INFECTION AND IMMUNITY, Apr. 1982, p. 407-416 Vol. 36, No. 10019-9567/82/040407-10$02.00/0

Anatomical and Immunological Responses of RabbitGallbladders to Bacterial Infections

ROBERT H. GILMAN,l*t CHARLES YOUNG,' RUTH BULGER,2 RICHARD B. HORNICK,' ANDBARBARA GREENBERG'

Division of Infectious Disease' and Department of Pathology,2 University of Maryland School of Medicine,Baltimore, Maryland 21201

Received 11 September 1981/Accepted 25 November 1981

To study the sequential morphological and immunological response of therabbit gallbladder to bacterial infection and to compare the inflammatory respons-es with different pathogens, gallbladders were infected with Streptococcusfaecalis and two strains of Escherichia coli, one of which produced enterotoxin.Gallbladder infection was produced either by intravenously injecting bacteria intorabbits with a small liver infarct or by injecting bacteria directly into thegallbladder of normal rabbits. The percentage of gallbladders infected intrave-nously with a nonenterotoxigenic E. coli strain was 86% at 1 week, 70% at 3weeks, and 15% at 6 weeks. Epithelial necrosis and leukocyte infiltration wereprominent 1 week after infection. At 3 and 6 weeks after infection, there was cryptdistortion and increased mucus secretion in the epithelium as shown by periodicacid-Schiff staining. The lamina propria was infiltrated with mononuclear cells,many of which were plasma cells. Myofibroblasts (contractile fibroblasts) werealso identified on transmission microscopy. In addition to these changes, toxigen-ic E. coli produced subepithelial capillary dilation in the villus core. Morphologi-cal changes (excluding toxin-associated changes) were related to the duration ofinfection rather than to the specific species of infecting bacteria. Infected gall-bladders studied by immunofluorescence had a greater than 50-fold increase inplasma cells compared with control cells. In addition, the number increased withthe duration of infection. Immunoglobulin A cells were the major cell type ingallbladders infected by intravesical injection, whereas immunoglobulin G cellspredominated in gallbladders infected intravenously. The gallbladder appears tomount a local immune response to bacterial infection.

Over 400,000 cholecystectomies are per-formed per year (unpublished data for 1979 fromthe National Hospital Discharge Survey, Na-tional Center for Health Statistics), and 4 to 10%of these operations are due to acute acalculouscholecystitis (16, 43). The significance of thisdisease is further exemplified in the pediatric agegroup in which 40% of the cholecystectomies area result of acute acalculous cholecystitis (17).Classically, acute acalculous cholecystitis wasassociated with acute infections, in particular,typhoid fever (32), scarlet fever (38), and lepto-spirosis (2). More recently, it has been encoun-tered in patients with extensive burns and mas-sive trauma in association with bacteremia (31,37).There are multiple factors which precipitate

the onset of acute acalculous cholecystitis.Glenn suggests that bacteremia, in the presenceof associated biliary stasis, may be one of the

t Present address: Baltimore City Hospitals, Division ofGeographic Medicine, Baltimore, MD 21224.

important etiological factors irrespective of thespecific organism or its site of origin (16).Although the histological changes of acute

acalculous cholecystitis in humans have beendescribed, the pathogenic process has not beenestablished. Some animal studies have attempt-ed to examine the sequential response of theinfected gallbladder to specific organisms, suchas Salmonella typhi (5, 9, 29), Vibrio cholerae(24, 36), or Streptococcus species (33). Thehistological changes in the rabbit gallbladderinfected in typhoid bacilli and Streptococcusspecies encompass the range of changes report-ed in the infected human gallbladder (9, 29, 33).In humans, biliary infection is most commonlycaused by normal aerobic microflora (3, 26). Inthese studies, bacterial cholecystitis was pro-duced either by direct injection of bacteria intothe gallbladder (by needle or indwelling cathe-ter) (24, 29, 36) or by systemic infection ofbacteria in numbers sufficient to produce endo-toxin shock (5, 9, 10, 29, 33). The resultinginfection was temporary and associated with an

407

on April 15, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

408 GILMAN ET AL.

inflammatory infiltrate of the gallbladder muco-sa (29, 36).The morphological changes found in gallblad-

ders infected with either S. typhi, Streptococcusspecies, or V. cholerae have been assumed to bespecific for these organisms (27, 29, 33, 36).However, it is unlikely that the gallbladder mu-cosa responds specifically to different bacterialorganisms. Rather, the gallbladder, an analog ofthe intestine, might respond to bacterial infec-tion by mounting a local immune response.Although biliary transport of immunoglobulin A(IgA) immunoglobulin has recently been demon-strated (4), the ability of the biliary tract itself toproduce a local immune reaction has not beenshown. The morphological events which occurafter bacterial infection of the gallbladder mightbe in part related to the stimulation of localimmunity by bacterial antigens.To further elucidate the sequential changes in

the infected gallbladder, a rabbit model was usedto describe the pathological process and theimmune response of the gallbladder to bacterialinfection. To determine whether the gallbladderresponds stereotypically to bacterial infection,gallbladders were infected with three bacterialagents: Escherichia coli, Streptococcusfaecalis,and E. coli H10407, which is enterotoxin pro-ducing. Because E. coli H10407 simulates chol-era infection and produces higher infection ratesin our model, it was used to observe changes inthe gallbladder secondary to toxin production.

