Heteronuclear NMR Studies of Metabolites ...

Magnetic Resonance in Medicine 42:442–453 (1999)

Heteronuclear NMR Studies of Metabolites Produced by Cryptococcus neoformans in Culture Media: Identification of Possible Virulence Factors William A. Bubb,1* Lesley C. Wright,2 Michelle Cagney,2,3 Rosemary T. Santangelo,2 Tania C. Sorrell,2 and Philip W. Kuchel1 The yeast, Cryptococcus neoformans var. neoformans is a major contributor to the morbidity and mortality experienced by the immunosuppressed population. With a view to providing better treatment, identification of cryptococcal virulence factors is an important goal, with most effort to date directed toward the significance of structural variations in the polysaccharide capsule. The present work describes the characterization of supernatants obtained from cryptococcal cultures. This was achieved by thorough identification of the spin systems of individual metabolites through both homonuclear and heteronuclear NMR experiments that circumvented the difficulties imposed by limited dispersion and a range of concentrations in different cultures covering more than 3 orders of magnitude. More than 30 metabolites, including amino acids, alditols, nucleosides, acetate, and ethanol, were identified by their 1H and 13C chemical shifts and observed long-range correlations. The possible contribution of some detected substances to the pathogenicity of Cryptococcus neoformans is discussed. Magn Reson Med 42:442–453, 1999. r 1999 Wiley-Liss, Inc. Key words: Cryptococcus neoformans; NMR; metabolites; virulence factors

The yeast, Cryptococcus neoformans var. neoformans is a major contributor to the morbidity and mortality experienced by the immunosuppressed population, either through human immunodeficiency virus (HIV) infection, chemotherapy of cancer patients and transplant recipients, or innate deficiencies (1). Treatment to date has been suboptimal, with a high rate of deaths and relapses recorded; new treatments could be developed through the study of cryptococcal virulence factors. One virulence factor, the rigid polysaccharide capsule produced by the fungus, has been studied extensively. In particular, artificial neural network analysis of 1H-NMR data for structural reporter groups of the glucuronoxylomannan capsules obtained from a large number of C. neoformans isolates has led to the suggestion of eight principal chemotypes, which might be associated with distinct biological properties of the organism (2). Other virulence factors assist cryptococci in combating oxidants released by phagocytes during the immune response. Cryptococci contain phenol oxidase, which uses polyphenolics from 1Department

of Biochemistry, University of Sydney, NSW, Australia. for Infectious Diseases and Microbiology, University of Sydney, Westmead Hospital, Westmead, NSW, Australia. 3Department of Infectious Diseases, University of Sydney, Westmead Hospital, Westmead, NSW, Australia. Grant sponsor: Australian Research Council. *Correspondence to: Dr. W.A. Bubb, Department of Biochemistry, University of Sydney, NSW 2006, Australia. E-mail: [email protected] Received 21 December 1998; revised 26 May 1999; accepted 27 May 1999. 2Centre

r 1999 Wiley-Liss, Inc.

the environment to produce melanin in the cryptococcal cell wall (3); superoxide dismutase, which degrades superoxide produced by phagocytes (4); and mannitol, which has been assigned antioxidant properties and the ability to confer increased tolerance by the yeast of heat and osmotic stresses (5). Recently we discovered that phospholipases (B, lysophospholipase, and lysophospholipase transacylase) are produced by clinical and environmental isolates of C. neoformans (6). Phospholipase B production was associated with pathogenicity in a mouse model system, with high phospholipase-producing strains associated with higher mortality rate and dissemination of infection to the brain than low or intermediate phospholipase producers (7). While studying the effects of the crude enzyme extracts on human neutrophils, we found that these cells were damaged by heatinactivated preparations, and sought to identify the components that might be responsible. Many NMR techniques have recently become available for the simplification of spectra and identification of components in complex mixtures such as biofluids. These include spectral dispersion based on diffusion weighting (8), selective excitation of frequencies (9) or functional groups (10), and strategies based on recognition of differences from standard spectral patterns, either by conventional NMR studies (11) or using flow techniques (12). Further developments include the coupling of NMR with other instrumentation such as high-performance liquid chromatography (HPLC) and mass spectrometry (13). Despite these advances, NMR identification of specific substances in mixtures remains dependent on comparison of spectral patterns with those produced by authentic samples (14) or the recognition that the values of certain parameters, generally chemical shifts and coupling constants, are essentially coincident with literature data (11). Since authentic samples are not always available and, if added for verification, inevitably compromise part of the sample under investigation, the latter approach is clearly preferable. While there have been numerous studies involving monitoring of specific processes by heteronuclear NMR (e.g., 15), sensitivity considerations have dictated that most literature determinations of the structures of components of biofluids have been based on 1H NMR data (16), with 13C data only used as supplementary information in some studies (14), for the characterization of specific classes of compounds in extracts (17,18), or to circumvent limited dispersion of 1H chemical shifts when samples have been available in sufficiently large quantities (19). Because of their complexity and the lack of precedents for their composition in the literature, we sought to charac-

442

NMR Study of Cryptococcus neoformans Metabolites

terize cryptococcal supernatants by complete identification of spin systems through both homonuclear and heteronuclear experiments. The concentrations of identified components have been estimated by reference to both one(1D) and two-dimensional (2D) NMR data, and the possible contribution of some detected substances to the pathogenicity of the fungus is considered.

443

and STG-1 strains (containing equal amounts of protein) was diluted 1 : 2 in harvesting buffer, adjusted to pH 5.5, and added to neutrophils (2 ⫻107) in a final volume of 1.5 ml. Samples were incubated at 37°C for 90 min with gentle agitation. Control cells were incubated in harvesting buffer (pH 5.5). Estimation of Neutrophil Viability by Flow Cytometry

METHODS Cryptococcal Strains Two strains of Cryptococcus neoformans var. neoformans, BL-1 and STG-1, were supplied by Dr. Sharon Chen of Westmead Hospital. These are clinical isolates originating from lung biopsies of two patients and were chosen as representatives of high and low phospholipase-producing strains, showing high and low virulence, respectively, in mice (7). Both strains produce capsules of similar size and exhibit similar growth kinetics in vitro, but BL-1 produces less melanin than STG-1, under the same set of conditions (unpublished observation). The fungi were normally cultured on Sabauroud dextrose agar (SDA) plates at 30°C, subcultured every 2 weeks and stored at 4°C. Preparation of Supernatants Enzyme-containing supernatants were prepared according to the method of Chen et al. (6). Briefly, both cryptococcal isolates were grown to confluence on SDA for 72 hr at 30°C, washed and resuspended in harvesting buffer (10 mM imidazole, 2 mM CaCl2, 2 mM MgCl2, 56 mM D-glucose, in 0.9% NaCl, pH 5.5) for incubation (20–24 hr at 37°C). BL-1 cells were also cultured slowly on SDA plates at 4°C, scraped from the plates and processed as above, but entirely at 4°C, or cultured normally but incubated for 24 hr in distilled water instead of harvesting buffer, or in yeast nitrogen broth instead of harvesting buffer. Cells were removed by centrifugation twice at 12,000 ⫻ g for 15 min. The cell-free supernatant was assayed for total protein (BIO-RAD assay kit, BIO-RAD, Ryde, NSW, Australia) and stored at ⫺70°C until required for NMR analysis. Isolation of Neutrophils Neutrophils were isolated from healthy Blood Bank volunteers as previously described (20). Leukocytes were sedimented with 3.5% (w/v) dextran and neutrophils were separated from other cell types by centrifugation on FicollPaque. Residual erythrocytes were removed by hypotonic lysis and neutrophils were resuspended in calcium- and magnesium-free phosphate buffered saline [PBS(⫺)]. Cell purity (ⱖ90%) was monitored by Coulter counter (Coulter, Hialeah, FL, USA) and viability assessed by propidium iodide and fluorescein isothiocyanate (FITC)-annexin V binding (see below). Incubation of Neutrophils With Cryptococcal Supernatant Neutrophils (108) were incubated with shaking in 5 ml Hanks’ balanced salt solution (HBSS) for 1 hr at 37°C then treated with cryptococcal supernatant which had been heated at 80°–90°C for 10 min to destroy enzyme activity. A volume of cryptococcal supernatant from each of the BL-1

