Accepted Manuscript Characterization of lithium coordination sites with magic-angle spinning NMR A. Haimovich, A. Goldbourt PII: DOI: Reference:

S1090-7807(15)00030-0 http://dx.doi.org/10.1016/j.jmr.2015.02.003 YJMRE 5603

To appear in:

Journal of Magnetic Resonance

Received Date: Revised Date:

18 September 2014 28 January 2015

Please cite this article as: A. Haimovich, A. Goldbourt, Characterization of lithium coordination sites with magicangle spinning NMR, Journal of Magnetic Resonance (2015), doi: http://dx.doi.org/10.1016/j.jmr.2015.02.003

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Characterization of lithium coordination sites with magic-angle spinning NMR A. Haimovich1 and A. Goldbourt1* 1

School of Chemistry, Raymond and Beverly Sackler Faculty of Exact Sciences, Tel Aviv University, Ramat Aviv 69978, Tel Aviv, Israel *corresponding author: email: [email protected]; Phone: 972-3-6408437; Fax 972-3-6409293;

ABSTRACT Lithium, in the form of lithium carbonate, is one of the most common drugs for bipolar disorder. Lithium is also considered to have an effect on many other cellular processes hence it possesses additional therapeutic as well as side effects. In order to quantitatively characterize the binding mode of lithium, it is required to identify the interacting species and measure their distances from the metal center. Here we use magic-angle spinning (MAS) solid-state NMR to study the binding site of lithium in complex with glycine and water (LiGlyW). Such a compound is a good enzyme mimetic since lithium is four-coordinated to one water molecule and three carboxylic groups. Distance measurements to carbons are performed using a 2D transferred echo double resonance (TEDOR) MAS solid-state NMR experiment, and water binding is probed by heteronuclear high-resolution proton-lithium and proton-carbon correlation (wPMLG-HETCOR) experiments. Both HETCOR experiments separate the main complex from impurities and non-specifically bound lithium species, demonstrating the sensitivity of the method to probe the species in the binding site. Optimizations of the TEDOR pulse scheme in the case of a quadrupolar nucleus with a small quadrupole coupling constant show that it is most efficient when pulses are positioned on the spin-1/2 (carbon-13) nucleus. Since the intensity of the TEDOR signal is not normalized, careful data analysis that considers both intensity and dipolar oscillations has to be performed. Nevertheless we show that accurate distances can be extracted for both carbons of the bound glycine and that these distances are consistent with the X-ray data and with lithium in a tetrahedral environment. The lithium environment in the complex is very similar to the binding site in inositol monophosphatase, an enzyme associated with bipolar disorder and the putative target for lithium therapy. A 2D TEDOR experiment applied to the bacterial SuhB gene product of this enzyme was designed to probe direct correlations between lithium, the enzyme inhibitor, and the closest carboxyl carbons of the binding site. At this point, the chemical shift of the bound carboxyl groups in this 29kDa enzyme could be determined. Keywords: magic angle spinning; solid-state NMR; lithium; TEDOR; HETCOR; metal coordination site; PMLG

1. Introduction Lithium has been a prominent treatment for bipolar disorder since the discovery of its therapeutic action towards mania[1]. Lithium has also been suggested to have implications in regards to Alzheimer's disease, Parkinson's and Huntington's diseases and other neurodegenerative diseases[2]. For those reasons it is of importance to understand the binding mode of lithium to its biological targets, which are associated with these conditions.

One such target is the enzyme inositol monophosphatase

(IMPase), for which over-activity was linked with bipolar disorder [3–8]. In order to understand the binding mode of lithium we acquire structural information about lithium and its surrounding environment using magic-angle spinning solid state NMR (MAS ssNMR). Lithium NMR has been extensively used for the study of batteries and super-capacitors relying on the large knight shift in such materials[9]. Lithium ssNMR has also been used for the study of polyelectrolytes[10,11]. There is a variety of ssNMR techniques to acquire structural information, most of which rely on recoupling of the heteronuclear dipolar interaction. Heteronuclear correlations (HETCOR) can be obtained using polarization transfer techniques such as double cross-polarization

(DCP)[12,13]

and

proton-aided

polarization

transfer[14,15]

(XHH[X/Y]); however, these methods are not quantitative, in particular when quadrupolar spins are involved, and in the latter protons are not always available for efficient transfer. Inter nuclear distances to quadrupolar spins can be accurately and efficiently measured by REDOR (Rotational Echo Double Resonance)[16] and its variant[17]. Yet such experiments lack the advantage of spectral distinction between different chemical sites of one of the coupled species. Additional techniques such as TEDOR[18]

(Transferred-Echo

Double

Resonance)

and

HS/MQC[19–21]

(Heteronuclear Single/Multiple Quantum Coherence) recouple the dipolar interaction and can be implemented as 1D or spectrally-resolved two-dimensional (2D) experiments. Therefore, they have the potential to provide quantitative distance information while maintaining site resolution. Very few examples of quantitative, siteresolved, distance measurements to quadrupolar spins have been demonstrated, in particular when the interaction is of small to medium strength (10-200 kHz). In a recent study, the dipolar HMQC technique was used for measuring the shortest heteronuclear 27Al(spin-5/2)-31P distances in the aluminophosphate VPI-5[22].

