FOURIER TRANSFORM RAMAN SPECTROSCOPYOF BIOLOGICAL

Nov 1, 1990 - FOURIER TRANSFORM RAMAN SPECTROSCOPYOF BIOLOGICAL MATERIALS. Ira W. Levin ,. E. Neil Lewis. Anal. Chem. , 1990, 62 (21), ...
0 downloads 0 Views 7MB Size
FOURIER TRANSFORM RAMAN SPECTROSCOPY OF BIOLOGICAL MATERIALS Ira W. Levin and E. Neil Lewis Laboratory of Chemical Physics National Institute of Diabetes and Digestive and Kidney Diseases National Institutes of Health Bethesda, MD 20892

Since its discovery more than 60 years ago, the spontaneous Raman effect has enjoyed spectacular success in applications spanning the physical, chemical, biological, and engineering fields. Following the initial excitement generated primarily by physicists in the 1930s, Raman spectroscopy was used almost exclusively by chemists for about 20 years for vibrational studies of polyatomic molecules. In the early to middle 1960s, visible light lasers as convenient sources of radiation stimulated an enormous amount of interest in Raman spectroscopy. As conventional and resonance-enhanced Raman studies proliferated throughout the 1960s and 1970s, new nonlinear Raman phenomena, distinguished by a plethora of colorful acronyms from CARS to RIKES, were developed and applied to systems of chemical interest. In 1986, when many Raman specThis article not subject to U.S. copyright Published 1990 American Chemical Society

troscopists were sensing the serenity of a rapidly maturing field, several investigators (1-3) successfully demonstrated the coupling of Raman-scattered radiation to a Michelson interferometer, replacing the traditional monochromator with a multiplexing instrument. These studies stimulated scientific and economic interest from both the vibrational spectroscopic community and

REPORT

the manufacturers of FT-IR spectrometers. In the past few years, near-IR FTRaman spectroscopy has evolved as a viable technique for obtaining vibrational Raman spectra of many chemical systems that are either intractable to visible laser excitation or can benefit from the capabilities of interferometric instrumentation (4-9). Although applications to biophysical and biochemical materials have not developed as rapidly as those pertaining to traditional chemical characterizations, FT-Raman spectroscopy holds great promise for providing vibrational data on vast numbers of biomolecular systems that are not immediately amenable to general Raman and IR techniques. FT-Raman spectroscopy The advantages of FT-Raman spectroscopy over conventional dispersive Raman techniques, which employ visible laser excitation and photomultiplier detection, include longwavelength excitation to frustrate or eliminate sample fluorescence and improved instrumental performance of FT spectrometers over standard monochromators. The first advantage should not be understated; for

ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990 · 1101 A

REPORT typical industrial or biological samples, the presence of a dominant fluorescence signal induced by either visible or UV laser radiation often precludes measurement of the weakly scattered Raman light. Because the intensity of the Raman signal is inversely proportional to the fourth power of the excitation wavelength, the near-IR 1064-nm radiation from the Nd:YAG lasers generally used in FT-Raman instrumentation is less efficient by factors of 18-40 in generating Raman data than, for example, argon-ion sources delivering 514.5-nm emission. In the FT-Raman experiment, this decrease in sensitivity is somewhat compensated for by the interferometer's intrinsic throughput and multiplex characteristics, referred to as Jacquinôt's and Fellgett's advantages, respectively. To achieve an effective multiplex advantage, however, we assume that the experiment is detector noise limited, as is the case for the currently available cooled photoconductive germanium (Ge) and indium gallium arsenide (InGaAs) devices. (Although future improvements are expected in detector performance, existing photoconductors now function near their theoretical detection limits). Finally, the capability for precise spectral frequency measurements, which is also implicit in interferometric instrumentation (Connes' advantage), can influence the decision to use FTRaman equipment rather than scanning, dispersive instruments. For example, in using dispersive instrumentation, one must be sure to maintain the fidelity of the resulting spectral data, particularly after lengthy signal averaging efforts to increase the signalto-noise characteristics of weakly scattering samples. Furthermore, frequency precision is critical in data manipulations involving spectral subtractions. For neat liquids, dilute solutions, or polycrystalline samples, the ability to rapidly record high-quality reproducible Raman spectra with a near-IR Nd:YAG laser and FT instrumentation is as straightforward as the conventional procedures involving visible laser excitation and dispersive monochromators. Another advantage over dispersive instruments accrues in the use of an interferometer to record moderately high-resolution (0.25 cm -1 ) Raman spectra. Once a sample is aligned with the entrance aperture of the interferometer in the FT experiment, changes in spectral resolution can be made using only computer keystrokes. (For dispersive instruments, matching the alignment of a sample capillary to an entrance slit corresponding to even 1-cm-1 spectral resolution is neither

