INFECTION AND IMMUNITY, June 1990, p. 1664-1670 Vol. 58, No. 60019-9567/90/061664-07$02.00/0Copyright © 1990, American Society for Microbiology

Production of the Hexitol D-Mannitol by Cryptococcus neoformansIn Vitro and in Rabbits with Experimental MeningitisBRIAN WONG,'* JOHN R. PERFECT,2 SANDRA BEGGS,' AND KATHLEEN A. WRIGHT2

Division of Infectious Diseases, Department of Internal Medicine, University of Cincinnati College of Medicine,Cincinnati, Ohio 45267,1 and Division of Infectious Diseases, Department of Medicine,

Duke University School of Medicine, Durham, North Carolina 277102

Received 5 December 1989/Accepted 2 March 1990

We studied the ability of Cryptococcus neoformans to produce the hexitol D-mannitol in vitro and in rabbitswith experimental meningitis. Twelve of twelve human isolates of C. neoformans produced D-mannitol in yeastnitrogen base plus 1% glucose and released D-mannitol into the medium. In a pilot study, pooled cerebrospinalfluid (CSF) from cortisone-treated rabbits given 3 x 107 C. neoformans H99 intracisternally contained moreD-mannitol (identified by gas chromatography and enzymatically) than CSF from normal controls orcortisone-untreated rabbits with self-limited meningitis. In a second experiment, cortisone-treated rabbitsgiven C. neoformans intracisternally had significantly higher CSF D-mannitol concentrations than controlsgiven cortisone alone at 4, 6, and 8 days after infection. Moreover, log1o CSF D-mannitol correlated well withloglo CSF CFU (r = 0.81) and log1o CSF cryptococcal antigen titers (r = 0.78). Lastly, the initial volume ofdistribution and elimination half-life of D-mannitol given intracisternally to normal rabbits suggested thatD-mannitol was distributed in total CSF and was removed by CSF bulk flow. Thus, C. neoformans producesD-mannitol in vitro and in vivo, and D-mannitol is a quantitative marker for experimental cryptococcalmeningitis. D-Mannitol produced by C. neoformans may also contribute to brain edema and interfere withphagocyte killing by scavenging hydroxyl radicals.

The prevalence of cryptococcal meningoencephalitis hasincreased markedly in recent years, primarily because thehuman immunodeficiency virus epidemic has greatly ex-panded the population of patients with profoundly deficientcellular immunity. A serious problem among patients withthe acquired immune deficiency syndrome (AIDS) and cryp-tococcosis is that most of these patients relapse after receiv-ing antifungal therapy that cures most patients with otherunderlying diseases (11, 28). This problem is compounded bythe fact that AIDS patients who will eventually relapsecannot be differentiated reliably from those who will not byclinical or available laboratory criteria. It has therefore beenproposed that all AIDS patients with cryptococcal meningi-tis should receive prolonged or even lifelong suppressivetherapy with the toxic antifungal agent amphotericin B (11,28, 29). Much of the drug toxicity, inconvenience, andexpense that this approach necessarily entails might beavoided if better quantitative methods for assessing theeffects of treatment on fungal load were available.

It has long been known that many fungi produce largeamounts of acyclic polyols (23), and recent studies haveshown that Candida and Aspergillus species produce suffi-cient amounts of their polyol metabolites in infected mam-malian hosts to cause increased body fluid concentrations.For example, the Candida species responsible for almost allcases of human infection produce large amounts of thepentitol D-arabinitol in vitro (2, 3, 10), and animals (24) andhumans (3, 9, 10, 20, 25, 27) with invasive candidiasis havemore D- or DL-arabinitol in the serum than uninfectedcontrols. It is also known that Candida albicans producesD-arabinitol in infected animals directly in proportion tofungal load (24). Similarly, the Aspergillus species responsi-ble for most cases of human aspergillosis produce largeamounts of the hexitol D-mannitol in vitro (S. J. Flaherty,

* Corresponding author.

E. M. Bernard, E. McKinney, B. Wong, and D. Armstrong,Abstr. Annu. Meet. Am. Soc. Microbiol. 1984, F7, p. 293).Moreover, rats with experimental aspergillosis have higherD-mannitol levels in serum and tissue than uninfected con-trols, and these levels correlate well with severity of infec-tion as assessed by histology (26).

In 1968, Onishi and Suzuki (14) reported that a strain ofCandida neoformans produced D-mannitol in culture, andthe study by Flaherty et al. (Abstr. Annu. Meet. Am. Soc.Microbiol. 1984) confirmed this observation in a few addi-tional strains. However, it is not known whether all oralmost all strains of C. neoformans produce D-mannitol invitro or whether any C. neoformans strain can produceappreciable amounts of D-mannitol in vivo. Therefore, wemeasured D-mannitol production by cultures of 12 humanisolates of C. neoformans. We also analyzed cerebrospinalfluid (CSF) samples from rabbits with experimental crypto-coccal meningitis to assess the ability of C. neoformans toproduce D-mannitol in vivo.

