Studies of organic nitrogen (N) cycling and uptake by plants have focused on protein amino acids, but the soil solution includes organic N compounds from many other compound classes.
The two aims of this study were to characterize the 30–50 most abundant molecules of small (< 250 Da), nonpeptide organic N in the soil solution from six soils, and to determine if two ecologically disparate species (nonmycorrhizal Banksia oblongifolia and mycorrhizal Triticum aestivum) have the ability to take up intact molecules of three quaternary ammonium compounds (betaine, carnitine and acetyl-carnitine).
Protein amino acids were dominant components of the pool of small nonpeptide organic N in all soils. The most abundant other compound classes were quaternary ammonium compounds (1–28% of nonpeptide small organic N) and nonprotein amino acids (3–19% of nonpeptide small organic N). B. oblongifolia and T. aestivum took up intact quaternary ammonium compounds from dilute hydroponic solution, while T. aestivum growing in field soil took up intact quaternary ammonium compounds injected into soil.
Results of this study show that the pool of organic N in soil is more diverse and plants have an even broader palate than is suggested by most of the literature on organic N.
Over recent decades, many experiments have established that plants have the ability to take up intact amino acids from soil (Chapin et al., 1993; Jones & Darrah, 1993; Warren, 2006; Näsholm et al., 2009). The most common method for estimating uptake is to supply plants or soil with amino acids in which C or N are isotope-labelled (e.g. 13C and 15N, or 14C) and subsequently estimate uptake from uptake of isotope label (Näsholm et al., 1998; Näsholm & Persson, 2001; Warren, 2009b) or intact isotope-labelled amino acid molecules (Sauheitl et al., 2009a; Warren, 2012). A key aspect of studies using isotope labeling is that choices need to be made regarding which small organic N molecules are tested and which are not. In principle, the choice ought to be guided by knowledge of which forms of organic N occur in soil.
Experiments from the late 19th century until the 1970s examined uptake of many different forms of organic N (e.g. protein amino acids, nonprotein amino acids, nucleic acids, ureides, quaternary ammonium compounds, purine and pyrimidine bases) (Hutchinson & Miller, 1912; Paungfoo-Lonhienne et al., 2012), whereas in the last 20 yr the literature has focused almost exclusively on plant uptake of protein amino acids (Chapin et al., 1993; Jones & Darrah, 1993; Warren, 2006). For example, a recent review stated: ‘The concept of plant organic N nutrition relies, to a large degree, on studies of amino acids. Thus, amino acid N is in many cases used as a synonym for organic N’(Näsholm et al., 2009). Soil and/or the soil solution was long ago recognized as containing many organic N compounds in addition to protein amino acids (e.g. amino sugars, purine and pyrimidine bases, nucleic acids; e.g. see historical review: Paungfoo-Lonhienne et al., 2012), but the recent focus on protein amino acids arose because studies of the soil solution suggested amino acids were ‘the major constituent of low molecular weight dissolved organic N’ (Jones et al., 2005).
The apparent dominance of the soil solution by protein amino acids may reflect the fact that studies used targeted analytical approaches (sensu Patti et al., 2012) focused on protein amino acids plus a few common nonprotein amino acids (e.g. GABA, citrulline, ornithine) (Kielland, 1995; Turnbull et al., 1996; Andersson & Berggren, 2005; Warren, 2008; Jämtgård et al., 2010; Farrell et al., 2011a; Inselsbacher et al., 2011). Indeed, a recent broader exploration of the pool of small organic N found that, in addition to protein amino acids, the soil solution from a subalpine soil contained quaternary ammonium compounds, nonprotein amino acids, heterocyclic compounds derived from aromatic amino acids, amines, and sugar amines (Warren, 2013). Some of the single most abundant molecules were quaternary ammonium compounds, and the pool of quaternary ammonium compounds was c. 25% of the size of the pool of protein amino acids (Warren, 2013).