Annexin V-Fluos (Boehringer-Mannheim Australia, Castle Hill, NSW, Australia) and propidium iodide (Sigma, St. Louis, MO, 50 µg/ml) were diluted 50-fold in incubation buffer (10 mM HEPES/NaOH, pH 7.4, containing 140 mM NaCl and 5 mM CaCl2) to produce the labeling solution. Duplicate or triplicate samples of neutrophils (106) were washed in PBS(⫺), resuspended in 100 µl of labeling solution, and left at room temperature for 10–15 min. Incubation buffer (400 µl) was added and the fluorescence was read on an EPICS Profile II Analyzer (Coulter, Hialeah, FL). Forward angle light scatter and side scatter were measured for each sample. Fluorescence was quantified using an excitation wavelength of 488 nm and a 530 nm bandpass filter for fluorescein and a ⬎575 nm bandpass filter for propidium iodide. Data were collected as log FL1 and FL2 measurements, respectively. NMR Samples Except for volatile components, NMR assignments were made from spectra acquired for solutions obtained by lyophilization of 2 ml of cryptococcal supernatant and redissolving the residue in 500 µl of D2O. Volatile metabolites were identified and all concentrations estimated from 550 µl unprocessed cryptococcal supernatants dispensed in an NMR tube, with a coaxial capillary containing 1 mM sodium 3-trimethylsilyl-2,2,3,3-tetradeuteropropionate (TSP-d4, sodium salt; Commissariat a` l’E´nergie Atomique, Gif-sur-Yvette, Cedex, France) in D2O as field-frequency lock. NMR Experiments All spectra were obtained with a Bruker AMX-600 NMR spectrometer, using a broadband inverse-detection xyzgradient probe at a 1H frequency of 600.14 MHz, with the variable temperature unit set to 25°C. Chemical shifts are expressed relative to the anomeric resonance of ␣-glucose (1H, ␦ 5.233; 13C, ␦ 92.9) (14) and generally represent the mid-points of cross-peaks in the heteronuclear singlequantum coherence (HSQC) experiment. For 13C–1H correlation experiments with D2O solutions, standard Bruker (UXNMR 970101) pulse programs were used without modification. Otherwise, programs were altered to provide off-resonance presaturation as previously described (21). Spectral widths were routinely 5–6 kHz for 1H, and 16–20 kHz for one-bond, and 30–33 kHz for long-range C–H correlation experiments. All 2D experiments incorporated a relaxation/presaturation delay of 1.7–2 sec in a recycle time of 2–2.2 sec. The majority of resonance assignments were made with the data obtained from an xyz-gradient homonuclear correlation spectroscopy (COSY) experiment acquired with 440 increments of 2K data points and 36 scans per increment, a

444

gradient HSQC experiment acquired with globally optimized alternating-phase rectangular pulses (GARP) decoupling for 1012 increments of 3124 data points and 40 scans per increment, and a gradient heteronuclear multiple-bond correlation (HMBC) experiment acquired with 1024 increments of 4K data points and 32 scans per increment. The one-bond and long-range correlation experiments were optimized for 1JCH of 145 Hz and nJCH of 6.25 Hz, respectively. Resonances associated with certain metabolites present in relatively high concentrations were characterized from a gradient HSQC-total correlation spectroscopy (HSQC-TOCSY) experiment acquired with a mixing time of 80 msec for 920 increments of 2K data points and 32 scans per increment. Metabolites with 1H resonances coupled to 31P were identified from an heteronuclear multiplequantum coherence (HMQC) experiment acquired with 128 increments of 4K data points and 4 scans per increment and an HMQC-TOCSY experiment acquired with a mixing time of 139 msec for 128 increments of 4K data points and 96 scans per increment. Both experiments were optimized for JPH of 7 Hz. Two-dimensional spectra were routinely processed with one degree of zero-filling in F1 and a ␲/2 shifted sine bell in each dimension. Assessments of concentrations were obtained from fullyrelaxed 1H spectra acquired over 64K data points for a spectral width of 6410 Hz, with solvent presaturation for 3 sec following a relaxation delay of 22 sec. Overlapping peaks were identified and the concentrations of alditols other than glycerol estimated from 13C–1H gradient HMQC spectra, typically acquired with 512 increments of 2K data points and 64 scans per increment. Statistical Calculations All statistical calculations were performed using the computer program Instat version 1.13 (GraphPad, San Diego, CA). RESULTS Damage to Human Neutrophils by Cryptococcal Supernatants The nonenzymatic effects of cryptococcal supernatants from the highly virulent, high phospholipase-producing strain BL-1 and the low-virulence, low phospholipaseproducing strain STG-1 were compared to determine if there were factors, in addition to phospholipases, which might contribute to the observed differences in virulence. Compared with similar numbers of necrotic cells in control experiments for each strain, and notwithstanding the highly variable increase in the apoptotic population of neutrophils due to aging (22), the data in Table 1 show that heat-inactivated supernatants from both strains reduced cell viability substantially. That from BL-1 produced highly significant increases in the numbers of both apoptotic and necrotic neutrophils, while that from STG-1 produced more necrotic neutrophils. NMR Identification of Metabolites in Cryptococcal Supernatants NMR assignments for all identified metabolites from both BL-1 and STG-1 strains are summarized in Table 2, which

Bubb et al. Table 1 The Effects of Heat-Inactivated Cryptococcal Supernatants on the Viability of Human Neutrophils* Treatment

Percentage distribution of cells Necrotic

Normal

Apoptotic

Control BL-1

3.4 ⫾ 0.7 12.9 ⫾ 3.2a

71.7 ⫾ 3.2 48.9 ⫾ 7.3a

24.6 ⫾ 2.9 37.8 ⫾ 4.9a

Control STG-1

4.8 ⫾ 1.1 24.0 ⫾ 9.6b

55.0 ⫾ 4.5 40.3 ⫾ 7.2c

40.1 ⫾ 3.5 35.3 ⫾ 5.4

*Neutrophils were incubated in HBSS for 1 hr, followed by 1.5 hr in harvesting buffer, pH 5.5 (controls) or heat-inactivated BL-1 or STG-1 supernatants. The results are expressed as the means and SEM of eight experiments, in duplicate or triplicate for the BL-1 supernatants; five experiments in duplicate or triplicate were performed for the STG-1 control and four experiments for the test. Each experiment, control and test, was carried out on a single batch of cells but different donors were used for individual experiment pairs. aSignificantly different from the controls, P ⬍ 0.005 by the paired, 2-tail, t-test. bDifferent from the control, P ⫽ 0.034 by the unpaired, 2-tail t-test. cDifferent from the control, P ⫽ 0.04 by the unpaired, one tail t-test.