TEDOR has been proposed by Schaefer and co.[18] in order to deal with the complication in interpreting REDOR experiments when signal contributions of a natural abundance detected spin background exist since only the dipolar-coupled spins are selected by a coherence transfer process between the two nuclei. In TEDOR, an excited I-spin is evolved by recoupling its dipolar interaction with a coupled S-spin. Recoupling is obtained by the application of rotor synchronized π pulses. The coherent magnetization is transferred to the S-spin via a pair of π/2 pulses applied to both spins and finally the transferred coherence is turned into observable S magnetization using rotor synchronized π pulses. Repeating the experiment for different numbers of applied π pulses creates the dipolar-modulated signal pattern. One dimensional and multidimensional TEDOR experiments have been implemented on quadrupolar nuclei with large quadrupole moments, such as 27Al & 11B, sometimes incorporating MQMAS (for high-resolution detection of the quadrupolar spin) and z-filter techniques[23–26]. These experiments have been mainly utilized to determine connectivities; however, no quantitative analysis has been performed and to date an analytic TEDOR solution is available only for a spin 1/2 pair[27]. The acquisition of high-resolution 1H solid-state NMR spectra has been proposed initially by Lee and Goldburg[28] and evolved to more complex and efficient phase and amplitude modulated schemes[29–31]. Vega and co-workers have proposed to use windowed phase-modulated Lee-Goldburg (wPMLG) experiments in order to obtain high-resolution

1

H spectra at moderate-to-high spinning speeds within a 1D

experiment[32]. wPMLG can also be efficiently applied to the indirect dimension of a 2D HETCOR experiment in order to obtain proton-carbon or other spin correlations. In particular, the high-resolution proton spectrum can reveal water signals[33] and thus it is suitable to study water binding to lithium.

In this work we show that quantitative TEDOR measurements between the small quadrupole moment nucleus 7Li (spin 3/2) and 13C (spin 1/2), combined with wPMLGHETCOR experiments and chemical shift perturbations, provide qualitative and quantitative information about the binding site of lithium in a bio-mimetic lithiumglycine complex containing water in the binding site (LiGlyW), whose crystal structure was reported before[34]. Preliminary data show the potential application to a

lithium-inhibited enzyme, inositol monophosphatase (IMPase), which is one of the therapeutic targets of lithium-carbonate in the treatment of bipolar disorder.

2. Results and Discussion 2.1 Optimization of Li-C TEDOR distance measurements using simulations: In order to determine the most efficient and quantitative TEDOR pulse sequence, different versions of the experiment were designed by employing the pulses on either the carbon channel, the lithium channel or both (Figure 1). As was demonstrated before [35], the nature of the dipolar oscillations (maximal signal and the first zero) varies with the preparation time. In the current simulations, the TEDOR preparation time was set to 8 rotor periods (n = 8), corresponding to a mixing time of nTr=571 μs (MAS of 14 kHz) and for each of six versions of the experiment, a TEDOR curve was simulated as a function of the reconversion time mTr. As can be seen in Figure 2, when finite pulses are considered, pulse sequences 5, 4 and 1 gave the best results, with the dynamics of TEDOR 1 (TED1) largely modified since the pulses are positioned at a different location within the rotor period. When ideal pulses are used, all pulse sequences perform similarly well and better (see '3-I'); however, for quadrupolar spins such a situation is mostly impractical even for nuclei with small quadrupolar coupling constants such as 7Li. TED2, TED3, and TED6, which exhibited reduced performance, employ either single (TED2) or pairs (TED3, TED6) of πpulses on the lithium channel every rotor period, during the reconversion period. That radio-frequency (rf) pulses affect the dipolar oscillations, is not surprising. It was already shown that final pulse effects have impact on the effective dipolar interaction[36,37] in REDOR-type irradiation both in the absence and presence of a quadrupolar spin. In the case of a quadrupolar spin, average Hamiltonian theory calculations predicts that the dipolar frequency is affected by the quadrupolar interaction; its effective value is influence by the length of the pulse and the quadrupolar interaction strength[36]. While the single-crystal oscillations are faster, their integrated sum increases slowly as can be seen for TED3 and TED6. Another significant effect is the generation of high order multiple-quantum terms; such terms effectively contribute to REDOR dephasing; however, in the case of TEDOR, these terms cannot be converted back to observable magnetization hence they significantly reduce the sensitivity of the experiment. As the quadrupolar coupling increases, these

terms become dominant and the efficiency drops below detection levels. In such cases other transfer mechanisms are required, or central-transition selective pulses can be used with the accompanied sensitivity limitations (only part of the Zeeman manifold contributes to the signal). While most sequences compensate for the lithium chemical shift offset by the application of an echo pulse, when using a sequence that lacks chemical shift refocusing, as TED1, even the small shift of 0.5 ppm used for the lithium spin affects the TEDOR dynamics. Since as discussed above and as shown in Fig. 2b, the efficiency of the experiment depends on the actual value of the quadrupolar coupling constant (Cq), in order to quantitatively determine the

7

Li-X inter-nuclear distance, Cq was determined

independently. For LiGlyW, Cq takes a value of 30 kHz (see Experimental Methods Section).

Figure 1. (a) A general scheme of the TEDOR pulse sequence with 13C excitation and 7Li detection. Blocks A and C represent the preparartion section (n rotor periods), blocks B and D represent the reconvertion section (m rotor periods). The vertical black bars indicate π/2 pulses. It is assumed that 13C polarization is excited either by CP or directly. (b) A summary of the different πpulse train schemes of the TEDOR sequences that were simulated. '-': no pulses; '+': πpulse train; 'e' a single echo pulse; in this case this pulse is omitted from the coupled spin. In all cases two πpulses are applied every rotor period and are positioned as indicated in the "Synchronization" entry.

