Automated 2D IR spectroscopy using a mid-IR pulse shaper and application of this technology to the human islet amyloid polypeptide

Edited by Robin M. Hochstrasser, University of Pennsylvania, Philadelphia, PA, and approved March 28, 2007
September 4, 2007
104 (36) 14197-14202

Abstract

The capability of 2D IR spectroscopy to elucidate time-evolving structures is enhanced by a programmable mid-IR pulse shaper that greatly improves the ease, speed, and accuracy of data collection. Traditional ways of collecting 2D IR spectra are difficult to implement, cause distorted peak shapes, and result in poor time resolution and/or phase problems. We report on several methods for collecting 2D IR spectra by using a computer-controlled germanium acoustooptic modulator that overcomes the above problems. The accuracy and resolution of each method is evaluated by using model metal carbonyl compounds that have well defined lineshapes. Furthermore, phase cycling can now be employed to largely alleviate background scatter from heterogeneous samples. With these methods in hand, we apply 2D IR spectroscopy to study the structural diversity in amyloid fibers of aggregated human islet amyloid polypeptide (hIAPP), which is involved with type 2 diabetes. The 2D IR spectra reveal that the β-sheet fibers have a large structural distribution, as evidenced by an inhomogeneously broadened β-sheet peak and strong coupling to random coil conformations. Structural diversity is an important characteristic of hIAPP because it may be that partly folded peptides cause the disease. This experiment on hIAPP is an example of how computer generation of 2D IR pulse sequences is a key step toward automating 2D IR spectroscopy, so that new pulse sequences can be implemented quickly and a diverse range of systems can be studied more easily.
Two-dimensional infrared (2D IR) spectroscopy is becoming a very useful tool for probing the fast structural dynamics of chemical and biological systems (13). Analogous in many ways to 2D NMR spectroscopy, 2D IR spectroscopy probes molecular structures by means of vibrational frequencies, couplings, and transition dipole angles (48). Environmental dynamics are probed through lineshape analysis (911). The combination of structural sensitivity and fast time resolution (fs/ps) makes this technique and its variants (3, 1214) especially adept at monitoring the dynamics of evolving structures or the kinetics of chemical reactions (1518). Although 2D IR spectroscopy is a powerful technique, implementing it is technically challenging. In a typical 2D IR spectrometer, each pulse has its own optical path composed of several mirrors to route the beam to the sample and a mechanical delay stage that controls the time delay by changing the physical length of the optical path. All of the pulses have the same frequencies and shapes unless additional optics are added, such as an etalon to narrow a pulse bandwidth (4, 19) or a second mixing crystal to generate pulse sequences with two center frequencies (20, 21). Thus, implementing even the simplest pulse sequence is a tremendous amount of work. This is in sharp contrast to 2D NMR spectroscopy, in which NMR pulse sequences are easily programmed with precisely set frequencies, time delays, phases, and intensities. As a result, it is straightforward in NMR to change the pulse sequences or add additional pulses to create the best spectrum for measuring the desired information. Here, we take a step closer to making 2D IR more like 2D NMR spectroscopy. Using a mid-IR pulse shaper recently developed by our group (22, 23), we can now programmably generate 2D IR pulse sequences with controlled phases, delays, and shapes. Using traditional 2D IR methods as a starting point, we improve on their accuracy, speed, and ease of implementation with several alternative pulse sequences optimized to maximize the structural content of the spectra. We test these pulse sequences on two metal carbonyl systems and then apply them to study amyloid fibers formed from the human islet amyloid polypeptide (hIAPP).
