Serosal But Not Mucosal Endotoxin Exposure Increases IntestinalPermeability in Vitro in the Rat

N. E. OSMAN, B. WESTRO¨ M & B. KARLSSONDept. of Animal Physiology, Lund University, Lund, Sweden

Osman NE, Westro¨m B, Karlsson B. Serosal but not mucosal endotoxin exposure increases intestinalpermeability in vitro in the rat. Scand J Gastroenterol 1998;33:1170–1174.

Background:Microbial endotoxins are normally present in the gut, usually without apparent harmfuleffects, whereas systemically administered endotoxin impairs the mucosal barrier function. Our aim was toinvestigate whether in vitro exposure to bacterial lipopolysaccharide (LPS) could affect the intestinalbarrier properties of the rat small intestine.Methods:Small-intestinal segments from rats were mounted inUssing diffusion chambers, and the mucosal to serosal permeation of the marker molecules bovine serumalbumin (BSA) and51Cr-ethylenediaminetetraacetic acid (EDTA) was measured after addition of LPS tothe mucosal or serosal side. Results:Mucosal exposure to LPS (0.01, 0.05, 0.25 mg/ml) had no effects onthe permeation of BSA and51Cr-EDTA, whereas when added to the serosal side at 0.05 or 0.25 mg/ml,LPS increased the marker permeation.Conclusion:Serosal LPS exposure in vitro increased the intestinalpermeability to the different-sized markers, whereas mucosal LPS did not, indicating that the mechanismsleading to intestinal barrier impairment can be initiated in the intestinal wall itself.

Key words:Bacterial translocation; bovine serum albumin; endotoxin;51Cr-ethylenediaminetetraaceticacid; mucosal barrier; small intestine

Borje Karlsson, Ph.D., Dept. of Animal Physiology, Lund University, Helgonava¨gen 3B, S-223 62 Lund,Sweden (fax:�46 46 222 4539)

The gastrointestinal mucosa, in parallel to its absorptivefunctions, constitutes a defense barrier that prevents noxiousagents contained in the gut lumen, such as bacteria andbacterial products, and antigenic macromolecules fromescaping and spreading into extraintestinal tissues and organs(1–3). However, small amounts of macromolecules maypermeate the gut mucosa in healthy experimental animals andhumans, and even endogenous gut bacteria can pass throughthis barrier to infect mesenteric lymph nodes and systemicorgans, a process known as bacterial translocation (3).

Gram-negative bacteria in the gut lumen are known toconstantly shed their outer membrane fragments, and en-dotoxins are therefore produced in large amounts, surpris-ingly without obvious harmful effects (4). An increasedleakage of bacterial endotoxins from the intestinal lumen hasbeen found to occur in conjunction with a wide variety ofclinical conditions involving intestinal trauma and inflamma-tion (1, 5). Moreover, intravenous administration of endotox-in results in increased intestinal molecular permeability andbacterial translocation in humans (6), pigs (7), and rats (8).The tight junctions of the intestinal epithelium might be a sitefor leakage of microbial agents, as found by ultrastructuralanalysis of mouse ileal epithelium after parenteral endotoxinadministration (9). Bacterial endotoxins increase gut perme-ability (6, 9), alter host immune defenses (10), and arerelatively common in patients at increased risk of developingenteric infections (2).

The aim of the present study was to experimentallyinvestigate whether mucosal (luminal) or serosal (blood-side) exposure to the bacterial endotoxin lipopolysaccharide(LPS) could affect the intestinal barrier properties. This wasdone by measuring the mucosal to serosal permeability of thetwo marker molecules51Cr-ethylenediaminetetraacetic acid(EDTA) and bovine serum albumin (BSA) in the rat smallintestine, using the Ussing diffusion chamber model.

MATERIALS AND METHODS

AnimalsMale rats (Rattus norvegicus; Sprague–Dawley strain,

Møllegaard, Skensved, Denmark), weighing 300–350 g,were kept on chopped wood bedding in polycarbonate cagesunder a 12-h day–night rhythm at 20� 2°C and a relativehumidity of 50%� 10%. The rats had free access to rat chow(Altromin 1324, Altromin International, Lage, Germany) andtap water. The study was approved by the Lund UniversityEthical Review Committee on Animal Experiments.

