INFECTION AND IMMUNITY, JUlY 1992, p. 2758-27680019-9567/92/072758-11$02.00/0Copyright © 1992, American Society for Microbiology

Salmonella choleraesuis and Salmonella typhimurium Associatedwith Liver Cells after Intravenous Inoculation of Rats Are

Localized Mainly in Kupffer Cells and Multiply IntracellularlyNDUBISI A. NNALUE,lt* ALEXANDER SHNYRA,1 KJELL HULTENBY,2 AND ALF A. LINDBERG'

Department of Clinical Bacteriology F:821 and The Clinical Research Centre,2Karolinska Institute, Huddinge Hospital, S-141 86 Huddinge, Sweden

Received 2 December 1991/Accepted 16 April 1992

Male Sprague-Dawley rats were inoculated intravenously with Salmonella choleraesuis or Salmonellatyphimurium and used over 3 consecutive days to produce highly enriched (>95% homogenous) preparationsof Kupffer and mononuclear cells (KC), liver endothelial cells (LEC), and hepatocytes. The methods involvedcollagenase perfusion of the liver in situ, differential centrifugation of liver cells over a Percoll gradient, andselective attachment of the cells to plastic or to culture dishes coated with collagen. The different cellpreparations were then assayed for the number and location, intracellular or extracellular, of associated viablebacteria. Most of the viable bacteria recovered were associated with KC and were mainly intracellular. Theintracellular bacteria in KC from rats infected with either bacterial strain increased about 20- to 50-fold over2 days. Some of the bacteria associated with LEC and in some experiments with hepatocytes also survivedtreatment with gentamicin and increased in number with time. Intracellular bacteria were readily visualizedin KC by light microscopy and transmission electron microscopy. On rare occasions, bacteria were seen withinLEC from rats infected with S. choleraesuis but not from those infected with S. typhimurium. Microcolonies ofS. typhimurium but not of S. choleraesuis were occasionally found on the surface of some LEC. Bacteria werenot seen within or on the surface of hepatocytes by transmission or scanning electron microscopy. Theintegration of microscopic and viability data suggested that most intracellular S. choleraesuis organisms in KChad been killed whereas most intracellular S. typhimurium organisms were viable.

The diseases caused by Salmonella species such as ty-phoid fever and severe enteritis complicated by septicemiaremain a global public health problem. Systemic infectionsbegin with the penetration of the intestinal epithelium bysalmonellae and their subsequent dissemination throughoutthe reticuloendothelial system, where they multiply, espe-cially in the liver and spleen. Research over decades hasgenerated and tested a variety of typhoid vaccines (25, 51).Some of these vaccines are widely used, but the incidence ofSalmonella infections remains very high. However, thesurface components of Salmonella species (e.g., lipopoly-saccharides and outer membrane proteins) are stronglyantigenic and elicit good humoral as well as cellular re-sponses in humans and experimental animals (5, 20, 22, 23,24, 39, 44). One puzzling aspect of salmonellosis is the lackof a good correlation between the immunological responseand protective immunity (7, 13, 32). These facts underline adeficit in our understanding of the pathogenesis of systemicSalmonella infections.The earliest studies have identified Salmonella species as

facultative intracellular pathogens, able not only to survivebut also to multiply within the phagocytic cells of their hosts(7, 30). Bacteria which were apparently intact have beendemonstrated in the macrophages of infected animals (45-47), and experiments in cultured macrophages showed thatthe bacteria can survive or multiply within these cells (6, 16,50). Bacterial multiplication in the reticuloendothelial sys-tem during infection has been generally presumed to occur,

* Corresponding author.t Present address: Department of Medical Microbiology, Faculty

of Medicine and Health Sciences, United Arab Emirates University,P.O. Box 17666, Al Ain, United Arab Emirates.

at least partly, within macrophages, but this has never beendirectly demonstrated. The ability of Salmonella species tomultiply or even survive within phagocytic cells, and hencetheir status as intracellular parasites, has been questioned onthe basis of evidence obtained from electron microscopicobservations of the livers of mice infected with a virulentstrain of Salmonella typhimurium, SR-11; this has beeninterpreted to mean that this strain did not multiply intra-cellularly and was rapidly killed once engulfed by macro-phages (19, 27, 28).Most previous studies on the multiplication of salmonellae

in the livers of infected animals have been based on thedetermination of bacterial counts in the entire organ and, insome cases, on the examination of tissue slices by electronmicroscopy. We have taken a different approach in investi-gating early events in the livers of rats after infection witheither of two Salmonella species known to differ in their 0antigens as well as in other properties associated with viru-lence (26, 32, 40). Rats were inoculated intravenously withenough virulent bacteria to initiate a progressive but notrapidly lethal infection. At intervals, selected animals wereanesthetized and their livers were perfused with collagenase-containing medium to allow separation of the different cells.The different cell types, hepatocytes (HEP), Kupffer cells(KC), and liver endothelial cells (LEC), were studied todetermine the numbers and localization of associated bacte-ria.

MATERIALS AND METHODS

Bacteria, their cultivation and animal inoculation. Thevirulent wild-type Salmonella strains used were S. choler-aesuis subsp. kunzendorff strain SL2824 (34) and S. typhi-

2758

Vol. 60, No. 7

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

CELL-ASSOCIATED SALMONELLAE IN RAT LIVER 2759

murium SL3201 (37). They were grown overnight in Lbroth, kept overnight at 4°C while viable counts weredetermined by plating, and then used for the inoculation ofmale Sprague-Dawley rats (ALAB Laboratorietjanst, Sol-lentuna, Sweden). The rats were anesthetized with ether andinoculated with appropriate doses of bacteria in 0.4 ml ofsaline via the lateral marginal vein. For each experiment,three rats were injected with one of the above strains andwere sacrificed on 3 successive days. A total of 12 rats wereused for each bacterial strain.

