A high‐temperature water vapor equilibration method to determine non‐exchangeable hydrogen isotope ratios of sugar, starch and cellulose

Abstract The analysis of the non‐exchangeable hydrogen isotope ratio (δ2Hne) in carbohydrates is mostly limited to the structural component cellulose, while simple high‐throughput methods for δ2Hne values of non‐structural carbohydrates (NSC) such as sugar and starch do not yet exist. Here, we tested if the hot vapor equilibration method originally developed for cellulose is applicable for NSC, verified by comparison with the traditional nitration method. We set up a detailed analytical protocol and applied the method to plant extracts of leaves from species with different photosynthetic pathways (i.e., C3, C4 and CAM). δ2Hne of commercial sugars and starch from different classes and sources, ranging from −157.8 to +6.4‰, were reproducibly analysed with precision between 0.2‰ and 7.7‰. Mean δ2Hne values of sugar are lowest in C3 (−92.0‰), intermediate in C4 (−32.5‰) and highest in CAM plants (6.0‰), with NSC being 2H‐depleted compared to cellulose and sugar being generally more 2H‐enriched than starch. Our results suggest that our method can be used in future studies to disentangle 2H‐fractionation processes, for improving mechanistic δ2Hne models for leaf and tree‐ring cellulose and for further development of δ2Hne in plant carbohydrates as a potential proxy for climate, hydrology, plant metabolism and physiology.

However, recent studies show the great potential of δ 2 H values of plant compounds to retrospectively determine hydrological and climatic conditions (Anhäuser, Hook, Halfar, Greule, & Keppler, 2018;Gamarra & Kahmen, 2015;Hepp et al., 2015Hepp et al., , 2019Sachse et al., 2012), as well as to disentangle metabolic and physiological processes (Cormier et al., 2018;Estep & Hoering, 1981;Sanchez-Bragado, Serret, Marimon, Bort, & Araus, 2019;Tipple & Ehleringer, 2018) such as the proportional use of carbon sources (i.e., fresh assimilates vs. storage compounds) for plant growth (Lehmann, Vitali, Schuler, Leuenberger, & Saurer, 2021;Zhu et al., 2020). Enabling the analysis of δ 2 H ne of NSC, especially sugar at the leaf level, will make it possible to study processes and environmental conditions which are shaping the 2 H-fractionation of carbohydrates at a much higher time resolution compared to the analysis of δ 2 H ne of cellulose. New routines and high-throughput analytical methods for δ 2 H ne values of NSC are thus needed to enable widespread application in earth and environmental sciences.
The difficulty of establishing reliable methods for δ 2 H ne values of NSC and cellulose is mainly caused by the presence of oxygen-bound hydrogen atoms (H ex ) that can freely exchange with hydrogen atoms of the surrounding liquid water and water vapor. The interference of H ex greatly affects the analysis of δ 2 H ne , which retains useful information on climate, hydrology, metabolism and physiology. The oldest method of measuring δ 2 H ne is to derivatize hydroxyl groups with nitrate esters, using a mixture of either H 2 SO 4 or H 3 PO 4 with HNO 3 (Alexander & Mitchell, 1949;Boettger et al., 2007;DeNiro, 1981;Epstein et al., 1976). However, the nitration process requires a large sample amount, is labour intensive, uses hazardous derivatization reactions and leads to thermally unstable products. A newer derivatization method to measure δ 2 H ne in sugars is using N-methyl-bistrifluoroacetamide to replace H ex with trifluoroacetate derivatives, which are measured by gas chromatography -chromium silver reduction/high-temperature conversion-IRMS (GC-CrAg/HTC-IRMS) (Abrahim et al., 2020). This method still relies on a large sample amount of >20 mg extracted NSC, a relatively long measuring time, and the limitation of measuring only one element per analysis. Potential alternative methods that work without derivatization and use smaller amounts of material are based on water vapor equilibration, which sets H ex to a known isotopic composition that allows the determination of δ 2 H ne by mass balance (Cormier et al., 2018;Filot et al., 2006;Sauer et al., 2009;Schimmelmann, 1991;Wassenaar & Hobson, 2000). However, established water vapor equilibration methods are mainly calibrated for analysis of δ 2 H ne values of complex molecules such as cellulose, keratin and chitin (Schimmelmann et al., 1986;Wassenaar & Hobson, 2000) and whether these methods can also be used for the analysis of δ 2 H ne in NSC remains to be shown. The main purpose of this study was therefore to establish a high-throughput hot water vapor equilibration method to determine δ 2 H ne of NSC, based on already established protocols for cellulose (Sauer et al., 2009 (Cernusak et al., 2016). The leaf samples were immediately transferred to gas-tight 12 ml glass vials ('Exetainer', Labco, Lampeter, UK, Prod. No. 738W), stored on ice until the harvest was complete (≤2 hr), and then at À20 C in a freezer until further use (Appendix 1).