(This work was presented in part at the 29th InterscienceConference on Antimicrobial Agents and Chemotherapy,Houston, Tex., 17 to 20 September 1989.)

MATERIALS AND METHODSIn vitro studies. We studied the abilities of 12 human

isolates of C. neoformans to produce D-mannitol in vitro andto release it into the extracellular environment as follows.These strains included C. neoformans H99 (serotype A) and11 strains from the Mycology Laboratory at the Universityof Cincinnati Hospital (4 type A, 1 type B, 2 type C, 1 typeD, and 3 of unknown serotype). Each isolate was inoculatedat 105 yeast cells per ml in yeast nitrogen base plus 1%glucose. The cultures were incubated in air with slowshaking at 37°C, and samples were removed after 12, 24, 36,48, 72, and 96 h. The cells were enumerated in a hemacy-tometer. Colony counts were determined by culturing serial

1664

on January 13, 2019 by guesthttp://iai.asm

.org/D

ownloaded from

D-MANNITOL PRODUCTION BY C. NEOFORMANS 1665

10-fold dilutions on Sabouraud glucose agar at 37°C for 24 to48 h. Extracellular D-mannitol in the cell-free supernatantswas measured by gas chromatography (GC). The samplesobtained at 24, 48, and 96 h were also heated to 100°C for 10min to release intracellular polyols (2), and total D-mannitol(intracellular and extracellular) was measured.

Experimental cryptococcal meningitis. The ability of C.neoformans H99 to produce D-mannitol in vivo was studiedby analyzing CSF from rabbits with cryptococcal meningitis(16). New Zealand White rabbits (2 to 3 kg) were housed inindividual cages and received Purina rabbit chow and waterad lib. Beginning 1 day prior to infection and daily thereafter,each rabbit received either 2.5 mg of cortisone acetate per kgintramuscularly or no cortisone. On the day of infection, therabbits were anesthetized with 100 to 150 mg of ketamineand 15 to 25 mg of xylazine intramuscularly, and 0.3 ml of asuspension of 108 C. neoformans H99 yeast cells per ml of0.015 M phosphate-buffered saline was injected into thecisternum magnum. The rabbits were reanesthetized, andCSF was obtained by cisternal puncture at intervals there-after.

Fungal colony counts were determined by culturing serial10-fold dilutions of 0.1 ml of CSF in phosphate-bufferedsaline on Sabouraud agar (containing 100 p,g of chloram-phenicol per ml) at 37°C for 24 to 48 h. The capsularpolysaccharide antigen of C. neoformans was measured bytesting serial twofold dilutions of CSF in sterile water(starting at 1:20) by latex agglutination (Wampole Laborato-ries, Cranbury, N.J.). D-Mannitol was measured by GC inthe cell-free supernatants of the CSF samples.

In a pilot study, D-mannitol was measured in pooled CSFsamples (approximately five to eight rabbits per pool) thathad been stored at -70°C for as long as 3 years. Thesesamples included (i) CSF specimens from four groups ofuninfected normal rabbits, (ii) serial CSF specimens fromcortisone-treated rabbits with experimental cryptococcalmeningitis, and (iii) serial CSF specimens from two groups ofnon-cortisone-treated rabbits with experimental cryptococ-cal meningitis. CSF colony counts had previously beendetermined in the samples from the infected rabbits.

D-Mannitol was next measured in CSF from individualcortisone-treated rabbits with experimental cryptococcalmeningitis and from uninfected controls. Nine rabbits re-ceived cortisone daily and 3 x 107 C. neoformans H99 yeastcells intracisternally as described above, and 11 controlsreceived cortisone alone. Approximately 0.5 ml of CSF was

obtained from as many infected and control rabbits as was

technically feasible on the day before and 2, 4, 6, and 8 daysafter infection. Fungal colony counts and cryptococcal anti-gen titers were determined in the specimens from the in-fected rabbits.

Distribution and elimination of intracisternal D-mannitol.Six rabbits were anesthetized and given 0.275 p.mol (50 pug)of D-mannitol in 0.3 ml of phosphate-buffered saline byintracisternal injection. D-Mannitol was measured in CSFobtained after 1, 2, 4, 8, 24, and 48 h and in serum obtainedafter 2, 4, 8, 24, and 48 h. The apparent volumes ofdistribution and elimination half-lives were determined fromplots of log D-mannitol concentrations versus time (14).