Perfusion and separation of HEP, KC, and LEC. Themethod described by Seglen (41) was used for the isolation ofHEP. Perfusion was done in situ via the portal vein withCa2+- and Mg2"-free Hanks balanced salt solution (HBSS)at 37°C and a flow rate of 40 ml/min. During the first 10 minof perfusion, normal HBSS was used and the perfusate wasallowed to waste. Subsequently, HBSS containing 0.05%collagenase and 50 mM HEPES (N-2-hydroxyethylpipera-zine-N'-2-ethanesulfonic acid) was substituted and the per-

fusion continued with recirculation of the medium for ca. 25min and continuous pH adjustment to ca. 7.5 with 1 MNaOH. The cell suspension generated was filtered through anylon gauze (mesh width, 50 ,um), and the filtrate was

centrifuged for 5 min (50 x g) to pellet HEP. The pellet wassubjected to two washes (50 x g, 5 min) to remove dead cellsand debris, resulting in a final suspension consisting of atleast 95% HEP. The HEP were found to be 80 to 90% viableas estimated by trypan blue (4 mg/ml) exclusion and weregenerally used immediately for the determination of bacterialcounts or were fixed for electron microscopy.The supernatant obtained after the first centrifugation of

the slurry generated by perfusion of the liver with HBSS-collagenase was used for the preparation of KC and LEC bycentrifugation over a two-step Percoll gradient, as describedpreviously (43). A stock of isotonic Percoll solution was firstprepared by mixing 90 ml of Percoll with 10 ml of a

10-fold-concentrated phosphate-buffered saline (PBS) stock.Next, the gradient was prepared in 50-ml centrifuge tubes,with the bottom layer (15 ml; density, 1.066 g/ml) made up of7.5 ml of PBS and 7.5 ml of isotonic Percoll solution and thetop layer (20 ml; density, 1.037 g/ml) composed of 15 ml ofPBS and 5 ml of isotonic Percoll solution. The HEP-depletedsupernatant was subjected to a second centrifugation (800 xg, 4°C, 10 min), and the pellet was resuspended in 40 ml offresh HBSS. Ten milliliters of the suspension was applied toeach gradient and centrifuged (400 x g, 4°C, 20 min). A bandat the boundary of the Percoll layers was carefully aspiratedfrom each tube (5 ml). This is an enriched fraction containingmainly LEC and some KC, but it is practically free of debris,HEP, and erythrocytes. The majority of KC, distributedthroughout the lower Percoll cushion, were harvested bydilution (threefold) in PBS and subsequent centrifugation(800 x g, 4'C, 10 min). The pellet was resuspended in culturemedium, RPMI 1640, with or without gentamicin (100 ,ug/ml).

In the final steps of cell separation, KC were selectivelyremoved from the fractions by selective attachment ontoplastic. The resuspended pellet of KC was plated at 106 cellsper cm' in six-well cluster dishes (Becton Dickinson). KCwere allowed to attach to the plastic for 15 min and were

washed several times with PBS to remove unattached cells,mainly LEC. The KC present in the enriched fraction ofLEC were further depleted by incubation in petri dishes for15 min. The LEC, which do not attach to plastic, remainedin the supernatant and were collected and plated (2 ml perwell at 2 x 106 cells per ml) in six-well tissue culture dishes

coated with rat tail collagen (3 pug/cm2). After a 2-h incuba-tion, the plates were washed with PBS to remove dead andother nonadherent cells, and fresh medium, RPMI 1640 withor without gentamicin (100 ,ug/ml), was added to each well.

Determination of viable counts in liver cell populations. TheLEC and KC from infected animals were plated with orwithout gentamicin to allow the determination of total andintracellular bacterial counts. Wells plated without gentami-cin were used as soon as possible after isolation for adetermination of the counts of viable bacteria in the prepa-ration. Wells containing gentamicin were incubated for 90min to allow killing of all extracellular bacteria before use forthe determination of intracellular counts. In either case, theculture medium was aspirated and the cells were washedthree times with PBS. Then, 100 1±l of 1% Triton X-100 inPBS was added to each well and incubated for 5 min at 37°C;the resulting medium was agitated on a microplate mixer(Titertek) to assure complete lysis of the cells. The resultingsuspension was diluted in PBS and plated on L agar.

Light microscopy. Cells attached to the bottom of plastictissue culture dishes were stained by Giemsa stain for lightmicroscopy. They were examined and photographed underoil immersion with the Ultraphot Universal Photomicro-scope (Carl Zeiss, Oberkochen, Germany).TEM and SEM. Preparation of cells for transmission

electron microscopy (TEM) (21) and scanning electron mi-croscopy (SEM) (42) was done by standard procedures.Liver sinusoidal cells (KC and LEC) on a cover glass andHEP in suspension were incubated for 1 h at 37°C and thenovernight at 4°C in fixative (2.5% glutaraldehyde in 0.1 Mcacodylate buffer [pH 7.3], 0.1 M sucrose). For TEM, thecells were washed after fixation with 0.15 M cacodylatebuffer and LEC and KC were scraped off the cover glasswith a wooden peg. The cells were transferred to microcen-trifuge tubes and postfixed for 1 h at 4°C in 1% osmiumtetroxide-0.15 M cocadylate buffer (pH 7.4)-3 mM CaCl2.After centrifugation, the cells were dehydrated in a series ofethanol solutions of increasing concentrations, the last ofwhich contained 1% uranyl acetate. Finally, the cells weredehydrated in acetone and embedded in LX-112 epoxy resin(Polysciences, Inc., Warrington, Pa.). Sections were cutwith a diamond knife and contrasted by treatment with 1%uranyl acetate and then lead citrate. The sections were thenexamined in a Philips 420 electron microscope at 100 kV.The preparation of cells for SEM was basically the same asthat described for TEM with a few minor differences. Afterfixation, the sinusoidal cells were not scraped off the coverglass but were rinsed in distilled water and then subjected todehydration as described above. HEP were attached to a0.1-,um-pore-size nylon membrane (Sartorius, Gottingen,Germany) pretreated with poly-L-lysine (0.5 mg/ml) beforebeing dehydrated. After dehydration, the cells were im-mersed in tetramethylsilane and air dried as described pre-viously (10). The dried specimens were sprayed with gold-palladium (Polaron; Wortford, England) to give a 10-nmcoating and then examined in a Philips 501 SEM at 15 kV.

