Environ. Sci. Technol. 2008, 42, 1430–1437
Molecular and Structural Characterization of Dissolved Organic Matter from the Deep Ocean by FTICR-MS, Including Hydrophilic Nitrogenous Organic Molecules T H O R S T E N R E E M T S M A , * ,† A N J A T H E S E , † MICHAEL LINSCHEID,‡ JERRY LEENHEER,§ AND ALEJANDRO SPITZY| Department of Water Quality Control, Technical University of Berlin, Sekr KF 4, Strasse des 17 Juni 135, 10623 Berlin, Germany, Department of Chemistry, Humboldt-Universität zu Berlin, Brook-Taylor-Strasse 2, 12489 Berlin, Germany, U.S. Geological Survey, Box 25046, Mail Stop 408, Building 95, Federal Center, Denver, Colorado 80225, and Institute of Biogeochemistry and Marine Chemistry, University of Hamburg, Bundesstrasse 55, 20146 Hamburg, Germany
Received August 28, 2007. Revised manuscript received December 10, 2007. Accepted December 17, 2007.
Dissolved organic matter isolated from the deep Atlantic Ocean and fractionated into a so-called hydrophobic (HPO) fraction and a very hydrophilic (HPI) fraction was analyzed for the first time by Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) to resolve the molecular species, to determine their exact masses, and to calculate their molecular formulas. The elemental composition of about 300 molecules was identified. Those in the HPO fraction (14C age of 5100 year) are very similar to much younger freshwater fulvic acids, but less aromatic and more oxygenated molecules are more frequent. This trend continues toward the HPI fraction and may indicate biotic and abiotic aging processes that this material experienced since its primary production thousands of years ago. In the HPI fraction series of nitrogenous molecules containing one, two, or three nitrogens were identified by FTICR-MS. Product ion spectra of the nitrogenous molecules suggest that the nitrogen atoms in these molecules are included in the (alicyclic) backbone of these molecules, possibly in reduced form. These mass spectrometric data suggest that a large set of stable fulvic acids is ubiquitous in all aquatic compartments. Although sources may differ, their actual composition and structure appears to be quite similar and largely independent from their source, because they are the remainder of intensive oxidative degradation processes.
Introduction Over the past years (ultra-) high-resolution Fourier transform ion cyclotron resonance mass spectrometry (FTICR-MS) * Corresponding author fax: +49-30-8412-3685;
[email protected]. † Technical University of Berlin. ‡ Humboldt-Universität zu Berlin. § U.S. Geological Survey. | University of Hamburg. 1430
9
e-mail:
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008
involving electrospray ionization has contributed substantially to dissolved organic matter (DOM) research. With its high mass resolving power it is, yet, the only technique to determine individual molecular species out of complex DOM mixtures, and with its mass accuracy it enables the calculation of molecular formulas for these species. Most molecular formulas published so far are derived from DOM isolates enriched by hydrophobic interactions with a stationary phase made up either from polymeric XAD material (1, 2), which has been used in the classical fulvic acid enrichment processes, or from silica-based C18 stationary phases (3). Owing to these enrichment steps, the material isolated so far was limited in polarity and dominated by fulvic acids, the so-called hydrophobic acids in the XAD isolation procedure (4-6), which are however already quite polar and highly water soluble. More hydrophilic DOM escaped from these enrichment procedures and has not yet been accessible to electrospray ionization mass spectrometry (ESI-MS). The marine DOM budget comprises about 700 × 1015 g of carbon, and the majority of this material is stored in the deep oceans (7). Despite the importance of this enormous mass of organic matter, our knowledge of the chemical nature of DOM from the deep oceans is limited. This is particularly due to greater difficulties in isolating this material than for that in freshwater environments: deep marine DOM resides some kilometers below the ocean’s surface, in a very diluted stage (below 1 mg/L of dissolved organic carbon) and in a very saline environment (35 g/L). Recently published FTICR-MS data on marine DOM originate from isolates enriched by C18 solid phase extraction (SPE) from locations in the south Atlantic Ocean (Weddell Sea (8)) and a Brazilian estuary (9). Previously, ultrafiltration has been used to enrich organic matter of higher molecular mass including colloidal matter (so-called UDOM) from the Pacific Ocean (10), but such isolates necessarily lack low molecular weight components. As for isolates from freshwater environments, these studies did not report any nitrogenous molecules but were limited to molecules consisting of carbon, hydrogen, and oxygen. Supposely this was not due to the absence of dissolved organic nitrogen in marine waters but a tribute to the enrichment process that excluded highly polar low molecular weight DOM components. Only for one freshwater isolate some nitrogen-containing molecular formulas have been reported (3). Besides its benefits in providing molecular formula information for thousands of molecules of complex DOM isolates, ESI-MS is being critically discussed with respect to its potential selectivity. Not all classes of compounds are equally well detected by ESI-MS, whether in the positive or the negative mode. Thus, one has to assume that not all constituents of a chemically diverse DOM mixture are equally represented in its ESI-MS spectrum. In terms of the molecular weight of DOM molecules, a decreasing sensitivity of ESIMS with increasing molecular size of the molecules has been proven by SEC-MS coupling (11). Unfortunately, the extent of selectivity of ESI-MS toward chemically differing DOM constituents cannot be experimentally verified or compensated without exact knowledge of the identity of larger numbers of molecules in such isolates and their availability. To date, such knowledge is only provided by ESI-MS itself. On this basis the interpretation of ESI-MS data should be done with care and without generalizing beyond the molecules that ESI-FTICR-MS shows us. Only recently a procedure previously used to isolate DOM from the Great Salt Lake (12) has been adapted to marine DOM and applied to water of 4000 m depth in the Angola 10.1021/es7021413 CCC: $40.75
