Self-Assembly of Peptide− Porphyrin Complexes Leads to pH

Oct 2, 2009 - Department of Chemistry and Biochemistry, Rowan UniVersity, 201 Mullica Hill Road,. Glassboro, New Jersey 08028. ReceiVed: June 10 ...
0 downloads 0 Views 1MB Size
J. Phys. Chem. B 2009, 113, 14439–14447

14439

Self-Assembly of Peptide-Porphyrin Complexes Leads to pH-dependent Excitonic Coupling Darius Kuciauskas* and Gregory A. Caputo* Department of Chemistry and Biochemistry, Rowan UniVersity, 201 Mullica Hill Road, Glassboro, New Jersey 08028 ReceiVed: June 10, 2009; ReVised Manuscript ReceiVed: August 27, 2009

Using absorbance, fluorescence, resonance light scattering, and circular dichroism spectroscopy, we studied the self-assembly of the anionic meso-tetra(4-sulfonatophenyl)porphine (TPPS42-/4-) and a cationic 22-residue polypeptide. We found that three TPPS42-/4- molecules bind to the peptide, which contains nine lysine residues in the primary sequence. In acidic solutions, when the peptide is in the random-coil conformation, TPPS42bound to the peptide forms excitonically coupled J-aggregates. In pH 7.6 solutions, when the peptide secondary structure is partially R-helical, the porphyrin-to-peptide binding constants are approximately the same as in acidic solutions (∼3 × 106 M-1), but excitonic interactions between the porphyrins are insignificant. The binding of TPPS42-/4- to lysine-containing peptides is cooperative and can be described by the Hill model. Our results show that porphyrin binding can be used to change the secondary structure of peptide-based biomaterials. In addition, binding to peptides could be used to optimize porphyrin intermolecular electronic interactions (exciton coupling), which is relevant for the design of light-harvesting antennas for artificial photosynthesis. Introduction Naturally occurring photosynthetic antennas are complexes of peptides, chlorophylls, and other light-absorbing molecules. Photosynthetic pigments are rarely covalently linked to peptides or proteins but rather are positioned in well-defined structures by a combination of electrostatic interactions, hydrogen bonding, and metal-ion ligation.1 It is challenging to mimic lightharvesting antennas using self-assembly chemistry because, in addition to structural considerations, one needs to control excitonic interactions between the light-absorbing compounds. Peptides that efficiently bind porphyrins have been designed; however, in such cases, the porphyrin excitonic interactions were not strong, as typically only one porphyrin binding site was available in the peptide2-5 or the distance between porphyrin binding sites was larger than required for excitonic interactions.6 In detailed studies, it was shown that porphyrins can bind to polypeptide templates7,8 and that, in some cases, these systems show porphyrin excitonic interactions.8-11 However, it is difficult to determine binding stoichiometry because such polypeptide templates are polydisperse. Protein-porphyrin,12-16 porphyrinmicelle,17 and porphyrin-dendrimer18 assemblies have also been investigated. To study multiple porphyrin binding to specific peptide sites, we used the synthetic peptide shown in Scheme 1. This 22residue peptide follows the amphiphilic R-helical design common in antibacterial/antimicrobial peptides. When adopting an R-helical secondary structure, the peptide can be effectively split into two halves: a nonpolar face and a polar, cationic face. Porphyrins, such as meso-tetra(4-sulfonatophenyl)porphine (TPPS42-/4-, Scheme 1), are expected to bind to the polar face of the helix and, if in close contact, could have strong intermolecular (excitonic) interactions. The presence of nine lysine residues suggests that binding of multiple porphyrin molecules to the peptide could be possible. In principle, by controlling the lysine position(s) in the peptide sequence, one * E-mail: [email protected] (D.K.), [email protected] (G.A.C.).

could influence the excitonic interactions between the porphyrins. Peptides such as that shown in Scheme 1 have disordered structures in aqueous solution, but interactions with templates, such as phospholipid bilayers,19 metal ions,20 or porphyrins,2,5 were shown to induce changes in the peptide secondary structure. Therefore, in addition to self-assembly of porphyrin aggregates in the presence of polypeptide templates, we are also exploring changes in the peptide secondary structure following TPPS42-/4- binding. In this study, we used optical spectroscopy to investigate the self-assembly of peptide-porphyrin complexes and found that porphyrin excitonic interactions depend on pH and peptide secondary structure. Experimental Section Materials and Methods. All chemicals except as specified below were from Aldrich. meso-Tetra(4-sulfonatophenyl)porphine dihydrochloride was purchased from Frontier Scientific and used as received. Peptides were synthesized using standard 9-fluorenylmethyloxycarbonyl (Fmoc) solid-phase synthesis procedures using Fmoc-amino acids (Bachem) and a rink-amide solid support (EMD Chemicals). Cleavage and side-chain deprotection was performed using a cleavage cocktail (92.5% trifluoroacetic acid/ 2.5% H2O/2.5% triisopropylsilane/2.5% ethane dithiol, v/v/v/ v) with stirring for 2-3 h at room temperature. Cleaved peptides were precipitated in cold diethyl ether, pelleted via centrifugation, dissolved in neat trifluoroacetic acid (TFA), reprecipitated in diethyl ether, and the resulting precipitate was pelleted. The pellet was then placed under a vacuum for 8 h and stored at 4 °C. The peptide pellet was dissolved in 90:10 water/acetonitrile (v/v) containing 0.1% TFA prior to purification. An aliquot of dissolved peptide (50-100 µL) was loaded onto a Zorbax 300SB-C3 column using an Agilent 1100 high-performance liquid chromatography (HPLC) instrument. Peptides were eluted using a linear gradient of water/acetonitrile containing 0.1% TFA, and peaks were identified using matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS). HPLC

10.1021/jp905468y CCC: $40.75  2009 American Chemical Society Published on Web 10/02/2009