The sample material was dried using a cryogenic water distillation method (West, Patrickson, and Ehleringer (2006), crumbled with a spatula (dicotyledon species) or cut with scissors (monocotyledon species) into small pieces, and 100 mg of the fragmented material was separated for cellulose extraction. The remaining leaf material was then ball-milled to powder (Retsch MM400, Retsch, Haan, Germany) for NSC extraction.

| Cellulose and starch nitration, and isotopic analysis of the nitrated products
Nitrates of cellulose and starch without exchangeable H were used as reference material to assess the δ 2 H ne values derived from the hot water vapor equilibration method. Nitration of cellulose and starch standards was performed following the method of Alexander and Mitchell (1949), using a mixture of P 2 O 5 and 90% HNO 3 . δ 2 H values of nitrated cellulose and starch were analysed with a TC/EA-IRMS system, using a reactor filled with chromium as described by Gehre et al. (2015). Reference materials for δ 2 H measurements of cellulose and starch nitrates were the IAEA-CH-7 polyethylene foil (PEF; International Atomic Energy Agency, Vienna, Austria) for a first offset correction and the USGS62, USGS63 and USGS64 caffeine standards (United States Geological Survey, Reston, Virginia, U.S.A.) (Schimmelmann et al., 2016) for the final normalization.

| Preparation of leaf cellulose and NSC for δ 2 H ne analysis
Every compound (i.e., sugars, starch and cellulose) was extracted once per sample. Cellulose (hemicellulose) was extracted from 100 mg of the fragmented leaf material in F57 fibre filter bags (made up of polyester and polyethylene with an effective pore size of 25 μm; ANKOM Technology, Macedon NY, U.S.A.). In brief, the samples were washed twice in a 5% sodium hydroxide solution at 60 C, rinsed with deionized water, washed 3 times for 10 hr in a 7% sodium chlorite solution, which was adjusted with 96% acetic acid to a pH between 4 and 5, and subsequently rinsed with boiling hot deionized water, and dried overnight in a drying oven at 60 C. The neutral sugar fraction ('sugar', a mixture of sugars, typically glucose, fructose, sucrose and sugar alcohol [Rinne, Saurer, Streit, & Siegwolf, 2012]) were extracted from 100 mg leaf powder and further purified using ion-exchange cartridges, following established protocols for carbon and oxygen isotope analyses (Lehmann et al., 2020;Rinne et al., 2012). This is needed to separate the sugar from other water-soluble compounds such as amino acids which would alter the resulting δ 2 H ne values (Schmidt, Werner, & Eisenreich, 2003). Starch was extracted from the remaining pellet of the sugar extraction via enzymatic digestion following the established method for carbon isotope analysis (Richter et al., 2009;Wanek et al., 2001). The same protocol was used to hydrolyse the commercial starch standards. Aliquots of the extracted sugar (including those derived from starch) were pipetted in 5.5 Â 9 mm silver foil capsules (IVA Analysentechnik GmbH & Co. KG, Germany, Prod. No. SA76981106), frozen at À20 C, freeze-dried, folded into cubes and packed into an additional silver foil capsule of the same type, folded again and stored in an exicator at low relative humidity (2-5%) until isotope analysis.