2008 American Chemical Society
Published on Web 02/01/2008
Basin, Atlantic Ocean (13). This multistep procedure involved ultrafiltration to remove colloidal matter, followed by XAD-8 adsorption of the permeate to isolate the so-called hydrophobic acids. Highly hydrophilic marine DOM was isolated for the first time from the XAD filtrate (13). We here report results of the first ultrahigh-resolution mass spectrometric investigation of the two major fractions of deep marine DOM from this isolation procedure. Using FTICR-MS the elemental composition of hundreds of molecules in both fractions is determined, including series of highly hydrophilic nitrogenous molecules that have not been determined before. Low-resolution hyphenation mass spectrometry is used to complement elemental information on these molecules with structural information.
Materials and Methods Marine DOM Isolation. DOM was isolated from deep seawater of the Angola Basin, South Atlantic Ocean, collected from a depth of approximately 4000 m in 2004. This water was subjected to a comprehensive enrichment and fractionation approach that is described elsewhere (13). Briefly, colloidal matter was removed by ultrafiltration on board and the permeate subjected to XAD-8 enrichment with elution by acetonitrile/water (75/25), yielding the hydrophobic fraction (HPO fraction). The XAD-8 permeate was reduced to dryness by evaporation of the water and desalted in a multistep procedure involving zeotrophic distillation of water from acetic acid, ultimately yielding the hydrophilic and neutrals fraction (HPI). The HPO and the HPI fraction were the major fractions of the ultrafiltration permeate in terms of the amount of dissolved organic matter (28% and 4% of the organic carbon of the sample) and are further investigated in this study. A freshwater fulvic acid, Suwannee River Fulvic Acid (SRFA) of the International Humic Substances Society (14) was analyzed for comparison. Ultrahigh-Resolution MS. Although we have previously used online size-exclusion chromatography coupled to FTICR-MS (15), this system was not employed for these marine isolates because it would have had to be operated with a mass resolution below 100000 to gain sufficient sensitivity. Here, where the amount of isolate was limited and the presence of nitrogenous organic matter in one of the fractions was likely, high sensitivity and high mass resolution was required. Therefore a small volume of sample dissolved in MeOH/H2O was introduced by infusion (flow of 10 µL/ min) and the MS operated with a resolution of 100000–200000 (m/∆m at fwhm for m/z 400). A Finnigan LTQ FTICR-MS (Thermo Electron Co., Bremen, Germany) linear ion trap FTICR hybrid mass spectrometer with a 6 T superconducting magnet was used with the following instrumental settings: spray voltage, -4.3 kV; shealth gas, 15 arbitrary units; transfer capillary temperature, 230 °C; transfer capillary voltage, -47 V; tube lens voltage, -190 V; number of injected ions (ICR cell), 1–3 × 106. In all experiments negative ions were aquired in the range m/z 200-600. The instrument was externally calibrated each day by the standard procedure using ascorbic acid and Ultramark 1620 as calibrants. MS data were analyzed with Xcalibur version 1.4 SR1. Data Treatment. Between 40 and 50 spectra were summed into one spectrum. Molecular formula determinations proceeded in two phases. Initially, molecular formulas were calculated for fulvic acid molecules from the uncorrected exact mass data assuming that the number of O was >2, of H was >4, of C was >4, and of double bond equivalents (DBE) was g 1 for odd electron anions. The fulvic acid signals occurring at all odd m/z values were then used as internal calibrants to ascribe elemental compositions to many other signals of odd and even mass anions. For these calculations C, H, O, and N were allowed. Of the formulas proposed by
the Xcalibur software, the one with the lowest mass deviation relative to the calibrant signals was selected. In the internally calibrated spectra, the deviation of most ions remained below (0.5 mDa. Identification was further supported by the regular 36.4 mDa distance found for most molecular series at one nominal mass. The total number of molecular formulas determined in the HPO and HPI isolates remained comparatively low (about 300 in the range m/z 320–420). They accounted for approximately 85% of the ions occurring in that mass range with sufficient sensitivity. The small number of detected ions was due to the limited amount of material available and the poor ionization efficiency of the HPI fraction. Product Ion Spectra. Low-resolution product ion spectra were recorded by infusion of the solutions into a triple quadrupole mass spectrometer (Quattro LC, Micromass, Manchester, U.K.) operated in the negative ion mode and using electrospray ionization. The capillary voltage was 2.9 kV, the cone voltage was 25 V, source temperature was 120 °C, desolvation temperature was 100 °C, nebulizer gas flow was 100 L/h, and the drying gas flow was set to 600 L/h. Argon pressure in the collision cell was kept at 1.0 × 10-3 bar, and the collision energy was varied between 12 and 25 eV.
