Structural Properties of an Artificial Protein That Regulates the

Feb 20, 2007 - John L. Kulp, III,Tamiko Minamisawa,Kiyotaka Shiba,Margaret Tejani, .... Chiou, Jinghua Guo, Susan N. Coppersmith, and P. U. P. A. Gilb...
0 downloads 0 Views 284KB Size
Download PDF
Langmuir 2007, 23, 3857-3863

3857

Structural Properties of an Artificial Protein That Regulates the Nucleation of Inorganic and Organic Crystals John L. Kulp, III,† Tamiko Minamisawa,‡ Kiyotaka Shiba,‡ Margaret Tejani,† and John Spencer Evans*,† Laboratory for Chemical Physics, New York UniVersity, New York, New York 10010, and Department of Protein Engineering, Cancer Institute, Japanese Foundation for Cancer Research, and CREST, JST, Koto, Tokyo 135-8550 Japan ReceiVed August 17, 2006. In Final Form: December 26, 2006 Technological advances have facilitated the generation of artificial proteins that possess the capabilities of recognizing and binding to inorganic solids and/or controlling nucleation processes that form inorganic solids. However, very little is known regarding the structure of these interesting polypeptides and how their structure contributes to functionality. To address this deficiency, we report structural investigations of an artificial protein, p288, that self-assembles and controls the nucleation of simple salts and organic compounds into dendrite-like crystals. Under aqueous conditions at low pH and in the presence of high salt, p288 is conformationally labile and exists primarily as a random coil conformer in equilibrium with other undefined secondary structures, including polyproline type II and β turn. We note that p288 can fold into either a partial β strand (at neutral pH) or a predominantly R helical (in the presence of TFE) conformation. Solid-state 13C-15N NMR experiments also reveal that p288 in the lyophilized, hydrated state possesses some degree of nonrandom coil structure. These results indicate that p288 is conformationally labile but can undergo conformational transitions to a more stable structure when water solvent loss/displacement occurs and protein concentrations increase. We believe that conformational instability and the ability to adopt different structures as a function of different environmental conditions represent important molecular features that impact p288 supramolecular assembly and crystal nucleation processes.

Introduction In nature, the nucleation and growth of inorganic solids (biominerals) is sometimes controlled by proteins.1-5 Using these protein-based biological systems as models, genetic-based techniques6-10 have emerged wherein repetitive polypeptide sequences have been created and used to construct artificial proteins that recognize inorganic solids6-10 or control nucleation.6,7 These artificial proteins represent potentially interesting model systems for studying and understanding the impact of primary and secondary structure on crystal recognition and nucleation activity. In one particular example, a micro genebased technology11 was employed to select a protein, p288 (116 * To whom correspondence should be [email protected]. † New York University. ‡ Japanese Foundation for Cancer Research.

addressed.

E-mail:

(1) (a) Lowenstam, H. A.; Weiner, S. On Biomineralization; Oxford Press: New York, 1989. (b) Mann, S.; Webb, J.; Williams, R. J. P., Eds. Biomineralization: Chemical and Biochemical PerspectiVes; VCH: Weinheim, Germany, 1989. (2) Levi-Kalisman, Y.; Falini, G.; Addadi, L.; Weiner, S. J. Struct. Biol. 2001, 10, 4372-4337. (3) Weiss, I. M.; Kaufmann, S.; Mann, K.; Fritz, M. Biochem. Biophys. Res. Commun. 1999, 267, 17-21. (4) Weiss, I. M.; Tuross N.; Addadi, L.; Weiner, S. J. Exp. Zool. 2002, 293, 478-491. (5) Michenfelder, M.; Fu, G.; Lawrence, C.; Weaver, J. C.; Wustman, B. A.; Taranto, L.; Evans, J. S.; Morse, D. E. Biopolymers 2003, 70, 522-533. Ibid 2004, 73, 299. (6) Shiba, K.; Honma, T.; Minamisawa, T.; Nishiguchi, K.; Noda, T. EMBO Rep. 2003, 4, 148-153. (7) (a) Sarikaya, M.; Tamerler, C.; Jen, A. K.-Y.; Schulten, K. S.; Baneyx, F. Nat. Mater. 2003, 2, 577-585. (b) Brown, S. Nat. Biotechnol. 1997, 15, 269272. (c) Brown, S. Proc. Natl. Acad. Sci. U.S.A. 1992, 89, 8651-8655. (8) Naik, R. R.; Brott, L.; Carlson, S. J.; Stone, M. O. J. Nanosci. Nanotechnol. 2002, 2, 1-6. (9) Gaskin, D. J. H.; Starck, K.; Vulfson, E. N. Biotechnol. Lett. 2000, 22, 1211-1216. (10) (a) Whaley, S. R.; English, D. S.; Hu, E. L.; Barbara, P. F.; Belcher, A. M. Nature 2000, 405, 665-668. (b) Lee, W. S.; Mao, C.; Flynn, C. E.; Belcher, A. M. Science 2002, 296, 892-895. (11) Shiba, K.; Takahashi, Y.; Noda, T. J. Mol. Biol. 2002, 320, 833-840.

