Peptide

Jul 4, 2008 - Charlotte J. C. Edwards-Gayle , Francesca Greco , and Ian W. ... Gheorghe Stan , Frank W. DelRio , Robert I. MacCuspie , and Robert F. C...
0 downloads 0 Views 1MB Size
8210

Langmuir 2008, 24, 8210-8214

Effect of PEG Crystallization on the Self-Assembly of PEG/Peptide Copolymers Containing Amyloid Peptide Fragments Ian W. Hamley* and Marta J. Krysmann Department of Chemistry, UniVersity of Reading, Reading RG6 6AD, United Kingdom ReceiVed February 21, 2008. ReVised Manuscript ReceiVed April 23, 2008 The effect of poly(ethylene glycol) PEG crystallization on β-sheet fibril formation is studied for a series of three peptide/PEG conjugates containing fragments modified from the amyloid β peptide, specifically KLVFF, FFKLVFF, and AAKLVFF. These are conjugated to PEG with Mn ) 3300 g mol-1. It is found, via small-angle X-ray scattering, X-ray diffraction, atomic force microscopy, and polarized optical microscopy, that PEG crystallinity in dried samples can disturb fibrillization, in particular cross-β amyloid structure formation, for the conjugate containing the weak fibrillizer KLVFF, whereas this is retained for the conjugates containing the stronger fibrillizers AAKLVFF and FFKLVFF. For these two samples, the alignment of peptide fibrils also drives the orientation of the attached PEG chains. Our results highlight the importance of the antagonistic effects of PEG crystallization and peptide fibril formation in PEG/peptide conjugates.

1. Introduction PEGylation is a technique used to improve the biocompatibility of proteins and peptides. It leads to reduced biodegradation rates, reduced uptake by antibodies and antigens, and improved solubility.1–3 Poly(ethylene glycol), PEG, provides steric stabilization due to the coil-like barrier provided by the polymer in solution. PEG has excellent properties including low toxicity and low biodegradability. We have recently been investigating the self-assembly of a series of copolymers of PEG with fragments of the amyloid beta (Aβ) peptide.4 The fragment is based on sequence Aβ(16-20), KLVFF, which has been shown to be critical in the formation of β-sheet amyloid fibrils.5 During the course of our research, we have noticed a pronounced effect of PEG crystallization on the morphology of dried films as prepared for spectroscopic and microscopic investigation. The present paper summarizes these findings. We find that PEG crystallization can disrupt the peptide fibrillar secondary structure or not, depending on the fibrillization strength of the peptide. Three copolymers are studied comprising PEG with Mn ) 3300 and KLVFF, FFKLVFF, or AAKLVFF. The inclusion of additional hydrophobic units enhances the tendency for fibrillization, which is rather weak in the KLVFF homopolypeptide. Peptide FFKLVFF is insoluble in water but dissolves in methanol, in which it has a strong fibrillization tendency.6 Peptide AAKLVFF is water soluble and forms welldefined fibrils,7 whereas KLVFF is a weak fibrillizer in aqueous solution.8–10 * To whom correspondence should be addressed. E-mail: I.W.Hamley@ reading.ac.uk. (1) Zalipsky, S. AdV. Drug DeliV. ReV. 1995, 16, 157–182. (2) Veronese, F. M. Biomaterials 2001, 22, 405–417. (3) Harris, J. M.; Chess, R. B. Nat. ReV. Drug DiscoVery 2003, 2, 214–221. (4) Krysmann, M. J.; Hamley, I. W.; Castelletto, V.; Noirez, L. AdV. Mater. 2008, in press. (5) Hamley, I. W. Angew. Chem., Int. Ed. 2007, 46, 8128–8147. (6) Krysmann, M. J.; Castelletto, V.; Hamley, I. W. Soft Matter 2007, 3, 1401– 1406. (7) Castelletto, V.; Hamley, I. W.; Harris, P. J. F. Biochim. Biophys. Acta, submitted for publication. (8) Tjernberg, L. O.; Callaway, D. J. E.; Tjernberg, A.; Hahne, S.; Lillieho¨o¨k, C.; Terenius, L.; Thyberg, J.; Nordstedt, C. J. Biol. Chem. 1999, 274, 12619– 12625. (9) Gordon, D. J.; Tappe, R.; Meredith, S. C. J. Pept. Res. 2002, 60, 37–55. (10) Krysmann, M. J.; Castelletto, V.; Kelarakis, A.; Hamley, I. W.; Hule, R. A.; Pochan, D. J. Biochemistry 2008, 47, 4597–4605.