Figure 2. (a) Simulated TEDOR intensities for the different pulse sequences shown in figure 1 plotted as a function of the reconversion time mTr (Tr=71.4 μs). (b) The maximum intensity of a TEDOR curve as a function of the quadrupoler coupling constant Cq. The preparation time (nTr) was set to 571 μs (optimized by simulations of TEDOR 5, not shown here). The reconversion time was varied and the intensity was extracted from the maximum of the curve. All simulations were done employing δ7Li=0.5 ppm and with non-ideal pulses (ν1C=ν1Li=50 kHz) except for 'TEDOR 3 ideal'.

2.2 Li-C distances from 2D TEDOR experiments on LiGlyW: In order to obtain distance information in a site-selective manner, allowing to exclusively monitor Li-C couplings, and following the simulated results, two-dimensional (2D) TEDOR experiments were implemented, in which the

13

C evolution period was inserted directly after the cross-

polarization step[38]. The implementation of TEDOR 5, in which all pulses are applied to the 13

C spin, is shown in Fig. 3.

Figure 3. Pulse sequence for 2D TEDOR 5: 13C magnetization generated by 1H-13C CP is allowed to evolve in t1 and consequently subjected to the TEDOR 5 sequence. Rotor syncronized πpulses (empty bars) are applied to the 13 C channel during both the preparation and reconversion periods with the exception of echo pulses applied to 7Li. Polarization is transffered to 7Li via two simoultaneous π/2 pulses (black bars). The following phase cycle was employed: φଵ  xyxy, ଶ  x , ଷ  y y , ସ  xxx x y y yy , φହ  yxyyxyxx , ଺  yyy y xxx x , ୰ୣୡ  xx x xy yyy . φଵ and φହ were incremented within the pulse sequence. The 2D version of TEDOR 4 had pulses on lithium during the preparation period and a single echo pulse on the carbon channel.

LiGlyW is a layered material, in which lithium is bound to three carboxyl groups and a water molecule, while the alpha-carbon and the amino group are located between the layers[34]. The shortest distance between two lithium atoms in LiGlyW is 3.1 Å and the strong 7Li-7Li homonuclear interaction (~600 Hz) causes spectral broadening and hampers accurate data analysis. Therefore, we performed our experiments on a 6Lienriched sample prepared with only 20% 7Li. An example of two 13C-7Li 2D TEDOR 5 spectra of LiGlyW are presented in Figure 4. The bound lithium signal appears at 0.6 ppm and is correlated with the carbonyl carbon at 176.5 ppm and with Cαat 45.0 ppm. Weak signals corresponding to non-specifically bound lithium species, or to minor impurities appear at a lithium shift of ~0 ppm, carbonyl shifts of 175.0 and 176.2 ppm and Cαshifts of 44.6 and 45.9 ppm. The lithium dimension in those spectra is very similar to the lithium spectrum of SuhB IMPase, also exhibiting unbound and bound lithium species[39].

Figure 4. Two separate 2D 7Li-13C TEDOR 5 of LiGlyW showing (a) carbonyl region; (b) aliphatic region. For (a) the carrier frequency was set at the carbonyl region, the spectrum was acquired with n=8 (571 μs) and m=24 (1714 μs) and the strong signal at δ13C=176.5/δ7Li=0.6 ppm corresponds to the bound species; for (b) the carrier was set to the aliphatic region (45 ppm), the spectrum was acquired with n=8 (571 μs) and m=28 (2 ms) and the bound glycine Cαsignal appears at δ13C=45.0/δ7Li=0.6 ppm. In both experiments, 140 pts were acquired in the indirect dimension with a total experimental time of ~1.25 hrs.

Several such 2D datasets were acquired by employing TEDOR sequences 2, 4 and 5 on a [20%-7Li, N.A-13C] LiGlyW. For each sequence a set of 2D experiments were performed using a constant n and different m values, and every dataset was fit by simulating with different distances. In Figure 5 we show the two-dimensional χ2(I,D) minimization for three data sets (D is the dipolar interaction, I is the scaling value for the simulations results), and from the skyline projections of the minimum χ2 value

along the inter-nuclear distance axis, the best-fit distance was extracted. The best-fit simulations with their corresponding experimental data points are presented in the bottom. We note here that in order to fit a TEDOR curve, data beyond the maximum must be obtained and simulated otherwise the error increases considerably, the reason being that TEDOR curves have the shape of a squared sine function[35] multiplied by an intensity factor that accounts for the lack of a reference experiment; at short times, those functions can be approximated by a linear function and given an arbitrary intensity factor, almost any distance can be safely fitted to the data. On the other hand, at long times, multi-spin effects and eventually relaxation may affect the dipolar oscillations, producing additional errors. Despite the fact that the different experimental conditions resulted in different sensitivity to the distance and intensity parameters, all our results indicate an average Li-CO distance of 2.9±0.15 Å, in agreement with the crystal structure data (shortest distances of 2.8 and 2.9 Å), and a typical distance for lithium in a tetrahedral environment[40]. We further tested the accuracy of other TEDOR versions and other preparation periods that not necessarily utilized the full xy4 cycle. Despite having a reduced intensity, those measurements (Fig. S2) gave distances of 2.9 Å (TEDOR 4, n=8, 857 μs) and 3.0 Å (TEDOR 2, n=8, 857 μs), in agreement with the optimal implementation shown in Fig. 5.

