1035
Interaction of Citrobacter diversus Strains with HEp-2 Epithelial and Human Umbilical Vein Endothelial Cells Charles R. Woods, Jr., Edward O. Mason, Jr., and Sheldon L. Kaplan
Myers-Black Section of Pediatric Infectious Diseases. Department of Pediatrics. Baylor College of Medicine. and C. T. Parker Laboratory. Texas Children's Hospital. Houston
Citrobacter diversus is a gram-negative enteric bacterium that causes sporadic and epidemic meningitis in neonates [1-5]. In two recent series, C. diversus caused meningitis in neonates as often as did Salmonella species: Each accounted for 6% of cases [6, 7]. In our institution from 1984 through 1989, C. diversus caused 3 of20 cases ofgram-negative meningitis in infants 30-fold from 2 h to 4 h (P < .007). In other experiments in which monolayers were incubated with bacteria for 2 h, washed extensively, incubated with new medium without bacteria for 2 h, then incubated with gentamicin (data not shown), -60% of bacteria recovered from the 4-h invasion assay were shown to have adhered during the first 2 h of incubation. Fimbrial phase variation by C. diversus strains . The 4-h invasion results for strain 2988 in the kinetics experiments in figure 4 differ from those reported in tables 2 and 4. C. diversus strains undergo fimbrial phase variation (confirmed by TEM for this strain and by loss of yeast cell agglutination by this and other strains). Kinetics studies were done with cultures of 2988 that had only moderate ability to agglutinate yeast cells. Full loss offimbriae by strain 2988 was associated with a decrease in 4-h invasiveness from 3.9% ± 1.1% to 0.38% ± 0.15% (P < .004; two experiments in triplicate). A similar decrease in invasiveness of strain 4277 also was observed when ability to agglutinate yeast cells was lost or diminished. Intracellular replication and survival ofC. diversus in HEp2 cells. Preliminary experiments and TEM findings (not
Downloaded from http://jid.oxfordjournals.org/ at UQ Library on July 16, 2015
Figure 2. Transmission electron micrograph of HEp-2 cell after incubation with C. diversus strain 2988 for 4 h, showing bacterium in process of invasion, which appears to involve endocytosis.
Woods et al.
1040
JID 1992; 166 (November)
Figure 3. Transmission electron
shown) suggested that C. diversus strains are capable of replication within HEp-2 cells, at least during the first 2 h of intracellular residence. There was no significant change , however, in the percentage of organisms recovered after a 2-h incubation with bacteria, followed by 2 h with gentamicin at 120 ~g/mL, compared with this same 4-h period followed by Table 3. Adhesion of C. diversus strains to HEp-2 Cells. Strain
Fimbriae
Invasion
2988 4277 3078 4406 4509 4036
MS MS MS MS NF NF
High High Low Low Low Low
% adhesion
120 ± 138 ± 20 ± 109 ± 49 ± 34 ±
40' (8) 88 (6) 15 (8) 56 (6) 19 (6) 21 (8)
NOTE. MS = man nose sensit ive; NF = not fimbriated or man nose resistant. Invasion by strains 2988 and 4277 (high) was significa ntly greater than the other strains (low). which were not different from each other. Data a re %o f initial inoculum . mean ± SD (n) . n = total slides. minimum ofthree experiments in duplicate. • 2988. 4277 or 4406 vs, 4509. 4036 . or 3078. all P < .0 16; 4509 vs. 3078. P < .01; all other P not significant (Mann-Whitney rank sum test).
an additional 4 h with gentamicin at 12 ~g/mL (0.025% ± 0.020% vs. 0.015% ± 0.009%, P not significant; results offour experiments in triplicate). Four strains, including 2988, then were evaluated for survival within HEp-2 cells after a 4-h incubation period (figure 5). Well sets incubated for 4 h, followed by 2 h with gentamicin at 120 ~g/mL, were compared with sets treated identically followed by an additional 18 h with gentamicin at 12 ~g/mL. Strains 2988, 4277, and 4406 showed decreases in intracellular survival over this time period (P < .015, .005, and .001, respectively), while strain 4509 showed an increase in the number of organisms recovered (P < .016). Results with agents that alter bacterial or HEp-2 cell functions . Cytochalasin D was used to inhibit microfilament-dependent endocytosis by mammalian cells. Nalidixic acid, rifarnpin, and chloramphenicol were used to inhibit DNA replication (inhibition of bacterial DNA gyrase), RNA transcription (inhibition of DNA-dependent RNA polymerase), and protein translation (inhibition of ribosomal function) , respectively. aMM was used to block mannose receptors of type I fimbriae. These studies were done with strain 2988. Experimental wells were compared only with same-
Downloaded from http://jid.oxfordjournals.org/ at UQ Library on July 16, 2015
micrograph of HEp-2 cell after incubation with C. diversus strain 4509 for 4 h, showing single intracellular organism (arrow). Comparison with figure I shows difference in ability to invade HEp-2 cells by strains 2988 and 4509.