There have been relatively few studies to date concerning the interplay between PEG crystallization and peptide secondary structure formation. Prior work on the solid state structure of PEG/peptide block copolymers has been reviewed.11 Previous studies on PEG/peptides have been limited mainly to studies on conjugates of PEG with synthetic polypeptides such as PBLG [poly(γ-benzyl-L-glutamate)], in which the secondary structure (usually R-helical) is retained on crystallization.12–18 Concerning PEG/peptide copolymers with nonsynthetic peptide sequences, Ro¨sler et al. investigated the effect of PEG crystallization on the solid state self-assembly of PEG/peptide diblock and PEG/peptide/ PEG triblocks containing peptides with 18 residues from a biomimetic “switch” sequence.19 The β-strand peptide secondary structure was retained even when PEG crystallized. Smeenk and co-workers have investigated fibril formation by a series of PEG/ peptide/PEG block copolymers in which the central peptide domain is based on the [(AG)3EG]10 sequence.20 Peptide β-sheet formation was unaffected by conjugation to PEG. Only for high molar mass PEG (5000 g mol-1) could an effect on fibril formation be observed, in particular a reduction in fibril length. In summary, the prior work on PEG/peptides with biomimetic peptide sequences indicates retention of β-sheet secondary structure when PEG crystallizes, just as the secondary structure is retained when PEG crystallizes in PEG/peptides with synthetic homopolypeptide sequences. The present study is the first study in which the competition between PEG crystallization and peptide fibrillization is investigated for a series of rationally designed short peptides in which (11) Klok, H.-A.; Lecommandoux, S. AdV. Polym. Sci. 2006, 202, 75–111. (12) Cho, C.-S.; Kim, S.-W.; Komoto, T. Makromol. Chem. 1990, 191, 981– 991. (13) Floudas, G.; Papadopoulos, P.; Klok, H.-A.; Vandermeulen, G. W. M.; Rodriguez-Hernandez, J. Macromolecules 2003, 36, 3673–3683. (14) Parras, P.; Castelletto, V.; Hamley, I. W.; Klok, H.-A. Soft Matter 2005, 1, 284–291. (15) Cho, C.-S.; Jo, B.-W.; Kwon, J.-K.; Komoto, T. Macromol. Chem. Phys. 1994, 195, 2195–2206. (16) Cho, C.-S.; Jeong, Y.-I.; Kim, S.-H.; Nah, J. W.; Kubota, M.; Komoto, T. Polymer 2000, 41, 5185–5193. (17) Cho, I.; Kim, J.-B.; Jung, H.-J. Polymer 2003, 44, 5497–5500. (18) Tanaka, S.; Ogura, A.; Kaneko, T.; Murata, Y.; Akashi, M. Macromolecules 2004, 37, 1370–1377. (19) Ro¨sler, A.; Klok, H.-A.; Hamley, I. W.; Castelletto, V.; Mykhaylyk, O. O. Biomacromolecules 2003, 4, 859–863. (20) Smeenk, J. M.; Scho¨n, P.; Otten, M. B. J.; Speller, S.; Stunnenberg, H. G.; van Hest, J. C. M. Macromolecules 2006, 39, 2989–2997.

10.1021/la8005426 CCC: $40.75  2008 American Chemical Society Published on Web 07/04/2008

Effect of PEG Crystallization on PEG/Peptides

Figure 1. HPLC data for the three PEG/peptide copolymers studied.

the fibrillization tendency is controlled by addition of hydrophobic residues and increases in the order KLVFF < AAKLVFF < FFKLVFF. The PEG/peptide conjugates are model materials with low polydispersity that provide well-ordered materials on multiple length scales, as confirmed by fiber X-ray diffraction and small-angle X-ray scattering.