Hamm, Lim, and Hochstrasser (4) collected the first 2D IR spectrum in 1998 by using a hole-burning approach (Fig. 1a). They monitored the changes in the intensity of the protein carbonyl stretch band (amide I band) while using an etalon to scan a narrowband mid-IR pulse across the protein absorption spectrum. Changes in intensity indicate coupled vibrational modes, which appear as cross-peaks when plotted in a 2D representation. In 2000, Hochstrasser and coworkers (9) implemented a pulsed version of 2D IR spectroscopy, based on simulations by Zhang et al. (12), that is analogous in many ways to modern pulsed 2D NMR spectroscopy (Fig. 1b). The pulsed version uses a sequence of femtosecond mid-IR pulses to coherently excite the sample. The resulting photon echo (or free induction decay) is measured and Fourier-transformed to give the 2D IR spectrum. Both versions of 2D IR spectroscopy are now being used in many research laboratories around the world. As the field has evolved, the strengths and weaknesses of each approach have become apparent; the pulsed version is more versatile and has better time resolution than the hole-burning method, but it is also harder to accurately implement and the spectra must be carefully phased. We have used a mid-IR pulse shaper to realize several alternative methods for collecting 2D IR spectra that overcome many of the weaknesses of the two original methods. With this shaper, new pulse sequences are straightforward to implement, spectra are properly phased, and high time resolution is retained. In essence, we have combined the strengths of the two original methods while adding many unique capabilities.
Fig. 1.
Pulse sequences for different 2D IR methods. (a and b) The traditional pulse sequence for measuring 2D IR spectra by hole-burning with an etalon (a) and by a heterodyned photon echo (b). (ce) The pulse sequences we implement here use pump pulses shaped to be a time-reversed etalon (c), a Gaussian (d), and a pulse pair (e).
The mid-IR pulse shaper used here was designed recently, based on similar devices that operate in the visible frequency range (24). Femtosecond mid-IR pulses that enter the shaper have the phases and amplitudes of their spectrum tailored so that they exit with the desired shape. In this manner, one pulse can be turned into a train of pulses with computer-controllable characteristics, as is done in NMR. Here, we describe the use of our shaper to collect 2D IR spectra in several ways. First, we reproduce the original hole-burning method for collecting 2D IR spectra by mimicking the pulse shape created by an etalon (25). Etalons narrow the bandwidth of the mid-IR pulses as desired, but also cause the pulses to have exponentially long tails in time (Fig. 1a), which can distort the 2D IR spectra, as we show. Second, we improve on this method by using pulses with different time and frequency shapes that cause less spectral distortion and have better signal strengths (Fig. 1 c and d). Third, we implement a variation of the pulsed method for collecting 2D IR spectra. Pulsed methods are preferable to hole-burning because they have the best possible time resolution, which is a necessary attribute when fast structural changes, such as solvent exchange dynamics, need to be measured (11, 16, 17). However, the traditional pulsed method, unlike its hole-burning counterpart, does not automatically generate absorptive spectra or spectra that are properly phased. Instead, two separate 2D IR spectra must be collected by using different pulse sequences and be added together to generate absorptive features and optimal spectral resolution (26). Performing this addition is tricky and requires an appropriate pump-probe spectrum (27), which does not exist for all pulse sequences. Furthermore, implementing the traditional pulsed method is more challenging because it requires four separate optical paths, one for each pulse in the pulse sequence. Our version of the pulsed method uses two beams rather than four, one of which passes through our shaper to generate a femtosecond pulse pair (Fig. 1e). Because the pulses are collinear and their relative time delays are perfectly set, absorptive 2D IR spectra are automatically generated with proper phases. Thus, our collinear method retains the best aspects of both techniques: properly phased absorptive spectra with ultrafast time resolution.
Here, we first apply our approaches to a model system, W(CO)6, that has a narrow absorption band and a large anharmonicity. This system allows us to compare the resolution, phase, and spectral distortions for each method. We then measure the 2D IR spectrum of hIAPP fibrils in D2O. hIAPP fibers are involved in the disease mechanism of type 2 diabetes but are difficult to study with traditional 2D IR techniques because the fibers scatter light, which obscures the desired signal. We collected a 2D IR spectrum of hIAPP fibers by phase-cycling the pump pulses to shift the observed frequency of the scatter, leaving a high-resolution spectrum of hIAPP. The resulting 2D IR spectra are representative of β-sheet conformations with vibrational modes localized over several strands, although strong inhomogeneous broadening indicates significant structural variation. Phase-cycling capabilities allow 2D IR spectra to be better applied to heterogeneous samples such as fibers and membranes.