2.5 | δ 2 H ne analysis of cellulose and NSC using a hot water vapor equilibration method One microgram of commercial starch or cellulose standard was packed into 3.3 Â 5 mm silver foil capsules (IVA, Prod. No. SA76980506), which led to a total peak area between 20 and 30-V seconds (Vs) of each IRMS analysis. For sugar standards, one mg was transferred first into a 5.5 Â 9 mm silver foil capsule (IVA), and additionally packed in a second capsule of the same size and folded again. The reason for the double packing was the observation that sugar samples became liquefied and rinsed out of single-packed capsules during the hot water vapor equilibration, which led to a loss of sample and to negative impacts on the analysis of δ 2 H ne in sugars. Such rinsing was prevented by double packing and had no negative impact on the drying time of the sugars (Appendix 2). The double packing did not have a negative impact on the equilibration itself, as indicated by the high x e of the sugars (Table 1). All packed samples were stored in an exicator at low relative humidity (2-5%) until isotope analysis.
All samples were equilibrated in a home-built The inlet was connected to a stainless steel tube (i.e., feeding capillary, BGB, Switzerland), which was leading out of the oven where a santoprene pump tubing was fitted into a peristaltic pump (Appendix 6).
The end of the santoprene pump tubing was inserted into a 50 ml falcon tube containing the equilibration water. The peristaltic pump pro- For testing the reproducibility of the adapted method, triplicates of each type of cellulose and sugar samples were equilibrated independently on separate days following a standardized sample sequence (Appendix 7), in total three times with Water 1 (δ 2 H = À160‰) and three times with Water 2 (δ 2 H = À428‰). For starch and digested starch, triplicates were equilibrated only once with Water 1 and once with Water 2.
Subsequently, all samples (still hot) were immediately transferred into a Zero Blank Autosampler (N.C. Technologies S.r.l.), which was installed on a sample port of a high-temperature elemental analyser system. The latter was coupled via a ConFlo III interface to a Delta Plus XP IRMS (TC/EA-IRMS, Finnigan MAT, Bremen, Germany). It is crucial to transfer the samples as fast as possible and still hot from the equilibration chamber to the autosampler to avoid any isotopic reequilibration of the sample with air moisture and water absorption.
The autosampler carousel was evacuated to 0.01 mbar and afterwards filled with dry helium gas to 1.5 bar to avoid any contact with ambient water (vapor). The samples were pyrolysed in a reactor according to Gehre, Geilmann, Richter, Werner, and Brand (2004) 2.6 | Calculation of non-exchangeable hydrogen isotope ratio (δ 2 H ne ) According to Filot et al. (2006), the %-proportion of exchanged hydrogen during the equilibrations [x e , Equation (2)] can be calculated as: where δ 2 H e1 and δ 2 H e2 are the δ 2 H values of the two equilibrated samples, δ 2 H w1 and δ 2 H w2 are the δ 2 H values of the two waters used, α e-w is the fractionation factor of 1.082 for cellulose (Filot et al., 2006). While α e-w needs to be adapted for different compounds and fractions with different functional groups (Schimmelmann, 1991 Statistical analyses (one-way ANOVA and Tukey posthoc test) were performed using R version 3.6.3 (R. Core.Team, 2021 (Werner & Brand, 2001) is applied, that is, all samples are prepared and measured in the same way.
Besides, the calculated x e values of the IAEA-CH-7 reference material without any H ex were close to 0 throughout all measurements, denoting the absence of absorbed water on the surface of each compound, as well as the analytical reproducibility for all δ 2 H ne values of cellulose, was high as indicated by a standard deviation of 0.8 to 1.9‰ for three repetitions.