Experimental protocolStarting at 0900–1000 h, a laparotomy was done under

ether anesthesia, and a 50-cm-long segment of the smallintestine, ending 5 cm proximal to the ileocecal connection,was removed. The segment was further divided into 4-cm-long pieces and immediately immersed in a modified Krebs–

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y SU

NY

Sta

te U

nive

rsity

of

New

Yor

k at

Sto

ny B

rook

on

10/3

1/14

For

pers

onal

use

onl

y.

Ringer buffer, pH 7.4 (110.0 mM NaCl2, 3.0 mM CaCl2,5.5 mM KCl, 1.4 mM KH2PO4, 29.0 mM NaHCO3, 5.7 mMNa-pyruvate, 7.0 mM Na-fumarate, 5.7 mM Na-glutamate,and 13.4 mM glucose), and oxygenated with 95% O2 and 5%CO2 at room temperature. Each piece was cut along themesenteric border and mounted as a sheet, with an exposedarea of 1.78 cm2, between the mucosal and serosal reservoirsof Ussing diffusion chambers (Precision Instrument Design,Los Altos, Calif., USA), modified in accordance with Grass &Sweetana (11). The serosal and mucosal reservoirs were filledwith 5 ml Krebs–Ringer buffer that was continuouslyoxygenated and circulated by means of a gas lift at atemperature of 37°C throughout the incubations. The experi-ments started (t = 0) within 40 min after the induction ofanesthesia, when the buffer in the mucosal reservoirs wasreplaced with 5 ml buffer containing the different-sizedmolecular markers BSA (25 mg/ml, A-4503, Sigma, Mw67,000) and51Cr-labeled EDTA (3mCi/ml, DuPont, BadHomburg, Germany, Mw 358), and the buffer in the serosalreservoirs was replaced with 5 ml fresh buffer. At the sametime LPS at different concentrations (0, 0.01, 0.05, and0.25 mg/ml, L-2630, Sigma, St. Louis, Mo., USA) was addedto either the serosal (n = 6 rats) or the mucosal reservoirs(n = 6 rats), to study its effects on the mucosal to serosalmarker permeability in the small intestine. Every 20 min for2 h, 1-ml samples were taken from the serosal reservoirs formarker analysis and replaced with 1 ml fresh buffer. Theviability of the intestinal segments throughout the experimentwas ascertained as previously reported (12).

Marker molecule analysis51Cr-EDTA radioactivity in 1-ml samples was measured

for 120 sec in a well-type gamma counter (LKB, Bromma,Sweden).

BSA concentration was measured with a two-step sand-wich enzyme-linked immunosorbent assay (ELISA) (13). Inshort, a 96-microwell ELISA plate (Maxisorp, Nunc, Ros-kilde, Denmark) was coated with chicken anti-BSA (65–437;ICN Biomedicals Inc., Costa Mesa, Calif., USA) diluted1:500 in 0.1 M NaHCO3, pH 8.3 (200ml/well), and incubatedovernight at�4°C. In another 96-microwell incubation plate(Nunc) 50ml of the samples or standards—that is, 4–1000 ng/ml of BSA (A-7638, Sigma)—were added to each welltogether with 100ml of specific rabbit anti-BSA antiserum(Dakopatts A/S, Glostrup, Denmark) diluted 1:10,000 in 0.01M phosphate-buffered saline (PBS), pH 7.2, plus 0.05%Tween 20, and incubated overnight at�4°C. After washingof the ELISA plate, 3� 5 min, with 0.05% Tween in 0.9%NaCl, 50ml PBS, and 150ml mixture from the incubationplate were added to respective wells of the ELISA plate andincubated for 1 h at room temperature. After being washed,alkaline phosphatase-conjugated swine anti-rabbit immuno-globulins (D306, Dakopatts) diluted 1:1000 in PBS–Tweenwere added (200ml/well) and incubated for 1 h at roomtemperature. Finally, substrate, 0.8 mg/ml ofp-nitrophenyl

phosphate (Sigma) in 0.1 M glycine buffer, pH 10.4, wasadded (200ml/well), and the color reaction followed at450 nm (iEMS reader MF, Labsystems, Helsinki, Finland).

Calculations and statisticsThe apparent intestinal permeability coefficients (Papp) of

the marker molecules in the intestinal epithelia werecalculated from the equation; Papp (cm/sec� 10–6) = dc/dt� V/(A � C0), where dc/dt is the change of the serosalconcentration during 60–120 min (mol/l/sec), V is the volumein the reservoirs (cm3), C0 is the initial marker concentrationin the mucosal reservoirs (mol/l), and A is the exposedintestinal area in the chamber (cm2) (11).