routine nor trivial.) Despite the general advantages and utility of near-IR FT-Raman spectroscopy, the technique has developed more slowly in biology than in chemistry. This more deliberate advancement within the biological disciplines stems from several fundamental problems, including the intrinsic fragility and weakly scattering nature of most native preparations; the ubiquitous and unrelenting fluorescent properties, often representative of biological samples, which can unaccountably affect baseline contours and relative spectral intensities; the difficulties in both monitoring and controlling temperatures of aqueous dispersions; and the rather different aggregation characteristics encountered in intact cellular and model biomolecular systems that make preparation of each new biological sample a separate alignment and focusing effort. The benefits of near-IR laser radiation accrue rapidly when dealing with

biological samples—particularly when persistent fluorescing contaminants are introduced during lengthy sample handling and purification procedures. Because biomolecular samples tend to be fragile, dilute, and weakly scattering aggregates, the intensity of the Raman scattering can be increased dramatically by pelleting the biological suspension directly in the capillary sample tube by ultracentrifugation, a nontrivial exercise. (During centrifugation, the hydrated pellet, which is ultimately illuminated by the laser in the FTRaman experiment, will experience a force of about 150 000-170 000 g. This necessary sample packing effect can be compared with normal methods for visible laser approaches, in which the pelleting of lipid suspensions requires only a hematocrit centrifuge operating at a force of about 10 000 g.) In many cases, the ultracentrifugation step is required not only to record a spectrum effectively, but is necessary even to observe a signal.

Interference filter

ο"

-IH

Nd: YAG laser

y^

Beam

Optical "N fiber

Qo Sample 1 c t > Lens

Aperture

Interferometer Collimator

Filter

Transfer mirror / Moving mirror

/ Beam splitter Fixed mirror

I

Mirror (to detector)

Filters

Detector

Microcomputer

Computer Plotter

Figure 1. Schematic of the FT-Raman experiment using a 90° scattering geometry.

1102 A · ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990

Disadvantages arise when using near-IR laser radiation with samples dispersed in aqueous media. The overtone region of water absorbs radiation from both the Rayleigh signal at the exciting wavelength (9394.5 cm -1 for the Nd:YAG laser) and the longer Raman scattered wavelengths corresponding to the C-H stretching modes (6500 cm - 1 in absolute wavenumbers) of the sample. In addition to the loss of Raman scattered intensity for the C-H stretching vibrations, reproducible temperature measurements become a critical concern. In these cases, one resorts to the use of either reference cells or internal temperature calibrations that are based on the spectral sensitivity of the water spectrum. Temperature control can also be maintained by manipulating the incident laser spot size or by using fiber-optic bundles to carry the exciting and scattered radiation in conjunction with sample cooling techniques. Experimental overview

Either a 90° or a 180° backscattering geometry can be used to collect the Raman emission, which is subsequently coupled to a Michelson interferometer. We have used primarily the classic right-angle sampling geometry (Figure 1), initially adopting this approach to accommodate a variety of sampling accessories and optical arrangements (6-9). For samples capable of absorbing 1064-nm Nd:YAG illumination, such

as dilute aqueous biological dispersions, a backscattering geometry (Figure 2) may be more appropriate. Although a single fiber-optic bundle is shown bringing near-IR radiation to the sample in Figure 1, a second fiberoptic collection bundle passing between the sample and the IR source compartment can also be used. In this case, the Raman scattered radiation emanating from the end of the collection fiber replaces the interferometer's IR source, maximizing the coupling of the scattered radiation to the instrumental optics. Removal of both fibers quickly provides the 90° scattering geometry. Whenever even minor changes occur in the arrangement of the optical components or sampling accessories, with respect to the 90° instrumentation, a new phase spectrum must be obtained to ensure accurate spectral representation. Data processing of the enormous numbers of spectra generated is facilitated by using removable disk cartridge systems and a laboratory computer network. The most critical aspects of the FTRaman technique using Nd:YAG laser excitation lie in efficiently rejecting the extremely intense Rayleigh and reflected signals prior to detection and in choosing suitable, sensitive near-IR detectors. In the 90° system, up to three dielectric notch or narrow bandpass filters act to discriminate against the Rayleigh scattering, and a solid-state cooled InGaAs detector provides the