Interaction of C. diversus with Cells
lID 1992;166 (November)
10
100 - - % Adhesion - 0 - % Invasion
3.0
t:
t:
0
0
ii
·iii
CD
as >
.s:::: 10 'tJ c::(
.5 .1
0~
1
2
0
3
5
4
0~
C
0
·iii as >
.E tfl.
1.0
.01
0.0 0
Time (h)
day controls. Not all agents were tested in each experiment, but controls were run each day. All concentrations of cytochalasin D inhibited invasion of HEp-2 cells by strain 2988 (figure 6; all P < .00 I). There was a trend toward a dose response between the 0.5- and 2.0-llg/ mL doses (P < .07). Adhesion was not affected by a dose of 0.5Ilg/mL versus controls (44% ± 12% vs. 52% ± 28%, Pnot significant ). The mannose analogue aMM (table 4) significantly re-
10r-------------------.
C
4509
·iii as c>
42n
0
.1 2988 4406
6
12
18
24
Time (h) Figure 5. Intracellular survival of 4 C. diversus strains. Strains 4277, 2988, and 4406 survived intracellulariy but were recovered in decreasing numbers over time (P < .005, .015, and .001, respectively; Mann-Whitney rank sum test). However, strain 4509 increased over time (P < .016), suggesting ability to replicate intracellularly. Data are mean ± SD of three experiments in triplicate.
0.5
2
Figure 6. Effect of cytochalasin D on invasion ofHEp-2 cells by C. diversus strain 2988 during 4-h incubation period. All three concentrations of cytochalasin D decreased invasion relative to controls (all P < .001; Mann-Whitney rank sum test). Trend toward dose response between 0.5- and 2.0-Jlg/mL concentrations was seen (P < .07). Data are mean ± SD of three experiments in triplicate.
duced adhesion to and invasion of HEp-2 cells by strain 2988 (P < .027 and P < .001, respectively). Growth of the bacteria in the medium above monolayers was not affected by this agent. Of the antibiotics, rifampin and chloramphenicol significantly reduced invasion ofHEp-2 cells by strain 2988 (both P < .00 I), but nalidixic acid had no effect (table 4). Neither rifampin nor chloramphenicol affected the ability of strain 2988 to adhere to HEp-2 cells.
Discussion C. diversus is an important neonatal pathogen for its devastating effects more than for its prevalence. In the series studied by Graham and Band [I], >80% of infants with C. diversus meningitis at follow-up had died or were mentally retarded. Frequency ofbrain abscess formation by this organism probably contributes to the high prevalence of poor outcome. C. diversus can occur in epidemics in nurseries [2-4] and recently was found to be the predominant nosocomial pathogen in one neonatal intensive care unit [34]. We chose two cell lines of different embryologic origin to study the sequential pathogenesis of C. diversus infections. Before eNS invasion, bacteria must cross an epithelial mucosal barrier to reach the bloodstream [35]. In neonates, likely sites of this event include the upper respiratory tract, intestinal mucosa, and umbilical stump [36]. HEp-2 cells, originally derived from a respiratory tract carcinoma, may be more like undifferentiated than respiratory epithelium. Ability to invade these and other high-passage, tumor-derived
Downloaded from http://jid.oxfordjournals.org/ at UQ Library on July 16, 2015
Figure 4. Time course of adhesion to and invasion of HEp-2 cells by C. diversus strain 2988. Both processes increased I log order over 2 h (P < .00 I for adhesion, P < .007 for invasion; Mann-Whitney rank sum test). Invasion lagged behind adhesion. Only a small fraction of adherent organisms was internalized by HEp-2 cells during time periods studied. Data are mean ± SD ofat least four experiments in duplicate for adhesion and in triplicate for invasion.
~ 0
1041
Woods et al.