14440

J. Phys. Chem. B, Vol. 113, No. 43, 2009

Kuciauskas and Caputo

SCHEME 1: Helical-Wheel Schematic and Sequence of Porphyrin-Binding Peptide and Structure of meso-Tetra(4-sulfonatophenyl)porphine

fractions containing peptide were pooled and lyophilized for storage. Experimental stock solutions of peptides were prepared by dissolving lyophilized peptide in an appropriate buffer, and the concentration was calculated using ε280 ) 5560 M-1 cm-1. The porphyrin concentration (at pH 7) was calculated using ε413 ) 5.33 × 105 M-1 cm-1. HPLC-grade water was from VWR. Tris(hydroxymethyl)aminomethane hydrochloride (Tris · HCl, 10 mM) buffer was used for the studies at pH 7.6, 10 mM citrate buffer was used for studies at pH 3.6. For experiments at pH 1.8, acidity was adjusted using HCl. Peptide-porphyrin complexes were prepared and studied at 25 °C. When stored in the dark at 4 °C, all samples were stable for at least 1 week. Spectroscopy. Absorption spectra were measured with a Perkin-Elmer Lambda 35 spectrometer. Circular dichroism (CD) spectra were measured with a Jasco J-810 spectropolarimeter using 1-cm-path-length quartz cuvettes for porphyrin CD measurements (concentration ≈ 4 µM) and 1-mm-path-length quartz cuvettes for peptide CD measurements (concentration ≈ 40 µM). For peptide CD measurements, molar ellipticity, θ, was calculated using the equation

θ)

θobs 1 10lc N

(1)

where θobs is the CD signal (in mdeg), l is the optical path length, c is the concentration (M), and N is the number of peptide bonds in a polypeptide (N ) 22). A Jobin Yvon Fluoromax-2 spectrometer was used for fluorescence emission and resonance light scattering (RLS) measurements. Fluorescence emission

spectra were not corrected. In RLS experiments, the emission and excitation bandwidth was 1 nm, and the excitation and emission monochromators were synchronously scanned from 350 to 600 nm. Origin 7.0 software was used for data analysis. Spartan and Gaussian 03W computational software programs were used for molecular modeling. Results and Discussion Absorption, RLS, and Fluorescence Emission Spectra Indicate Peptide-Porphyrin Binding and Porphyrin Excitonic Interactions. To study peptide-porphyrin self-assembly, we used optical spectroscopy. The inset in Figure 1A shows porphyrin absorption spectra in aqueous solutions. At pH 7.6, the Soret band of TPPS44- has a maximum at 413 nm. Four weaker Q bands appear in the range 515-633 nm (Table 1). Because of protonation of the two pyrroles at lower pH, the symmetry of the molecule changes from D2h to D4h (Scheme 1), and as a result, the Soret band for porphyrin diacid, TPPS42-, shifts to 434 nm. Notably, the largest spectral change is evident in the Q-band region. TPPS44- (D2h symmetry) has distinct Qx and Qy bands; x and y transition dipole moments are indicated in Scheme 1. For Qx, the 0-0 vibronic transition is at 633 nm, and the 0-1 vibronic transition is at 580 nm. For Qy, the 0-0 vibronic transition is at 552 nm, and the 0-1 vibronic transition is at 515 nm. The diacid TPPS42- has two stronger Q bands (at 645 and 594 nm) because the x and y directions are structurally equivalent in the case of D4h symmetry. Porphyrin absorption spectra in aqueous solutions are attributed to monomers. Fluorescence, RLS, and CD data also show that, under these conditions (porphyrin concentration < 5 µM, pH 1.8-7.6), aggregation of TPPS42- or TPPS44- is not observed.

Figure 1. Peptide-porphyrin (A) absorption and (B) RLS spectra at pH 1.8 (black), pH 3.6 (red), and pH 7.6 (green). Peptide/porphyrin concentrations (µM) were (A) 0.50/3.75 and (B) 0.25/3.75. Insets show absorption and RLS spectra of porphyrin alone at the same pH.

Self-Assembly of Peptide-Porphyrin Complexes

J. Phys. Chem. B, Vol. 113, No. 43, 2009 14441

TABLE 1: Summary of Spectroscopic Data for Peptide-Porphyrin Complexes pH 1.8 2-

TPPS4

pH 3.6 2- a

2-

pep-TPPS4

TPPS4

pH 7.6 2- a

4-

pep-TPPS4

Soret (nm) Q bands (nm) isosbestic points (nm)

434 594, 645

491 706 452, 662

Absorption 434 490 594, 645 707 452, 664

emission (nm) width (cm-1) Stokes shift (nm)

672 920 27

720b 560 14

Fluorescence 672 ∼670, 718b 920 920, 560 27 11

RLS amplitudec

2.0 × 104

3.5 × 106

porphyrin bands (nm) peptide bandsd (nm)

-

416, 432, 488, 496 199

Resonance Light Scattering 5.0 × 104 3.5 × 106 Circular Dichroism 417, 431, 488, 498 198, 229

TPPS4

pep-TPPS44-

a

413 515, 552, 580, 633

413 517, 554, ∼583, ∼635 425

645, 704 640 12

645, 704 640 ∼10

4.0 × 104

3.0 × 105

-

195, 210, 225

a Porphyrin concentration ) 3.75 µM, peptide concentration ) 0.5 µM. b Aggregate fluorescence emission measured with λex ) 490 nm. When excitation at the isosbestic point was used, monomer emission was much stronger than the J-aggregate emission. c Porphyrin concentration ) 3.75 µM, peptide concentration ) 0.20-0.25 µM. d Peptide concentration ) 40-50 µM.

The Frenkel exciton model deriving from four dominant molecular excited states has been used to explain optical spectra of TPPS42- J-aggregates.21 TPPS42- J-aggregates exhibit characteristic absorption bands at 490 and 707 nm.22 Figure 1A shows that, when 0.5 µM peptide is added to a solution containing 3.75 µM porphyrin at pH 1.8 or 3.6, the absorption spectra change in a manner consistent with TPPS42- J-aggregate formation. The narrowing of the absorption bands indicates excitonic coupling; spectral narrowing is more evident for the 490-nm band because of its higher oscillator strength/extinction coefficient. Because TPPS42- molecules do not form J-aggregates at pH 1.8-3.6 in aqueous solutions (inset in Figure 1A), aggregation can be attributed to porphyrin binding to peptides. Electrostatic interactions between negatively charged porphyrin sulfonate groups and cationic amino acid side chains have been shown to be important for TPPS4 binding to proteins and polypeptides.2,7,12,13,23 For the polypeptide studied in this work, only lysine side chains are positively charged, suggesting that the porphyrins bind to lysine residues. The binding stoichiometry analysis described in a subsequent section further supports this conclusion. Excitonic coupling most likely occurs between several porphyrins bound to the same peptide. It is known that different templates, such as peptides,9 micelles,17 and nanoparticles,24 can lead to TPPS42- J-aggregate formation at less acidic pH than is required to form such aggregates in water (pH < 1).22 Selfassembly of TPPS42- J-aggregates is driven by excitonic, electrostatic, and hydrogen-bonding interactions. In porphyrin J-aggregates, molecules are arranged in a “slipped-deck-ofcards” pattern, with the anionic sulfonate groups of one monomer interacting with the positively charged pyrroles (at the porphyrin center) of the second monomer (see Scheme 2).25,26 Based on high-resolution scanning tunneling microscopy (STM) imaging, the distance between the TPPS42- macrocycles in a J-aggregate was estimated to be 0.34 nm.26 Whereas atomic force microscopy (AFM)/STM imaging showed that TPPS42J-aggregates could be very large (nanorods several micrometers long),25-27 a much shorter coherence delocalization length was obtained from optical studies of TPPS42- J-aggregates (10-100 molecules).28 This suggests that the excitonic delocalization length could differ significantly from the aggregate’s physical size, and studies of smaller aggregates with well-defined sizes are relevant for understanding the photophysical properties of such systems.