MATERIALS AND METHODSExperimental design. Gallbladder infection was pro-

duced by two methods. In method 1, a sterile liverinfarct was produced by cautery and then infected byan intravenous injection of bacteria. This methodatraumatically produces a high rate of associated gall-bladder infection and avoids morphological changes inthe gallbladder caused by direct trauma or by endotox-in shock. This indirect method of gallbladder infectionwas used for all of the morphological studies and someof the immunological studies. Immunological changeswere also examined by method 2, in which gallblad-ders were infected by direct injection of bacteria intothe gallbladder lumen. Both methods used coccidia-free New Zealand white female rabbits weighing 1.5 to2.0 kg.

Liver infarction and intravenous infection. Rabbitswere shaved, pretreated with atropine, and anesthe-tized with ether before operation. At laparotomy,electrocautery was performed on a lobe of the liver toproduce a burn 1 cm in diameter. Incisions wereclosed by layers with 3-0 chromic suture and skinstaples. After 24 h, an inoculum of bacteria wasinjected via the marginal ear vein, as described below(bacteremia was not present 24 h after inoculation asevidenced by sterile blood cultures). Animals weresacrificed after 1, 3, or 6 weeks. This procedureproduced infected bile in most animals. Control ani-mals were separated into two groups: (i) cauterizedrabbits inoculated intravenously with saline and sacri-

ficed 1, 3, or 6 weeks later, and (ii) sham-operated(noncauterized) animals injected with a bacterial in-oculum and examined 1 week later. Rabbits with liverinfarcts who were infected experimentally, but whosebile was not infected at the time of sacrifice (1, 3, and 6weeks post-inoculation), also functioned as a controlgroup. In this group, however, it was not determinedwhether biliary infection had never taken place or hadbeen present and then spontaneously cleared.

Infection by direct gallbladder injection. Each rabbitwas anesthetized with ether, the gallbladder was ex-posed, and the bacterial inoculum was injected in avolume of 0.15 ml with a 26-gauge needle. Leakagewas controlled by applying mild pressure. At 15 minafter inoculation, 1 ml of ear blood was taken forculture. Control rabbits had saline injected into thegallbladder. The experimental and control rabbitswere sacrificed 1 week after inoculation. Gallbladderbile was aspirated, and the colony-forming units(CFU) per milliliter were determined by dilution.

Bacteria. Three strains of bacteria were used: (i) HS,a noninvasive nontoxigenic E. coli strain isolated fromnormal human feces (11); (ii) H10407, a well-character-ized strain (11, 39) of enterotoxigenic E. coli (positivefor heat-labile and heat-stable enterotoxin) isolated inDacca, Bangladesh, from a patient with diarrhea; and(iii) S. faecalis, isolated from a patient with enterococ-cal vertebral osteomyelitis. All bacteria were washedtwo times in saline before injection.

Rabbits were infected intravenously with 1-ml vol-umes of 1 x 108 to 4 x 108 CFU of E. coli strains or 1x 109 to 2 x 109 CFU of S. faecalis. Rabbits wereinfected by direct injection into the gallbladder with E.coli HS at a dose of either 1 x 108 or 1 x 105 CFU ineach of four rabbits.

Tissue processing, bacterial quantification, and histo-pathology. Rabbits were rapidly sacrificed at varioustimes with intravenous phenobarbitol, and tissue spec-imens were obtained aseptically. After the bile wasaspirated for bacterial culture, the gallbladder wasremoved, and a longitudinal section was taken fromthe free surface for histology. Sections of liver infarctand normal liver were dissected and homogenized in40 ml of saline; 10-fold dilutions were plated onMacConkey agar and phenylethyl alcohol plates, andviable bacteria were counted. Gram-negative bacteriawere identified by the triple sugar iron reaction andseroagglutination. S. faecalis colonies were identifiedby Gram smear and a positive reaction in Pfizerselective enterococcus broth.

Gallbladder specimens were cut lengthwise; onepiece was fixed in 10% buffered Formalin, and anotherportion was used for fluorescent-antibody studies.Paraffin sections of Formalin-fixed tissue were thenstained with hematoxylin and eosin or periodic acid-Schiff. Rabbit granulocytes all have large granules andstain eosinophilic with hematoxylin and eosin. Theterm heterophile, employed by Bloom and Fawcett (6)to describe rabbit granulocytes, is used in this paper.Ten gallbladders, examined by scanning and trans-

mission electron microscopy, were fixed by vascularperfusion with a solution containing 2% formaldehydeand 2.5% glutaraldehyde, buffered with 0.04 M potas-sium phosphate (22). Perfusion fixation maintains thegallbladder in an expanded state and minimizes mor-phological artifacts caused by a collapse of the gall-bladder before fixation (30). After 5 min of perfusion,

INFECT. IMMUN.

on April 15, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

BACTERIAL INFECTION OF THE GALLBLADDER 409

the liver was removed. The gallbladder was transect-ed, placed in fixative for an additional 4 to 24 h, storedin 0.1 M potassium phosphate with 0.2 M sucrose, andthen processed for scanning (1) and transmission mi-croscopy.

