Anal Chem Technical Note: Resolving enantiomers of 2-hydroxy acids by NMR

Abstract

Biologically important 2-hydroxy carboxylates such as lactate, malate and 2-hydroxyglutarate exist in two enantiomeric forms, that cannot be distinguished under achiral conditions. The D and L (or R, S) enantiomers have different biological origins and functions, and therefore there is a need for a simple method for resolving, identifying and quantifying these enantiomers. We have adapted and improved a chiral derivatization technique for NMR, which needs no chromatography for enantiomer resolution, with >90% overall recovery. This method was developed for 2-hydroxyglutarate (2HG) to produce diastereomers resolvable by column chromatography. We have applied the method to lactate, malate and 2HG. The limit of quantification was determined to be about 1 nmol for 2HG with coefficients of variation <5%. We also demonstrated the method on an extract of a renal carcinoma bearing an isocitrate dehydrogenase-2 (IDH2) variant that produces copious quantities of 2HG, and showed that it is the D enantiomer that was exclusively produced. We also demonstrated in the same experiment that the lactate produced in the same sample was the L enantiomer.

Graphical Abstract

Most biochemical compounds contain at least one chiral center and the chiral enantiomers interact with proteins differently ¹. Although most protein amino acids are found in the L configuration (the exception being glycine which is achiral), D amino acids such as D-Ala, D-Ser and D-Asp also occur in various organisms ^2,3. Similarly many organic acids are chiral, or in some cases prochiral (e.g. citrate) such that a chiral enzyme (e.g. aconitase) can discriminate the different ends of a prochiral molecule ⁴. There are several biologically important 2-hydroxy acids that exist as enantiomeric pairs, which are produced by the reduction of the corresponding 2-oxo acid, such as lactate (from pyruvate), malate (from oxalacetate) and 2-hydroxyglutarate (from 2-oxoglutarate). Although both the D and L forms of 2-hydroxyglutarate (2HG) occur naturally, and are associated with rare acidurias ⁵, the compound is also produced at high concentrations by mutations in the genes IDH1/2, which changes the catalysis of the freely reversible NADP⁺-dependent oxidative decarboxylation of isocitrate to 2-oxoglutarate, to the preferred reduction of 2OG to 2HG. The latter reaction was observed in gliomas and acute myeloid leukemia, and more recently in other cancers ⁶. These specific gain-of-function mutations in the genes that encode IDH1 and 2 produce D-2HG ^7,8, which has been shown to be a potent inhibitor of 2OG-dependent dioxygenases such as the epigenetic TET enzyme and prolyl hydroxylases ^9,10 as 2HG is structurally very similar to 2OG. More recently it has also been shown that even non-variant oxidoreductases can catalyze the reduction of 2OG to 2HG under certain conditions, e.g. both LDH and MDH produce L-2HG ^{11–13 14}, whereas PHGDH produces the D enantiomer ¹⁵.

L-lactate (L-2-hydroxypropanoate) is the enantiomer produced by mammalian LDH-A and LDH-B whereas some bacteria can produce D-lactate, and high blood D-lactate concentration is associated with short bowel syndrome ¹⁶. D-lactate is also produced from glyoxal by the action of glyoxylase ^17,18. Similarly, L-malate (L-2-hydroxybutanoate) is the usual form of malate produced by the Krebs cycle, but D-malate dehydrogenase and a racemase are found in bacteria ¹⁹. Therefore, it is important to be able not only to measure the concentration of such metabolites, but also to delineate the preferred enantiomer in biological systems.

Enantiomers cannot be distinguished under achiral conditions by any physical method. To resolve enantiomers, chiral discrimination must be used e.g. via enzymes specific for one enantiomer, chiral chromatography ²⁰, or converting the enantiomers to diastereomers followed by chromatographic separation ^14,21,22. It is also practical to directly discriminate enantiomers by NMR using a chiral shift reagent or after conversion to diastereomers ¹⁸. Several chiral reagents have been used in chiral chromatography to separate the enantiomers, including heptakis (2,3-di-O-methyl-6-O-tert.butyldimethylsilyl)-beta-cyclodextrin²³, tert-butylcarbamoyl-quinine and −quinidine (Chiralpak QN-AX and QD-AX)²⁴, ristocetin A glycopeptide antibiotic²⁵ and Vancomycin^26,27. Chiral solvating reagents may also help resolve racemic enantiomers by ¹³C NMR spectroscopy with²⁸ or without DNP²⁹. However, DNP methods are typically very expensive to use or the reagents may cause severe line-broadening issues in the NMR spectra.