Results and Discussion About 500 L of seawater collected at a depth of 4000 m in the Angola Basin have been subjected to a comprehensive procedure to isolate its DOM (13). Of the seven fractions of DOM obtained by this procedure, two major fractions could be analyzed by FTICR-MS. In the hydrophobic fraction (28% of organic carbon) anions in the range m/z 300–500 were detected with an intensity maximum around m/z 400 (Figure 1b). The mass spectrum of the highly hydrophilic fraction (Figure 1c) (4% of organic carbon) shows anions in the range m/z 250–450 and a skewed intensity distribution with a maximum around m/z 360. As compared to the HPO fraction, the signal intensity of the HPI fraction was much weaker and the region of maximum shifted toward lower m/z values. From these FTICR-MS data, molecular formulas were calculated, and ions consisting solely of C, H, and O were detected in large numbers for both fractions. In the HPI fraction, a series of molecules bearing 1-3 nitrogen atoms was detected also. The presence of molecules with a higher N-number is likely from the data, but mass accuracy did not permit the derivation of unambiguous molecular formulas for these ions. The following presentation, therefore, focuses on the C, H, and O molecules and on those bearing 1–3 nitrogen atoms. Reconstructed mass spectra show the signal intensities of all ions with identified elemental composition with 0, 1, 2, or 3 nitrogens (Figure 2). Nonhetero Molecules. Anions in the mass range m/z 320 to m/z 420 provided signal intensities sufficient for exact mass determination. A total number of 120 molecular formulas consisting solely of C, H, and O were derived for the HPO fraction and 81 for the HPI fraction in this mass range (Figure 2a,b). For comparison 215 formulas were calculated for SRFA under the same instrumental conditions in this mass range. The elemental composition of the molecules determined in the hydrophobic fraction shows a substantial similarity to freshwater isolates as only four out of its 120 molecular formulas were not found in the SRFA isolate. This confirms 13C NMR data of this HPO fraction which have been found to be nearly identical to those of a freshwater isolate from the Colorado River (13). A comparison with the UDOM isolate of the Pacific Ocean (10) is possible only for a limited mass range (m/z 423–455) because the Pacific isolate is dominated by material of higher molecular weight retained by ultrafiltration, while low VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1431
FIGURE 1. FTICR-MS scan spectra (ESI negative) of (a) Suwannee River fulvic acid (SRFA) isolated from freshwater and of dissolved organic matter from the deep Atlantic Ocean, separated into (b) a hydrophobic fraction (HPO) and (c) a hydrophilic fracton (HPI). Inserts show the extended mass spectra for m/z 397.00–397.25. In panels a and b, the numbers denote the number of C/H/O atoms of the anions. In panel c they denote the number of C/H/O/N atoms of the anions. molecular weight DOM was isolated here. For this limited mass range, however, about 2/3 of the molecular formulas of the HPO fraction of deep Atlantic Ocean DOM agree to formulas found in the Pacific Ocean isolate. This is a remarkable agreement if one considers that these materials were isolated in different decades at two stations separated by some ten thousands of kilometers distance, that different isolation procedures were used (XAD vs ultrafiltration), and that elemental formulas in that study were derived from sodiated cations (10) while deprotonated anions were analyzed here. Compared to the freshwater isolate, however, the molecular composition of this marine HPO fraction is much narrower. This is visible at the level of the integer mass, as 1432
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008