SCHEME 2: Illustration of Binding between r-Helical Peptide and Three Porphyrins

As shown in Figure 1A, J-aggregates are not formed at pH 7.6 (there are no excitonic 490- and 707-nm absorption bands). Instead, the amplitude of the Soret band decreases, and the Q bands become broader. At this neutral pH, pyrroles of TPPS4 are not protonated (the TPPS44- molecule has a charge of 4-), and it is known that such protonation is necessary for J-aggregate formation.22,25,26 To determine whether TPPS44- bind to peptides at pH 7.6, we used resonance light scattering (RLS).29,30 The RLS intensity is expected to increase if aggregates (larger than monomers) are present in solution. In addition, RLS intensity reflects the excitonic coupling between the monomers in an aggregate.29,30 As shown in Figure 1B, RLS spectra are very different for solutions containing only porphyrin compared to peptideporphyrin solutions. In solutions that contained only porphyrins, RLS spectra are broad regardless of pH. (Strong absorption by the 413-434-nm Soret band leads to somewhat lower RLS intensity in the Soret-band region.) When porphyrins bind to peptides and form J-aggregates at pH 1.8 and 3.6, the RLS intensity increases by a factor of about 100 (Table 1). RLS is strongest in the J-aggregate absorption region (470-520 nm), which is in agreement with strong resonance polarization in this wavelength range. The peptide-porphyrin RLS spectrum at pH 7.6 is different from the RLS spectra for J-aggregates: it is broader and somewhat less intense. However, the RLS spectrum for peptide-porphyrin complexes at pH 7.6 has an amplitude that is ∼10 times larger than that of the corresponding spectrum for porphyrins (same concentration). The observed increase in RLS amplitude suggests aggregation (porphyrin binding to peptides) at pH 1.8, 3.6, and 7.6, which is supported by the fluorescence emission quenching data described below.

14442

J. Phys. Chem. B, Vol. 113, No. 43, 2009

Kuciauskas and Caputo

Figure 2. (A) Porphyrin fluorescence emission spectra measured with 452-nm excitation (black, pH 1.8; red, pH 3.6) and 425-nm excitation (green, pH 7.6). Inset: Porphyrin fluorescence emission quenching when peptide concentration is increased. The TPPS44-/2- concentration was constant at 3.75 µM. Emission amplitudes were determined at 672 nm (pH 1.8 and 3.6) and 645 nm (pH 7.6). (B) Peptide-porphyrin (0.6 µM/3.75 µM) fluorescence emission spectra measured with excitation at 490 nm at pH 1.8 (black) and at pH 3.6 (red). Fits of fluorescence emission bands with one (pH 1.8) and two (pH 3.6) Gaussian functions are also shown. Inset: Tryptophan fluorescence emission spectrum (λex ) 280 nm, peptide/ porphyrin ) 15 µM/5 µM).

In aqueous solutions, TPPS44- monomers exhibit a fluorescence emission quantum yield of 0.08-0.10.12 Such a high emission yield is typical for meso-substituted porphyrins and indicates that phenylsulfonate substituents are not strongly electronically coupled to the porphine macrocycle. Figure 2A shows porphyrin fluorescence emission spectra in acidic and neutral aqueous solutions. Emission is attributed to TPPS44- (pH 7.6, two vibronic emission bands at 645 and 704 nm) and TPPS42- (pH 1.8 and 3.6) monomers. The different vibrational structures of the emission bands at neutral and acidic pH can be attributed to the protonation of the macrocycle, leading to a change of the molecular symmetry. The Stokes shift is smaller for D2h-symmetric porphyrin than for D4h-symmetric porphyrin (12 vs 27 nm), and a similar relationship holds for the widths of the major emission bands (640 vs 920 cm-1, Table 1). As shown below, the wavelengths of 452 and 425 nm correspond to isosbestic points in the peptide-TPPS44-/2absorption spectra; therefore, these wavelengths were used to measure the fluorescence emission spectra of peptide-porphyrin complexes. The peptide-TPPS44-/2- fluorescence emission spectra are very similar to the porphyrin monomer spectra in Figure 2A, but the emission intensity is reduced at both neutral and acidic pH. No fluorescence spectral shifts were observed, which suggests that emission can be attributed to porphyrin monomers. The inset in Figure 2A shows porphyrin emission quenching due to increased concentration of peptides. The decrease in porphyrin fluorescence emission intensity upon binding to peptides could be related to changes in radiative and nonradiative decay constants (in the case of excitonic aggregates) and to intermolecular energy transfer.8,31,32 In acidic solutions, fluorescence emission is quenched ∼75fold when the peptide concentration reaches 1.5 µM (the fluorescence emission quantum yield decreases from 10-1 to 1.3 × 10-3). At pH 7.6, the emission intensity decreases approximately 13-fold when the peptide concentration increases to 1 µM, whereas at higher peptide concentrations, the TPPS44emission becomes stronger. More complex changes in fluorescence emission at pH 7.6 indicate differences in TPPS44- binding to peptides, which is discussed below. However, in all cases, the largest changes in fluorescence emission intensity are observed when more than three porphyrins are present per peptide (three porphyrins per nine lysine residues). The emission intensity is approximately constant (and low) when there are

