Vol. 59, No. 4

INFECTION AND IMMUNITY, Apr. 1991, p. 1352-1358 0019-9567/91/041352-07$02.00/0 Copyright C 1991, American Society for Microbiology

Pathophysiology of Citrobacter diversus Neonatal Meningitis: Comparative Studies in an Infant Mouse Model ALEJANDRO L. SORIANO, 2t ROBERT G. RUSSELL,2'3 DAVID JOHNSON,4 ROSANNA LAGOS,2t IANCU SECHTER,5 AND J. GLENN MORRIS, JR.2* Division of Neonatology, Department of Pediatrics,' Division of Geographic Medicine, Department of Medicine, and Center for Vaccine Development,2 Department of Pathology and Program of Comparative Medicine,3 and Division of Infectious Diseases, Department of Medicine,4 University of Maryland School of Medicine, Baltimore, Maryland 21201,

and Government Central Laboratories, Ministry of Health, Jerlusalem, Israel5 Received 7 November 1990/Accepted 21 January 1991

Citrobacter diversus is a cause of devastating neonatal meningitis, with illness characterized by formation of multiple brain abscesses. We developed an infant mouse intracranial inoculation model to evaluate the pathophysiology of C. diversus neonatal infections. Eighteen of 26 strains inoculated intracranially at a dose of ca. 3.3 x 103 CFU caused >50% mortality in 2-day-old mice. No correlation was seen between the epidemiologic characteristics of a strain and its rate of mortality. When seven C. diversus isolates (four isolates from patients with meningitis, three from non-central nervous system [CNS] sites) were further evaluated, mortality was significantly correlated with bacteremia. The initial lesion in the CNS was a suppurative ventriculitis beginning 1 to 2 days postinoculation. Subsequent ventriculomegaly was associated with ventriculitis and periventricular abscessation. Brain lesions were seen with all strains, although strains of low virulence (as measured by having no bacteremia and low mortality) caused less-severe damage. An age-related susceptibility to C. diversus brain lesions was demonstrated, with 5-day-old mice showing a significant reduction in, and 8-day-old mice being apparently resistant to, infection and CNS damage. Our data indicate that C. diversus has a propensity to cause abscess formation in the neonatal mouse brain, with characteristic pathologic findings; however, the factors that determine whether a strain will cause meningitis in a human infant remain to be identified.

determining whether there was an age-related susceptibility to infection in infant mice, as has been described in humans (11, 22).

Citrobacter diversus is a gram-negative bacillus that is a rare but devastating cause of neonatal meningitis, occurring sporadically or in epidemics (5, 10, 12, 19, 22, 25, 31, 32). The fatality rate of patients with meningitis is reported to be 34%, with 90% of survivors having various degrees of mental retardation (11, 13). In contrast to patients with other bacterial causes of neonatal meningitis such as Escherichia coli K-1 strains and group B Streptococcus strains, 77% of C. diversus meningitis patients develop brain abscesses (11). Although the clinical and epidemiologic features of C. diversus infections in neonates have been well described, we lack a good understanding of the associated virulence mechanisms. The importance of putative virulence factors (16) in development of disease remains to be determined, and the role of host factors, such as age, is unclear. The pathogenic mechanisms involved in cerebral abscess formation have also been controversial (7, 15, 17). To assess the pathophysiology of C. diversus neonatal meningitis, we and others have investigated a number of possible animal models (17, 23a); we report here results obtained after intracranial inoculation of the bacterium into infant mice. With our infant mouse model we sought to determine whether there were differences among strains in pathogenicity and whether these differences correlated with epidemiologic background or with standard markers such as serotype, biotype, plasmid profile, or outer membrane protein (OMP) profile. We were also interested in evaluating progression of the central nervous system (CNS) lesions and

MATERIALS AND METHODS

Bacterial strains. Twenty-six isolates of C. diversus were examined (Table 1). This includes 11 isolates from cases of neonatal meningitis, 3 isolates associated with bacteremia, and 12 isolates cultured from other sites. The organisms were obtained from different geographic locations within the United States and Israel. E. coli K-12 strain HB101 was used as a negative control. All isolates were stored at -70°C in L broth with 15% glycerol. C. diversus isolates were serotyped for both 0 and H antigens at the Government Central Laboratories, Jerusalem, Israel, by previously described methods (1). Biotyping was based on the fermentation of dulcitol, rhamnose, sucrose, and sorbose (26). Plasmid profiles were determined by using an alkaline plasmid extraction technique (2). For selected strains, OMPs were extracted and prepared by using a modification of the method described by Sears and Richardson (21, 28). Strains for OMP extraction were grown in L broth for 16 h at 37°C; proteins were visualized by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) by using discontinuous gels (30) with stacking and resolving gels of 4% (pH 6.9) and 11% (pH 8.7), respectively. Intracranial challenge of neonatal mice. The study was conducted in four parts. A preliminary experiment was conducted to establish a dose-response curve and to determine an appropriate inoculum size for subsequent studies. Two-day-old mice were challenged with C. diversus 52