It is possible that quaternary ammonium compounds could be a source of N for plants, given that they can be abundant in the soil solution and plants possess transporters for quaternary ammonium compounds (Breitkreuz et al., 1999). However, at the time of writing, there was only one report of higher plants taking up quaternary ammonium compounds from soil (Audley & Tan, 1968), and with the exception of one site (Warren, 2013), little is known about the relative abundance of quaternary ammonium compounds and other compound classes that comprise the pool of small organic N. The aims of this study were to explore the abundance of quaternary ammonium compounds in a range of soils, and determine if plants can take up quaternary ammonium compounds. Capillary electrophoresis-mass spectrometry (CE-MS) was used for untargeted profiling of nonpeptide small organic N molecules (including quaternary ammonium compounds) in the soil solution of six soils. Uptake of three isotope-labelled quaternary ammonium compounds (betaine, carnitine and acetyl-carnitine) was contrasted with uptake of three protein amino acids (glycine, alanine and arginine). Uptake experiments used two ecologically divergent species: mycorrhizal wheat (Triticum aestivum L., Poaceae) a fast-growing annual grass species; and nonmycorrhizal Banksia (Banksia oblongifolia Cav., Proteaceae) a slow-growing perennial shrub that commonly occurs in nutrient-poor coastal heathlands of eastern Australia. To allow plants to express a preference for different N forms without complications from microbial competition and diffusion in soil, uptake from solution was examined with seedlings of wheat and Banksia in hydroponic solutions containing a mixture of the six forms of N at a low, field-relevant concentration of 10 μmol N l−1 each (Svennerstam et al., 2011). To examine intact uptake from soil, the six isotope-labelled N forms were injected individually into soil containing wheat plants and uptake was determined after 1 and 24 h. Intact uptake of N forms was determined from the amounts of isotope-labelled molecule that were measured within plant tissues (Persson & Nasholm, 2001; Sauheitl et al., 2009a; Warren, 2012).
Materials and Methods
Methanol, acetonitrile and formic acid were LC/MS (Optima) grade from Fisher Chemical (Scoresby, Victoria, Australia). Ammonium formate (Acros Organics), ammonium hydroxide (28–30% NH3, Sigma) and iodomethane (Sigma) were analytical grade. Sixty-three standards of small N-containing molecules were prepared from their free acids or salts purchased from Sigma. α-N-methyl-histidine, α-N,N-dimethyl-histidine and N-methyl-proline were from Chem-Impex (Chem-Impex International, Wood Dale, IL, USA). All standards of chiral amino acids were L-enantiomers.
Synthesis of hercynine
In soils from two sites, one of the most abundant peaks did not match any available chemical standards, but had a tandem mass spectrum very similar to a molecule putatively identified (via de novo interpretation) as hercynine (Nα,Nα,Nα-trimethyl-L-histidine) (Baran et al., 2010). Hercynine is not commercially available, and thus, to confirm identification, hercynine was synthesized according to Reinhold et al. (1968). In brief, 0.5 g of α-N,N-dimethyl-L-histidine was dissolved in 13.7 ml of LC/MS grade methanol, 1.15 ml of NH4OH (28–30% NH3) was added to raise the pH to 9, and then 0.27 ml of iodomethane was added. The mixture was left to stand overnight at room temperature, then evaporated under a stream of dry nitrogen gas, dissolved in minimal volume of water and purified by anion exchange.
Soil collection and extraction of soil solution
Measurements of small organic N in the soil solution were made on six contrasting soils: a subalpine grassland with native perennial grasses and annual herbs; a nutrient-poor coastal heathland dominated by slow-growing and stress-tolerant species from the Proteaceae, Myrtaceae and Epacridaceae; a crop of triticale (x Triticosecale); a temperate/subtropical grassland dominated by Themeda triandra; and two plantations with 1-yr-old seedlings of the fast-growing perennial tree Eucalyptus globulus on contrasting geologies (Table 1). Note that the pool of small organic N in the subalpine grassland has been reported previously (Warren, 2013), but samples were reanalysed for the present study. At each site, three to five replicate soil samples were collected from 0 to 15 cm. Soil samples were c. 200 ml volume each, while at the heathland and triticale site, a larger volume of soil was collected to enable growth of Banksia and wheat seedlings. Soils were collected between early autumn and spring. In all cases, soils were sufficiently wet to be able to extract soil solution, but all soils, with the exception of the two E. globulus plantations, had been exposed to varying degrees of water stress during the preceding 6 months. Samples were kept at 4°C during transport and storage. Within 2 d of collection, samples of the soil solution < 3 kDa were extracted from soil by centrifugation (20 ml volume, 3 kDa molecular weight cutoff, Vivaspin 20, Sartorius, Goettingen, Germany), as described previously (Warren, 2013). Blanks comprising ultrapure water were carried through the same procedure.
Table 1. Location and characteristics of the seven soils examined in this study
pH was determined with fresh soil in 1 : 5 ratio with ultrapure water. Pools of organic and inorganic N are for soil solution extracted centrifugally from soil collected (0–15 cm) from six field sites. Data are also shown for soil in which wheat seedlings had grown for 8 wk. Data are dissolved organic N < 3 kDa (DON) determined by persulfate digestion. Standard colourimetric methods were used to measure nitrate and ammonium. Small (<250 Da) nonpeptide organic N is the sum of molecules positively identified by capillary electrophoresis-mass spectrometry (CE-MS). Data area means (SE) of three to five replicates. asl, altitude above sea level.
