Cost-efficient and user-friendly ¹⁷O/¹⁸O labeling procedures of fatty acids using mechanochemistry

Abstract

Two mechanochemical procedures for ¹⁷O/¹⁸O-isotope labeling of fatty acids are reported: a carboxylic acid activation/hydrolysis approach and a saponification approach. The latter route allowed first-time enrichment of important polyunsaturated fatty acids (PUFAs) including docosahexaenoic acid (DHA). Overall, a total of 9 pure labeled products were isolated in high yields (≥80%) and with high enrichment levels (≥37% average labeling of C=O and C-OH carboxylic oxygen atoms), under mild conditions, and in short time (<half day).

---

Fatty acids (FAs) represent a substantial class of biomolecules with multiple applications. Because of their biological importance, they are widely studied in the fields of lipidomics and metabolomics, as well as in human nutrition research, especially the essential ω-3 and ω-6 PUFAs¹ like arachidonic acid (AA) or DHA.² Furthermore, because of their amphiphilic character they are also used in materials science, for example as surfactants in nanoparticles synthesis, where they can direct the final shape and properties of nanoparticles.^3,4

The labeling of free FAs and their ester derivatives in stable isotopes (²H, ¹³C and ¹⁷O/¹⁸O) is of high interest as it allows to better understand the structure and reactivity of the complex (bio)molecules and materials systems. For example, by deuteration, the influence of the chain length of free FAs on the phase behaviour of stratum corneum membranes could be explored by ²H NMR.⁵ Using ¹³C labeling, in vivo¹³C NMR spectroscopy could be used to study the metabolism of polyunsaturated fatty acid methyl esters in the neonatal rat brains.⁶ When labeled in ¹⁸O, FA derivatives could be used as internal standards in mass spectrometry (MS) for metabolites quantitation or to elucidate reaction mechanisms.^7–9 Lastly, regarding ¹⁷O labeling, it is only very recently that we demonstrated that the ¹⁷O enrichment of free FAs could be of great interest for (nano)materials science applications, by providing unique insight into the structure of metal soaps and in the surface reactivity of functionalized nano-particles.¹⁰

Despite the great potential that ¹⁷O/¹⁸O-labeled FAs could offer to the research community, their use remains scarce compared to ²H and ¹³C labeled species. One of the reasons to this comes from the limited access to ¹⁷O/¹⁸O-labeled FAs, which are not commercially available and require to be synthesized. The very few oxygen enrichment protocols found in the literature almost exclusively focus on the ¹⁸O-labeling of specific FA derivatives used as standards in GC-MS analyses. In these protocols, the carboxylic oxygen atoms are labeled via hydrolysis of the corresponding esters (under basic or enzymatically catalysed conditions), or by acid catalysed oxygen exchange in labeled water.^7,8,11 These syntheses engage large excesses (>500 equiv.) of H₂O* in order to produce only μg quantities of labeled products. Although such low amounts of product are adequate for applications in GC-MS analyses, they are absolutely insufficient for the synthesis of (nano)materials, and a simple scale-up of the protocols would be very costly considering the price of enriched water (up to ∼2000 € for 1 mL of 90%-¹⁷O H₂O*, and ∼120 € for 1 mL of >98%-¹⁸O H₂O*).

In order to address the need for efficient and scalable ¹⁷O/¹⁸O labeling protocols which would be generally applicable to a broad range of FA substrates, we have developed alternative synthetic approaches using mechanochemistry,¹² where the ¹⁷O/¹⁸O enrichment is performed at ambient temperature and pressure using only few μL of H₂O*.¹³ Recently, we adapted this methodology for the first time for the synthesis of ¹⁷O and ¹⁸O-labeled oleic and stearic acids, and also scaled-up producing up to 1 g of ¹⁷O-labeled acids for less than 280 €.¹⁰

In this paper, we first demonstrate how by adapting our previous strategy, the scope of ¹⁷O/¹⁸O-labeled free FAs can be extended to other saturated FAs with the chain-length varying between C12 to C18 and also to PUFAs containing up to four double bonds, and how detailed analyses of reaction media can help determine working conditions for future syntheses. Then, we demonstrate how an original and efficient labeling scheme based on saponification of the ester derivatives can be used for the ¹⁷O/¹⁸O-labeling of the longest and highly sensitive PUFAs with five and six double bonds (EPA and DHA), thereby expanding the scope of approaches for enriching FAs.

