|
|
Results and Discussion
Bulk Chemical Characterization of Solid Phase Extracted
and Ultrafiltered DOM. Generally, DOC concentrations
fluctuate with discharge (17), exhibiting low levels at low
flow and vice versa, indicating that DOC concentrations
reflect well river hydrodynamics. In this study, the amount
of surface waterDOMretained by solid phase extraction and
ultrafiltration did not correlate with total DOM (Table
1).
Solid-state 13C NMR spectra of DOM fractions extracted
by the two techniques exhibit a clear difference (Figure 1A-D
versusE-H), especiallywhencomparingDOMfractions from
spring 2000 in which the DOM isolates were from the same
water sample (Figure 1D,H). Due to time and sample
limitations, solution-state NMR studies could be only
performed on a limited number of samples. However, the
solid-state NMR spectra (Figure 1) as well as additional
spectra presented elsewhere by Kaiser et al. (in preparation)
showed that the chemical composition of DOM did not
significantly change spatially or temporally.
One-Dimensional (1-D) Solid-StateNMR.The 1-D solidstateramp13CNMRspectra
ofDOMisolated by both methods
(Figure 1)showbroad resonances.Someof the spectra exhibit
some sharp peaks that are uncharacteristic of similar
published spectra of DOM and may be caused by contaminants
(artifacts). The sharp-clipped peaks at 33ppmobserved
in the solid phase extracted DOM spectra (Figure 1A-D)
may be from bleed of the C18 phase. However, these sharp
peaks are narrow and likely constitute <2% of the sampled
carbon. Ultrafiltered DOM spectra E, F, and G show intense
peaks at 56 ppm not observed in solid phase extracted
materials. These peaks probably resulted from the production
of methyl esters during the drying of the methanol extract
or may be from residual sorbed methanol. The peak at 0
ppm observed in the spectra of ultrafiltered DOM was
probably from methyl siloxanes of undetermined origin.
Spectra obtained by both extraction techniques exhibit
four major functional groups of organic compounds: aliphatics,
carbons adjacent toOorN(such as in carbohydrates,
amino acids, esters, alcohols, etc.), aromatics, and carboxyls-/
aliphatic amides. The chemical shift of more defined signals
indicate specific functional groups or structures, such as the
following: alkyl groups (12-25 ppm), methylene (29-35
ppm), sugars, aliphatic methine, alcohols, methoxyl and
amino carbons (45-90 ppm), possible anomeric sugars (90-
110 ppm), aromatics and alkenes (110-140 ppm), aromatic
carbon adjacent to oxygen (140-160 ppm), carboxylates,
aliphatic amides (160-190 ppm), and carbonyls and ketones
(190-230 ppm).
Except for sample spectrum B, little seasonal variability
is observed (Figure 1). Any such seasonal variability is
certainly less than differences observed between the two
extraction techniques. All spectra display the same carboxylic
acid/aliphatic amide (160-190 ppm) signal intensities. The
solid phase extracted DOM samples contain high aliphatic
carbon contents, high carbohydrate/alcohol contents, and
relatively low aromatic carbon concentrations. Ultrafiltered
DOM spectra have considerably greater aromatic carbon
contents, lower aliphatic carbon contents, and a higher
abundance of methoxy/amino group carbons compared with
spectra of solid phase extracted DOM. Furthermore, the
ultrafiltered samples exhibit minor peaks for carbohydrate/
alcohol carbons and appear to have a lower aliphatic carbon
content than solid phase extracted DOM. The carbon
resonances for aromatics from ultrafiltered DOM showed
phenolic signals, possibly indicating the contribution of
lignin.
Multidimensional Solution-State NMR. A combination
of 1- and 2-D solution-state NMR techniques were applied
to the solid phase extracted and ultrafiltered DOM. Conventional
1-D protonNMRspectra were obtained in DMSOd6
and then reacquired after addition of D2O, allowing
identification of exchangeable functionalities that disappeared
in the presence ofD2O (18, 19). We also performed
homonuclear Total Correlation Spectroscopy (TOCSY) and
Heteronuclear Multiple Quantum Coherence (HMQC) experiments.