Results

Shown in Fig. 2 are four series of 2D IR spectra collected for W(CO)6 solvated in hexane. Each series is generated by using a different pump shape. W(CO)6 has three degenerate modes at 1,983 cm−1 that are created by antisymmetric stretches of the metal carbonyl groups. The intrinsic linewidth of the asymmetric stretch band is 4.4 cm−1 FWHM and is Lorentzian-shaped. Although solvent interactions can break the degeneracy of these three modes, the interactions are small enough in hexane (solvent) that we treat the modes identically. The first series of 2D IR experiments mimic the spectra that would be generated had an etalon been used to narrow the pump spectrum. An etalon generates a Lorentzian-like spectrum according to
where R is the reflection coefficient of the mirrors in the etalon and Δ is the path difference between each proceeding reflection (25). To generate the 2D IR spectra in Fig. 2a, we set the bandwidth to 4.6 cm−1, scanned ω0, and plotted the change in absorption of the probe with and without the pump. 2D IR spectra were collected at three different pump-probe delay times: T = 2,000, 1,000, and 200 fs. At T = 2,000 fs, a negative and positive peak pair is observed (blue and red, respectively, in Fig. 2). The negative peak appears along the diagonal with ωpump = ωprobe = 1,983 cm−1, which is the fundamental frequency of W(CO)6. The positive peak is shifted by Δωprobe = 13.3 cm−1, which is the anharmonicity of the W(CO)6 antisymmetric stretch (28). The FWHMs of the peaks are 13.4 cm−1 along ωpump and 5.3 cm−1 along ωprobe (16.5 and 7.5 cm−1, respectively, for the first overtone). This spectrum has many desirable characteristics. The spectrum is properly phased, without any postdata corrections, and the peaks have absorptive lineshapes. However, the peaks are broader than the intrinsic W(CO)6 linewidth, especially along ωpump, and the spectra exhibit aberrations that become especially grievous for T = 200 fs. With regard to the widths, the width along ωprobe is a convolution of the intrinsic linewidth with the monochromator resolution (2.2 cm−1), whereas the width along ωpump involves a convolution with the pump bandwidth (4.6 cm−1). The monochromator resolution can be improved with a different grating, but narrowing the pump bandwidth leads to larger aberrations because the aberrations are caused by the pump pulse overlapping in time with the probe pulse. An etalon narrows the pump pulse bandwidth, but doing so causes the pulse to have an exponential-like decay in the time domain (Fig. 1a). For the pump bandwidth of 4.6 cm−1 used here, the exponential tail has a time constant of 1.2 ps. Thus, the spectra become distorted for T = 1,000 and 200 fs because the probe overlaps with the exponential tail of the pump. Distortion can be avoided by collecting spectra with T delays longer than the exponential decay, but the signal strength decreases proportionately because of population relaxation. Thus, many applications are best done with T delays as short as possible.
Fig. 2.
2D IR spectra of W(CO)6 dissolved in hexane, obtained with pump pulses shaped as a traditional etalon (a), a time-reversed etalon (b), a Gaussian (c), and a pulse pair (d) for T = 2.0, 1.0, and 0.2 ps. Fifteen contours were drawn, from minimum to maximum of normalized intensities.
On the basis of the characteristics of the etalon-style 2D IR pump-probe spectra, we have explored several other pump shapes to minimize the pump and probe temporal overlap. With our pulse shaper, we can generate any frequency- and phase-tailored pump shape within the confines of the mid-IR bandwidth and the shaper resolution (22, 23). Shown in Fig. 2b are three 2D IR spectra collected by using a “time-reversed” etalon, generated by changing the sign of the exponential in Eq. 1. The pump pulse frequency distribution is identical for the normal vs. the time-reversed etalon. The difference is that the shape of the pulse is reversed in the time domain (Fig. 1c). Thus, the pump and probe pulses do not overlap until much shorter time delays, resulting in smaller peak distortions. Because T can be smaller without causing distortions, the signal strength is also stronger for a time-reversed etalon pulse shape. The improvement in signal strength is not very dramatic for W(CO)6 because the vibrational lifetime is 140 ps, but for the amide I band of a peptide or protein, for which the vibrational lifetime is ≈1,400 fs, a 3× enhancement in signal strength would occur for T = 200 vs. 1,800 fs.