Soil subsequently used for growing Banksia oblongifolia seedlings.
Soil subsequently used for growing Triticum aestivum.
World reference base
36.1S,148.3E, 1500 m asl
34.1S, 151.1E, 25 m asl
34.0S, 150.6E, 75 m asl
34.0S, 150.6E, 75 m asl
41.2S, 147.8E, 500 m asl
41.2S, 147.3E, 200 m asl
From triticale site
Precipitation (mm yr−1)
Potential evaporation (mm yr−1)
30 cm of dark, highly organic A1 horizon overlying weakly developed B horizon
10–20 cm of coarse quartz sand overlying sandstone bedrock
0.5–1.0 m of sandy clay loam, overlying clay
0.5–1.0 m of clay loam, overlying clay
Dark coarse sandy loam grading into dark yellowish sandy clays
Brown loam overlying clay
Shale and alluvium
Shale and alluvium
Siltstone and mudstone
Native perennial grasses and annual herbs
Coastal heathland. Predominantly Proteaceae and Mytraceae spp.
Native perennial grassland dominated by Themeda triandra
Plantation of 1-yr-old Eucalyptus globulus seedlings
Plantation of 1-yr-old Eucalyptus globulus seedlings
NO3 (μmol N l−1)
NH4 (μmol N l−1)
DON < 3 kDa (μmol N l−1)
Small nonpeptide organic N (μmol N l−1)
Small nonpeptide organic N (% of DON < 3 kDa)
Seeds of B. oblongifolia were germinated and grown in heathland soil; seeds of Triticum aestivum were germinated and grown in soil collected from the triticale crop (Table 1). Plants were grown one-per-pot in 210 ml plastic tubes (50 mm square × 125 mm high) in a fully sunlit glasshouse at the University of Sydney (Camperdown, New South Wales, Australia). Banksia was grown for 1 yr before experimentation, whereas wheat was grown for 8–10 wk. Seedlings were watered every 1–4 d to near field capacity.
N uptake from mixed solutions
Uptake of N was determined from intact roots placed in isotope-labelled solutions for 2 h, essentially as described previously (Warren, 2009a). Measurements were made on four replicate 1-yr-old Banksia seedlings (mean fresh weight = 4.5 g) and four replicate 10-wk-old wheat seedlings (mean FW = 3.5 g) fed isotope-labelled N or water (controls). N uptake was determined from uptake of uniformly (98–99 atom%) 13C- and 15N-labelled glycine, l-alanine and l-arginine (Isotec inc., Miamisburg, OH, USA) and deuterium-labelled (99 atom% D) l-carnitine-d3 (methyl-d3), betaine-d9 and acetyl-l-carnitine-d3 (methyl-d3) (CDN isotopes, Pointe-Claire, Quebec, Canada). Plants were simultaneously offered all six isotope-labelled N sources at 10 μmol N l−1 each so as to provide a ‘choice’ of different N sources. In addition to N, uptake solutions contained 0.01 M CaCl2 for membrane stability. To measure uptake, seedlings were removed from soil, and roots were gently washed to remove adhering soil. The entire root system was immersed in the uptake solution and seedlings were kept in their natural (i.e. upright) position. During uptake experiments, air temperature was 20–25°C and photosynthetically active radiation (PAR) was 750–1500 μmol m−2 s−1. After 2 h of incubation, the root system was cut off. Shoots were frozen in liquid nitrogen (N); roots were rinsed with 50 mmol l−1 KCl and then ultrapure water before being frozen in liquid N. N uptake was calculated from the amounts of intact isotope-labelled molecules detected within roots and shoots (see details later). A pilot study using seedlings pretreated with the protonophore carbonyl cyanide m-chlorophenylhydrazone established that passive uptake was below detection limits. Hence, uptake reported here represents true uptake rather than passive uptake that might occur, for example, following damage of roots or as a result of adsorption of isotope label to the root surface.