First, we looked into expanding the range of free FAs which can be labeled in ¹⁷O/¹⁸O focusing on other FAs which are not only of synthetic interest, but also through which we could learn more about the labeling mechanisms, using activation/hydrolysis (Table 1a). Here, the enrichment consists of two mechanochemical steps: activation with small excess of 1,1′-carbonyldiimidazole (CDI, 1.1 equiv.), followed by hydrolysis of the activated acyl–imidazole intermediate using labeled water (2–3 equiv.).¹⁰ For each substrate, the oxygen isotope enrichment protocol was first optimized using H₂¹⁸O, due to the lower purchasing price of ¹⁸O-enriched water, before producing ¹⁷O-labeled equivalents. The course of the reaction was monitored by FT-IR looking at the shift of carbonyl vibration band. The homogeneity of the reaction mixture was controlled by analysing samples from different parts of the reaction vessel at each synthetic step. After complete hydrolysis, products were recovered from the jar using non-labeled water, the medium was acidified and the labeled FAs were finally recovered by extraction with ethyl-acetate (see ESI† for detailed protocols). Enriched products were isolated in good synthetic yields (≥80%) and with high enrichment levels (≥39% average per carboxylic oxygen, for the ¹⁸O-labeled products). The purity of isolated phases was confirmed by solution ¹H and ¹³C NMR spectroscopy and LCMS analyses. The ¹⁷O-labeled solid FAs were also analysed by ¹⁷O NMR, showing that on average, the ¹⁷O-label is distributed equally among the two oxygen sites in the isolated product (see ESI,† Fig. B1-13, B2-13 and B3-13).

Table 1 Fatty acids labeled in ¹⁷O/¹⁸O by mechanochemistry using (a) CDI-activation/hydrolysis, (b) saponification

For error bars see detailed Table in ESI (Table D-1). Average isolated yield (n = 3–4), ¹⁸O-EL = average enrichment level of carboxylic oxygen atoms using 97%-¹⁸O water (n = 2–3), ¹⁷O-EL = enrichment level using 90%-¹⁷O water (n = 1). Both oxygen atoms (C=O and C–OH) have the same probability of being enriched in the isolated product (only one is shown in red here for each labeled FA molecule). * For SA: K₂CO₃ (1 equiv.) was added at the hydrolysis step. Full work-up procedures leading to the pure labeled products are described in ESI.

---

For the labeling of saturated FAs, concerning the activation step, full conversion was achieved for all molecules within 30 min of milling at 25 Hz. The hydrolysis step, in contrast, was found to be strongly substrate dependent, and needed to be optimized for each acid separately. Indeed, it was observed that the milling time required to achieve complete hydrolysis highly correlates with the increasing chain-length of activated acids. Specifically, milling for 1 h at 30 Hz was sufficient to fully hydrolyse lauroyl-imidazole (C12), whereas the complete hydrolysis of palmitoyl-imidazole (C16) was achieved only after 3 h at 30 Hz. In the case of activated stearic acid (C18), full conversion was not observed even after 6 h of milling at 30 Hz, and could only be achieved by adding K₂CO₃ (1 equiv.) (Table 2).