Figure 2. Schematic illustrating a 180° backscattering sample configuration.

necessary detection capability for this region of the spectrum (6). Although efficient Rayleigh line rejection can be obtained by placing the filters perpendicular to the optical axis, an accompanying loss of Raman intensity—particularly for small wavenumber shifts— becomes evident. By carefully rotating one or two of the filters, one can tune the Raman instrument to record spectra closer to the exciting line. For example, we can obtain spectra down to 100 cm -1 (9). To extend the high-frequency limits of the spectrum, the temperature of the detector can be increased. Increasing the detector temperature from —196 °C to —38 °C extends the frequency range from ~2990 cm"1 to ~3300 cm"1 (9). Under these conditions, we experience only a marginal degradation in the signal-to-noise ratio. A suitable mixing of cold and ambient-temperature nitrogen gas provides a wide range of detector temperatures stable to ±1 °C and precludes the necessity of constructing separate thermoelectric coolers for the detector element. Rapid improvements will be forthcoming in both filter and detector technology for extending the high and low frequency limits of the Raman spectrum collected by FT techniques. Membrane biophysical studies

Conventional bilayer assemblies. To understand the conformational and dynamic properties contributing to the complex physiological and pharmacological processes occurring at biological interfaces, a detailed architectural view of cellular membranes at the molecular level is required. For almost two decades, the most productive hypothesis for generating and extending membrane models has been the fluid mosaic construct, in which amphipathic lipids form a bilayer matrix for the expression of a variety of peripheral and integral membrane components (10). Although most molecular bilayer information pertains to the cellular plasma membrane, the membranes of intracellular organelles exhibit varying properties, functions, and molecular compositions, further demonstrating the richness and complexities intrinsic to biomembranes. In discussing applications designed to elucidate the properties of biomembranes, we will outline several related studies from which both advantages and snares of the FTRaman approach can be seen. Among the many physical techniques used to investigate the structural, dynamic, thermodynamic, and functional properties of biomembranes, vibrational spectroscopy provides an extraordinarily sensitive, noninvasive

ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990 · 1103 A

REPORT means for probing the many intra- and intermolecular interactions defining the membrane matrix (11). Unlike artificial membranes, biological membrane assemblies that have been subjected to several isolation and purification steps yet remain sufficiently free from fluorescence effects to allow Raman spectra to be recorded with visible laser excitation are rare. Figure 3 shows the FT-Raman spec-

trum of a bovine synaptic plasma membrane preparation obtained with 1.6 W of near-IR Nd:YAG laser power at a 4-cm -1 spectral resolution. For this sample preparation, which contains integral bilayer proteins and a peripheral protein clathrin coat surrounding the lipid vesicle, it is necessary to pellet the dilute aqueous membrane dispersion received from the cell biologist in the same sample capillary that is ulti-



I I J I

Figure 3. FT-Raman spectra of an aqueous dispersion of bovine synaptic membranes in (a) the 2900-cm" 1 C-H stretching mode region and (b) the 1700650-cm~1 spectral region showing attainable frequency precision of the spectra. (Peaks marked by an asterisk contain small contributions from nonfiltered laser lines.) 1104 A · ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990