Fluorescent-antibody technique. In initial experi-ments, gallbladder material was dipped in phosphate-buffered saline (pH 7.4) and fixed immediately in 95%cold ethanol (4°C) for 18 h. In later studies, to reducebackground fluorescence, all gallbladder tissue waswashed for 18 h in cold agitated phosphate-bufferedsaline (7). After washing, tissues were processed bythe Sainte-Marie technique, modified by deparaffinat-ing with two 5- to 8-min xylene baths (35).Immunoglobulin cell types. Freeze-dried goat antise-

rum, monospecific against the heavy chain of rabbitIgA, IgG, and IgM conjugated with fluorescein isothio-cyanate (Cappel Laboratories), was used for directstaining of plasma cell immunoglobulin types. Controlslides for these experiments were prepared by (i)staining with a fluorescein conjugated goat antiserumproduced in rabbits, and (ii) using nonfluorescein-conjugated goat antiserum to the heavy chain of rabbitIgA, IgG, or IgM to specifically block the reactionwith the fluorescent antiserum. Fluorescent cells infive or more representative fields in the gallbladderlamina propria were counted at a magnification of 400,and the mean count per field was then calculated. Allcounting was done on coded specimens. The finalcount was obtained by subtracting the mean cell countof unblocked slides from the mean cell count presentin control slides (consistently under 15% of the total)which had been blocked by prior treatment withunlabeled antisera.Serum antibody. Serum antibody titer in rabbits

inoculated with E. coli HS was determined by thepassive hemagglutination technique (25).

RESULTSMicrobiological results. Both infarcted liver

tissue and bile obtained from cauterized rabbitsinoculated intravenously with saline were ster-ile. Injection of E. coli H10407 into sham-operat-ed noncauterized rabbits frequently producedbiliary infection (five of seven rabbits, 71%).However, only one of seven and none of fivesham-operated controls inoculated with E. coliHS and S. faecalis, respectively, had biliaryinfection.The number of rabbits with infected bile de-

creased significantly with time (Table 1). TheCFU per milliliter of bile from gallbladders stillinfected at 3 and 6 weeks were comparable tothe CFU per milliliter of bile from infectedgallbladders studied over 1 week.

Histology. All control animals (saline injectedand sham operated) without biliary infection hadhistologically normal gallbladders. The normalhistology of the rabbit gallbladder has beenpreviously well described (20, 41). We alsonoted droplets positive for periodic acid-Schiffstaining in the apex of some epithelial cells.Also, scanning microscopy showed, betweenthe rugal folds, small pits resembling those of theguinea pig gallbladder (Fig. 1).

Rabbits with infected infarcts and sterile bile.Gallbladders of rabbits with culture-positive in-farcts whose bile contained less than 10 CFU/mlwere normal when examined 1 week after inocu-lation. However, among animals with culture-positive infarcts and noninfected bile, at 3 and 6

TABLE 1. Bacterial infection of liver infarcts and gallbladder bile in cauterized rabbitsa

No. of rabbits with Infected infarct mean No. of rabbits with Infected bile meanBacteria Days of infected infarcts/ bacterial count (log infected bile/total bacterial count (log

infection total no. of rabbits CFU/g)b no. of rabbits CFU/ml)'

E. coli HS 7-10 14/14 7.7 ± 0.5 12/14 6.4 ± 1.820-24 13/14 5.1 ± 1.1 11/14 6.5 ± 1.942-44 8/13d 4.5 ± 1.5 2/13d 7.0 ± 0.8

E. coli H10407 7-10 5/5 7.2 ± 0.3 4/5 8.4 ± 1.520-24 2/2 6.3 ± 0.1 2/2 8.2 ± 0.1

S.faecalis 7-10 9/11 7.0 ± 1.8 6/11 7.7 ± 1.020-24 10/11 7.0 ± 0.3 4/10 5.2 ± 1.642-44 3/9d 4.5 ± 0.5 1/9

Infarct alone 7-10 0/5 0/520-24 0/4 0/442-44 0/5 0/5

a Sham-operated noncauterized control rabbits intravenously inoculated with bacteria and sacrificed 1 weekafter infection had the following results: E. coli HS, one of seven animals and S. faecalis, none of five hadinfected bile; in contrast, five of seven sham-operated rabbits inoculated with E. coli H10407 had infected bilewith a mean bacterial count of 7.2 + 0.5 log CFU ml.

b Values are means ± standard deviation.c Values are means ± standard deviation of eight rabbits.d There was a significant difference P < 0.05 (chi-square test) between the number of infected infarcts and bile

at 7 to 10 and at 42 to 44 days in rabbits inoculated with either E. coli HS or S. faecalis.