We also collected 2D IR spectra by using a Gaussian-shaped pump pulse, shown in Fig. 2c. For these spectra, we set the FWHM of the Gaussian spectrum close to the etalon-style pump (FWHM = 7.0 cm−1) and the phase to zero, so that the pulse is transform-limited with 2.2 ps FWHM. The resulting spectra are similar in some respects to the time-reversed etalon. The peaks have the correct phases and little distortion, even at T = 200 fs. The main difference is that the peak shapes along ωpump now have Gaussian, rather than Lorentzian, profiles and thus better resolution. The change in profile occurs because the measured spectrum is a convolution of the natural system response with the pump pulse (see Discussion). Because the natural linewidth of W(CO)6 (4.4 cm−1) is much narrower than the FWHM of the pump pulses, the measured spectrum closely resembles the shape of the pump. Gaussian-shaped pump pulses are a better choice than Lorentzians because Gaussians decay to baseline more quickly. A comparison of the peak profiles along ωpump is shown in supporting information (SI) Fig. 4.
The fourth method we report here more closely resembles the pulsed 2D IR method of Hochstrasser and coworkers (6, 9, 12), except that the experiment is performed in a pump-probe beam geometry. In this method, the pulse shaper is used to create two transform-limited Gaussian pulses. Rather than scanning the frequencies, as was described above, the 2D IR spectrum is instead generated by collecting the signal as a function of the time delay between the two pulses, τ, and Fourier-transforming it to give the ωpump axis. To mimic the way that pulse delays are typically generated using translation stages, the phases of the two pump pulses inside their respective envelopes were held constant (ϕ1 = ϕ2 = 0). The resulting spectra are shown in Fig. 2d for τ = 0–10,000 fs, in 14-fs steps. As in the hole-burning methods, the spectra are automatically phased and have absorptive lineshapes, but now the narrowest W(CO)6 pump linewidth is measured and there is no discernable distortion in the peaks, even at T = 200 fs. Thus, we are able to generate the ideal 2D IR spectrum in a straightforward manner. As is explained in more detail below, this pulsed method retains the phasing and absorptive features of the pump-probe spectra because both rephasing and nonrephasing signals are collected simultaneously and the time-zeros of the pulse delays are perfectly set, resulting in a properly phased spectrum. Moreover, femtosecond pump pulses allow linewidths closest to the intrinsic resolution to be measured. We also point out that the peak shapes are symmetrical about ωpump and that there are no spurious ghost images, indicating that the shaper time resolution is sufficiently accurate. Asymmetric shapes and ghost images are common problems that occur when 2D IR spectra are collected by using translation stages to increment the time delays (29, 30). In SI Fig. 5, examples of 2D IR spectra that exhibit diagonal and cross-peaks are shown for a Ni metal dicarbonyl collected by using the Gaussian and pulse-pair pump methods.
Finally, we present 2D IR spectra of hIAPP fibrils in D2O, as shown in Fig. 3 along with the FTIR spectrum of the same sample. The 2D IR spectrum was collected using the pulse-pair method described above, with a slight alteration. hIAPP forms fibrils ≈100 nm in diameter and up to several microns in length (31). As a result, the sample scatters the mid-IR pump beam, which interferes with the desired spectrum. Scatter can be subtracted or filtered from the data after the experiment is complete, so long as the scatter phase and intensity are constant throughout data collection. This is not always the case, especially during folding studies in which aggregation or fiber elongation occurs. Rather than postcorrecting our data, we instead remove scatter during the experiment by cycling the phase of the pump pulses. For the hIAPP spectrum shown in Fig. 3, we incremented the pump pulse phases by Δϕ1 = Δϕ2 = π/15 for each τ delay while leaving the relative phase unchanged, ϕ1 − ϕ2 = 0. Because the scatter depends on the absolute phase, whereas the nonlinear signal depends on the relative phase, the observed frequency of the scatter is shifted away from the desired 2D IR spectrum. The features of this spectrum are discussed below.
Fig. 3.
FTIR and 2D IR spectrum of the amide I transition of hIAPP fibrils in D2O. The 2D IR spectrum was collected by using the pulse-pair method, which was also phase-cycled to reduce scatter. The dotted line in the FTIR panel represents the spectrum of the probe pulse.