Figure 5. Chi-square analysis of (a) TEDOR 5 with n=6 (429 μs); (b) TEDOR 5 with n=8 (571 μs); (c) TEDOR 4 with n=6. The contour plots in the center represent χ2 for simulations of different dipolar interactions (D) scaled by different intensity (I) factors. The top graphs represent a skyline projection of the minimum χ2 value for any given inter-nuclear distance and the bottom graphs show an overlay of the best-fit simulations and simulations that show a deviation of 0.2 Å from the best fit. For TEDOR 5 the best fit is for (a) a dipolar interaction of 500 Hz (2.9 Å) with an I-factor of 1.8 and (b) D = 500 Hz (2.9 Å) with an I-factor of 1.7. For TEDOR 4 the best fit is also for D = 500 Hz (2.9 Å) with an I-factor of 1.85. A mutual fit of the three experiments done by summing the chi-square values of all fits (shown in Fig. S1) is also consistent with a distance of 2.9 Å.

LiGlyW has an additional methylene group that is located slightly further from the lithium center. We performed additional experiments, positioning the carrier frequency directly in the aliphatic region, in order to measure this distance as well. The results shown in Fig. 6 are consistent with a shortest Li-Cαdistance of 3.1 Å, within agreement with the crystallographic distance of 3.25 Å and consistent with it being slightly more remote than the carbonyl carbons. We further verified these results by two means; we performed a 1D 13C-detected TEDOR experiments (Fig. S3) assuming that the contribution from unbound species is negligible (see intensity differences at δ13C = 45.0 ppm in Fig. 4) and we performed {13C}7Li REDOR experiments (Fig. S4). In both

cases, experiments were performed on a 20%-13Cα/20%-7Li labelled sample. The results were 3.2 Å (TEDOR) and 3.1 Å (REDOR), again consistent with X-ray data.

Figure 6. Chi-square analysis of (a) TEDOR 5 with n=6 (429 μs); (b) TEDOR 5 with n=8 (571 μs). The contour plots in the center represent χ2 for simulations of different dipolar interactions (D) scaled by different intensity (I) factors. The top graphs represent a skyline projection of the minimum χ2 value for any given inter-nuclear distance and the bottom graphs show an overlay of the best-fit simulations and simulations that show a deviation of 0.2 0.3 Å from the best fit. For (a) the best fit is for a dipolar interaction of 400 Hz (3.1 Å) with an I-factor of 2 and for (b) D = 400 Hz (3.1 Å) with an I-factor of 1.7.

2.3 Water-Lithium interactions in the coordination site: While TEDOR results report on the proximity of lithium and the carboxylate groups, the lithium environment contains a single water molecule in LiGlyW, similarly to the coordination proposed for the inhibitory site in IMPase. Water may play a critical role in the mode of action of lithium since it is well established that in the active form Mg2+ species bind between

three and five water molecules. The proximity of water to lithium and to carbons in the binding site can be probed by acquiring 1H-7Li and 1H-13C heteronuclear correlation (HETCOR) experiments provided that a high-resolution spectrum of the protons can be obtained by decoupling the strong 1H-1H homonuclear interactions. In Figure 7a we show the supercycled wPMLG proton spectrum of LiGlyW overlayed with that of α-glycine. The ammonium group in LiGlyW shifts significantly to higher field, probably due to the presence of the counter ion (‫ ) ି ݈ܥ‬between the layers. It also shows an additional high-frequency shoulder, which is a result of impurities in the sample, as will be shown below. The splitting between the methylene lines is smaller than in pure glycine and a new peak at 4.7 ppm can be assigned to the water signal. The 2D wPMLG-HETCOR spectra are shown in Figure 7b. While several species can be observed, the 1H-7Li spectrum clearly indicates that water are only bound to the lithium species at a shift of 0.6 ppm, and to the CH2 signal at 45.0 ppm. The additional two CH2 lines at 44.6 and 45.9 ppm are impurities and species that do not contain water. The carboxyl region of the same spectrum, shown in the Supplementary Material, also shows a bound signal (176.5 ppm) and an additional proton spectrum coupled to the weak carbonyl line at 175.0 ppm (also seen in Figure 4), which lacks the water resonance. The lithium signal at 0 ppm is either an impurity or non-specifically bound lithium. Slices taken at the chemical shift of the main signal of LiGlyW are similar to the 1D spectrum shown in Fig. 7a but lack the shoulder of the amine proton resonance that can be traced to the impurity signals. These spectra demonstrate the sensitivity of the method to separate and identify the main site of interest. While the HETCOR experiment acquired here cannot yet provide accurate distances, it is clear that the detected water are close to the bound lithium, since no water correlation to the non-specifically bound lithium species is observed.

1 7 ௫௫ҧ Figure 7. (a) 1H  ௠௠ ഥ spectra of glycine and of LiGlyW acquired at a spinning speed of 11.5 kHz. (b) H- Li 1 13 7 13 wPMLG-HETCOR and a slice through the bound-lithium site. (c) H- C wPMLG-HETCOR showing Li- Cα correlations and a slice through the bound-lithium site.

2.4 Effect of binding on the chemical shifts: In the proton spectra above, it is shown that the proton chemical shifts of bound glycine are varied upon complexation with lithium and that the ammonium signal shifts to a higher field. It is also evident that the lithium chemical shift is slightly increased upon binding. When the

13

C spectra of glycine and

LiGlyW are compared (Figure 8), an up-field shift of both carbons in LiGlyW (ΔδCO=2 ppm, ΔδCα=0.6 ppm) is also observed relatively to the pure glycine signals (δCO,gly=178.4 ppm, δCα,gly=45.6 ppm) upon binding to lithium. At this stage, we do not attempt to explain the trend and size of those shifts; however, with the advance in DFT calculations, such shifts may in the future be correlated with the structural properties of the lithium coordination sites.