Alternatively, diastereomers can be generated from enantiomers via a chemical reaction. Various reagents have been reported to react with 2HG to produce such diastereomers, including (D)-2-butanol with acetic anhydride³⁰, (S)-(+)-3-methyl-2-butanol³¹, N-(p-toluenesulfonyl)-L-phenylalanylchloride (TSPC)³², diacetyl-L-tartaric anhydride (DATAN)³³ and Diazomethane + ethyl chloroformate³⁴. All of these studies used chromatography coupled with mass spectrometry to resolve the enantiomers. Here we report a method that couples DATAN derivatization of the OH group of hydroxy acids to produce diastereomers ^14,21 with NMR analysis without chromatographic separation. We used this method to generate NMR-distinguishable diastereoisomers of lactate, malate and 2HG as standards and in biological extracts, based on Scheme 1. It provides a non-destructive and highly reproducible means to analyze enantiomers. We also improved the efficiency and speed of the derivatization reaction by adopting a focused-beam microwave assisted synthesis approach.

Scheme 1.

Reaction scheme for 2HG with DATAN

Experimental Section

Reagents and standards

D,L 2-hydroxyglutarate, D-2-hydroxyglutarate and L-2-hydroxyglutarate, and DATAN ((+) O,O’-diacetyl-L-tartaric anhydride) were purchased from Sigma Aldrich (St. Louis, MO). D-lactic acid, D,L lactic acid, L-malic acid were purchased from Acros (Carlsbad, CA), and D,L malate was obtained from Tokyo Chemical Industries (Portland, OR). All other reagents were of analytical grade or better.

Derivatization with DATAN

As acid anhydrides react readily with water, all reactions must be carried out under dry conditions. Standards or biological extracts were dissolved in 285 μL DI water + 1.2 mL methanol + 15 μL 10 mM lactate (original method) or without lactate and lyophilized to complete dryness in a screw cap Eppendorf centrifuge tube. The DATAN reactions are typically catalyzed using a small amount of lactate. However, in most, if not all, biological systems, lactate is already present. We found that adding lactate is unnecessary.

For derivatization of standards with DATAN (as 50 mg/mL), a molar ratio of 2:3 (2-hydroxy acids: DATAN) was used as in the original reports. As acid anhydrides will react with amino-groups (e.g. in all amino acids) in addition to OH groups (e.g. in Tyr, Ser, Thr, malate, and lactate) in biological extracts, a sufficient excess of the derivatizing agent is needed to ensure complete reaction. Thus, the hydroxyl groups in these biological samples were overestimated as 50 times of the concentration of the dominant lactate.

DATAN solution was freshly prepared by dissolving 50 mg dry DATAN in 1 mL 4:1 HPLC grade acetonitrile (ACN):glacial acetic acid (gHAc) (i.e. 0.8 mL ACN + 0.2 mL gHAc), and 200 μL were immediately added to the lyophilized standards and/or extract, sealed with the screw cap, and heated to 70°C on a preheated dry block. After 2 h of heating, the sample was removed, spun at 15,000 g for 5 minutes, and the supernatant was lyophilized overnight.

D,L 2HG was also dissolved in DMEM medium and derivatized as above. D and L 2-HG were separately derivatized with DATAN to be used as authentic spike-in standards.

Optimization of DATAN derivatization by microwave heating

With media samples, the dry block heating for 2 h gave rise to a brown product, presumably reflecting breakdown. To minimize such breakdown and speed up the overall process, we adopted microwave heating with a focused beam CEM Discover microwave reactor (CEM Corp, Matthews, NC) under a nitrogen atmosphere as previously described ³⁵.

Freshly dissolved DATAN in 4:1 HPLC grade ACN: gHAc was prepared as 50 mg/ml for reaction with standards and 25 mg/mL for reaction with media or tissue extracts. 200 μL of the DATAN solution was added to the dried standards or extracts and transferred to a 500 μL glass insert. The insert was placed inside a 10 mL glass vial and the space between the insert and vial was filled with the same 4:1 ACN/gHAc solution to the same level as those inside the small insert. The large glass vial was sealed by a silicone cap with a needle vent and placed under nitrogen gas flow for 20 min to remove oxygen. The reaction was initiated by heating to 70°C in a microwave-assisted synthesis reactor for 7 min. After DATAN derivatization, the solution was yellow. The solution was transferred into Eppendorf tubes and centrifuged at 20,000 g for 20 min, and the supernatant was lyophilized overnight.