2. Experimental Section Materials. The PEGylated conjugates were synthesized by solid phase peptide synthesis using standard FastMoc (Fmoc-9-fluorenylmethyloxycarbonyl protecting group/HBTU activation) chemistry. The peptides were synthesized on a 0.25 mmol scale using a fully automated peptide synthesizer (433A, Applied Biosystems, U.K.) which allowed for direct conductivity monitoring of Fmoc deprotection. PEGylated TentaGel PAP resin was used, with 0.21 mmol/g substitution of PEG3000 to provide C-terminal PEGylation upon cleavage from the resin. The peptide was assembled from the C-terminus toward the N-terminus and was attached to the solid support at the C-terminal by the R-carbonyl group of the amino acid. Peptide attached to resin was obtained from the synthesizer. In the cleavage step, a mixture of 94% trifluoroacetic acid (TFA), 1% trimethylbromosilane, and 5% thioanisole was used. The samples were occasionally shaken at room temperature for approximately 4 h, and the insoluble resin was washed with TFA. The peptide solution was dropped into ice cold diethyl ether to effect precipitation. The solid peptide was washed with diethyl ether and dried. During the cleavage, the side chain protecting group (Boc) were removed by TFA. The crude peptides were purified by reverse phase high performance liquid chromatography (RP-HPLC; Perkin-Elmer 200) using a C18 preparative column (Macherey Nagel) for 30 min with a flow rate of 4 mL/min. A mobile phase of water with 0.1% TFA and acetonitrile with 0.1% TFA was used. Sample elution was monitored using a UV/vis detector operating at 220 nm. Molecular weights were determined by electrospray ionization mass spectrometry (ES-MS). The structures were confirmed by NMR and Fourier transform infrared (FT IR) spectroscopy. PEG/peptide purity was confirmed by HPLC; a single peak was observed and indicated copolymer purity of >90% (Figure 1). MS-ES and MALDI-TOF indicated that for PEG m/z ) 3300 g mol-1 and m/z for the peptide fragments was determined to be 652.8 (KLVFF, expected 653.4), 795 (AAKLVFF, expected 796.2), and 947.97 (FFKLVFF, expected 947.2). Fitting matrix-assisted laser desorption ionization time-offlight (MALDI-TOF) spectra with shape envelopes based on the Schulz-Zimm distribution suggests Mw/Mn < 1.03 for these PEG/ peptide conjugates. Fourier Transform Infrared spectroscopy. IR spectra including amide bands were recorded on a FT IR spectrometer equipped with

Langmuir, Vol. 24, No. 15, 2008 8211 a DTGS detector. A solid film of dry PEGylated peptide was deposited on a CaF2 plate by drying a 5 wt % aqueous solution. Spectra were scanned 64 times over the range of 4000-400 cm-1. Spectral Manager for Windows was used for data acquisition and handling. X-Ray Diffraction (XRD). X-ray diffraction was performed on stalks prepared by drying filaments of the copolymers. An aqueous solution (10 wt %) of peptide copolymer was suspended between the ends of wax-coated capillaries and dried.21 The stalks were mounted (vertically) onto the goniometer of an Oxford Instruments Gemini X-ray diffractometer. Small-Angle X-ray Scattering (SAXS). SAXS experiments were performed on station 2.1 at the Synchrotron Radiation Source, Daresbury Laboratory, U.K. Stalks prepared for XRD were mounted directly using kapton tape in the X-ray beam. A two-dimensional RAPID area detector was used to acquire SAXS patterns. Wideangle X-ray scattering (WAXS) data were obtained using a curved one-dimensional multiwire gas detector. Since no orientation was observed in the SAXS patterns, data were reduced to one-dimensional form using software BSL, with appropriate background subtraction. The wavenumber q ) 4π sin θ/λ (scattering angle 2θ, wavelength λ ) 1.5 Å) scale was calibrated using wet collagen (rat tail tendon). Polarized Optical Microscopy (POM). Samples (5 wt % in water) were placed onto glass microscope slides and dried. Images between crossed polarized were obtained with an Olympus CX-41 microscope. Atomic Force Microscopy (AFM). Samples for AFM were prepared by absorbing passively aliquots of 25 µL of PEG/peptide suspensions in methanol onto a freshly cleaved mica sheet. The samples were then left to dry at room temperature for a few minutes. AFM images were obtained with an NTEGRA instrument (NTMDT, Moscow, Russia). The samples were scanned in intermittent contact mode by using a gold-coated silicon cantilever (NT-MDT, Moscow, Russia) with a nominal spring constant of 5.5 N/m and a nominal resonance frequency of 150 Hz. All the AFM experiments were performed in air at room temperature. The raw AFM images were deconvoluted by using the NT-MDT deconvolution software to take into account the shape of the conical tip and then processed by using the NT-MDT image processing software.

