Quantitative chemical exchange saturation transfer with hyperpolarized nuclei (qHyperCEST): Sensing xenon-host exchange dynamics and binding affinities by NMR M. Kunth, C. Witte, and L. Schröder
Citation: J. Chem. Phys. 141, 194202 (2014); doi: 10.1063/1.4901429 View online: http://dx.doi.org/10.1063/1.4901429 View Table of Contents: http://aip.scitation.org/toc/jcp/141/19 Published by the American Institute of Physics
THE JOURNAL OF CHEMICAL PHYSICS 141, 194202 (2014)
Quantitative chemical exchange saturation transfer with hyperpolarized nuclei (qHyper-CEST): Sensing xenon-host exchange dynamics and binding affinities by NMR M. Kunth,a) C. Witte, and L. Schröderb) ERC Project BiosensorImaging, Leibniz-Institut für Molekulare Pharmakologie (FMP), 13125 Berlin, Germany
(Received 23 July 2014; accepted 30 October 2014; published online 21 November 2014) The reversible binding of xenon to host molecules has found numerous applications in nuclear magnetic resonance studies. Quantitative characterization of the Xe exchange dynamics is important to understand and optimize the physico-chemical behavior of such Xe hosts, but is often challenging to achieve at low host concentrations. We have investigated a sensitive quantification technique based on chemical exchange saturation transfer with hyperpolarized nuclei, qHyper-CEST. Using simulated signals we demonstrated that qHyper-CEST yielded accurate and precise results and was robust in the presence of large amounts of noise (10%). This is of particular importance for samples with completely unknown exchange rates. Using these findings we experimentally determined the following exchange parameters for the Xe host cryptophane-A monoacid in dimethyl sulfoxide in one type of experiment: the ratio of bound and free Xe, the Xe exchange rate, the resonance frequencies of free and bound Xe, the Xe host occupancy, and the Xe binding constant. Taken together, qHyper-CEST facilitates sensitive quantification of the Xe exchange dynamics and binding to hydrophobic cavities and has the potential to analyze many different host systems or binding sites. This makes qHyper-CEST an indispensable tool for the efficient design of highly specific biosensors. © 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4901429] I. INTRODUCTION
The nuclear magnetic resonance (NMR) signal of the noble gas isotope 129 Xe is extremely sensitive to its molecular environment, due to the high polarizability of its large electron cloud, resulting in remarkably large chemical shifts.1 Xe has thus been used as an atomic probe for different molecular environments.2, 3 In X-ray crystallography, Xe has found applications for defining and characterizing hydrophobic sites in different proteins. For instance, Xe has served as a gas probe in protein crystals for identification of putative diffusion channels of O2 .4, 5 In general, such mechanisms include aspects of exchange dynamics which, though inaccessible to x-ray crystallography, are accessible through solution NMR. Hence, NMR based techniques can go beyond the pure mapping of such binding sites. Similar related applications are valuable for the characterization of molecular hosts (such as cryptophane-A monoacid (CrAma )) that have been used for the design of smart Xe NMR biosensors6–10 for which dissolved Xe is used. Beyond its use as a biosensor for molecular imaging, Xe has also been used to sense model biomembrane fluidity.11 However, when working with Xe in solution, identification of resonances from bound Xe using conventional NMR spectroscopy can be impaired by the limited solubility of Xe
a) Electronic mail: [email protected]. Telephone: +49 30 947 93 279. b) Author to whom correspondence should be addressed. Electronic mail:
[email protected]. Telephone: +49 30 947 93 121.