exemplarily shown for m/z 397 (insert in Figure 1). Only three ions with the same integer m/z value are visible in the HPO fraction as compared to five in the freshwater isolate. This lower number of detectable ions may partly be due to a generally lower intensity of the HPO spectrum. The comparison of the m/z 397 ions of SRFA and HPO also shows that the intensity maximum in the HPO fraction is shifted toward the left, to ions with a lower mass defect, i.e. to molecules with more oxygen and less hydrogen (C18H21O10 (Figure 1b) compared to C19H25O9 (Figure 1a)). For larger data sets such differences in the distribution of molecular formulas can be well visualized by plotting the C-number of each molecule against its molecular mass. In such diagrams the molecular formulas arrange in islands, and all molecules of one such island are characterized by a constant sum of their carbon and oxygen atoms. This sum increases from one island to the next (15, 16). Molecules with the same integer mass plot in almost vertical lines; their small mass difference of 36.4 mDa corresponds to the mass difference between CH4 and O. In the horizontal direction, the increase in mass within one island is due to the stepwise inclusion of pairs of hydrogen (+2H; + 2015.7 mDa). In this way the molecular compositions of C,H,O-only ions in the range m/z 320–425 of the three isolates studied here are compared in Figure 3. Each dot in the figure corresponds to one molecular formula detected in the freshwater fulvic acid isolate (SRFA). As mentioned above hardly any molecular formula found in the HPO and HPI fraction had not been detected in the freshwater isolate. The molecules of the HPO fraction (14C age of 5100 year (13)) occupy the lower right area of the freshwater islands (Figure 3) only, indicating that primarily the molecules with less carbon (i.e., more oxygen) and more hydrogen are present. This trend continues in the hydrophilic fraction. Here the intensity maxima of each 14 mass unit period are characterized by more oxygens as compared to the HPO fraction (Figure 2a,b). Correspondingly, these molecules plot toward even lower C-numbers in each island (Figure 3) and remain on its right side (high hydrogen content). These molecular formulas explain that the difficulties in isolating such hydrophilic material are not only due to its high oxygen content (high polarity) but also to its higher aliphatic character. The higher degree of hydrogenation of the HPI fraction compared to the HPO fraction is in agreement with the 13C NMR data of this fraction (13). Both these trends, the relative increase in oxygen and hydrogen content of the molecules from the young freshwater isolate toward the much older marine DOM of the deep Atlantic Ocean, appear to be characteristic for aging processes of aquatic DOM. The preferential removal of unsaturated molecules with low oxygen content has been observed by FTICR-MS upon irradiation of freshwater fulvic acids with visible light (17). The same trend could be found when fulvic acids were oxidized by ozone (18). Moreover, newly formed ozonation products could be detected that were extremely oxidized and aliphatic and in which the carbon skeleton appeared to be electronically deactivated and sterically shielded by carboxylate groups against further oxidative attack (18). Thus, photochemically induced as well as chemical oxidation processes preferentially remove more reduced and aromatic DOM molecules, leaving highly oxygenated and aliphatic molecules behind, and they produce highly carboxylated molecules that are deactivated against further oxidation. Nitrogenous Molecules. As visible already from the density of ions in the scan spectrum (Figure 1c), the hydrophilic fraction is very rich in nitrogenous molecules. This is consistent with its C/N ratio of 12 as compared to a ratio of 40 in the HPO fraction (13). From the FTICR-MS data, molecular series with a total of 110 molecules with one,
FIGURE 2. Reconstructed mass spectra showing the intensity distribution of identified anions in the mass range m/z 320–360. Molecules containing solely C, H, and O: (a) HPO and (b) HPI fraction. Nitrogenous molecules of the HPI fraction also containing (c) N1, (d) N2, and (e) N3 molecules. Note that due to the limited graphical resolution most vertical lines are overlays of several ions. The molecular formula of the most prominent ion in C/H/O (a, b) or C/H/O/N (c-e) is indicated at the signal tops. two, and three nitrogens per molecule could be identified with high accuracy, with a mean mass error of e0.05 mDa for all except one series and a standard deviation of 0.2–0.4 mDa (Table 1). The average C/N ratio calculated from the molecular formulas is 15 for the N1, 7.6 for the N2, and 5.2 for the N3 molecules. A characteristic mass difference of 11.2 mDa was found in the mass spectra of many odd m/z ion envelopes which is due to the formal replacement of CO by N2 between molecules consisting only of C, H, and O and those with two nitrogens (Table 1, Figure 1c). The same distance between pairs of signals also was found for even m/z ions, between ions containing one and three nitrogens (Table 1).