Uptake of individual N forms from soil
Experiments on uptake of N from soil were conducted with wheat plants that were 8 wk old (mean FW = 3.3 g) and were in pots containing c. 200 ml of soil from the triticale field site. There were three replicate seedlings for each of the six N forms (Gly, Ala, Arg, betaine, carnitine, acetyl-carnitine) and two harvests (1 and 24 h) plus an additional six seedlings that received water and were harvested after 1 h (controls). A comparatively large amount of N was added so as to avoid complications such as differential pool dilution resulting from the differing native concentrations of the six N forms in soil. Pots received 5 ml of 1.2 mM N (5.95 μmol) as one of the six isotope-labelled N forms or water. Seedlings were not watered for 1–2 d before solutions were injected into the soil. This let the soil dry down to 10 ml below field capacity, and solutions could be injected without any loss resulting from drainage. Labelled substrates were uniformly (98–99 atom%) 13C- and 15N-labelled glycine, l-alanine and l-arginine (Isotec Inc.) and deuterium-labelled (99 atom% D) l-carnitine-d3 (methyl-d3), betaine-d9 and acetyl-l-carnitine-d3 (methyl-d3)(CDN isotopes). Solutions were injected into the soil between 10:00 and 12:00 h (noon) when the air temperature was 20–25°C and PAR was 750–1500 μmol m−2 s−1. Five mililitres of solution were injected into the soil in five 1 ml aliquots using a four-sideport Cass needle (18G × 150 mm long, Victor-G & Company, Kanpur, India). Seedlings were harvested 1 and 24 h after addition of N or water. Shoots were promptly frozen in liquid N, whereas roots were first rinsed with 50 mmol l−1 KCl and then ultrapure water before being frozen in liquid N. N uptake was calculated from the amounts of intact isotope-labelled molecules detected within roots and shoots (see details later). Samples of the soil solution were extracted from soil of the control plants as described earlier and then analysed for small organic N molecules as for field-collected soils.
Extraction of metabolites from leaf and root samples
A c. 400 mg subsample of frozen leaf or root material was placed into a grinding jar precooled in liquid N and then ground to a fine powder with a mixer mill (30 s at 25 Hz with a 5 mm stainless steel ball bearing) (TissueLyser; Qiagen, Doncaster, Victoria, Australia). One hundred to 200 mg of frozen and ground material was weighed into a 2 ml microfuge tube, 1.0 ml of ice-cold methanol was added and the tube was mixed for 2 min (end-to-end shaking at 10 Hz) and then centrifuged (5 min at 16 837 g). A 500 μl aliquot of supernatant was fractionated and deproteinized by adding 300 μl of chloroform, 500 μl of water and centrifuging (5 min at 16 837 g). The aqueous supernatant was kept at 4°C for no longer than 24 h before being used for quantification of amino acids and quaternary ammonium compounds.
Capillary electrophoresis-mass spectrometry was used for profiling of small organic N in the soil solution and for quantification of amino acids and quaternary ammonium compounds in leaf and root aqueous extracts, essentially as described previously (Warren, 2013). CE-MS was performed with a capillary electrophoresis system (P/ACE MDQ, Beckman-Coulter, Fullerton, CA, USA) equipped with a bare fused silica capillary (50 μm i.d. × 100 cm long) interfaced via a coaxial sheath-flow sprayer (G1607A, Agilent, Waldbronn, Germany) to an ion trap mass spectrometer (AmaZon SL, Bruker Daltonics, Bremen, Germany). Sheath liquid of 50% (v/v) methanol with 0.1% (v/v) formic acid was delivered at 4 μl min−1 by a syringe pump (NE-1002X Microfluidics Syringe Pump, New Era Pump Systems, Farmingdale, NY, USA) driving a 10 ml PTFE-tipped gas-tight syringe (SGE, Ringwood, Victoria, Australia). Ion source parameters were as described previously (Warren, 2013).
Profiling of small organic N in the soil solution
Samples of soil solution were concentrated by evaporating 1.0–2.0 ml of soil solution to c. 25 μl under reduced pressure (Vacufuge; Eppendorf, Hamburg, Germany) and then adding 25 μl of 200 mM ammonium formate (pH 10) in 50% (v/v) acetonitrile, and 4.0 μl of an internal standard (10 μg ml−1 methionine sulfone). Samples were injected at 3 psi for 30 s and separated with an electrolyte of 2 M formic acid with 20% (v/v) methanol under 30 kV positive polarity. The mass spectrometer was set to a scan a range of 50–250 m/z in enhanced resolution mode (8100 u s−1) and data were recorded as the average of five scans. Between runs, the capillary was flushed with electrolyte for 10 min (50 psi).
The general procedures for identification and quantification have been described previously (Warren, 2013), so only brief details are given here. For metabolite identification, MS2 and in some cases MS3 spectra were obtained by reanalysing standards and samples under the same electrophoretic conditions. MS3 spectra were obtained only for those molecules in which uninformative losses of water or ammonia dominated MS2 spectra. ‘Known’ molecules were identified based on comparison of migration times, [M+H]+, MS2 and MS3 (where available) with 63 authentic standards run under the same conditions on the same instrument. ‘Unknowns’ were then searched against a CE-MS library of [M+H]+ and migration time for 364 cationic organic molecules (Baran et al., 2006) and online MS/MS databases Metlin (Smith et al., 2005), HMDB (Wishart et al., 2007) and MassBank (Horai et al., 2010). Mass spectrometer parameters (scan range, fragmentation parameters etc.) were optimized for small nonpeptides, and thus the method was incapable of unbiased profiling or ID of peptides. For these reasons, data are reported for nonpeptides only.