DATAN derivatization and analysis of a Renal Carcinoma Extract

We analyzed an extract of a renal cell carcinoma that harbors an IDH2 mutation, which was previously shown to produce high concentrations of 2HG ⁶. After NMR analysis of the underivatized extract, the sample was lyophilized and then derivatized using the microwave heating as described above, and analyzed by 1D and 2D NMR. After recording spectra, DATAN-derivatized L-2HG was spiked in at approximately equimolar concentrations as the original 2HG and the NMR spectra were re-recorded.

DATAN derivatization of media from non-small cell lung carcinoma tissue slices

Non-small cell lung cancer (NSCLC) patient tissue slices were collected and cultured as reported previously ³⁶. The medium samples after 24 hours of culturing were collected and derivatized. The overall hydroxy acid metabolites were overestimated as 50 times of the concentration of lactate, which was determined by 1D ¹H NMR before derivatization. The corresponding amount of DATAN were used to ensure complete derivatization.

Quantification limits

We determined the NMR response at 16.45 T as a function of 2HG concentration before and after derivatization with DATAN. A series of D-2HG standard samples (~100 nmols, 25 nmols, 5 nmols, 1 nmols and 0.5 nmols) were prepared in four replicates in each group. After lyophilization, three of the replicates were derivatized with 50 μL 25 mg/mL DATAN in the same procedures as described as above. The last one was used as a reference to determine the precise amount of D-2HG before the reaction.

NMR analysis

Lyophilized DATAN derivatives were dissolved in 55 μL (for analysis with 1.7 mm NMR tube) or 120 μL (for analysis with 3 mm Shigemi NMR tube) 100 mM sodium phosphate buffer in D₂O, pD 7, containing 0.5 mM DSS-d6 as chemical shift reference and internal concentration standard.

The reconstituted solution was centrifuged (20,000 g, 20 min) and the cleared supernatant was loaded into an NMR tube. NMR spectra were recorded at 14.1 T on an Agilent DD2 spectrometer equipped with a 3 mm inverse triple resonance HCN cold probe.1D ¹H Presat spectra were recorded with an acquisition time of 2 s and a relaxation delay of 4 s during which the residual HOD resonance was suppressed by a weak rf field. double-quantum filtered COSY (DQFCOSY) spectra were recorded with acquisition times of 1 s in t₂ and 40 ms in t₁ with 8 transients per t₁ increment (experimental time = 2 h). NMR spectra were analyzed using MNOVA software (Mestrelab Research, Santiago de Compostela). 1D spectra were apodized using a 0.5 Hz line broadening exponential and a cosine squared function, followed by phasing, baseline correction using 3-order Bernstein polynomials and referencing to internal DSS. 2D spectra were zerofilled, apodized with a 1 Hz line broadening exponential and a cosine squared function in both dimensions, phased and baseline corrected.

Results and Discussion

(i). Standards

Figure 1A shows the NMR spectra of derivatized D,L 2HG (racemic mixture of 1:1), which clearly shows the presence of two sets of resonances of equal intensity, corresponding to the diastereomers of the D and L enantiomers of 2-HG. None of the original 2HG was detectable, indicating 100% conversion. As expected the chemical shifts of the derivatized forms differed substantially from the original D,L mixture, which comprises a single set of resonances (Table 1 and Fig. S1). To assign the resonances for each diastereomer, we also recorded spectra of the individually derivatized enantiomers (Fig. 1B); the more upfield shifted resonances (ca. 0.07 ppm) corresponded to the L-isomer (Table 1).

Figure 1. NMR spectra of DATAN-derivatized 2HG.

D,L 2HG and D-2HG were derivatized with DATAN as described in the Experimental Section. NMR spectra were recorded at 14.1 T, 20°C.

A. DQCOSY and 1D ¹H NMR spectra of the D,L mixture showing resolution of the enantiomers especially at the 2 and 4 positions (cf Table 1).

B. DQCOSY and 1D ¹H NMR spectra of D-2HG showing a single product corresponding to the D-enantiomer (cf Table 1)

Table 1.