D-Mannitol measurements. D-Mannitol in serum was mea-

sured by GC as described previously (26), and it was

measured in culture supernatants and CSF as follows. ax-Methylmannoside and a-methylglucoside (1.03 p.mol [200p.g] per ml of culture supernatant or 0.103 p.mol [20 p.g] perml of CSF) were added as internal standards to 0.2 ml ofculture supernatant or to 0.025 to 0.2 ml of CSF (depending

on the amount available). The samples were deproteinizedby adding 1.0 ml of acetone and by centrifugation at 1,000 xg for 5 min. The supernatants were evaporated to dryness inan N2 stream at 55°C, and the trimethylsilyl ether derivativeswere formed by adding 0.1 ml of trimethylsilylimidazole and0.1 ml of N,N-dimethylformamide to the sample residuesand by heating to 50°C for 15 min.The derivatives were extracted into 0.20 to 0.25 ml of dry

hexane. The extracts of derivatized culture supernatants (2RI) were analyzed by GC with a fused silica CP-Sil 5CBcolumn (25 m by 0.32 mm; 1.2-pm film thickness;Chrompack, Inc., Raritan, N.J.), helium carrier at 35 cm/s,splitless injector at 250°C (septum purge and inlet splittervalves closed for the first 30 s after infection), and a flameionization detector at 275°C. The column oven temperaturewas 120°C for 1 min after injection, after which it wasincreased by 20°C/min to 180°C and by 3°C/min to 240°C.The hexane extracts of derivatized CSF (2 pul) were analyzedwith a fused silica SPB-5 column (60 m by 0.32 mm; 1.0-p.mfilm thickness; Supelco, Inc., Bellefonte, Pa.), helium carrierat 40 cm/s, splitless injector at 250°C (septum purge and inletsplitter valves closed for the first 30 s after injection), and aflame ionization detector at 275°C. The oven temperaturewas 120°C for 1 min after injection, after which it wasincreased by 25°C/min to 195°C and by 3°C/min to 240°C.

D-Mannitol was quantified in culture supernatants bycomparing its peak area with that of a-methylmannoside andin CSF by comparing its peak area with that of the better-resolved internal standard. Relative response factors weredetermined daily from standard curves derived from fivesamples of culture medium or three samples of poolednormal rabbit CSF to which known amounts of D-mannitolwere added. The standard curves were linear throughout therelevant ranges.

Selected CSF samples were also analyzed after treatmentfor 30 min at 21°C with the Klebsiella pneumoniae NAD:oxidoreductase D-arabinitol dehydrogenase, rabbit musclelactate dehydrogenase, sodium pyruvate, and NAD as de-scribed previously (25). D-Arabinitol dehydrogenase con-verts D-mannitol to D-fructose by oxidizing the C2 hydroxylgroup (13), and it depletes serum of at least 98% of D-mannitol when NAD is regenerated by a coupled reaction(25, 26). Since the only known substrates of this enzyme areD-arabinitol and D-mannitol (13), disappearance of a peakthat coelutes with D-mannitol from D-arabinitol dehydroge-nase-treated specimens confirms that the compound of inter-est is D-mannitol (26).

Statistical methods. The Mann-Whitney test was used tocalculate the significance of differences between groups, andthe least-squares method was used for linear regressionanalyses.

RESULTS

In vitro studies. All of the C. neoformans strains studiedproduced D-mannitol in culture. The range of extracellularD-mannitol concentrations at 96 h was 140 to 2,700 p.M (25 to500 ,ug/ml). Differences in growth rates were responsible formuch of the variation in D-mannitol production; strains thatgrew faster generally produced and released more D-man-nitol than those that grew more slowly. D-Mannitol produc-tion by most strains continued after the cultures stoppedgrowing; indeed, several strains produced and released moreD-mannitol during the stationary phase than during logarith-mic growth. Lastly, higher proportions of total D-mannitolwere released into the culture medium as the cultures aged;

VOL. 58, 1990

on January 13, 2019 by guesthttp://iai.asm

.org/D

ownloaded from

1666 WONG ET AL.

8 r

0a

V)

U

0

0-J

0

.

5

1000

100

10

A IIi -

B

0.

1--.-.,i0 1 2 3 4

Days after inoculationFIG. 1. D-Mannitol production by C. neoformans in vitro.

Twelve strains of C. neoformans originally isolated from humanswere inoculated at 105 cells per ml in yeast nitrogen base plus 1%glucose and cultured at 37°C. (A) Mean ± standard error cell countsin samples removed at the intervals shown. (B) Mean ± standarderror concentrations of extracellular (0) and intracellular (0) D-mannitol per milliliter of culture medium.

extracellular D-mannitol represented 0.54 ± 0.24 of totalD-mannitol at 24 h, 0.83 ± 0.12 at 48 h, and 0.95 ± 0.19 at 96h (means ± standard deviation [SD]) (Fig. 1).

Identification of D-mannitol in CSF. When the pooled CSFspecimens were analyzed by GC, a compound with the sameretention time as authentic D-mannitol was found in muchlarger amounts in the samples from the cortisone-treated,infected rabbits than in those from normal uninfected ornon-cortisone-treated, infected rabbits. Treatment of theCSF samples with D-arabinitol dehydrogenase, lactate de-hydrogenase, NAD, and sodium pyruvate resulted in thealmost complete disappearance of the compound of interest,thus confirming its identification as D-mannitol (Fig. 2).