Discussion

All of the methods described here for generating 2D IR spectra rely on the same principles (32). At least three interactions are required between the pulse sequence and the sample to generate a 3rd- or higher-order signal field. If each pulse in the sequence interacts with the sample just once, then a minimum of three pulses are necessary. Alternatively, a 2D IR spectrum can be generated from only two pulses if one pulse interacts with the sample twice. Either way, the probe pulse is one of these three interactions, which also serves to heterodyne the emitted field for phase-sensitive detection. The resulting heterodyned signal is Fourier-transformed by the monochromator and collected in the frequency domain. The measured signal can be written as
where E1, E2, and Eprobe are the three necessary electric fields; Rn(τ, T, t) are the vibrational and orientational responses of the system that contains the desired molecular information; and ⊗s represent convolutions of the electric fields with Rn(τ, T, t). In the first three methods described above that use various forms of hole-burning (Fig. 1 a, c, and d), the pump beam interacts twice with the sample, accounting for both E1 and E2 in Eq. 2. The second frequency dimension is generated by narrowing the bandwidth of the pump pulse and scanning its center frequency, thereby integrating the signal in Eq. 2 for τ = 0. In the fourth method (Fig. 1e), the pulses all have broad bandwidths and the 2D IR spectrum is instead generated by Fourier-transforming the signal along τ by means of
The molecular response is a sum of so-called rephasing and nonrephasing signals:
where each term in the summation corresponds to a different Feynman pathway that creates a peak in the 2D IR spectrum. To generate absorptive spectra, Rnrephase and Rnnonrephase must be measured, phased, and added (26, 27). When three pulses are used that each have a unique wavevector, such as in a boxcar configuration, then Rnrephase and Rnnonrephase must be measured separately because they appear in spatially distinct directions as a result of phase-matching. In the pump-probe phase-matching geometry used here, Rnrephase and Rnnonrephase are measured simultaneously because they are emitted in the same direction. Furthermore, because Eprobe serves as both an excitation and a heterodyning pulse, and because the time-zeros of the E1 and E2 pulses are perfectly set, the measured signal is correctly phased.
Because all four methods presented here measure the same molecular response, Rn(τ, T, t), the 2D IR spectra all contain the same dynamical and structural information. The differences arise in the convolution of the molecular responses with the envelopes of the electric fields E1 and E2. In the first three methods, E1 and E2 are generated by the same pump pulse whose bandwidth is narrowed to about the homogeneous linewidth of the sample. Thus, the time evolution of the molecular response is convoluted over temporally extended pump pulses, limiting the time resolution of the technique to the temporal width of the pump pulse. The differences in the first three methods arise in the temporal envelopes of the pump pulses, as shown in Fig. 1. For the etalon-style method, the E1 and E2 envelopes are exponential decays. Thus, for a given T delay such as T = 2,000 fs, the measured signal contains contributions to the molecular response for all 2,000 ≥ T ≥ 0, albeit with smaller contributions. When the exponential tail extends past T = 0, then other Feynman pathways contribute to the signal and are responsible for the peak distortions shown in Fig. 2 along with nonresonant effects. The advantage of the reverse-etalon method over a standard etalon is that the pulse envelope extends away from the probe pulse rather than toward it. This orientation has two advantages. First, the T delay can be much smaller without contributions from nonresonant signals and other molecular responses. Second, the signal from the exponential tail for T ≥ 2,000 has diminishing importance because of population relaxation, whereas the signal from the exponential tail of a traditional etalon is enhanced because smaller T times correspond to stronger signal strengths. Distortion and unwanted Feynman pathways also contribute to Gaussian pump pulses, but only at small T delays when the pump and probe pulses overlap. The other advantage of Gaussian over etalon-style pulse shapes is that the convolution results in 2D IR spectra with Gaussian rather than Lorentzian lineshapes. Gaussian lineshapes are preferable because their profiles do not extend over as large a frequency range as Lorentzians of the same FWHM. Eliminating long Lorentzian tails improves the spectral resolution in 2D IR spectra of peptides and other molecules with multiple absorption bands in which the long tails can spectrally interfere. Many other pulse shapes might be useful as well.
In contrast to the first three frequency-scanning methods, the fourth method operates solely in the time domain. Because E1 and E2 are collinear, the Rnrephase and Rnnonrephase signals are emitted collinearly with the probe pulse, resulting in absorptive spectra, as in the first three methods. However, the pulse pairs in the pump beam are transform-limited and have the maximum bandwidth possible, resulting in optimal time resolution. With the fourth method, structural dynamics can be measured to ≈50 fs accuracy (limited by the time duration of our mid-IR pulses), as compared with 2–3 ps for the first three methods. This method also gives the best frequency resolution because the molecular response can be measured for many picoseconds, resulting in a high-resolution Fourier transform. Furthermore, measuring the spectra in this manner alleviates many experimental uncertainties associated with the traditional four-beam setup. In the four-beam setup, the temporal overlaps cannot be perfectly calibrated, resulting in improperly phased spectra (26, 27). The pump-probe style of pulsed 2D IR spectroscopy alleviates this difficulty. Thus, this method has the best possible combination of time resolution, frequency resolution, and phase accuracy.

Advantages of Pulse-Shaping 2D IR Spectroscopy.

Besides its improved accuracy and ease of implementation, 2D IR spectroscopy by means of pulse-shaping has several other advantages. First, data collection is much more rapid. In addition to the obvious advantage of only having to collect a single spectrum rather than two (a rephasing and a nonrephasing spectrum), in a traditional setup most of the data collection time is spent moving the translation stages to increment the time delays. Because there are no moving parts and the time delay can be incremented for each laser shot, data collection is extremely efficient. We can collect a high signal-to-noise hIAPP spectrum in as little as ≈70 s, and this rate could be reduced further with a higher repetition-rate laser system. Second, more complicated pulse sequences can now be explored easily. The experiments presented here are all based on third-order responses of the sample to the incident electric fields. Trains of five pulses could be used to measure higher-order correlations, non-Gaussian frequency fluctuations, and 3D IR spectra, which have recently been reported (14, 30, 33). Generating the two additional pulses necessary to implement these experiments is trivial with our pulse shaper, but can be very tedious using traditional optics. Third, phase-cycling can also now be applied to select particular Feynman paths (34), to isolate higher-order responses, or to average-out background scatter, as was done with hIAPP. Removing scatter should be especially advantageous in heterogeneous membrane systems. Fourth, it should be possible to simplify and more precisely measure the 2D IR spectra of complex molecules by actively controlling vibrational populations (35). We have recently found that we can selectively populate vibrational levels in a metal carbonyl system with properly shaped and intense mid-IR pulses (D.B.S., S.-H.S., and M.T.Z., unpublished data). Finally, our approach of using shaped beams to collect multidimensional spectra is not limited to the IR. Phase stability is much harder to achieve in 2D electronic spectroscopy because of the shorter wavelengths, but it should be possible to collect 2D electronic spectra straightforwardly by using the methods described here with common and commercially available visible pulse shapers (3740).