The same method was also applied to analyse δ 2 H ne of NSC ( which is comparable to the precision of derivatization methods (Dunbar & Schmidt, 1984: 1.9‰;Augusti et al., 2008: 2 and 10‰; Abrahim et al., 2020: 0.4 and 3.6‰). As no nitrated sugars were available due to the safety problems with sugar nitration, we could not calculate the accuracy. We, however, can assume that the accuracies for sugars should be in a comparable range as those derived from digested starch (À8.0 and À2.0‰). The reproducibility of the results for all tested commercial sugars ranged between 4.0 and 8.6‰ for three repetitions. The x e of the different sugars ranged between 34.1 and 53.5% and was thus similar or very close to x e.pot , which gives further confidence in the reliability of the method for sugars. The smaller deviation of x e from x e.pot for sugars than for cellulose might be explained by the dissolution of the sugars during the hot water vapor equilibration, leading to a breakdown of the crystal structure of the sugars. This might have facilitated a complete exchange of H ex with the water vapor in sugars, that is, not feasible for cellulose (Sauer et al., 2009;Schimmelmann, 1991 (Leaney, Osmond, Allison, & Ziegler, 1985;Sternberg, Deniro, & Ajie, 1984). While the observed variation in δ 2 H ne of NSC and cellulose among the photosynthetic pathways are unlikely to be explained solely by differences in leaf water 2 H enrichment (Kahmen, Schefuß, & Sachse, 2013;Leaney et al., 1985), higher leaf water δ 2 H values might partially contribute to higher δ 2 H ne of NSC and cellulose in CAM plants compared to C 3 plants (Smith & Ziegler, 1990 isotope effects during metabolic processes (Cormier et al., 2018;Cormier, Werner, Leuenberger, & Kahmen, 2019). Our results are supported by a previous study (Luo & Sternberg, 1991;Schleucher et al., 1999), showing that nitrated starch was more 2 H-depleted than nitrated cellulose within the same autotrophic photosynthetic tissue, which can be interpreted as another proof for the reliability of the new method for δ 2 H ne values of NSC. The high variability in 2 Hfractionation in the sequence from sugars to starch to cellulose (Table 2) between all tested species indicates high variability in common 2 H-fractionation processes, which is also supported by recent studies (Cormier et al., 2018;Sanchez-Bragado et al., 2019). Thus, the variability in 2 H-fractionation may find application in future plant physiological studies, investigating stress responses or short-and long-term carbon dynamics. We assume that δ 2 H ne of NSC are susceptible to diel or seasonal changes in environmental conditions such as temperature and light intensity due to their short turnover time (Fernandez et al., 2017;Gibon et al., 2004). The variability in 2 Hfractionation between different species might also be important if multiple tree species are used during the establishment of tree-ring isotope chronologies in dendroclimatological studies (Arosio, Ziehmer-Wenz, Nicolussi, Schlüchter, & Leuenberger, 2020).
In conclusion, we show that a hot water vapor equilibration method originally developed for cellulose can be adapted for accurate, precise and reproducible analyses of δ 2 H ne in non-structural carbohydrates (NSC) such as sugar and starch. By applying the method for compounds from different plant species, we demonstrated that this analytical method can now be used to estimate 2 Hfractionation among structural and NSC and to distinguish plant material from plants with different photosynthetic pathways. It should be noted that the method presented herein enables analysis of δ 2 H ne of bulk sugar and sugar derived from digested starch and is therefore not compound-specific nor position-specific. Yet, our δ 2 H ne method allows us to measure NSC samples in high-throughput and we thus expect that it will help to identify important 2 Hfractionation processes. These findings could then eventually be studied in more detail with compound-specific methods (GC-IRMS [Abrahim et al., 2020]) or methods giving positional information (NMR [Ehlers et al., 2015]). We therefore expect that the method will find widespread applications in plant physiological, hydrological, ecological and agricultural research to study NSC fluxes and plant performance, and the beverage and food industry, to distinguish between sugars of different origins, which could be applied to check if a certain product is altered by the addition of low-cost supplements. We also expect that the method can help to improve mechanistic models for 2 H distributions in organic material (Roden, Lin, & Ehleringer, 2000;Yakir & DeNiro, 1990). The method may further help, in combination with other hydrogen isotope proxies (e.g., fatty acids, n-alkanes or lignin methoxy groups), researchers to better understand metabolic pathways and fluxes, shaping the hydrogen isotopic composition of plant material.