Quantification of isotope-labelled molecules in leaf and root extracts
Only very small amounts of isotope-labelled molecules were taken up from solutions, and thus it was necessary to maximize detection limits while filtering out background signal from endogenous molecules present at three to six orders of magnitude greater concentrations. High sensitivity and specificity of detection were achieved by injecting a large amount of sample (typically injection = 4 psi × 90 s), separating by CE as described previously, and then detecting isotope-labelled molecules by ion trap mass spectrometry with pseudo-multiple reaction monitoring (MRM). Standard curves were constructed between 1 nM and 10 μM (arginine, betaine, carnitine, acetyl-carnitine) or 10 nM and 100 μM (glycine and alanine).
In the experiment on N uptake from soil, leaf and root extracts contained two orders of magnitude more isotope-labelled N than in the experiment on solutions. Hence, adequate sensitivity and specificity were achieved by injecting samples at 3 psi for 90 s, and then separating and detecting by CE-MS as described for the soil solution (see ‘Profiling of small organic N in the soil solution’ above). Standard curves were constructed between 100 nM and 100 μM.
A potential complication with the use of deuterium-labelled molecules is that deuterium can exchange with hydrogen in the solvent. Preliminary experiments established that D/H exchange was negligible for the samples and measurements in this experiment. Extracts of roots and leaves from unlabelled Banksia and wheat were spiked with 10 μM of deuterium-labelled betaine, carnitine and acetyl-carnitine. Extracts were analysed immediately and then stored for 4 wk at 4°C. After 4 wk storage, there was no evidence of a decrease in abundance of deuterium-labelled molecules or appearance of nonlabelled isotopologues.
Small organic N molecules in the soil solution
There were large differences among soils in absolute and relative amounts of dissolved organic N (< 3 kDa), nitrate and ammonium (Table 1). In all soils, the pool of dissolved organic N (< 3 kDa) was larger than pools of inorganic N (i.e. sum of nitrate or ammonium, Table 1). In E. globulus 1, the pool of DON < 3 kDa was five to six times smaller than for any other site, and a modest pool of nitrate comprised most of the N. The pool of nitrate was larger than ammonium in four soils (Themeda grassland, E. globulus 1, E. globulus 2, wheat soil), whereas in the remaining three soils nitrate was smaller than ammonium (subalpine, heathland, triticale crop).
Sixty-five nonpeptide small organic N molecules from 12 compound classes were identified and quantified by CE-MS (Supporting Information, Table S1). Between 31 and 52 molecules were identified in individual soil samples. All samples also contained at least another 20 molecules that could not be identified, but, with only one exception, unidentified molecules had small peak areas and would have made a minor contribution to small organic N. The only highly abundant unidentified molecule was in the heathland soil where a molecule with MH+ = 194.1 had one of the largest peak areas and was probably one of the 10 most abundant molecules (assuming the unidentified molecule had a typical response factor).
The pool of small nonpeptide organic N generally accounted for 0.4–6% of dissolved organic N < 3 kDa and was 14–83% of the size of the pool of inorganic N. Arranging small organic N from the most to the least abundant molecule revealed that concentrations of individual molecules spanned approximately three orders of magnitude (Fig. 1). Owing to the (more or less) logarithmic decrease of concentration with increasing molecule number, 93–99% of the pool of small nonpeptide organic N was contained within the 25 most abundant molecules.
In addition to differences in size of the total pool of small nonpeptide organic N (Table 1), there were differences in the compound classes making up the pool. Of the 12 compound classes detected in the soil solution (Table S1), the quantitatively most important were protein amino acids (52–88% of nonpeptide small organic N), quaternary ammonium compounds (1–28% of nonpeptide small organic N) and nonprotein amino acids (3–19% of nonpeptide small organic N) (Fig. 2). Protein amino acids were a dominant feature of all soils, with at least 19 protein amino acids detected in all samples. Twelve quaternary ammonium compounds were identified and quantified, but the number detected at any one site varied between a maximum of nine and a minimum of three. Thirteen nonprotein amino acids were identified, with three being detected at most sites (GABA, citrulline, β-alanine).