0021-9606/2014/141(19)/194202/9/$30.00
and by the consequently low total NMR signal intensity. This is particularly true for exchanging spins systems which occur when Xe interacts with host molecules. In such systems deriving the Xe binding and exchange constants is of particular interest but has only been performed at relative high host concentrations12–14 and as such has limited use for molecular imaging experiments and the investigation of hosts with low binding constants. An excellent technique to overcome this limitation is the chemical exchange saturation transfer (CEST, originally proposed for protons15 ) approach. In CEST NMR, an abundant pool (free Xe in solution, pool A, see Fig. 1(a)) is used to detect a dilute pool (bound Xe, pool B). This is achieved by initially depleting the spin polarization of pool B (i.e., the magnetization is “saturated”) for a certain duration by a (selective) radiofrequency saturation pulse. This induced depolarization is transferred to pool A via chemical exchange followed by detection of the large pool A for which the spin polarization is cumulatively decreased. A reference measurement without the radiofrequency pulse reveals the intensity of the saturation transfer. Critically, the selective manipulation of pool B prior to detection of pool A is easy to achieve with Xe due to its large chemical shift range. This results in an effective amplification of the signal of Xe bound to the host. Moreover, in combination with spin hyperpolarized Xe (produced by spin exchange optical pumping16 ), a significantly increased dynamic range between the initial condition and the depolarized final state is achieved. This technique is called Hyper-CEST17 and allows for the detection of extremely low concentrations of bound Xe.18–21
141, 194202-1
© 2014 AIP Publishing LLC
194202-2
(a)
Xe kAB
0
(b)
Mth
-20 20 CESTeffect
-40 40
offresonant
intensity
kBA
Xe
0
J. Chem. Phys. 141, 194202 (2014) δB on-resonant
δA
M(ωSat)/M0
1
Kunth, Witte, and Schröder
-60 60
host / binding site
-80 80 -100 100 -120 120 -140 140 -160 160 -180 180 saturation frequency /ppm (c)
off-resonant
η.Mth
onresonant
CESTeffect off-resonant on-resonant
0
tSat
time / s
0
Mth
tSat
time / s
FIG. 1. (a) z-Spectrum of free Xe (at δ A ) and Xe bound to a host molecule or binding site (orange curve; signal at δ B well shifted from that of free Xe (blue line)). The Xe exchange rate from pool B to A is kBA and vice versa. Comparison of CEST signal build-up (b) with thermally polarized protons (1 H-CEST) and (c) hyperpolarized Xe (Hyper-CEST). After saturation, in 1 H-CEST, the system relaxes back to (higher) thermal magnetization Mth while in HyperCEST, the system starts from a hyperpolarized system (enhancement factor η ∼ 104 , M0 = η · Mth ) and relaxes to (negligible) thermal magnetization Mth .
The CEST effect can be modeled by the BlochMcConnell (BM) equations22, 23 which describe the time evolution of the detected macroscopic magnetizations given by the spin polarization of A and B under the effect of a saturation pulse. The parameters included therein allow for quantification of the exchange dynamics at low host concentrations. These equations have been previously used to study Xe host-guest interactions but without considering the intrinsic decay of the hyperpolarization.21, 24 Such evaluation is incomplete especially when utilizing long CEST durations to detect low host concentrations or to achieve high spectral resolution with weak/narrow pulses. A more comprehensive way to perform quantitative Hyper-CEST analysis (qHyperCEST) is therefore still lacking. Whereas the BM equations are solved numerically (for which various approaches have been discussed25–27 ), a simplified analytical solution has been derived from the BM equations specifically for hyperpolarized nuclei, the so-called full Hyper-CEST (FHC) solution.28 A further approximation of the FHC solution is an exponential decay with a Lorentzian lineshaped rate,28 which has already been used to qualitatively describe experimental Hyper-CEST data.29 Further experimental validation of the FHC model and its application to quantitative studies is therefore necessary. Our work aims to present a qHyper-CEST approach which delivers comprehensive information about the Xe-host exchange dynamics and binding properties. We begin by introducing the necessary mathematical tools and the general procedure of qHyper-CEST, followed by briefly describing the BM equations and the FHC solution. We further present an approximation to the FHC solution. This analytical expression gives uncomplicated insight into the complex HyperCEST dynamics which becomes important in planning the appropriate set of experiments to perform. We then use simulated data to explore the sensitivity of our qHyper-CEST approach to noise and choice of saturation parameters. These