Figure 8. 1D 1H-13C CPMAS experiments of glycine (blue) and LiGlyW (red). Upon binding to lithium, carbon shifts move to lower frequencies.

2.5 Li-CO correlations in the enzyme IMPase observed by 2D TEDOR: We applied a 2D TEDOR 2 experiment to a fully labelled

13

C SuhB IMPase sample prepared with

MgCl2 and LiCl. In the spectrum presented in Figure 9, a cross peak of bound lithium is observed at (0.4, 178.5) ppm and an additional cross peak is observed at (0.0, 176.5) ppm. Simulations suggest that the recoupling time applied here allows for detection of only the closest carbon within the signal-to-noise of the spectrum (~5.5). Consequently, the carbonyl chemical shift of the bound carbonyls at 178.5 ppm separates them from all other carbonyl signals in this 267-residue-long enzyme. The additional lithium signal at 0 ppm has been shown before to belong to non-specifically bound lithium species since its dephasing due to

13

C carbons could not be reproduced

with any defined set of 13C spins, only using a distribution of distances[39]. Many prior studies[41–43] have proposed that in the inhibitory binding site of IMPase Mg2+ ions are ligated by water and by several aspartate carboxylic groups[41]. Relative to aspartate average chemical shifts reported in the BMRB, we can again observe an upfield shift, as in LiGlyW. Also, the lithium chemical shift is at 0.4 ppm, shifted from the free lithium signal at 0 ppm. It remains to be tested by calculations whether such chemical shift changes are correlated to the lithium environment and coordination.

Figure 9. 2D 13C-7Li TEDOR 2 with n = m = 4 (364 μs) applied to a 267-residue-long SuhB bacterial inositol monophosphatase prepared with activating amounts of MgCl2 and inhibitory amounts of LiCl. The carrier frequency was set at 115 ppm for 13C and at 0 ppm for 7Li. A signal of lithium in the IMPase binding site appears at (0.4, 178.5) ppm. 122 pts were acquired in the indirect dimension, with 256 scans taken at each point. The total experimental time was ~43.5 hrs. The spectrum was processed with a 30 Hz exponential broadening in the direct dimension and a Lorentz-to-Gauss transformation in the indirect dimension.

3. Summary & Conclusion In this work we utilized MAS ssNMR for the characterization of lithium coordination in the enzyme mimetic complex LiGlyW, in which lithium is coordinated to three carboxyl groups and a single water molecule. The site configuration was analysed by measuring accurately the distance to the coordinated carboxyl groups and by identifying the bound water molecules. Our results are in good agreement with the Li coordination described in the crystal structure of LiGlyW and the distances are consistent with lithium in a tetrahedral environment. Accurate distances were also measured between lithium and the methylene group suggesting that the experimental approach is valid and accurate. Application of TEDOR to the enzyme inositol monophosphatase, the putative target of lithium therapy in bipolar disorder, enabled us to observe and extract the carbon shifts of the carboxyl groups in the binding site through their correlations to Li-7 since the signals we detect cannot result from a dipolar coupling constant that is weaker than 300-500 Hz (a distance longer than 3.5 Å). In addition, they reconfirm the existence of non-specifically bound lithium species. The TEDOR technique, used to measure the Li-C distances, was optimized towards a quantitative analysis of the dipolar modulation curve. It was shown that the efficiency of the experiment depends on the position of the inversion pulses, which preferentially are applied to the half-spin. However, we have also shown that although positioning

the pulses on the quadrupolar nucleus result in reduced efficiency, the accuracy of the determined distances are not affected. Therefore in cases were the half-spin has a very large CSA it might be desirable to apply the pulses on the quadrupolar spin; if Cq is known, accurate distances can still be obtained. Following the measurements shown here, an error of no more than ±0.15 Å can be safely assumed. Spectroscopically, changes in chemical shifts of the involved nuclei were observed upon binding. For LiGlyW as well as for IMPase, a similar trend was observed for changes in the chemical shifts of lithium and carbon upon binding; the bound lithium was shifted to a higher chemical shift in comparison to free lithium and carbons were shifted to a lower chemical shift relative to native glycine for LiGlyW, and relative to BMRB average values of aspartate for IMPase. The two techniques described here are basic elements in the analysis of metal coordination sites in inorganic and bioinorganic materials, and in enzymes. TEDOR allows for accurate distance measurements while maintaining site selectivity towards the bound species and the proton HETCOR technique allows the detection of coordinating water molecules. Such studies can be corroborated and verified with DFT calculations and additional methods that probe the orientation of the metal will provide a complete picture of its binding environment.

4. Experimental Methods 4.1 Preparation of a lithium-glycine-water complex (LiGlyH(H2O)+Cl-, LiGlyW): The complex was prepared following the procedure described by Müller et al [34]. A 2:1 molar ratio of a mixture of 7LiCl/6LiCl (20/80) and natural abundance (or partially 13Cenriched) glycine was dissolved in purified water and filtered. The solution was heated on a water bath until a white solid material appeared. The solution was then cooled to room temperature, the solid material was filtered out, washed with absolute ethanol, and dried on a Büchner funnel. The compound is termed [20%-7Li/N.A.-13C]-LiGlyW or simply LiGlyW. The three closest Li-C crystal structure distances are (in Å): (i) LiCO 2.8 (amounts to a dipolar coupling constant of 530 Hz), 2.9, 2.9 (480 Hz); (ii) LiCα3.25 (340 Hz), 4.2, 4.2 (158 Hz).