Corresponding author. t Present address: Trover Clinic, Madison, KY 42431. t Present address: Hospital Roberto del Rio, Santiago, Chile. *

1352

VOL. 59,

PATHOPHYSIOLOGY OF C. DIVERSUS NEONATAL MENINGITIS

1991

1353

TABLE 1. Characteristics of 26 C. diversus strains studied and their respective mortality outcome Strain

Meningitis-associated strains 52 4511 836 4408 4406 666-546 19266 4409

6947 3078 871-3851 Septicemia-associated strains 22525 53114 4963

Source

Biotype

Meningitis/Maryland Meningitis/Kentucky Meningitis/Maryland Meningitis/Louisiana Meningitis/Delaware Meningitis/Israel Meningitis/Maryland Meningitis/Texas Meningitis/Maryland Meningitis/Texas Brain abscess/Israel

E D E A E D E E A A E

Blood/Maryland Blood/Maryland Blood/Oklahoma

E C E

Plasmid profilea

Serotype

Mortalityb

3:a:2 3:a:3:a:2 2:a:2 3:a:2 3:a:2 3:a:2 2:a:2

5:i:2:a:2 3:a:2

16/21 (76) 9/9 (100) 11/11 (100) 12/12 (100) 12/12 (100) 15/15 (100) 11/11 (100) 5/40 (13) 6/9 (67) 3/12 (25) 4/10 (40)

3:a:15:a:2 1:a:-

8/10 (80) 8/8 (100) 6/18 (33)

I XX I XXI

Il XXII II XXIII II XXIV XXV

1I II II

Strains isolated from other sites 20871 C II 15' 13/13 (100) Cervix/Maryland II C 15:a:2 11/11 (100) 896-109 Vagina/Israel XV 15C 10/10 (100) C 26357 Rectum/Marylandd II 3:a:2 11/11 (100) 30488 E Wound/Maryland II 4/10 (40) E 14545 Breast/Maryland XXVI 11:a:2 9/9 (100) 75 C Rectum/Marylandd XXVII 5/13 (38) A 1:c:745-72 Eye/Israel XIII 11/11 (100) 15:a:2 C 27447 Rectum/Marylandd XVIII 15' 10/10 (100) Umbilical cord/Marylandd C 26359 XXVIII 4/11 (40) 12:a:2 D 663-972 Eye/Israel IV 3/11 (27) A 2:a:2 896-2744 Urine/Israel 11/11 (100) I 3:a:2 E 2742 Hands/Marylandd.e a Designation of plasmid profile follows that used previously by Morris et al. (22); strains with profiles that had not been previously identified were numbered consecutively, beginning with XX. b Number of pups that died/number tested. Numbers in parentheses represent percent mortality. ' Data on flagellar antigen type not available. d Asymptomatic patients. e Strain associated with meningitis outbreak.

(isolated from an infant with meningitis during a major nursery outbreak in Baltimore [19]) in doses of 101 to 107 CFU in 10-fold serial dilutions; each dilution was administered to 10 mice from a single litter. Experiment 1 was subsequently undertaken with all 26 C. diversus isolates to determine the mortality rate with each strain in 2-day-old infant mice. In experiment 2, we selected seven isolates from experiment 1 to evaluate bacteremia and brain pathology in 2-day-old mice. Experiment 3 evaluated the effects of age on mortality and the development of CNS lesions in 5- and 8-day-old pups by using C. diversus 52. Challenge organisms were prepared by inoculating single colonies from an overnight Luria agar (L agar) plate into L broth and incubating the cultures for 18 h at 37°C. Bacterial concentrations of the inocula were determined by relating optical density at 600 nm to CFU per milliliter; all determinations were confirmed by plate count assays done in duplicate. Pathogen-free pregnant (12 to 14 days gestation) female CD-1 mice were obtained from Charles River Breeding Laboratories. The dams were examined twice daily to detect births. The time of day (morning or late afternoon) that litters were subsequently challenged with the inoculum at 2, 5, or 8 days old corresponded to whether pups were delivered during the day or overnight so that the actual age of litters differed by no more than 12 h at the time of challenge. All mice in a litter were inoculated with the same strain (or,

