Interactive comment on “ Optimizing sample pretreatment for compound-specific stable carbon isotopic analysis of amino sugars in marine sediment ”

Amino sugars are quantitatively significant constituents of soil and marine sediment, but their sources and turnover in environmental samples remain poorly understood. The stable carbon isotopic composition of amino sugars can provide information on the lifestyles of their source organisms and can be monitored during incubations with labeled substrates to estimate the turnover rates of microbial populations. However, until now, such investigation has been carried out only with soil samples, partly because of the much lower abundance of amino sugars in marine environments. We therefore optimized a procedure for compound-specific isotopic analysis of amino sugars in marine sediment, employing gas chromatography–isotope ratio mass spectrometry. The whole procedure consisted of hydrolysis, neutralization, enrichment, and derivatization of amino sugars. Except for the derivatization step, the protocol introduced negligible isotopic fractionation, and the minimum requirement of amino sugar for isotopic analysis was 20 ng, i.e., equivalent to ~8 ng of amino sugar carbon. Compound-specific stable carbon isotopic analysis of amino sugars obtained from marine sediment extracts indicated that glucosamine and galactosamine were mainly derived from organic detritus, whereas muramic acid showed isotopic imprints from indigenous bacterial activities. The δ 13 C analysis of amino sugars provides a valuable addition to the biomarker-based characterization of microbial metabolism in the deep marine biosphere, which so far has been lipid oriented and biased towards the detection of archaeal signals.


Introduction
Amino sugars are significant sedimentary components that are mostly derived from microorganisms and invertebrates.For example, the four major amino sugars, glucosamine (GlcN), galactosamine (GalN), mannosamine (ManN) and muramic acid (MurA), accounted for up to 12 % of total organic carbon (TOC) in grassland soils in North America (Amelung et al., 1999) and ∼ 2 % of TOC in coastal Peruvian surface sediments (Niggemann and Schubert, 2006).Amino sugars are preserved in the form of biopolymers such as peptidoglycan, chitin, and lipopolysaccharides.Because of the minor contribution of amino sugars from meso-and macro-organisms (Simpson et al., 2004), the proportion of amino sugar-derived C in the total C pool has been used frequently as a proxy for microbial contributions to soil organic matter, and the ratios between different amino sugars have been employed as indicators of microbial community compositions (e.g., Guggenberger et al., 1999;Zhang et al., 1999;Amelung, 2001;Glaser et al., 2004).In marine environments, amino sugars could be derived from microbes, phytoplankton and animals.Prokaryotic biomass is thought to be the major source of GlcN and GalN in marine sediment (e.g., Niggemann and Schubert, 2006;Langerhuus et al., 2012) and seawater (Benner and Kaiser, 2003); however, most amino sugars in seawater are not likely associated with peptidoglycan (Aluwihare et al., 2005;Aluwihare and Meador, 2008).Planktonic bacterial sources may contribute to the sedimentary MurA pool, and thus may complicate the use of MurA as a biomarker for sedimentary bacteria.Recently, the investigation of amino sugars has been extended to the marine deep Published by Copernicus Publications on behalf of the European Geosciences Union.
biosphere by Lomstein et al. (2012), who used MurA as a tool for indirectly quantifying endospores in subseafloor sediment.
The stable carbon isotopic composition (δ 13 C) of individual cellular biomarkers can reveal the carbon metabolism and/or lifestyles of the source organisms in natural environments (e.g., Lin et al., 2010;Schubotz et al., 2011), and incubations with 13 C-labeled substrates have been used to probe microbial communities by tracking the production of labeled biomarkers over time (e.g., Veuger et al., 2006;Kellermann et al., 2012;Lin et al., 2013).The latter technique has been widely employed in soil science to investigate the formation and/or turnover kinetics of amino sugars (e.g., Glaser et al., 2005;Decock et al., 2009;Bai et al., 2012).However, to date, the isotopic composition of amino sugars in marine sediment remains poorly explored, partly because of the much lower concentration of amino sugars in marine sediment compared to soil, which results in the need for an efficient pretreatment procedure to enable precise isotopic determination.
Compound-specific δ 13 C analysis of amino sugars is commonly performed via isotope ratio mass spectrometry (IRMS) after separation of compounds by either gas chromatography (GC; Glaser and Gross, 2005), liquid chromatography (LC; Bodé et al., 2009) or, more recently, ion chromatography (IC; Dippold et al., 2014).Compared with LC-IRMS, the GC-based method is less sensitive to adverse effects of the sample matrix on detection (McCullagh, 2010;Morrison et al., 2010;Rinne et al., 2012), and requires smaller amounts of amino sugars.The accuracy and precision of the GC-based method, though compromised by the introduction of carbon during derivatization (Glaser and Gross, 2005), has been shown to sufficiently resolve molecular isotopic differences caused by diverse biogeochemical processes in marine sediments (cf.Lin et al., 2010).Therefore, we consider GC-IRMS an attractive alternative to LC-IRMS for the isotopic analysis of amino sugars in deep marine sediments.A substantial effort has been made during the past few decades to optimize the pretreatment procedure for GC-based quantification of amino sugars, which requires three major steps: acid hydrolysis, purification, and derivatization, with hydrolysis being the key step for releasing amino sugars from biopolymers.Besides the most frequently used hydrolysis protocol, namely 6 M hot hydrochloric acid (HCl; Zhang and Amelung, 1996), less destructive procedures involving either hot trifluoroacetic acid (TFA; Neeser and Schweizer, 1984) or sulfuric acid (H 2 SO 4 ; Fox et al., 1983) have been proposed for the simultaneous extraction of neutral and amino sugars.Different purification protocols, such as neutralization with a base solution (Zhang and Amelung, 1996), precipitation of excess acid (Cowie and Hedges, 1984;Neeser and Schweizer, 1984), and deionization of hydrolysate (Cowie and Hedges, 1984), to name a few, have been used to reduce the content of acid and salts that are known to interfere with amino sugar derivatization.Conversion of amino sugars for GC analysis has been achieved via derivatization into alditol acetates (AA; Fox et al., 1983), aldononitrile acetates (ANA; Guerrant and Moss, 1984), or O-methyloxime acetates (Neeser and Schweizer, 1984).However, to perform isotopic analysis of amino sugars at trace levels, a systematic evaluation of these various methods with regard to the product recovery and pretreatment reproducibility is necessary.
The goal of this study is to devise a pretreatment protocol for GC-based isotopic analysis of amino sugars in subseafloor sediments.Since deep sediment samples are generally severely limited in size and often contain substantially lower TOC than surface sediments, a protocol for sensitive isotopic analysis of amino sugars in trace amounts is critical.Hence, we tackled three major analytical tasks.(1) We systematically evaluated existing pretreatment methods for amino sugar analysis to select the method that gave the highest recoveries of products from marine sediments.(2) We introduced the use of a new type of solid-phase extraction (SPE) into the pretreatment protocol and demonstrated enhanced recoveries compared with existing methods.(3) We developed a preparative high-performance liquid chromatography (HPLC) method to enrich amino sugars to a desired concentration for proper isotopic assessment.We then applied our protocol to determine the stable carbon isotopic compositions of amino sugars from the selected subseafloor sediment samples.