4.2 Preparation of inositol monophosphatase (IMPase): A sample of

13

C fully labeled

wt SuhB bacterial IMPase with LiCl and MgCl2 was prepared as described previously[39].

4.3 Simulations: All TEDOR simulations were performed using SIMPSON[44]. For sequence optimization, simulations of different TEDOR sequences employing lithium detection were performed for an isolated 7Li-13C spin pair using a dipolar coupling constant of 500 Hz, a nuclear quadrupole coupling constant (Cq) of 60 kHz, and a spinning rate of 14 kHz. An offset of 0.5 ppm was applied to lithium and

13

C

was assumed on resonance. For the analysis of LiGlyW data, simulations were performed for different dipolar coupling constants taking into account the experimental offsets of both nuclei and the quadrupolar frequency of lithium (Cq ~ 30 kHz, determined by fitting the sideband pattern obtained from a single-pulse experiment acquired at a spinning speed of 2 kHz). The CSA of the carbonyl was taken to be 70 ppm (as determined by fitting the sideband pattern obtained from a CPMAS experiment acquired at a spinning speed of 2 kHz). 4.4 Data analysis and distance extraction from TEDOR results: The experimental TEDOR data points were normalized to the maximum signal intensity and due to scaling differences between the experimental and the simulated intensities (TEDOR does not have a reference experiment), the simulated data points were multiplied by different intensity factors (I-factor). The minimum chi-square was then calculated on a two-dimensional parameter space as follows: 

,   ,    Δ,  

1

Here Iexp,k is the normalized kth experimental cross peak intensity, ΔIexp,k is the experimental error of the normalized data point calculated from the noise level, and Isim,k is the simulated data point intensity multiplied by the intensity factor. χ2 values were divided by 103.

4.5 Solid-state NMR experiments: All NMR experiments were performed using a Bruker Avance-III NMR spectrometer operating at a magnetic field of ~14.1 T (1H Larmor frequency of 600.0 MHz) using a wide-bore 4 mm probe operating in

1

H−7Li−13C triple-resonance mode, with 7Li and

13

C Larmor frequencies of 233.2 and

150.9 MHz, respectively. 7Li chemical shifts were referenced to a 1 M solution of LiCl in purified H2O (18.2 MΩ) at 0 ppm and 13C shifts were referenced to adamantane CH2 line at 40.48 ppm. For all experiments, carbon and lithium πand π/2 pulses were applied using an rf field of 50 kHz and 60-63 kHz, respectively. Various versions of 2D transferred echo double resonance (TEDOR) experiments, which differ in the position of the pulses, were applied to the enzyme-mimetic compound LiGlyW, using a spinning rate of 14 KHz and a temperature set to -10 °C (with the exception of TEDOR 2, -20 °C). The carrier frequency was set either at the carbonyl or at the aliphatic region for

13

C according to the desired correlation. Proton

decoupling was employed at 80 kHz using the swf-tppm scheme[45]. For every dataset the number of rotor periods prior to magnetization transfer, n, was selected to be a constant (set at either 6 or 8 rotor periods, corresponding to 429 and 571 μs, respectively) and the number of rotor periods after magnetization transfer, m, was varied. For IMPase, a single 2D 7Li -13C TEDOR spectrum was recorded at 2 ˚C with a spinning rate of 11 KHz. The carrier frequency was set at 115 ppm for

13

C and at 0

ppm for 7Li. The parameters n and m were set to 4, corresponding to a mixing time of 364 μs. Supercycled wPMLG[46] experiments were performed on a [20%-7Li/20%-13Cα]LiGlyW sample, using a spinning rate of 11.5 KHz and a temperature set to -20 °C. wPMLG was calibrated using a glycine crystal by selecting for the best resolved CH2 splitting. 1D experiments performed at different offsets were used to calculate the scaling factor, which was 0.49, and to verify its linearity. The scale was then referenced using the known amine proton shifts of glycine (ߜேுయ = 8 ppm). In all 1D and 2D wPMLG-HETCOR experiments, a window of 6 μs was used, a PMLG pulse length of 1.2 μs, and the 10x2 phases followed the supercycling routine[46,47] with the first 10 phases taking the values φ=339.22, 297.65, 256.08, 214.51, 172.94, 352.94, 34.51, 76.08, 117.65, 159.22 and the next 10 taking the reverse order. A complete compilation of experimental and processing parameters appears in the Supplementary Material. Acknowledgements

This study was supported by the Binational Science Foundation (BSF grant 2011077 to AG and Tatyana Polenova) and by the national institute for psychobiology in Israel. The NMR spectrometer was funded in part by the Tel Aviv Center for Nanosciences and Nanotechnology. The plasmid for SuhB IMPase was kindly provided by Prof. Mary Roberts from Boston College. We thank Hadar Ivanir for her help with setting up the wPMLG experiment.

6. References [1]

J. Cade, Lithium salts in the treatment of psychotic excitement., Med. J. Aust. 2 (1949) 349–351.

[2]

W. Young, Review of Lithium Effects on Brain and Blood, Cell Transplant. 18 (2009) 951–975.

[3]

C.J. Phiel, P.S. Klein, Molecular targets of lithium action, Annu. Rev. Pharmacol. Toxicol. 41 (2001) 789–813.