1042
JID 1992; 166 (November)
Table 4. Effect of inhibitors on adhesion and invasion ofHEp-2 cells by C. diversus strain 2988. % of adhesion*
Inhibitor
% of invasion"
Experimental
Control
p
Experimental
Control
p
48 ± 23
82 ± 31
Interaction of Citrobacter diversus Strains with HEp-2 Epithelial and Human Umbilical Vein Endothelial Cells Charles R. Woods, Jr., Edward O. Mason, Jr., and Sheldon L. Kaplan
Myers-Black Section of Pediatric Infectious Diseases. Department of Pediatrics. Baylor College of Medicine. and C. T. Parker Laboratory. Texas Children's Hospital. Houston
Citrobacter diversus is a gram-negative enteric bacterium that causes sporadic and epidemic meningitis in neonates [1-5]. In two recent series, C. diversus caused meningitis in neonates as often as did Salmonella species: Each accounted for 6% of cases [6, 7]. In our institution from 1984 through 1989, C. diversus caused 3 of20 cases ofgram-negative meningitis in infants 30-fold from 2 h to 4 h (P < .007). In other experiments in which monolayers were incubated with bacteria for 2 h, washed extensively, incubated with new medium without bacteria for 2 h, then incubated with gentamicin (data not shown), -60% of bacteria recovered from the 4-h invasion assay were shown to have adhered during the first 2 h of incubation. Fimbrial phase variation by C. diversus strains . The 4-h invasion results for strain 2988 in the kinetics experiments in figure 4 differ from those reported in tables 2 and 4. C. diversus strains undergo fimbrial phase variation (confirmed by TEM for this strain and by loss of yeast cell agglutination by this and other strains). Kinetics studies were done with cultures of 2988 that had only moderate ability to agglutinate yeast cells. Full loss offimbriae by strain 2988 was associated with a decrease in 4-h invasiveness from 3.9% ± 1.1% to 0.38% ± 0.15% (P < .004; two experiments in triplicate). A similar decrease in invasiveness of strain 4277 also was observed when ability to agglutinate yeast cells was lost or diminished. Intracellular replication and survival ofC. diversus in HEp2 cells. Preliminary experiments and TEM findings (not
Downloaded from http://jid.oxfordjournals.org/ at UQ Library on July 16, 2015
Figure 2. Transmission electron micrograph of HEp-2 cell after incubation with C. diversus strain 2988 for 4 h, showing bacterium in process of invasion, which appears to involve endocytosis.
Woods et al.
1040
JID 1992; 166 (November)
Figure 3. Transmission electron
shown) suggested that C. diversus strains are capable of replication within HEp-2 cells, at least during the first 2 h of intracellular residence. There was no significant change , however, in the percentage of organisms recovered after a 2-h incubation with bacteria, followed by 2 h with gentamicin at 120 ~g/mL, compared with this same 4-h period followed by Table 3. Adhesion of C. diversus strains to HEp-2 Cells. Strain
Fimbriae
Invasion
2988 4277 3078 4406 4509 4036
MS MS MS MS NF NF
High High Low Low Low Low
% adhesion
120 ± 138 ± 20 ± 109 ± 49 ± 34 ±
40' (8) 88 (6) 15 (8) 56 (6) 19 (6) 21 (8)
NOTE. MS = man nose sensit ive; NF = not fimbriated or man nose resistant. Invasion by strains 2988 and 4277 (high) was significa ntly greater than the other strains (low). which were not different from each other. Data a re %o f initial inoculum . mean ± SD (n) . n = total slides. minimum ofthree experiments in duplicate. • 2988. 4277 or 4406 vs, 4509. 4036 . or 3078. all P < .0 16; 4509 vs. 3078. P < .01; all other P not significant (Mann-Whitney rank sum test).
an additional 4 h with gentamicin at 12 ~g/mL (0.025% ± 0.020% vs. 0.015% ± 0.009%, P not significant; results offour experiments in triplicate). Four strains, including 2988, then were evaluated for survival within HEp-2 cells after a 4-h incubation period (figure 5). Well sets incubated for 4 h, followed by 2 h with gentamicin at 120 ~g/mL, were compared with sets treated identically followed by an additional 18 h with gentamicin at 12 ~g/mL. Strains 2988, 4277, and 4406 showed decreases in intracellular survival over this time period (P < .015, .005, and .001, respectively), while strain 4509 showed an increase in the number of organisms recovered (P < .016). Results with agents that alter bacterial or HEp-2 cell functions . Cytochalasin D was used to inhibit microfilament-dependent endocytosis by mammalian cells. Nalidixic acid, rifarnpin, and chloramphenicol were used to inhibit DNA replication (inhibition of bacterial DNA gyrase), RNA transcription (inhibition of DNA-dependent RNA polymerase), and protein translation (inhibition of ribosomal function) , respectively. aMM was used to block mannose receptors of type I fimbriae. These studies were done with strain 2988. Experimental wells were compared only with same-
Downloaded from http://jid.oxfordjournals.org/ at UQ Library on July 16, 2015
micrograph of HEp-2 cell after incubation with C. diversus strain 4509 for 4 h, showing single intracellular organism (arrow). Comparison with figure I shows difference in ability to invade HEp-2 cells by strains 2988 and 4509.