AA, MW ) 15 162 Da, pI ) 8.5, Figure 1), that induces dendritic or periodic crystallization in evaporative in vitro assays.6 Although not highly charged, this protein was found to be soluble and monomeric at pH e 4.0, and self-assembly and crystal nucleation are noted in solution at pH 4.0 or above.6 It is believed that the protein forms supramolecular assemblies prior to crystal nucleation and that it is the supramolecular assembly that controls crystal growth via a diffusion limiting mechanism.6 The assembly and nucleation capabilities of p288 have been attributed to the repetitive sequence blocks within the protein (Figure 1), which, in turn, are suspected of forming repetitive structural motifs within this protein.6 However, very little is known regarding the structure of p288 and the relationship between p288 structure and the protein assembly or nucleation processes. To achieve a better understanding of the p288 structurefunction relationship, we investigated the conformation of this artificial protein under conditions similar to those utilized in in vitro crystallization assays.6 We find that the p288 protein exists as an unstructured, conformationally labile polypeptide, similar to the structural state found in other biomineral associated polypeptide sequences.1-5,12,13 The unfolded state of p288 is maintained over a range of NaCl concentrations, indicating that high salt conditions, such as those found in the crystallization assay systems, do not induce conformational transitions in this protein.6 However, at neutral pH, or in the presence of the organic solvent 2,2,2-trifluoroethanol (TFE) at pH 2.0, we find that the conformation of p288 can be stabilized into a more structured (12) (a) Kim, I. W.; Morse, D. E.; Evans, J. S. Langmuir 2004, 20, 1166411673. (b) Wustman, B. A.; Morse, D. E.; Evans, J. S. Biopolymers 2004, 74, 363-376. (c) Wustman, B. A.; Weaver, J. C.; Morse, D. E.; Evans, J. S. Langmuir 2003, 19, 9373-9381; Ibid 20, 277. (13) (a) Du, C.; Falini, G.; Fermani, S.; Abbott, C.; Moradian-Oldak, J. Science 2005, 307, 1450-1454. (b) Goto, Y.; Kogure, E.; Takagi, T.; Aimoto, S.; Aoba, T. J. Biochem. 1993, 113, 55-60. (c) Aichmeyer, B.; Margolis, H. C.; Sigel, R.; Yamakoshi, Y.; Simmer, J. P.; Fratzl, P. J. Struct. Biol. 2005, 151, 239-249.

10.1021/la062442f CCC: $37.00 © 2007 American Chemical Society Published on Web 02/20/2007

3858 Langmuir, Vol. 23, No. 7, 2007

Figure 1. Primary sequence of p288. Magenta and blue regions indicate the two integral sequence repeats and their location within p288. Taken from ref 6.

state (i.e., R helix in the presence of TFE, βstrand at pH 7.5). Similarly, in the bulk, hydrated, lyophilized state, solid-state NMR experiments reveal that p288 possesses some degree of nonrandom coil secondary structure. These findings indicate that the structure of this protein is inherently unstable in aqueous solution but can achieve stabilization and adopt different types of secondary structures under certain conditions, and this instability may be a driving force for p288 supramolecular assembly formation. Materials and Methods Preparation of p288 Protein. Protein 288 was expressed from plasmid pYT288 in Escherichia coli, and protein preps were purified as described previously.6,14 The 13C-15N-labeled p288 was prepared by using BIOEXPRESS-MIN (Cambridge Isotope Laboratories, Inc., Andover, MA). Previous work6 used STBL2 (DE3) as a host E. coli strain for expressing p288, but we noticed that the cells produced little amount of p288 in the synthetic medium. From pilot tests with several strains, we found that E. coli XL1Blue (DE3), which was constructed by lysogenizing λDE3 phage (Novagen, Madison, WI) into XL1Blue (Stratagene, La Jolla, CA) (Y. Takahashi, unpublished), expressed a fair amount of p288 in the conditions used. The pYT288/ XL1Blue (DE3) was first grown in Luria-Bertani medium14 containing 50 µg of carbenicillin (Sigma, St. Louis, MO) at 37 °C for 12 h, and then 1 mL of the culture was re-inoculated into 1 L of synthetic labeling medium. When OD660 reached 0.35 absorption units, isopropyl-β-D-thiogalactopyranoside (IPTG, TAKARA Co., Kyoto, Japan) was added at a final concentration of 0.5 mM to induce the expression of p288. After 3 h of incubation, cells were harvested, and the protein was purified by using TALON resin (Clontech, Palo Alto, CA) under the denaturing conditions as described previously.6 The eluted proteins (approximately 6 mL) were dialyzed against 500 mL of 50 mM Tris-acetate, pH 4.0, 100 mM NaCl, and 1 mM EDTA 3 times and then dialyzed against 500 mL of 1% formic acid 3 times at 4 °C. The dialyzed sample was concentrated to the volume of 1.8 mL by using Centriprep (cut-off MW ) 10 000, Millipore Co., Bedford, MA) and was lyophilized. From 1 L of bacterial culture, approximately 10 mg of the labeled p288 was purified. Circular Dichroism Experiments. CD spectra were obtained for p288 using an AVIV 60 CD Spectrometer (60DS software version 4.1t). The CD spectrometer was previously calibrated with d10camphorsulfonic acid. All CD experiments were performed at 5 °C to slow down conformational exchange processes in this protein. The protein was dissolved and diluted to final concentrations of 2, 4, 8, 16, 20, and 50 µM in either 100 µM NaH2PO4, pH 2.0, or 50 mM Tris-acetate/100 mM NaCl nucleation buffer, pH 4.0. For studies conducted at pH 7.5, we utilized final concentrations of p288 of 2.5 and 5.0 µM in 100 µM Tris-HCl buffer, pH 7.5. Under these conditions and protein concentrations, turbidity measurements at 380 nm revealed no evidence of protein precipitation (data not shown); however, note that the presence of precipitation was observed for (14) Sambrook, J.; Fritsch, E. F.; Maniatis, T. Molecular Cloning: A Laboratory Manual, 2nd ed.; Cold Spring Harbor Laboratory: Cold Spring Harbor, NY, 1989.