[4]

M.J. Berridge, C.P. Downes, M.R. Hanley, Lithium amplifies agonist-dependent phosphatidylinositol responses in brain and salivary-glands, Biochem. J. 206 (1982) 587–595.

[5]

R.S. Jope, M.B. Williams, Lithium and brain signal-transduction systems, Biochem. Pharmacol. 47 (1994) 429–441.

[6]

M.J. Berridge, R.F. Irvine, Inositol phosphates and cell signaling, Nature. 341 (1989) 197–205.

[7]

L.M. Hallcher, W.R. Sherman, The effects of lithium ion and other agents on the activity of myo-inositol-1-phosphatase from bovine brain, J. Biol. Chem. 255 (1980) 896–901.

[8]

M.J. Berridge, C.P. Downes, M.R. Hanley, Neural and developmental actions of lithium: A unifying hypothesis, Cell. 59 (1989) 411–419.

[9]

F. Blanc, M. Leskes, C.P. Grey, In situ solid-state NMR spectroscopy of electrochemical cells: batteries, supercapacitors, and fuel cells., Acc. Chem. Res. 46 (2013) 1952–1963.

[10]

S. Causemann, M. Schönhoff, H. Eckert, Local environment and distribution of alkali ions in polyelectrolyte complexes studied by solid-state NMR., Phys. Chem. Chem. Phys. 13 (2011) 8967–8976.

[11]

T.K.-J. Köster, L. van Wüllen, Cation–anion coordination, ion mobility and the effect of Al2O3 addition in PEO based polymer electrolytes, Solid State Ionics. 181 (2010) 489–495.

[12]

J. Schaefer, R.. McKay, E.. Stejskal, Double-cross-polarization NMR of solids, J. Magn. Reson. 34 (1979) 443–447.

[13]

M. Baldus, A.T. Petkova, J. Herzfeld, R.G. Griffin, Cross polarization in the tilted frame: assignment and spectral simplification in heteronuclear spin systems, Mol. Phys. 95 (1998) 1197–1207.

[14]

M. Wilhelm, H. Feng, U. Tracht, H.W. Spiess, 2D CP/MAS 13C isotropic chemical shift correlation established by 1H spin diffusion., J. Magn. Reson. 134 (1998) 255– 260.

[15]

M. Etzkorn, A. Böckmann, A. Lange, M. Baldus, Probing molecular interfaces using 2D magic-angle-spinning NMR on protein mixtures with different uniform labeling, J. Am. Chem. Soc. 126 (2004) 14746–14751.

[16]

T. Gullion, J. Schaefer, Rotational-echo double-resonance NMR, J. Magn. Reson. 81 (1989) 196–200.

[17]

A. Goldbourt, Distance Measurements to Metal Ions and Other Quadrupolar Spins by Magic Angle Spinning Solid State NMR, Isr. J. Chem. 54 (2014) 125–135.

[18]

A.W. Hing, S. Vega, J. Schaefer, Transferred-echo double-resonance NMR, J. Magn. Reson. 96 (1992) 205–209.

[19]

O. Lafon, Q. Wang, B. Hu, F. Vasconcelos, J. Trébosc, S. Cristol, et al., Indirect detection via spin-1/2 nuclei in solid state NMR spectroscopy: application to the observation of proximities between protons and quadrupolar nuclei., J. Phys. Chem. A. 113 (2009) 12864–12878.

[20]

Z. Gan, 13C/14N heteronuclear multiple-quantum correlation with rotary resonance and REDOR dipolar recoupling., J. Magn. Reson. 184 (2007) 39–43.

[21]

J. Trebosc, B. Hu, J.P. Amoureux, Z. Gan, Through-space R3-HETCOR experiments between spin-1/2 and half-integer quadrupolar nuclei in solid-state NMR., J. Magn. Reson. 186 (2007) 220–227.

[22]

X. Lu, J. Trébosc, O. Lafon, J.-P. Amoureux, Measurement of the shortest heteronuclear distances in multiple-spin systems using constant-time correlation NMR methods, CrystEngComm. 15 (2013) 8713–8726.

[23]

C.A. Fyfe, K.C. Wong-Moon, Y. Huang, H. Grondey, K.T. Mueller, Dipolar-Based 27Al -> 29Si Solid-State NMR Connectivity Experiments in Zeolite Molecular Sieve Frameworks, J. Phys. Chem. 99 (1995) 8707–8716.

[24]

C.A. Fyfe, K.T. Mueller, H. Grondey, K.C. Wong-Moon, Dipolar dephasing between quadrupolar and spin- nuclei. REDOR and TEDOR NMR experiments on VPI-5, Chem. Phys. Lett. 199 (1992) 198–204.

[25]

D.H. Brouwer, J.-M. Chézeau, C.A. Fyfe, Solid state NMR investigation of the structure of AlPO4-14A, Microporous Mesoporous Mater. 88 (2006) 163–169.

[26]

G. Tricot, J. Trebosc, J.P. Amoureux, Measurement of through-space connectivities between spin-1 / 2 and quadrupolar nuclei in solid-state NMR : the TEDORMQMAS method, Magn. Reson. Chem. (2007) 187–191.

[27]

K.T. Mueller, Analytic solutions for the time evolution of dipolar-dephasing NMR signals, J. Magn. Reson. Ser. A. 113 (1995) 81–93.