results are used to guide our experiments and as an example to compare the fitting performance of the BM equations and the FHC solution. As a result we demonstrate the sensitivity of this method for quantifying the Xe exchange dynamics for cryptophane-A monoacid (CrAma ) in dimethyl sulfoxide (DMSO), in particular the ratio of bound and free Xe, fB , the Xe exchange rate, kBA . In addition to this, the resonance frequencies of free and bound Xe, δ A and δ B , the longitudinal A can also be and transverse relaxation times of free Xe, T1,2 determined. Finally, we determine the previously unquantified binding constant, KA , of Xe to CrA and the host occupancy, β, as an exemplary Xe host-guest system in this solvent. II. MATHEMATICAL TOOLS A. General qHyper-CEST approach
Accessing Xe-host exchange dynamic and binding parameters through qHyper-CEST requires acquisition of zspectra (see Fig. 1(a)). Each data point within a z-spectrum is collected by saturating at a specific frequency, followed by detection of the resulting z-component of the magnetization of the large pool A. To cover a whole chemical shift range this is repeated multiple times by iterative variation of the specific frequency (saturation frequency in ppm). All detected magnetizations M(ωsat ) are referenced to the initially available magnetization M0 . The characteristic shape of the z-spectrum is influenced by all of the above mentioned exchange paramA . eters: fB = [Xe@host]/[Xe in solution], kBA , δ A,B , and T1,2 For the experimental conditions used here, the ratio of bound and free Xe, fB , was
Citation: J. Chem. Phys. 141, 194202 (2014); doi: 10.1063/1.4901429 View online: http://dx.doi.org/10.1063/1.4901429 View Table of Contents: http://aip.scitation.org/toc/jcp/141/19 Published by the American Institute of Physics
THE JOURNAL OF CHEMICAL PHYSICS 141, 194202 (2014)
Quantitative chemical exchange saturation transfer with hyperpolarized nuclei (qHyper-CEST): Sensing xenon-host exchange dynamics and binding affinities by NMR M. Kunth,a) C. Witte, and L. Schröderb) ERC Project BiosensorImaging, Leibniz-Institut für Molekulare Pharmakologie (FMP), 13125 Berlin, Germany
(Received 23 July 2014; accepted 30 October 2014; published online 21 November 2014) The reversible binding of xenon to host molecules has found numerous applications in nuclear magnetic resonance studies. Quantitative characterization of the Xe exchange dynamics is important to understand and optimize the physico-chemical behavior of such Xe hosts, but is often challenging to achieve at low host concentrations. We have investigated a sensitive quantification technique based on chemical exchange saturation transfer with hyperpolarized nuclei, qHyper-CEST. Using simulated signals we demonstrated that qHyper-CEST yielded accurate and precise results and was robust in the presence of large amounts of noise (10%). This is of particular importance for samples with completely unknown exchange rates. Using these findings we experimentally determined the following exchange parameters for the Xe host cryptophane-A monoacid in dimethyl sulfoxide in one type of experiment: the ratio of bound and free Xe, the Xe exchange rate, the resonance frequencies of free and bound Xe, the Xe host occupancy, and the Xe binding constant. Taken together, qHyper-CEST facilitates sensitive quantification of the Xe exchange dynamics and binding to hydrophobic cavities and has the potential to analyze many different host systems or binding sites. This makes qHyper-CEST an indispensable tool for the efficient design of highly specific biosensors. © 2014 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4901429] I. INTRODUCTION
The nuclear magnetic resonance (NMR) signal of the noble gas isotope 129 Xe is extremely sensitive to its molecular environment, due to the high polarizability of its large electron cloud, resulting in remarkably large chemical shifts.1 Xe has thus been used as an atomic probe for different molecular environments.2, 3 In X-ray crystallography, Xe has found applications for defining and characterizing hydrophobic sites in different proteins. For instance, Xe has served as a gas probe in protein crystals for identification of putative diffusion channels of O2 .4, 5 In general, such mechanisms include aspects of exchange dynamics which, though inaccessible to x-ray crystallography, are accessible through solution NMR. Hence, NMR based techniques can go beyond the pure mapping of such binding sites. Similar related applications are valuable for the characterization of molecular hosts (such as cryptophane-A monoacid (CrAma )) that have been used for the design of smart Xe NMR biosensors6–10 for which dissolved Xe is used. Beyond its use as a biosensor for molecular imaging, Xe has also been used to sense model biomembrane fluidity.11 However, when working with Xe in solution, identification of resonances from bound Xe using conventional NMR spectroscopy can be impaired by the limited solubility of Xe
a) Electronic mail: [email protected]. Telephone: +49 30 947 93 279. b) Author to whom correspondence should be addressed. Electronic mail:
[email protected]. Telephone: +49 30 947 93 121.