D-Mannitol production by C. neoformans in vivo. Amongthe rabbits studied in retrospect, the CSF D-mannitol con-centrations corresponded well with severity of infection.The CSF colony counts declined over time in the non-cortisone-treated rabbits given C. neoformans intracister-nally, and CSF D-mannitol concentrations in these rabbitswere no higher than the values in uninfected controls. Incontrast, the CSF colony counts and CSF D-mannitol con-centrations increased over time in the cortisone-treatedrabbits given C. neoformans intracisternally (Fig. 3). Theseresults suggested that C. neoformans produced D-mannitolin proportion to fungal load, but specimens from cortisone-treated controls were unavailable.

Therefore, we next studied individual cortisone-treatedrabbits given C. neoformans H99 intracisternally and corti-

cI)CD)0

CO)

c0

U1)

13

A

12

Ki3)

B

1 2

3

1 1~~~~~~

15 17 19 21

Retention Time (min)FIG. 2. Enzymatic degradation of D-mannitol in CSF. A com-

pound with the same retention time by capillary GC as that ofauthentic D-mannitol was found in increased amounts in pooled CSFfrom cortisone-treated rabbits 7 days after intracistemal administra-tion of 3 x 107 cells of C. neoformans H99 (A). This compounddisappeared almost completely following treatment with the K.pneumoniae enzyme D-arabinitol dehydrogenase, rabbit musclelactate dehydrogenase, NAD, and sodium pyruvate (B), thus con-firming that it was D-mannitol. Peaks: 1, a-methylmannoside; 2,a-methylglucoside; 3, D-mannitol.

sone-treated, uninfected controls. The fungal colony counts,cryptococcal antigen titers, and D-mannitol concentrationsin the infected and control rabbits are summarized in Fig. 4.The CSF D-mannitol concentrations rose substantially overtime in the infected rabbits but not in the controls. The meanCSF D-mannitol concentrations (+SD) in the infected andcontrol rabbits, respectively, were as follows: 1.91 ± 0.52and 1.84 ± 0.34 ,uM on day -1 (n = 9 and 11, respectively;P = 0.675); 3.01 ± 0.56 and 2.60 ± 0.43 ,uM on day 2 (n = 7and 9; P = 0.136); 7.03 ± 6.21 and 2.20 ± 0.40 ,uM on day 4(n = 6 and 8; P = 0.008); 36.9 ± 21.6 and 2.14 ± 0.38 ,uM onday 6 (n = 5 and 8; P = 0.002); and 53.7 ± 37.4 and 2.43 ±0.66 ,uM on day 8 after infection (n = 4 and 10; P = 0.001).Among the infected rabbits, the CSF D-mannitol concen-

trations correlated well with severity of infection as assessedby CSF fungal colony counts and CSF cryptococcal antigentiters (r = 0.81 for log10 D-mannitol versus log1o CFU, n = 21

-

4i

r

I

.1

INFECT. IMMUN.

7p~

6F

Ir

on January 13, 2019 by guesthttp://iai.asm

.org/D

ownloaded from

D-MANNITOL PRODUCTION BY C. NEOFORMANS 1667

E0~4-

U0

0r-

4

8A0

/ E4-0

0T-

31-

2 .

7

6

5

4

3

A

0cz

100

100

10I

1

B 0

0 0

o/~~~~~--

G1)

0)

0T-

4 1

21

100

0 2 4 6 8 10 3

Days after infectionFIG. 3. CSF colony counts and D-mannitol concentrations in

rabbits studied in retrospect. Groups of approximately five to eightrabbits received (i) either 2.5 mg of cortisone acetate per kg per dayintramuscularly beginning on day -1 (0) or no cortisone (A, [) and(ii) 3 x 107 C. neoformans H99 cells intracisternally on day 0. Theserial colony counts and D-mannitol concentrations in the pooledCSF samples are shown in panels A and B, respectively. The mean+ 2 SD CSF D-mannitol concentrations in four pools of normalrabbit CSF are also shown (0).

paired observations; r = 0.78 for log10 D-mannitol versusloglo antigen titer, n = 20) (Fig. 5).

Distribution and elimination of D-mannitol in CSF. Sincethe rate at which a compound appears in a body compart-ment cannot be estimated from its concentration unless itsdistribution and rate of clearance are also known, we studiedthe distribution and elimination of D-mannitol in CSF. Figure6 shows the CSF D-mannitol concentrations in four rabbitsgiven 0.275 Rmol of D-mannitol intracistemally; the resultsin two rabbits were excluded because sufficient sampleswere unobtainable or because of incomplete intracisternalinjection of D-mannitol. Clearance of exogenous D-mannitolfrom the CSF was biphasic; approximately 90% disappearedwithin 4 h (apparent volume of distribution = 2.9 ml;elimination half-life = 1.35 h), and the remainder disap-peared more slowly. The mean D-mannitol concentration(+SD) in serum was 1.1 ± 0.51 ,uM at 2 h and did not changesignificantly thereafter.