Standards and environmental samples
The amino sugar standards, derivatization reagents, and all other chemicals used in this study were purchased from Sigma-Aldrich Chemie GmbH (Munich, Germany) or Merck KGaA (Darmstadt, Germany).The Supelclean ™ ENVI-Carb ™ Plus SPE cartridges and accessories were obtained from Sigma-Aldrich Chemie GmbH or Carl Roth GmbH (Karlsruhe, Germany).
A batch of surface sediment for method optimization was collected from the upper tidal flat of the Wadden Sea near Wremen, Germany, in February 2010 (53 • 38 0 N, 8 • 29 30 E).In addition, four marine sediment samples of different types were selected for stable carbon isotopic analysis of individual amino sugars.The seep sample was recovered from the continental margin off Pakistan (site GeoB 12315-9, 1-2 cm below the seafloor, cmbsf) during RV Meteor cruise M74/3 in November 2007.It was located within the lower part of the oxygen minimum zone and associated with dense microbial mats from sulfide-oxidizing bacteria (Fischer et al., 2012).Two subseafloor samples were retrieved from the Marmara Sea (site GeoB 15104-2) and the Black Sea (site GeoB 15105-2) during RV Meteor cruise M84/1 (DARCSEAS) in February 2011 (Zabel et al., 2013).The sample from the Marmara Sea was collected at 2.88 mbsf and was located in a sapropel layer deposited under suboxic bottom-water conditions (Çagatay et al., 2000).The Black Sea sample was collected at 1.57 mbsf, where the sediment was highly diluted by terrigenous components and showed a lack of bioturbation.The deepest sediment sample was recovered from the Peru Margin (Ocean Drilling Program leg 201, 1229A-3H-2, 16.6 mbsf), and had a high TOC content (3.8 %).