in dose-response studies, with the same dilution of the same strain) and were then returned to their natural mothers. One to three litters were inoculated with each strain or dilution in each of the experiments. Bacteria in a volume of 100 p.1 were inoculated intracranially into the right parietal area by penetrating the skull with a 27-gauge needle attached to a 1-ml tuberculin syringe. The site of inoculation was 3 mm behind the right eye and 2 mm from the sagittal suture. In all experiments, inoculations were performed by the same investigator (A.L.S.). For studies of mortality, animals were observed for a maximum of 7 days. Blood cultures and histopathology. Blood cultures were obtained at the time of sacrifice by cardiac puncture after the skin was disinfected with povidone and then with 70% isopropyl alcohol. Samples (100 ,ul) of blood were inoculated onto L agar plates. Identity of isolates from positive blood cultures was confirmed by use of the API system (Analytab

Products, Plainview, N.Y.). For pathologic examination, mice

were decapitated and the brains were collected and fixed in 10% neutral buffered Formalin. Coronal sections of the brain were processed for paraffin embedding. The 5-p.m sections were stained with hematoxylin and eosin. Histologic interpretation was undertaken by one investigator (R.G.R.) who was blinded to the identification of the isolate and the age of the mouse at the time of inoculation and sacrifice. To histologically evaluate the extent of the inflammation and damage in the brain

1354

SORIANO ET AL.

INFECT. IMMUN. TABLE 3. Rates of mortality and bacteremia for seven C. diversus isolates

10

1

100

1000

10000

100000

1000000

10000000

inoculum size (CFU) FIG. 1. Results of initial dose-response studies with C. diversus 52, showing percent mortality as a function of inoculum size. Ten mice from a single litter were inoculated with each dilution tested.

lesions and to assess sampling variability, serial sections were prepared from the brains of selected mice with representative lesions. RESULTS

Preliminary dose-response studies with C. diversus 52. The

results of the preliminary experiment using intracranial challenge of strain 52 in 2-day-old mice are shown in Fig. 1. The dose of 3.3 x io3 used in all subsequent experiments was the calculated 85% lethal dose. Experiment 1: mortality in 2-day-old mice (26 isolates). After intracranial inoculation in 2-day-old mice, 15 of the 26 C. diversus isolates caused 100% mortality, 3 isolates caused mortality of 67 to 80%, and the remaining 8 isolates had mortalities of 6 to 40% (Table 1). There was no correlation between mortality and the site from which the isolate was recovered (CNS versus blood versus other sites), the geographic location from which the isolate was obtained, or the plasmid profile of the isolate (Tables 1 and 2). Strains of TABLE 2. Mortality outcome of 26 C. diversus isolates by epidemiologic characteristics and source Epidemiologic characteristic or source

Mortality in:

50o

3 0 1 4

2 7 2 7

2 3 1 0 0 1 0 1

0 1 9 1 1 0 6 0

Strain

Source

Bac-

Mortalityb

Meningitis-associated strains 52 19266 4409 6947

Meningitis/Maryland Meningitis/Maryland Meningitis/Texas Meningitis/Maryland

6/6 3/5 0/6 4/5

21/21 (100) 12/12 (100) 2/14 (14) 7/8 (88)

Strains isolated from other sites 896-109 75 663-972

Vagina/Israel Rectum/Maryland Eye/Israel

3/4 3/5 0/6

12/12 (100) 15/15 (100) 0/8 (0)

a Number of pups with bacteremia/number tested. Results are from animals sampled on days 2, 4, and 6. b Number of pups that died/number tested. Numbers in parentheses represent percent mortality. As these experiments were conducted separately from the initial studies of mortality, there are differences in the reported mortality rates, compared with those shown in Table 1.

biotype C, serotype 15, were significantly more likely to cause 100% mortality (P = 0.01; Fischer's exact test). However, none of the biotype C, serogroup 15 strains were from patients with meningitis. Experiment 2: mortality, bacteremia, and brain histopathology in 2-day-old mice (seven isolates). (i) Bacteremia and mortality. Seven C. diversus isolates were selected for further study (Table 3). Mortality studies were repeated for these isolates. Results were comparable to those obtained in experiment 1: strains 52, 19266, 6947, 896-109, and 75 caused high mortality, while minimal mortality was seen with strains 4409 and 663-972 (Fig. 2). When analyzed statistically, significant heterogeneity in mortality rates was observed among litters inoculated with the same strain for strains 52 and 663-972. For strain 52, one litter (inoculated as part of experiment 1) had 6 deaths of 11 infants, while the remaining litters tested had 10 deaths of 10 infants (experiment 1), 13 deaths of 13 infants (experiment 2), and 8 deaths of 8 infants (experiment 2) (P < 0.001, G test for heterogeneity). For strain 663-972, there were 4 deaths of 11 infants (experiment 1) and 0 deaths of 8 infants (experiment 2) (P = 0.025, G test for heterogeneity). However, even with this heterogeneity, mortality rates for litters inoculated with the high mortality