mately used in the FT-Raman experiment. The membrane is hydrated in an aqueous medium, making heating effects from the 1064-nm laser line substantial. To prevent diffusion of the biomembrane out of the exciting beam and to prevent protein denaturation, the sample is cooled to approximately —40 °C by gently streaming cold nitrogen gas across the capillary. After a two-and-a-half-hour acquisition time, the unsmoothed spectrum (shown in Figure 3) is again representative of an ordered form of a natural membrane. Intact biomembranes often remain liquid crystalline to temperatures below even —20 °C and thus are not routinely examined as a lipid gel phase, as are model systems. The low-temperature phase is specifically characterized by the intensity patterns in the 2845and 2880-cm"1 acyl chain methylene C-H stretching mode regions and in the 1085- and 1129-cm"1 acyl chain C-C stretching mode regions. The frequency values on the spectra illustrate Connes' advantage in the determination of vibrational Raman frequencies with interferometric instrumentation. Because the accuracy of these frequencies depends on measured displacements from the exciting line, the absolute wavenumber of the Nd:YAG fundamental is measured at the same resolution of the experiment, but without the rejection filters. A precise and accurate measurement of an undistorted lineshape of the exciting fundamental thus can be obtained. If spectra of these dilute, fragile membranes could be recorded dispersively with a scanning monochromator and visible laser excitation, signal averaging times would be so great—even with the intensity advantage of the incident visible radiation—that reliable frequency measurements would probably not be obtainable because of the mechanical limitations of the monochromator. Precise frequency data are important because subtle membrane reorganizations are reflected in frequency shift information; shifts of tenths of a wavenumber are significant, for example, in characterizing gel to liquid crystalline phase transitions. Figure 4 shows the spectra of a model lipid bilayer system consisting of 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) and lanosterol in both the gel (a and b) and liquid crystalline (c and d) phases. Lanosterol, a precursor of various sterols in animals, is used as a membrane perturbant. POPC, with one saturated and one unsaturated chain, mimics a general mammalian membrane. The two phases exhibit markedly different intensity and fre-

REPORT quency patterns. Because lanosterol is a bulkier molecule than cholesterol, we have compared temperature-dependent spectra of liposomes reconstituted separately with lanosterol and cholesterol to deduce the modulating effects of sterols on the dynamic and packing properties of the membranes. Interdigitated chain bilayers: lipid/antibiotic interactions. The fluid mosaic model serves as the working hypothesis for modeling membrane behavior, although the initial simplistic view of integral proteins floating in a sea of lipids has been refined. We now recognize, for example, that proteinprotein contacts within the membrane are common and that the protein influences its surrounding matrix lipids through the formation of annular rings of boundary lipids. Recently, the suggestion and demonstration of explicit domain structures

within the bilayer have begun to provide a basis for further understanding the role of various lipid-lipid and lipid-protein interactions in modifying membrane function. Among the different types of domains available to the bilayer matrix, partially and completely interdigitated bilayer lipid chains are of current biophysical interest (12). In interdigitated chain bilayers, the lipid hydrocarbon chain moieties penetrate across the bilayer center; for completely interdigitated systems, the terminal methyl groups from the hydrophobic region of an opposing monolayer pack near the membrane's more polar interfacial region. In this packing motif, saturated chain diacylphospholipids, for example, exhibit four hydrocarbon chains per head group instead of the usual two chains per head group in the conventional bilayer. As a consequence of chain interdigitation, model

membrane systems show a decrease in the thickness of the bilayer and a concomitant ordering of the lipid chains. Consider the interaction of a series of macrolide polyenes, amphotericin B, amphotericin A, and nystatin Ai (Figure 5), with a model bilayer composed entirely of the saturated chain diacyl lipid 1,2-dipalmitoylphosphatidylcholine (DPPC). These polyene antibiotics represent a class of natural products with potent antifungal properties. Although it is clear that the mechanism involving antibiotic-cellular interaction occurs at the level of the membrane interface, the nature of the interaction by which cell death is induced by these antibiotics is not fully understood. One common hypothesis suggests that the formation of transmembrane polyene-sterol channels may be involved in allowing cellular components to leak out by increasing the

Figure 4. FT-Raman spectra of an aqueous dispersion of 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) bilayer assemblies containing 18 mol % lanosterol in the gel (a and b) and liquid crystalline (c and d) phases. 1106 A

ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990

membrane's permeability characteris­ tics. Our approach is to examine the morphological response of DPPC bilayers, reconstituted both with and without cholesterol, to the binding ac­ tivity of amphotericin Β and ampho­ tericin A. Figure 6 shows both the conventional dispersive Raman spectrum and the FT-Raman spectrum of the DPPC/amphotericin Β complex in the C-H stretching mode region. Because the hydrophobic conjugated polyene por­ tions of the antibiotic ring lead to high­ ly colored yellow-orange compounds, the selective resonance or preresonance enhancement and fluorescent ef­

fects induced by visible and UV laser radiation preclude a complete charac­ terization of the various possible mem­ brane complexes. Thus the upper curve in Figure 6, obtained with argon-ion radiation, is virtually featureless in the important 2900-cm -1 spectral region of the DPPC bilayer. In contrast, the near-IR FT-Raman spectrum reflects the bilayer complex in its low-tempera­ ture gel phase. The spectral region in Figure 6 is dominated by the two lipid chain methylene features at ~2850 and 2880 cm - 1 , which correspond to the CH 2 symmetric (in-phase) and antisymmet­ ric (out-of-phase) stretching modes, re­