DISCUSSION

The objective of this study was to determine whether C.neoformans produces D-mannitol in culture and in infectedanimals. We found that cultures of 12 of 12 human isolates ofC. neoformans produced D-mannitol. D-Mannitol productionby C. neoformans differed from D-arabinitol production byCandida spp. in that most of the C. neoformans strains

0-5*E_

EILi

10

B

C

*

-1 2 4 6 8

Days after infectionFIG. 4. CSF colony counts (A), CSF cryptococcal antigen titers

(B), and CSF D-mannitol concentrations (C) in rabbits with crypto-coccal meningitis. Nine rabbits received cortisone acetate and C.neoformans H99 cells on day 0 (0) (see legend to Fig. 3 for detailedmethods), and 11 controls received cortisone alone (0). Data shownare means ± standard errors; *, P < 0.01 versus correspondingcontrols; n = 9, 7, 6, 5, and 4 infected and 11, 9, 8, 8, and 10 controlrabbits, respectively, on days -1, 2, 4, 6, and 8. undet, Antigen titerof <1:20.

produced more D-mannitol during the stationary than duringthe logarithmic phase of the cultures, whereas C. albicans isa net producer of D-arabinitol during the logarithmic phaseand a net utilizer of D-arabinitol thereafter (2). We also foundthat C. neoformans released most of the D-mannitol itproduced into the medium. C. albicans also releases most ofthe D-arabinitol it produces into the medium (2), but most ofthe D-mannitol found in 48-h cultures of several Aspergillusspecies was intracellular (Flaherty et al., Abstr. Annu. Meet.Am. Soc. Microbiol. 1984). These differences in the relativeproportions of intracellular and extracellular polyols mayexplain in part why rabbits with cryptococcal meningitis andrats with invasive candidiasis (24) had much higher bodyfluid polyol levels than rats with disseminated aspergillosis(26).The ability of C. neoformans H99 to produce D-mannitol

in vivo was studied in rabbits with experimental meningitis.In a pilot study of stored CSF specimens, the CSF D-

VOL. 58, 1990

6r

5 [

13,o-,..,Al--"21..

a

el-

1

0

on January 13, 2019 by guesthttp://iai.asm

.org/D

ownloaded from

1668 WONG ET AL.

100r A

10

1

S S

0.0*0

0

r = 0.81

B.

.

r = 0.78

2 3 4 5 6 7 8 2 3 4 5

Log10 cfu per ml Log 1o antigen titerFIG. 5. Evidence that logl0 CSF D-mannitol concentrations were highly correlated with the loglo CSF colony counts (A) and the logl0 CSF

cryptococcal antigen titers (B) in rabbits with experimental cryptococcal meningitis (see legend to Fig. 4 for methods).

mannitol leveN corresponded well with severity of infection.There was no more D-mannitol in the CSF of non-cortisone-treated rabbits with self-limited infection than in normalcontrols. In contrast, the CSF D-mannitol levels rose sub-stantially in cortisone-treated rabbits with progressive infec-tion. In a subsequent experiment in which individual rabbitsgiven cortisone and C. neoformans H99 were compared withcontrols given cortisone alone, the CSF D-mannitol concen-trations were much higher in the infected rabbits. Moreover,the CSF D-mannitol levels in infected rabbits rose directly inproportion to fungal load as assessed by CSF colony countsor CSF cryptococcal antigen titers.We also examined the distribution of D-mannitol within

the CSF and its rate of elimination to facilitate interpretationof CSF D-mannitol concentrations. We found that (i) approx-imately 90% of D-mannitol given intracisternally was clearedduring the first 4 h, (ii) the initial apparent volume ofdistribution approximated total CSF, and (iii) the initialelimination half-life was consistent with removal by bulkCSF flow. D-Mannitol was cleared more slowly after 4 h;possible explanations include uptake and slow release bytissues adjacent to the CSF or slow elimination from anotherbody compartment that is in equilibrium with the CSF.

100

:6

a0

E

10

-;_ _ --

- - - -3

0 4 8 24

Hours after injectionFIG. 6. D-Mannitol clearance from CSF. Normal rabbits re-

ceived 0.275 ,Imol of D-mannitol intracisternally, and D-mannitol inthe CSF was measured at the intervals shown. The CSF D-mannitolconcentrations declined rapidly for the first 4 h (solid regression line)and more slowly thereafter (dashed regression line). Data shown aremeans ± standard errors; n = 4. The standard errors were too smallto plot and remained within the symbols designating the means.

