Multiparametric Optimization of 31P NMR Spectroscopic Analysis of

May 5, 2010 - The quality of NMR spectra in general and of spectra to be used for analysis of compound mixtures in particular is essentially defined b...
0 downloads 11 Views 1MB Size
Download PDF
Anal. Chem. 2010, 82, 5441–5446

Multiparametric Optimization of 31P NMR Spectroscopic Analysis of Phospholipids in Crude Tissue Extracts. 2. Line Width and Spectral Resolution Norbert W. Lutz* and Patrick J. Cozzone Centre de Re´sonance Magne´tique Biologique et Me´dicale, UMR CNRS 6612, Faculte´ de Me´decine de la Timone, Universite´ de la Me´diterrane´e, 13005 Marseille, France The quality of NMR spectra in general and of spectra to be used for analysis of compound mixtures in particular is essentially defined by two basic parameters: signal-tonoise ratio and spectral resolution. The latter is determined by signal dispersion (chemical shift differences) and line widths. The present study focuses on multiparametric optimization of spectral resolution in 31P NMR spectra of phospholipids from brain tissue extracts. This report presents, for the first time, a systematic and comprehensive study of phospholipid 31P NMR line widths as a function of four experimental parameters: (i) extract concentration, (ii) concentration of a chelating agent, (iii) pH of the aqueous component of the solvent system, and (iv) temperature of the NMR measurement. Theoretical underpinnings of observed line width variations (transversal relaxation effects) are briefly discussed. In conjunction with an analogous, concurrently published report on chemical shift effects in the same tissue extract system, this multiparametric line width study provides a complete set of methodological guidelines for (i) generating well-defined tissue extracts, and (ii) choosing matched and optimized measurement conditions for highly reproducible and well-resolved 31P NMR spectra of brain phospholipids. This study also offers a comprehensive database and a strategy for rational and efficient optimization of phospholipid spectra from other tissue extracts. 31

P NMR phospholipid (PL) analysis of tissue extracts is, under appropriate circumstances, an intrinsically quantitative, fast, and accurate method for measuring PL profiles and can be applied to crude extracts without any need for prior PL separation.1 For the most frequently used PL solutions based on methanol-chloroform-water solvent systems, the effects of varying the water and methanol content on line width (LW) of PL signals have been previously determined.2,3 However, little * Corresponding author. Phone: +33 491 324476. Fax: +33 491 256539. E-mail: Norbert.Lutz@ univmed.fr. (1) Sotirhos, N.; Herslof, B.; Kenne, L. J. Lipid Res. 1986, 27, 386–392. (2) Edzes, H. T.; Teerlink, T.; van der Knaap, M. S.; Valk, J. Magn. Reson. Med. 1992, 26, 46–59. (3) Branca, M.; Culeddu, N.; Fruianu, M.; Serra, M. V. Anal. Biochem. 1995, 232, 1–6. 10.1021/ac100515y  2010 American Chemical Society Published on Web 05/05/2010

attention has been paid to the influence that other extract components may exert on line shape and signal separation. Part 1 of this multiparametric optimization study4 has shown that, in addition to avoiding saturation and polarization transfer effects, four experimental parameters need to be adjusted to obtain highly reproducible 31P NMR spectra from tissue PL extracts: (i) extract concentration, (ii) chelator concentration, (iii) pH of the aqueous component of the solvent mixture, and (iv) measurement temperature. A well-defined one-phase methanolchloroform-water solvent mixture was employed in Part 1, which was focused on relative and absolute peak positions determined by chemical shifts. However, the potential to separately quantitate a high number of PL depends not only on the distance between peak maxima, but also on the LW, since both together determine the spectral resolution that can be achieved. For this reason, we present here in Part 2, an analogous study aimed at characterizing the effects of the aforementioned experimental parameters on PL line width. As in Part 1, the main goal of this work is to provide the information and strategies needed for systematic optimization of PL analysis by 31P NMR spectroscopy. In addition, some basic physical mechanisms are discussed that control LW as a function of the aforementioned experimental parameters. Finally, a compilation of practical rules that integrates insights derived from both chemical shift and LW studies is provided. EXPERIMENTAL SECTION Sample Preparation, Acquisition of NMR Spectra. All experimental details have been exhaustively described in Part 1 of this study.4 Processing and Evaluation of NMR Spectra. The spectra processing procedure has been essentially described in Part 1 of this study. In addition, Bruker’s TOPSPIN Gaussian/Lorentzian fit procedure (MDCON) was used to determine LW () fwhh, full width at half height). LW were analyzed for phosphatidylcholine (PtdC) signals only, since the other PL peaks either were composed of multiple poorly resolved resonances, significantly overlapped with neighboring signals, or were extremely weak. These circumstances rendered precise determination of LW virtually impossible for signals other than PtdC. By contrast, the PtdC peak represents a single strong resonance that does not (4) Lutz, N. W.; Cozzone, P. J. Anal. Chem. 2010, DOI: 10.1021/ac100514n.