Biotype A

C D E

Serogroup 01 02 03 05 011 012 015 Unknown

Source CNS Other sites

3 5

8 10

52

19266

4409

6947

896-109

75

663-972

strain number FIG. 2. Cumulative mortality (%) in 2-day-old mice, by day after inoculation, for seven selected C. diversus strains. Symbols: _, days 1 to 2; 1 , days 3 to 4; , days 5 to 6.

PATHOPHYSIOLOGY OF C. DIVERSUS NEONATAL MENINGITIS

VOL. 59, 1991 ,0

P

1355

110

W

o1

45bc

'PII

.&

.,. 'In p

A

IF* -w

9p

o

4, f

a-

Vo v

dir~~~~ 4' I -

-_

lb

FIG. 3. Response of 2-day-old mouse 2 days postinoculation with C. diversus 896-109. Moderate numbers of neutrophils are present in the lateral ventricles. The ependyma is intact. Hematoxylin and eosin stain. Magnification, x950.

strain 52 were significantly higher than rates for litters inoculated with low mortality strain 4409 (P = 0.03) or 663-972 (P = 0.04, t test using Arcsin transformed data). Additional pups were sacrificed on days 2, 4, and 6 postinoculation, and blood samples were obtained for culture. Bacteremia, if present, was always detected on day 2 postinoculation; only animals inoculated with strain 52 were still bacteremic by day 6. The two strains that did not cause bacteremia (4409 and 663-972) were the two strains that caused minimal mortality; a significant correlation was seen between overall rates of bacteremia and mortality (r = 0.8, P = 0.01; Spearman correlation coefficient). (ii) Brain histopathology. On days 1 to 2, mice inoculated with 4409 and 663-972 (the two low mortality strains) had localized acute necrosis of the cortex at the site of inoculation without infiltration of neutrophils or abscessation. Mice inoculated with strains 52, 19266, 6947, 896-109, and 75 (the high mortality strains) had acute ventriculitis with numerous neutrophils in the lateral ventricles (Fig. 3) and suppurative meningitis. In addition, mice inoculated with strains 19266 and 6947 developed severe purulent ventriculitis with abscessation, destruction of the ependyma, ventriculomegaly, and infiltration of neutrophils into the adjacent parenchyma with formation of bilateral periventricular abscesses (Fig. 4). Bacteria were located intracellularly and extracellularly. On days 3 to 4 postinoculation, there were no lesions in the brains of mice inoculated with strain 663-972. Mice inoculated with strain 4409 showed unilateral abscessation involving the lateral ventricle and expanding locally into the adjacent parenchyma. This was associated with destruction of the ependyma and ventriculomegaly. In contrast, mice inoculated with the other five strains (52, 896-109, 19266, 6947, and 75) showed severe bilateral abscessation involving the ventricles and expanding into the adjacent parenchyma accompanied by ventriculomegaly. In addition to abscessation, strains 52 and 6947 also caused extensive adjacent cortical damage with edema, malacia, and cavitation (Fig. 5). On days 5 to 6 postinoculation, mice inoculated with strain 663-972 had localized abscessation involving the cortex and

FIG. 4. Response of 2-day-old mouse 2 days postinoculation with C. diversus 6947. Marked ventriculomegaly is seen, with destruction of the ependyma and extension of the severe purulent inflammation into the periventricular brain parenchyma. There is well-demarcated brain abscessation (arrows). Hematoxylin and eosin stain. Magnification, x300.

ventricles (no abscessation was found in mice challenged with this strain and examined at earlier time intervals). No brain lesions were found in surviving mice inoculated with strain 4409. There were no survivors from the group of mice inoculated with strain 896-109. The mice inoculated with three (19266, 6947, and 75) of the remaining four strains showed well-demarcated periventricular abscesses with early glial scarring at the perimeter (Fig. 6). The brains of mice inoculated with strain 52 showed severe purulent exudate in the ventricles, extensive cortical damage and abscessation accompanied by destruction of the ependyma

I

.*,

8 4

,,

.

U.t..'-,;. *

*.

t I

-.

.

* .L .

.

a

- .um. 0

4

.

5A

S

..rZ w ..