The intensity distribution of the three series of nitrogenous DOM molecules of the HPI fraction is visible from the reconstructed mass spectra (Figure 2c-e). They resemble the same pattern as the C,H,O-only molecules (Figure 2a,b). These mass spectral data outline that DOM isolates are not diverse, almost accidential mixtures of molecules but that they exhibit a high degree of regularity. All these nitrogenous molecules carry oxygens in large number, comparable to but not exceeding the C,H,O-only molecules (Figure 2). Elemental compositions of larger series of nitrogenous aquatic DOM molecules derived from FTICR-MS analyses are reported here for the first time. Molecular formulas carrying up to five nitrogens were recently reported from a freshwater isolate (3), but the data are for molecules beyond VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1433
FIGURE 3. C-Number versus molecular mass diagram of all C,H,O-only molecular formulas for the mass range m/z 320–425 of all identified ions in the three samples. For HPO and HPI only those molecules that are not found in SRFA are marked by symbols. Numbers at the top of each island denote the “island number” ()∑ C+O). the m/z range considered in this study and, thus, cannot be compared. This scarcity of data on nitrogenous DOM from marine as well as other aquatic compartments is primarily due to the hydrophilic nature of this material which had previously hampered its enrichment from freshwater and even more so from marine waters. Product Ion Spectra and Possible Structures. HPO Fraction. The determination of their elemental composition is only the first step in elucidating the nature of the DOM molecules detected by FTICR-MS. The next is to obtain structural information by hyphenation mass spectrometry. Most previous product ion spectra of aquatic DOM were generated in negative mode by either low-resolution triplequad-MS or ion-trap MS (19-21) or by time-of-flight (TOF-) MS with higher mass resolution (2, 22, 23). Spectra were dominated by successive losses of 44 Da (CO2) and parallel loss of 18 Da (H2O) (Table S1), leading to fragments of decreasing mass and intensity (Figure 4a,b). These fragmentations have been interpreted as being indicative of carboxylate groups and a limited number of hydroxy groups. Indeed, it has been shown for some fulvic acid molecules by Q-TOF-MS that the number of oxygens in their molecular formulas and the number or decarboxylations in the MS/MS spectra matched well, implying that most of the oxygens in those fulvic acid molecules occurred as carboxylate moieties (23). Product ion spectra of marine DOM have not been published before. The product ion spectra of molecular ions of the marine HPO fraction resemble the freshwater fulvic acids (SRFA) (Figure 4a,b), suggesting that these molecules coincide not
FIGURE 4. Product ion spectra generated from m/z 297 in negative ion mode from (a) SRFA and (b) the HPO fraction of the deep oceanic DOM; product ion spectra of m/z 383 in positive ion mode from (c) the HPO fraction and (d) the HPI fraction of this sample. only in elemental composition but also in molecular structures. These mass spectrometric findings agree to the 13C NMR data of the HPO fraction (13). Obviously a large set of fulvic acid molecules is ubiquitously distributed and can be found in freshwater environment (SRFA) as well as in the deep Atlantic Ocean (Angola Basin, this study) and possibly also in the Pacific Ocean (10). Considering the average 14C age of the two isolates (1400–5000 year) (13) these ought to be very stable molecules. Additionally, product ion spectra of HPO ions were recorded in the positive mode (Figure 4c), and they confirm the structural concept: up to two molecules of water are
TABLE 1. Accuracy and Precision of Mass Determination for Nitrogenous Compounds Relative to Their C,H,O-Only Neighbors in the Range m/z 320–420 formula differencea N1 N 1c N2 N3 N 3c N3d
-2C-5H+N+O -C-O+2N -3C-5H+3N -C-O+2N
theoretical mass differencea (mDa) variousb 958.86 11.23 variousb 970.10 11.23
measured mass differencea (mDa) 958.90 ( 0.17 11.18 ( 0.20 970.26 ( 0.25 11.21 ( 0.26
accuracy (mDa)
n
+0.02 ( 0.23 +0.04 ( 0.17 -0.05 ( 0.18 -0.02 ( 0.39 +0.15 ( 0.25 -0.02 ( 0.26
30 14 30 22 7 13
a To the next C,H,O-only anion. b Summarized over different C,H,O-only neighbors. Thus a constant difference in neither the elemental composition nor the mass distance between the C,H,O-only molecule and the nitrogenous compound can be given. c Also included in the data one line above. d Distance to the closest N1 compound.
1434
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008
eliminated from the cations, while one carboxylate group is expelled as formic acid (Table S2). On the basis of the elemental composition and the unique homogeneous product ion spectra, several structure proposals have been elaborated for fulvic acid molecules from freshwater environment that all show aromatic, alicyclic, and aliphatic (hydroxy-) polycarboxylates (1, 16, 20, 22, 23). They also hold true for marine DOM. Although such structure proposals may appear speculative, the product ion spectra of fulvic acids enable exclusion of all those (sub-) structures that would lead to prominent fragments of the carbon skeleton of the molecules. Among those structures that can be excluded are (a) flavonoids and other structures containing heterocyclic monounsaturated six-membered ring systems, e.g. pyoverdins, because they undergo retro-Diels–Alder scission, leading to prominent fragment ions at significantly lower m/z values than the precursor ion (24, 25), (b) esters, because those tend to eliminate the alcohol as a neutral molecule, and (c) ethers, where also one of the C-O bonds is preferentially broken (26). Moreover lactones or quinones do not seem to be prominent, because no significant loss of CO (-28 Da) (27) was discernible in the product ion spectra. Thus, on the basis of molecules amenable to ESI-MS it appears that these DOM molecules are structurally homogeneous and remarkably poor in the diversity of functional groups. HPI Fraction. The low-resolution triple-quadrupol mass spectrometer used for recording the product ion spectra cannot discriminate between ions with the same integer mass. Thus, one product ion spectrum is the superimposition of the product ion spectra of all precursor ions with the same integer mass. For the HPI fraction, which exhibits strong signals from nitrogenous molecules (Figures 1 and 2), product ion spectra of precursor ions with odd m/z value are influenced by both N2 and N0 