Identification of LEC and KC by ovalbumin accumulationand rosetting of erythrocytes. LEC were identified in cultureby the selective uptake of fluoresceinamine-conjugated oval-bumin prepared as described previously (2). This solutionwas added to cultures in RPMI 1640 medium to a finalconcentration of 100 ,ug/ml and incubated for 2 h at 37GC.The culture wells were rigorously washed with HBSS andthen examined under a Nikon Labophot fluorescence micro-scope to identify the LEC which fluoresce as a result of theaccumulation of this conjugate. Similarly, KC in culture

VOL. 60, 1992

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

2760 NNALUE ET AL.

10 8

10 7

0

= i 6-

UD

10

10 8

10 7

10 6

10 5

10

A

._

5).5)

1 2 3 4 5

Days

B

1 2 3 4 5

Days

were distinguished from LEC on the basis of their ability toform rosettes with sheep erythrocytes fixed with glutaralde-hyde as previously described (38). Preparations of KC were

incubated with the modified sheep erythrocytes for 30 min at37°C, washed with HBSS, and then examined microscopi-cally to determine their purity. This method does not allow a

distinction of KC from mononuclear cells.Peroxidase staining. KC were distinguished from other

mononuclear cells by staining for endogenous peroxidase.Cultures were fixed for 5 min in glutaraldehyde as describedabove for electron microscopy, stained with 1% diaminoben-zidine tetrahydrochloride (Fluka, Buchs, Switzerland)-0.02% hydrogen peroxide in 0.1 M Tris, pH 7.6, for 30 min atroom temperature (22°C) (53), and examined by light micros-copy. KC characteristically show diffuse brown staining ofthe cytoplasm and a dark perinuclear rim from peroxidaseactivity, whereas mononuclear cells show granular staining(3).

RESULTS

Cell preparations. The methods we used for separation arewell established and reproducible and give high yields ofhomogenous preparations of the different liver cell types(43). Nevertheless, we rigorously monitored every prepara-tion for purity by microscopy as well as by specific tests forthe different cell types. The tests showed that the cellseparation procedure functioned well in uninfected as well as

10 8

10 7

10 6

10 5

10 4

C

0 1 2 3 4 5

Days

FIG. 1. Kinetics of bacterial growth in the livers and/or spleensof rats after i.v. infection with 5 x 107 CFU of virulent salmonellae.(A) Results from combined livers and spleens of rats infected with S.choleraesuis (0) or S. typhimurium (0); (B) comparative growth ofS. choleraesuis in liver (l) and spleen (-); (C) comparative growthof S. typhimurium in liver (O) and spleen (-). Each datum pointrepresents the geometric mean of results from three rats. Error barsindicate standard deviations.

infected animals and that essentially pure preparations ofeach kind of cell were obtained. Preparations of KC were 85to 92% homogenous and contained monocytes (3 to 10%) aswell as other leukocytes and LEC (5%). The LEC prepara-tions were at least 95% pure, as determined by the ability ofthe cells to accumulate fluoresceinamine-labeled ovalbumin.The contaminants were mainly KC and monocytes (1 to 5%)as determined by the ability of these cells to form rosetteswith sheep erythrocytes. The yields of KC (1 x 107 to 1.4 x107) and LEC (1.8 x 107 to 2.2 x 107) from infected rats werewithin the range of values expected from normal animals(22). The HEP preparations were also essentially pure(>95%). The contaminants included KC, LEC, and fat-storing cells, which were readily identified by their autoflu-orescence (from vitamin A). The presence of these cells inHEP preparations is not indicative of incomplete separationof liver cells but is more likely a consequence of trapping bythe reaggregation of HEP, as has been previously demon-strated (52).

Bacterial distribution and multiplication in liver cells. Al-though the mouse is the most widely used model for exper-imental salmonellosis, we chose rats for this study becausetheir larger size more readily accommodated our need forregular, successful perfusion and cell separation. The appli-cability of rats for experimental studies of Salmonella infec-tion has also been shown (12, 17, 18). Our first infectionexperiments were used for determining the appropriate doseof each strain sufficient to initiate progressive infection inmost or all of the rats inoculated, but not enough to leadrapidly to death. A dose of 5 x 106 CFU injected intrave-nously (i.v.) was found insufficient to initiate progressiveinfection. Very few bacteria were recovered from liver cellson days 1 to 3, and animals not used for cell separationremained healthy. A higher dose of 2 x 107 CFU was tried,but the recovery of bacteria from rats was quite irregular andtwo rats not sacrificed for cell separation survived until day14. Next, we tested a dose of 5 x 107 CFU, and this resultedin a progressive infection by S. choleraesuis SL2824 or S.typhimunum SL3201. The bacterial counts recovered from

.5

U

I-

5)

INFECT. IMMUN.

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

CELL-ASSOCIATED SALMONELLAE IN RAT LIVER 2761

TABLE 1. Distribution and localization of salmonellae in different liver cells

Mean (SD) CFU recovered'

Species and day Kc LEC HEP

Total Intracellular Total Intracellular Total Intracellular

S. choleraesuis1 1.7 x 104 (1.2) 6.5 x 103 (2.4) 2.5 x 103 (2.1) 1.2 x 103 (2.0) 2.0 x 103 (1.5) 3.4 x 102 (2.4)2 1.5 x 105 (1.2) 8.4 x 104 (1.3) 1.1 x 104 (1.6) 3.7 x 103 (1.8) 5.5 x 104 (1.2) 3.0 x 104 (1.1)3 4.3 x 105 (1.3) 3.5 x 105 (1.2) 7.0 x 104 (1.3) 3.2 x 104 (1.3) 9.5 x 104 (1.8) 6.7 x 104 (1.6)

S. typhimunum1 2.2 x iO4 (1.7) 1.2 x io4 (2) 6.5 x 103 (1.4) 5.9 x 103 (1.5) 4.6 x 103 (1.1) 3.7 x 103 (1.2)2 2.5 x 104 (1.5) 1.2 x 104 (1.5) 4.9 x 102 (1.5) 4.2 x 102 (1.7) 1.1 x 104 (1.4) 1.1 x 104 (1.6)3 3.3 x 105 (1.1) 2.4 x 105 (1.1) 4.5 x 103 (1.5) 3.6 x 103 (1.3) 5.4 x 104 (1.3) 4.3 x 104 (1.3)

a Values represent the geometric means and standard deviations (SD) of CFU recovered per million host cells in three separate experiments. SDs are expressedgeometrically.