Data are presented as mean� standard deviation (s).ANOVA, followed by the Student–Newman–Keuls test,was used for statistical evaluation.

RESULTS

Generally, the permeability of the low-molecular marker51Cr-EDTA was higher than that of the high-molecularmarker BSA (Table I). Addition of LPS in increasingconcentrations to the mucosal or serosal side in Ussingdiffusion chambers affected the mucosal to serosal passageof both marker molecules in the small intestine in differentways.

Mucosal exposure to LPS at concentrations of 0.01, 0.05,and 0.25 mg/ml did not have any effects, compared with thecontrol, when the cumulative passage of the markers wasstudied during 2 h (Figs. 1A, 2A) or when the marker passageswere expressed as the apparent permeability coefficients(Papp) obtained between 60 and 120 min (Table I). However,when LPS was added to the serosal side, the permeabilityincreased significantly for both marker molecules, at con-centrations of 0.05 and 0.25 mg/ml, whereas at a lower

Table I. The apparent permeability coefficient, Papp (cm/s� 10ÿ6),for the marker molecules51Cr-labeled ethylenediaminetetraaceticacid (EDTA) and bovine serum albumin (BSA) obtained afterexposure to increasing concentrations of lipopolysaccharide (LPS)(0.01, 0.05 and 0.25 mg/ml) at the mucosal or serosal side of thesmall intestine in Ussing diffusion chambers

Groups 51Cr-EDTA BSA

Control 4.8� 1.5 (n = 32) 0.011� 0.006 (n = 28)Mucosal LPS

0.01 mg/ml 3.8� 1.4 (n = 15) 0.008� 0.008 (n = 15)0.05 mg/ml 4.8� 1.4 (n = 14) 0.020� 0.021 (n = 13)0.25 mg/ml 5.3� 1.2 (n = 14) 0.019� 0.012 (n = 8)

Serosal LPS0.01 mg/ml 5.4� 1.5 (n = 18) 0.018� 0.010 (n = 17)0.05 mg/ml 6.1� 1.4 * (n = 18) 0.023� 0.009 * (n = 18)0.25 mg/ml 6.0� 1.9 * (n = 18) 0.034� 0.032 * (n = 17)

Values are given as mean� standard deviation;n = number oftissues.

* Statistically significant difference (P< 0.05) compared withcontrols.

LPS Effects in Vitro on Intestinal Permeability 1171

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y SU

NY

Sta

te U

nive

rsity

of

New

Yor

k at

Sto

ny B

rook

on

10/3

1/14

For

pers

onal

use

onl

y.

concentration (0.01 mg/ml) no effect was found (Table I,Figs. 1B, 2B).

DISCUSSION

The results of the present in vitro study using isolated small-intestinal segments from the rat in Ussing diffusion chambersshow that the bacterial endotoxin LPS, when administered atthe luminal side of the intestinal mucosa, had no effect on theintestinal permeability of the different-sized marker mole-cules51Cr-EDTA and BSA, whereas addition of LPS to theserosal side resulted in increased permeability. The firstfinding suggests that the intestinal barrier function wasmaintained after luminal LPS exposure. Our results are

supported by other studies that have shown that bacterialendotoxins administered to the luminal side of the intestinalmucosa do not give any morphologic or functional effects onisolated intestinal segments or on cultured epithelial cells(4, 14, 15). Even in vivo, oral administration of large amountsof LPS did not induce adverse systemic reactions in man(4). Moreover, the transmucosal passage of LPS is lowor negligible, as shown in the everted gut sac model andin intact intestine in vivo (16). The significant protectivecapacity of normal intestinal mucosa to LPS exposureconsists of specific secretory IgA antibodies in the intestinalmucus (17), as IgA may coat most of the aerobic gram-negative bacteria in the intestine (18) containing LPS. Inaddition, the mucus layer protects the mucosal surface againstendotoxin attachment and penetration (19, 20). A primaryassault, like acute damage or inflammation, might benecessary to break the intestinal barrier properties andenhance the transmucosal transport of bacterial endotoxins.