provides chemical shift data for resonances that were observed in spectra of the untreated supernatants and for some additional assigned resonances that were only observed in the concentrated samples. Since the complexity of the mixtures generally precluded comparison of spectra after addition of authentic samples, assignments of resonances to particular structures were based on comparison of both 1H and 13C chemical shifts with literature data, following identification of all detectable correlations. Nonvolatile metabolites were characterized from the spectra obtained from concentrated solutions in D2O (Figs. 1 and 2) and volatile components from the spectra of untreated supernatants (Fig. 3). Chemical shifts for all amino acid resonances are consistent with 1H (23) and 13C (24) literature data, as are (1H, 13C) shifts for lactate and acetate (16,24), ethanol, acetaldehyde and succinate (25,24), ␥-aminobutyric acid (GABA) (26,24) and glycine betaine (betaine) (16,27). The 1D 31P NMR spectrum was dominated by a strong resonance with no 1H correlations, assumed to be inorganic phosphate; minor resonances were just distinguishable from the baseline. However, a 31P–1H HMQC experiment revealed a peak on the shoulder of the inorganic phosphate resonance which had correlations with more than one proton and a similar pattern for a low-intensity 31P multiplet at 0.5 ppm to lower frequency; these observations are consistent with the presence of two secondary phosphates. The extended spin systems identified from a 2D HMQC-TOCSY experiment established that the phosphorus resonances were associated with glycerophosphorylcholine (GPC) and glycerophosphorylethanolamine (GPE) (28), for which several 13C chemical shifts are known (15) to be coincident. A 13C–1H HMQC-TOCSY experiment with a relatively short TOCSY mixing time simplified the most congested part of the HSQC/HMQC data to show only subspectra of certain metabolites present in relatively high concentrations, including both glycerol (14) and mannitol (29,30). Despite extensive overlap of both 1H and 13C resonances, the latter references indicate that each of the common alditols should have unique 13C–1H correlation spectra, and this


445

Table 2 NMR Assignments for Cryptococcus neoformans Metabolites* ␦ 1H

Molecule

HMBCa

Assignment

0.9(0) 0.94 0.95 0.96 0.97 0.99 1.01 1.05 1.18 1.23 1.27/1.47 1.33

unassigned isoleucine unassigned leucine leucine valine isoleucine valine ethanol unassigned isoleucine lactate

1.33 1.41 1.43–1.54 1.48

threonine unassigned lysine alanine

CH3

1.67–1.77 1.70 1.71 1.73 1.77 1.91 1.91

leucine arginine leucine lysine unassigned GABA lysine

␤,␤8-CH2 ␥-CH2 ␥-CH ␦-CH2 JH,H to 3.07 ␤-CH2 ␤-CH2

1.92 1.93

arginine acetate

␤-CH2 CH3

1.99 2.12 2.14 2.24 2.27 2.31

isoleucine unassigned methionine acetaldehyde valine GABA

␤-CH CH3 CH3 ␤-CH2 ␣-CH2

2.59c

succinate

CH2

2.65

methionine

␥-CH2

2.69

aspartate

␤-CH

2.82 3.02 3.03 3.06

aspartate lysine GABA tyrosine

␤8-CH ⑀-CH2 ␥-CH2 ␤-CH

3.06 3.07 3.13

unassigned unassigned phenylalanine

␤-CH

3.20

tyrosine

␤8-CH

3.21

choline

CH3

3.22 3.23 3.25

unassigned GPC arginine

CH3 ␦-CH2

3.25 3.27 3.28 3.29

␤-glucose betaine unassigned phenylalanine

CH-2 CH3

3.30 3.31 3.40

GPE unassigned ␤-glucose

PO-CH2-CH2N

␦-CH3

␤-C, ␥-C

␦-CH3 ␦8-CH3 ␥-CH3 ␤8-CH3 ␥8-CH3 CH3

␦8-CH3, ␤-C, ␥-C ␦-CH3, ␤-C, ␥-C ␥8-CH3, ␣-C, ␤-C ␣-C, ␤-C, ␥-C ␥-CH3, ␣-C, ␤-C C-1

␥-CH2 CH3

␥,␥8-CH2 CH3

␤8-CH

CH-4

C-2 CO2H ␣-C, ␤-C ␣-C CO2H ␥-C ␤-C, ␥-C, ⑀-C 47.2 ␣-C, ␥-C ␣-C, ␥-C CO2H

CO2H ␥-C ␤-C, ␥-C CO2H 13CH 2 CO2H CH3 ␤-C CO2H CO2H ␥-C, ␦-C ␣-C, ␤-C C-5 C-4

C-5 C-4 C-4, C-5 CO2H 13CH 3 N-CH2 13CH , N-CH 3 2 ␤-C, ␥-C C ⫽ NH C-1, C-3 13CH , CH 3 2 53.0, 77.9 C-4, C-5 CO2H

53.0, 78.2 C-3, C-5, C-6

␦ 13C 21.7 11.95 19.3 21.7 22.9 17.5 15.5 18.8 17.6 19.7 25.3 20.3 183.2 20.9 17.3 22.3 17.0 176.5 40.6 27.1 25.0 27.2 26.2 24.4 30.7 175.3 28.4 23.9 182.0 36.7 15.0 14.8 31.0 29.8 35.1 182.2 31.2 179.6 29.7 30.5 37.3 178.4 37.3 40.1 39.9 36.3 127.5 39.7 47.2 37.2 135.8 36.3 174.8 54.7 68.2 41.4 54.8 41.3 157.7 75.0 54.2 53.0 37.2 174.7 40.8 53.0 70.4

Peak no.b

m1

m2, m3, m4 m5, m6 m7, m8

m9

1 2

446

Bubb et al.

Table 2 (Continued) ␦ 1H

Molecule

3.41 3.46 3.47 3.49 3.54 3.56 3.57

␣-glucose ␣,␣-trehalose ␤-glucose ␤-glucose ␣-glucose glycerol glycine

CH-4 CH-4 CH-5 CH-3 CH-2 CH-1A CH2

3.60

threonine

␣-CH

3.62

valine

␣-CH

3.62d 3.62d 3.62 3.62 3.65 3.65 3.66 3.66 3.66 3.68 3.68d 3.68d 3.68 3.68 3.69 3.72 3.72 3.72 3.74 3.74 3.76 3.76 3.76 3.77 3.78 3.78 3.78 3.79

GPC GPE glucitol erythritol glycerol ␣,␣-trehalose ethanol glucitol glucitol isoleucine GPC GPE mannitol GPC erythritol ␤-glucose glycerate ␣-glucose leucine glucitol ␣-glucose lysine mannitol ␣,␣-trehalose erythritol glucitol glycerol alanine

HO-CHH8 HO-CHH8 CH-1A CH-1A CH-1B CH-2 CH2 CH-6A CH-4 ␣-CH HO-CHH8 HO-CHH8 CH-1A PO-CH2-CH2N CH-2 CH-6A CH-3A CH-3 ␣-CH CH-1B CH-6A ␣-CH CH-2 CH-6A CH-1B CH-5 CH-2 ␣-CH

3.79

arginine

␣-CH

3.80 3.81 3.81 3.82 3.82 3.82 3.83 3.84 3.84 3.84 3.85

mannitol cytidine uridine guanosine glycerate ␣,␣-trehalose glucitol ␣-glucose adenosine ␣-glucose serine