the different types of liver cells were not only high enoughfor accurate quantitation on day 1 but they also increasedsubsequently. Five rats inoculated with 5 x 107 CFU of S.choleraesuis died 4 or 5 days later, and of six rats given thesame dose of S. typhimuiium, three died on day 5 or 6 andthree survived. These data are, in general, consistent withprevious studies. Edwards et al. (12) recently reported thatthe 50% lethal dose for Sprague-Dawley rats of a strain of S.typhimurium is 109 CFU. In another study, 106 CFU of S.typhimunum injected i.v. did not cause illness in LEW rats,whereas 108 CFU led to progressive infection and rapiddeath (18). It is evident from our data that there was only asmall difference (10-fold or less) between the doses of ourSalmonella strains that were quickly eliminated and dosesthat caused death. We, therefore, investigated the kinetics ofbacterial growth in the liver and spleen in order to bettervisualize the infection process. The bacteria localized andmultiplied progressively in the liver and spleen, as is nor-mally observed in experimental models of salmonellosis.The viability curves obtained with either S. choleraesuis orS. typhimurium were virtually identical (Fig. 1). About 1.5 to4% of the initial inoculum of either species was recoveredfrom the liver and spleen on day 1. The counts decreasedanother three- to sevenfold on day 2 but subseq7uentlyincreased exponentially to about 3 x 107 to 7 x 10 CFU(about equal to the initial inoculum) on day 4.The results of six experiments on the recovery of bacteria

from the livers of rats infected with S. choleraesuis or S.typhimurium (three experiments with each strain) are shownin Table 1. On day 1 after the i.v. administration of S.choleraesuis, the total bacterial count associated with KC(1.7 x 104) was seven- to ninefold more than that associatedwith LEC (2.5 x 103) or HEP (2.0 x 103). Many of thebacteria in KC (40%), LEC (48%), and HEP (18%) survivedtreatment with gentamicin. The fact that living mammaliancells are poorly permeable to gentamicin and consequentlythat bacteria which survive gentamicin treatment under theconditions we used are intracellular in living cells is wellestablished. We hereinafter refer to bacteria recovered afterthe treatment of cell preparations with gentamicin as intra-cellular, although they may not necessarily be located withinthe predominant cell type in a specified preparation. On days2 and 3, the intracellular bacteria in KC preparations suc-cessively increased 11- and 5-fold, respectively, while thosein LEC preparations increased 3- and 2-fold, respectively.The intracellular bacteria in the HEP preparations increasedca. 86-fold on day 2 and only ca. 2-fold on day 3. On each

day of infection, the intracellular bacteria in KC prepara-tions were 3 to 21 times greater in number than those in theLEC or HEP preparations. Also, an over-50-fold increase inthe number of intracellular bacteria in KC preparationsoccurred between day 1 and day 3. Since our KC prepara-tions were practically devoid of other liver cell types andcontained only a maximum of 5% of other leukocytes, ourdata would support a conclusion that most of the intracellu-lar bacteria in KC preparations were actually within KC.The results of the experiments with S. typhimurium were

similar to those obtained with S. choleraesuis. However,there was a decrease (LEC) or only a minimal increase (lessthan threefold in KC or HEP) in bacterial counts betweenday 1 and day 2 after infection with S. typhimurium. Thebacterial counts increased significantly on day 3 in KC (by 13to 20 times) and LEC (9 times), while a somewhat lesserincrease occurred in HEP (4 times). As observed in experi-ments with S. choleraesuis, the treatment of cells withgentamicin had little or no effect on the bacterial countsassociated with all three cell types. Thus, S. typhimuriumwas mainly intracellular, at least in KC. On days 1 and 2,intracellular bacteria were at least as high in number in KCas they were in LEC or HEP preparations, but on day 3, KCcontained 6 to 67 times more intracellular bacteria than LECor HEP. The number of intracellular bacteria in KC in-creased 20-fold between days 1 and 3.

Light microscopic observations. Aliquots of cell prepara-tions were stained by the Giemsa method and examined bylight microscopy. In rats infected with S. choleraesuis, onlyabout 5% of the KC contained bacteria (generally one or twobacteria per cell) on day 1 (data not shown). The percentageof infected cells increased to ca. 10% on day 2, and each cellcontained from one to several bacteria (data not shown). Byday 3, however, most KC (50 to 85%; average, 63%) wereinfected and contained up to 50 bacteria per cell (Fig. 2A).Many KC were literally full of bacteria, and some of thebacterial clusters were in obvious vacuoles (Fig. 2B). Few orno bacteria were seen at the edges of the cells, an indicationthat few bacteria were on the cell surface. In most cases, theintracellular bacteria were in large clusters. A cluster ofbacteria within a cell could, in theory, originate either by thelocal multiplication of bacteria or by the coalescence ofbacteria accumulated from the extracellular milieu. As mostKC were infected by day 3 and infected cells usuallycontained many bacteria, the recovery of a total of only 3.3x 105 CFU from 106 KC was much lower than expected. Weinfer that most of the intracellular bacteria visualized micro-

VOL. 60, 1992

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

2762 NNALUE ET AL.

A

1:q1

-~~~~~~

_1.

4._.wI#,*F

B

A

f '...

:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.iJ 4

FIG. 2. Photomicrograph showing clusters of intracellular bacteria (arrowheads) in KC isolated from rats on day 3 after infection withS. choleraesuis. The bacteria were generally in vacuoles, some of which are quite obvious (B).

scopically were no longer viable, although they were mor-phologically intact. KC from infected rats contained a largenumber of empty vacuoles, and some of them were enlarged.These features were not seen in KC from uninfected rats orin LEC from infected as well uninfected rats and might beindicative of degenerative processes.The LEC were not as extensively infected as the KC,

which agrees with bacterial viability recovery data (Fig. 3).It seems that many of the bacteria recovered from the LECfraction were contributed by contamination from infectedKC (Fig. 3A). However, on rare occasions (about 1 in 1,000cells observed), a bona fide LEC was seen infected by a fewbacterial cells (Fig. 3B). After Giemsa staining, HEP insuspension were opaque and did not allow visualization ofany intracellular bacteria. Bacteria were not seen on thesurface of HEP (data not shown).The features of cells from rats infected with S. typhimu-

num differed from those of rats infected with S. choleraesuisin several respects. KC infected with S. typhimurium con-tained far fewer bacteria than were seen in those infectedwith S. choleraesuis. The bacteria were generally scatteredthroughout the cell rather than in clusters, but an occasionalKC with clusters of intracellular bacteria was seen (Fig. 4A).Even on day 3 after infection, many KC from rats inoculatedwith S. typhimurium remained uninfected (Fig. 4B). Thenumber of viable bacteria recovered from KC on all dayswould seem to match the degree of infection observedmicroscopically. This indicates that many or all of theintracellular bacteria were viable. As was the case with S.