Kulp et al. p288 at protein concentrations >5 µM at pH 7.5. This was also confirmed using dynamic light scattering measurements. For polyproline type II CD temperature studies (5-80 °C in 5° increments), we utilized 8 µM in 100 µM NaH2PO4, pH 2.0. Similarly, NaCl concentration experiments (50 µM Tris-acetate containing either 50, 100, 250, 500, or 750 mM NaCl) were performed on a 8 µM p288 sample. Finally, 2,2,2-trifluoroethanol (TFE) stabilization experiments utilized p288 protein stock solutions, TFE (99.8%, Acros Chemicals, Geel, Belgium), NaH2PO4 buffer stock, and p288 stock solutions to make final p288 concentrations of 8 µM in 100 µM NaH2PO4, pH 2.0, with specific volume percentages of TFE (i.e., 0, 5, 10, 20, 30, 40, 50, 75, and 90% v/v). For all spectra, wavelength scans were conducted from 260 to 185 nm with appropriate background buffer subtraction, using a total of 3 scans with a 1 nm bandwidth and 0.5 nm/s scan rate with 1 s averaging. In all CD spectra, the mean residue ellipticity [θM] is expressed in deg cm2 dmol-1.5,12 Solution-State Nuclear Magnetic Resonance Experiments. We encountered solubility limitations with p288 at pH 4.0 and 7.5 under typical conditions required for high field solution NMR spectroscopy (i.e., protein concentrations >50 M) as evidenced by turbidity measurements at 380 nm (data not shown). For this reason, solution NMR experiments were performed on 94 µM U-13C-15N labeled p288 samples at 5 °C, pH 2.0, conditions that provided ideal solubility and polypeptide stability for NMR. NMR experiments were acquired on Bruker narrow bore AVANCE-800 and 700 NMR spectrometers outfitted with 5 mm four-channel x,y,z-PFG cryoprobes. Proton, carbon, and nitrogen spin system assignments for the aqueous p288 monomeric samples were obtained at 800 MHz using triple resonance HNCACB experiments.15 For comparative studies involving the monomeric pH 2.0 aqueous and pH 2.0/90% v/v TFE p288 samples, we performed 15N-1H, 13C-1H HSQC,16 and 15N-1H HSQCTROSY, NOESY experiments17 at 700 MHz. Acquisition and processing parameters are provided in the appropriate figure legends. All solution NMR data processing was performed using the Bruker TOPSPIN software, and spectral assignments were obtained using the Sparky software package.18 1H and 13C NMR chemical shifts were referenced via internal d4-TSP and tetramethylsilane (TMS), respectively. 15N NMR chemical shifts are referenced from external 15NH Cl, with the center of the 15N frequency axis set to 116.7 ppm. 4 Solid-State NMR Experiments. We were interested in performing solid-state NMR experiments on p288 aggregates rescued from crystallization assays. However, given the gel-like state of aggregated p288 and the associated bulk water, high -speed magic-angle spinning of this type of sample poses numerous problems (e.g., expelled water, gel collapse) that could have a serious impact on the resulting dataset. Hence, we elected to perform NMR experiments on a hydrated, uniformly labeled 13C/15N p288 lyophilized sample (5 mg) that represents the bulk state and qualitatively mimics the situation that occurs when evaporative loss takes place in p288 in vitro assays, triggering higher protein concentrations and subsequent assembly.6 Hydration of the lyophilized sample was achieved by adding a minimal volume (5 L) of 50 mM Tris-acetate/100 mM NaCl, pH 4.0, nucleation buffer to the sample to allow swelling but prevent excess moisture buildup that would compromise high speed magic-angle spinning techniques. This mixture was incubated for 12 h at 25 °C, and then the mixture was packed into a 5 mm Bruker MAS rotor for NMR studies. Solid-state heteronuclear experiments (15) (a) Wittekind, M.; Mueller, L. J. Magn. Reson. B 1993, 101, 201-205. (b) Muhandiram, D. R.; Kay, L. E. J. Magn. Reson. B 1994, 103, 203-216. (16) (a) Palmer, A. G., III; Cavanagh, J.; Wright, P. E.; Rance, M. J. Magn. Reson. 1991, 93, 151-170. (b) Kay, L. E.; Keifer, P.; Saarinen, T. J. Am. Chem. Soc. 1992, 114, 10663-10665. (c) Schleucher, J.; Schwendinger, M.; Sattler, M.; Schmidt, P.; Schedletzky, O.; Glaser, S. J.; Sorensen, O. W.; Griesinger, C. J. Biomol. NMR 1994, 4, 301-306. (17) (a) Zhu, G.; Kong, X. M.; Sze, K. H. J. Biomol. NMR 1999, 13, 77-81. (b) Czisch, M.; Boelens, R. J. Magn. Reson. 1998, 34, 158-160. (c) Pervushin, K.; Wider, G.; Wuethrich, K. J. Biomol. NMR 1998, 12, 345-348. (d) Meissner, A.; Schulte-Herbrueggen, T.; Briand, J.; Sorensen, O. W. Mol. Phys. 1998, 96, 1137-1142. (e) Weigelt, J. J. Am. Chem. Soc. 1998, 120, 10778-10779. (f) Rance, M.; Loria, J. P.; Palmer, A. G., III J. Magn. Reson. 1999, 136, 91-101. (18) Goddard, T. D.; Kneller, D. G. SPARKY 3, version 3.110; University of California, San Francisco: San Francisco, 2004.