CH-3 H-5A8 CH-5A8 CH-5A8 CH-3B CH-5 CH-6B CH-6B CH-5A8 CH-5 ␣-CH

3.85 3.85 3.86 3.86 3.87 3.87 3.88d 3.88d 3.89 3.90 3.90 3.91

glucitol glucitol ␣,␣-trehalose ␣,␣-trehalose methionine mannitol GPC GPE guanosine aspartate ␤-glucose cytidine

CH-3 CH-2 CH-6B CH-3 ␣-CH CH-1B CHH8OP CHH8OP CH-5B8 ␣-CH CH-6B H-5B8

HMBCa

Assignment

C-3, C-5, C-6 C-3, C-5, C-6 C-1 C-1, C-2, C-4 C-3 13C-1, C-2 CO2H ␤-C CO2H ␥,␥8-CH3, ␤-C CO2H N-CH2, HO-CH N-CH2, HO-CH C-2, C-3 C-2 13C-1, C-2 C-3 C-2 ␦,␦8-CH3, ␤-C N-CH2 N-CH2 C-2, C-3 PO-CH2, N-CH3 C-2 C-4, C-5 C-2 ␤-C C-3 ␤-C, ␥-C, CO2H C-1, C-3

C-1 CO2H ␤-C, ␥-C CO2H C-1, C-2

␤-C CO2H C-4e C-4e

C-3

␦ 13C

Peak no.b

70.5 70.5 76.8 76.6 72.3 63.3 42.3 173.1 61.2 173.5 61.2 175.0 62.8 62.8 63.3 63.4 63.3 71.9 58.3 63.7 71.9 60.3 62.8 62.8 64.1 66.8 72.8 61.6 65.0 73.6 54.2 63.3 61.4 55.4 71.7 61.4 63.4 71.7 72.9 51.4 176.6 55.0 175.0 70.1

3 4 5 6 7 8A 9

61.5 62.2 65.0 73.0 63.7 61.4 62.3 72.3 57.2 173.1 70.5 73.8 61.4 73.4 54.8 64.1 67.4 67.4 62.2 53.0 61.6

10 11 12A 12A 13A 14A 8B 15 16 17A 18 19 12B 12B 20A 21 22 23A 24A 25 26 13B 27A 28 29 30A 14B 31 32 33 34 35 36A 37A 24B 38 17B 27B 39A 40 41 42 43 30B 44 20B 45A 45A 37B 23B


447

Table 2 (Continued) ␦ 1H

Molecule

HMBCa

Assignment

3.91 3.91

uridine betaine

CH-5B8 CH2

3.91d 3.91d 3.92 3.95 3.96d 3.96d 3.97 3.99 4.00 4.02c 4.09 4.12 4.13 4.13 4.21 4.23 4.23 4.27 4.30 4.31 4.33 4.36 4.41 4.43 4.65 4.73 4.80 5.20 5.23 5.89(8) 5.90(3)

GPC GPE adenosine tyrosine GPC GPE serine unassigned phenylalanine lactate glycerate GPE cytidine uridine cytidine uridine guanosine threonine adenosine cytidine GPC uridine guanosine adenosine ␤-glucose guanosine adenosine ␣,␣-trehalose ␣-glucose uridine guanosine

HO-CH HO-CH CH-5B8 ␣-CH CHH8OP CHH8OP ␤-CH2 ␣-CH CH-2 CH-2 PO-CH2-CH2N CH-48 CH-48 CH-38 CH-38 CH-48 ␤-CH CH-48 CH-28 PO-CH2-CH2N CH-28 CH-38 CH-38 CH-1 CH-28 CH-28 CH-1 CH-1 CH-5 CH-18

5.90(5) 5.91(5)

cytidine uridine

CH-18 CH-18

6.06 6.07

cytidine adenosine

CH-5 CH-18

6.89 7.18

tyrosine tyrosine

CH-6, 68 CH-5, 58

7.32 7.35 7.42 7.85 7.87

phenylalanine phenylalanine phenylalanine cytidine uridine

CH-5, 58 CH-7 CH-6, 68 CH-6 CH-6

8.00 8.24 8.34 9.68

guanosine adenosine adenosine acetaldehyde

CH-8 (J, 216 Hz) CH-2 (J, 204 Hz) CH-8 (J, 216 Hz) H-1

C-38 CH3 CO2H

CO2H HO-CH HO-CH ␣-C, CO2H C-4, CO2H

C-18, C-58 C-18

C-18, C-48 C-18, C-58 C-18, C-58 C-18 C-18

C-6 C-28, C-8 C-4 C-28, C-6 C-2 C-28, C-8 C-4 C-4 ␤-C C-7 ␤-C, 13C-58, 5 C-4, 13C-68, 6 C-2 C-2 C-4 C-4 C-4

␦ 13C

Peak no.b

61.5 67.0 170.0 71.5 71.5 62.3 57.0 67.4 67.4 61.0 43.3 56.9 69.7 74.2 62.7 84.6 85.0 70.1 70.2 86.2 66.7 86.7 74.5 60.3 74.5 71.3 71.5 96.7 74.3 74.5 94.0 92.9 103.1 88.5 152.0 91.0 90.2 152.4 97.0 89.1 149.3 116.7 131.6 155.6 130.2 129.3 129.9 142.6 142.6 166.9 138.7 153.3 141.5

36B 46 47 47 39B 48 45B 45B 49 50 51 52 53 54 55 56 57 58 59 60 61

*Chemical shifts represent the midpoints of CH correlations and are relative to the anomeric resonance of ␣-glucose at ␦ 5.233 for 1H and ␦ 92.9 for 13C. Except for ethanol, acetaldehyde, and succinate, whose shifts were determined from 1D proton and HMQC spectra obtained for unconcentrated samples, assignments refer to an HSQC spectrum obtained for a freeze-dried sample of BL-1 reconstituted in D2O. aEntries designated 13C refer to correlations with chemically equivalent 12C-attached protons; chemical shifts are given for those resonances not observed in HSQC or HMQC experiments. bNumbers refer to the HSQC and HMQC correlations of Figs. 1 and 3, respectively, and numbers prefaced with m to the HMBC correlations of Fig. 2. cChemical shift significantly different for different samples. dCoincidence of 1H resonances confirmed from 1H-31P HMQC and HMQC-TOCSY experiments. 13C correlations may be associated with either or both molecules. eEither or both of these correlations.

448

FIG. 1. The section of a gradient-selected HSQC spectrum, obtained for a concentrated solution of BL-1 in D2O, covering the region of the 1H spectrum which includes carbohydrate resonances and affords the poorest dispersion. The numbers refer to correlations identified in Table 2; no, denotes correlations which were not observed in the untreated supernatants. Experimental details are provided in Methods.

allowed for the identification of erythritol and glucitol. The presence of the symmetrical ␣,␣-trehalose was suggested by the characteristic coincidence of one- and multiple-bond 13C–1H correlations for the anomeric proton and confirmed from literature data (31). Glucose (14) was present in high concentration in the buffer in which spectra for most samples were acquired. The concentration of glycerate, was not sufficient to yield any HMBC correlations, even for a D2O solution, and the assignment is based on comparison of the 1H (16) (associated through a COSY spectrum) and 13C (32) chemical shifts with literature data. Although present in relatively low concentrations, the identification of nucleosides was facilitated by the disper-

FIG. 2. Part of a gradient-selected HMBC spectrum obtained for a concentrated solution of BL-1 in D2O, illustrating the dynamic range achieved. The detected long-range correlations, labeled m1–m9 (Table 2), all have intensities in a 1H spectrum (shown above the contour plot) below the levels of the 13C satellites (s) associated with the strong acetate resonance (A). Experimental details are provided in Methods.