These results were similar to those of Prockop et al. (19),who showed that (i) 8.1% of ['4C]mannitol given intraven-tricularly or intracisternally to rabbits remained in the CSFafter 6 h and (ii) mannitol was cleared slightly faster thancompounds that are cleared by unidirectional CSF bulk flow.The authors concluded that mannitol exits the CSF primarilyby bulk flow but that some is eliminated by other means suchas "slow penetration into brain tissue." Although we did notmeasure D-mannitol clearance from the CSF of infectedrabbits, Prockop and Fishman (18) have shown that[14C]mannitol entered the CSF more rapidly after intrave-nous administration and disappeared from the CSF more

rapidly after intracisternal administration in dogs with exper-imental pneumococcal meningitis than in uninfected con-

trols. Taken together, our results and those of Prockop et al.imply that continuous production of large amounts of D-

mannitol within the central nervous system is necessary formaintenance of high CSF D-mannitol concentrations.These findings strongly suggest that C. neoformans H99

can produce large amounts of D-mannitol in vivo as well as

in vitro. An alternative explanation for the results is that theexcess D-mannitol in the CSF of infected rabbits may havebeen produced by the host as part of the inflammatory orimmune response. We consider this unlikely, because (i)D-mannitol is not a known product of mammalian metabo-lism, (ii) the CSF D-mannitol levels were highly correlatedwith fungal load, and (iii) intracisternal inoculation of C.neoformans H99 results in lower CSF leukocyte counts andless intense meningeal inflammation in rabbits given corti-sone than in cortisone-untreated controls (8, 16). On thebasis of these considerations, we conclude that the excessD-mannitol observed in the CSF of the infected rabbits wasproduced in vivo by C. neoformans.One implication of these results is that D-mannitol is a

quantitative marker for experimental cryptococcal meningi-tis in rabbits. It remains to be determined whether theseresults also apply to humans. It is not yet known, forexample, whether the variable abilities of different C. neo-formans strains to produce and release D-mannitol in vitroimply that different strains also produce variable amounts ofD-mannitol in vivo. We also do not yet know whetherD-mannitol production or release by C. neoformans orD-mannitol clearance by an infected host is influenced inhumans by factors such as host inflammatory or immuneresponses, renal failure (which is associated with increased

_011

:2_

I

INFECT. IMMUN.

I

on January 13, 2019 by guesthttp://iai.asm

.org/D

ownloaded from

D-MANNITOL PRODUCTION BY C. NEOFORMANS 1669

serum and CSF D-mannitol concentrations [17, 22]), orimmunosuppressive or antifungal drugs.

Nevertheless, it is likely that CSF D-mannitol measure-ments should be more useful for some purposes than forothers. Cryptococcosis differs from most other opportunisticmycoses in that reliable initial diagnostic methods are al-ready available. C. neoformans can be isolated from the CSFof almost all infected patients, and the latex agglutinationtest for cryptococcal capsular polysaccharide antigen is areliable and widely used initial diagnostic test. Moreover,increased CSF D-mannitol concentrations were not observedin the non-cortisone-treated infected rabbits or in the corti-sone-treated infected rabbits until 4 days after infection. Ittherefore appears unlikely that CSF D-mannitol measure-ments will improve our ability to diagnose cryptococcalmeningitis initially.On the other hand, it is very difficult to assess the severity

of cryptococcal meningitis by current diagnostic methods,especially during treatment (1, 5, 7). For example, steriliza-tion of the CSF within 2 weeks, pretreatment CSF crypto-coccal antigen titers, and changes in CSF cryptococcalantigen titers over time correlated poorly with final outcomein a recent large treatment trial in patients with cryptococcalmeningitis and a variety of underlying diseases (7). Persis-tence of viable fungi outside the subarachnoid space andprolonged antigen shedding by dead or dying fungi wereprobably responsible for these findings.

In contrast, only actively metabolizing cryptococci pro-duce D-mannitol, and dead fungi are unlikely to releasesubstantial amounts of D-mannitol because little is con-served intracellularly. Also, since D-mannitol is clearedrapidly from the CSF, changes in fungal viability or meta-bolic activity should result in prompt changes in CSF D-mannitol concentrations. Thus, serial CSF D-mannitol mea-surements may provide a more precise means for quantifyingthe load of metabolically active fungi than is currentlyavailable, at least when the CSF D-mannitol concentrationsare elevated initially. This approach should be most useful inpatients in whom treatment is least likely to be successful(such as those with AIDS). In such patients, serial CSFD-mannitol measurements might eventually be used to indi-vidualize the dosages or durations of initial or suppressiveantifungal therapy. This approach may also be helpful forassessing the likelihood of relapse or for comparing theeffects of different forms of treatment on fungal viability ormetabolic activity or both.The results of the present study may also have pathoge-