In contrast, blood-side exposure of endotoxin in vivo afterdeposition of bacteria or their endotoxins in the peritoneal

Fig. 1. Mucosal to serosal passage (mean� standard deviation) of51Cr-labeled ethylenediaminetetraacetic acid (EDTA) across thesmall intestine during mucosal (A) or serosal (B) exposure tolipopolysaccharide (LPS) (0.01, 0.05, 0.25 mg/ml) for 2 h in Ussingdiffusion chambers. * Statistically significant difference (P< 0.05)compared with control.

Fig. 2. Mucosal to serosal passage (mean� standard deviation) ofbovine serum albumin (BSA) across the small intestine duringmucosal (A) or serosal (B) exposure to lipopolysaccharide (LPS)(0.01, 0.05, 0.25 mg/ml) for 2 h in Ussing diffusion chambers.* Statistically significant difference (P< 0.05) compared withcontrol.

1172 N. E. Osman et al.

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y SU

NY

Sta

te U

nive

rsity

of

New

Yor

k at

Sto

ny B

rook

on

10/3

1/14

For

pers

onal

use

onl

y.

cavity or after intravenous injection (6–8) results in an injuredintestinal epithelium and an impaired barrier function. In ourstudies, serosal exposure to LPS in Ussing diffusion chambersresulted in increased intestinal passage of both small and largemarker molecules. This in vitro model thus shows similarresults as in vivo, although in vitro there is less influence fromthe luminal contents and no systemic influence. Thesefindings indicate that the mechanisms leading to increasedmucosal permeability and decreased intestinal barrier proper-ties are local acute effects that can be initiated in the intestinalwall itself.

The mechanisms of LPS action on the serosal side may beby macrophage activation after interaction of the endotoxinwith an LPS-binding protein, which then binds specifically tothe CD14 receptor (21). If this macrophage pathway isimportant, it may involve release of cytokines, such as tumornecrosis factor-a and interleukin-1 (10). Recent results (22)indicate that LPS may also activate the epithelial cells directlyby binding to a receptor after forming a complex with solubleCD14. In vivo challenge with endotoxin has been shown todisrupt the tight or occluding junctions between intestinalepithelial cells (9). Moreover, it was recently reported thatadministration of LPS to the basolateral surface of IEC-6intestinal cells disrupted the tight junctions, whereas LPSadministration to the apical surface did not alter the barrierfunction (15). Our results, with an increased51Cr-EDTApermeability, are in line with these observations, since thisinert low-molecular marker has been widely used to testparacellular passage (23). The concomitantly increasedpermeability for the macromolecular marker BSA mightalso be indicative of an increased paracellular leakage,although under normal conditions this protein marker mainlyuses a transcytotic route for epithelial passage in the gut (1).In an attempt to more closely investigate intestinal regionaldifferences in sensitivity to LPS, some preliminary experi-ments showed that the proximal small intestine appearedmore sensitive than the distal small intestine and colon. Thisobservation is in line with a study showing that administrationof nonlethal doses of endotoxin to adult rats caused moresevere alterations of the intestinal mucosa in jejunum than incolon (24).

The observation in the present study that LPS affects theintestinal barrier function only when it interacts with theserosal side could result in a wide variety of pathologicconditions in vivo once endogenous bacterial LPS has passedthe intestinal epithelium. An increased molecular permeation,even though there are no straight connections, might berelated to an increased bacterial translocation, which mayresult in sepsis, septic shock, and multiple system organfailure (5).

ACKNOWLEDGEMENTS

Our sincere thanks are due to Mrs Inger Mattsson for herexpert technical assistance. Financial support was kindly

given by Director Albert Pa˚hlsson’s Research Foundation,Malmo, Sweden; the Royal Physiographic Society, Lund,Sweden; and the Foundation of Hierta-Retzius, RoyalSwedish Academy of Sciences, Stockholm, Sweden.

REFERENCES

1. Sanderson IR, Walker WA. Uptake and transport of macro-molecules by the intestine: possible role in clinical disorders (anupdate). Gastroenterology 1993;104:622–39.

2. Deitch EA, Berg R, Specian R. Endotoxin promotes thetranslocation of bacteria from the gut. Arch Surg 1987;122:185–90.

3. Berg RD, Garlington AW. Translocation of certain indigenousbacteria from the gastrointestinal tract to the mesenteric lymphnodes and other organs in a gnotobiotic mouse model. InfectImmun 1979;23:403–11.