Artificial Protein Regulates Crystal Nucleation

Figure 2. Far UV CD spectra of the p288 protein. (A) Concentrationdependent experiments at 100 µM NaH2PO4, pH 2.0; 50 mM Trisacetate/100 mM NaCl nucleation buffer, pH 4.0; and 100 µM TrisHCl buffer, pH 7.5. (B) At 100 µM NaH2PO4 pH 2.0 with varying volume percentages of TFE. Protein concentrations are noted on the spectra. In all CD spectra, mean residue ellipticity [θM] is expressed in deg cm2 dmol-1. (1-D 13C-15N CP/MAS) were collected at 750 MHz using a widebore triple resonance HCN MAS probe. Temperature was controlled at 0 ( 0.5 °C, and the rotor spin rate was regulated at 10 kHz and controlled by the Bruker MAS controller to fluctuate less than (1 Hz. 2-D 13C-13C and 15N-13C correlation experiments were obtained using rotary-assisted proton spin diffusion (RADMIX)19b,c or double cross-polarization (DCP)19a experiments. Acquisition and processing parameters are given in the figure legends. Adamantane (relative to TMS) and solid 15NH4Cl were used as an external standard for 13C and 15N chemical shifts, respectively, with the center of the 15N frequency axis set to 116.7 ppm.

Results Probing the Structure and Stability of p288 under Aqueous Conditions. p288 crystallization assays are typically conducted under acidic aqueous conditions (i.e., pH 4.0). To determine the secondary structure of p288 under these conditions, we performed CD experiments with p288 in 50 mM Tris-acetate/100 mM NaCl nucleation buffer, pH 4.0. As shown in Figure 2A, p288 exhibits a negative π-π* (-) transition band centered at 198-200 nm at pH 4.0 over a protein concentration range of 4-20 µM. This ellipticity band is consistent with the presence of turn, extended, loop, polyproline type II, or other labile structures that exist in equilibria with random coil conformations.12 At p288 concentrations similar to those utilized in crystallization assays (i.e., 50 µM),6 we noted a slight red shift of the π-π* transition band (19) (a) Baldus, M.; Petkova, A. T.; Herzfeld, J.; Griffin, R. G. Mol. Phys. 1998, 95, 1197-1200. (b) Bennet, O. K.; Griffin, R. G.; Vega, J. J. Chem. Phys. 1992, 98, 8624-8628. (c) Takegoshi, K.; Nakamura, S.; Terao, T. Chem. Phys. Lett. 2001, 344, 631-663.

Langmuir, Vol. 23, No. 7, 2007 3859

to 202 nm, which indicates that a concentration-dependent shift is occurring toward nonrandom coil structures, possibly containing β turn conformations.12 We note that the pKa for Asp carboxylate groups in proteins typically ranges between 2.0 and 5.0 pH units.20 Thus, we also performed parallel CD experiments with p288 at pH 2.0 (Figure 2A), where the p288 solubility is ideal for NMR experiments and where a greater percentage of p288 Aspcarboxylate groups would be expected to exist in the protonated state as compared to pH 4.0.20 Here, we find that p288 at pH 2.0 possesses conformational traits that are nearly identical to those observed at pH 4.0. From these results, we conclude that the conformation of p288 at low pH is structurally labile and poorly organized, and this suggests that p288 may adopt an unfolded labile structure under reported assay conditions.6 NMR experiments conducted with p288 also confirm the presence of unstructured conformations. 2-D HSQC experiments at pH 2.0 (Figure 3) indicate that this 116 AA protein exhibits poor spectral dispersion and significant 13C-15N-1H chemical shift overlap. This severe overlap problem precludes obtaining complete amino acid spin assignments as well as sequential assignments. Thus, our use of NMR spectroscopy in this paper is restricted to qualitative interpretations. The 3-D 15N-1H HSQC NOESY experiments reveal the absence of intraresidue and interresidue NOEs between backbone NH, CHR, and CHβ hydrogen atoms (Figure 4). Combined with the HSQC observations, these NOESY features are atypical of globular, folded proteins but are typical of unfolded, labile polypeptides that experience conformational exchange on the NMR time scale that leads to relaxational broadening effects and loss of NMR resonances.22 Thus, p288 appears to possess a labile, unfolded structure under aqueous conditions at low pH, and it is evident that no particular secondary structure preferences exist within this protein under these conditions. Probing the Effect of Neutral pH, Temperature, and Ionic Strength on p288 Structure. Having established the baseline conformational state of p288 at low pH, we now turn to perturbational studies aimed at determining the molecular basis for p288 supramolecular assembly. We examined three perturbations: pH, temperature, and salt concentration. As shown in Figure 2A, we performed CD experiments at a higher pH (7.5) where we would expect all p288 Asp-carboxylate groups to exist in a deprotonated state and a certain percentage of His imidazole groups to exist in a neutrally charged state (pKa ) 6.0-7.0).20 At pH 7.5, we observe two (-) ellipticity bands for p288 at pH 7.5 over 2.5 and 5.0 µM polypeptide concentrations: one centered near 192-195 nm (π-π* transition) and a second centered near 212 nm (n-π* transition). Because of aggregation problems, we were unable to study p288 at higher protein concentrations at pH 7.5. The π-π* transition band is consistent with an unfolded, random coil-like conformation;12 however, the broad n-π* transition band is consistent with a structured state that includes a β strand and an R helix.21 From these results, we conclude that the structure of p288 is sensitive to deprotonation and can transform to a more ordered conformation under these conditions. Next, we pursued temperature variation experiments as a means (20) (a) Brown, L. S.; Needleman, R.; Lanyi, J. K. Biochemistry 1999, 38, 6855-6861. (b) Kuhlman, B.; Luisi, D. L.; Young, P.; Raleigh, D. P. Biochemistry 1999, 38, 4896-4903. (c) Huyghues-Despointes, B. M. P.; Baldwin, R. L. Biochemistry 1997, 36, 1965-1970. (21) (a) Lockwood, N. A.; van Tankeren, R.; Mayo, K. H. Biomacromolecules 2002, 3, 1225-1232. (b) Yang, W. Y.; Larios, E.; Gruebele, M. J. Am. Chem. Soc. 2004, 125, 16220-16227. (22) (a) Fiebig, K. M.; Schwalbe, H.; Buck, M.; Smith, L. J.; Dobson, C. M. J. Phys. Chem. 1996, 100, 2661-2666. (b) Logan, T. M.; Theriault, Y.; Fesik, S. W. J. Mol. Biol. 1994, 236, 637-648. (c) Smith, L. J.; Bolin, K. A.; Schwalbe, H.; MacArthur, M. W.; Thornton, J. M.; Dobson, C. M. J. Mol. Biol. 1996, 255, 494-506.