Structural Diversity of hIAPP Fibers.

hIAPP consists of 37 amino acids and forms long fibers in the β-cells of the pancreas. Although the disease mechanism is not understood, there is much interest in the structural diversity of hIAPP because aggregated, prefibrillar forms of hIAPP are more cytotoxic than fully formed fibers (41). Two-dimensional IR spectroscopy provides a good means for monitoring structural diversity because structural heterogeneity causes broad distribution of couplings and hydrogen bond strengths that are reflected in the shapes and intensities of the diagonal and cross-peaks. The 2D IR spectrum of the amide I band of hIAPP (the carbonyl stretch of the peptide backbone) consists of a doublet with a negative peak at ωpump = ωprobe = 1,620 cm−1 (Fig. 3). This peak corresponds to the antisymmetric stretch mode (α-) of a β-sheet. There is also a long negative feature extending to 1,670 cm−1 along the diagonal that is created from random coil conformations. Cross-peaks appear along ωpump at ωprobe = 1,610 cm−1 that extend up to 1,660 cm−1, indicating that the random coil and β-sheet conformations are coupled. It is common to observe cross-peaks that are more intense on one side of the diagonal than the other in 2D IR spectra when multiple conformations are present (42). An atomic-level structure of hIAPP fibrils does not exist, but it is known that they have a cross β-strand conformation similar to amyloid fibers associated with other diseases, such as Alzheimer's (43, 44). The β-sheet structure of amyloid fibers is consistent with the features in the 2D IR spectrum. However, the 2D IR spectra also provide strong evidence that the fibrils are structurally disordered. The β-sheet peaks at 1,620 cm−1 are elongated along the diagonal, which is indicative of a large inhomogeneous distribution caused by a diverse structural ensemble (45). Furthermore, cross-peaks to a symmetric β-sheet stretch mode are also absent, which is a prominent feature in typical β-sheet proteins such as Con A (45), again suggesting large amounts of disorder. It is possible that the fibers are undergoing dynamical β-sheet rearrangements that intermix the β-sheet and random coil conformations (46). Dynamics like this would be consistent with the broad and intense cross-peaks that appear between the β-sheet and random coil conformations and indicate strong but disordered vibrational coupling. Further experiments that use polarization and isotope labeling will help to ascertain more quantitative information. Quantitative simulations will also require better characterization of vibrational coupling between stacked β-sheets, although coupling across 2D β-strands is rather well understood (36, 45, 47). Nonetheless, this 2D IR spectrum indicates that hIAPP amyloid fibers have a large structural distribution of secondary, and perhaps quaternary, stacking structures that create strong cross-peaks and elongated diagonal peaks in the 2D IR spectrum. It will be of much interest to use these features as structural markers in future experiments that monitor the time-resolved folding of hIAPP into amyloid fibers.

Methods

Femtosecond mid-IR pulses are generated by difference frequency mixing the signal and idler beams from an optical parametric oscillator that is pumped by a 45-fs Ti:sapphire regenerative amplifier running at 1 kHz. The resulting pulses have bandwidths >320 cm−1 at 5 μm and >220 cm−1 for 6-μm (<55-fs) pulses with energies of 4–5 μJ. About 50 nJ are used for the probe beam; the remaining light passes through the pulse shaper to generate the pump beam. The shaper consists of a pair of diffraction gratings, a pair of cylindrical mirrors, and a Ge acoustooptic modulator (AOM) set in a standard 4-f geometry (22). Pulses are shaped by generating an acoustic wave in the AOM with a 300 million samples per second arbitrary waveform generator (AWG). The acoustic wave deflects particular frequencies with set amplitudes and phases, depending on the desired pulse shape. The shaper has 190 resolvable frequency elements under our current optical setup, although this can be expanded to 500 elements with tighter focusing. The resulting pump pulses are ≈0.8 μJ at ≈6 μm and ≈1.5 μJ at ≈5 μm. Because it takes only 10 μs for the acoustic waves to propagate through the AOM, compared with the 1-ms duty cycle of the laser system, a different pump shape can be generated for each laser shot. The AWG has 1 Mbyte of memory, which is sufficient to hold 338 different pulse shapes that can be cycled automatically without pausing data collection to upload new waveforms. In the pulsed method described above, pulse pairs with time separations of up to 35 ps can be generated. The probe is dispersed in a monochromator and collected on a 32-channel array detector to give a frequency resolution of 2.2 or 5.2 cm−1 with gratings of 150 and 75 lines per millimeter, respectively. The samples have a thickness of 56 μm. For the metal carbonyl compounds, the concentration is set to give an optical density of 0.3. hIAPP was obtained from Bachem (Torrance, CA), dialysized to remove residual trifluoroacetic acid, solvated in D2O for an optical density of 0.1 (1 mM with a 56-μm sample thickness), and allowed to equilibrate into folded amyloid fibers. All spectra are collected with pump and probe pulses vertically polarized.