VOL. 36, 1982

on April 15, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

410 GILMAN ET AL.

E

MWIVI

FIG. 1. Scanning micrograph of a normal rabbit gallbladder on cross-section; pitting (arrow) and smoothhexagonal epithelial cells (E) are seen. Although not clearly seen at this magnification, the epithelial cells are

covered by microvilli (x540).

weeks after inoculation, 2 of 8 and 6 of 19gallbladders, respectively, had mild lamina pro-pria lymphocytic infiltration, increased lympho-cyte infiltration, cuboidal metaplasia of the epi-thelium, and crypt distortion. The gallbladdersof the remaining animals resembled those ofnoninfected controls.

Rabbits with infected gallbladders. Gallblad-ders infected with E. coli HS and S. faecaliswere similar 7 to 10 days after inoculation.Acute inflammatory changes predominated ingallbladders infected for 1 week. Prominent find-ings were focal necrosis with adjacent cuboidalmetaplasia of the normal columnar epithelium(Fig. 2) and marked infiltration of the surfaceepithelium with heterophiles. Occasionally, theinfiltration was severe enough to produce epi-thelial cysts filled with clumps of eosinophilicleukocytes. There was an increase in dropletspositive for periodic acid-Schiff staining in theepithelial cells without marked crypt distortion.In one gallbladder infected with E. coli HS, theinflammatory changes were so severe that theywere associated with submucosal vessel throm-

9

FIG. 2. Scanning micrograph of a gallbladder oneweek after infection with E. coli HS. The villus tip isnecrotic (N) and the adjacent epithelium is cuboidal(arrow). The lamina propria is infiltrated with hetero-philes, mononuclear cells, and plasma cells (x400).Note the contrast with the scanning micrograph of anormal gallbladder (Fig. 1).

INFECT. IMMUN.

on April 15, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

BACTERIAL INFECTION OF THE GALLBLADDER

_

4

4,4~~~~~~~~~~o

,.- ,- ....e,-

.S.---f

FIG. 3. Scanning micrograph of a gallbladder 1 week after infection with E. coli HS. The normally confluenthexagonal epithelial cell outline has a cobblestone appearance due to marked variation in the size and shape ofthe epithelial cells. There is increased pitting and hyperplastic hillocks (H) of the surface epithelium. The villuscore is packed with cells (arrow) (x 360). Note the contrast with the scanning micrograph of a normal gallbladder(Fig. 1).

bosis with coagulation necrosis of the mucosa.The lamina propria, normally containing onlyscattered fibroblasts and mononuclear cells, wasnow infiltrated with heterophiles, mononuclear

FIG. 4. Semithin section of a gallbladder villus,examined 1 week after infection with E. coli H10407(enterotoxigenic-producing strain). Marked capillarydilation (C) and interstitial edema (E) are evident(x180).

cells, and plasma cells. This was reflected onscanning microscopy (one gallbladder infectedwith E. coli HS) as marked distension andflattening of the mucosal folds associated withthe infiltration of the core of the villus withinflammatory cells (Fig. 3). The following pointsare of interest to the morphologist. The microvil-li were not significantly changed. Electron mi-croscopy revealed cells with the appearance ofmyofibroblasts in the lamina propria. These cellshave previously been reported in contractinggranulation tissue (34). Bacteria were not seen inthe sections studied by transmission microsco-py-

Gallbladders infected with E. coli H10407 (anenterotoxin-producing strain) examined 1 weekafter inoculation had the changes describedabove, but in addition, were characterized bymarked enlargement of subepithelial capillarieswith interstitial edema of the villi (Fig. 4). Theenlarged capillaries did not appear damaged. Onscanning microscopy, this was reflected astensely swollen rugae in which the epithelialcells were stretched out, flattened, and whoseintervening cell margins were poorly defined.Further detail showed the presence of increasedpits and normal microvilli. Transmission elec-

411VOL. 36, 1982

.e... .:.

PIIL

e, i.-T

A&

on April 15, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

412 GILMAN ET AL.

FIG. 5 Scanning micrograph of a gallbladder 3weeks after infection with E. coli HS. An immaturelymphoid follicle (L) is present in the lamina propriaand intestinal metaplasia of the epithelium (arrow) ispresent (x180).

tron micrographs (not shown) demonstrated anincreased number of apical epithelial mucusdroplets and widened lateral intracellularspaces, a condition observed in gallbladderswhich are actively transporting fluid (12, 23).At 3 weeks after inoculation, gallbladders

infected with S. faecalis, and E. ccli HS no

longer had epithelial necrosis. The predominantchanges were evidence of increased mucus se-cretion and an inflammatory infiltrate. This infil-trate was more intensive and contained a greaterproportion of mononuclear cells compared withthe infiltrate of gallbladders infected for 1 week.Epithelial hyperplasia and crypt distortion wasnoted (Fig. 5). Epithelial cells filled with period-ic acid-Schiff-stained granules were especiallyprominent in the epithelial crypts with intestinalmetaplasia. Intraepithelial leukocytes were in-creased as were plasma cells and other mononu-clear cells in the lamina propria. The muscularisand submuscularis were thickened and showedincreased mononuclear infiltration. As noted at1 week, the myofibroblasts in the lamina propriawere still present on electron microscopy (Fig.6).