3860 Langmuir, Vol. 23, No. 7, 2007

Kulp et al.

Figure 4. 15N-1H-1H HSQC NOESY overlay spectra, 100 µM NaH2PO4 pH 2.0 aqueous (black) vs 100 µM NaH2PO4 pH 2.0 with 90% v/v TFE (gray). Eight scans collected with 2048 increments in ω3 (1H), 64 increments in ω2 (15N), and 140 increments in ω1 (1H) with a recycle delay of 1 s. Spectral windows were 11 ppm in ω3 and ω1 and 30 ppm in ω2 with centered frequencies of 4.7 and 118 ppm, respectively. For both TFE and aqueous p288 samples, hard pulses were 9.8 µs at -6 dB for 1H and 33.9 µs at -5 dB for 15N. During processing, a sine squared apodization function was used. Size of spectrum was 256 × 256 points, with zero filling in the ω2 and ω1 dimensions. Figure 3. p288 2-D HSQC overlay spectra. (A) 13C-1H HSQC overlay, 100 µM NaH2PO4 pH 2.0 (black) vs 100 µM NaH2PO4 pH 2.0 with 90% v/v TFE (gray). (B) 15N-1H HSQC overlay, 100 µM NaH2PO4 pH 2.0 (black) vs 100 µM NaH2PO4 pH 2.0 with 90% v/v TFE (gray). For 13C-1H HSQC 2-D experiments, the following acquisition and processing parameters were employed: 32 scans per increment (256 total increments) with 2048 increments in ω2 (1H) and 256 increments in ω1 (13C), recovery delay ) 1 s, spectral window ) 11 ppm in ω2 and 75 ppm in ω1 with carrier frequencies centered at 4.7 and 37 ppm, respectively. Pulse widths: 1H ) 9.8 µs at -6 dB and 13C ) 11.6 µs at -5 dB. Decoupling was performed during acquisition using the GARP decoupling pulse train. Z-axis gradient parameters were all 1 ms duration, sine 100 type, with gradient ratios of 50, 80, and 8.1% for gpnam 1, 2, and 3, respectively. During processing, a sine squared apodization function was used. The size of the spectrum was 1024 × 1024 points, with zero filling in the ω1 dimension. For 15N-1H HSQC 2-D experiments, acquisition and processing parameters were identical except for the following: spectral windows were 11 ppm in ω2 and 30 ppm in ω1 with carrier frequencies centered at 4.7 and 118 ppm, respectively. Pulse widths: 1H ) 9.8 µs at -6 dB and 15N ) 33.9 µs at -5 dB. Z-axis gradient parameters were all 1 ms duration, sine 100 type, with gradient ratios of 80, 20.1, 11, and -5% for gpnam 1, 2, 3, and 4, respectively.

of probing the presence of a polyproline type II structure within p288 (Figure 5). Although our p288 samples lack the typical (+) n-p* 220 nm ellipticity band associated with a polyproline type II (PPII) secondary structure (Figure 2),23,24 we were keenly aware that p288 does contain Pro and Pro-Pro regions and a significant percentage of other PPII forming amino acids (i.e., Ala, Ile; Figure 1).23 To verify or refute the presence of PPII (23) (a) Chen, K.; Liu, Z. G.; Kallenbach, N. R. Proc. Natl. Acad. Sci U.S.A. 2004, 101, 15352-15357. (b) Cubellis, M. V.; Caillez, F.; Blundell, T. L.; Lovell, S. C. Proteins: Struct., Funct., Genet. 2005, 58, 880-892. (c) Zagrovic, B.; Lipfert, J.; Sorin, E. J.; Millett, I. S.; van Gunsteren, W. F.; Doniach, S.; Pande, V. S. Proc. Natl. Acad. Sci U.S.A. 2005, 102, 11698-11703. (24) (a) Gokce, I.; Woody, R. W.; Anderluh, G.; Lakey, J. H. J. Am. Chem. Soc. 2005, 127, 9700-9701. (b) Lam, S. L.; Hsu, V. L. Biopolymers 2002, 69, 270-281. (c) Eker, F., Griebenow, K.; Cao, X., Nafie, L. A.; Schweitzer-Stenner, R. Biochemistry 2004, 43, 613-621. (d) Tamburro, A. M.; Bochicchio, B.; Pepe, A. Biochemistry 2003, 42, 13347-13362.