;8i '*

~

a

_w _d

,-

~

' 1:

:X

*

.

?

*

ALi

.&1I*

Is

is

*.

w b

. .

A

X

zM z W IL,_ *t t *

^

4

FIG. 5. Response of 2-day-old mouse 3 days postinoculation with C. diversus 52. Malacia and neutrophil infiltration in the cerebral cortex are present adjacent to abscessation. Hematoxylin and eosin stain. Magnification, x950.

1356

t

f

INFECT. IMMUN.

SORIANO ET AL. '.,o i

.

I'

* y

-..F

4"

wE

%4**

7:V E~~ * O t *R

5

#

's

so

i ~~~A.OA

dS..^@'4@

S@4tt

FIG. 6. Response of 2-day-old mouse 6 days postinoculation with C. diversus 6947. Glial scar formation (arrows) is present at the perimeter of a well-demarcated brain abscess in the ventricles and adjacent parenchyma. Hematoxylin and eosin stain. Magnification, x950.

and necrosis, and large numbers of intracellular and extracellular bacteria (Fig. 7). Adjacent to the abscess there was malacia and gliosis without evidence of early glial scarring. The severity of meningitis was variable in the animals examined. In some mice, no meningitis was observed. In others, mild to moderate suppurative meningitis involved the area immediately overlying the abscess and was both dorsal and dependent in the brain stem and cerebellar areas. Except for focal malacia at the site of inoculation, no brain lesions were noted in the control mice inoculated with sterile media or E. coli HB1O1 in doses as high as i07 CFU. The ability of the different strains to cause bacteremia, mortality, and CNS lesions did not correlate with the presence of any

.W:A

A

A.

il t

ddgsLl:

FIG. 7. (A) Large numbers of intracellular bacteria (arrowheads) and (B) extracellular bacteria (arrows) in the brain of a 2-day-old mouse 6 days postinoculation with C. diversus 52. The mouse had a severe purulent ventriculitis with abscessation and malacia in the adjacent parenchyma. Hematoxylin and eosin stain. Magnification, x 5,000.

FIG. 8. Response of an 8-day old mouse 4 days postinoculation with C. diversuis 52 showing a localized meningeal abscess. Hematoxylin and eosin stain. Magnification, x300.

specific OMP band on SDS-PAGE or with any one plasmid or plasmid profile (data not shown). Experiment 3: mortality and brain histopathology in 5- and 8-day-old mice inoculated with C. diversus 52. (i) Mortality and bacteremia. Six (26%) of 23 mice (composing two litters) inoculated with strain 52 at the age of 5 days died. The mortality rates among these litters were significantly lower than the rates among litters inoculated with this strain at the age of 2 days (P = 0.012; t test using Arcsin transformed data). Bacteremia was evaluated in six mice inoculated at the age of 5 days; only one was bacteremic. There was no mortality (14 mice) or bacteremia (3 mice) among mice inoculated with strain 52 at the age of 8 days. Mortality rates among litters inoculated at the age of 8 days were significantly lower than those among litters inoculated at the age of 2 days (P = 0.008; t test using Arcsin transformed data). (ii) Brain histopathology. At 2 days postinoculation of 5-day-old mice with strain 52, there was marked bilateral suppurative ventriculitis and ventriculomegaly with large numbers of bacteria. In contrast to the extensive abscessation and malacia observed in 2-day-old mice on days 3 to 6 postinoculation, the mice inoculated at the age of 5 days showed localized necrosis in the superficial cortex at the site of inoculation accompanied by diffuse moderate meningitis at days 3 to 4 postchallenge. The focal lesion was healing with scarring on days 5 to 6 postinoculation, and there was meningitis with infiltration of low to moderate numbers of neutrophils and mononuclear cells. As in 2-day-old mice, no brain lesions other than focal malacia at the site of inoculation were seen in 5-day-old mice inoculated with E. ccli HB101 or sterile culture media. Mice inoculated at the age of 8 days examined on day 2 postinfection had localized cortical necrosis at the site of inoculation. On day 4 after inoculation there was a unilateral localized abscess in the meninges which did not extend into the brain parenchyma (Fig. 8). At 6 days postinfection, the

VOL. 59, 1991

PATHOPHYSIOLOGY OF C. DIVERSUS NEONATAL MENINGITIS

1357

local lesion at the site of inoculation was healing with early scar tissue.

tigators have suggested that the occurrence of disease is associated with the presence of a 32- or 69-kDa OMP (16, 34); this suggestion was made on the basis of differences