3. Results We consider first the macroscopic superstructure as revealed by polarized optical microscopy (POM). Representative images are shown in Figure 2. The image for KLVFF-PEG in Figure 2a shows a characteristic spherulite structure from PEG crystallization. In contrast, the images in Figure 2b and c for AAKLVFFPEG and FFKLVFF-PEG, respectively, do not exhibit a spherulitic structure. There is birefringence from an ill-defined liquid crystal texture. This already points to an interplay between fibrillization of the peptide segment and crystallization of PEG, which was investigated further using electron microscopy and X-ray scattering. Atomic force microscopy also reveals the presence of a spherulitic structure resulting from PEG crystallization for KLVFF-PEG (Figure 3). A spherulitic structure was not seen for AAKLVFF-PEG or FFKLVFF-PEG, both of which exhibited fibrillar structures. The small-angle X-ray scattering data (Figure 4) contains a pronounced Bragg peak for KLVFF-PEG with a peak position q* ) (0.37 ( 0.02) nm-1, corresponding to d ) (17 ( 0.5) nm. In contrast, there is no Bragg peak for either AAKLVFF-PEG or FFKLVFF-PEG. This is consistent with the absence of spherulites in the POM images. Differential scanning calorimetry (DSC) data (not shown) confirm the melting of PEG for all samples at around 55 °C, confirming that it is crystalline for all samples as studied at room temperature (also confirmed by X-ray diffraction, vide infra). The d spacing observed provides evidence for chain folding of PEG. The extended chain length (21) Makin, O. S.; Serpell, L. C. Fibre Diffr. ReV. 2004, 12, 29–35.

8212 Langmuir, Vol. 24, No. 15, 2008

Hamley and Krysmann

Figure 4. SAXS data obtained for stalks used for fiber X-ray diffraction (Figures 5 and 6): (+) KLVFF-PEG, (9) AAKLVFF-PEG, and (O) FFKLVFF-PEG. The same results were obtained for bulk powders.

Figure 5. FT IR data for dried films. Figure 2. POM images from dried films of (a) KLVFF-PEG, (b) AAKLVFF-PEG, and (c) FFKLVFF-PEG.

Figure 3. AFM height image for a KLVFF-PEG film, showing a spherulitic structure.

of PEG3000 is 19.8 nm, and the length of KLVFF is 5 × 0.35 nm ) 1.8 nm, assuming a usual β-strand residue repeat distance, giving a total length of 21.6 nm. Since this is slightly larger than

the observed d spacing, there may be interdigitation of polymer chains or partial chain folding of the PEG. It appears however to be nonintegral chain folding. Figure 5 presents FT IR data in the amide I region. The peaks at 1685 and 1620 cm-1 are associated with a β-sheet structure.22,23 The peak at 1685 cm-1 is often associated with an antiparallel β-sheet structure. It could also be due to turns or bends,24 although this seems unlikely in such a short peptide. The lower frequency peak is red-shifted for FFKLVFF-PEG, which may reflect enhanced hydrogen bonding24 in this sample compared to the others. It may be noted that sample KLVFF-PEG still retains signals indicating a β-sheet structure despite the effect of PEG crystallization noted above. This suggests that the β-sheet structure may be present locally (as probed by FT IR); however, longer range fibrils are disrupted by PEG crystallization as confirmed by X-ray diffraction, AFM, and POM. The main focus of the FT IR experiments was on the amide I band, which is sensitive to secondary structure. However, considering as a representative example sample AAKLVFF-PEG, additional peaks were located in the amide A band at 3282 cm-1 and amide B band at 3088 cm-1, both resulting from N-H stretching vibrations.23,24 A broad absorption at 2880 cm-1 was due to the CH2 asymmetric (22) Haris, P.; Chapman, D. Biopolymers 1995, 37, 251–263. (23) Stuart, B. Biological Applications of Infrared Spectroscopy; Wiley: Chichester, 1997. (24) Barth, A.; Zscherp, C. Q. ReV. Biophys. 2002, 35, 369–430.