In gallbladders infected with E. coli H10407,marked capillary distension remained 3 weeksafter inoculation, but was less prominent thanafter 1 week. Except for the increase in subep-ithelial capillary size, gallbladders infected withE. coli H10407 were similar to E. coli HS-infected gallbladders (Fig. 7).

Rabbits with infected bile at 6 weeks showedpathological changes similar to the rabbits withinfected bile at 3 weeks. However, there wasincreased crypt distortion and intestinal meta-

--N -

FIG. 6. Transmission electron micrography of the lamina propria of a gallbladder examined 3 weeks afterinfection with E. coli HS. Two myofibroblasts (M) containing numerous fine filaments and dense bodies arepresent and appear to be contracting as evidenced by the irregular wavy cell outline (x1,030).

INFECT. IMMUN.

on April 15, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

BACTERIAL INFECTION OF THE GALLBLADDER

FIG. 7. Scanning micrograph of gallbladder examined 3 weeks after infection with E. coli H10407.Hyperplastic hillocks (H) containing pits (P), disparate cell size, and frequent blurring of cell margins (E) arepresent (x 560).

plasia, a higher degree of lymphocytic infil-tration in the walls of the tissue, and moreextensive lymphoid follicle formation in the sub-epithelial layers in the rabbits with infected bileat six weeks. Also, perivascular lymphocyticcuffing of blood vessels and lymphocyte epitheli-al transmigration were more prevalent than ob-served in the rabbits with infected bile at 3weeks.

Fluorescent-antibody studies from intravenous-ly infected rabbits. No differences in counts ofantibody-containing cells were found betweenspecimens immediately fixed and those washedfor 18 h and then fixed. These results have beencombined and treated as a single unit. Gallblad-ders from cauterized controls injected with sa-line had very few cells containing IgG, IgA, andIgM. Similar results were found in noninfectedgallbladders of sham-operated rabbits injectedintravenously with E. coli or S. faecalis (Table2). Four rabbits intravenously injected with E.coli HS had culture-positive liver infarcts withculture-negative bile (noninfected gallbladders);gallbladders from these animals had few plasmacells. Increased counts occurred as the intervalbetween inoculation and sacrifice increased.

All intravenously infected gallbladders had

high counts of immunoglobulin-containing cells.Gallbladders from rabbits with bile and the in-farct cultures positive for E. coli had a signifi-cantly increased plasma cell population com-pared with noninfected controls (Table 2). Thenumber of IgG cells exceeded the number of IgAcells, and the number of IgM cells was thesmallest. The number of plasma cells of allimmunoglobulin types was higher 3 weeks afterinfection than 1 week after infection. Gallblad-der IgG/IgA cell ratios did not significantlychange with an increased duration of infection.No correlation was found between serum anti-

body titers of E. coli HS and the number ofplasma cells present in gallbladders infectedwith this organism. Six rabbits with bile infectedwith E. coli HS, sacrificed 1 week after inocula-tion, had a geometric mean serum antibody titerto E. coli of 1:320 + 1:24. A similar group ofeight rabbits with infected bile sampled 3 weeksafter inoculation had a mean serum antibodytiter of only 1:38 ± 1:38 (P < 0.01). Saline-injected controls and rabbits with infected in-farcts but uninfected bile were examined 6weeks after inoculation. Both groups had meanserum antibody titers below 1:20.

S. faecalis-infected gallbladders also had a

413VOL. 36, 1982

on April 15, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

414 GILMAN ET AL.

predominance of IgG-containing cells. Totalcounts tended to be lower than rabbits infectedsimilarly with E. coli.

Fluorescent-antibody studies from intralumi-nally infected gallbladders. In animals given 1 x108 CFU of E. coli HS intravesically, bacteremia(>103 CFU/ml) was present in all animals sam-

pled 15 min after injection. In the lower dosegroup, 5 x 105 CFU of, only two of four rabbitswere bacteremic 15 min post-inoculation. Bothof these rabbits had less than 15 CFU of bacteriaper ml cultured from their blood. No differencewas found in the number or the proportion ofimmunoglobulin cell types in the gallbladders ofrabbits injected with the high- or low-dose in-oculum.At 1 week post-infection with E. coli HS,

gallbladders infected intravesically had moreIgA and fewer IgG cells than did gallbladdersinfected by the intravenous route (Table 2). IgAcells were the major immunoglobulin cell type ingallbladders infected by intravesical injection,whereas IgG cells predominated in gallbladdersinfected intravenously (Table 2).

DISCUSSIONGallbladders were infected without direct

trauma or systemic shock by first infarcting theliver before intravenous injection of bacteria.Confusion caused by anatomical changes in thegallbladder occurring during the production ofinfection was thereby avoided. This model alsoallowed biliary colonization to occur with spe-cies and strains of bacteria, such as E. coli HS

and S. faecalis, which normally would not pro-duce biliary infection after a nonlethal dose.