DISCUSSION In this study we describe a mouse model in which abscessation was produced in the brain after intracranial inoculation of low doses of C. diversus. No CNS lesions were seen in animals inoculated with sterile media or with E. coli HB101, emphasizing the etiologic role of C. diversus. In prior work in our laboratory we were unsuccessful in developing models for C. diversus by using intraperitoneal inoculation of adult mice (with and without iron loading), subcutaneous inoculation of infant rats, and oral inoculation of infant mice (23a). Kline et al. (17) reported an infant rat model for C. diversus meningitis by using simultaneous intraperitoneal and intranasal inoculation. We found this to be cumbersome, requiring large numbers of animals (ca. 90 infants per strain to show a statistically significant difference in mortality between the two strains in Kline's study) and an excessively large inoculum size to produce infection. The histopathologic changes of purulent ventriculitis and adjacent parenchymal abscessation documented in this rat model were similar to the brain lesions that we observed in our infant mouse model. Our model also appears to mimic the human infection in several important respects: abscessation occurred only in the periventricular areas, there was a lack of involvement of the brain stem and cerebellum, and age-related susceptibility was demonstrated. In investigations of outbreaks of C. diversus neonatal meningitis there has been a suggestion that certain strains (as identified by biotype, serotype, plasmid profile, and/or chromosomal restriction endonuclease digest) possess increased virulence; this suggestion was made on the basis of the observation that only one strain of several present in a neonatal nursery was associated with occurrence of disease (10, 11, 22). Our data indicate that there are clear differences in virulence among C. diversus strains after CNS inoculation. Despite some litter-to-litter variability, it was possible to demonstrate statistically significant differences in mortality rates between high and low mortality strains. In our second experiment (seven strains) we found that the two strains with low mortality rates produced no bacteremia and had only localized abscessation, in contrast to the severe abscessation and malacia observed with the more virulent strains. This is in keeping with the work by Kline et al. (17), in which differences in mortality, bacteremia, and occurrence of meningitis were noted when two strains (one from a child with meningitis and the other from an apparent tracheal colonization) were compared. While we found that strains differed in virulence in our model, we were unable to correlate virulence with the source of the isolate (i.e., CNS versus blood versus other sites). This may be a function of the relatively small number of strains included in the study or may reflect difficulties in using data on source of isolation to characterize clinical virulence of strains (i.e., all strains isolated from an asymptomatic person may not be avirulent). Alternatively and perhaps more likely, there may be factors other than virulence after CNS inoculation that determine whether a C. diversus strain will cause disease in a human infant. It has not been possible to consistently associate meningitis cases with any particular C. diversus serogroup or biotype (12). While strains of biotype C, serotype 15, were more likely to kill mice in our model, none of these isolates were from human patients with meningitis. Previous inves-

noted between strains from persons with and without meningitis. In our model (in contrast to the work by Kline et al. [16, 17]) we saw no association between the presence of any OMP and patterns of mortality or CNS lesions. In our histopathologic studies, the development of lesions in 2-day-old mice was initiated by purulent ventriculitis which caused destruction of the ependyma, exposing the periventricular parenchyma to bacterial invasion and subsequent abscessation. Depending on the virulence of the strain, the abscesses became delineated by early glial scarring or progressed into extensive cortical damage. The descriptions of brain pathology in human neonates infected with C. diversus include brain abscessation and necrosis with hemorrhage in the white matter of the cerebral hemispheres (7, 8, 18). One author suggested that the hemorrhage and malacia in the white matter were due to septic vasculitis (7). Our observation of the absence of vasculitis and thrombosis suggests that infarction does not play a role in the pathogenesis of the abscess in the infant mouse model. Although hemorrhage was not seen in the brains of the infant mice, the mice inoculated with some strains had severe malacia indicative of the propensity for destructive necrosis by C. diversus. Cerebral hemorrhage in human neonates may reflect species differences in the response of the neonatal brain to injury or may be a consequence of the route of inoculation. The lack of involvement of other areas of the brain may indicate that C. diversus has a tropism for the periventricular area or the ependymal layer, as has been described for E. coli (24), and one could speculate that C. diversus recognizes specific binding sites in the neonatal mouse brain. Further studies with this model may help to define the pathogenic mechanism(s) involved. In this study we looked at the chronological development of brain lesions beginning 2 days after inoculation in 2-, 5-, and 8-day-old mice. The severity of disease was age related, with absence of bacteremia and with lower mortality and lower numbers of mice with brain abscessation in 5- and 8-day-old mice than with 2-day-old mice. The brain histology in the older mice showed localization of the ventriculitis and meningitis without marked destruction of the periventricular area by abscessation. The reason for this age-related susceptibility to C. diversus infection is not known. Possible explanations could include a change in the biochemical characteristics, postnatal neuroanatomical development, or host defense capabilities of the infant mouse brain between 2 and 5 days of age (6, 9, 14, 24). Neonatal animals and humans have been shown to have impaired host defense mechanisms, such as diminished levels of complement, profound abnormalities in phagocytic function, and deficiency in immunoglobulin M and immunoglobulin A and some classes of immunoglobulin G (14, 20). Experiments to manipulate and enhance host defenses have been done with group B Streptococcus strains and Listeria monocytogenes by using gamma globulin, neutrophil transfusions, and gamma interferon (3, 4, 27). Further studies are needed to assess the relevance of such interventions in C. diversus infections. In summary, our intracranial inoculation model provides a means for evaluating progression of intracranial lesions associated with C. diversus infections. Because of the direct route of inoculation, this model cannot be used to address questions relating to the initial route of infection (although use of scalp electrodes, which has been noted in some