Table 2 Melting points of non-labeled saturated FAs and their activated intermediate mixtures, and hydrolysis times used

Fatty acid	M.p. fatty acid [°C]	M.p. active. interm. mix. [°C]	Hydrolysis time^a [h]
a Time necessary for complete hydrolysis at 30 Hz. b With the addition of K2CO3 (1 equiv.); see Table D-3 (ESI) for further details.
LauA (C12)	43.4–46.3	58.4–62.3	1
MA (C14)	54.2–56.0	66.3–70.9	2.5
PA (C16)	62.4–64.4	73.5–78.1	3
SA (C18)	67.3–71.0	77.0–82.0	3^b

---

To further explain such differences, the melting points of the media isolated after activation were measured. These were found to follow the same trend as for the starting acids, with the melting points increasing with the length of the organic chain (Table 2). Additionally, it was found that the higher these melting points, the longer the time needed for hydrolysis. The evolution of the physical aspect of the reaction mixture is also worth mentioning: only when the reaction mixture had an oily aspect after hydrolysis was the hydrolysis found to be complete (see ESI,† Fig. D1). It can be hypothesized that the long milling times at 30 Hz required for the full conversion of the longer chain fatty acyl-imidazoles enabled to reach a temperature high enough to change the physical state of the reaction mixture, resulting in higher mobility of the molecules, which may accelerate the hydrolysis. These observations could be related to the progressive temperature increase during the ball-milling which has been described previously.^14–16

For the labeling of PUFAs containing up to four double bonds, the same two steps protocol was applied. Compared to saturated FAs, no difference in reactivity related to the chain length was observed during hydrolysis, and overall reaction times were reduced. In particular, all PUFAs were completely activated in 20 min of ball-milling at 25 Hz and fully hydrolysed in 90 min of milling at 25 Hz. The reduction of reaction times is most probably caused by the different physical state of non-labeled starting compounds, with unsaturated FAs being viscous oils at ambient temperature. No impurities were detected in the isolated enriched products. However, slightly lower enrichment levels were determined (39–43%, compared to 44–47% for saturated fatty acids, for ¹⁸O-labeled products), which can be explained by the choice of the milling equipment used for the labeling of PUFAs (see ESI,† Fig. D2). Hence, overall, we have proven here that the CDI-activation procedure can be easily optimized and generally applied to a variety of saturated and unsaturated FAs, allowing pure ¹⁷O/¹⁸O-enriched compounds to be produced in high yields.

In order to reach other important PUFAs, such as eicosapentaeonic acid (EPA) and docosahexaenoic acid (DHA), and to further extend the substrate scope, a new enrichment procedure was developed: a one-pot mechanochemical saponification (Table 1b).^17,18 This appeared as a potentially attractive strategy considering that many fatty acids are commercially available in the form of esters. In this enrichment protocol, ethyl-ester derivatives of fatty acids were introduced into the reaction jar together with labeled water (3 equiv.) and sodium ethoxide (1.5 equiv.). After only 30 min of milling, full conversion was observed by FT-IR. In this one-pot synthesis, labeled hydroxide anions were most probably first formed in situ via acid–base reaction between labeled water and sodium ethoxide, and subsequently reacted with the ester function in the substrate, producing the labeled molecule (as carboxylate) and ethanol as the only side-product. Using the same work-up as for CDI-activation/hydrolysis, all labeled molecules were obtained in high yield (≥93%) with high enrichment levels (≥45% average per carboxylic oxygen, for ¹⁸O-labeled products). The purity of PUFAs prepared via saponification was confirmed by ¹H and ¹³C solution NMR and by LCMS analyses, and no degradation was observed (see ESI†). Hence, this work demonstrates for the first time that saponification reactions are highly attractive enrichment protocols for oxygen labeling of carboxylic functions of FAs, starting from the ester derivatives.