Gallbladders were infected with S. faecalis, E.coli HS, and E. coli H10407, which producesenterotoxin. The same histological changes, ex-cept for those related to toxin production, wereobserved with each bacterial species. Althoughnot reported here, infection with other bacterialspecies and strains (S. typhi, V. cholerae, andthree strains of E. coli) produced similarchanges. The observed histological and morpho-logical changes were consistent with changeswhich occur secondary to inflammation. In addi-tion, the gallbladder appears to respond to bac-terial infection with a local immune response

which probably accounts for the uniformity ofthe observed histological change.

Specific histological changes were related tothe duration of infection and not to the type ofbacterial agent. Initially, all three bacterialagents produced epithelial necrosis and a hetero-philic inflammatory response. The pathologicallesions observed at 3 and 6 weeks were similarand were associated with (i) distorted cryptswith increased mucus secretion, and (ii) mono-nuclear cell infiltration. Gallbladders infectedwith an enterotoxigenic producing strain of E.coli also had subepithelial capillary enlargementand villus edema.

Like Wienstock and Bonneville (41), we sawapical secretory vesicles in normal rabbit epithe-lial cells; other investigators have noted theirabsence (20). Gallbladders examined 3 and 6weeks after infection had a marked increase inmucus secretion compared with the mucus se-

TABLE 2. Immunoglobulin cell types in gallbladders of control rabbits and rabbits with infected bile

No. of Culture Mean no. of cells per field

Group Inoculuma Route rabbits of Culture (x400)b Mean IgGIsampled infarct of bile IgG IgA IgM

Control Saline Systemic 4 - - 1.0 0.5 0.0Saline Intravesical 5 NA"d 0.4 2.8 12E. coli HS Systemic 2 + - 0.4 0.8 0.4

Infected E. coli HS Systemic 14e + + 77 ± 56 46 + 29 16 + 17 1.95 ± 1.13bile

E. coli HS Intravesical 4 + + 45 ± 28 63 ± 28 19 ± 17 0.66 ± 0.28(1 X 105)

E. coli HS Intravesical 4 NA + 58 ± 13 94 ± 20 NDf 0.64 ± 0.18a E. coli HS, except where noted, was injected at a concentration of 1 x 108 CFU. All rabbits were sampled 1

week after infection.b Values are means ± standard deviation.c The ratio of IgG to IgA plasma cells for each gallbladder was calculated, and the mean ratio ± standard

deviation of each group of gallbladders was then determined. The ratio of IgG to IgA for gallbladders inoculatedintravesically with E. coli HS is 0.65 ± 0.22. This ratio is significantly lower (p < 0.05) than the ratio 1.95 ± 1.13found in gallbladders of rabbits inoculated with E. coli HS systemically and sacrificed at 1 week.

d NA, Not applicable.Includes two rabbits with liver infarcts who were intravenously injected with S x 106 CFU of bacteria on the

day of operation.f ND, Not done.

INFECT. IMMUN.

on April 15, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

BACTERIAL INFECTION OF THE GALLBLADDER 415

cretion from noninfected gallbladder. This reac-tion may be a nonspecific response to inflamma-tion.

Myofibroblasts present in contracting granula-tion tissue (15, 34) were seen in infected gall-bladders and appear to be part of the inflamma-tory response of the gallbladder.

Gallbladders infected with E. coli H10407 andV. cholerae (personal observation) producedsecretory changes that have also been describedin gallbladders actively transporting fluid (23).Cholera toxin injected into rabbit gallbladdersactivates adenyl cyclase and increases the trans-port of water and solute (28). Presumably, thesame mechanism is responsible for the changesseen in gallbladders infected with E. coliH10407.

Bacterial infection or immunization of thegallbladder may provide a useful model in whichto study local immunity. The normal (coccidia-free) rabbit gallbladder and its effluent bile aresterile. The mucosa contains only a few plasmacells, and as a result of this low cellular back-ground, changes in the population of plasmacells due to immunization or infection are readi-ly visable and easily measured.The rabbit gallbladder appears to respond to

bacterial infection with a local immune re-sponse. Evidence of this is the sequence of theinflammatory reaction, the presence of IgA plas-ma cells, the development of lymphoid folliclesand lymphocytic cuffing around vessels, and theuniformity of the pathological response of gall-bladders to different bacteria.

Bacterial infection provoked a vigorous plas-ma cell response in the gallbladder mucosa.Gallbladders infected by intravenous inoculationhad more IgG than IgA plasma cells. In contrast,IgA plasma cells predominated in gallbladdersinfected intravesically.

Antigen applied to mucosal surfaces stimu-lates a local immune response which is primarilyIgA in character (40). Rabbits with infectedgallbladders had a proliferating antigen (bacte-ria) in contact with the gallbladder epithelium,resulting in an increase of IgA plasma cells. Inchronic nonbacterial cholecystitis of humans,most plasma cells found in the gallbladder arealso of the IgA type (8, 18).We also found an increase in IgG cells in the

gallbladders of rabbits infected by both the intra-venous and intravesical routes. Systemic bacter-emia and probable sensitization of the spleenand other nongut lymphoid tissue followed injec-tion with both methods. It was not determinedwhether IgG cells in the gallbladder originatedthere or in peripheral lymph nodes and thespleen.