Figure 5. Structures of amphotericin B, amphotericin A, and nystatin A i .

spectively. In monitoring bilayer re­ sponses and membrane perturbations, the peak height intensity ratios I2850/ I288O and I2935/I288O, determined as a function of temperature, provide use­ ful indices for gauging lipid packing ef­ fects and determining relative bilayer order/disorder parameters. In particu­ lar, the I2850/I2880 index reflects primar­ ily interchain interactions whereas the I2935/I2880 intensity ratio measures ef­ fects originating from changes in intrachain trans/gauche isomerization su­ perimposed on the chain-chain inter­ actions (11). Comparisons of the chain-chain in­ teractions between pure multilamellar DPPC assemblies and bilayer complex­ es of amphotericin A and B, in the pres­ ence and absence of cholesterol, are summarized in Figure 7 in terms of the relevant I2850A288O intensity parame­ ters. The head and tail of each arrow, representing a specific liposomal as­ sembly or complex, correspond to tem­ peratures of approximately 41.5 °C and 25 °C, respectively. The values of the I2850/I288O indices at these two tempera­ tures provide a measure of the intermolecular lipid chain order for the aque­ ous dispersion of each liposome. (Be­ cause of slight differences in the physical antibiotic-lipid packing char­ acteristics, the low gel-phase tempera­ tures may deviate several degrees from the 25 °C DPPC reference bilayer. The possible small discrepancies in tem­ perature are not problematic, however, because we are concerned primarily with the trends displayed in the fig­ ure.) Thus for the pure DPPC multi­ lamellar system, the I2850/I2880 peak height intensity ratio of about 0.77 for the 25 °C gel phase is characteristic of a bilayer whose chains are hexagonally packed. A value of about 0.98 reflects a normal liquid crystalline DPPC phase. (The phase transition temperature T m between the gel and liquid crystalline states of aqueous, uncomplexed DPPC dispersions is 41 °C.) Both amphotericin A and Β signifi­ cantly order the bilayer chains (lower values of I2850/I2880); in particular, the 25 °C spectra reflect intensity ratios of 0.67 and 0.65, respectively, for the two complexes. Because other membrane surface-active species (such as poly­ myxin B, another cyclic antibiotic) are known from X-ray diffraction studies to induce bilayer chain interdigitation, and because the observed I2850/I2880 ra­ tios in the 25 °C regime correspond to Raman data for known interdigitated systems, we interpret the lipid-ordering properties of these two polyene macrolide antibiotics as originating from the induced interdigitated chain mem­ brane morphology. In contrast to the

ANALYTICAL CHEMISTRY, VOL. 62, NO. 2 1 , NOVEMBER 1, 1 9 9 0 · 1 1 0 7 A

REPORT

Figure 6. Dispersive Raman spectrum (top) and FT-Raman spectrum (bottom) of polycrystalline 1,2-dipalmitoylphosphatidylcholine (DPPC) bilayers containing am­ photericin B. (DPPC: Amphotericin Β mole ratio equals 24:1.)

Figure 7. Ordering effects of polyene antibiotics on DPPC multilayers. The I2850/I2880 intensity ratio represents the interchain order parameter for the various liposomes at ~25 °C (tail of arrow) and 41.5 °C (head of arrow).