Hydrolysis tests
Hydrolysis tests were performed in triplicate in 40 ml glass tubes by adding different acids to ∼ 5 g of freeze-dried Wadden Sea sediment.The tubes were filled with nitrogen before being sealed with Teflon-coated screw caps to prevent oxidation of sugars at high temperature during hydrolysis.The evaluated hydrolytic conditions were as follows.
1.The sample was kept at room temperature for 2 h after the addition of 12 M H 2 SO 4 , followed by dilution to 1.2 M H 2 SO 4 and heating at 105 • C for 3 h (modified from Cowie and Hedges, 1984).
3. The sample was treated with 6 M HCl at 105 • C for 8 h (Zhang and Amelung, 1996).
After hydrolysis, the samples were cooled to room temperature, spiked with 100 µg of myo-inositol as an internal standard, and centrifuged at 800 × g for 5 min.After being passed through combusted glass microfiber filters (GF/F, 70 mm , Whatman ™ ), the hydrolysates obtained from treatments (2) and (3) were evaporated to dryness with a rotary evaporator (45 • C, under vacuum) and re-dissolved in 4 ml of MilliQ water.

Neutralization and desalting
Samples hydrolyzed with H 2 SO 4 were neutralized by adding finely ground barium hydroxide (Ba(OH) 2 ; Cowie and Hedges, 1984), whereas TFA-treated samples were freezedried overnight to remove the acid.Hydrolysates liberated with HCl were initially subjected to the procedure described by Zhang and Amelung (1996), with slight modification.The acidic solution was adjusted to pH 6.5-7.0 with 1 M potassium hydroxide (KOH) and centrifuged to remove the precipitates.The supernatant was evaporated to dryness under N 2 , the condensates were re-dissolved in 2 ml methanol (MeOH), and amino sugars were collected in the supernatant after centrifugation.Because treatments with HCl gave the best recoveries (see below), we went on to explore other neutralization and desalting methods that are compatible with HCl.A standard mixture containing 20 µg of each amino sugar was treated with 1 ml of 6 M HCl, and the following three methods were performed in triplicate to compare the yields with those of Zhang and Amelung (1996).
1. Silver carbonate (Ag 2 CO 3 ) was added gradually to the mixture to neutralize the HCl.The silver chloride (AgCl) precipitate was removed by centrifugation (cf.Neeser and Schweizer, 1984).
2. After neutralization with a 1 M KOH solution, the mixture was desalted by percolating the solution through a glass column filled with 3 g of pre-conditioned Dowex 50WX8 cation exchange resin (100-200 mesh, H + form), as described by Amelung et al. (1996).The column was washed with 10 ml of MilliQ water to remove excess salts, and amino sugars were subsequently eluted with 10 ml of 2 M ammonium hydroxide (NH 4 OH).
3. We evaluated for the first time the applicability of Supelclean ™ ENVI-Carb ™ Plus SPE cartridges in carbohydrate analysis.ENVI-Carb ™ Plus is a microporous amorphous carbon molecular sieve developed for the extraction of highly polar compounds from water.Its predecessor, Supelclean ™ ENVI-Carb ™ , has been used to extract oligosaccharides in aqueous samples (Itoh et al., 2002).Prior to use, the SPE cartridge was preconditioned sequentially with 10 ml methylene chloride (DCM), 20 ml MeOH, and 15 ml MilliQ water.The hydrolysate containing amino sugars was neutralized with 1 M KOH and desalted by pulling through the SPE cartridge coupled with an SPE manifold (Carl Roth GmbH, Karlsruhe, Germany) under vacuum.The SPE cartridge was then dried for 10 min and eluted under vacuum in the reversed direction with 5 ml MeOH, followed by 5 ml DCM/MeOH (1 : 1, v : v) to recover the amino sugars.
All the desalted products were evaporated under a stream of N 2 , lyophilized overnight, and converted to GC-amenable derivatives for analysis.Quantities and recoveries of individual amino sugars were determined via calibration curves that were generated from pure standards.The Wadden Sea sediment was also used to assess the efficiencies of the neutralization and desalting methods for environmental samples.