Effect of PEG Crystallization on PEG/Peptides

Langmuir, Vol. 24, No. 15, 2008 8213

Figure 6. X-ray fiber diffraction images of (a) KLVFF-PEG, (b) AAKLVFF-PEG, and (c) FFKLVFF-PEG. Table 1. Observed and Calculated (Most Intense) Reflections for PEG XRDa hkl

|F(hkl)| (calc)

020 021 110 1j13 101 111 120 004 2j02 1j24 112 2j04 032 1j32 2j12

26.47 24.14 31.39 29.16 72.63 57.56 197.26 54.18 87.18 61.30 145.82 58.52 194.04 171.97 148.13

2θ (λ ) 1.54 Å, obs) 13.42 14.55 15.0 18.45 19.12 22.0 (broad) 22.0 (broad) 22.0 (broad) 23.32 (broad) 23.32 (broad) 23.32 (broad) 23.32 (broad) 23.32 (broad)

2θ (λ ) 1.54 Å, calc) 13.57 14.67 15.10 15.70 17.34 18.63 19.17 22.37 22.52 22.22 23.03 23.06 23.30 23.47 23.54

a

Experimental taken from data for AAKLVFF-PEG. Calculation using PEG unit cell data from ref 27.

stretching bands and the CH3 symmetric stretching bands. Peaks at 1551 and 1457 cm-1 were assigned to the amide II band, associated with N-H bending coupled with C-N stretching.25 The CN and CO stretching bands correspond to broad peaks at 1207 cm-1. Peaks in the region 699-839 cm-1 are due to deformation vibrations of aromatic side groups. Peaks at 1299, 1250, 950, and 845 cm-1 are associated with crystalline PEG.26 Figure 6 contains X-ray diffraction data obtained for dried stalks (prepared by drying a thread of concentrated solution between the ends of two waxed capillaries). For KLVFF-PEG, there is no indication of the classical cross-β structure observed for amyloid. The equatorial reflection arising from β-sheet spacing (10-12 Å depending on peptide side group packing) is absent. A reflection at 4.7 Å that could arise from the peptide backbone spacing within a β-strand is present, but this coincides with a strong reflection from crystalline PEG. In fact, all of the observed diffraction rings can be assigned to the crystal structure of PEG (vide infra, Table 1). The fiber diffraction patterns for AAKLVFFPEG and FFKLVFF-PEG in Figure 6 show multiple diffraction rings that can be indexed to the usual monoclinic unit cell of crystalline PEG.27 However, in contrast to KLVFF-PEG, there is evidence for a cross-β amyloid structure due to the presence of equatorial reflections. These are quite weak and broad, especially for AAKLVFF-PEG, but are nonetheless present (as shown in the image with enhanced contrast in Figure 7) and indicate stacking of β-sheets along the fibril axis. The corre(25) Hiramatsu, H.; Kitagawa, T. Biochim. Biophys. Acta 2005, 1753, 100– 107. (26) Waku, T.; Matsusaki, M.; Kaneko, T.; Akashi, M. Macromolecules 2007, 40, 6385–6392. (27) Takahashi, Y.; Tadokoro, H. Macromolecules 1973, 6, 672–675.

Figure 7. Overexposed image from Figure 6b to highlight equatorial reflections from the cross-β structure.