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

5441

overlap with any other peak, save two extremely weak signals that do not significantly affect the PtdC line shape. Apparent activation energies, Ea, have been obtained from the Arrhenius equation for T *2 , that is, from the slope characterizing the linear relationship between ln(1/T*2 ) and the inverse temperature, 1/T, where T*2 ) 1/(π × LW). THEORETICAL BASIS Effects of Phospholipid Complexation on 31P NMR Line Width in Crude Extracts. LWs are controlled by the chemical composition of PL samples in conjunction with NMR experimental parameters. In analogy to processes influencing chemical shift (Part 1 of this study), the processes determining the PL line width are rather complex and difficult to discern for impure reverse micelles and microemulsions5 that may occur in crude tissue extracts. Divalent and, to a lesser degree, trivalent metal cations readily form complexes with the phosphate moiety in PL. This is the principal mechanism for cation extraction into the chloroform/ methanol phase. Since many metal cations in mammalian tissue are paramagnetic (e.g., Mn2+, Fe2+, Fe3+, Co2+, Cu2+), these exchanging ligands broaden PL 31P resonances due to spin-spin (electron spin-nuclear spin) relaxation. Furthermore, exchange processes with diamagnetic metal cations (e.g., Mg2+, Ca2+) also contribute to line broadening in phosphates.6 These line-broadening effects can be abrogated by adding chelating agent, CDTA (trans-1,2-cyclohexyldiaminetetraacetic acid, cesium salt), which forms extremely stable chelates with the aforementioned cations. CDTA is to be preferred over the more commonly used EDTA (ethylenediaminetetraacetate) because stability constants of binary CDTA complexes are higher when compared to EDTA complexes (e.g., log K ) 12.8 and 10.6 for CDTA and EDTA, respectively, for Ca2+ at pH 7.4).7,8 To minimize complex formation of monovalent metal cations with the phosphate moieties of PL, an appropriate counterion has to be chosen for CDTA. The largest nonradioactive alkali metal ion, Cs+, is widely used for this purpose. Due to their ionic character, CDTA molecules are associated with polar solvent molecules, that is, a solvation sphere consisting of water and methanol. Hence, solubility and complex formation of CDTA also depend on the ratio of CDTA molecules to water and methanol molecules in conjunction with pH, which is of particular importance at high CDTA concentrations. Overall, optimization of PL analysis by 31P NMR spectroscopy has to take into account both CDTA-dependent LW and chemical shift effects. In addition to the divalent cations mentioned above, cation exchange effects may also be due to monovalent alkali cations, such as K+ and Na+,9 that are highly concentrated in the intracellular and extracellular space of tissues, respectively. Although most of these ions do not enter the chloroform/ methanol phase during tissue extraction, a small fraction is introduced through binding to PL and residual proteins. (5) Langevin, D. Annu. Rev. Phys. Chem. 1992, 43, 341–369. (6) McCain, D. C. In Phosphorus NMR in Biology; Burt, C. T., Ed.; CRC Press: Boca Raton, 1987; pp 26-61. (7) Harrison, D. G.; Long, C. J. Physiol. 1968, 199, 367–381. (8) Schwarzenbach, G.; Gut, R.; Anderegg, G. Helv. Chim. Acta 1954, 37, 937– 957. (9) Merchant, T. E.; Glonek, T. Lipids 1992, 27, 551–559.