Bubb et al.

FIG. 3. Section of a gradient-selected HMQC spectrum obtained for the fresh, unconcentrated BL-1 supernatant documented in Table 3. The figure shows that the level of detail of Fig. 1 (boxed) could be achieved for untreated samples and that consideration of heteronuclear correlations permits a much more thorough assessment of contributions to the 1H spectrum than is available from a fully-relaxed 1D experiment (shown above the contour plot). Both spectra were obtained with a coaxial capillary containing D2O for field-frequency lock; other experimental details are provided in Methods. The numbers refer to correlations identified in Table 2; nc, denotes correlations which were not observed in concentrated samples.

sion of their chemical shifts which, after allowing for solvent and pH effects in guanosine and adenosine (33), are consistent with recently published literature data (34), except for the 13C chemical shifts of C28 and C38, where our results support the finding (35) that the original assignments must be interchanged. Comparison of STG-1 and BL-1 Supernatants Determination of Metabolite Concentrations In order to confirm the identity of all substances contributing to each integrated region of the 1H spectrum, gradientselected HMQC spectra were acquired for each supernatant. The superposition of 1H resonances identified by this approach is illustrated in Fig. 3 for the unconcentrated BL-1 supernatant in buffer. Metabolite concentrations are given in Table 3; not included are substances only identified in freeze-dried, concentrated D2O solutions. The spectrum confirms the rich array of information available from NMR analysis of untreated supernatants. With the species contributing to each region of a 1H spectrum identified, the concentrations of many metabolites could be determined through integration of fully-relaxed 1D spectra, by consideration of appropriate sums and differences. Otherwise, some concentrations were estimated from peak heights of similar signals in similar molecules, as indicated in Table 3. The concentrations of alditols other than glycerol were estimated from comparison of volumes of resolved HMQC cross-peaks with those of glycerol. While present at significant levels in some freeze-dried samples, only traces of ␣,␣-trehalose were detected in unconcentrated supernatants. Similarly, glycerate was not detected at all in the latter. Several samples contained


449

Table 3 Comparison of Metabolite Production for Different Cryptococcus neoformans Incubations* Signal

Molecule

Assignment

BL-1 H2O

BL-1 4°C

BL-1 buffer

STG-1 buffer

0.96a 1.01a 1.05 1.18 1.33 1.33 1.48 1.93 2.31 2.59 2.65 2.69 3.02b 3.21 3.23 3.25 3.27 3.30 3.56 3.57b 3.78/63.4c 3.83/63.7c 3.87/64.1c 3.97/61.0a 5.20/94.1 5.23d 6.06 6.89/71.8c 7.32–7.42 7.87 8.00 8.24/8.34e 9.68

leucine isoleucine valine ethanol lactate threonine alanine acetate GABA succinate methionine aspartate lysine choline GPC arginine betaine GPE glycerol glycine erythritol glucitol mannitol serine ␣,␣-trehalose glucose cytidine tyrosine phenylalanine uridine guanosine adenosine acetaldehyde

␦-CH3 ␤-CH3 ␥8-CH3 CH3 CH3 CH3 CH3 CH3 ␣-CH2 CH2 ␥-CH2 ␤-CH ⑀-CH2 CH3 CH3 ␦-CH2 CH3 CH2 H-1A CH2 CH-1B CH-6B CH-1B ␤-CH2 H-1 H-1 H-5 H-6,68/H-5,58 HAROM H-6 H-8 H-2/H-8 H-1

1 1 1 48 2 2 5 124 4 1 0.4 0.9 ⬍7 0.2 2 3 2 0.8 21 1 6 4 8 1 t 21 0.1 0.6 1 0.6 0.4 1.0 0.2

nd nd nd 72 nd nd nd 108 0.8 1 nd nd nd nd t nd 0.1 nd 3 nd nd nd nd nd nd — nd nd nd nd nd nd t

1 1 2 62 2 1 5 172 4 1 0.3 0.9 ⬍10 0.3 2 2 3 0.7 23 1 7 4 8 1 nd — 0.2 0.5 1 0.8 0.6 2 0.1

0.5 nd 0.6 63 4 0.9 5 215 8 2 0.3 0.6 ⬍4 0.4 1 nd 3 1 13 1 6 6 4 nd t — nd 0.2 0.5 0.6 nd 0.9 0.1

*Unless indicated otherwise, concentrations (mM) were estimated from integration of fully-relaxed 1H spectra after subtracting independently measured contributions where necessary. Samples were freshly prepared, i.e., not concentrated by freeze-drying. The buffer was harvesting buffer pH 5.5 (see Methods). nd, not detected; t, trace amount only although detected in both 1D and 2D spectra. aEstimated from peak height compared with the valine ␥8-CH doublet, with allowance for relative number of protons and signal multiplicity. 3 bIn addition to the dominant lysine and GABA resonances, minor peaks were evident. cEstimated from HMQC cross-peak volume compared with that of glycerol CH-1 . A dExcept for the water-treated sample, glucose was a major constituent of the buffer. eAverage of both integrals.

weak signals at ␦ 1H, 2.05–2.4/␦ 13C, 25–35 which were plausibly associated with glutamate and glutamine, but the assignments could not be confirmed due to the lack of further correlations. The substances present in greatest amount in both preparations are acetic acid, which we have reported previously in BL-1 supernatant (6) and ethanol. Both supernatants produced sufficient acid to reach pH ⬇ 3.8. STG-1 supernatant contained about double the amount of lactate and GABA, and half the amount of glycerol and amino acids, (arginine was not detected), compared with BL-1. Mannitol and nucleosides were present in greater amounts in BL-1 supernatant. Surprisingly, no evidence was found of capsular components in either supernatant despite evidence in the literature of antigenic, capsular components being detectable in human blood and cerebrospinal fluid (36). In addition, the presence of lipids could not be established unequivocally.

Influence of Culture Conditions on Metabolite Production To ascertain if the components of the BL-1 supernatant differed under conditions of osmotic stress or low metabolic rate, the supernatants were produced in either distilled water or from cells held at 4°C. The cells incubated for 24 hr in water produced a supernatant qualitatively almost identical to that obtained in harvesting buffer (Table 3), and confirmed the production of glucose by the cryptococci, presumably as a response to osmotic stress. In contrast, very few metabolites were produced from cells maintained at 4°C (Table 3). Ethanol and acetate were again dominant, and succinate, GPC, betaine, and glycerol were present, but no amino acids were detected. These results suggest that active yeast metabolism is required to produce most of the metabolites present in the BL-1 supernatant. The components produced when the supernatant was made from cells incubated overnight in yeast nitrogen