more than three lysine residues per TPPS42-. This stoichiometric ratio suggests that TPPS42-/4 binds to three lysine residues. When a sufficient number of binding sites (more than three lysine residues per TPPS42-/4-) are available, most porphyrins are bound to peptides, and the fluorescence emission intensity no longer changes upon further addition of peptide. This model also suggests strong TPPS42-/4- binding to peptides, which leads to a low free porphyrin monomer concentration when the peptide/porphyrin ratio is greater than 1:3; similar conclusions are reached from other studies described below. Through molecular modeling, absorption and CD binding analysis, and analytical ultracentrifugation, it was previously shown that three TPPS44- sulfonate groups could bind to three lysine residues in an R-helical peptide.2,5 Our data appear to be in good agreement with this model and suggest the same binding stoichiometry for TPPS42- in acidic solutions. To measure the fluorescence emission of porphyrin Jaggregates, we used an excitation wavelength of 490 nm, as this wavelength corresponds to the strongest J-aggregate absorption band (Figure 1A). As shown in Figure 2B, at pH 1.8, peptide-TPPS42- complexes have a narrow emission band at 720 nm. (The shoulder at ∼670 nm can be attributed to porphyrin monomer emission; Figure 2A.) Using a Gaussian function to fit the data in Figure 2B yields an approximate width of the J-aggregate emission band as 560 cm-1, whereas the width of the monomer emission band under the same conditions is 920 cm-1. At pH 3.6, the peptide-TPPS42- fluorescence emission spectrum has two peaks at 670 and 720 nm. As shown in Figure 2B, the data could be satisfactorily fit with two Gaussian functions having maxima at these wavelengths. The width of the 670-nm band (920 cm-1) corresponds to that of porphyrin monomer, whereas the width of the 720-nm emission band corresponds to that of a porphyrin J-aggregate (560 cm-1). According to the exciton theory, the narrowing of the emission bands could be used to estimate the number of excitonically coupled monomers.33 Because 920 cm-1/560 cm-1 ≈ 31/2, the data suggest that three TPPS42- porphyrins are excitonically coupled. It is most likely that the three excitonically coupled porphyrins are bound to one peptide. This estimate is in excellent agreement with the fluorescence emission quenching data described above. Peptide-bound TPPS42- J-aggregates are likely to be different from TPPS42- J-aggregates formed in strongly acidic solutions in the absence of peptide; however,

Self-Assembly of Peptide-Porphyrin Complexes

J. Phys. Chem. B, Vol. 113, No. 43, 2009 14443

Figure 3. Absorption spectra recorded as the peptide concentration was increased from 0 to 4-15 µM at pH (A) 1.8, (B) 3.6, and (C) 7.6. The porphyrin concentration was 3.75 µM in all samples. In A and B, insets show absorbance changes at 434 nm (black), 490 nm (red), 645 nm (green), and 707 nm (blue). In C, the first inset shows absorption changes at 413 nm (black) and 633 nm (red). The second inset shows absorption changes in the Q-band region.

the exiton delocalization lengths (coherence aggregation numbers) in some studies were found to be similar.22,28 Peptide-Porphyrin Binding Stoichiometry and Binding Constants. Figure 3 shows changes in absorption spectra as the peptide concentration is increased. Increased absorption between 250 and 290 nm is attributed to tryptophan, the amino acid with the largest conjugated system in the peptide, whereas absorption at 300-800 nm is due to porphyrins. At pH 1.8 and 3.6, there are two isosbestic points in the Soret- and Q-band regions at 452 and 662-664 nm (Table 1). This result indicates that the conversion between the two porphyrin forms (free monomer and peptide-bound aggregate) is quantitative. At pH 7.6, there is one isosbestic point at 425 nm, whereas changes in the Q-band region are more complex. The inset in Figure 3C shows that the longest-wavelength Q band is at 633-635 nm when the peptide concentration is less than 0.5 µM (peptide/ porphyrin < 0.13). When the peptide concentration is equal to 0.5 µM (peptide/porphyrin ) 0.13), the Q band has two maxima at 635 and 652 nm. When the peptide concentration is greater than 1 µM (peptide/porphyrin > 0.13), only the 652-nm Q band is present. These data also suggest that, at pH 7.6, the conversion between free TPPS44- and peptide-bound TPPS44- is quantitative. The major difference between data at low pH (1.8 and 3.6) and data at pH 7.6 is that peptide-bound porphyrins are not excitonically coupled and no J-aggregate absorption bands are observed at pH 7.6. However, the quenching of the fluorescence emission, attributed to porphyrin binding to peptides, is similar at acidic and neutral pH (Figure 2A). The insets in Figure 3 indicate absorption changes at the maximum of the monomer Soret band, J-aggregate bands (A

and B), and Q band. In all cases, Soret- and Q-band absorption changes considerably until the peptide concentration reaches 1-1.5 µM. (At some wavelengths, smaller changes are observed when the peptide concentration increases to 4-15 µM.) Absorption changes significantly until there are approximately three lysine residues per TPPS42-/4-, or three porphyrins per peptide. This result is in excellent agreement with the fluorescence emission quenching data (shown in Figure 2A) and supports the hypothesis that three TPPS42-/4-porphyrins bind per peptide. The estimate that three porphyrins bind to one peptide is also in excellent agreement with the fluorescence emission spectral narrowing results at pH 1.8 and 3.6 (Figure 2B). Thus, the data indicate that the stoichiometry of peptide/porphyrin binding is 1:3 at both acidic and neutral pH, but that excitonic coupling between porphyrins depends on pH: it is significant for TPPS42but negligible for TPPS44-. Apparently, binding to peptide is not sufficient to induce excitonic coupling for TPPS44-. To estimate peptide-porphyrin binding constants, we fit the concentration dependence data to the modified Hill function12,17

A ) A0 + ∆A

[peptide]n Kn + [peptide]n

(2)

where A is absorbance, A0 is initial absorbance, ∆A is change in absorbance, n is a constant (Hill exponent), and K is a dissociation constant (K-1 is a binding constant). Absorbance in both the Soret- and Q-band regions was used in the analysis. The results are summarized in Table 2. We found that the