Abbreviation

hIAPP
human islet amyloid polypeptide.

Acknowledgments

This work is supported by National Science Foundation Grant CHE0350518, National Institutes of Health Grant R21AI064797, The Packard Foundation, and The Beckman Foundation. S.-H.S. acknowledges the Kwanjeong Educational Foundation for a fellowship.

Supporting Information

Adobe PDF - 00804Fig4.pdf
Adobe PDF - 00804Fig4.pdf
Adobe PDF - 00804Fig5.pdf
Adobe PDF - 00804Fig5.pdf

References

1
S Mukamel Annu Rev Phys Chem 51, 691–729 (2000).
2
NH Ge, RM Hochstrasser PhysChemComm 5, 17–26 (2002).
3
MH Cho PhysChemComm 5, 40–58 (2002).
4
P Hamm, MH Lim, RM Hochstrasser J Phys Chem B 102, 6123–6138 (1998).
5
S Woutersen, P Hamm J Phys Chem B 104, 11316–11320 (2000).
6
MT Zanni, S Gnanakaran, J Stenger, RM Hochstrasser J Phys Chem B 105, 6520–6535 (2001).
7
O Golonzka, M Khalil, N Demirdoven, A Tokmakoff Phys Rev Lett 86, 2154–2157 (2001).
8
AT Krummel, MT Zanni J Phys Chem B 110, 13991–14000 (2006).
9
MC Asplund, MT Zanni, RM Hochstrasser Proc Natl Acad Sci USA 97, 8219–8224 (2000).
10
P Mukherjee, I Kass, I Arkin, MT Zanni Proc Natl Acad Sci USA 103, 3528–3533 (2006).
11
JJ Loparo, ST Roberts, A Tokmakoff J Chem Phys 125, 194521 (2006).
12
WM Zhang, V Chernyak, S Mukamel J Chem Phys 110, 5011–5028 (1999).
13
EC Fulmer, P Mukherjee, AT Krummel, MT Zanni J Chem Phys 120, 8067–8078 (2004).
14
EC Fulmer, F Ding, P Mukherjee, MT Zanni Phys Rev Lett 94, 067402 (2005).
15
HS Chung, M Khalil, AW Smith, Z Ganim, A Tokmakoff Proc Natl Acad Sci USA 102, 612–617 (2005).
16
YS Kim, RM Hochstrasser Proc Natl Acad Sci USA 102, 11185–11190 (2005).
17
JR Zheng, K Kwak, J Asbury, X Chen, IR Piletic, MD Fayer Science 309, 1338–1343 (2005).
18
C Kolano, J Helbing, M Kozinski, W Sander, P Hamm Nature 444, 469–472 (2006).
19
V Cervetto, J Helbing, J Bredenbeck, P Hamm J Chem Phys 121, 5935–5942 (2004).
20
IV Rubtsov, JP Wang, RM Hochstrasser J Phys Chem A 107, 3384–3396 (2003).
21
K Heyne, N Huse, ETJ Nibbering, T Elsaesser Chem Phys Lett 369, 591–596 (2003).
22
SH Shim, DB Strasfeld, EC Fulmer, MT Zanni Opt Lett 31, 838–840 (2006).
23
SH Shim, DB Strasfeld, MT Zanni Opt Express 14, 13120–13130 (2006).
24
MA Dugan, JX Tull, WS Warren J Opt Soc Am B 14, 2348–2358 (1997).
25
G Hernandez Fabry–Perot Interferometers (Cambridge Univ Press, New York, 1986).
26
M Khalil, N Demirdoven, A Tokmakoff Phys Rev Lett 90, 047401 (2003).
27
JB Asbury, T Steinel, MD Fayer J Lumin 107, 271–286 (2004).
28
T Witte, JS Yeston, M Motzkus, EJ Heilweil, KL Kompa Chem Phys Lett 392, 156–161 (2004).
29
V Volkov, R Schanz, P Hamm Opt Lett 30, 2010–2012 (2005).
30
F Ding, P Mukherjee, M Zanni Opt Lett 31, 2918–2920 (2006).
31
ET Jaikaran, CE Higham, LC Serpell, J Zurdo, M Gross, A Clark, PE Fraser J Mol Biol 308, 515–525 (2001).
32
S Mukamel Principles of Nonlinear Spectroscopy (Oxford Univ Press, New York, 1995).
33
F Ding, MT Zanni Chem Phys, in press.
34
PF Tian, D Keusters, Y Suzaki, WS Warren Science 300, 1553–1555 (2003).
35
D Abramavicius, S Mukamel J Chem Phys 120, 8373–8378 (2004).
36
S Hahn, SS Kim, C Lee, M Cho J Chem Phys 123, 084905 (2005).
37
WP de Boeij, MS Pshenichnikov, DA Wiersma Chem Phys 233, 287–309 (1998).
38
SMG Faeder, DM Jonas J Phys Chem A 103, 10489–10505 (1999).
39
T Brixner, T Mancal, IV Stiopkin, GR Fleming J Chem Phys 121, 4221–4236 (2004).
40
JC Vaughan, T Hornung, KW Stone, KA Nelson J Phys Chem A, in press.
41
J Janson, RH Ashley, D Harrison, S McIntyre, PC Butler Diabetes 48, 491–498 (1999).
42
MT Zanni, NH Ge, YS Kim, RM Hochstrasser Proc Natl Acad Sci USA 98, 11265–11270 (2001).
43
M Sunde, C Blake Adv Protein Chem 50, 123–159 (1997).
44
R Tycko Curr Opin Struct Biol 14, 96–103 (2004).
45
N Demirdoven, CM Cheatum, HS Chung, M Khalil, J Knoester, A Tokmakoff J Am Chem Soc 126, 7981–7990 (2004).
46
SA Petty, SM Decatur Proc Natl Acad Sci USA 102, 14272–14277 (2005).
47
W Zhuang, D Abramavicius, S Mukamel Proc Natl Acad Sci USA 102, 7443–7448 (2005).