Biliary infection was short-lived and de-creased sharply by 6 weeks. Similar results have

been found when S. typhi or V. cholerae wereused to infect normal rabbit gallbladders (29,36).The mechanism by which the biliary tract

clears bacteria is, however, unclear. In rats,immunization of the intestine with bacteria ledto the production of specific IgA antibody whichwas transported across the rat hepatocytes andthen concentrated in the bile (19). IgA has beenshown to interfere with the attachment of bacte-ria to epithelial receptors (42) and to facilitatebacterial agglutination (14). Also, as suggestedby this study and studies with V. cholerae (24),biliary clearance of bacteria may occur throughthe development of a local immune response andthe associated production of specific antibody.Evidence for the latter mechanism is the demon-stration by immunofluorescence of plasma cellsin E. coli HS-infected gallbladders which specifi-cally bind E. coli HS antigen (personal observa-tion). Whether either of the above mechanismsis responsible for biliary bacterial clearance stillremains to be demonstrated.As the reaction of the normal rabbit gallblad-

der to infection has been described, it is nowpossible to use this model to investigate themore complex interrelationship of calculi, infec-tion, and the consequent immunopathology.

ACKNOWLEDGMENTSThe advice of D. Nalin, R. B. Sack, N. Pierce, F. Koster,

and M. Schuster is appreciated.Additional support for this study came from K. Schutz and

J. B. Phu and D. Sara.

LITERATURE CITED1. Anderson, T. F. 1951. Technique for preservation of three

dimensional structure in preparing specimens for electronmicroscopy. Trans. N.Y. Acad. Sci. Ser. 130:11-13.

2. Barton, L. L., M. B. Escoledo, J. P. Keating, and J. L.Ternberg. 1973. Leptospirosis with acalculous cholecysti-tis. Am. J. Dis. Child. 126:350-352.

3. Bergan, T., I. Dobog, and I. Lhavag. 1979. Bacterialisolates in cholecystitis. Scand. J. Gastroenterol. 14:625-631.

4. Birbeck, M. S., P. Cartwright, J. G. Hall, E. Orlans, andJ. Peppard. 1979. The transport by hepatocytes ofimmunoglobulin A from blood to bile visualized by auto-radiography and electron microscopy. Immunology37:477-484.

5. Blachstein, A. G. 1891. Intravenous inoculation of rabbitswith the Bacillus coli communis and the Bacillus typhiabdominalis. Bull. Johns Hopkins Hosp. 2:96-103.

6. Bloom, W., and D. Faweett. 1962. p. 75. Textbook ofhistology, 8th ed. The W. B. Saunders Co., Philadelphia,Pa.

7. Bradtzaeg, P., P. Bakllen, 0. Fawn, and P. S. Hoel. 1974.Immunohistochemical characterization of local immuno-globulin formation in ulcerative colitis. Gastroenterology66:1123-1136.

8. Chen, S. T., and T. Tobe. 1974. Cellular sites of immuno-globulins. V. An immunological study of the humangallbladder. Digestion 10:184-190.

9. Chrlstansen, C. R., M. Nielson, and K. F. Meyer. 1921.Do carrier strains differ from strains isolated from ordi-nary typhoid cases? J. Infect. Dis. 28:394-407.

10. Cushing, H. 1899. Observations upon the origin of gall-

VOL. 36, 1982

on April 15, 2020 by guest

http://iai.asm.org/

Dow

nloaded from

416 GILMAN ET AL.

bladder infections and upon the experimental formation ofgallstones. Bull. Johns Hopkins Hosp. 10:166-170.

11. DuPont, H. L., S. B. Formal, R. B. Hornick, M. J. Snyder,J. P. Libonati, D. G. Sheahan, E. H. LaBrec, and J. P.Kalas. 1971. Pathogenesis of Escherichia coli diarrhea. N.Engl. J. Med. 285:1-9.

12. Evett, R. D., J. A. Higgins, and A. L. Brown. 1964. Thefine structure of normal mucosa in human gallbladder.Gastroenterology 47:49-60.

13. Freston, J. W., I. A. Bouchier, and J. Newman. 1969.Biliary mucous substances in dihydrocholesterol inducedcholelithiasis. Gastroenterology 57:670-678.

14. Fubara, E. S., and R. Freter. 1973. Protection againstenteric bacterial infection by secretory IgA antibodies. J.Immunol. 111:395-402.

15. Gabbiani, G., and G. Majno. 1972. Dupuytrens contrac-ture: fibroblast contraction? An ultrastructure study. Am.J. Pathol. 66:131-138.