sponding peak positions are d ) 9.8 Å for AAKLVFF-PEG and a doublet d ) 17.6 Å, d ) 11.3 Å for FFKLVFF-PEG (the latter may indicate an orthorhombic unit cell28,29 for the well-developed β-sheet structure of FFKLVFF-PEG). The expected 4.7 Å reflection along the meridian again may be superposed on a strong reflection from PEG. The diffraction with this d spacing is oriented along the meridian as expected, but this may also be due to orientation of the PEG crystalline stems as discussed below. The diffraction rings can be indexed to the PEG crystal unit cell, as exemplified by data for AAKLVFF-PEG in Table 1. The SAXS data in Figure 4 were obtained with the same fibrils used for fiber X-ray diffraction confirming the association between the crystallization of PEG and the disruption of the cross-β structure for KLVFF-PEG. SAXS data were also obtained for bulk powder samples, with identical results. The X-ray diffraction patterns for AAKLVFF-PEG and FFKLVFF-PEG show orientation of the strong diffraction peaks. The peak at 2θ (λ ) 1.54 Å) ) 22° shows two-fold anisotropy with an enhancement of intensity on the meridian. The peak at 2θ (λ ) 1.54 Å) ) 22° exhibits enhancement of intensity at (27° with respect to the meridian (Figure 8). Similar orientation has been reported for crystallized PEG-containing block copolymers in the case of a polycrystal in which the PEG crystal stems (along c axis of unit cell) are oriented but there is cylindrical (fiber) symmetry around the c axis.30 This suggests that the orientation of the fibers within the dried stalks of our PEG/ peptide copolymers has also resulted in alignment of the PEG crystal stems along the fibril axis. (28) Makin, O. S.; Serpell, L. C. J. Mol. Biol. 2004, 335, 1279–1288. (29) Makin, O. S.; Atkins, E.; Sikorski, P.; Johansson, J.; Serpell, L. C. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 315–320. (30) Zhu, L.; Cheng, S. Z. D.; Calhoun, B. H.; Ge, Q.; Quirk, R. P.; Thomas, E. L.; Hsiao, B. S.; Yeh, F.; Lotz, B. J. Am. Chem. Soc. 2000, 122, 5957–5967.

8214 Langmuir, Vol. 24, No. 15, 2008

Figure 8. Azimuthal (φ defined in Figure 6) intensity variation in the second ring of the XRD pattern for AAKLVFF-PEG.

4. Discussion and Summary In summary, crystallization of PEG disrupts long-range β-sheet formation for sample KLVFF-PEG but not for samples AAKLVFF-PEG and FFKLVFF-PEG. Locally, the β-sheet structure appears to be retained for KLVFF-PEG as indicated by FT IR, but the cross-β structure is absent from the X-ray fiber diffraction data, suggesting that β-sheet fibrils are disrupted by PEG crystallization. We believe that the results are related to the fibrillization capacity of the peptides. KLVFF is a weak fibrillizer (there are conflicting reports in the literature about whether it fibrillizes,9 although we have recently presented compelling evidence that it does form fibrils).10 The presence of additional hydrophobic residues in AAKLVFF and FFKLVFF appears to enhance fibrillization capacity. AAKLVFF fibrillizes in water and methanol,31 and FFKLVFF is insoluble in water but forms (31) Krysmann, M. J.; Castelletto, V.; McKendrick, J. M. E.; Hamley, I. W.; Stain, C.; Harris, P. J. F. Langmuir, in press.

Hamley and Krysmann

fibrils in methanol.6 PEGylation confers water solubility to the PEG/peptide conjugates and both FFKLVFF-PEG4 and AAKLVFF-PEG32 form fibrils in aqueous solution. The additional phenylalanine units in FFKLVFF are predicted to enhance fibril formation. This is confirmed by computer analysis of sequences from protein databases, which reveals that F residues are preferentially associated with β-sheet regions.33,34 Among our PEG/peptides, the fiber X-ray diffraction data for the FFKLVFFPEG sample indeed exhibits the clearest cross-β amyloid structure. Our results point to the competing effects of secondary structure formation and crystallization of PEG in our peptide/polymer conjugates. Strategies based on this approach can be envisaged in which the morphology of crystalline polymers is modified by attachment of short peptides (as for our heptapeptides). Alternatively, peptide secondary structure can be altered by conjugation to a crystallizing polymer. It is worth noting that the pronounced effects of PEG crystallization should not be neglected when interpreting transmission electron microscopy (TEM), XRD, or spectroscopic data from dried films of PEG/peptides. The crystallization of PEG during drying means that aggregate structures in solution will be disrupted, and so these techniques cannot be used to interpret such self-assemblies. Acknowledgment. We are grateful to Adam Squires and Yu Gan (both at the University of Reading) for assistance with preparation of stalks and fiber diffraction experiments and to Dr Elisabetta Canetta (University of Surrey) for assistance with AFM experiments. LA8005426 (32) Hamley, I. W.; et al. In preparation. (33) Hutchinson, E. G.; Sessions, R. B.; Thornton, J. M.; Woolfson, D. N. Protein Sci. 1998, 7, 2287–2300. (34) Pawar, A. P.; DuBay, K. F.; Zurdo, J.; Chiti, F.; Vendruscolo, M.; Dobson, C. M. J. Mol. Biol. 2005, 350, 379–392.