1358

SORIANO ET AL.

outbreak reports [10, 19], might create a situation analogous to direct inoculation in some human infants). Results in the model do not correlate with the clinical source of the strain (i.e., whether a strain was isolated from a patient with meningitis), further suggesting that occurrence of illness in human infants is dependent on factors other than virulence of the organism after CNS inoculation. However, the model does appear to be of value in assessing pathophysiologic mechanisms involved in C. diversus CNS infections and may help to provide some understanding of why this and other gram-negative organisms (29, 33) cause such devastating brain lesions in human neonates. ACKNOWLEDGMENTS We thank Steve Wasserman for statistical assistance. This work was supported in part by a grant from the University of Maryland Medical Biotechnology Center. REFERENCES 1. Altman, G., I. Sechter, I. Braunstein, and C. B. Gerichter. 1984. Citrobacter koseri isolated in Israel, 1972-83. Israel J. Med. Sci. 20:1056-1060. 2. Birnboim, H. C., and J. Doly. 1979. A rapid alkaline extraction procedure for screening recombinant plasmid DNA. Nucleic Acids Res. 7:1513-1523. 3. Chen, Y., A. Nakane, and T. Minagawa. 1989. Recombinant murine gamma interferon induces enhanced resistance to Listeria monocytogenes infection in neonatal mice. Infect. Immun. 57:2345-2349. 4. Christensen, R. D., G. Rothstein, H. B. Anstall, and B. Bybee. 1982. Granulocyte transfusion in neonates with bacterial infection, neutropenia, and depletion of mature marrow neutrophils. Pediatrics 70:1-6. 5. Curless, R. G. 1980. Neonatal intracranial abscess: two cases caused by Citrobacter and a literature review. Ann. Neurol. 8:269-272. 6. Finne, J. 1982. Occurrence of unique polysialosyl carbohydrate units in glycoproteins of developing brain. J. Biol. Chem. 257:11966-11970. 7. Foreman, S. D., E. E. Smith, N. J. Ryan, and G. R. Hogan. 1984. Neonatal Citrobacter meningitis: pathogenesis of cerebral abscess formation. Ann. Neurol. 16:655-659. 8. Friede, R. L. 1973. Cerebral infarcts complicating neonatal leptomeningitis. Acta Neuropathol. 23:245-253. 9. Glode, M. P., A. Sutton, E. R. Moxon, and J. B. Robbins. 1977. Pathogenesis of neonatal Escherichia coli meningitis: induction of bacteremia and meningitis in infant rats fed E. coli Kl. Infect. Immun. 16:75-80. 10. Graham, D. R., R. L. Anderson, F. E. Ariel, N. J. Ehrenkranz, B. Rowe, H. R. Boer, and R. E. Dixon. 1981. Epidemic nosocomial meningitis due to Citrobacter diversus in neonates. J. Infect. Dis. 144:203-209. 11. Graham, D. R., and J. D. Band. 1981. Citrobacter dix'ersus brain abscess and meningitis in neonates. JAMA 245:1923-1925. 12. Gross, R. J., and B. Rowe. 1983. Citrobacter koseri (syn. C. diversus): biotype, serogroup and drug resistance patterns of 517 strains. J Hyg. Camb. 90:233-239. 13. Gwynn, C. M., and R. H. George. 1973. Neonatal Citrobacter meningitis. Arch. Dis. Child. 48:455-458. 14. Hill, H. R. 1985. Host defenses in the neonate: prospects for enhancement. Semin. Perinatol. 9:2-11. 15. Kline, M. W. 1988. Citrobacter diversus and brain abscess in

INFECT. IMMUN.