netic implications. We have shown that D-mannitol accumu-lates in the CSF of infected rabbits despite an efficientclearance mechanism. Since D-mannitol is not metabolizedby mammalian cells and since it crosses the blood-brainbarrier by passive diffusion alone (18), it is probably clearedless efficiently from brain tissue than from CSF. This sug-gests that there may be much more D-mannitol in tissuesimmediately surrounding actively metabolizing cryptococcithan in CSF. One consequence of D-mannitol accumulationin brain tissue may be increased tonicity and edema. Cere-bral edema is common in cerebral cryptococcosis (4), but itspathogenesis is poorly understood. Our results suggest thatfungal production of a low-molecular-weight solute that isnot easily cleared may be a contributing factor. It is alsoknown that oxygen-dependent products of phagocytic cells(e.g., H202) kill C. neoformans in vitro (6). Moreover,D-mannitol is an effective scavenger of hydroxyl radicals,and D-mannitol inhibits the abilities of hydroxyl radicalsgenerated by xanthine oxidase and acetaldehyde to kill

Staphylococcus aureus (10) and of monocytes and macro-phages to kill Toxoplasma gondii (12). Thus, accumulationof D-mannitol in infected tissues may also contribute topathogenesis by interfering with optimal killing of C. neofor-mans by phagocytes.

In summary, we have shown that C. neoformans canproduce large amounts of D-mannitol in vitro and in vivo.These results suggest that serial CSF D-mannitol measure-ments may be useful for quantifying fungal load in crypto-coccal meningitis and that fungal D-mannitol production maycontribute to the pathogenesis of cryptococcosis.

ACKNOWLEDGMENTSJudith C. Rhodes supplied C. neoformans isolates used in this

study. Marcia Hobbs and Miguel Castellanos provided valuabletechnical assistance.

This study was supported in part by Public Health Service grantsAI-23938 and AI-28392 from the National Institute of Allergy andInfectious Diseases.

LITERATURE CITED1. Bennett, J. E., W. E. Dismukes, R. J. Duma, G. Medoff, M. A.

Sande, H. Gallis, J. Leonard, B. T. Fields, M. Bradshaw, H.Haywood, Z. A. McGee, T. R. Cate, C. G. Cobbs, J. F. Warner,and D. W. Alling. 1979. A comparison of amphotericin B aloneand combined with flucytosine in the treatment of cryptococcalmeningitis. N. Engl. J. Med. 301:126-131.

2. Bernard, E. M., K. J. Christiansen, S. Tsang, T. E. Kiehn, andD. Armstrong. 1982. Rate of arabinitol production by pathogenicyeast species. J. Clin. Microbiol. 16:353-359.

3. Bernard, E. M., B. Wong, and D. Armstrong. 1985. Stereoiso-meric configuration of arabinitol in invasive candidiasis. J.Infect. Dis. 151:711-715.

4. Diamond, R. D. 1985. Cryptococcus neoformans, p. 1460-1468.In G. L. Mandell, R. G. Douglas, Jr., and J. E. Bennett (ed.),Principles and practice of infectious diseases. John Wiley &Sons, Inc., New York.

5. Diamond, R. D., and J. E. Bennett. 1974. Prognostic factors incryptococcal meningitis: a study in 111 cases. Ann. Intern.Med. 80:176-181.

6. Diamond, R. D., R. K. Root, and J. E. Bennett. 1972. Factorsinfluencing killing of Cryptococcus neoformans by human leu-kocytes in vitro. J. Infect. Dis. 125:367-376.

7. Dismukes, W. E., G. Cloud, H. A. Gallis, T. M. Kerkering, G.Medoff, P. C. Craven, L. G. Kaplowitz, J. F. Fisher, C. R.Gregg, C. A. Bowles, S. Shadomy, A. M. Stamm, R. B. Diasio, L.Kaufman, S. J. Soong, W. C. Blackwelder, and the NationalInstitute of Allergy and Infectious Diseases Mycoses Study Group.1987. Treatment of cryptococcal meningitis with combinationamphotericin B and flucytosine for four as compared with sixweeks. N. Engl. J. Med. 317:334-341.

8. Durack, D. T., and J. R. Perfect. 1985. Experimental model ofcryptococcal meningitis: a model for study of defense mecha-nisms, p. 71-82. In M. S. Sande, A. L. Smith, and R. K. Root(ed.), Bacterial meningitis. Churchill Livingstone, New York.

9. Gold, J. W. M., B. Wong, E. M. Bernard, T. E. Kiehn, and D.Armstrong. 1983. Serum arabinitol concentrations and arabini-tol/creatinine ratios in invasive candidiasis. J. Infect. Dis.147:504-514.

10. Kiehn, T. E., E. M. Bernard, J. W. M. Gold, and D. Armstrong.1979. Candidiasis: detection by gas-liquid chromatography ofD-arabinitol, a fungal metabolite, in human serum. Science206:577-580.