450

broth (pH 7.0) were similar to those produced in harvesting buffer (data not shown). DISCUSSION NMR Characterization of Mixtures As is evident from the projection above Fig. 3 and data presented in Table 3, 1D proton spectra of supernatants were dominated by the resonances of a small number of metabolites present in relatively high concentrations. The t1 noise associated with these resonances produced very poor quality 2D spectra with coherence selection by phase cycling, but gradient selection improved spectral quality sufficiently to permit the identification of all but a few resonances of relatively minor intensity. The F1 resolution, which was achievable for these mixtures of metabolites of relatively low molecular weight, is shown in Fig. 1, there still being ample signal after 1000 increments of an HSQC experiment. For the biofluids examined here, selective excitation (9) is clearly not essential to achieving adequate resolution in the 13C dimension. The multiple-bond correlations revealed in Fig. 2 show that even the relatively insensitive cross-peaks associated with proton multiplets may be detected on the shoulders of intense singlets of metabolites present in much higher concentration, to provide an extensive array of information in support of each structural assignment (Table 2). Heteronuclear correlation experiments are often rejected for the simplification of complex mixtures because of their relatively low sensitivity compared with exclusively 1H approaches. Here we have demonstrated that the additional dispersion of the 13C dimension offers the potential to characterize complex mixtures of metabolites present in a very high range of concentrations. Of particular importance for mixtures of metabolites, the HMBC experiment provides the capacity to identify attached carbonyl groups, and ready recognition of certain groups of chemically equivalent nuclei, such as the 12CH2–13CH2 moiety in succinate, 12CH–O–13CH in ␣,␣-trehalose, and (12CH3)2– 13CH in choline derivatives, all of which display strong 3 multiple-bond correlations at the same 1H/13C frequencies as the corresponding one-bond C–H correlation. Where there is considerable overlap in 1D spectra, the fingerprint established by identification of both short- and long-range heteronuclear, in addition to homonuclear, correlations provides a more rigorous proof of structure than is available from the more laborious approaches of obtaining 1D spectra for individual metabolites or adding authentic samples. The resonances of alditols and saccharides, in particular, have such restricted dispersion in 1D spectra that these latter approaches have limited applicability. The methodology adopted here has permitted the identification of several of these compounds despite the presence of high concentrations of glucose in the buffer. An interesting procedure for the characterization of binary mixtures of alditols by training artificial neural networks to recognize 1H NMR spectral patterns has been described recently (37) but would not appear to be readily adaptable to complex mixtures of metabolites such as those encountered here. As indicated in Table 2, few of the relatively weak unassigned resonances could be associated with more than one additional resonance. In particular, it was not possible

Bubb et al.

to identify a network of correlations for several weak low-frequency resonances that are conceivably due to lipids (14,18). No additional correlations could be associated with the putative trimethylammonium groups at ␦ 3.28 and ␦ 3.31, one of which might represent ergothioneine (38), but other regions of the spectra do not provide sufficient evidence for a conclusive assignment. Similarly, unidentified peaks at ␦ 1.77/26.2, ␦ 3.06/39.7, and ␦ 3.07/47.2 are plausibly associated with polyamines, but there was insufficient evidence to associate them with the structures of substances likely to be found in biofluids (17). Not surprisingly, some additional resonances were observed in the spectra of the freeze-dried, concentrated samples that were obtained with the additional advantage that there was no need for a capillary which occupied active volume. However, only glycerate and ␣,␣-trehalose provided sufficient spectral detail to permit structural assignments. Only the anomeric resonances for the latter were observed in fresh supernatants, whereas relative intensities in some concentrated samples appeared abnormally large. Since chemical modification during concentration seems unlikely, it is possible that ␣,␣-trehalose production by the fungus is highly variable. NMR responses for glycerate in freeze-dried samples were consistent with it being just below the detection threshold in unconcentrated supernatants, but its presence therein cannot be unequivocally asserted. Unassigned cross-peaks observed in the spectra of concentrated samples have not been included in Table 2, since without identification they cannot be unequivocally deemed to be associated with compounds produced by the fungus. Specifically, the reasonably intense peak seen in Fig. 1 at ␦ 3.71/70.4 was not observed in any of the fresh supernatants and, based on chemical shift data alone, is conceivably due to polyethyleneglycol (28). In inverse-detected heteronuclear experiments, strong multi-proton singlets inevitably give rise to relatively strong correlations, but further tracing of the associated spin system is hindered by the much weaker signals of greater multiplicity due to relatively fewer protons, as is evident for the trimethylammonium groups noted above. Similarly, the C-5 resonance of cytidine, which is dispersed over two proton chemical shifts, was below the detection threshold, while the ␥-carbon of methionine was only detected in concentrated samples by contrast with the methyl-carbon that was also observed in the untreated supernatants. As indicated in Table 3, both cytidine and methionine were present at submillimolar concentrations in untreated supernatants. Notwithstanding the large range of metabolite concentrations, comparison of integrated intensities with that of a single reference concentration was deemed to be appropriate, since resolved 13C satellites of some major metabolites had the expected integrated intensities compared with their 12C parents. Estimates of the concentrations of alditols other than glycerol from HMQC cross-peak volumes are likely to be much less reliable. However, despite the recognized limitations associated with the measurement of peak-volumes in 2D NMR spectra (39), the concentration estimates are considered to be more meaningful than qualitative descriptions of peak intensities. Test spectra with comparable digital resolution, the same window functions, and known concentrations of glycerol and man-


nitol showed that the error in estimating concentrations from comparison of HMQC cross-peak volumes could be as high as 30%. Further refinement of the procedure should be possible but was not considered justified in the present context. Biological Significance Many of the compounds listed in this report as released into the medium by Cryptococcus neoformans have not been previously observed in fungal cultures. Those which have been reported for Cryptococcus include ethanol (40), mannitol (5), and acetic acid (6). Of these compounds, only mannitol has been documented as correlating with cryptococcal virulence (5). As an intracellular component, it is believed to confer resistance to heat and osmotic stresses and to scavenge extracellular hydroxyl radicals produced by phagocytes, thus avoiding oxidative killing. Furthermore, the large amounts of mannitol accumulating in heavily infected brain tissues may increase the osmolarity of those tissues, thereby contributing to cerebral edema (41). Ethanol A previous report (40) has shown that glucose-based fermentation in C. neoformans is of minor importance compared with that in Candida albicans and Saccharomyces cerevisiae. The highest amount of ethanol produced by one strain of C. neoformans was 0.064 % (w/v). Ethanol levels of 0.0004% (w/v) have been reported in spinal fluid inoculated with cryptococci (42). The ethanol production in our supernatants was much greater [⬇0.3% (w/v)]. Nilsson et al. (43) found that 0.79% (w/v) ethanol impaired leukotriene generation and leukotriene-induced functional responses in human neutrophils, such as aggregation, oxidative metabolism, chemotaxis, migration, and elastase release. Cell viability measured by Trypan blue exclusion was not decreased by 0.79% (w/v) ethanol treatment. Thus ethanol formation by our strains of cryptococci could attenuate the immune response, and contribute to the pathogenicity of the organism. Acetic Acid The source of acetic acid in cryptococcal cultures is unknown, but it could result from the oxidation of acetaldehyde (a side-product of the Embden-Meyerhof pathway), or from glycerol/fatty acids released from lipid metabolism [cryptococci contain large stores of intracellular lipid droplets (unpublished observations)]. Certainly the acidification of the environment of cryptococcal cells (pH 3.8 under our conditions) would be important when these cells cluster within tissues in vivo to form cryptococcomata. Many mammalian host cell types could be damaged by this acidification alone, but more significantly, the cryptococcal phospholipases are active at these low pH values (6) and may initiate or exacerbate membrane damage.