Information & Authors

Information

Published in

Go to Proceedings of the National Academy of Sciences
Proceedings of the National Academy of Sciences
Vol. 104 | No. 36
September 4, 2007
PubMed: 17502604

Classifications

Submission history

Received: January 29, 2007
Published online: September 4, 2007
Published in issue: September 4, 2007

Keywords

  1. femtosecond spectroscopy
  2. infrared spectroscopy
  3. pulse shaping
  4. protein structure

Acknowledgments

This work is supported by National Science Foundation Grant CHE0350518, National Institutes of Health Grant R21AI064797, The Packard Foundation, and The Beckman Foundation. S.-H.S. acknowledges the Kwanjeong Educational Foundation for a fellowship.

Notes

This article is a PNAS Direct Submission.
This article contains supporting information online at www.pnas.org/cgi/content/full/0700804104/DC1.

Authors

Affiliations

Sang-Hee Shim
Department of Chemistry, University of Wisconsin, 1101 University Avenue, Madison, WI 53706
David B. Strasfeld
Department of Chemistry, University of Wisconsin, 1101 University Avenue, Madison, WI 53706
Yun L. Ling
Department of Chemistry, University of Wisconsin, 1101 University Avenue, Madison, WI 53706
Martin T. Zanni [email protected]
Department of Chemistry, University of Wisconsin, 1101 University Avenue, Madison, WI 53706

Notes

*To whom correspondence should be addressed. E-mail: [email protected]
Author contributions: S.-H.S. and M.T.Z. designed research; S.-H.S., D.B.S., and Y.L.L. performed research; Y.L.L. contributed new reagents/analytic tools; S.-H.S. and D.B.S. analyzed data; and S.-H.S., D.B.S., and M.T.Z. wrote the paper.

Competing Interests

The authors declare no conflict of interest.

Metrics & Citations

Metrics

Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.


Citation statements




Altmetrics

Citations

If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

Cited by

    Loading...

    View Options

    View options

    PDF format

    Download this article as a PDF file

    DOWNLOAD PDF

    Get Access

    Login options

    Check if you have access through your login credentials or your institution to get full access on this article.

    Personal login Institutional Login

    Recommend to a librarian

    Recommend PNAS to a Librarian

    Purchase options

    Purchase this article to access the full text.

    Single Article Purchase

    Automated 2D IR spectroscopy using a mid-IR pulse shaper and application of this technology to the human islet amyloid polypeptide
    Proceedings of the National Academy of Sciences
    • Vol. 104
    • No. 36
    • pp. 14175-14543

    Media

    Figures

    Tables

    Other

    Share

    Share

    Share article link

    Share on social media