11. Kovacs, J. A., A. A. Kovacs, M. Polis, C. Wright, V. J. Gill,C. U. Tuazon, E. P. Gelmann, H. C. Lane, R. Longfield, G.Overturf, A. M. Macher, A. S. Fauci, J. E. Parrillo, J. E.Bennett, and H. Masur. 1985. Cryptococcosis in the acquiredimmunodeficiency syndrome. Ann. Intern. Med. 103:533-538.

12. Murray, H. W., B. Y. Rubin, S. M. Carriero, A. M. Harris, andE. A. Jaffee. 1983. Human mononuclear phagocyte antiproto-zoal mechanisms: oxygen-dependent vs oxygen-independent

VOL. 58, 1990

on January 13, 2019 by guesthttp://iai.asm

.org/D

ownloaded from

1670 WONG ET AL.

activity against intracellular Toxoplasma gondii. J. Immunol.134:1982-1988.

13. Neuberger, M. S., R. A. Patterson, and B. S. Hartley. 1979.Purification and properties of Klebsiella pneumoniae D-arabitoldehydrogenase. Biochem. J. 183:31-42.

14. Onishi, H., and T. Suzuki. 1968. Production of D-mannitol andglycerol by yeasts. Appl. Microbiol. 16:1847-1852.

15. Park, G. D. 1986. Pharmacokinetics, p. 67-102. In R. Spector(ed.), The scientific basis of clinical pharmacology. Principlesand examples. Little, Brown, & Co., Boston.

16. Perfect, J. R., D. R. Selwyn, M. B. Lang, and D. T. Durack.1980. Chronic cryptococcal meningitis. A new experimentalmodel in rabbits. Am. J. Pathol. 101:177-193.

17. Pitkanen, E., A. Bardy, A. Pasternack, and C. Servo. 1976.Plasma, red cell and cerebrospinal fluid concentrations of man-nitol and sorbitol in patients with severe chronic renal failure.Ann. Clin. Res. 8:368-373.

18. Prockop, L. D., and R. A. Fishman. 1968. Experimental pneu-mococcal meningitis. Permeability changes influencing the con-centration of sugars and macromolecules in cerebrospinal fluid.Arch. Neurol. 19:449-463.

19. Prockop, L. D., L. S. Schanker, and B. B. Brodie. 1962. Passageof lipid-insoluble substances from cerebrospinal fluid to blood.J. Pharmacol. Exp. Ther. 135:2676-2670.

20. Roboz, J., R. Suzuki, and J. F. Holland. 1980. Quantification ofarabinitol in serum by selected ion monitoring as a diagnostictechnique in invasive candidiasis. J. Clin. Microbiol. 12:594-602.

21. Rosen, H., and S. J. Klebanoff. 1979. Anion-generating system.A model for the polymorphonuclear leukocyte. J. Exp. Med.

149:27-39.22. Servo, C., J. Palo, and E. Pitkanen. 1977. Polyols in the

cerebrospinal fluid and plasma of neurological, diabetic anduraemic patients. Acta Neurol. Scand. 56:111-116.

23. Touster, O., and D. R. D. Shaw. 1962. Biochemistry of theacyclic polyols. Physiol. Rev. 42:181-225.

24. Wong, B., E. M. Bernard, J. W. M. Gold, D. Fong, A. Silber,and D. Armstrong. 1982. Increased arabinitol levels in experi-mental candidiasis in rats: arabinitol appearance rates, arabini-tol/creatinine ratios, and severity of infection. J. Infect. Dis.146:346-352.

25. Wong, B., and K. L. Brauer. 1988. Enantioselective measure-ment of fungal D-arabinitol in the sera of normal adults andpatients with candidiasis. J. Clin. Microbiol. 26:1670-1674.

26. Wong, B., K. L. Brauer, R. R. Tsai, and K. Jayasimhulu. 1989.Increased amounts of the Aspergillus metabolite D-mannitol intissues and serum of animals with experimental aspergillosis. J.Infect. Dis. 160:95-103.

27. Wong, B., and M. Castellanos. 1989. Enantioselective measure-ment of the Candida metabolite D-arabinitol in human serumusing multidimensional gas chromatography and a new chiralphase. J. Chromatogr. 495:21-30.

28. Zuger, A., E. Louie, R. S. Holzman, M. S. Simberkoff, and J. J.Rahal. 1986. Cryptococcal disease in patients with the acquiredimmunodeficiency syndrome. Diagnostic features and outcomeof treatment. Ann. Intern. Med. 104:234-240.

29. Zuger, A., M. Schuster, M. S. Simberkoff, J. J. Rahal, and R. S.Holzman. 1988. Maintenance amphotericin B for cryptococcalmeningitis in the acquired immunodeficiency syndrome (AIDS).Ann. Intern. Med. 109:592-593.

INFECT. IMMUN.

on January 13, 2019 by guesthttp://iai.asm

.org/D

ownloaded from