451

as energy stores and intracellular stress protectants. Deletion or mutation of the glycerol-3-phosphate dehydrogenase genes in Saccharomyces cerevisiae resulted in the inability to produce glycerol and grow in hyperosmolar media and high temperatures (e.g., 42°C) (44). While small amounts of glucitol and erythritol were detected in cryptococcal supernatants, the large amounts of glycerol produced from BL-1 could, in combination with mannitol, produce a hyperosmotic environment in surrounding host tissues resulting in cellular damage. These effects would be less likely in the presence of the STG-1 strain. GABA The secretion of GABA from fungi has not previously been reported. In mammalian systems GABA is a neurotransmitter that acts through receptors to produce neural depression, and could be responsible for neurological dysfunction in patients with cryptococcal meningo-encephalitis. The observation that encapsulated cryptococci adhere to glial cells (45) indicates that the yeast can gain close proximity to neurons. GABA is degraded by GABAketoglutarate transaminase, which produces succinate, another compound found to be present in cryptococcal supernatants (Table 3). Choline Derivatives The observation of GPC and choline in cryptococcal supernatants is novel, and may have physiological consequences. The GPC component of Streptococcus pneumoniae has been demonstrated to be antigenic (46). It is possible that specific antibodies are made against the GPC of cryptococcal origin. The protective nature of anti-GPC antibodies from cryptococci has not been assessed but is probably minimal as anti-GPC IgG antibodies do not seem to be protective against pneumococcal infection. However, much of the human population carry GPC antibodies (46) and cross-reactions to the antigen produced by cryptococcal cells would stimulate an immune response. Nucleosides Adenosine, in particular, interacts with various cell types including neutrophils, where it has an anti-inflammatory effect. It prevents the stimulation of neutrophils to produce superoxide and inhibits actin polymerization required for adhesion, shape change, and phagocytosis (47). It also exerts a protective effect on neutrophils during reperfusion (48). Less adenosine appeared to be produced from STG-1 cells. Arginine This amino acid was not detected in STG-1 supernatants. Arginine is a precursor of nitric oxide and may have anti-inflammatory effects by preventing superoxide production in neutrophils (49). It prevents neutrophil accumulation during lung inflammation and in gastric reperfusion injury (50).

Polyols

SUMMARY AND CONCLUSIONS

Acyclic polyols such as glycerol and D-arabinitol have been reported in other fungi (19) and have been proposed to act

Application of gradient-selected heteronuclear NMR techniques has provided the first extensive fingerprinting of

452

metabolites produced by Cryptococcus neoformans in culture media. Several identified metabolites have the capacity to damage mammalian cells or to suppress the immune response. Supernatants from the two cryptococcal strains in this study produced different effects on neutrophils, with strain STG-1 tending to induce more necrosis than BL-1. Metabolites, such as glycerol, mannitol, adenosine, and arginine, may contribute to this phenomenon. Reduced necrosis of neutrophils in the presence of cryptococci and suppression of neutrophil stimulation could avoid production of an inflammatory response by the host and allow the immunologically ‘‘silent’’ invasion of tissues by the fungus. We conclude that metabolically active cryptococci produce, in addition to enzymes (7), a concentrated complex mixture of potentially damaging compounds that may be identified and estimated by NMR analysis without sample pretreatment or concentration. Pulsed-field-gradient based heteronuclear experiments permit the characterization of metabolites present in a large range of concentrations over a limited range of chemical shifts; some of these substances plausibly contribute to the virulence of the fungus, and their effects on phagocytic cells are the subject of continuing investigations.

REFERENCES 1. Kwon-Chung KJ, Bennett JE. Medical mycology. Philadelphia: Lea and Febiger; 1992. 866 p. 2. Cherniak R, Valafar H, Morris LC, Valafar F. Cryptococcus neoformans chemotyping by quantitative analysis of 1H nuclear magnetic resonance spectra of glucuronoxylomannans with a computer-simulated artificial neural network. Clin Diagn Lab Immunol 1998;5:146–159. 3. Jacobson ES, Tinnell SB. Antioxidant function of fungal melanin. J Bacteriol 1993;175:7102–7104. 4. Hamilton AJ, Holdom MD. Biochemical comparison of the Cu, Zn, superoxide dismutases of Cryptococcus neoformans var. neoformans and Cryptococcus neoformans var. gattii. Infect Immun 1997;65:488– 494. 5. Chaturvedi V, Flynn T, Niehaus WG, Wong B. Stress tolerance and pathogenic potential of a mannitol mutant of Cryptococcus neoformans. Microbiology 1996;142:937–943. 6. Chen SCA, Wright LC, Santangelo RT, Muller M, Moran VR, Kuchel PW, Sorrell TC. Identification of extracellular phospholipase B, lysophospholipase and acyltransferase produced by Cryptococcus neoformans. Infect Immun 1997;65:405–411. 7. Chen SCA, Muller M, Zhou JZ, Wright LC, Sorrell TC. Phospholipase activity in Cryptococcus neoformans: a new virulence factor? J Infect Dis 1997;175:414–420. 8. Dalvit C, Bo¨hlen JM. Analysis of biofluids and chemical mixtures in non-deuterated solvents with 1H diffusion-weighted PFG phasesensitive double-quantum NMR spectroscopy. NMR Biomed 1997;10: 285–291. 9. Willker W, Flo¨gel U, Leibfritz D. Ultra-high-resolved HSQC spectra of multiple-13C-labeled biofluids. J Magn Reson 1997;125:216–219. 10. Liu M, Mao X-A, Ye C, Nicholson JK, Lindon JC. Three-dimensional maximum-quantum correlation HMQC NMR spectroscopy (3D MAXYHMQC). J Magn Reson 1997;129:67–73. 11. Holmes E, Foxall PJD, Nicholson JK, Neild GH, Brown SM, Beddell CR, Sweatman BC, Rahr E, Lindon JC, Spraul M, Neidig P. Automatic data reduction and pattern recognition methods for analysis of 1H nuclear magnetic resonance spectra of human urine from normal and pathological states. Anal Biochem 1994;220:284–296. 12. Spraul M, Hofmann M, Ackermann M, Nicholls AW, Damment SJP, Haselden JN, Shockcor JP, Nicholson JK, Lindon JC. Flow injection proton nuclear magnetic resonance spectroscopy combined with pattern recognition methods: implications for rapid structural studies and high throughput biochemical screening. Anal Commun 1997;34:339– 341. 13. Shockcor JP, Unger SE, Wilson ID, Foxall PJD, Nicholson JK, Lindon JC.

Bubb et al.

14.

15.

16.

17.

18.

19.

20.

21.

22.

23.

24. 25. 26.

27.

28.

29.

30. 31.

32.

33.

34.