choleraesuis, many of the CFU recovered from the LECfraction seemed to be due to contamination by infected KC(data not shown). Bacteria were also not seen on the surfaceof HEP.SEM and TEM. When examined by SEM, KC from

infected animals (Fig. 5A and B) had the appearance oftypical macrophages. They showed surface folds and finger-like processes at the central part of the cell and flattenedtonguelike processes at the periphery. An extensive searchthrough many preparations did not reveal bacteria on thesurfaces of these cells. This would support evidence fromlight microscopy and from a determination of viable countswhich showed that many or all of the bacteria associatedwith KC were intracellular. The cells in the LEC fractionalso had the expected morphology (Fig. 5C) and lacked theextensive folds and processes associated with KC. Bacteriawere not seen on the surfaces of most of these cells. Despitean extensive and careful search, no bacteria were seen onthe surfaces of LEC from rats infected with S. choleraesuis,whereas some of the LEC (about 2 in 1,000) from ratsinfected with S. typhimunum had microcolonies of bacteriaon their surfaces (Fig. 5D). HEP isolated from uninfectedanimals were found by SEM to be generally intact (Fig. 6A),while those from infected animals contained some damagedcells (Fig. 6B) with broken surfaces that revealed roundedsubcellular structures. These rounded structures are notbacteria but are apparently mitochondria, as indicated byTEM (data not shown). The absence of bacteria on the

INFECT. IMMUN.

.1i.

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

CELL-ASSOCIATED SALMONELLAE IN RAT LIVER 2763

S

0

"A0FIG. 3. Photomicrograph of LEC from a rat infected with S. choleraesuis showing the absence of intracellular bacteria from most LEC

but the presence of a large intracellular cluster in a KC contaminant (A) and a rare LEC containing a small cluster (arrowhead) of intracellularbacteria (B).

A

t p jt4 -

90I

i1.yi

I0Al

t I

Y43

S.

q&.

FIG. 4. KC isolated from rats on day 3 after infection with S. typhimurium showing the presence of relatively few intracellular bacteria.The bacteria in KC were generally scattered throughout the cell, but a few small clusters (arrowhead) were also seen (A) and many KCremained uninfected (B).

VOL. 60, 1992

do

AA

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

2764 NNALUE ET AL.

FIG. 5. SEMs of KC (A and B) and LEC (C and D) from infected rats. Bacteria were absent from the surfaces of most of these cells, butan occasional microcolony was seen on the surface of LEC (D). Bars, 7.0 (A and C) and 0.7 (B and D) ,um.

surface of HEP as seen by SEM shows that bacteria did notmultiply attached to the cell surface.The different cell fractions from infected animals were also

examined by TEM. Nearly all of the bacteria seen in KCfrom rats infected with S. choleraesuis showed signs ofdegeneration, as defined by Guo et al. (15), such as centralvacuolation, peripheral condensation, surface compression,and discontinuity of the cell envelope (Fig. 7A, arrowhead).Some morphologically intact bacteria were also seen in someKC (Fig. 7B). The results were different with KC from ratsinfected with S. typhimurium, in which most of the intra-cellular bacteria seemed undamaged and viable (Fig. 7C). Allintracellular bacteria, S. choleraesuis or S. typhimunum,were present in vacuoles and bounded by membranes.Intracellular membrane-bound bacteria were seen in LECpreparations from animals infected with S. choleraesuis(Fig. 8) but not in those infected with S. typhimurium. Thebacteria appeared intact but were sometimes found in thesame vacuole with some ill-defined debris, possibly resultingfrom disintegrated bacteria. Intracellular bacteria were not

seen in HEP infected with S. choleraesuis or S. typhimu-ium, despite a rigorous and careful search through manypreparations. Some KC containing bacteria were seen in theHEP preparation, apparently entrapped by reaggregatedHEP (data not shown). These most likely have contributedpart of the gentamicin-resistant fraction of the bacteriaassociated with HEP.

DISCUSSION

Our data show that, during the first 3 days after the i.v.infection of rats with Salmonella, viable bacteria can berecovered from preparations of each liver cell type aftertreatment with gentamicin to kill extracellular bacteria. In allexperiments, more bacteria per host cell survived gentami-cin treatment in the KC than in the LEC or HEP prepara-tions. Microscopic evidence suggests that most of the intra-cellular bacteria in the LEC and HEP preparations resultedfrom contamination with infected KC. That the intracellularbacteria associated with KC preparations were actually

INFECT. IMMUN.

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

CELL-ASSOCIATED SALMONELLAE IN RAT LIVER 2765

FIG. 6. SEMs showing intact HEP from normal rats (A) and damaged ones from infected rats (B). Magnification, x4,700.

within KC was microscopically demonstrated as well as

deduced from the fact that KC preparations were not con-taminated by other liver cell types and were only minimallycontaminated by leukocytes other than mononuclear cells.We did not observe bacteria on the surface of KC, althoughsome of the bacteria associated with KC preparations were

susceptible to gentamicin and were apparently extracellular.Such bacteria are few (less than one per KC), however, andmight have been hidden in the extensive folds and processes

of infected KC. They might also have been within damagedcells or attached to extracellular debris and thus were

exposed to gentamicin. The number of intracellular bacteriain KC from rats infected with S. choleraesuis increased morethan 50-fold in 2 days, whereas those from rats infected withS. typhimurium showed about a 20-fold increase in the same

period. Intracellular S. choleraesuis in KC increased innumber within 48 h postinfection, whereas intracellular S.typhimurium increased in number only after 48 h. After thei.v. inoculation of rats with either bacterial species, the totalbacterial counts in the liver and spleen declined betweendays 1 and 2 but began to increase exponentially by day 3.Thus, in rats infected with S. choleraesuis, the total bacterialcounts in the liver decreased from day 1 to day 2 whileintracellular bacteria in KC increased. These findings mayindicate that kinetic data obtained at the organ level may notreflect events at the cellular level.