Interaction of C. diversus with Cells
lID 1992;166 (November)
10
100 - - % Adhesion - 0 - % Invasion
3.0
t:
t:
0
0
ii
·iii
CD
as >
.s:::: 10 'tJ c::(
.5 .1
0~
1
2
0
3
5
4
0~
C
0
·iii as >
.E tfl.
1.0
.01
0.0 0
Time (h)
day controls. Not all agents were tested in each experiment, but controls were run each day. All concentrations of cytochalasin D inhibited invasion of HEp-2 cells by strain 2988 (figure 6; all P < .00 I). There was a trend toward a dose response between the 0.5- and 2.0-llg/ mL doses (P < .07). Adhesion was not affected by a dose of 0.5Ilg/mL versus controls (44% ± 12% vs. 52% ± 28%, Pnot significant ). The mannose analogue aMM (table 4) significantly re-
10r-------------------.
C
4509
·iii as c>
42n
0
.1 2988 4406
6
12
18
24
Time (h) Figure 5. Intracellular survival of 4 C. diversus strains. Strains 4277, 2988, and 4406 survived intracellulariy but were recovered in decreasing numbers over time (P < .005, .015, and .001, respectively; Mann-Whitney rank sum test). However, strain 4509 increased over time (P < .016), suggesting ability to replicate intracellularly. Data are mean ± SD of three experiments in triplicate.
0.5
2
Figure 6. Effect of cytochalasin D on invasion ofHEp-2 cells by C. diversus strain 2988 during 4-h incubation period. All three concentrations of cytochalasin D decreased invasion relative to controls (all P < .001; Mann-Whitney rank sum test). Trend toward dose response between 0.5- and 2.0-Jlg/mL concentrations was seen (P < .07). Data are mean ± SD of three experiments in triplicate.
duced adhesion to and invasion of HEp-2 cells by strain 2988 (P < .027 and P < .001, respectively). Growth of the bacteria in the medium above monolayers was not affected by this agent. Of the antibiotics, rifampin and chloramphenicol significantly reduced invasion ofHEp-2 cells by strain 2988 (both P < .00 I), but nalidixic acid had no effect (table 4). Neither rifampin nor chloramphenicol affected the ability of strain 2988 to adhere to HEp-2 cells.
Discussion C. diversus is an important neonatal pathogen for its devastating effects more than for its prevalence. In the series studied by Graham and Band [I], >80% of infants with C. diversus meningitis at follow-up had died or were mentally retarded. Frequency ofbrain abscess formation by this organism probably contributes to the high prevalence of poor outcome. C. diversus can occur in epidemics in nurseries [2-4] and recently was found to be the predominant nosocomial pathogen in one neonatal intensive care unit [34]. We chose two cell lines of different embryologic origin to study the sequential pathogenesis of C. diversus infections. Before eNS invasion, bacteria must cross an epithelial mucosal barrier to reach the bloodstream [35]. In neonates, likely sites of this event include the upper respiratory tract, intestinal mucosa, and umbilical stump [36]. HEp-2 cells, originally derived from a respiratory tract carcinoma, may be more like undifferentiated than respiratory epithelium. Ability to invade these and other high-passage, tumor-derived
Downloaded from http://jid.oxfordjournals.org/ at UQ Library on July 16, 2015
Figure 4. Time course of adhesion to and invasion of HEp-2 cells by C. diversus strain 2988. Both processes increased I log order over 2 h (P < .00 I for adhesion, P < .007 for invasion; Mann-Whitney rank sum test). Invasion lagged behind adhesion. Only a small fraction of adherent organisms was internalized by HEp-2 cells during time periods studied. Data are mean ± SD ofat least four experiments in duplicate for adhesion and in triplicate for invasion.
~ 0
1041
Woods et al.
1042
JID 1992; 166 (November)
Table 4. Effect of inhibitors on adhesion and invasion ofHEp-2 cells by C. diversus strain 2988. % of adhesion*
Inhibitor
% of invasion"
Experimental
Control
p
Experimental
Control
p
48 ± 23
82 ± 31