Combined HPLC, NMR spectroscopy, and ion-trap mass spectrometry with application to the detection and characterization of xenobiotic and endogenous metabolites in human urine. Anal Chem 1996;68:4431– 4435. Nicholson JK, Foxall PJD, Spraul M, Farrant RD, Lindon JC. 750 MHz 1H and 1H–13C NMR spectroscopy of human blood plasma. Anal Chem 1995;67:793–811. Klein G, Martin J-B, Cotter DA, Satre M. Multinuclear NMR spectroscopy of the cellular slime mold Polysphondylium pallidum—monitoring of the encystment and excystment processes. Eur J Biochem 1992;204:847–856. Agar NS, Rae CD, Chapman BE, Kuchel PW. 1H NMR spectroscopic survey of plasma and erythrocytes from selected marsupials and domestic animals of Australia. Comp Biochem Physiol 1991;99B:575– 597. Willker W, Flo¨gel U, Leibfritz D. A 1H/13C inverse 2D method for the analysis of the polyamines putrescine, spermidine and spermine in cell extracts and biofluids. NMR Biomed 1998;11:47–54. Willker W, Leibfritz D. Assignment of mono- and polyunsaturated fatty acids in lipids of tissues and body fluids. Magn Reson Chem 1998;36: S79–S84. Dijkema C, Kester HCM, Visser J. 13C NMR studies of carbon metabolism in the hyphal fungus Aspergillus nidulans. Proc Natl Acad Sci (USA) 1985;82:14–18. May GL, Wright LC, Groot Obbink KL, Byleveld PM, Garg ML, Ahmed ZI, Sorrell TC. Increased saturated triacylglycerol levels in plasma membranes of human neutrophils stimulated by lipopolysaccharide. J Lipid Res 1997;38:1562–1570. Urashima T, Bubb WA, Messer M, Tsuji Y, Taneda Y. Studies of the neutral trisaccharides of goat (Capra hircus) colostrum and of the oneand two-dimensional 1H and 13C NMR spectra of 68-N-acetylglucosaminyllactose. Carbohydr Res 1994;262:173–184. Savill JS, Wyllie AH, Henson JE, Walport MJ, Henson PM, Haslett C. Macrophage phagocytosis of aging neutrophils in inflammation— programmed cell death in the neutrophil leads to its recognition by macrophages. J Clin Invest 1989;83:865–875. McDonald CC, Phillips WD. Proton magnetic resonance spectra of proteins in random-coil configurations. J Am Chem Soc 1969;91:1513– 1521. Kalinowski H-O, Berger S, Braun S. Carbon-13 NMR spectroscopy. Chichester: John Wiley and Sons; 1994. 776 p. Pretsch E, Clerc T, Seibl J, Simon W. Tables of spectral data for structure determination of organic compounds. Berlin: Springer-Verlag; 1983. Re´my C, Aru´s C, Ziegler A, Lai ES, Moreno A, Le Fur Y, De´corps M. In vivo, ex vivo, and in vitro one- and two-dimensional nuclear magnetic resonance spectroscopy of an intracerebral glioma in rat brain: assignment of resonances. J Neurochem 1994;62:166–179. Prior BA, Hewitt E, Brandt EV, Clarke A, Mildenhall JP. Growth, pectate lyase production and solute accumulation by Erwinia chrysanthemi under osmotic stress: effect of osmoprotectants. J Appl Bacteriol 1994;77:433–439. Moreno A, Aru´s C. Quantitative and qualitative characterization of 1H NMR spectra of colon tumors, normal mucosa and their perchloric acid extracts: decreased levels of myo-inositol in tumours can be detected in intact biopsies. NMR Biomed 1996;8:33–45. Hawkes GE, Lewis D. 1H Nuclear magnetic resonance spectra and conformations of alditols in deuterium oxide. J Chem Soc Perkin Trans II 1984;2073–2078. Angyal SJ, Le Fur R. The 13C-NMR spectra of alditols. Carbohydr Res 1980;84:201–209. Bock K, Defaye J, Driguez H, Bar-Guilloux E. Conformations in solution of ␣,␣-trehalose, ␣-D-glucopyranosyl ␣-D-mannopyranoside, and their 1-thioglycosyl analogs, and a tentative correlation of their behaviour with respect to the enzyme trehalase. Eur J Biochem 1983;131:595–600. Serianni AS, Clark EL, Barker R. Carbon-13-enriched carbohydrates. Preparation of erythrose, threose, glyceraldehyde, and glycolaldehyde with 13C-enrichment in various carbon atoms. Carbohydr Res 1979;72: 79–91. Corfu` NA, Tribolet R, Sigel H. Comparison of the self-association properties of the 58-triphosphates of inosine (ITP), guanosine (GTP), and adenosine (ATP)—further evidence for ionic interactions in the highly stable dimeric [H2(ATP)]24⫺ stack. Eur J Biochem 1990;191:721– 735. SantaLucia Jr J, Shen LX, Cai Z, Lewis H, Tinoco Jr I. Synthesis and NMR of RNA with selective isotopic enrichment in the bases. Nucleic Acids Res 1995;23:4913–4921.

NMR Study of Cryptococcus neoformans Metabolites 35. Mantsch HH, Smith ICP. Fourier-transformed 13C NMR spectra of polyuridylic acid, uridine, and related nucleotides—the use of 31POC13C couplings for conformational analysis. Biochem Biophys Res Commun 1972;46:808–815. 36. Mitchell TG, Perfect JR. Cryptococcosis in the era of AIDS—100 years after the discovery of Cryptococcus neoformans. Clin Microbiol Rev 1995;8:515–548. 37. Amendolia SR, Doppiu A, Ganadu ML, Lubinu G. Classification and quantitation of 1H NMR spectra of alditols binary mixtures using artificial neural networks. Anal Chem 1998;70:1249–1254. 38. Xu J, Yadan JC. Synthesis of L-(⫹)-ergothioneine. J Org Chem 1995;60: 6296–6301. 39. Weiss GH, Kiefer JE, Ferretti JA. Accuracy and precision in the estimation of internuclear distances for structure determinations. J Magn Reson 1992;97:227–234. 40. Pappagianis D, Marovitz R. Studies on ethanol production by Cryptococcus neoformans. Sabouraudia 1966;4:250–255. 41. Wong B, Perfect JR, Beggs S, Wright KA. Production of the hexitol D-mannitol by Cryptococcus neoformans in vitro and in rabbits with experimental meningitis. Infect Immun 1990;58:1664–1670. 42. Dawson DM, Taghavy A. A test for spinal-fluid alcohol in torula meningitis. N Engl J Med 1963;269:1424–1425. 43. Nilsson E, Edenius C, Lindgren JA. Ethanol affects leukotriene generation and leukotriene-induced functional responses in human polymorphonuclear granulocytes. Scand J Clin Lab Invest 1995;55:589–596.

453 44. Albertyn J, Hohmann S, Thevelein JM, Prior BA. GPD1, which encodes glycerol-3-phosphate dehydrogenase, is essential for growth under osmotic stress in Saccharomyces cerevisiae, and its expression is regulated by the high-osmolarity glycerol response pathway. Mol Cell Biol 1994;14:4135–4144. 45. Merkel GJ, Scofield BA. Comparison between in vitro glial cell adherence and internalization of non-encapsulated and encapsulated strains of Cryptococcus neoformans. J Med Vet Mycol 1994;32:361–372. 46. Gray BM, Dillon Jr HC, Briles DE. Epidemiological studies of Streptomyces pneumoniae in infants: development of antibody to phosphocholine. J Clin Microbiol 1983;18:1102–1107. 47. Zalavary S, Bengtsson T. Adenosine inhibits actin dynamics in human neutrophils: evidence for the involvement of cAMP. Eur J Cell Biol 1998;75:128–139. 48. Zhao ZQ, Todd JC, Sato H, Ma XL, Vinten-Johanson J. Adenosine inhibition of neutrophil damage during reperfusion does not involve K(ATP)-channel activation. Am J Physiol 1997;273:H1677–H1687. 49. Rodenas J, Mitjavila MT, Carbonell T. Nitric oxide inhibits superoxide production by inflammatory polymorphonuclear leukocytes. Am J Physiol 1998;274:C827–C830. 50. Naito Y, Yoshikawa T, Matsuyama K, Yagi N, Arai M, Nakamura Y, Kaneko T, Yoshida N, Kondo M. Neutrophils, lipid peroxidation and nitric oxide in gastric reperfusion injury in rats. Free Radic Biol Med 1998;24:449–502.