infancy: epidemiology, pathogenesis, and treatment. J. Pediatr. 113:430-434. 16. Kline, M. W. 1988. Characterization of Citrobacter diversus strains causing neonatal meningitis. J. Infect. Dis. 157:101-105. 17. Kline, M. W., S. L. Kaplan, E. P. Hawkins, and E. 0. Mason. 1988. Pathogenesis of brain abscess formation in an infant rat model of Citrobacter diversus bacteremia and meningitis. J. Infect. Dis. 157:106-112. 18. Levy, R. L., and R. L. Saunders. 1981. Citrobacter meningitis and cerebral abscess in early infancy: cure by moxolactam. Neurology 31:1575-1577. 19. Lin, F. C., W. F. Devoe, C. Morrison, J. Libonati, P. Powers, R. J. Gross, B. Rowe, E. Israel, and J. G. Morris, Jr. 1987. Outbreak of neonatal Citrobacter diversus meningitis in a suburban hospital. Pediatr. Infect. Dis. J. 6:50-55. 20. McCracken, G. H., and H. F. Eichenwald. 1971. Leukocyte function and the development of opsonic and complement activity in the neonate. Am. J. Dis. Child. 121:120-126. 21. McDade, R. L., and K. L. Johnston. 1980. Characterization of serologically dominant outer membrane proteins of Neisseria gonorrhoeae. J. Bacteriol. 141:1183-1191. 22. Morris, J. G., F. C. Lin, C. B. Morrison, R. J. Gross, R. Khabbaz, K. 0. Maher, B. Rowe, E. Israel, and J. P. Libonati. 1986. Molecular epidemiology of neonatal meningitis due to Citrobacter diversus: a study of isolates from hospitals in Maryland. J. Infect. Dis. 154:409-414. 23. Morris, J. G., B. D. Tall, K. L. Kotloff, and I. Sechter. 1988. Carriage of Citrobacter diversus among young children in Baltimore. Pediatr. Infect. Dis. J. 7:294-296. 23a.Morris, J. G., Jr., A. C. Wright, P. K. Wood, J. Michalski, and D. E. Johnson. 1986. Abstr. Annu. Meet. Am. Soc. Microbiol. 1986, B 119, p. 44. 24. Parkkinen, J., and T. K. Korhonen. 1988. Binding sites in the rat brain for E. coli S fimbriae associated with neonatal meningitis. J. Clin. Invest. 81:860-865. 25. Ribeiro, C. D., P. Davis, and D. M. Jones. 1976. Citrobacter koseri meningitis in a special care baby unit. J. Clin. Pathol. 29:1094-1096. 26. Richard, C., B. Brisou, and J. Lioult. 1972. Etude taxonomique de nouveau genre de la famille des enterobacteries. Ann. Inst. Pasteur 122:1137-1146. 27. Santos, J. I., A. 0. Shigeoka, and H. R. Hill. 1981. Protective efficacy of a modified immune serum globulin in experimental group B streptococcal infection. J. Pediatr. 99:873-879. 28. Sears, S. D., R. Richardson, C. Young, C. Parker, and M. M. Levine. 1984. Evaluation of immune response to outer membrane proteins of Vibrio cholerae. Infect. Immun. 44:439-444. 29. Shortland-Webb, W. R. 1968. Proteus and coliform meningoencephalitis in neonates. J. Clin. Pathol. 21:422-431. 30. Tacket, C. O., D. R. Maneval, and M. M. Levine. 1987. Purification, morphology, and genetics of new fimbrial putative colonization factor of enterotoxigenic Escherichia coli 0159: H4. Infect. Immun. 55:1063-1069. 31. Vogel, L. C., L. Ferguson, and S. P. Gotoff. 1978. Citrobacter infections of the central nervous system in early infancy. J. Pediatr. 93:86-88. 32. Williams, W. W., J. Mariano, M. Spurrier, H. D. Donnel, Jr., R. L. Breckenridge, Jr., R. L. Anderson, I. K. Wachsmuth, C. Thornsberry, D. R. Graham, D. W. Thibeault, and J. R. Allen. 1984. Nosocomial meningitis due to Citrobacter diversus in neonates: new aspects of the epidemiology. J. Infect. Dis. 150:229-235. 33. Willis, J., and J. E. Robinson. 1988. Enterobacter sakazakii meningitis in neonates. Pediatr. Infect. Dis. 7:196-199. 34. Wright, A. C., D. R. Maneval, J. E. Galen, and J. G. Morris. 1987. Abstr. Annu. Meet. Am. Soc. Microbiol. 1987, B 262, p. 68.