14444

J. Phys. Chem. B, Vol. 113, No. 43, 2009

Kuciauskas and Caputo

TABLE 2: Dissociation Constants and Hill Exponents for TPPS42-/4- Binding to Peptide pH 1.8

pH 3.6

pH 7.6

experiment

K (µM)

n

K (µM)

n

K (µM)

n

absorption, Soret band absorption, Q band fluorescence emission intensity

0.31 ( 0.04 0.31 ( 0.05 0.23 ( 0.04

2.4 ( 0.4 1.7 ( 0.4 1.5 ( 0.4

0.49 ( 0.02 0.41 ( 0.03 0.43 ( 0.02

2.5 ( 0.3 2.4 ( 0.4 2.6 ( 0.3

0.36 ( 0.03 0.37 ( 0.08

2.2 ( 0.4 1.6 ( 0.5

dissociation constants are approximately 0.3 µM, corresponding to binding constants of K-1 ≈ 3 × 106 M-1. Similar binding constants have been reported for TPPS4 binding to the proteins human serum albumin [HSA, K-1 ) (0.2-1) × 106 M-1], β-lactoglobulin [K-1 ) (0.2-1) × 106 M-1], and tubulin heterodimers (K-1 ) 1.1 × 106 M-1). Additionally, the K-1 values obtained for the peptide-porphyrin interactions in our study are on the same order as those found for TPPS4 binding to other lysine-containing polypeptides [K-1 ) (0.48-1) × 106 M-1].2,3,34 When considering the Hill coefficients, our results show higher values (n ) 1.5-2.6, depending on the data set; Table 2) when compared to published values for β-lactoglobulin and HSA (n ) 0.8-1.5).12 The values of the calculated Hill exponents are indicative of cooperativity in the binding event (a Hill coefficient of n > 1 indicates cooperative binding). Welldefined binding sites in the peptide studied here (cationic lysine residues fully exposed to the aqueous environment) might contribute to the more cooperative binding observed in our study. Interestingly, unlike for β-lactoglobulin and HSA, binding appears to be cooperative in both acidic and neutral solutions. Therefore, excitonic coupling between TPPS42- moieties in acidic solutions does not appear to change binding cooperativity significantly, and electrostatic interactions between sulfonate groups and lysine side chains might be most important for binding. Following binding to peptides, TPPS42-monomers are likely to be in close contact, which allows for excitonic coupling. As shown in Figure 3, at higher peptide concentrations, some absorption bands undergo additional shifts. (At pH 1.8 and 3.6, such shifts are in addition to much larger red shifts discussed earlier and attributed to TPPS42- J-aggregate formation.) For example, at pH 1.8 and 3.6, the Soret band blue shifts from 434 to 425 nm when the peptide concentration exceeds 1.5 µM. In addition, at pH 3.6 but not pH 1.8, the J-aggregate Soret band shifts from 491 to 495 nm, and both J-aggregate bands (495 and 707 nm) become significantly broader. Under some conditions, a blue shift of the Soret band to 420 nm indicates the formation of H-aggregates.26 However, TPPS42- J-aggregates also have a shoulder at 423 nm,26 so a blue shift to 425 nm at pH 1.8 could simply be related to the larger concentration of (quantitative conversion to) TPPS42- J-aggregates. Spectral broadening at pH 3.6 is more likely related to aggregation. In general, aggregation of cationic polypeptides is not significant until the concentration exceeds the millimolar range,35 whereas in our studies, the peptide concentration was always less than 50 µM. Therefore, aggregation is likely to be related to porphyrin binding effects. Finally, at pH 7.6 and when the peptide concentration is greater than 1.5 µM, the monomer Soret band blue shifts from 413 to 403 nm, the four Q bands shift to the red by 6-18 nm, and all bands broaden. A similar porphyrinconcentration-dependent blue shift and broadening was reported for TPPS44- monomers bound to R-helical peptide and was interpreted to indicate TPPS44- binding to peptide.2,5 Peptide-porphyrin binding was also analyzed using fluorescence emission quenching data shown in Figure 2A. Data obtained from the fluorescence studies are similar to absorption results. All binding constants and Hill exponents are summarized

in Table 2. The absorption and fluorescence data described above are consistent with 1:3 peptide/porphyrin binding and with the presence of porphyrin excitonic coupling at pH 1.8 and 3.6. It is also possible that several peptide-porphyrin assemblies are aggregated to larger structures; however, we do not see spectroscopic evidence for such association. For example, isosbestic points in absorption spectra indicate only one transition that corresponds to free and bound porphyrins. However, more detailed light scattering studies are necessary to investigate the possibility of larger-scale aggregation effects. If peptideporphyrin complexes were found to aggregate, such aggregation could be controlled by changing peptide sequence and, in particular, by using more polarizable residues on the nonpolar face of the peptide (Scheme 1).36,37 Peptide-Porphyrin Aggregates Have Chiral Structures at Acidic pH. TPPS42- J-aggregates (formed in acidic solutions for pH < 1) show only weak CD signals.22 However, when porphyrins are assembled on some matrixes, including polypeptides, strong porphyrin CD signals can be observed.5,9,12,17 Figure 4A shows that peptide-TPPS42- CD signals are strong when the peptide concentration is only 60-83 nM (peptide/porphyrin ) 0.02:1). Both the J-aggregate band at 490 nm and a monomer Soret band yield CD signals that have similar shapes, although the J-aggregate CD bands are stronger and narrower. Starting from porphyrin CD spectra, calculations based on exciton theory could be used to predict the geometry of porphyrin dimers and larger structures;21,38,39 however, such calculations are complex. Comparable CD spectra at pH 1.8 and 3.6 suggest that porphyrin aggregate structures are similar in both acidic solutions. The orders of positive and negative bands for peptide-induced TPPS42- CD spectra in Figure 4A are similar to those for TPPS42- aggregate CD spectra when bound to poly(D-lysine) templates9 and to HSA proteins in bis(2-ethylhexyl) sodium sulfosuccinate (AOT) reverse micelles,17 but the intensities of the peaks and their wavelengths differ. In particular, the negative and positive Cotton effects for the monomer Soret bands in Figure 4A are approximately equal, which suggests that porphyrins bind to peptides in a single conformation.5,6,38 In contrast, the amplitude of the negative Cotton effect for the J-aggregate band is larger than that for the positive Cotton effect, which is in agreement with a more complex orientation of excitonically coupled transition dipole moments in TPPS42J-aggregates.21,39 Porphyrin CD signals were not detected at pH 7.6. This indicates that there is no chiral porphyrin aggregate structure at neutral pH and suggests that excitonic interactions are necessary to achieve chiral porphyrin aggregate structure in this system. Figure 4B shows CD signal changes as the peptide concentration is increased. Although, overall, the spectra at pH 1.8 and 3.6 are comparable, indicative of similar porphyrin orientations in aggregates in acidic solutions (Figure 4A), the signal amplitudes change somewhat differently. The 480-490-nm signals (in the J-aggregate band region, Figure 4B top) have similar amplitudes when the peptide concentration is 1 µM. However, the pH 1.8 signal is approximately constant at 0.25-5 µM, whereas the signal at pH 3.6 is much stronger at 0.6 µM