ions, while those from even m/z precursors are almost exclusively due to ions with N1 and N3 because 13C isotope signals are much weaker. The few fragmentations visible in these product ion spectra, when recorded in the negative mode, however, were identical to those found in SRFA and the marine HPO fraction. No other prominent fragment ions were detected in any of the spectra recorded in the mass range of m/z 256–398 (Table S1), so that no additional information was obtained by negative ESI-MS on the nitrogen bearing structural element in these HPI molecules. This contrasts the behavior of molecules with highly oxidized nitrogen in atmospheric aerosol, where the presence of nitrate esters was clearly indicated by the loss of 63 Da and the formation of m/z 62 in the negative ion mode, (28) and it supports the suggestion that the nitrogen in HPI molecules occurs in reduced form. Such reduced nitrogen may not be seen in product ion spectra of anions. For example, the lactam substructure in penicillins does not fragment in the negative ion mode, as the fragmentation behavior is governed by the negative charge on one of the oxygens (29), while in positive ion mode the fragmentation of the β-lactam ring led to the most prominent fragments (30). Product ion spectra were recorded also from cations generated in the positive mode (Figure 4c,d). Besides the expulsion of carboxylate groups as formic acid, which was also observed in the HPO fraction, loss of acetic acid (-60 Da) or propanoic acid (-74 Da) is observed from some of the cations with a higher m/z value in the HPI fraction (Table S2). This reflects the higher aliphatic contribution in the HPI as compared to the HPO molecules, which was already recognized from the molecular formula composition (Figure 3). Recording of product ion spectra in the positive mode was performed with the hope that a protonation of the nitrogen in these molecules would support its involvement
in fragmentation processes and would provide evidence for its chemical environment. Although m/z values with a high contribution of nitrogenous cations were selected for fragmentation, such as m/z 341, 353, 383 (corresponding to m/z 339, 351, and 381 for anions in Figure 2d), and although cations with even m/z values that bear one or three nitrogens were also investigated (m/z 358, 372, 384; corresponding to m/z 356, 370, and 382 in Figure 2c,d), their fragmentation processes were almost identical to those of the nonnitrogenous HPO cations (Figure 4c,d; Table S2). Thus, even the positive ion mode provided no nitrogenspecific fragmentations or fragments, neither for odd nor for even m/z ions. This suggests that the nitrogen in the HPI fraction be located in the (alicyclic) skeleton of the molecules, which does not fragment, rather than in functional groups that are expelled easily. The lack of nitrogen-specific fragments may imply that the nitrogen does not occur in the form of amines or amides (25), because these tend to show prominent fragmentations of C-N bonds in positive mode, while carbamates show C-O bond cleavage (31) which did not occur either. Amines and other basic nitrogen moieties also can be excluded on the basis of the enrichment process, as the cation exchange resin used during desalination (13) would have retained them. From the MS data alicyclic nitrogenous compounds such as imides would be feasible, for which decarboxylation but no N-fragmentation has been observed in negative mode (32). It must be noted, however, that the lack of fragments originating from the cleavage of C-N bonds may be explainable by the large number of oxygens present in all these nitrogenous molecules. Typical O/N ratios range from 10/1 to 7/3 (Figure 2c-e), and it is conceivable that in such ions the oxygen atoms, occurring primarily as carboxylate and less so as hydroxy-groups, govern the fragmentation processes rather than the nitrogen(s). This phenomenon has been observed for azaspiracid toxins where series of ether bond cleavages occurred in positive ion mode, whereas the one tertiary amino-group incorporated in the backbone of the molecules did not experience any fragmentation (26). Such a lack of fragmentation in ESI-MS/MS could also be the reason why the deep marine DOM isolates appear structurally less diverse than by infrared and 13C NMR spectroscopy (13). Implications for DOM Sources. The mass spectral data suggest that DOM molecules from freshwater as well as from the HPO and HPI fraction of the marine isolate are quite homogeneous in both their elemental composition and their structures. Such an agreement could indicate that this fraction of DOM from the deep Atlantic Ocean, with a 14C age of several thousand years (13), is preserved terrigenous matter or originates from biological sources that occur on land as well as on sea. Considering the similarity also with soluble DOM from peat and groundwater (23), however, it is more likely that this material can originate from various sources and its similarity is due to the loss of most of its source signature and its functional groups during intensive reworking over the average 1500–5000 years that have past since the primary production of this material. Such material may be found in all aquatic compartments because these molecules are highly water soluble and comparatively stable against further oxidative degradation (16), factors that allow these molecules to reside in the aqueous phase for long periods of time. The shift in the relative frequency toward more aliphatic and more oxidized molecules from the freshwater isolate over the HPO to the HPI fraction may well indicate some aging process. Except for the presence of carboxylate and hydroxyl groups, the product ion spectra did not exhibit any fragmentation specific for structures of known classes of natural organic matter. Either such structures do not exist VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1435
in the aged DOM from the marine environment or they were not amenable to ESI-MS/MS determination. For the series of nitrogenous molecules, a comparison with material from other sources cannot be made, because such data are missing. In terms of the chemical nature of the nitrogenous molecules, mass spectrometric data suggest alicyclic reduced nitrogen species. The origin of these nitrogenous species is not clear yet. These could either be comparatively stable biogenic nitrogenous compounds, from which the skeleton remained, or the nitrogen could have been incorporated from inorganic nitrogen species into a non-nitrogenous precursor material (33).