Figure 5. (A) Far UV CD spectra of 8 µM p288 protein in 50 mM Tris-acetate, pH 4.0, as a function of NaCl concentration. (B) Thermal CD titration plot. Thermal titration was monitored by continuous measurements of 8 µM p288 θM 220 nm ellipticity band from 5 to 80 °C in 100 µM NaH2PO4, pH 2.0. Linear regression analysis was utilized for line fitting.

structure, CD experiments were conducted with p288 at pH 2.0 over 5-80 °C (Figure 5). Here, these experiments revealed that

Artificial Protein Regulates Crystal Nucleation

the θM 220 nm ellipticity does exhibit a weak biphasic transition near 45 °C (Figure 5). Typically, PPII structures exhibit a temperature-dependent behavior24 with a biphasic transition point near 50 °C for L-POLY(PRO).25 Thus, p288 exhibits a biphasic transition at a lower temperature, indicating that if there is any PPII structure within the sequence, it represents a very small proportion of the total secondary structure within p288 and is most likely limited to local regions where Pro and Pro-Pro residues reside (Figure 1). Finally, the creation of increasing ionic strength conditions during evaporative assays periods6 is an interesting stimulus that could trigger conformational transition(s) within p288 either prior to or during supramolecular assembly. To assess this possibility, we performed CD experiments on p288 samples containing 50 µM Tris-acetate buffer, pH 4.0, and varying concentrations of NaCl (50, 100, 250, 500, and 750 mM) (Figure 5), the most common salt utilized in the reported crystallization assay systems.6 We observe that the p288 π-π* transition ellipticity band remains centered at 198-200 nm over the range of 100-750 mM NaCl concentrations, which indicates that the global structure of p288 persists in an unfolded state under these conditions. Thus, it appears that high salt conditions do not significantly perturb the internal structure of p288. Surprisingly, we do note that at NaCl concentrations lower than those utilized in the crystallization assay systems (i.e., 50 mM), the p288 CD spectra exhibits two (-) ellipticity bands (Figure 5), one centered at 195 nm, corresponding to random coil-like conformations, and the other at 205 nm, which is associated with a β turn conformation.12 In conclusion, it would appear that lower ionic strength conditions are capable of triggering a conformational rearrangement in p288, leading to the formation of β turn secondary structures. This finding confirms original sequence prediction studies that indicated the existence of a β turn structure within the p288 sequence.6 Probing the Structure and Stability of p288 with Organic Solvents. In the preceding sections, we have established the labile nature of p288, and we note that unstable polypeptides can stabilize their internal folding or conformation via the use of organic solvents.26,27 These structure-inducing compounds provide useful probes of the tendency of polypeptides to adopt a structure in solution. One such solvent is 2,2,2-trifluoroethanol (TFE), which has been employed to induce and stabilize folded structures (i.e., β hairpins, R helices, molten globule-like folding intermediates) in conformationally labile polypeptides.26,27 In addition, this solvent is also known to disperse protein aggregates.28 As reported elsewhere, the effects of TFE on protein and peptide conformation and aggregate formation are likely to be the result of several mechanisms (e.g., TFE solvent clustering along the polypeptide backbone, perturbation of hydrophobichydrophobic interactions) whose relative importance depends upon the primary sequences and on the structures involved.26 To investigate the potential for p288 to form stable secondary structures and determine the effect of solvent water replacement on the p288 structure, we performed CD experiments in the presence of varying volume percentages of TFE at pH 2.0. This (25) Borza, D. B.; Tatum, F. M.; Morgan, W. T. Biochemistry 1996, 35, 19251934. (b) Wustman, B. A.; Morse, D. E.; Evans, J. S. Langmuir 2002, 18, 99019906. (c) Ma, K.; Kan, L. S.; Wang, S. Biochemistry 2001, 40, 3427-3438. (26) (a) Ragona, L; Catalano, M.; Zetta, L.; Longhi, R.; Fogolari, F.; Molinari, H. Biochemistry 2002, 41, 2786-2796. (b) Buck, M. Q. ReV. Biophys. 1998, 31, 297-355. (c) Luo, P.; Baldwin, R. L. Biochemistry 1997, 36, 8413-8421. (d) Hong, D. P.; Hoshino, M.; Kuboi, R.; Goto, Y. J. Am. Chem. Soc. 1999, 121, 8427-8433. (27) Kulp, J. L., III; Shiba, K.; Evans, J. S. Langmuir 2005, 21, 11907-11914. (28) Wecker, K.; Morellet, N.; Bouaziz, S.; Roques, B. P. Eur. J. Biochem. 2002, 269, 3779-3788.