1108 A·ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990

spectroscopic and diffraction data de­ rived from model systems, evidence for even partially interdigitated bilayers in native or intact membrane assemblies is limited (23). Despite this paucity of information, interdigitated chain ar­ rangements may represent a potential­ ly important biological membrane ar­ chitecture (13,14). As mentioned earlier, membrane cholesterol, an intrabilayer compo­ nent, is known to modulate the order/ disorder parameters of the lipid ma­ trix. A comparison of the I2850/I2880 or­ der parameters for the various lipo­ somes shown in Figure 7 demonstrates that cholesterol orders the liquid crys­ talline properties of DPPC multilayers. Although high concentrations of cho­ lesterol generally disorder the gel phase, the lower concentrations of ste­ rol in this experiment (18 mol %) exhib­ it less of an effect. When either ampho­ tericin A or Β with 18 mol % cholesterol is added, the I2850/Ï-2880 order/disorder parameters for the membrane chains indicate a gel-phase bilayer ordering that is comparable to systems without cholesterol. If a sterol-antibiotic complex were formed within the bilayer, the gelphase disorder would be substantially greater than that reported in Figure 7 because of the disrupting properties of the complex on the lipid chain packing arrangements. Thus, the data suggest that the sterol is expelled from the interior of the bilayer in the low-temperature phase; we again speculate about the possibility of interdigitated bilayer chains in these assemblies. Also, referring to Figure 7, we observe t h a t at higher temperatures (41.5 °C) the DPPC-cholesterol-amphotericin A complex is slightly more ordered than the reference DPPC-cholesterol systems. Surprisingly, the DPPC-cholesterol-amphotericin Β complex is as disordered as liquid crys­ talline DPPC multilayers. It appears that amphotericin A and B, whose con­ jugated polyene moieties differ slight­ ly, form rather dissimilar cholesterol complexes with the zwitterionic lipid substrate. That is, in the liquid crystal­ line phase, cholesterol binds externally more strongly to amphotericin Β and is removed from the bilayer. In the am­ photericin A complex, cholesterol may still remain within the bilayer, ordering the matrix slightly. Antibiotics in the presence of sterol, however, do not have quite the strong bilayer ordering properties observed in the binary lip­ id—antibiotic systems. (To minimize the differences in preparing the com­ plexes, binary and ternary systems were hydrated from lyophilized mix­ tures of the appropriate components.)

Figure 8. Near-IR absorption spectra of H 2 0 and D 2 0 superimposed on the gelphase, near-IR FT-Raman spectrum of a DPPC bilayer.

Figure 9. Schematic of thermostatted sample assembly for 180° backscattering geometry.

Temperature measurement in aqueous dispersions. In experiments involving lipid-antibiotic interactions,

the temperatures of the systems were measured with externally placed thermocouples in which the sampling ge-

ometry was calibrated against a known system (such as pure DPPC liposomes) whose temperature behavior has been well characterized. To exploit fully the FT-Raman methodology in elucidating membrane properties, as in the determination of temperature profiles, it is necessary to measure sample temperatures precisely. The problem of assessing temperatures of aqueous dispersions is illustrated in Figure 8. The absorption spectra of H2O and D2O are superimposed on the FT-Raman spectrum of DPPC. Although it is not shown, the exciting Nd:YAG line at 1064 nm falls in the wings of H2O overtone bands, and the resulting absorption of radiation by the aqueous medium leads to significant sample heating. Increasing the Raman scattered signal by increasing the incident Nd:YAG power levels can dramatically increase the temperature of the dispersion. (Because the overtone manifold is different for D 2 0, the heating effect is diminished for D 2 0 dispersions.) In an H 2 0 medium, moderate heating effects often lead to a channeling in the sample and subsequently to sample diffusion in the capillary. Also, the Raman scattered C-H stretching mode photons, which are in a particularly important spectral region for interpreting bilayer reorganizations, are absorbed by another H 2 0 overtone, leading to a decrease in the intensity of the observed C-H stretching mode vibrations. Thus one should hydrate model bilayer systems with D 2 0 when possible; most native systems, however, are best spectroscopically surveyed as H 2 0 dispersions. For these cases the loss in C-H stretching mode signal can often be tolerated; it is virtually impossible, however, to measure the sample temperature within a desired ±0.3 °C error with either externally or internally placed thermocouples. To obviate the temperature measurement difficulties discussed above, an "internal" standard is used. A small reference cell, containing an internal standard (usually D20) is placed in a thermostatted cell assembly such that the incident Nd:YAG beam passes first through the reference cell and then through the contiguous sample capillary with its biological dispersion in an H 2 0 medium. Thus the D 2 0 is in intimate thermal contact with the sample capillary. Figure 9 presents a schematic of the thermostatted cell fabricated for the 180° backscattering geometry. Because the D 2 0 stretching modes reflect a sensitive temperature dependence, and because these D 2 0 features occur in the 2500 cm -1 spectral window, library

ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990 · 1109 A

REPORT

method. (In this example the model bilayer assembly was hydrated directly with D2O.) Although fewer tempera­ ture points were recorded than are usu­ ally required for a complete profile, a temperature-dependent curve in terms of the I2850/I2880 order/disorder param­ eters may still be fit to a two-state mod­ el (15) to extract reliable phase transi­ tion temperatures T m . The central curves on the figure rep­ resent least-squares fits of the spectral data. The outer curves represent the calculated curves plus and minus three standard deviations, respectively. Al­ though the gel phase and liquid crystal­ line phase order/disorder parameters are correct for the externally placed thermocouple, the highly cooperative, narrow phase transition region charac­ teristic of pure DPPC liposomes is, in­ stead, quite broad, yielding a low T m . For the D2O internal temperature cali­ bration method, however, the correct

curves can easily be generated for de­ termining accurate sample tempera­ tures. (If the sample is dispersed in D2O, no reference cell is required.) A separate D2O reference cell is ideal for investigating lipid-protein complexes, because D2O in the sample capillary would exchange with peripheral mem­ brane protein segments. Using a thermostatted sample chamber and a Pel­ tier device for precise temperature con­ trol, we are able to generate library spectra with a dispersive instrument employing visible laser excitation. From data reduction techniques, tem­ perature measurements well within ±0.3 °C can be attained from the FTRaman D2O spectra excited with nearIR radiation. Figure 10 shows the differences in measuring the gel to liquid crystalline phase transition region for DPPC using an externally attached thermocouple and the D2O reference calibration

1.08 "

(a)

•^ - 4

1

τ

À^Z..\ t jA '

I.

1

a

"^T

= 34.6°C

m

j j

1 1 \

0.82 1D

1.08

-

ι

ι

ι

53

T(°C)

1

(b)



0

0 m CO _f\l

P~~Tm = 41.3°C

-

0.83

20

·-·?: ι

ι

ι

1

1

T(°C)

1

I

, 1

1

57

Figure 10. Temperature profile showing the change in intermolecular order of pure DPPC multilamellar assemblies using (a) an external thermocouple to determine spectral temperatures and (b) D20 as an internal temperature calibrant. 1110 A · ANALYTICAL CHEMISTRY, VOL. 62, NO. 21, NOVEMBER 1, 1990

DPPC temperature profile is obtained, exhibiting a sharp, first-order gel to liq­ uid crystalline phase transition at 41.3 °C. The ability to merge under computer control the operations of changing the sample temperature, by means of the solid-state Peltier heat pump, with sample equilibration and spectral ac­ quisition enables us to obtain routinely both ascending and descending tem­ perature profiles, which are subse­ quently processed through our labora­ tory computer network. This method­ ology has been successfully applied to a wide variety of model and intact mem­ brane dispersions in studies designed to illuminate the predominant molecu­ lar interactions occurring at biological interfaces. Conclusion We have attempted to convey both the flavor of using long-wavelength excita­ tion coupled to interferometric instru­ mentation and the idea that biophysicists and biochemists can provide in­ sight on biomolecular problems that prove unyielding to conventional vi­ brational spectroscopic approaches. For example, we have obtained near-IR FT-Raman spectra for a wide variety of systems ranging from human and calf lung surfactants to specific proteins re­ lated to cataract formation, the expres­ sion of Alzheimer's syndrome, and the insecticidal properties of particular ba­ cilli. In the past our general guide for ex­ amining biological samples has been to use dispersive instrumentation if the molecular dispersion was amenable to visible laser excitation with either photomultiplier or charge-coupled device detectors. Our techniques have im­ proved to the point that we are now going directly to the near-IR instru­ mentation to obtain Raman spectra of biomolecules. Despite the somewhat steep learning curve that may be re­ quired for handling many biological samples, the flexibility and inherent advantages of near-IR FT-Raman spectroscopy offer sufficient overall benefits to make it the vibrational spectroscopic approach of choice for most biomolecular preparations. For example, the ease of sample alignment with sets of fiber-optic bundles; the in­ terferometer's multiplex, frequency precision, and resolution advantages; and the ability to control and deter­ mine precisely the temperatures of di­ lute, fragile molecular aggregates all justify an optimistic outlook toward applying FT-Raman spectroscopy to even the more recalcitrant biological systems.