5442

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

Effects of Phospholipid Aggregation on 31P NMR Line Width in Crude Extracts. Intermolecular lipid aggregation, which is a function of tissue extract concentration, affects the LW of PL 31P resonances. As described in the context of chemical shift effects (Part 1 of this study), there is a direct relationship between the water/surfactant molar ratio and the size of reverse micelles.10,11 Since the amount of “free” water (not bound to the PL polar heads) increases relative to the amount of “bound” water with increasing micellar size,12 micellar size determines the structure of water molecules within the micelle.11 Consequently, water exchange with PL phosphate moieties varies as a function of micelle formation and PL concentration, and 31P NMR signal broadening is to be expected for intermediate exchange rates. In addition, varying intramicellar water/methanol ratios as a function of PL concentration4 may affect the 31P NMR signal LW through the interaction of solvent molecules with the polar PL heads (see Part 1). Furthermore, in addition to the mobility of intramicellar molecules, also PL structures themselves are linked to micellar size.12 The transition from micelles to more ordered structures,13 potentially containing PL bilayers,12 would render PLs less mobile due to less structure breaking solubilization. Therefore, 31P lines would broaden considerably in the process, potentially via the effects of chemical shift anisotropy (CSA).5 For free PL, and for mobile PL in small micelles and micelle-like aggregations, CSA is largely averaged out. However, CSA modulation by molecular tumbling provides a spin-lattice and a spin-spin relaxation mechanism in cases that correlation times are sufficiently long, in addition to the chemical exchange mentioned above. These mechanisms may become particularly significant at high magnetic fields. Effects of pH and Temperature on 31P NMR Line Width in Crude Extracts. The pH of the aqueous component of the solvent affects the PL 31P NMR signal LW14 by three different mechanisms previously described for chemical shift effects (see also Part 1): (i) Directly via protonation of the PL phosphate moiety.15 (ii) Indirectly via protonation of compounds interacting with cations that potentially complex with the PL phosphate moiety. The latter effect most certainly applies to CDTA, since the stability constant of metal chelates is known to be pHdependent. In all these cases, intermediate rates for cation exchange with the PL phosphate moiety would result in relatively broad PL 31P NMR signals, as described in the preceding paragraphs. Furthermore, (iii) indirect effects of pH on the molecular mobility and structure of micelle-like aggregations cannot be excluded. The temperature of the sample during spectrum acquisition markedly affects both the LW and chemical shift of PL 31P NMR (10) Bru, R.; Sanchez-Ferrer, A.; Garcia-Carmona, F. Biochem. J. 1995, 310, 721–739. (11) Marhuenda-Egea, F. C.; Piera-Velazquez, S.; Cadenas, C.; Cadenas, E. Archaea 2002, 1, 105–111. (12) Kumar, V. V.; Manoharan, P. T.; Raghunathan, P. J. Biosci. 1982, 4, 449– 454. (13) Lopez, O.; de la Maza, A.; Coderch, L.; Lopez-Iglesias, C.; Wehrli, E.; Parra, J. L. FEBS Lett. 1998, 426, 314–318. (14) Silvander, M.; Karlsson, G.; Edwards, K. J. Colloid Interface Sci. 1996, 179, 104–113. (15) Sappey-Marinier, D.; Letoublon, R.; Delmau, J. J. Lipid Res. 1988, 29, 1237– 1243.

Figure 1. PtdC line width as a function of measurement temperature, obtained by 31P NMR spectroscopy of rat brain tissue extracts at 162 MHz (9.4 T). Each diagram shows the data obtained for a particular extract preparation protocol. Line widths obtained for different concentrations (by serial dilution) are color-coded. Lines tend to be particularly narrow for low extract concentrations and low temperatures, notably at high CDTA concentrations in the aqueous component of the solvent. Line widths below 1 Hz were observed only for 1000 mM CDTA and for 200 mM CDTA at pH 7.33. Temperature-dependent variations increased with extract concentrations, and were largest for samples with the lowest CDTA concentrations.

signals.16 The mechanisms involved include effects on molecular mobility and cation exchange rates. On the basis of considerations concerning the mobility of individual molecules alone, one would expect line narrowing with increasing temperature. However, lines may, in fact, broaden in cases in which a temperature increase causes a transition from a slowexchange to an intermediate-exchange regime, notably for “encounter-controlled relaxation”5 or “exchange-limited relaxation”.17 In addition, an increase in the temperature may alter the structure and size of the lipid aggregations, and these variations9 may specifically affect the PL headgroup mobility18 and, consequently, induce changes of 31P NMR line widths. RESULTS AND DISCUSSION Global trends for 31P NMR Line Width of Phospholipids. In 31P NMR spectra of brain extracts, most PL resonances partially overlap, are too weak to permit precise LW measure(16) London, E.; Feigenson, G. W. J. Lipid Res. 1979, 20, 408–412.

ments, or both. However, the PtdC signal is particularly strong and does not overlap with major 31P NMR peaks. For this reason, we used PtdC to quantitatively determine the effects of experimental conditions on the line width of PL 31P signals from diesterified phosphate moieties. PtdC line widths increased with temperature for different extract and CDTA concentrations and for different pH values (Figure 1). Line widths were largest (8.48 Hz) for a combination of high extract concentration (943 mg/mL), low CDTA concentration (50 mM), and high sample temperature (297 K) at intermediate pH (7.63). By contrast, the narrowest lines (0.79 Hz) were obtained for a combination of low extract concentration (118 mg/mL), high CDTA concentration (1000 mM), and low sample temperature (277 K), again at intermediate pH (7.36). The temperature dependence increased with increasing extract concentration, in (17) Metz, K. R.; Dunphy, L. K. J. Lipid Res. 1996, 37, 2251–2265. (18) Tolman, C. J.; Kanodia, S.; Roberts, M. F. J. Biol. Chem. 1987, 262, 11088– 11096.