There was great diversity among sites in the most abundant small organic molecules in the soil solution (Fig. 3). There were several molecules that were among the 10 most abundant molecules at some sites, yet were either absent or minor components of small organic N at other sites (e.g. hercynine, ergothioneine, theanine, triethanolamine, urocanic acid). The triticale crop was unusual in the sense that all of the 10 most abundant molecules were protein amino acids. At all other sites, the 10 most abundant molecules included other compound classes. Some of the more noteworthy molecules that are not protein amino acids included the following:
Theanine (N5-ethyl-glutamine) was the second most abundant molecule at the heathland site, but was absent or a minor component of small organic N in other soils.
Hercynine was the second most abundant molecule in the Themeda grassland, the third most abundant molecule in the subalpine grassland, and among the 20 most abundant molecules for three other soils (heathland, E. globulus 1, wheat).
Quaternary ammonium compounds that are putative osmolytes (betaine, proline betaine, ectoine, hydroxyectoine) were among the 10 most abundant molecules in soils from Themeda grassland and wheat seedlings, and among the 20 or 25 most abundant molecules for all other soils, except for E. globulus 2. Carnitine and acetyl-carnitine were among the 10 most abundant molecules in the subalpine soil, and were present at lower concentrations in most other soils
Uptake of a mixture of N forms from solution by wheat and Banksia
After 2 h exposure to a mixed solution of six N forms at a concentration of 10 μmol N l−1 each, roots of Banksia and wheat contained detectable amounts of all six N forms (Table 2a). Substantial amounts of isotope-labelled quaternary ammonium compounds were detected in leaves, whereas little or no isotope-labelled glycine, alanine or arginine was detected in leaves.
Table 2. Uptake of isotope-labelled molecules by seedlings
U-13C,15N Gly (nmol g−1 FW)
U-13C,15N Ala (nmol g−1 FW)
U-13C,15N Arg (nmol g−1 FW)
Betaine-d9 (nmol g−1 FW)
Carnitine-d3 (nmol g−1 FW)
Acetyl-carnitine-d3 (nmol g−1 FW)
Data are leaf and root concentrations of isotope-labelled molecules in (a) Banksia oblongifolia (Banksia) and Triticum aestivum (wheat) placed for 2 h in a solution containing a mixture of six isotope-labelled forms of organic N at 10 μmol N l−1 each; and (b) Triticum aestivum seedlings grown in field soil that was injected with 5.95 μmol of one of the forms of N and then harvested 1 or 24 h later. Data are means (SE) of four replicates (a) or three replicates (b). nd, not detected, below detection limits.
a) N uptake from solutions
b) N uptake by wheat from soil
Uptake of single forms of N from soil by wheat
One hour after injection of isotope-labelled N into soil, roots of wheat seedlings contained detectable amounts of all six N forms (Table 2b). Amounts of intact labelled molecules detected in roots after 1 and 24 h were largest for betaine and smallest for arginine. Isotope-labelled quaternary ammonium compounds were detected in leaves, whereas no isotope-labelled glycine, alanine or arginine was detected in leaves at either harvest.
To place in context the amounts of N taken up from soil in 1 and 24 h (Table 2b), endogenous concentrations of unlabelled N were measured in control seedlings that received water (Table 3). Endogenous amounts of carnitine and acetyl-carnitine in leaves and roots were small (0.4–5.1 nmol g−1 FW) in comparison to amounts of betaine, glycine, alanine and arginine (179–2370 nmol g−1 FW). Owing to these massive differences in endogenous concentrations, the amount of isotope-labelled molecules present in tissues at the 24 h harvest (Table 2b) varied from < 50% of the size of the endogenous pool for the three amino acids and betaine to 10–70 times larger than the pool of endogenous carnitine and acetyl-carnitine.
Table 3. Leaf and root concentrations of six unlabelled nitrogen (N) forms in control seedlings of wheat (Triticum aestivum)
Gly (nmol g−1 FW)
Ala (nmol g−1 FW)
Arg (nmol g−1 FW)
Betaine (nmol g−1 FW)
Carnitine (nmol g−1 FW)
Acetyl-carnitine- (nmol g−1 FW)
Data are means (SE) of three replicates.
Small organic N molecules in the soil solution
A key finding of this study is that the pool of small, nonpeptide organic N in the soil solution is chemically diverse and not dominated solely by protein amino acids. This is an important finding because in recent decades protein amino acids have been the focus of studies on organic N (Näsholm et al., 2009), with amino acids viewed as the major pool of small organic N in the soil solution (Jones et al., 2005). In broad agreement, it was found that, in seven soils, protein amino acids accounted for 52–88% of nonpeptide small organic N and 19 protein amino acids were detected in all soils. However, N-containing organic molecules from an additional 11 compound classes were quantified, with the most abundant other compound classes being quaternary ammonium compounds and nonprotein amino acids (Fig. 2). Moreover, only in one of seven soils (the triticale crop) did protein amino acids comprise all 10 of the most abundant molecules (Fig. 3). The more common situation was for the group of 10 most abundant molecules to include quaternary ammonium compounds (e.g. betaine, hercynine), nonprotein amino acids (e.g. theanine), azoles (urocanic acid) or alkylamines (triethanolamine).