Langmuir, Vol. 23, No. 7, 2007 3861

allow us to compare our results with parallel NMR experiments performed at this same pH. As shown in Figure 2B, we note that the introduction of TFE has a pronounced effect on p288 conformation. We begin to observe a conformational transition in p288 at 10% v/v TFE as evidenced by a red shift in the π-π* transition band to 202 nm, which is consistent with the presence of a β turn structure.12 At TFE levels of 20% v/v or greater, we observe the appearance of two ellipticity (-) bands, one centered near 205-208 nm (corresponding to the π-π* transition) and the other at 218-222 nm (corresponding to the n-π* transition). These ellipticity bands, which are associated with an R helical structure,26,27 become more intense as the TFE content increases. The induction of folding was also verified by NMR studies conducted on a 90% v/v TFE p288 sample (Figures 3 and 4). Here, with the introduction of TFE, we observe two phenomena in the 2-D 13C-1H and 15N-1H HSQC NMR spectra of p288. First, backbone associated CR, CHR, NR, and NHR resonances experience frequency shifts in the presence of TFE. These backbone resonances are sensitive to secondary structure, and the resulting TFE-induced frequency shifts indicate that conformational transition(s) have occurred within p288 in the presence of TFE, as compared to aqueous conditions. The presence of TFE-induced folding is also evident from the 15N-1H HSQC NOESY spectra, where backbone-to-side chain dRN and dβN NOE connectivities are observed for the p288/TFE sample but not for the p288 aqueous sample (Figure 4). Because of our inability to assign sequential resonances, at this time, we cannot determine if these NOEs are intra- or interresidue in nature, nor can we assign any particular secondary structure from our TFE NMR data. However, we can infer from our CD and NMR data that the p288 protein is conformationally unstable at low pH but can be transformed into a folded structure by a structure stabilizing organic solvent such as TFE, suggesting that solvent water replacement by TFE can impact the structure of this protein. Conformation of p288 in the Bulk Solid State. The p288 protein forms supramolecular assemblies or aggregates within crystallization buffers as evaporation occurs.6 Unfortunately, due to aggregation problems encountered at higher protein and salt concentrations, it is not possible at present to use CD or solutionstate NMR to examine this scenario. Ideally, one would like to probe the conformation of p288 within these assemblies using high speed magic-angle spinning NMR techniques19 to determine if p288 undergoes conformational rearrangement under these conditions. However, a gel-like, aqueous sample such as an aggregate, when subjected to high speed rotation, presents a number of problems for solid-state NMR spectroscopy (i.e., expelling of sample water, sample heterogeneity, loss of sample spinning stability, gel collapse, and so on). Hence, to gain some insight into the structure of p288 at higher protein concentrations under conditions of solvent water loss, we elected to study 13C-15N- uniformly labeled p288 as a lyophilized sample that has been exposed to a minimal volume (5 L) of 50 mM Tris-acetate/100 mM NaCl, pH 4.0 buffer (Figure 6), thereby creating a very concentrated, hydrated sample. As shown in Figure 6, we were successful in applying 2-D RADMIX19b,c 13C-13C and double cross-polarization (DCP)19a 15N-13C solid-state NMR experiments to this sample. The 15N13C DCP correlation spectra reveals a broad frequency distribution of 15NR (89-132 ppm, centered at 112 ppm) and 13CR (41-64 ppm, centered at 47 ppm) backbone resonances of p288, with minor cross-peak contributions arising from side chain C-N spin correlations (i.e., from Arg guanidinium, Gln and Asn amide, His imidazole, and Trp indole; Figure 6B). The broad frequency distribution of 13CR and 13C carbonyls (160-183 ppm, centered

3862 Langmuir, Vol. 23, No. 7, 2007

Kulp et al. Table 1 a 15

NR (ppm)

observed calculated RC-1 calculated RC-2

112.0 120.6 118.8

13

CR (ppm)

13

C carbonyl (ppm)

47.0 57.5 56.6

165.0 176.2 175.9

a RC-1 and RC-2 ) theoretical random coil chemical shift values calculated using eqs 1 and 2, respectively. Adamantane (relative to TMS) and solid 15NH4Cl were used as an external standrad for 13C and 15 N chemical shifts, respectively, with the center of the 15N frequency axis set to 116.7 ppm.

hypothetical bulk average p288 15NR-13CR-13Ccarbonyl random coil chemical shifts

av obsd chemical shift )

∑[(A)(RC)] (1) (N)

av obsd chemical shift )

∑[(RC)] (2) (B)

Figure 6. (A) RADMIX 13C-13C solid-state homonuclear correlation spectra and (B) DCP 15N-13C solid-state heteronuclear correlation spectra for 5 mg of p288 protein sample hydrated in 50 mM Tris-acetate/100 mM NaCl, pH 4.0. In panel A, the 1H amplitude was linearly ramped from 80 to 100% during the first 1.5 ms crosspolarization period, and continuous wave decoupling was applied during the 13C chemical shift evolution period. TPPM decoupling was applied during the rotary-assisted proton spin diffusion mixing (300 ms) and detection periods. Spectral widths were 360 ppm in both dimensions, and the 13C frequency axis was centered at 100 ppm. A total of 64 scans was acquired per experiment, with 2048 total experiments, and a recovery delay of 3 s was employed. Dashed lines denote frequency regions for specific polypeptide 13C chemical groups. For panel B, the 1H amplitude was linearly ramped from 80 to 100% during the first 2.5 ms cross-polarization period, with the center of the ramping corresponding to a transfer ratio of 35 kHz 1H to 25 kHz 15N. Continuous wave heteronuclear 1H-15N decoupling was applied during the 15N chemical shift evolution period, t1. The second cross-polarization mixing time was 2.0 ms with 15N amplitude linearly ramped from 80 to 100%, with the 15N-13C transfer ratio being 25 kHz (15N at center of ramp) to 18 kHz (13C). A decoupling field strength of 70 kHz was applied. The 15N and 13C frequency axes were centered at 118 and 100 ppm, respectively. A total of 256 scans was acquired per experiment, with 1800 total experiments, with a recovery delay of 3 s. Side chain 15N and 13C resonances are denoted in the spectrum.

at 165 ppm) resonances is also confirmed by the RADMIX 13C13C correlation experiment (Figure 6A); here, we also detect 13C -13C , 13C -13C 13C-13C side chain R β β aromatic, and other correlations. Because of residual dipolar broadening effects, chemical shift anisotropy, and amino acid redundancy, individual amino acid spin systems cannot be identified at this time. However, we can use these experiments to extract qualitative information regarding the global conformation of p288 in the bulk state relative to a unfolded, random coil state. To determine if the observed 15NR-13CR chemical shifts for this p288 sample correlate with or deviate from protein database random coil conformations,29 we devised two equations for calculating the

where A is the number of each individual amino acid spin type (i.e., number of Arg, Ala, Ser, Leu, etc.), RC is the published random coil chemical shift value for each spin type,29 B is the the total number of unique amino acid spin types, and N is the total number of amino acids in p288. Eq 2 is considered to be less sequence-specific as compared to eq 1, but we include it here for comparison. As shown in Table 1, the average backbone chemical shifts obtained for p288 resonate upfield from the theoretical random coil 13C-15N backbone chemical shift value calculated via eqs 1 or 2. Upfield shifts are indicative of structured domains,29 indicating that the global conformation of bulk-state p288 is not entirely disordered but does possess some degree of nonrandom coil secondary structure, albeit unknown at this time. These findings suggest that p288, under conditions of high protein concentration and solvent water depletion, can exist in a structured state.