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

5443

Figure 2. Apparent activation energy derived from the temperature dependence of the PtdC line width in phospholipid 31P NMR spectra for different brain extract formulations. Negative Ea values indicate the presence of exchange-limited relaxation (transition from a slow-exchange to an intermediate-exchange regime). Ea increases with extract concentration in a nonlinear fashion. Logarithmic curves fitted to experimental data clearly showed a very good correlation for 1000 and 200 mM CDTA, with somewhat lower r values for the low-pH 200 mM CDTA sample (see Results and Discussion).

Figure 3. 31P NMR line widths of PtdC peaks from rat brain extracts, as a function of extract concentration (milligrams brain tissue per milliliter extract solvent used for NMR measurement). The three diagrams on the left represent samples at pH 7.4; the two diagrams on the right represent samples at pH 8.0. Data are presented for three (two) different CDTA concentrations at pH 7.4 (8.0). Data points are color-coded for different measurement temperatures, as indicated in the insets.

particular where CDTA concentrations were relatively high (200 or 1000 mM). However, for these two CDTA levels, PtdC line widths were almost independent of temperature at the lowest extract concentration (118 mg/mL) (Figure 1). Influence of Temperature on 31P NMR Line Width. For all pH values and CDTA concentrations studied, PtdC line widths for samples of different extract concentrations converged to low values as the sample temperature decreased (Figure 1). Thus, at 277 K, the line widths covered narrow ranges between 1.35 and 2.60 Hz for the highest, and between 0.79 and 1.6 Hz for the lowest extract concentration employed. By contrast, at 297 K, line widths ranged more widely: between 3.40 and 8.48 Hz for the highest and between 1.05 and 4.4 Hz for the lowest extract concentration employed (Figure 1). As a result, the PtdC resonance is generally narrowest when brain extracts are measured at low temperature. Assuming a near-linear temperature dependence of PtdC line widths (linear correlation coefficients r varied between 0.97 and 1.00; data not shown), the slope was at a maximum (0.29 Hz/K) for the combination of highest extract and lowest CDTA concentrations (pH 7.63). The reverse also holds true: the temperature dependence was at a minimum (0.01 Hz/K) for the combination 5444

Analytical Chemistry, Vol. 82, No. 13, July 1, 2010

of lowest extract and highest CDTA concentrations (pH 7.36). These trends are also reflected by the apparent activation energies, Ea, derived from the temperature-dependence of the LWs (Figure 2). At 277 K, extract and CDTA concentrations as well as pH have considerably less influence on the LW than they have at higher temperatures. Figures 3 and S1 and S2 of the Supporting Information illustrate these points. For Figures 3 and S1, samples with similar pH values have been pooled so as to result in two groups (label “pH 8.0”: 7.96 ± 0.11; label “pH 7.4”: 7.37 ± 0.20; means ± s.d.). Influence of Extract Concentration on 31P NMR Line Width. A linear relationship (r ) 1.00) between PtdC line width and extract concentration was determined for 1000 mM CDTA at pH 7.4 (r ) 1.00), and a near to linear relationship (r ) 0.99) for 200 mM CDTA at pH 7.4 and 8.0 (Figure 3). Slopes ranged from 0.67 to 1.24 Hz/(g/mL) for 1000 mM CDTA (pH 7.4) and from 0.55 to 3.31 Hz/(g/mL) for 200 mM CDTA (pH 8.0), with maximum slopes obtained for the highest temperature (297 K), For 50 mM CDTA and pH 8.0, a sigmoid curve was obtained, characterized by upper plateaus starting at about 500 mg/mL

extract concentration (Figure 3). Maximum LWs were markedly higher for pH 7.4 (>8 Hz) than for pH 8.0 (∼6 Hz). Furthermore, the pattern of the initial LW increase with extract concentration observed for 50 mM CDTA at both pH values suggested transition from a lower plateau. Finally, Figure 3 demonstrates that LWs for different temperatures tended to converge to low values at low extract concentrations, a trend that increased with increasing CDTA concentration, with a LW of ∼1 Hz for the highest CDTA concentration (1000 mM). Moreover, Ea increased with extract concentration in a nonlinear fashion (Figure 2). For high CDTA concentrations, excellent results were obtained for logarithmic fits (r ) 1.000 for 1000 mM CDTA at pH 7.36; r ) 0.994 and 0.969 for 200 mM CDTA at pH 7.88 and 7.33, respectively). For 50 mM CDTA, the range of Ea variation was much smaller; fit results were poorer than for higher CDTA concentrations and were strongly pHdependent. Influence of CDTA Concentrations on 31P NMR Line Width. Overall, boosting the CDTA concentration from 50 to 200 mM caused a major drop in the PtdC line width (by up to >3 Hz), but a further increase to 1000 mM had much less effect (decrease consistently by