0021-9606/2014/141(19)/194202/9/$30.00
and by the consequently low total NMR signal intensity. This is particularly true for exchanging spins systems which occur when Xe interacts with host molecules. In such systems deriving the Xe binding and exchange constants is of particular interest but has only been performed at relative high host concentrations12–14 and as such has limited use for molecular imaging experiments and the investigation of hosts with low binding constants. An excellent technique to overcome this limitation is the chemical exchange saturation transfer (CEST, originally proposed for protons15 ) approach. In CEST NMR, an abundant pool (free Xe in solution, pool A, see Fig. 1(a)) is used to detect a dilute pool (bound Xe, pool B). This is achieved by initially depleting the spin polarization of pool B (i.e., the magnetization is “saturated”) for a certain duration by a (selective) radiofrequency saturation pulse. This induced depolarization is transferred to pool A via chemical exchange followed by detection of the large pool A for which the spin polarization is cumulatively decreased. A reference measurement without the radiofrequency pulse reveals the intensity of the saturation transfer. Critically, the selective manipulation of pool B prior to detection of pool A is easy to achieve with Xe due to its large chemical shift range. This results in an effective amplification of the signal of Xe bound to the host. Moreover, in combination with spin hyperpolarized Xe (produced by spin exchange optical pumping16 ), a significantly increased dynamic range between the initial condition and the depolarized final state is achieved. This technique is called Hyper-CEST17 and allows for the detection of extremely low concentrations of bound Xe.18–21
141, 194202-1
© 2014 AIP Publishing LLC
194202-2
(a)
Xe kAB
0
(b)
Mth
-20 20 CESTeffect
-40 40
offresonant
intensity
kBA
Xe
0
J. Chem. Phys. 141, 194202 (2014) δB on-resonant
δA
M(ωSat)/M0
1
Kunth, Witte, and Schröder
-60 60
host / binding site
-80 80 -100 100 -120 120 -140 140 -160 160 -180 180 saturation frequency /ppm (c)
off-resonant
η.Mth
onresonant
CESTeffect off-resonant on-resonant
0
tSat
time / s
0
Mth
tSat
time / s
FIG. 1. (a) z-Spectrum of free Xe (at δ A ) and Xe bound to a host molecule or binding site (orange curve; signal at δ B well shifted from that of free Xe (blue line)). The Xe exchange rate from pool B to A is kBA and vice versa. Comparison of CEST signal build-up (b) with thermally polarized protons (1 H-CEST) and (c) hyperpolarized Xe (Hyper-CEST). After saturation, in 1 H-CEST, the system relaxes back to (higher) thermal magnetization Mth while in HyperCEST, the system starts from a hyperpolarized system (enhancement factor η ∼ 104 , M0 = η · Mth ) and relaxes to (negligible) thermal magnetization Mth .
The CEST effect can be modeled by the BlochMcConnell (BM) equations22, 23 which describe the time evolution of the detected macroscopic magnetizations given by the spin polarization of A and B under the effect of a saturation pulse. The parameters included therein allow for quantification of the exchange dynamics at low host concentrations. These equations have been previously used to study Xe host-guest interactions but without considering the intrinsic decay of the hyperpolarization.21, 24 Such evaluation is incomplete especially when utilizing long CEST durations to detect low host concentrations or to achieve high spectral resolution with weak/narrow pulses. A more comprehensive way to perform quantitative Hyper-CEST analysis (qHyperCEST) is therefore still lacking. Whereas the BM equations are solved numerically (for which various approaches have been discussed25–27 ), a simplified analytical solution has been derived from the BM equations specifically for hyperpolarized nuclei, the so-called full Hyper-CEST (FHC) solution.28 A further approximation of the FHC solution is an exponential decay with a Lorentzian lineshaped rate,28 which has already been used to qualitatively describe experimental Hyper-CEST data.29 Further experimental validation of the FHC model and its application to quantitative studies is therefore necessary. Our work aims to present a qHyper-CEST approach which delivers comprehensive information about the Xe-host exchange dynamics and binding properties. We begin by introducing the necessary mathematical tools and the general procedure of qHyper-CEST, followed by briefly describing the BM equations and the FHC solution. We further present an approximation to the FHC solution. This analytical expression gives uncomplicated insight into the complex HyperCEST dynamics which becomes important in planning the appropriate set of experiments to perform. We then use simulated data to explore the sensitivity of our qHyper-CEST approach to noise and choice of saturation parameters. These
results are used to guide our experiments and as an example to compare the fitting performance of the BM equations and the FHC solution. As a result we demonstrate the sensitivity of this method for quantifying the Xe exchange dynamics for cryptophane-A monoacid (CrAma ) in dimethyl sulfoxide (DMSO), in particular the ratio of bound and free Xe, fB , the Xe exchange rate, kBA . In addition to this, the resonance frequencies of free and bound Xe, δ A and δ B , the longitudinal A can also be and transverse relaxation times of free Xe, T1,2 determined. Finally, we determine the previously unquantified binding constant, KA , of Xe to CrA and the host occupancy, β, as an exemplary Xe host-guest system in this solvent. II. MATHEMATICAL TOOLS A. General qHyper-CEST approach
Accessing Xe-host exchange dynamic and binding parameters through qHyper-CEST requires acquisition of zspectra (see Fig. 1(a)). Each data point within a z-spectrum is collected by saturating at a specific frequency, followed by detection of the resulting z-component of the magnetization of the large pool A. To cover a whole chemical shift range this is repeated multiple times by iterative variation of the specific frequency (saturation frequency in ppm). All detected magnetizations M(ωsat ) are referenced to the initially available magnetization M0 . The characteristic shape of the z-spectrum is influenced by all of the above mentioned exchange paramA . eters: fB = [Xe@host]/[Xe in solution], kBA , δ A,B , and T1,2 For the experimental conditions used here, the ratio of bound and free Xe, fB , was