Light microscopy supported our quantitative observationof progressive increase in the number of viable bacteriaassociated with KC. The proportion of infected cells as wellas the number of intracellular bacteria associated with eachinfected cell increased progressively. Most of the KC fromrats infected with S. choleraesuis were infected by day 3 andcontained mainly clusters of apparently intact bacteria. Webelieve that these clusters in most cases have resulted fromthe intracellular multiplication of S. choleraesuis rather thanfrom the coalescence of bacteria phagocytosed at differenttimes. Even if the latter were the case, the presence ofclusters would indicate that S. choleraesuis SL2824 can

remain morphologically intact long after phagocytosis. One

other finding in support of our hypothesis that S. cholerae-suis had multiplied intracellularly is our observation that,while some KC contained large numbers of intracellularbacteria, others were apparently uninfected. This suggests afluctuation phenomenon consistent with evidence that mac-rophages differ in the ability to kill Salmonella species (35,40) and that salmonellae can replicate in at least somemacrophages or spleen cells (1, 4, 9, 11). It is unlikely to bethe result of a lack of contact between some KC and bacteriaover a 3-day period of rapidly progressive infection (andinflammation). It seems that most of the intracellular S.choleraesuis organisms seen microscopically were dead,since the average number of viable bacteria per KC wasmuch less than would be expected from microscopic data.This is buttressed by our electron microscopic observationthat most intracellular S. choleraesuis organisms showedsome evidence of degeneration and conforms with ourexpectation on the basis of the well-proven ability of phago-cytic cells to kill salmonellae (29, 33, 48, 49).

S. typhimurium differed from S. choleraesuis because itshowed a lesser tendency to form intracellular clusters ofbacteria and was seen in fewer KC at comparable levels ofinfection (as determined by viable counts). The fact that theviable counts of S. typhimurium recovered from KC corre-lated with the microscopic estimates indicated that mostintracellular S. typhimurium bacteria were alive. The resultsof TEM support this deduction because most of the bacteriaseen were apparently intact. Taken together, our data sug-gest that S. typhimurium SL3201 has a lesser capacity thanS. choleraesuis SL2824 to multiply intracellularly but abetter ability to survive within KC. This was also the mainconclusion of two recent studies based on the properties ofSL3201 and SL2824 in mice and of their aromatic-dependentmutants in mouse peritoneal macrophages in vitro (32, 33).

Intracellular bacteria were not seen within LEC from ratsinfected with S. typhimurium but were seen only on rareoccasions (about 1 in 1,000 cells examined) in LEC from ratsinfected with S. choleraesuis. Our LEC preparations werecontaminated with KC (2 to 5%), some of which contained

VOL. 60, 1992

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

2766 NNALUE ET AL.

..

FIG. 8. TEM of an LEC with intracellular S.Magnification, x 16,000.

choleraesuis.

..,.tk.

FIG. 7 TEMs of infected KC Many intracellular S. cholerae-

suis organisms (arrowhead) showed evidence of degeneration (A),

but some were apparently intact (B). Most intracellular S. typhimu-rium organisms were intact (C). Magnifications: x 11,500 (A),x26,000 (B), and x9,100 (C).

clusters of intracellular bacteria. We surmise that most of thebacteria associated with LEC were in infected KC and thatvery few LEC had intracellular Salmonella infections. Onrare occasions, S. typhimurium but not S. choleraesuis wasfound growing as microcolonies on the surface of LEC. It isnoteworthy that perfusion studies in mice have shown thatLEC as well as KC can trap S. typhimunum (48) and that thisis mediated, at least in part, through binding to carbohydratereceptors (31). The rarity of these findings notwithstanding,they might represent properties of significance for pathogen-esis.

Lin et. al. (27) have reported electron microscopic evi-dence for the invasion and multiplication of salmonellaewithin HEP in vivo and suggested that HEP might provide asafe haven where salmonellae may multiply inaccessible toinfiltrating phagocytes (28). Although moderate numbers ofthe bacteria associated with HEP survived gentamicin treat-ment, we did not visualize salmonellae within HEP by lightmicroscopy, TEM, or SEM. On the contrary, KC containingintracellular bacteria were seen in the HEP preparations.Therefore, while we do not exclude the presence of intra-cellular bacteria in HEP, the data show that many intracel-lular bacteria in HEP preparations were actually in KCpresent as contaminants.Although infections by Salmonella species have been

studied for decades using in vitro and in vivo models,

INFECT. IMMUN.

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

CELL-ASSOCIATED SALMONELLAE IN RAT LIVER 2767

various aspects of pathogenesis remain quite controversial(13, 19). As recently pointed out in a review, the classifica-tion of Salmonella species as facultative intracellular para-sites has not been indisputably substantiated by experimen-tal evidence (19). This is an area in which we envisage thatthe experimental model we describe might prove quiteuseful. Many of the studies that have directly examinedevents in the liver during Salmonella infection have beenbased either on gross determinations in the entire organ oron electron microscopy examination of thin liver sections.The major drawback of both systems is that they provide avery limited appreciation of events in the entire organ. Grossdeterminations lack resolution and do not show events insufficient detail, while direct electron microscopy studiesvisualize only thin sections of extremely minute areas oftissue and consequently generate data that are not repre-sentative of the entire organ. Our method not only avoidsthese drawbacks but also avoids the problem of guessworkwith respect to bacterial viability.Much evidence indicates uncontrolled extracellular multi-

plication of salmonellae at the terminal stages of experimen-tal infections (19, 32, 33). The failure to visualize manyintracellular salmonellae in macrophages during experimen-tal infection of mice has been reported (15, 19). These havebeen interpreted to mean that Salmonella species do notmultiply or survive within phagocytic cells. Our data bycontrast show a large increase (20- to 50-fold) in the numberof viable intracellular bacteria associated with KC over 2days of infection. It is noteworthy that large doses ofbacteria are often used for studies of the location of bacterialmultiplication in vivo in order to obtain results before theonset of the immune response. In some studies, as much asan estimated 200,000 x 50% lethal dose of a virulent strain ofS. typhimurium was administered i.v. to highly susceptiblemice (28). This most likely does not reflect the natural courseof events. Bacteria (except obligate intracellular parasites)are known to multiply much faster extracellularly thanintracellularly (8), and thus any unphagocytosed fraction ofan inoculum would quickly replicate and become dominant.Therefore, uncontrolled extracellular multiplication wouldbe a natural result of systemic challenge with a massive doseof virulent bacteria and is, perhaps, irrelevant to the ques-tion of facultative intracellular parasitism.

Naturally acquired systemic Salmonella infections mostlikely begin with the entry of a few bacteria into the tissuesor systemic circulation. We consider it unlikely that theterminal, fulminant stage of any disease, as is often rapidlyinduced in many experimental systems, would truly depictthe decisive events in pathogenesis. Perhaps, the ability tosurvive and multiply in phagocytic cells is relevant only tothe early stages of disease initiated from a small inoculum orto other situations such as the prolonged persistence ofbacteria in vivo (36). Careful studies of events in thesesettings would seem to be a logical path towards the dissec-tion of host-parasite interactions in the pathogenesis ofsystemic salmonellosis. The model we describe would seemideal for such investigations as well as for the role of humoraland cellular factors in protection against Salmonella infec-tion.