Derivatization and purification
Amino sugar standards were transformed to alditol acetate (AA) or aldononitrile acetate (ANA) derivatives in triplicate following the methods of Fox et al. (1983) or Guerrant and Moss (1984), respectively.After being converted to ANA derivatives, amino sugars extracted from the environmental samples were further purified with a self-packed silica gel column (0.5 g; Kieselgel, 0.06-0.2mm; Carl Roth GmbH, Karlsruhe, Germany; cf.Lin et al., 2010).The best recovery (> 95 % of each amino sugar as an ANA derivative) was achieved by elution with 8 ml hexane/ethyl acetate (1 : 4, v : v).The eluent was evaporated to dryness under a N 2 stream and re-dissolved in hexane/ethyl acetate (1 : 1, v : v) prior to analysis.
Amino sugars were thereafter separated into two fractions, the first (F1; 5.0-7.6 min) containing the three basic amino sugars, and the second (F2; 7.6-9.6min) MurA.Fractions collected during preparative HPLC were evaporated to dryness and converted to GCamenable derivatives to monitor separation efficiency and the isotopic fractionation effect.With this method, we also obtained the ManN-free and MurA-free sedimentary matrices (i.e., F2 and F1, respectively), which were used to determine the accessible amino sugar concentrations for isotopic analysis (see below).

Accessible amino sugar concentrations for GC-IRMS analysis
To determine the accessible concentration range of amino sugars for compound-specific δ 13 C analysis, 10-140 ng of the two minor amino sugars found in natural marine environments (i.e., ManN and MurA) were converted to ANA derivatives and injected as pure standards into the GC-IRMS in triplicate.Additionally, pure ManN or MurA was spiked to the ManN-free or MurA-free sedimentary matrix, to validate the δ 13 C values for realistic conditions.These sedimentary matrices were obtained from the preparative HPLC separation of the Peru Margin sample.Monitored by GC-mass spectrometry (MS) in selective ion monitoring (SIM) mode, the concentration of the corresponding analyte in the matrix was below the detection limit (< 0.05 ng on column).

Instrumentation
Quantification of amino sugars was accomplished using an Agilent 6890N GC instrument coupled to an Agilent 5973 inert Mass Selective Detector (MSD) with an electron impact (EI) source, whereas compound-specific isotopic analysis of amino sugars was performed with a Trace GC Ultra instrument coupled to a Delta Plus XP isotope ratio mass spectrometer via Combustion Interface III (Thermo Finni-gan MAT GmbH, Bremen, Germany).Two columns with different polarities, an Optima 17MS column (30 m × 0.25 mm, 0.25 µm film thickness; Macherey-Nagel GmbH & Co. KG, Düren, Germany) and an Rxi-5ms column (30 m × 0.25 mm, 0.25 µm film thickness; Restek GmbH, Bad Homburg, Germany), were applied to optimize the separation of sugar derivatives.
The ANA method produced the most stable derivatives and satisfying results (see below); the corresponding GC operation conditions were specified here in detail.Separation of ANA derivatives in the Optima 17MS column was modified from Glaser and Gross (2005).The injector temperature was 250 • C, and helium was used as a carrier gas at a constant flow rate of 1 ml min −1 .The GC temperature program was initiated at 80 • C and held for 4 min, then increased to 250 • C at 30 • C min −1 and held for 10 min, and finally raised to 280 • C at 3 • C min −1 and held for 5 min.The GC parameters for the Rxi-5ms column were as follows: injector temperature, 250 • C; carrier gas, helium; flow rate, 1.1 ml min −1 .The oven temperature was kept at 70 • C for 1 min, ramped up to 230 • C at 20 • C min −1 , held for 20 min, and further increased at 20 • C min −1 to 300 • C and held for 5 min.The mass spectrometer was programmed in SIM mode to target specific mass fragment ions of the derivatives for quantification.The selected ions were m/z 187 and 289 for GlcN, GalN and ManN, and m/z 236 and 356 for MurA.
Separation and detection of amino sugars were achieved with an Agilent 6130 MSD single quadrupole mass spectrometer coupled to an Agilent 1200 Series HPLC system via a multimode ion source in electrospray ionization mode (Agilent Technologies Deutschland GmbH, Böblingen, Germany) equipped with an Econosphere NH 2 column (250 mm × 4.6 mm, 5 mm particle size; Alltech Associates Inc., Deerfield, IL, USA).
Stable carbon isotopic compositions of the pure amino sugars, internal standards and the acetylation reagent were determined independently by a Flash 2000 organic elemental analyzer coupled to a Delta V Plus isotope ratio mass spectrometer (EA-IRMS) via a ConFlo IV interface (Thermo Finnigan MAT GmbH).The EA-IRMS was operated in C, N mode and controlled by the Finnigan Isodat 3.0 software.The oxidation and reduction reactors were heated to 999 • C and 680 • C, respectively.Helium was used as a carrier gas at a flow rate of 100 ml min −1 , while the oxidation reagent oxygen had a flow rate of 200 ml min −1 .The CO 2 reference gas pulse was introduced two times (20 s each) at the beginning of each run.Quadruplicate measurements by EA-IRMS generated highly accurate δ 13 C values of each compound.