The pool of small, nonpeptide organic N in soil is not only chemically diverse, but also variable among soils. Previous studies have suggested that the composition of the soil amino acid pool is affected little by soil type (Bremner, 1966; Stevenson, 1982), whereas in the present study the composition of the pool of small organic N varied enormously among soils. This variation was evident at the level of compound classes (Fig. 2) and individual molecules (Fig. 3). For example, among soils, the pool of quaternary ammonium compounds varied from 1 to 28% of small organic N, while the pool of nonprotein amino acids varied from 3 to 19% of small organic N. At the molecular level, molecules that were highly abundant in some soils were, in some cases, absent or minor components in other soils (e.g. hercynine, ergothioneine, theanine, triethanolamine, urocanic acid). Moreover, none of the 10 most abundant molecules were common to all seven soils (Fig. 3).
Large variation in the profile of small organic N among soils probably reflects a combination of variation in the plant community, microbial community and abiotic stresses. Abiotic stress may well be a key driver, given that in all soils except for the two E. globulus soils one can see the signature of abiotic stress, with many of the most abundant metabolites being compatible solutes (osmolytes) of microbial or plant origin, such as proline, ectoine and hydroxyectoine, and quaternary ammonium compounds, including betaine, hercynine, γ-butyrobetaine, carnitine and proline betaine (Lippert & Galinski, 1992; Hasegawa et al., 2000; Wood et al., 2001). By contrast, two recent studies did not find large constitutive or inducible amounts of N-containing osmolytes in a seasonally dry grassland soil (Boot et al., 2013) or a prairie soil exposed to a laboratory water stress treatment (Williams & Xia, 2009). Clearly, before conclusions can be drawn regarding functional significance, laboratory manipulations and seasonal field measurements are required to determine the role of the putative osmolytes in tolerance of water stress and other abiotic stresses.
The present study focused on nonpeptides, but other studies indicate that peptides may comprise a large proportion of the pool of small organic N (Farrell et al., 2011b; Hill et al., 2011, 2012). In the soils examined here, it is probable that peptides also accounted for a large proportion of the pool of organic N. For example, nonpeptide organic N (<250 Da) was only 0.4–6% of dissolved organic N < 3 kDa (Table 1), and thus peptides could comprise a maximum of 94–99% of the pool of organic N < 3 kDa. However, direct quantification is required because the pool of organic N between 250 and 3 kDa may comprise multiple compound classes and is not necessarily dominated by peptides.
There are several limitations to analysing pools of small organic N in centrifugal extracts:
Artefacts associated with collection, storage and extraction of soils mean that measured concentrations may not reflect in situ concentrations (Jones & Shannon, 1999; Inselsbacher & Näsholm, 2012). For example, mineralization during collection, storage and extraction probably means that centrifugal extracts overestimate concentrations of inorganic N and underestimate organic N.
Pools of molecules in the soil solution may not reflect accurately (relative or absolute) availability as a result of differential adsorption to the soil stationary phase (Jones et al., 2005).
The most general limitation is that pools of N provide no information about fluxes or processes. We already have information on cycling of protein amino acids (e.g. see reviews by Jones et al., 2005; Näsholm et al., 2009) and small peptides (Farrell et al., 2011b; Hill et al., 2012). A major challenge for future experiments will be to determine processes governing ecosystem cycling of nonprotein amino acids, quaternary ammonium compounds and other compound classes comprising the pool of small organic N.
Uptake of protein amino acids and quaternary ammonium compounds
Direct experimental evidence suggests that plants are capable of accessing at least some of the quaternary ammonium compounds that occur in soil (Table 2). Uptake was estimated from quantification of isotope-labelled molecules within plant tissues, and thus one may be certain that the molecules were taken up intact (Persson & Nasholm, 2001; Sauheitl et al., 2009a; Warren, 2012). Soil microbes are probably strong competitors for quaternary ammonium compounds (Wood et al., 2001), but evidently plants are not competitively excluded because wheat plants growing in soil were able to take up intact quaternary ammonium compounds. An important caveat is that experiments involved addition of a comparatively large amount of isotope label, and thus results are probably indicative of uptake from a pulse of quaternary ammonium compounds (e.g. as might occur following rewetting of dry soil) rather than steady-state uptake. Uptake of quaternary ammonium compounds by plants may well be rather common given that the tree Hevea brasiliensis can take up the quaternary ammonium compound ergothioneine (Audley & Tan, 1968), Arabidposis has transporters for quaternary ammonium compounds (Breitkreuz et al., 1999), while Banksia and wheat can take up betaine, carnitine and acetyl-carnitine from soil.