Self-Assembly of Peptide-Porphyrin Complexes

J. Phys. Chem. B, Vol. 113, No. 43, 2009 14445

Figure 4. (A) Circular dichroism spectra for peptide-TPPS42- in acidic solutions. Porphyrin concentration ) 3.75 µM; peptide concentration ) 0.083 µM (pH 1.8, black), 0.060 µM (pH 3.6, red), and 0.60 µM (pH 3.6, green). (B) Porphyrin CD signal amplitudes at 488-489 nm (top), 460 nm (middle), and 416-417 nm (bottom) for pH 1.8 (9) and pH 3.6 (b).

Figure 5. UV circular dichroism spectra for peptide-porphyrin complexes at (A) pH 1.8 (peptide/TPPS42- ) 1:0, black; 1:0.4, red; 1:1, green; 1:3, blue), (B) pH 3.6 (peptide/TPPS42- ) 1:0, black; 1:1, red; 1:2, green; 1:3, blue), and (C) pH 7.6 (peptide/TPPS44- ) 1:0, black; 1:0.5, red; 1:1, green; 1:5, blue). The peptide concentration was 40 µM. The inset in B shows pH 3.6 peptide CD spectrum with 50% TFE (v/v).

and decreases at higher peptide concentrations. Similar changes in the pH 3.6 signal amplitude occur in the monomer-band region at 410-420 nm (Figure 4B bottom). CD spectral changes are the most complex in the 440-480-nm region and suggest that the aggregate structure is changing when multiple porphyrins are binding per peptide. Because the Soret bands arising from the TPPS monomer and J-aggregate overlap at 440-480 nm, such overlap could lead to a more complex concentration dependence for the CD signals. In all cases, major changes in spectra (porphyrin structural rearrangements) appear to be complete when the stoichiometry of peptide/porphyrin binding is 1:3, supporting the hypothesis that three porphyrins bind to one peptide. Analysis of Peptide Secondary Structure. Circular dichroism spectra in the UV region (190-260 nm) are commonly used to determine peptide/protein secondary structure. A peptide adopting an R-helical secondary structure will exhibit characteristic negative bands at 222 and 208 nm and a positive band at 193 nm. Alternatively, disordered polypeptides and denatured proteins have very low ellipticity above 210 nm and negative bands near 195 nm. As such, the molar ellipticity at 222 nm is frequently used to determine R-helix content in a peptide. For short peptides, when estimating the number of residues in an R-helical conformation, molar ellipticity needs to be corrected for peptide length effects.40 Figure 5 shows peptide and peptide-porphyrin CD spectra in acidic and neutral pH. In all cases, when porphyrins are not present, peptides adopt a random-coil conformation (spectra have a characteristic negative band at 195-200 nm and a negligible CD signal at >220 nm). When the porphyrin concentration is increased, the negative band at 195 nm becomes

smaller (less negative), and the negative signal at 220-230 nm begins to appear. At pH 1.8 and 3.6, the negative signal at ∼220 nm is weaker than 195-nm band. Trifluoroethanol (TFE) and water mixtures are often used to stabilize R-helices and were employed to determine the maximal helical propensity of the peptide used in this study. Increasing the TFE concentration in the aqueous solution until the peptide CD spectrum no longer changes [typically, up to 50% (v/v) TFE] is commonly assumed to indicate that the peptide has assumed a fully R-helical conformation.40 The inset in Figure 5B shows the peptide pH 3.6 CD spectrum with 50% (v/v) TFE. Although molar ellipticity at 222 nm is not as large as is typically observed for R-helices, this result could be related to the relatively short polypeptide length. Comparison of the peptide CD spectra with 50% (v/v) TFE and peptide-TPPS42- spectra at pH 1.8 and 3.6 indicates that porphyrin binding in acidic solutions leads to very little, if any, helical content in the peptide. At pH 7.6, the shape of the final peptide-porphyrin CD spectrum suggests that the peptide has adopted partial R-helical conformation (negative bands at ∼210 and ∼225 nm, positive band at ∼195 nm). Comparison of the θ222 values for the peptide-porphyrin complex and the peptide in 50% (v/v) TFE indicates that the R-helix content of the peptide is ∼25% (when three porphyrins are bound to nine lysine residues). We conclude that addition of porphyrins to acidic and neutral solutions influences the polypeptide conformation; however, only in neutral solutions does the peptide becomes partially R-helical. For coiled-coil designed peptides, TPPS44- binding leads to a fully R-helical structure.2,5 Such coiled-coil peptides, however, already have a partially R-helical structure without porphyrins.5