Hydrolysis and neutralization methods
values only for ANA derivatives of amino sugars, which exhibited the highest yields.The δ 13 C values and the corresponding standard deviations (s Total ) of individual amino sugars were calculated following the procedure described by Glaser and Gross (2005), with a slight modification.3-Omethyl-D-glucopyranose (3-O-Me-Glc) was applied as an internal standard to correct the isotopic fractionation that is known to occur during acetylation of the hydroxyl groups (Lin et al., 2010).A compound-specific factor F to correct for fractionation during derivatization was introduced by Glaser and Gross (2005, Eq. 1).
where N is the number of C atoms of the amino sugar derivative (N Der ), the original amino sugar (N AS ), and the acetyl group (N Acet ) of acetic anhydride used for derivatization.We determined the F factor experimentally using ANA derivatives of seven external standard solutions in the concentration range of 1-20 µg per batch for ManN and MurA, and 1-60 µg per batch for the other standards.

Hydrolytic conditions for releasing amino sugar monomers
Table 1 summarizes the results of the hydrolysis tests.Hydrolysis with 6 M HCl provided the highest recoveries of amino sugars from the Wadden Sea surface sediment sample, followed by H 2 SO 4 and TFA.GlcN and GalN were the most abundant amino sugars, with concentrations being one order of magnitude higher than those of ManN and MurA.
Based on these results, further work aimed to optimize the neutralization and desalting steps that followed the hot HCl treatment.We observed that the HCl method yielded lower recoveries of neutral sugars compared with the other two protocols (data not shown), a result in agreement with the previous finding of Amelung et al. (1996), and suggesting the occurrence of a dehydration reaction between HCl and the monosaccharides.

Comparison of neutralization and desalting methods after hydrolysis by HCl
Recoveries of amino sugar standards ranged from 77 to 98 % using the procedure proposed by Zhang and Amelung (1996), i.e., neutralization with 1 M KOH and desalting with MeOH (Table 2).We did not observe significant differences in amino sugar recoveries using standard compounds after employing either the ENVI-Carb ™ Plus SPE cartridge or the method described by Zhang and Amelung (1996).However, for the surface sediment sample, recoveries using the former method exceeded those of the latter by factors of 1.5 to 2 (Table 1), suggesting that the SPE method is less sensitive to the presence of sedimentary matrices.Moreover, we noted that the SPE cartridge allowed for excellent recoveries of neutral hexoses such as glucose, galactose, and mannose (data not shown).The other two methods tested in this study yielded less satisfying results.Ag 2 CO 3 reacts with HCl to form CO 2 and AgCl, which has very low solubility in water and eliminates the need for a desalting step.However, with the exception of ManN, major losses of amino sugars were observed (Table 2), presumably owing to interactions between amino sugars and the AgCl-Ag 2 CO 3 mixture.Cation exchange using the Dowex 50WX8 cation exchange resin allows for the separation of amino sugars (in the NH 4 OH fraction) from neutral sugars (in the H 2 O fraction).Recoveries of amino sugars by this method were highly variable (Table 2), partly because of the incomplete evaporation of NH 4 OH, which might affect the efficiency of ANA derivatization.Nevertheless, further evaluation should be performed in future work using a recently optimized cation exchange-based procedure (Indorf et al., 2013).

Derivatization of amino sugars
Although the AA derivatization method has been employed previously to investigate amino sugars (Fox et al., 1995(Fox et al., , 1996)), it was rather time consuming (∼ 1 d; cf.Fox et al., 1983), and the derivative of MurA exhibited poor stability compared with the corresponding basic amino sugar derivatives (data not shown).The addition of N-methylimidazole as a catalyst helped to accelerate the acetylation of basic amino sugars, but it failed to produce an appreciable signal peak for the MurA derivative (Whiton et al., 1985).By contrast, the acetylation time of the ANA derivatization method was reasonably low (20 min), and the derivatives remained stable at −20 • C for up to 1 yr, as confirmed by repeated injections of amino sugar ANA derivatives over the course of 12 months (data not shown).Such long-term stability of derivatives is essential for isotope determination.Each amino sugar, including MurA, yielded a single, well-resolved chromatographic peak during GC analysis, with either an Rxi-5ms or an Optima 17MS column, when the injected amount was less than 120 ng per amino sugar.We therefore concluded that the ANA method was superior to the AA approach for GC-based analysis of amino sugars, and subsequently implemented this step in our method optimization.Like Glaser and Gross (2005), we also noted that the elu-tion order of ANA derivatives differed between the two GC columns (Fig. 1a, b).The availability of two alternative separations was advantageous for determining the δ 13 C values of the minor components, which co-eluted with major peaks when using the Optima 17MS column, but were separable when using the Rxi-5ms column for environmental samples.However, additional samples with different matrix effects should be tested to determine the column type with the best peak separation.