References

(12) Mcintosh, T. J.; Daniel, R. F.; Simon, S.A. Biochim. Biophys. Acta 1983, 731, 109-14. (13) Lewis, E. N.; Levin, I. W.; Steer, C. J. Biochim. Biophys. Acta 1989, 986, 16166. (14) Levin, I. W.; Thompson, T. E.; Barenholz, Y.; Huang, C. Biochemistry 1985,24, 6282-86. (15) Kirchoff, W. H.; Levin, I. W.; J. Res. Nat. Bur. Stand. 1987,92,113-28.

NIH in 1963, his interests have ranged from absolute intensity studies and normal coordinate analyses to eluci­ dating the lipid-lipid and lipid-protein interactions governing reorgani­ zations in biological membranes. Cur­ rent interests include application of IR and Raman spectroscopies to char­ acterize novel lipid morphologies.

(1) Hirschfeld, T.; Chase, D. B. Appl. Spectrosc. 1986,40,133-37. (2) Chase, D. B. J. Am. Chem. Soc. 1986, 108, 7485-88. (3) Jennings, D. E.; Weber, Α.; Brault, J. W. Appl. Opt. 1986,25, 284-90. (4) Zimba, C. G.; Hallmark, V. M.; Swalen, J. D.; Rabolt, J. D. Appl. Spectrosc. 1987, 41, 721-26. (5) Jennings, D. E.; Weber, Α.; Brault, J. W. J. Mol. Spectrosc. 1987,126,19-28. (6) Lewis, E. N.; Kalasinsky, V. F.; Levin, I. W. Appl. Spectrosc. 1988,42,1188-93. (7) Lewis, E. N.; Kalasinsky, V. F.; Levin, I. W. Anal. Chem. 1988,60, 2658-61. (8) Lewis, E. N.; Kalasinsky, V. F.; Levin, I. W. Anal. Chem. 1988,60, 2306-09. (9) Lewis, E. N.; Kalasinsky, V. F.; Levin, I. W. Appl. Spectrosc. 1989,43,156-59. (10) Nicolson, G. L.; Singer, S. J. Science 1972,/75, 720-31. (11) Levin, I. W. In Advances in Infrared and Raman Spectroscopy; Clark, R.J.M.; Hester, R.E., Eds.; Wiley-Heyden: New York, 1984; Vol. 11, pp. 1-48.

Ira W. Levin, deputy chief of the Lab­ oratory of Chemical Physics and chief of the section on Molecular Biophysics at the National Institute of Diabetes and Digestive and Kidney Diseases at NIH, received a B.S. degree from the University of Virginia and a Ph.D. from Brown University. Since joining

E. Neil Lewis received a B.S. degree and a Ph.D. from the Polytechnic of Wales in the United Kingdom. His re­ search interests include the use of Ra­ man and IR spectroscopies for study­ ing biomembranes and related macromolecular systems of biophysical relevance.

We would like to thank the Perkin-Elmer Corpo­ ration for the generous loan of an FT-Raman spec­ trometer. We would also like to acknowledge the participation of R. G. Adams and V. F. Kalasinsky in these studies.

Malvern - the complete answer to particle characterization Malvern offer the most comprehensive and advanced range of particle sizing and zeta potential equipment available anywhere.

Zeta potential and submicron particle sizing - the Malvern ZetaSizer 3 offers the ulti­ mate in PCS particle sizing using unique multi-angle analysis for reliable, de­ tailed size distributions from 3 nanometres to 3 microns. Its electro­ phoresis capabilities extend from strong electrolyte media to non-polar liquids such as hydrocarbons with full zeta potential distributions measured m seconds For characterizing colloidal dispersions, it has no equal.

Direct dry powder analysis with MasterSizer - already the world's most popular laser par­ ticle sizer, its new Dry Powder Feeder makes it even quicker and more versatile. Overall size range is 0.1-600 microns and MasterSizer is the only one to offer genuine Mie analysis.

The Malvern API AeroSizer uses aerodynamic techniques combined with advanced laser technology to give extremely high resolution particle size analysis of powders from 0.5 to 200 microns

Malvern Instruments Inc. 10 Soultwlie Rd. Southbotough. MA 01772. USA Tel (508) 480 0200 Fax (508) 460 9692 Malvern Instruments Limited Spring Lane South. Marvem. Worcs WR14 1AQ. UK Telephone 10684) 892456 Telex 339679 Fax (0684| 892789

ANALYTICAL CHEMISTRY, VOL. 62, NO. 2 1 , NOVEMBER 1, 1990 · 1111 A