[28]

M. Lee, W. Goldburg, Nuclear-Magnetic-Resonance Line Narrowing by a Rotating rf Field, Phys. Rev. 140 (1965) A1261–A1271.

[29]

E. Vinogradov, P.K. Madhu, S. Vega, High-resolution proton solid-state NMR spectroscopy by phase-modulated Lee–Goldburg experiment, Chem. Phys. Lett. 314 (1999) 443–450.

[30]

D. Sakellariou, A. Lesage, P. Hodgkinson, L. Emsley, Homonuclear dipolar decoupling in solid-state NMR using continuous phase modulation, Chem. Phys. Lett. 319 (2000) 253–260.

[31]

P.K. Madhu, High-resolution solid-state NMR spectroscopy of protons with homonuclear dipolar decoupling schemes under magic-angle spinning., Solid State Nucl. Magn. Reson. 35 (2009) 2–11.

[32]

E. Vinogradov, P.K. Madhu, S. Vega, Proton spectroscopy in solid state nuclear magnetic resonance with windowed phase modulated Lee–Goldburg decoupling sequences, Chem. Phys. Lett. 354 (2002) 193–202.

[33]

A. Goldbourt, E. Vinogradov, G. Goobes, S. Vega, High resolution heteronuclear correlation NMR spectroscopy between quadrupolar nuclei and protons in the solid state., J. Magn. Reson. 169 (2004) 342–350.

[34]

G. Müller, G. Maier, M. Lutz, Lithium coordination to amino acids and peptides. Synthesis, spectroscopic characterization and structure determination of lithium complexes of neutral and anionic, Inorganica Chim. Acta. 218 (1994) 121–131.

[35]

A. Hing, S. Vega, J. Schaefer, Measurement of Heteronuclear Dipolar Coupling by Transferred-Echo Double-Resonance NMR, J. Magn. Reson. Ser. A. 103 (1993) 151– 162.

[36]

I. Sack, A. Goldbourt, S. Vega, G. Buntkowsky, Deuterium REDOR: principles and applications for distance measurements, J. Magn. Reson. 138 (1999) 54–65.

[37]

C.P. Jaroniec, B.A. Tounge, C.M. Rienstra, J. Herzfeld, R.G. Griffin, Recoupling of heteronuclear dipolar interactions with rotational-echo double-resonance at high magic-angle spinning frequencies., J. Magn. Reson. 146 (2000) 132–9.

[38]

E.R. van Eck, W.S. Veeman, Spin density description of rotational-echo doubleresonance, transferred-echo double-resonance and two-dimensional transferred-echo double-resonance solid state nuclear magnetic resonance., Solid State Nucl. Magn. Reson. 2 (1993) 307–315.

[39]

A. Haimovich, U. Eliav, A. Goldbourt, Determination of the Lithium Binding Site in Inositol Monophosphatase, the Putative Target for Lithium Therapy, by Magic-AngleSpinning Solid-State NMR, J. Am. Chem. Soc. 134 (2012) 5647–5651.

[40]

U. Olsher, R.M. Izatt, J.S. Bradshaw, N.K. Dalley, Coordination Chemistry of Lithium Ion: A Crystal and Molecular Structure Review, Chem. Rev. 91 (1991) 137–164.

[41]

R. Gill, F. Mohammed, R. Badyal, L. Coates, P. Erskine, D. Thompson, et al., Highresolution structure of myo-inositol monophosphatase, the putative target of lithium therapy, Acta Crystallogr. Sect. D-Biological Crystallogr. 61 (2005) 545–555.

[42]

Y.L. Wang, K.A. Stieglitz, M. Bubunenko, D.L. Court, B. Stec, M.F. Roberts, The structure of the R184A mutant of the inositol monophosphatase encoded by suhB and implications for its functional interactions in Escherichia coli, J. Biol. Chem. 282 (2007) 26989–26996.

[43]

N. Singh, A.C. Halliday, M. Knight, N.A. Lack, E. Lowe, G.C. Churchill, Cloning, expression, purification, crystallization and X-ray analysis of inositol monophosphatase from Mus musculus and Homo sapiens., Acta Crystallogr. Sect. F. Struct. Biol. Cryst. Commun. 68 (2012) 1149–52.

[44]

M. Bak, J.T. Rasmussen, N.C. Nielsen, SIMPSON: a general simulation program for solid-state NMR spectroscopy. 2000., J. Magn. Reson. 213 (2011) 366–400.

[45]

R.S. Thakur, N.D. Kurur, P.K. Madhu, Swept-frequency two-pulse phase modulation for heteronuclear dipolar decoupling in solid-state NMR, Chem. Phys. Lett. 426 (2006) 459–463.

[46]

M. Leskes, P.K. Madhu, S. Vega, Supercycled homonuclear dipolar decoupling in solid-state NMR: toward cleaner 1H spectrum and higher spinning rates., J. Chem. Phys. 128 (2008) 052309.

[47]

M. Leskes, S. Steuernagel, D. Schneider, P.K. Madhu, S. Vega, Homonuclear dipolar decoupling at magic-angle spinning frequencies up to 65kHz in solid-state nuclear magnetic resonance, Chem. Phys. Lett. 466 (2008) 95–99.

Graphical abstract

Highlights • • • •

7Li-13C accurate distances are acquired on a lithium complex using quadrupolaroptimized TEDOR 1H-13C and 1H-7Li wPMLG HETCOR report on water in the lithium coordination site Lithium binding induces chemical shift changes TEDOR applied to the enzyme inositol monophosphatase separates bound and unbound species