Performance of the preparative HPLC procedure
MurA was the primary target compound when developing the preparative HPLC protocol, because it serves as a specific biomarker for bacteria, but is typically low in abundance in subseafloor sediments (Fig. 1c).ManN, another minor amino sugar, could not be isolated from GlcN and GalN using this approach.Nevertheless, preparative HPLC was beneficial for isotopic analysis of ManN, due to partial removal of the sample matrix, which allowed us to concentrate the sample into a small volume for GC injection.We observed minimal losses of GalN and MurA after preparative HPLC, but recoveries of GlcN and ManN were only 87 and 71 %, respectively (Table 3).The F2 and waste fractions contained less than 1 % of the original GlcN and ManN, suggesting that GlcN and ManN were either lost via preferential adsorption in the column, or that derivatization of these two amino sugars was hampered by the presence of ammonium formate originating from the HPLC eluents.

Summary of the protocol and isotope data assessment
Figure 2 summarizes the optimized pretreatment procedure.
In brief, amino sugars are released from marine sediments by hot HCl.The hydrolysate then undergoes a series of purification steps, including filtration with combusted glass fiber filters, evaporation under vacuum, neutralization with a KOH solution, and desalting with a ENVI-Carb ™ Plus SPE cartridge.After freeze-drying overnight, amino sugars in the extract are transformed to ANA derivatives, and further purified by a self-packed silica gel column.The recoveries of the whole procedure, estimated from parallel tests (n = 3) using a mixture of amino sugar standards, were 111.6 ± 1.9 %, 95.0 ± 9.2 %, 47.0 ± 3.8 % and 102.5 ± 2.6 % for GlcN, GalN, ManN, and MurA, respectively.The lower recovery of ManN may result from successive preferential losses during each step and is probably related to its steric structure, but the underlying mechanism is not yet fully understood.
It is established that derivatization with acetic anhydride to ANA derivatives is associated with an isotope effect (Glaser and Gross, 2005;Decock et al., 2009); this effect can be corrected with the use of a derivatization standard such as 3-O-Me-Glc (e.g., Lin et al., 2010).After processing authentic amino sugar standards using our proposed preparation procedure (Fig. 2), and correcting for the isotope effect, the δ 13 C values measured by GC-IRMS (δ 13 C AS, GC ) were in good agreement with those obtained by EA-IRMS (δ 13 C AS, EA ; Table 4), indicating negligible isotopic fractionation during the other sample pretreatment steps.The standard deviations of repeated injections for GC-IRMS were less than 1 ‰, but the total errors were up to 1.4 ‰, i.e., about 1 ‰ greater than those reported for the HPLC-based method developed for soils (Bodé et al., 2009).The total errors, which are derived from the addition of C atoms and fractionation during derivatization, impose constraints on the isotopic resolving power of our method, and should be taken into account during data interpretation.Despite this disadvantage, the much lower requirement of C (see below) renders the GC-based method an attractive avenue for the analysis of amino sugars in trace amounts.
Table 4. δ 13 C values of amino sugar standards analyzed by EA-IRMS (δ 13 C AS, EA ) or according to the pretreatment procedure described in Fig. 2 prior to GC-IRMS analysis (δ 13 C AS, GC ).s EA and s GC stand for standard deviations of EA-or GC-IRMS measurements, respectively.s Total is the total analytical error of δ 13 C values of individual amino sugars calculated according to the error propagation (n = 3).For samples that exhibited low signals of MurA and/or ManN due to the high background generated by the sedimentary matrix or a high abundance in the adjacent peak of GalN (i.e., the Peru Margin sample), amino sugar extracts were separated into two fractions (i.e., basic amino sugar fraction and MurA fraction) via the preparative HPLC procedure described above (Fig. 2), which could be further concentrated for GC-IRMS analysis.Tests using authentic amino sugar standards confirmed that this additional step did not introduce significant isotopic fractionation, as the δ 13 C values (Table 3) deviated from those obtained by EA-IRMS (Table 4) by less than 0.6 ‰.