14446

J. Phys. Chem. B, Vol. 113, No. 43, 2009

Finally, to learn about peptide properties in solution, we measured tryptophan fluorescence emission spectra. Tryptophan (Trp) fluorescence emission exhibits a dramatic sensitivity to the polarity of the environment in which it is located. Trp exposed to an aqueous solution exhibits an emission maximum at ∼355 nm, whereas for Trp surrounded by less polar media (for example, when the polypeptide is embedded to lipid bilayer), the emission maximum blue shifts to 320-330 nm.41 A representative tryptophan emission spectrum measured with 280-nm excitation is shown in the inset of Figure 2B. The emission λmax value at 355 nm suggests that tryptophan is surrounded by water for all peptide conformations. If the peptide is assumed to be R-helical, this result is expected from the peptide design. As shown in Scheme 1, if the peptide adopts an R-helical secondary structure, all nine lysine residues are on one (hydrophilic) face of the peptide, whereas aromatic and less polar residues, including tryptophan, are on the other more hydrophobic face. For a disordered/random coil conformation, this result is also indicative that tryptophan does not participate in aggregation or other hydrophobic intra- or intermolecular interactions. Summary and Molecular Structure Considerations. We studied the self-assembly of cationic peptides and anionic porphyrins in acidic and neutral solutions. Under all conditions tested, we observed stoichiometric binding of three TPPS42-/4molecules to nine lysine residues, which corresponds to the number of lysines in the peptide sequence. (Based on our spectroscopic data, we cannot exclude peptide-porphyrin aggregation to larger structures; however, excitonic coupling is evident only between three porphyrins that are presumably bound to the same peptide.) Depending on the conditions and experimental method, the binding constants between peptides and porphyrins are in the range (0.23-0. 49 µM)-1. Binding constants on this order of magnitude are typical for porphyrin binding to peptides and proteins. The binding was also shown to be cooperative, exhibiting a Hill exponent ranging from 1.5 to 2.6. In neutral solutions, TPPS44- binding to peptide induces a partial R-helical secondary structure in the polypeptide. In acidic solutions, peptide circular dichroism data also suggest peptide conformational changes. As shown in Scheme 1, in the sequence of the peptide, there are nine lysine residues arranged in three groups: residues (1, 2, 5), (9, 12, 13), and (15, 16, 19). The first and third groups have lysine residues at positions i, i + 1, and i + 4, whereas the second group has lysine residues at positions i, i + 3, and i + 4. In all groups, there are two adjacent lysine residues, and the third lysine is separated by two other amino acids. Molecular modeling was used to show that, in all three groups, the distance between the protonated amino groups (-NH3+) of lysine residues can range from 1 to 1.5 nm. [The flexible -(CH2)4linkage between the peptide backbone and the lysine amino group allows for the variation of this distance.) NMR studies of an identical peptide motif in the magainin 2 peptide also showed conformational flexibility of the lysine residues.42 The distance between porphyrin anionic sulfonate groups (SO3-) is also approximately 1.5 nm, suggesting that each porphyrin could be bound to each group of amino acids. When a 1:3 binding stoichiometry is assumed, three porphyrins should be bound to the three amino lysine clusters described above. It was already shown that, in an R-helical peptide, lysine residues at positions i, i + 4, and i + 8 also bind three TPPS44- sulfonate groups.2,5 Our results suggest that similar binding could be obtained using a shorter peptide sequence with i, i + 1, and i + 4 lysine residues. Our results also suggest that several such lysine groups

Kuciauskas and Caputo binding TPPS44- could be placed relatively close to each other on the polypeptide chain without steric or electrostatic hindrance. Very different spectroscopic properties were observed, and different structural interpretations were proposed for TPPS44-/2aggregation on poly[(DL, D, or L)-lysine] templates.8,9,43,44 For example, a 10-fold excess of lysine residues was required for complete aggregation of TPPS42- at pH 3.0.9 Well-defined rationally designed peptide sequences for porphyrin binding are probably preferable for establishing binding stoichiometry and binding constants. The results indicate that, in acidic solutions, peptide-bound porphyrins participate in excitonic interactions. Excitonic interactions are weak or negligible at neutral pH, when the peptide is partially R-helical. A possible interpretation of this result could be related to the peptide secondary structure. Although the structure illustrated in Scheme 2 is consistent with a TPPS42arrangement in J-aggregates25-27 and with TPPS44- orientation in respect of three lysine residues,2,5 it might be easier to achieve a similar porphyrin slipped-deck-of-cards25-27 arrangement when the peptide backbone is more flexible (disordered). In addition, at pH 7.6, the core of TPPS44- does not have a positive charge; such a charge appears to be necessary for the J-aggregate formation. However, the fluorescence emission is quenched equally strongly at pH 7.6 as in acidic solutions (Figure 2A), which indicates electronic coupling between the porphyrins. Studies of energy-transfer dynamics will be necessary to investigate quenching mechanisms and their dependence on the peptide secondary structure. Spectral changes in porphyrin absorption, RLS, CD, and fluorescence spectra suggest TPPS42- J-aggregate formation at pH 1.8 and 3.6. In acidic solutions, when the peptide backbone does not assume an R-helical conformation (and therefore is likely to be more flexible), a closer contact between peptidebound TPPS42- monomers could be possible. TPPS42- Jaggregate formation in acidic solutions might be related to such conformational flexibility of the peptide backbone. Whereas such aggregation might have a favorable enthalpy term at pH 1.8 and 3.6, entropy apparently favors monomers. When TPPS42monomers are bound to peptides, the entropic contribution is apparently reduced, and TPPS42- J-aggregates could be formed at higher pH. Although TPPS42- J-aggregates observed in our studies consist of a small number of monomers (three monomers are suggested by fluorescence emission narrowing and concentration dependence studies), their optical spectra are similar to the optical spectra of nanorods consisting of hundreds and thousands of TPPS42- molecules.25-27 This research opens several possibilities for further studies. First, by appropriately designing the peptide sequence, one can control the excitonic interactions between the porphyrins, including the number of excitonically coupled monomers. In such a manner, it is possible to investigate the size dependence of various optical properties of porphyrin aggregates and nanostructures. Second, photosynthetic antennas of all green plants and some photosynthetic bacteria consist of peptideporphyrin complexes. Self-assembly of such components offers a promising way to build artificial photosynthetic antennas with controlled excitonic interactions. Finally, cationic peptides with hydrophobic and hydrophilic interfaces often have antimicrobial and/or antibacterial properties.45 The presumed mechanism of cytotoxicity involves peptide binding to the lipid bilayer (cell membrane) and lysis due to peptide insertion. Some porphyrins and porphyrin aggregates are used in photodynamic therapy.