In comparison to previously published ¹⁸O-enrichment protocols in which both oxygen atoms in carboxylic function get labeled (enrichment >80% per carboxylic oxygen), the reaction mechanisms of both our mechanochemical enrichment protocols allow only one oxygen isotope per carboxylic function to be introduced. This is apparent from the MS spectra of ¹⁸O-labeled products, where the main molecular peak is shifted by two m/z units (p + 2), and from ¹³C solution NMR analyses, where the main resonance of carboxylic function appears shielded by ∼0.025 ppm in comparison with the non-labeled molecule due to the ¹⁸O isotope effect (see ESI†). However, when looking at the whole MS pattern of unsaturated FA enriched via CDI-activation/hydrolysis procedure, we have noticed that a molecular peak shifted by four m/z units (p + 4) exhibits higher intensity than expected from simulations, suggesting a non-negligible presence of doubly labeled molecules. This becomes even more obvious when MS spectra of ¹⁸O-labeled α-linolenic acid (ALA) prepared by the two mechanochemical protocols are compared. Indeed, as shown in Fig. 1, the ultra-high resolution FTICR mass spectra enabled us to resolve the “p + 4” peak (isotopic fine structures)¹⁹ and to differentiate the relative contribution of doubly labeled molecules in each method. More specifically, it was observed that in the case of saponification the “p + 4” peak mainly consists of molecules containing two ¹³C and one ¹⁸O isotope as expected, with only very little amount of doubly ¹⁸O-labeled product, whereas in the case of CDI-activation/hydrolysis, ∼10 times more of doubly ¹⁸O-labeled molecules was formed compared to saponification, representing the dominant contribution. It can be hypothesized that doubly labeled products result from a CDI-reactivation of already enriched acid molecules due to the presence of remaining unreacted CDI reagent in the reaction mixture.

Fig. 1 Ultra-high resolution mass spectra in ESI negative mode of α-linolenic acid (ALA) enriched in ¹⁸O (a) via CDI-activation/hydrolysis or (b) saponification procedure; relative content of doubly-labeled product is compared in p + 4 zoom on the right side; p = monoisotopic mass of ALA.

Overall, it was shown here that mechanochemistry is a versatile approach that stands out from currently available oxygen enrichment protocols for carboxylic acids, thanks to the simplicity, rapidity and cost-efficiency of proposed procedures. It was shown that the reaction conditions can be easily optimized to extend the scope of ¹⁷O/¹⁸O-labeled fatty acids using CDI-activation/hydrolysis. A variety of saturated as well as unsaturated free FAs was for the first time enriched in ¹⁷O/¹⁸O and isolated pure in high yields (≥80%), with no need for column chromatography, in less than a half day. Moreover, the newly developed mechanochemical saponification enabled to reach other important and highly sensitive PUFAs, including EPA and DHA. Using simple work-up, ¹⁷O/¹⁸O-enriched PUFAs were isolated pure, in high yields (≥93%) and with high enrichment levels (≥45% average per carboxylic oxygen, for ¹⁸O-labeling). Moreover, enrichment was found to be stable over a 1 year storage period (see ESI,† Table D-2). Although, in average, only one oxygen per carboxylic function is labeled, the enrichment levels provided by this methodology are perfectly suitable for (nano)materials science applications^20–22 for gaining deeper insight into the structure and reactivity of different systems via IR (using ¹⁸O-labeled species) or high-resolution ¹⁷O NMR analyses. More generally, saponification under ball-milling represents a new, simple, fast and cost-efficient ¹⁷O/¹⁸O labeling strategy, which has great potential to be applied for oxygen enrichment of other organic molecules containing carboxylic functions. Among other advantages, it enables expanding the substrate scope of carboxylic molecules which can be enriched in ¹⁷O, and may also offer more practical work-up procedures compared to the CDI activation/hydrolysis route. These are points we are currently looking into.

This project has received funding from the European Research Council under the European Union's Horizon 2020 research and innovation program (Grant Agreement No. 772204; 2017 ERC-COG, MISOTOP project). Access to a CNRS FTICR research infrastructure (FR3624) is gratefully acknowledged. Dr Nicolas Patris (LAMA laboratory of Hydrosciences Montpellier) is acknowledged for IR-MS analyses on labeled water batches, Dr Céline Crauste (IBMM, Montpellier) for providing EPA and DHA substrates, and Drs Thierry Durand and Jean-Marie Galano (IBMM, Montpellier) for discussions.

Conflicts of interest

There are no conflicts to declare.