Discussion Our present study indicates that the artificial p288 protein possesses the structural traits of an unfolded, conformationally labile polypeptide under conditions that exist at the start of the crystallization assay (Figures 2-5). Interestingly, these molecular features are also shared by a number of mineral associated polypeptides and proteins,12,30 and such features may convey an inherent internal instability within these proteins that acts as a driving force for stabilization via intermolecular interactions (e.g., the formation of p288 aggregates that subsequently control nucleation and/or crystal growth). The source of this instability is likely to be found within one or both of the two major types of repetitive sequence motifs (Figure 1, magenta and blue). As hypothesized in earlier studies,6 our current study supports the notion that the two major types of repetitive sequence blocks of p288 (Figure 1) are capable of forming repetitive structural motifs, as evidenced by the formation of a partial βstrand (Figure 2A), (29) (a) Wishart, D. S.; Bigam, C. G.; Holm, A.; Hodges, R. S.; Sykes, B. D. J. Biomol. NMR 1995, 5, 67-81. (b) Wishart, D. S.; Sykes, B. D.; Richards, F. M. J. Mol. Biol. 1991, 222, 311-333. (c) Schwarzinger, S.; Kroon, G. J. A.; Foss, T. R.; Chung, J.; Wright, P. E.; Dyson, H. J. J. Am. Chem. Soc. 2001, 123, 2970-2978. (d) Peti, W.; Smith, L. J.; Redfield, C.; Schwalbe, H. J. Biomol. NMR 2001, 19, 153-165. (e) Wang, Y. J. Biomol. NMR 2004, 30, 233-244. (30) (a) Matsushima, N.; Izumi, Y.; Aoba, T. J. Biochem. 1998, 123, 150156. (b) Renugopalakrishnan, V.; Strawich, E. S.; Horowitz, P. M.; Glimcher, M. J. Biochemistry 1986, 25, 4879-4887. (c) Beniash, E.; Simmer, J. P.; Margolis, H. C. J. Struct. Biol. 2005, 149, 182-190.

Artificial Protein Regulates Crystal Nucleation

predominantly Rhelical structures (Figures 2B, 3, and 4), partial PPII (Figure 5B), or β turns (Figures 2A and 5A) under specific solution scenarios. We believe that these repetitive sequence blocks and their conformation(s) are key to the p288 assembly process. Obviously, additional studies involving mimics of these two repetitive sequence blocks will provide information on their inherent structural features and how these respond to external perturbation. The existence of an unstructured p288 conformation under high salt conditions (Figure 5A) and the presence of a nonrandom coil secondary structure within p288 under low solvent water conditions (Figures 2-4 and 6 and Table 1) suggests that it is not high salt conditions but rather the loss of water solvents, along with increasing protein concentrations, that has a significant impact on the structure of p288 and possibly the self-assembly process as well. As shown in Figure 7, we propose a simple hypothetical model that incorporates these possibilities and explains important conformational events leading up to supramolecular assembly.6 We hypothesize that p288 exists in an unfolded state under the initial low pH and dilute conditions of the crystallization assays (Figure 7A). However, with the reduction in assay volume as a result of evaporation, protein and salt concentrations will increase and pH will decrease, and it is likely that the solvent environment around the p288 molecules will change in response to protein molecules coming closer together. Under these conditions, p288 molecules begin to interact, and these interactions involve reordering of the internal structure of the p288 monomer units in response to water solvent loss and external stabilization brought about by protein-protein interactions (Figure 7B).30 As the supramolecular assembly takes place and further loss of bulk solvent occurs, the p288 molecules undergo additional internal conformational transitions. The establishment of internal molecular ordering within p288 molecules becomes a stabilizing force against disassembly of the complex30 and allows the assembly to expand in three dimensions (Figure 7C). As discussed elsewhere,6 once formed, the p288 supramolecular complex now acts as a diffusion limiting barrier, in that there now exist small passages between and within protein molecules that restrict solute passage or movement. This limiting diffusion barrier, together with the presence of supersaturated conditions, affects the kinetics of nucleation in some way that leads to the formation of dendritic crystals within these assays.6 Obviously, there may be other mechanisms at work during the assembly and nucleation processes that are not accounted for in this model, such as the effects of lower pH and

Langmuir, Vol. 23, No. 7, 2007 3863

Figure 7. Proposed model of p288 assembly within in vitro crystallization assay systems. In panel A, the p288 molecules exist as a dilute species (50 µM) within an aqueous environment (solvent represented by blue spheres) at the start of the assay. The conformational state of these molecules is unfolded (represented in bright orange). Once evaporation occurs (B), the bulk solvent volume decreases, and the p288 molecules are now more concentrated in the media. Here, we propose that through-space associations occur between p288 molecules, leading to some degree of conformational change or organization (represented in dark orange) as well as the initial formation of protein intermolecular complexes. Once a critical p288 concentration level is reached via additional loss of solvent (C), the p288 molecules associate more strongly and form larger complexes. Here, we believe that the individual p288 molecules undergo further conformational rearrangement to a more ordered structure (represented in red), which further stabilizes the supramolecular assembly and allows propagation in three dimensions (note arrows).

higher salt concentrations. Likewise, we do not know what the internal structure of p288 is within these supramolecular assemblies or at what point conformational transition(s) occur(s) during the assay period. Additional experimentation will be required to elucidate this information. Acknowledgment. This study was supported by grants from the Army Research Office (DAAD19-02-1-0067) and the Department of Energy (DE-FG02-03ER46099). This paper represents contribution 32 from the Laboratory for Chemical Physics, New York University. LA062442F