TOCSYallows detection of proton bond couplings
in an entire spin system and HMQC correlates 1H and 13C
chemical shifts (over the range of one H-C bond). For both
methods, we observed very short T2 relaxation times, most
likely resulting from a combination of variable magnetic
susceptibility of the high molecular weight heterogeneous
samples, persistence of paramagnetic metals, exchange
processes, and rigidity. As a result, the 2-D spectra underestimate
the contributing molecules with very short T2
relaxation (broader signals), as these signals decay during
the 2-D pulse sequence.
Multidimensional Solution-State NMR of Solid Phase
ExtractedDOM.Four major spectral regions can be identified
(Figure 2A): (1) aromatics, (2) broad signal from water, (3)
signals from amino acids (protons on R-carbons, and various
â-,ç-carbons), sugars, methylene adjacent to ester and ether/
hydroxyl groups, and (4) various aliphatic units (9, 20, 21).
Addition of D2O (Figure 2B) caused the broad water signal
(region 2) to shift to a sharper signal (centered around 3.8
ppm) but otherwise had minimal impact on the spectrum.
Tentative assignments from the TOCSY and HMQC data
are given in figure captions 3A, 3B, and 4. The TOCSY
experiment confirms some of the major assignments (Figure
3). Strong cross-peaks in the aromatic region of the HMQC
(and also observed in theTOCSYdata, Figure 3) are consistent
with phthalate and phthalic acid. These compounds may
originate from storage in HDPE carboys or from natural
riverine organic compounds (20). Resonances in the 2-D
experiments are consistent with lignin, carbohydrates, and
aliphatic esters/acids/ether. Both the 13C solid-state (Figure
1) and 1H solution-state NMR spectra (Figure 2) support the
presence of these structures in the solid phase extractedDOM.
Lignin-derived methoxy carbons and protons are identified
as region 10 in Figure 4.
Multidimensional Solution-State NMR of Ultrafiltered
DOM. The 1-D 1H NMR spectrum of ultrafiltered DOM
dissolved in DMSO-d6 is shown in Figure 5A. Five major
regions can be identified: (1) amides, (2) ammonia, (3)
predominantly sugars, (4) protons on the R-carbon and sidechain
carbons of amino acids and possible contributions
from methine/methylene adjacent to aliphatic ester/ether/
hydroxyl and methine, (5) methylene units bridging lignin
aromatics and aliphatic structures (including resonances
from amino acid side-chains). The addition of D2O reduced
the intensity of region 1 (Figure 5B), suggesting the protons
are exchangeable and, therefore, consistent with amides
(18-20). Moreover, it unmasked resonances for aromatic
protons that can now be clearly observed (region 5).
The 2-D TOCSY (Figure 6) and HMQC (Figure 7) spectra
support assignments made from the 1-D spectra (Figure 5).
The 2-D experiments indicate the presence of peptides,
carbohydrates, protons in long-chain aliphatic structures,
aromatic protons, and ammonia. It is possible that the
aromatic protons are, in fact, part of the peptides, as there
is no evidence of methoxy (generally associated with lignintype
aromatics) in the HMQC spectra. Lignin-derived methoxy
gives a very characteristic and strong resonance at 3.7
ppm for proton and at 56 ppm for carbon in HMQC-type
experiments (22, 23), but it is not present in Figure 7. The
signal centered 56 ppm in the CPMAS spectrum (Figure
1H) may largely result from R-carbons in peptides which
also resonate around this chemical shift. The intensity of the
amide region in the 1H spectrum (Figure 5) indicates a
significant presence of protein/peptide-derived matter. Based
on the total proton intensity in the sample, an integration
of the amide region estimates that 4% of the total proton
intensity can be attributed to the N-H of amides. If we
consider the contribution of the associated side-chains in
each residue to the overall intensity, then the actual amino
acid/peptide contribution to the total 1H intensity of the
sample may be 16-32% (note that for every amino acid
there will be 4-8 additional protons other than the N-H of
amides and dependent on the exact amino acid unit). A C:N
ratio of 37 (Kaiser et al., in preparation) further suggests that
a mixture containing 20% peptide structures may exhibit a
C:N ratio of 30-40, assuming no other source of N. The exact
source of the ammonia is unclear, but its resonance is
apparent in many soil-derived humic-type materials dissolved
inDMSO-d6. It is possible that theammoniais sorbed/
entrapped within the DOM matrix and therefore may
originate from natural sources. However, ammonia also may
result from the hydrolysis of labile peptide structures during
the sample preparation for solution-stateNMRanalysis (see
methods). |
|
回复此楼