ACKNOWLEDGMENTS

This work was supported by the Swedish Medical ResearchCouncil (grant 16X-656).We thank Leon Rosenberg for critical reading of the manuscript.

REFERENCES1. Benjamin, W. H., Jr., P. Hall, S. J. Roberts, and D. E. Briles.

1991. The primary effect of the Ity locus is on the rate of growthof Salmonella typhimurium that are relatively protected fromkilling. J. Immunol. 144:3143-3151.

2. Blomhoff, R., B. Smedsrod, W. Eskild, P. E. Granum, and T.Berg. 1984. Preparation of isolated liver endothelial cells andKupffer cells in high yield by means of an enterotoxin. Exp. CellRes. 150:194-204.

3. Bouwens, L., and E. Wisse. 1985. Proliferation, kinetics and fateof monocytes in rat liver during a zymosan-induced inflamma-tion. J. Leukocyte Biol. 37:531-543.

4. Briles, D. E., W. Benjamin, Jr., B. Posey, S. M. Michalek, andJ. R. McGhee. 1986. Independence of macrophage activationand expression of the alleles of the Ity (immunity to typhimu-rium) locus. Microb. Pathog. 1:33-41.

5. Brown, A., and C. E. Hormaeche. 1989. The antibody responseto salmonellae in mice and humans studied by immunoblots andELISA. Microb. Pathog. 6:445-454.

6. Buchmeier, N. A., and F. Heffron. 1989. Intracellular survival ofwild-type Salmonella typhimurium and macrophage-sensitivemutants in diverse populations of macrophages. Infect. Immun.57:1-7.

7. Collins, F. M. 1974. Vaccines and cell-mediated immunity.Bacteriol. Rev. 38:371-402.

8. Collins, F. M. 1979. Cellular antimicrobial immunity. Crit. Rev.Microbiol. 7:27-91.

9. Collins, F. M., and S. G. Campbell. 1982. Immunity to intra-cellular bacteria. Vet. Immunol. Immunopathol. 3:5-66.

10. Dey, S., T. S. B. Baul, B. Roy, and D. Dey. 1989. A new rapidmethod for air-drying for scanning electron microscopy usingtetramethylsilane. J. Microsc. 156:259-261.

11. Dunlap, N. E., W. H. Benjamin, Jr., R. D. McCall, Jr., A. B.Tilden, and D. E. Briles. 1991. A 'safe-site' for Salmonellatyphimurium is within splenic cells during the early phase ofinfection in mice. Microb. Pathog. 10:287-310.

12. Edwards, C. K., Ill, L. M. Younger, R. M. Lorence, R. Dantzer,and K. W. Kelley. 1991. The pituitary gland is required forprotection against lethal effects of Salmonella typhimurium.Proc. Natl. Acad. Sci. USA 8:2274-2277.

13. Eisenstein, T. K., and B. M. Sultzer. 1983. Immunity to Salmo-nella infection, p. 261-296. In T. K. Eisenstein, P. Actor, and H.Friedman (ed.), Host defenses to intracellular pathogens. Ple-num Press, New York.

14. Friedman, R. L., and R. J. Moon. 1977. Hepatic clearance ofSalmonella typhimurium in silics-treated mice. Infect. Immun.16:1005-1014.

15. Guo, Y.-N., H. S. Hsu, V. R. Mumaw, and I. Nakoneczna. 1986.Electron microscopy studies on the bactericidal action of in-flammatory leukocytes in murine salmonellosis. J. Med. Micro-biol. 21:151-159.

16. Harrington, K. A., and C. E. Hormaeche. 1986. Expression ofthe innate resistance gene Ity in mouse Kupffer cells infectedwith Salmonella typhimurium in vitro. Microb. Pathog. 1:269-274.

17. Hougen, H. P., and E. T. Jensen. 1990. Experimental Salmo-nella typhimunum infection in rats. III. Transfer of immunitywith primed lymphocyte subpopulations. APMIS 98:1015-1021.

18. Hougen, H. P., E. T. Jensen, and B. Klausen. 1990. Experimen-tal Salmonella typhimurium infection in rats. I. Influence ofthymus on the course of infection. APMIS 97:825-832.

19. Hsu, H. S. 1989. Pathogenesis and immunity in murine salmo-nellosis. Microbiol. Rev. 53:390-409.

20. Kantele, A., H. Arvilommi, J. M. Kantele, and P. H. Makela.1991. Comparison of the human immune response to live oral,killed oral or killed parenteral Salmonella typhi TY21A vac-cines. Microb. Pathog. 10:117-126.

21. Knook, D. L., N. Blansjaar, and E. C. Sylvester. 1977. Isolationand characterization of Kupffer cells and endothelial cells fromrat liver. Exp. Cell Res. 109:317-329.

22. Knook, D. L., and E. C. Sylvester. 1976. Separation of Kupffercells and endothelial cells of the liver by centrifugal elutriation.Exp. Cell Res. 99:444-449.

VOL. 60, 1992

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

2768 NNALUE ET AL.

23. Kuusi, N., M. Nurminen, H. Saxen, and P. H. Makela. 1981.Immunization with major outer membrane protein (porin) prep-arations in experimental salmonellosis: effect of lipopolysaccha-ride. Infect. Immun. 34:328-332.

24. Levine, D. M., K. H. Wong, H. V. Reynolds, A. Sutton, and R. S.Northrup. 1975. Vi antigen from Salmonella typhosa and immu-nity against typhoid fever. II. Safety and immunogenicity inhumans. Infect. Immun. 12:1290-1294.

25. Levine, M. M., J. B. Kaper, R. E. Black, and M. L. Clements.1983. New knowledge on pathogenesis of bacterial entericinfections as applied to vaccine development. Microbiol. Rev.47:510-550.

26. Liang-Takasaki, C.-J., N. Grossman, and L. Leive. 1983. Salmo-nellae activate complement differentially via the alternate path-way depending on the structure of their lipopolysaccharide0-antigen. J. Immunol. 130:1867-1870.

27. Lin, F., H. S. Hsu, V. R. Mumaw, and I. Nakoneczna. 1989.Intracellular destruction of salmonellae in genetically resistantmice. J. Med. Microbiol. 30:79-89.

28. Lin, F., X. Wang, H. S. Hsu, V. R. Mumaw, and I. Nakoneczna.1987. Electron microscopic studies on the location of bacterialproliferation in the liver in murine salmonellosis. Br. J. Exp.Pathol. 68:539-550.