EA-IRMS
The δ 13 C values of ANA derivatives of ManN and MurA were consistent with the EA-IRMS reference values within the range of 20 to 140 ng per injection, i.e., equivalent to ∼ 8-56 ng of amino sugar carbon (Fig. 3), regardless of the pres- ence of a sedimentary matrix; we note that these two amino sugars are usually expected to have the lowest concentration in environmental samples (e.g., Guggenberger et al., 1999;Niggemann and Schubert, 2006;Carsten et al., 2012).At injected quantities below 20 ng, the peak height of the respective compound was usually below 100 mV, which precluded proper determination of the isotopic composition.The minimum threshold of 20 ng of amino sugar on GC-IRMS is one order of magnitude lower than the values reported for HPLC-IRMS (Bodé et al., 2009), and enables the isotopic analysis of amino sugars at trace levels, as required by their low abundances in subseafloor sediment.

Abundance and isotopic composition of individual amino sugars in marine sediment
Figure 4 summarizes the concentrations and δ 13 C values of individual amino sugars from four marine sediment samples determined according to the protocol described in Fig. 2. GlcN, accounting for 43.7-63.0% of the total amino sugars, was the most abundant amino sugar in all the investigated samples, followed by GalN (29.0-39.7 %) and ManN (4.4-12.4%), while MurA was always less than 5 %.The δ 13 C values of GlcN and GalN were similar, and fell between δ 13 C TOC and δ 13 C DIC in sediment collected from the Black Sea, the Marmara Sea, and the Peru Margin.In the cold seep sample from the Pakistan Margin, GalN was depleted in 13 C by 9 ‰ compared to GlcN.This sample was collected from a microbial mat composed of anaerobic methanotrophic archaea (ANME) group 2 and associated bacterial partners (Yoshinaga et al., 2012).These organisms were likely the predominant sources of TOC in the seep sample, which also exhibited the lowest δ 13 C TOC (−40.0 ‰).ManN exhibited δ 13 C values ranging from −31.5 to −10.9 ‰, and showed no consistent trend relative to δ 13 C TOC and δ 13 C DIC , or to the other amino sugars.MurA spanned the widest range in δ 13 C (from −46.0 to −13.6 ‰), and was more depleted than δ 13 C TOC in the Pakistan Margin, Black Sea, and Peru Margin samples.
GlcN and GalN are typically the major amino sugars found in sediment and preserved in the form of peptidoglycan, lipopolysaccharides, and pseudopeptidoglycan from prokaryotes (Schleifer and Kandler, 1972;Kandler et al., 1998;Madigan and Martinko, 2005).Moreover, GlcN has also been detected in most fungal cell walls and invertebrate exoskeletons as chitin (Müller et al., 1986;Liang et al., 2007;Davis et al., 2009).It is generally believed that the basic amino sugars persist after cell death, and are stable in soils (Glaser et al., 2004).Assuming a similar behavior in marine sediment, these amino sugars are likely derived from a diverse assemblage of pelagic and sedimentary organisms, and have been preserved in the sediment.The lack of a significant discrepancy between the δ 13 C values of GlcN and GalN (usually < 3.5 ‰) implies a common, likely allochthonous source in the Black Sea, Marmara Sea, and Peru Margin sediments.The distinct isotopic compositions of GlcN (−28.fraction of these amino sugars from autochthonous microbes that utilize 13 C-depleted carbon for biosynthesis, as is known from other biomolecules at similar sites (e.g., Lin et al., 2010;Schubotz et al., 2011).Further application of this method to define the heterogeneity of δ 13 C signatures of individual hexosamines in the downcore sediment profiles will help additionally to examine microbial activity in specific sediment horizons.
In contrast to the basic amino sugars with their diverse origins, MurA is a diagnostic bacterial biomarker, because it is exclusively derived from peptidoglycan in bacterial cell walls (Schleifer and Kandler, 1972), and is presumed to degrade rapidly after cell death (Moriarty, 1977).The relatively high abundance of MurA (4.2 % of total amino sugars) and its strong 13 C depletion (δ 13 C of −46.0 ‰) in the Pakistan Margin sample are consistent with our expectations of cold seep sediment that hosts microbial biomass fueled by methane (Hinrichs et al., 1999).Likewise, the 13 C-depleted MurA in the Black Sea sample is probably an indigenous signal resulting from bacteria that utilize relatively 13 C-depleted carbon at this depth (1.57mbsf), which is close to the sulfatemethane transition zone, based on this site's methane profile (Zabel et al., 2013).Although compound-specific isotopic analysis of amino sugars is more demanding than that of membrane lipids, these observations of putative indigenous bacterial signals in the form of MurA make this compound a valuable biomarker target for the isotopic analysis of microbial biomass in the deep marine biosphere, which so far has been based on lipids and intact cells, and biased towards detection of signals of the Archaea (cf.Biddle et al., 2006).The combined analysis of hydrolysable sugar-derived biomarkers with lipid biomarkers, with each group derived from different biosynthetic pathways and representing different preservation mechanisms, holds great potential for providing a more balanced view for the study of microbial life in subseafloor sediments.