16. Glenn, F. 1979. Actue acalculous cholecystitis. Ann.Surg. 189:458-465.

17. Gravier, V., G. W. Dorman, and T. Votteler. 1962.Gallbladder disease in infants and children. Surgery63:690-696.

18. Green, F. H. Y., and H. Fox. 1972. An immunofluorescentstudy of the distribution of immunoglobulin-containingcells in the normal and the inflamed human gallbladder.Gut 13:379-384.

19. Hall, J., E. Orlans, J. Reynolds, C. Dean, J. Peppar, L.Gyurel, and S. Hobbs. 1979., Occurrence of specificantibodies of the IgA class in the bile of rats. Int. Arch.All. Appl. Immunol. 59:75-84.

20. Hayward, A. F. 1966. An electron microscopic study ofdeveloping gallbladder epithelium in the rabbit. J. Anat.100:245-259.

21. Hayward, A. F., J. W. Freston, and I. A. D. Bouchier.1968. Changes in the ultrastructure of gallbladder epitheli-um in rabbits with experimental gallstones. Gut 9:550-556.

22. Karnowsky, M. J. 1965. Formaldehyde-glutaraldehydefixative of high osmolarity for use in electron microscopy.J. Cell. Biol. 27:13-14.

23. Kaye, G. I., H. 0. Wheeler, R. T. Whitlock, and N. Lane.1966. Fluid transport in the rabbit gallbladder. A com-bined physiological and electron microscopic study. J.Cell. Biol. 30:237-268.

24. Keclik, M., R. H. Wolf, 0. Felsenfeld, and H. F. Smetana.1970. Immunoglobulins and antibodies in gallbladder bile.Am. J. Gastroenterol. 54:19-29.

25. Lee, M. R., N. S. Ikari, W. C. Branche, and V. M. Young.1966. Microtiter bacterial hemagglutination technique fordetection of Shigella antibodies. J. Bacteriol. 91:463.

26. Lou, M. A., A. K. Mandal, J. L. Alexander, and H.

Thadepalli. 1977. Bacteriology of the human biliary tractand the duodenum. Arch. Surg. 112:965-967.

27. Mallory, T. B., and G. M. Lawson. 1931. Chronic typhoidcholecystitis. Am. J. Pathol. 7:71-75.

28. Mertens, R. B., H. 0. Wheeler, and S. F. Mayer. 1974.Effects of cholera toxin and phosphodiesterase inhibitorson fluid transport and cyclic adenosine 3' 5' monophos-phate concentrations in rabbit gallbladder. Gastroenterol-ogy 67:898-906.

29. Meyer, K. F., N. M. Nielson, and M. L. Feusier. 1921. Themechanism of gallbladder infections in laboratory ani-mals-experimental typhoid-paratyphoid carriers. J. In-fect. Dis. 28:456-509.

30. Mueller, J. C., A. L. Jones, and J. Long. 1972. Topograph-ic and subcellular anatomy of the guinea pig gallbladder.Gastroenterology 63:856-868.

31. Munster, A. M., M. N. Goodwin, and B. A. Pruln, Jr.1971. Acalculous cholecystitis in liver patients. Am. J.Surg. 122:591-594.

32. Reid, M. R., and P. Montgomery. 1920. Acute cholecysti-tis in children as a complication of typhoid fever. Bull.Johns Hopkins Hosp. 31:7-11.

33. Rosenow, E. C. 1916. The etiology of cholecystitis andgallstones and their production by the intravenous injec-tion of bacteria. J. Infect. Dis. 19:527-556.

34. Rudolph, R., W. J. McClure, and M. Woodward. 1979.Contractile fibroblasts in chronic alcoholic cirrhosis. Gas-troenterology 76:704-709.

35. Sainte-Marie, G. 1962. A paraffin embedding technique forstudies employing immunofluorescence. J. Histochem.10:250-256.

36. Schobl, 0. 1916. Further study on experimental choleracarriers. J. Infect. Dis. 19:145-152.

37. Shaw, R. C. 1970. Post-traumatic acute acalculous chole-cystitis in young males. Mil. Med. 135:210-214.

38. Strauss, R. G. 1969. Scarlet fever with hydrops of thegallbladder. Pediatrics 44:741-745.

39. Thorne, G. M., C. F. Deneke, and S. L. Gorbach. 1979.Hemagglutination and adhesiveness of toxigenic Esche-richia coli isolated from humans. Infect. Immun. 23:690-699.

40. Tomasi, T. B., Jr., and J. Blenenstock. 1968. Secretoryimmunoglobulins. Adv. Immunol. 9:1-96.

41. Wienstock, M., and M. A. Bonneville. 1971. Compart-ments rich in acidic carbohydrate-protein complexes withelectrolyte and water-transporting cells. Lab. Invest.24:355-367.

42. Willm, R. C., and R. J. Gibbons. 1972. Inhibition ofbacterial adherence by secretory immunoglobulin A amechanism of antigen disposal. Science 177:697-699.

43. Wolfson, W. L., and R. Rothenberg. 1936. Acute noncal-culous cholecystitis. J. Am. Med. Assoc. 106:1978-1980.

INFECT. IMMUN.

on April 15, 2020 by guest

http://iai.asm.org/

Dow

nloaded from