29. Lissner, C. R., R. N. Swanson, and A. D. O'Brien. 1983. Geneticcontrol of innate resistance of mice to Salmonella typhimurium:expression of the Ity gene in peritoneal and splenic macrophagesisolated in vitro. J. Immunol. 131:3006-3013.

30. Mackaness, G. B., R. V. Blanden, and F. M. Collins. 1966.Host-parasite relations in mouse typhoid. J. Exp. Med. 124:573-583.

31. Moon, R. J., and R. D. Leunk. 1983. Pilus mediated clearance ofSalmonella typhimunum by the perfused mouse liver, p. 251-259. In T. K. Eisenstein, P. Actor, and H. Friedman (ed.), Hostdefenses to intracellular pathogens. Plenum Press, New York.

32. Nnalue, N. A. 1990. Mice vaccinated with a non-virulent,aromatic-dependent mutant of Salmonella choleraesuis die fromchallenge with its virulent parent but survive challenge withSalmonella typhimurium. J. Med. Microbiol. 31:225-233.

33. Nnalue, N. A. Relevance of inoculation route to virulence ofthree Salmonella spp. strains in mice. Microb. Pathog., in press.

34. Nnalue, N. A., and B. A. D. Stocker. 1987. Tests of the virulenceand live-vaccine efficacy of auxotrophic and galE derivatives ofSalmonella choleraesuis. Infect. Immun. 55:955-962.

35. Nnalue, N. A., and B. A. D. Stocker. 1989. Vaccination route,infectivity and thioglycollate broth administration: effects onlive-vaccine efficacy of auxotrophic derivatives of Salmonellacholeraesuis. Microb. Pathog. 7:299-310.

36. O'Callaghan, D., D. Maskell, F. Y. Liew, C. S. F. Easmon, andG. Dougan. 1988. Characterization of aromatic- and purine-dependent Salmonella typhimurium: attenuation, persistence,and ability to induce protective immunity in BALB/c mice.Infect. Immun. 56:419-423.

37. Ornellas, E. P., and B. A. D. Stocker. 1974. Relation of lipo-polysaccharide character to P1 sensitivity in Salmonella typhi-murium. Virology 60:491-502.

38. Rabinovitch, M. 1967. Attachment of modified erythrocytes tophagocytic cells in the absence of serum. Proc. Soc. Exp. Biol.Med. 124:396-399.

39. Robertsson, J. A., C. Fossum, S. B. Svenson, and A. A. Lind-berg. 1982. Salmonella typhimurium infection in calves: specificimmune reactivity against 0-antigenic polysaccharide detect-able in in vitro assays. Infect. Immun. 37:728-737.

40. Saxen, H., I. Reima, and P. H. Makela. 1987. Alternativecomplement pathway activation by Salmonella 0 polysaccha-ride as a virulence determinant in the mouse. Microb. Pathog.2:15-28.

41. Seglen, P. 0. 1976. Preparation of isolated rat liver cells, p.29-85. In D. M. Prescott (ed.), Methods in cell biology, vol. 13.Academic Press, Inc., New York.

42. Shnyra, A., A. Bocharov, N. Bocharov, and V. Spirov. 1990.Large-scale production and cultivation of hepatocytes on Bio-silon microcarriers. Artif. Organs 14:421-428.

43. Smedsr0d, B., and H. Pertroft. 1985. Preparation of pure hepa-tocytes and reticuloendothelial cells in high yield from a singlerat liver by means of Percoll centrifugation and selective adher-ence. J. Leukocyte Biol. 38:213-230.

44. Svenson, S. B., and A. A. Lindberg. 1981. Artificial Salmonellavaccines: Salmonella typhimurium 0-antigen-specific oligosac-charide-protein conjugates elicit protective antibodies in rabbitsand mice. Infect. Immun. 32:490-496.

45. Takeuchi, A. 1967. Electron microscope studies of experimentalsalmonella infection. I. Penetration into the intestinal epitheliumby Salmonella typhimurium. Am. J. Pathol. 50:109-136.

46. Takeuchi, A. 1971. Penetration of the intestinal epithelium byvarious micro-organisms. Curr. Top. Pathol. 54:1-27.

47. Takeuchi, A., and H. Spinz. 1967. Electron microscope studiesof experimental salmonella infection in the pre-conditionedguinea pig. II. Response of the intestinal mucosa to the invasionby Salmonella typhimurium. Am. J. Pathol. 51:137-161.

48. van Dissel, J. T., J. J. M. Stikkelbroeck, B. C. Michel, P. C.Leijh, and R. van Furth. 1987. Salmonella typhimunum-specificdifference of intracellular killing by resident peritoneal macro-phages from Salmonella-resistant CBA and Salmonella-suscep-tible C57BL/10 mice. J. Immunol. 138:4428-4434.

49. van Dissel, J. T., J. J. M. Stikkelbroeck, W. Sluiter, P. C. J.Leijh, and R. van Furth. 1986. Differences in initial rate ofintracellular killing of Salmonella typhimunum by granulocytesof Salmonella-susceptible C57BL/10 mice and Salmonella-resis-tant CBA mice. J. Immunol. 136:1074-1080.

50. Vladoianu, I.-R., H. R. Chang, and J.-C. Perchere. 1990.Expression of host resistance to Salmonella typhi and Salmo-nella typhimurium: bacterial survival within macrophages ofmurine and human origin. Microb. Pathog. 8:83-90.

51. Wahdan, M. H., C. Serie, Y. Cerisier, S. Sallam, and R.Germanier. 1982. A controlled field trial of live Salmonella typhistrain Ty2la oral vaccine against typhoid: three year results. J.Infect. Dis. 145:292-295.

52. Wanson, J. C., D. Bernaert, and C. May. 1979. Morphology andfunctional properties of isolated and cultured hepatocytes, p.17-38. In H. Popper and F. Schaffner (ed.), Progress in liverdiseases, vol. 6. Grune and Stratton, New York.

53. Widmann, J. J., R. S. Cotran, and H. D. Fahimi. 1972. Mono-nuclear phagocytes (Kupffer cells) and endothelial cells. Iden-tification of two functional cell types in rat liver sinusoids byendogenous peroxidase activity. J. Cell Biol. 52:159-170.

INFECT. IMMUN.

on July 7, 2020 by guesthttp://iai.asm

.org/D

ownloaded from