Notes and references

1. R. K. Saini and Y.-S. Keum, Life Sci., 2018, 203, 255–267.
2. C. C. De Carvalho and M. J. Caramujo, Molecules, 2018, 23, 2583.
3. Y. Yang, H. Qin, M. Jiang, L. Lin, T. Fu, X. Dai, Z. Zhang, Y. Niu, H. Cao, Y. Jin, F. Zhao and X. Peng, Nano Lett., 2016, 16, 2133–2138.
4. H. Wang and Y. Lian, J. Alloys Compd., 2014, 594, 141–147.
5. X. Chen, S. Kwak, M. Lafleur, M. Bloom, N. Kitson and J. Thewalt, Langmuir, 2007, 23, 5548–5556.
6. S. C. Cun-nane, S. C. R. Williams, J. D. Bell, S. Brookes, K. Craig, R. A. Iles and M. A. Crawford, J. Neurochem., 1994, 62, 2429–2436.
7. W. C. Pickett and R. C. Murphy, Anal. Biochem., 1981, 111, 115–121.
8. E. Hill and R. C. Murphy, Biol. Mass Spectrom., 1992, 21, 249–253.
9. E. C. Izgu, A. Björkbom, N. P. Kamat, V. S. Lelyveld, W. Zhang, T. Z. Jia and J. W. Szostak, J. Am. Chem. Soc., 2016, 138, 16669–16676.
10. J. Špačková, C. Fabra, S. Mittelette, E. Gaillard, C.-H. Chen, G. Cazals, A. Lebrun, S. Sene, D. Berthomieu, K. Chen, Z. Gan, C. Gervais, T.-X. Métro and D. Laurencin, J. Am. Chem. Soc., 2020, 142, 21068–21081.
11. J. Y. Westcott, K. L. Clay and R. C. Murphy, Biomed. Mass Spectrom., 1985, 12, 714–718.
12. S. L. James, C. J. Adams, C. Bolm, D. Braga, P. Collier, T. Friščić, F. Grepioni, K. D. Harris, G. Hyett and W. Jones, Chem. Soc. Rev., 2012, 41, 413–447.
13. T.-X. Métro, C. Gervais, A. Martinez, C. Bonhomme and D. Laurencin, Angew. Chem., Int. Ed., 2017, 56, 6803–6807.
14. H. Kulla, S. Haferkamp, I. Akhmetova, M. Röllig, C. Maierhofer, K. Rademann and F. Emmerling, Angew. Chem., Int. Ed., 2018, 57, 5930–5933.
15. K. Užarević, N. Ferdelji, T. Mrla, P. A. Julien, B. Halasz, T. Friščić and I. Halasz, Chem. Sci., 2018, 9, 2525–2532.
16. B. P. Hutchings, D. E. Crawford, L. Gao, P. Hu and S. L. James, Angew. Chem., Int. Ed., 2017, 56, 15252–15256.
17. V. Štrukil, ChemSusChem, 2021, 14, 330–338.
18. Y. Yeboue, M. Jean, G. Subra, J. Martinez, F. Lamaty and T.-X. Métro, Org. Lett., 2021, 23, 631–635.
19. L. Z. Samarah, R. Khattar, T. H. Tran, S. A. Stopka, C. A. Brantner, P. Parlanti, D. Veličković, J. B. Shaw, B. J. Agtuca, G. Stacey, L. Paša-Tolić, N. Tolić, C. R. Anderton and A. Vertes, Anal. Chem., 2020, 92, 7289–7298.
20. T. N. Sut, S. Park, B. K. Yoon, J. A. Jackman and N.-J. Cho, Langmuir, 2020, 36, 5021–5029.
21. F. Modugno, F. Di Gianvincenzo, I. Degano, I. D. van der Werf, I. Bonaduce and K. J. van den Berg, Sci. Rep., 2019, 9, 5533.
22. G. Kwak, M. Seol, Y. Tak and K. Yong, J. Phys. Chem. C, 2009, 113, 12085–12089.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1cc02165f

---