Acknowledgments Funding of the work at TU Berlin by the German Research Council (DFG, Bonn; Re1290/5-3, 5-4) is gratefully acknowledged. The use of trade names is for identification purposes only and does not constitute endorsement by the U.S. Geological Survey.
Supporting Information Available Table S1 shows the fragments of odd and even m/z anions, and Table S2 shows the fragments of cations recorded by a low-resolution triple-quadrupole mass spectrometer. This material is available free of charge via the Internet at http:// pubs.acs.org.
Literature Cited (1) Stenson, A. C.; Marshall, A. G.; Cooper, W. T. Exact masses and chemical formulas of individual Suwannee River fulvic acids from ultrahigh resolution electrospray ionization Fourier transform ion cyclotron resonance mass spectra. Anal. Chem. 2003, 75, 1275–1284. (2) These, A.; Winkler, M.; Thomas, C.; Reemtsma, T. Determination of molecular formulas and structural regularities of low molecular weight fulvic acids by size-exclusion chromatography with electrospray ionization quadrupole time-of-flight mass spectrometry. Rapid Commun. Mass Spectrom. 2004, 18, 1777– 1786. (3) Kim, S.; Kaplan, L. A.; Hatcher, P. G. Biodegradable dissolved organic matter in a temperate and a tropical stream determined from ultra-high resolution mass spectrometry. Limnol. Oceanogr. 2006, 51, 1054–1063. (4) Leenheer, J. A. Comprehensive approach to preparative isolation and fractionation of dissolved organic carbon from natural waters and wastewaters. Environ. Sci. Technol. 1981, 15, 578– 587. (5) Aiken, G. R.; McKnight, D. M.; Thorn, K. A.; Thurman, E. M. Isolation of hydrophilic organic acids from water using nonionic macroporous resins. Org. Geochem. 1992, 18, 567–573. (6) Aiken, G.; Leenheer, J. Isolation and chemical characterization of dissolved and colloidal organic matter. Chem. Ecol. 1993, 8, 135–151. (7) Hedges, J. I.; Eglinton, G.; Hatcher, P. G.; Kirchman, D. L.; Arnosti, C.; Derenne, S.; Evershed, R. P.; Kögel-Knabner, I.; de Leeuw, J. W.; Littke, R.; Michaelis, W.; Rullkötter, J. The molecularlyuncharacterized components of nonliving organic matter in natural environments. Org. Geochem. 2000, 31, 945–958. (8) Koch, B. P.; Witt, M.; Engbrodt, R.; Dittmar, T.; Kattner, G. Molecular formulae of marine and terrigeous dissolved organic matter detected by electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Geochim. Cosmochim. Acta 2005, 69, 3299–3308. (9) Tremblay, L. B.; Dittmar, T.; Marshall, A. G.; Cooper, W. J.; Cooper, W. T. Molecular characterization of dissolved organic matter in a North Brazilian mangrove porewater and mangrovefringed estuaries by ultrahigh resolution Fourier TransformIon Cyclotron Resonance mass spectrometry and excitation/ emission spectroscopy. Mar. Chem. 2007, 105, 15–29. (10) Hertkorn, N.; Benner, R.; Frommberger, M.; Schmitt- Kopplin, P.; Witt, M.; Kaiser, K.; Kettrup, A.; Hedges, J. I. Characterization of a major refractory component of marine dissolved organic matter. Geochim. Cosmochim. Acta 2006, 70, 2990–3010. 1436
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 42, NO. 5, 2008
(11) These, A.; Reemtsma, T. Limitations of electrospray ionization of fulvic and humic acids as visible from size exclusion chromatography with organic carbon and mass spectrometric detection. Anal. Chem. 2003, 75, 6275–6281. (12) Leenheer, J. A.; Noyes, T. I.; Rostad, C. E.; Lee Davisson, M. Characterization and origin of polar dissolved organic matter from the Great Salt Lake. Biogeochemistry 2004, 69, 125–141. (13) Leenheer, J. A.; Spitzy, A. Isolation and characterization of dissolved organic matter in deep Atlantic Ocean seawater. Submitted for publication. (14) International Humic Substances Society. http://www.ihss.gatech.edu. (15) Reemtsma, T.; These, A.; Springer, A.; Linscheid, M. Differences in the molecular composition of fulvic acid size fractions detected by size-exclusion chromatography-electrospray ionization-on line-Fourier transform ion cyclotron resonance (FTICR-) mass spectrometry. Water Res. 2008, 42, 63–72. (16) Reemtsma, T.; These, A.; Springer, A.; Linscheid, M. Fulvic acids as transition state of organic matter: indications from high resolution-mass spectrometry. Environ. Sci. Technol. 