Self-Assembly of Peptide-Porphyrin Complexes Thus, using this approach, one can design novel membrane binding aggregates that could be useful for photodynamic therapy. Acknowledgment. Acknowledgment is made to the Donors of the American Chemical Society Petroleum Research Fund for partial support of this research. G.A.C. thanks William F. DeGrado at the University of Pennsylvania for use of the MALDI mass spectrometer. References and Notes (1) Ke, B. Photosynthesis: Photobiochemistry and Photobiophysics; Kluwer Academic Publishers: Dordrecht, The Netherlands, 2001. (2) Kovaric, B. C.; Kokona, B.; Schwab, A. D.; Twomey, M. A.; de Paula, J. C.; Fairman, R. J. Am. Chem. Soc. 2006, 128, 4166. (3) Takahashi, M.; Ueno, A.; Mihara, H. Chem.sEur. J. 2000, 6, 3196. (4) Rosenblatt, M. M.; Wang, J.; Suslick, K. S. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 13140. (5) Kokona, B.; Kim, A. M.; Roden, R. C.; Daniels, J. P.; PepeMooney, B. J.; Kovaric, B. C.; de Paula, J. C.; Johnson, K. A.; Fairman, R. Biomacromolecules 2009, 10, 1454. (6) McAllister, K. A.; Zou, H.; Cochran, F. V.; Bender, G. M.; Senes, A.; Fry, H. C.; Nanda, V.; Keenan, P. A.; Lear, J. D.; Saven, J. G.; Therien, M. J.; Blasie, J. K.; DeGrado, W. F. J. Am. Chem. Soc. 2008, 130, 11921. (7) Urbanova, M.; Setnicka, V.; Kral, V.; Volka, K. Biopolymers 2001, 60, 307. (8) Purrello, R.; Gurrieri, S.; Lauceri, R. Coord. Chem. ReV. 1999, 190192, 683. (9) Koti, A. S. R.; Periasamy, N. Chem. Mater. 2003, 15, 369. (10) Lauceri, R.; Gurrieri, S.; Bellacchio, E.; Contino, A.; Scolaro, L. M.; Romeo, A.; Toscano, A.; Purrello, R. Supramol. Chem. 2000, 12, 193. (11) Purrello, R.; Scolaro, L. M.; Bellacchio, E.; Gurrieri, S.; Romeo, A. Inorg. Chem. 1998, 37, 3647. (12) Andrade, S. M.; Costa, S. M. B. Biophys. J. 2002, 82, 1607. (13) Kano, K.; Watanabe, K.; Ishida, Y. J. Phys. Chem. B 2008, 112, 14402. (14) Valanciunaite, J.; Bagdonas, S.; Streckyte, G.; Rotomskis, R. Photochem. Photobiol. Sci. 2006, 5, 381. (15) Valanciunaite, J.; Poderys, V.; Bagdonas, S.; Rotomskis, R.; Selskis, A. J. Phys.: Conf. Ser. 2007, 1207. (16) Tian, F.; Johnson, E. M.; Zamarripa, M.; Sansone, S.; Brancaleon, L. Biomacromolecules 2007, 8, 3767. (17) Andrade, S. M.; Costa, S. M. B. Chem.sEur. J. 2006, 12, 1046. (18) Kubat, P.; Lang, K.; Janda, P.; Anzenbacher, P. Langmuir 2005, 21, 9714.

J. Phys. Chem. B, Vol. 113, No. 43, 2009 14447 (19) Ramamoorthy, A.; Thennarasu, S.; Lee, D.-K.; Tan, A.; Maloy, L. Biophys. J. 2006, 91, 206. (20) Signarvic, R. S.; DeGrado, W. F. J. Am. Chem. Soc. 2009, 131, 3377. (21) Vlaming, S. M.; Augulis, R.; Stuart, M. C. A.; Knoester, J.; van Loosdrecht, P. H. M. J. Phys. Chem. B 2009, 113, 2273. (22) Ohno, O.; Kaizu, Y.; Kobayashi, H. J. Chem. Phys. 1993, 99, 4128. (23) Shoichi Ikeda, T. N. G. E. Biopolymers 1991, 31, 1257. (24) Fujii, Y.; Hasegawa, Y.; Yanagida, S.; Wada, Y. Chem. Commun. 2005, 3065. (25) Rotomskis, R.; Augulis, R.; Snitka, V.; Valiokas, R.; Liedberg, B. J. Phys. Chem. B 2004, 108, 2833. (26) Friesen, B. A.; Nishida, K. R. A.; McHale, J. L.; Mazur, U. J. Phys. Chem. C 2009, 113, 1709. (27) Schwab, A. D.; Smith, D. E.; Rich, C. S.; Young, E. R.; Smith, W. F.; dePaula, J. C. J. Phys. Chem. B 2003, 107, 11339. (28) Ogawa, T.; Tokunaga, E.; Kobayashi, T. Chem. Phys. Lett. 2005, 408, 186. (29) Pasternack, R. F.; Collings, P. J. Science 1995, 269, 935. (30) Parkash, J.; Robblee, J. H.; Agnew, J.; Gibbs, E.; Collings, P.; Pasternack, R. F.; de Paula, J. C. Biophys. J. 1998, 74, 2089. (31) Gulbinas, V.; Karpicz, R.; Augulis, R.; Rotomskis, R. Chem. Phys. 2007, 332, 255. (32) Goncalves, P. J.; Aggarwal, L. P. F.; Marquezin, C. A.; Ito, A. S.; De Boni, L.; Neto, N. M. B.; Rodrigues, J. J.; Zilio, S. C.; Borissevitch, I. E. J. Photochem. Photobiol. A: Chem. 2006, 181, 378. (33) Knoester, J. J. Chem. Phys. 1993, 99, 8466. (34) Wang, J.; Rosenblatt, M. M.; Suslick, K. S. J. Am. Chem. Soc. 2007, 129, 14124. (35) Measey, T. J.; Schweitzer-Stenner, R. J. Am. Chem. Soc. 2006, 128, 13324. (36) Bryson, J. W.; Betz, S. F.; Lu, H. S.; Suich, D. J.; Zhou, H. X.; O’Neil, K. T.; DeGrado, W. F. Science 1995, 270, 935. (37) Degrado, W. F. Design of Peptides and Proteins. In AdVances in Protein Chemistry; Academic Press: New York, 1988; Vol. 39, p 51. (38) Pescitelli, G.; Gabriel, S.; Wang, Y.; Fleischhauer, J.; Woody, R. W.; Berova, N. J. Am. Chem. Soc. 2003, 125, 7613. (39) Didraga, C.; Knoester, J. J. Chem. Phys. 2004, 121, 946. (40) Luo, P.; Baldwin, R. L. Biochemistry 1997, 36, 8413. (41) Caputo, G. A.; London, E. Biochemistry 2003, 42, 3265. (42) Gesell, J.; Zasloff, M.; Opella, S. J. J. Biomol. NMR 1997, 9, 127. (43) Lauceri, R.; Campagna, T.; Raudino, A.; Purrello, R. Inorg. Chim. Acta 2001, 317, 282. (44) Purrello, R.; Bellacchio, E.; Gurrieri, S.; Lauceri, R.; Raudino, A.; Scolaro, L. M.; Santoro, A. M. J. Phys. Chem. B 1998, 102, 8852. (45) Giuliani, A.; Pirri, G.; Bozzi, A.; Di Giulio, A.; Aschi, M.; Rinaldi, A. Cell. Mol. Life Sci. (CMLS) 2008, 65, 2450.

JP905468Y