, V., Credi, A., Silvi, S., and Stoddart, J. F. (2004) A molecular elevator. Science 303, 1845-1849. (2) Muraoka, T., Kinbara, K., Kobayashi, Y., and Aida, T. (2003) Light-driven open-close motion of chiral molecular scissors. J. Am. Chem. Soc. 125, 5612-5613. (3) Norikane, Y., and Tamaoki, N. (2004) Light-driven molecular hinge: a new molecular machine showing a light-intensity-dependent photoresponse that utilizes the trans-cis isomerization of azobenzene. Org. Lett. 6 (15), 2595-2598. (4) Noji, H., Yasuda, R., Yoshida, M., and Kinosita, K. (1997) Direct observation of the rotation of F1-ATPase. Nature (London) 386, 299-302. (5) Finer, J. T., Simmons, R. M., and Spudich, J. A. (1994) Single myosin molecule mechanics: piconewton forces and nanometer steps. Nature (London) 368, 113-119. (6) Suzuki, H., Yonekura, K., and Namba, K. (2004) Structure of the rotor of the bacterial flagellar motor revealed by electron cryomicroscopy and single-particle image analysis. J. Mol. Biol. 337, 105113. (7) Liu, H., Schmidt, J. J., Bachand, G. D., Rizk, S. S., Looger, L. L., Hellinga, H. W., and Montemagno, C. D. (2002) Control of a biomolecular motor-powered nanodevice with an engineered chemical switch. Nat. Mater. 1, 173-177. (8) Reiersen, H., and Rees, A. R. (1999) An engineered minidomain containing an elastin turn exhibits a reversible temperature-induced IgG binding. Biochemistry 38, 14897-14905. (9) Yan, H., Park, S. H., Finkelstein, G., Reif, J. H., and LaBean, T. H. (2003) DNA-templated self-assembly of protein arrays and highly conductive nanowires. Science 301, 1882-1884. (10) Holowka, E. P., Pochan, D. J., and Deming, T. J. (2005) Charged polypeptide vesicles with controllable diameter. J. Am. Chem. Soc. 127, 12423-12428. (11) Ellerby, H. M., Lee, S., Ellerby, L. M., Chen, S., Kiyota, T., del Rio, G., Sugihara, G., Sun, Y., Bredesen, D. E., Arap, W., and Pasqualini, R. (2003) An artificially designed pore-forming protein with anti-tumor effects. J. Biol. Chem. 278, 35311-35316.
1624 Bioconjugate Chem., Vol. 18, No. 5, 2007 (12) Nautiyal, S., Woolfson, D. N., King, D. S., and Alber, T. (1995) A designed heterotrimeric coiled coil. Biochemistry, 34, 1164511651. (13) Nautiyal, S., and Alber, T. (1999) Crystal structure of a designed, thermostable, heterotrimeric coiled coil. Protein Sci. 8, 84-90. (14) Sciortino, F., Prasad, K. U., Urry, D. W., and Palma, M. U. (1993) Self-assembly of bioelastomeric structures from solutions: meanfield critical behavior and Flory-Huggins free energy of interactions. Biopolymers 33, 743-752. (15) Urry, D. W., and Parker, T. M. (2002) Mechanics of elastin: molecular mechanism of biological elasticity and its relationship to contraction. J. Muscle Res. Cell Motil. 23, 543-559. (16) Bressan, G. M., Pasquali-Ronchetti, I., Fornieri, C., Mattioli, F., Castellani, I., and Volpin, D. (1986) Relevance of aggregation properties of tropoelastin to the assembly and structure of elastic fibers. J. Ultrastruct. Mol. Struct. Res. 94, 209-216. (17) Urry, D. W. (1997) Physical chemistry of biological free energy transduction as demonstrated by elastic protein-based polymers. J. Phys. Chem. B 101, 11007-11028. (18) Meyer, D. E., and Chilkoti, A. (2002) Genetically encoded synthesis of protein-based polymers with precisely specified mo-
Fujita et al. lecular weight and sequence by recursive directional ligation: examples from the elastin-like polypeptide system. Biomacromolecules 3, 357-367. (19) Meyer, D. E., Trabbic-Carlson, K., and Chilkoti, A. (2001) Protein purification by fusion with an environmentally responsive elastinlike polypeptide: effect of polypeptide length on the purification of thioredoxin. Biotechnol. Prog. 17, 720-728. (20) Meyer, D. E., and Chilkoti, A. (1999) Purification of recombinant proteins by fusion with thermally-responsive polypeptides. Nat. Biotechnol. 17, 1112-1115. (21) Yamaoka, T., Tamura, T., Seto, Y., Tada, T., Kunugi, S., and Tirrell, D. A. (2003) Mechanism for the phase transition of a genetically engineered elastin model peptide (VPGIG)40 in aqueous solution. Biomacromolecules 4, 1680-1685. (22) Bilgic¸ er, B., and Kumar, K. (2004) De novo design of defined helical bundles in membrane environments. Proc. Natl. Acad. Sci. U.S.A. 101, 15324-15329. BC070120X