2006, 40, 5839–5845. (17) Kujawinski, E. B.; Del Vecchio, R.; Blough, N. V.; Marshall, G. C.; Kleinand, A. G. Probing molecular-level transformation of dissolved organic matter: insights on photochemical degradation and protozoan modification of DOM from electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Mar. Chem. 2004, 92, 23–37. (18) These, A.; Reemtsma, T. Structure dependent reactivity of low molecular weight fulvic acid molecules during ozonation. Environ. Sci. Technol. 2005, 39, 8382–8387. (19) McIntyre, C.; Batts, B. D.; Jardine, D. R. Electrospray mass spectrometry of groundwater organic acids. J. Mass Spectrom. 1997, 32, 328–330. (20) Leenheer, J. A.; Rostad, C. E.; Gates, P. M.; Furlong, E. T.; Ferrer, I. Molecular resolution and fragmentation of fulvic acid by electrospray ionization/multistage tandem mass spectrometry. Anal. Chem. 2001, 73, 1461–1471. (21) Reemtsma, T.; These, A. On-line coupling of size exclusion chromatography with electrospray ionization-tandem mass spectrometry for the analysis of aquatic fulvic and humic acids. Anal. Chem. 2003, 75, 1500–1507. (22) Plancque, G.; Amekraz, B.; Moulin, V.; Toulhoat, P.; Moulin, C. Molecular structure of fulvic acids by electrospray with quadrupole time-of-flight mass spectrometry. Rapid Commun. Mass Spectrom. 2001, 15, 827–835. (23) Reemtsma, T.; These, A. Comparative investigation of lowmolecular-weight fulvic acids of different origin by SEC-Q-TOFMS: new insight into structure and formation. Environ. Sci. Technol. 2005, 39, 3507–3512. (24) Zywicki, B.; Reemtsma, T.; Jekel, M. Analysis of commercial vegetable tanning agents by reversed-phase liquid chromatography-electrospray ionization tandem mass spectrometry and its application to wastewater. J. Chromatogr. A 2002, 970, 191–200. (25) Budzikiewicz, H.; Schäfer, M.; Fernandez, D. U.; Matthijs, S.; Cornelis, P. Characterization of the chromophores of pyoverdins and related siderophores by electrospray tandem mass spectrometry. BioMetals 2007, 20, 135–144. (26) Brombacher, S.; Edmonds, S.; Volmer, D. A. Studies on azaspiracid biotoxins. II. Mass spectral behaviour and structural elucidation of azaspiracid analogs. Rapid Commun. Mass Spectrom. 2002, 16, 2306–2316. (27) Crotti, A. E. M.; Fonseca, T.; Hong, H.; Staunton, J.; Galembeck, S. E.; Lopes, N. P.; Gates, P. J. The fragmentation mechanism of five-membered lactones by electrospray ionisation tandem mass spectrometry. Int. J. Mass Spectrom. 2004, 232, 271–276. (28) Reemtsma, T.; These, A.; Venkatachari, P.; Xia, X.; Hopke, P. K.; Springer, A.; Linscheid, M. Identification of fulvic acids and sulfate and nitrated analogues in atmospheric aerosol by electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Anal. Chem. 2006, 78, 8299–8304. (29) Rabbolini, S.; Verardo, E.; Da Col, M.; Gioacchini, A. M.; Traldi, P. Negative ion electrospray ionization tandem mass spectrometry in the structural characterization of penicillins. Rapid Commun. Mass Spectrom. 1998, 12, 1820–1826. (30) Adamus, G.; Kowalczuk, M. Electrospray multistep ion trap mass spectrometry for the structural characterisation of poly[(R,S)3-hydroxybutanoic acid] containing a β-lactam end group. Rapid Commun. Mass Spectrom. 2000, 14, 195–202.
(31) Wang, N.; Budde, W. L. Determination of carbamate, urea, and thiourea pesticides and herbicides in water. Anal. Chem. 2001, 73, 997–1006. (32) Reemtsma, T. Analysis of sulfophthalimide and some of its derivatives by liquid chromatography-electrospray ionizationtandem mass spectrometry. J. Chromatogr. A 2001, 919, 289– 297.
(33) Biers, E. J.; Zepp, R. P.; Moran, M. A. The role of nitrogen in chromophoric and fluorescent dissolved organic matter formation. Mar. Chem. 2006, 103, 46–60.
ES7021413
VOL. 42, NO. 5, 2008 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1437