The data presented here provide very strong evidence of intact uptake of quaternary ammonium compounds, but few insights into quantitative significance. To a large extent, this reflects the current state of knowledge on the uptake of protein amino acids: there is strong qualitative evidence for intact uptake but quantitative significance remains unresolved and controversial (Näsholm et al., 2009). The amounts of intact labelled molecules within plant tissues (e.g. Table 2) are affected by post-uptake metabolism and are best interpreted as being indicative of the minimum amount of intact molecules taken up (Warren, 2012). The massive differences among N forms in the distribution of isotope label between roots and leaves (Table 2) and endogenous concentrations of unlabelled isotopologues (Table 3) are strongly suggestive of differences in metabolism that will confound comparisons of N forms. It was not possible to estimate uptake from total amounts of tracer, because for quaternary ammonium compounds deuterium is the only commercially available isotope label. Quantification of total deuterium in bulk tissue is severely compromised by shifts in H/D values of up to 30% as a result of exchange of deuterium in carboxyl and hydroxyl groups with hydrogen in ambient water vapour (Epstein et al., 1976; Schimmelmann, 1991). For these reasons, it was not possible to compare quantitatively the uptake of quaternary ammonium compounds with that of protein amino acids, although this could be achieved if 13C-labelled quaternary ammonium compounds were synthesized and then used for uptake experiments.
Experiments on organic N uptake by plants are only just beginning to scratch the surface and uncover the amazing versatility of plants. To progress and design experiments that are ecologically realistic will be an enormous challenge for several reasons:
The distribution of concentrations of small organic N molecules (Fig. 1) means that to obtain a reasonable coverage of the pool (e.g. > 75%) will require uptake measurements on at least 10 molecules. To put this in perspective, current studies on organic N uptake from soil generally examine four or fewer isotope-labelled compounds (see data compilation by Sauheitl et al., 2009b), which in the seven soils examined here would account for only 33–69% of small organic N. The effective coverage may be greater if transporters have broad substrate specificity, but no conclusions can be drawn at present because almost nothing is known about transporters for 11 out of the 12 compound classes of small organic N detected in the soil solution.
The enormous variability among soils in the pool of small organic N (Figs 2 and 3) means that it is not possible to predict a priori which molecules will be the most abundant at any site. Thus for experiments to be as ecologically relevant as possible, the selection of isotope-labelled molecules for experimentation ought to be informed by measurements of the molecular profile of the pool of small organic N.
At least some of the most abundant molecules in the soil solution are not commercially available in isotope-labelled forms, and others are not even available as unlabelled standards (e.g. hercynine). Hence, experiments on the uptake of hercynine, ectoine and proline betaine will require synthesis of isotope-labelled molecules.
The results of this study show that the pool of organic N in soil is more diverse and plants have an even broader palate than is suggested by most of the literature on organic N. Studies of ecosystem cycling and plant uptake of organic N have historically focused on protein amino acids (e.g. Chapin et al., 1993; Jones & Darrah, 1993; Warren, 2006; Näsholm et al., 2009), while in recent years some of the focus has shifted to small peptides (Farrell et al., 2011b; Hill et al., 2011, 2012). Measurements of seven soils demonstrates that the pool of small nonpeptide organic N contains molecules from at least 12 compound classes, is not necessarily dominated by protein amino acids, and varies enormously between soils. Quaternary ammonium compounds were a conspicuous feature of many soils and it was demonstrated that two ecologically disparate plant species can take up intact molecules of three quaternary ammonium compounds (betaine, carnitine and acetyl-carnitine). Future experiments need to target those organic N molecules that are abundant in the soil solution, but that are completely uncharacterized in terms of roles in N cycling and plant nutrition. For example, quaternary ammonium compounds (e.g. proline betaine, hercynine), pyrimidines (ectoine and hydroxyectoine) and nonprotein amino acids (e.g. theanine) were abundant in the soil solution at one or more sites and ought to be the subject of further study.
This work was supported by a Future Fellowship from the Australian Research Council. Matthias Pelzing is thanked for assistance with CE-MS. Jacqui Simpson is thanked for collecting soil samples from Tasmania.