Conclusion
Methods for the carbon isotopic analysis of amino sugars have been developed for soils, but not for marine sediments, where amino sugar concentrations tend to be lower.We tested various steps in the workflow of amino sugar analysis in order to establish a robust protocol for the stable carbon isotopic analysis of amino sugars in marine sediments.A combination of the most effective steps, including a new SPE protocol for the post-hydrolysis clean-up and a new step for enrichment of amino sugars via preparative HPLC, resulted in a protocol optimized for GC-based isotopic analysis of amino sugars at trace levels (the limit of detection is 20 ng, equivalent to ∼ 8 ng of amino sugar C).The conversion of amino sugars to ANA derivatives results in GC-amenable analytes that have long-term stability and are suitable for isotopic analysis.Moreover, use of the protocol did not introduce sig-nificant isotopic fractionation during sample preparation, except for the derivatization step, which can be accounted for with a derivatization standard.Applying this protocol, we determined for the first time the carbon isotopic composition of amino sugars in marine sediment samples.The stable carbon isotopic values of basic amino sugars indicated a major contribution from organic detritus, whereas MurA was more sensitive to an indigenous and active bacterial community.This method thus enables investigation of the stable carbon isotopic compositions of amino sugars, and provides a valuable addition to the lipid-based characterization of microbial metabolism in the deep marine biosphere.

Figure 3 .
Figure 3. Stable carbon isotopic analysis of ManN (a) and MurA (b) in a range from 20 to 140 ng.Open circles represent the δ 13 C AS, GC values of pure standards.Solid squares designate the δ 13 C AS, GC values of the same standards spiked into sediment extract that did not contain the corresponding amino sugar.The dashed lines are the isotopic values of ManN and MurA determined by EA-IRMS, respectively (n = 3).The error bars represent the total analytical errors.

Figure 4 .
Figure 4. Concentrations and stable carbon isotopic values of amino sugars released from selected marine sediment samples.The depositional settings of the sediment samples are indicated in parentheses.Error bars indicate the total analytical errors (s Total ) of δ 13 C values of individual amino sugars calculated according to error propagation.The isotopic values of dissolved inorganic carbon (DIC) and total organic carbon (TOC) are also plotted for comparison.
2 ‰) and GalN (−37.2 ‰) in the Pakistan Margin cold seep sample can be explained best by an increasing www

2014 2.5 Preparation of amino sugar-enriched fractions by preparative HPLC
For samples containing low concentrations of amino sugars and/or elevated sedimentary matrices, amino sugar extracts were separated and enriched via preparative HPLC.A mixture of amino sugar standards and desalted hydrolysate of the Wadden Sea sediment sample were employed to develop the preparation protocol.The sample was re-dissolved in acetonitrile (ACN) / H 2 O (7 : 3, v : v) prior to injection.The flow rate was 1 ml min −1 , and the eluent gradient ramped steadily from 100 % eluent A to 100 % eluent B over 15 min, then held at 100 % eluent B for 15 min, followed by 10 min reequilibration with 100 % eluent A. Eluent A was composed of ACN / H 2 O / NH 4 OH / formic acid (90 : 10 : 0.1 : 0.2, v : v : v : v); eluent B was ACN / H 2 O / NH 4 OH / formic acid (30 : 70 www.biogeosciences.net/11/4869/2014/Biogeosciences, 11, 4869-4880,

Table 2 .
Recoveries and standard deviations of HCl-treated amino sugar standards after different neutralization and desalting procedures (n = 3).

Table 3 .
Recoveries and δ 13 C values of amino sugar standards after separation by preparative HPLC.Standard deviations of GC-IRMS measurements (s GC ) are reported for individual amino sugars, and the total analytical error of δ 13 C values (s Total ) was calculated according to error propagation (n = 3).
Schematic of the optimized procedure for isotope analysis of amino sugars.Each step has been validated and/or optimized in this study.