Chemical shifts of DATAN-derivatized 2-hydroxy carboxylic acids

DATAN-derivatized Compound 1	Atom	Shift (D) 2	Shift (L) 2	Underivatized compound 2
2-HG 20°C	C2H	5.14 (DD 3J = 7.8,4.7 Hz)	5.07 (DD 3J = 8.5,4.1 Hz)	3.98 (DD, 3J = 7.65,4.05 Hz
	C3H	2.22/2.28	2.16/2.21	1.96,1.81
	C4H	2.55	2.48	2.23,2.21
2HG in kidney cancer sample 15°C	C2H	5.22 (7.4, 4.6)	5.17 (8.5, 4.2)	3.98
	C3H	2.3/2.22	2.27/2.18	1.82/1.97
	C4H	2.55	2.5	2.24
Lactate 15°C	C2H	5.21 (Q, 3J= 7.14 Hz)	5.17 (Q, 3J= 7.14 Hz)	4.1 (Q)
	C3H	1.55 (D, 3J= 7.14 Hz)	1.52 (D, 3J= 7.14 Hz)	1.32 (D)
Mal 15°C	C2H	5.6 (T, 3J= 6 Hz)	5.59 (T, 3J= 6 Hz)	4.61 (DD, 3J= 6.8 Hz, 4.4 Hz)
	C3H	3.09 (D, 3J = 5.6 Hz)	3.06 (D, 3J= 5.6 Hz)	2.93 (DD, 3J= 4.45 Hz, 2J=16.5 Hz) 2.87 (DD, 3J= 7.1 Hz, 2J=16.5 Hz)

¹Samples were prepared as described in Methods; due to the different salt concentrations and pH after the derivatization, shifts can deviate to a small extent for different samples; a similar result was seen with D,L 2HG spiked into DMEM, which is a more representative matrix for biological samples with many interfering compounds (cf. Figure S2).

²DD-double doublet, T-Triplet, Q-quartet

Method improvement by microwave assisted reaction

We first replicated the derivatization from a previously reported method ²¹. However, we found that under the reported conditions, the reaction was not always complete after 2 hours of conventional heating (Figure S3A). Further, the reaction turned the solution dark brown, indicating some caramelization or side reactions. Indeed, at the 5 hour time point, the reaction was still incomplete and new NMR resonances were observed (Figure S3B). Thus, we adopted a microwave-assisted synthesis method, where a microwave was applied to greatly accelerate the reaction while the reaction temperature/pressure were precisely controlled and monitored. We found that using the same solvent, acetonitrile: acetic acid (4:1, v/v), the reaction was completed in 7 min without the addition of lactate to facilitate the reaction (Figure S3A). The reaction yielded a bright yellow transparent solution without signs of chemical degradation compared with the dark brown color as described in the original report ²¹(Figure S3B).

Quantification of enantiomeric purity, precision, accuracy and limits of quantification

To investigate whether the DATAN derivatization method can be applied for quantitative analysis of enantiomers. The D- and L-2HG mixture with ratios of 2:1 and 1:2 were prepared and aliquoted into four replicates, with one as a reference for the three used for derivatization with DATAN. After reaction, all four replicates in each group were reconstituted into the same volume of 100% D₂O phosphate buffer containing the same amount of d6-DSS. All NMR data were acquired under the exact same conditions. The D- and L-2HG derivatives were quantified by integrating the C2H peaks in 1D ¹H NMR spectra and normalizing to the area of the DSS peak, with correction for the partial saturation of the DSS as previously described ³⁷. As shown in Figure S4, the ratio of D to L remains the same after the reaction, although with some minor loss during the derivatization. The recovery was estimated to be ≈90%. The CoV for the ratio was 3–3.5 % in both cases indicating the consistency of the derivatization method. Furthermore, the accuracy estimated from the pre-derivatized spectra by normalizing to the known concentration of DSS was within 10 %.

To test the detection limits of NMR after derivatization, we prepared a series of different concentrations of the D-2HG standards to react with exactly the same amount of DATAN and quantified by NMR. We considered a signal to noise ratio greater than 10 to be adequate for reliable quantification in the 1D proton spectrum. Under our conditions, using a 14.1 T magnet equipped with a 3 mm inverse triple resonance cold probe with a sample volume of 55 μL in a 1.7 mm NMR tube, the detection limit was as low as 1 nmol before derivatization (Figure S5). In a typical biological sample in metabolomics study, where 1 million cells were extracted (approx. 1 pL per cell), this is equivalent to 1 mM concentration inside the cell. Within detection limits, the quantification response are linear comparing to the starting materials, as shown in Figure S5.

Determination of the 2HG enantiomers in a renal cell carcinoma

Figure S6 shows the TOCSY spectrum of the underivatized extract prepared from a renal cell carcinoma harboring a heterozygous p.R172M mutation in the IDH2 protein. We previously showed that the resonances at 4.0 (C2H) ppm, 2.23 ppm (C4H) 1.96/1.81 ppm (C3H) (Fig. S6) ppm belong to 2HG ⁶. After derivatization, the spectra were substantially different, with complete disappearance of the original resonance of lactate and 2HG, and the appearance of new resonances (Figure 2A). As only one set of resonances was observed in the derivatized samples, the 2HG produced from the mutant IDH2^R172M enzyme is a single enantiomer at δ=5.22 ppm, consistent with the D isomer (Table 1). The slight difference in chemical shifts from pure standards is attributable to the different temperature and matrix effects. We therefore spiked the sample with DATAN-derivatized L-2HG and re-recorded the NMR spectrum. We observed a new doublet of doublets at 5.17 ppm (Fig. 2B), confirming that the original sample contained essentially only D-2HG, as expected for the variant IDH2 enzyme.

Figure 2. DATAN-derivatized renal cell carcinoma extract.

The renal cell carcinoma cell extract 6 was derivatized with excess DATAN using the focused microwave at 70 °C for 7 minutes as described in the Experimental Section. NMR spectra were recorded at 14.1 T, 15°C.

A. DQCOSY + 1D ¹H NMR spectrum before spike in of DATAN-derivatized L-2HG

B. ¹H NMR spectra after spike in of DATAN-derivatized L-2HG

The enantiomeric purity of 2-HG was determined by estimating the ratio of the D,L peak areas of the derivatized compound in 1D spectra, as described for the standards. Since the L-isomer was below detectable limits, we estimate a minimum D:L 2-HG ratio of >10:1. The precision of the integration normalized to DSS in this sample pre-derivatization was CoV= 2%.

Similarly, lactate was present as a single L-enantiomer in the tissue extract, based on the chemical shifts and result of spiking with derivatized D,L-Lactate. This result ruled out a significant contribution from bacterial contamination or glyoxalase activity ¹⁸.

Analysis of lactate enantiomers in ex-vivo cancer patient tissue slice culture

Media from a NSCLC patient-derived ex vivo organotypic lung cancer slice cultures (OTC) were derivatized with excess DATAN and analyzed by NMR as described in Methods. The derivatized lactate also showed only one set of cross peaks in the 2D DQF-COSY spectrum (Figure S7A), indicating that only one enantiomer was present in the medium. After spiking with DATAN-derivatized L-lactate or D,L-lactate (Figure S7B), the L-lactate form was confirmed to be produced in the 24 hour culture of the ex vivo lung cancer OTC, presumably via glycolysis.

The application of DATAN derivatization on malate is also tested and revealed that it is also suitable for discrimination between D and L forms of malate (Figure S8). Therefore this method could potentially be expanded to distinguish other types of hydroxy acids enantiomers.

Conclusions

We improved the previously reported DATAN derivatization method ^14,21 for converting enantiomers of hydroxy acids to their diastereomers using a focus-beam microwave assisted synthesis method, which is simple, fast (7 min reaction time), and quantitative (100% conversion and absence of byproducts). We also bypassed the requirement for chromatography in LC-MS-based analysis ^14,21 by adopting ¹H NMR for analyzing the diastereomer products. In addition, the NMR method can simultaneously measure multiple analytes before and after derivatization for obtaining both total concentration and the concentration of individual D- or L-forms based on the D:L ratio in the derivatized state. Where resonances are resolved in 1D ¹H NMR spectra, absolute quantification (pre-derivatization) can be achieved to within better than 5% with adequate signal to noise ratio (>10) and enantiomeric ratios can be established in the 1D ¹H NMR spectra post derivatization using standard additions in about 30–60 minutes, depending on the concentration of the analytes. When 2D ¹H NMR is desired to achieve better resolution, we have shown that analytes present at ca. 20 nmol can be quantified for the extract of a small piece of human tumor tissue in 2–3 h acquisition at 14.1 T using a 1.7 mm tube in a 3 mm cryoprobe. For quantification in 1D ¹H spectra, we determined that that 0.5–1 nmole analyte must be present in the NMR tube to achieve an SNR of 10:1 in 50 minutes of acquisition under these conditions. This corresponds to the amount of analyte present in 10⁶ cells at ca. 1 fmol/cell (1 mM). Lower amounts could be analyzed by lengthier acquisitions, for example 4.5 h to detect a 3-fold lower amount.

The enantiomer resolution, which is proportional to the magnetic field strength, was highest for 2HG, but was sufficient for lactate and malate analysis.

Abbreviations:

ACN

acetonitrile

DATAN

(+) O,O’-diacetyl-L-tartaric anhydride

DNP

Dynamic Nuclear Polarization

DSS

2,2’dimethylsilapentane-5-sulfonate

gHAC

glacial acetic acid

2HG

2-hydroxyglutarate

2OG

2-oxyglutarate

IDH

isocitrate dehydrogenase

LDH

lactate dehydrogenase

MDH

malate dehydrogenase

PHGDH

phosphoglycerate dehydrogenase