4. van Deventer SJH, ten Cate JW, Tytgat GNJ. Intestinalendotoxemia: clinical significancy. Gastroenterology 1988;94:825–31.

5. Deitch EA. Bacterial translocation of gut flora. Proceedings ofNIH Conference on advances in understanding trauma and burninjury. J Trauma 1990;30:S184–9.

6. O’Dwyer ST, Mitchie HR, Ziegler TR, Revhaug A, Smith RJ,Wilmore DW. A single dose of endotoxin increases intestinalpermeability in healthy humans. Arch Surg 1988;123:1459–64.

7. Fink MP, Antonsson JB, Wang H, Rothschild HR. Increasedintestinal permeability in endotoxic pigs. Arch Surg 1991;126:211–8.

8. Herlbert DJ, Zhong R, Wang P. Intestinal permeability withhemorrhagic shock, surgical trauma, and endotoxemia. SurgForum 1989;40:93–5.

9. Walker RI, Porvaznik MJ. Disruption of the permeability barrier(zona occludens) between intestinal epithelial cells by lethaldoses of endotoxin. Infect Immun 1978;21:655–8.

10. Ogle CK, Mao JX, Wu JZ, Ogle JD, Alexander JW. Theproduction of tumor necrosis factor, interleukin-1, interleukin-6,and prostaglandin-E2 by isolated enterocytes and gut macro-phages: effect of lipopolysaccharide and thermal injury. J BurnCare Rehabil 1994;15:470–7.

11. Grass GM, Sweetana SA. In vitro measurements of gastro-intestinal tissue permeability using a new diffusion cell.Pharmaceut Res 1988;6:372–6.

12. Pantzar N, Westro¨m BR, Luts A, Lundin S. Regional small-intestinal permeability in vitro to different-sized dextrans andproteins in the rat. Scand J Gastroenterol 1993;28:205–11.

13. Kimm MH, Curtis GH, Hardin JA, Gall DG. Transport of bovineserum albumin across rat jejunum: role of the enteric nervoussystem. Am J Physiol 1994;266:G186–93.

14. Wells CL, Jechorek RP, Olmsted SB, Erlandsen SL. Effects ofLPS on epithelial integrity and bacterial uptake in the polarizedhuman enterocyte-like cell CaCo-2. Circ Shock 1993;40:276–88.

15. Kimura H, Sawada N, Tobioka H, Isomura H, Kokai Y, HirataK, et al. Bacterial lipopolysaccharide reduced intestinal barrierfunction and altered localization of 7H6 antigen in IEC-6 ratintestinal crypt cells. J Cell Physiol 1997;171:284–90.

16. Nolan JP, Hare DK, McDevitt JJ, Ali MV. In vitro studies ofintestinal endotoxin absorption. I. Kinetics of absorption in theisolated everted gut sac. Gastroenterology 1977;72:434–9.

17. Stokes CR, Soothil JF, Turner MW. Immune exclusion is afunction of IgA. Nature 1975;255:745–6.

18. Van Saene HKF, Van Der Waaij D. A novel technique fordetecting IgA coated potentially pathogenic microorganisms inthe human intestine. J Immunol Meth 1982;30:87–96.

19. Forstner JF, Forstner GG. Gastrointestinal mucus. In: JohnsonLR, editors. Physiology of the gastrointestinal tract. New York:Raven Press; 1994;Chapter 33:1255–83.

20. Lake AM, Bloch KJ, Walker WA. Anaphylactic release ofintestinal goblet cell mucus. Immunology 1980;39:173–8.

LPS Effects in Vitro on Intestinal Permeability 1173

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y SU

NY

Sta

te U

nive

rsity

of

New

Yor

k at

Sto

ny B

rook

on

10/3

1/14

For

pers

onal

use

onl

y.

21. Wright SD, Ramos RA, Tobias PS, Ulevitch RJ, Mathison JC.CD14, a receptor for complexes of lipopolysaccharide (LPS)and LPS binding protein. Science 1990;249:1431–3.

22. Pugin J, Schurer-Maly CC, Leturcq D, Moriarty A, Ulevitch RJ,Tobias PS. Lipopolysaccharide activation of human endothelialand epithelial cells is mediated by lipopolysaccharide-bindingprotein and soluble CD14. Proc Natl Acad Sci USA 1993;90:2744–8.

23. Bjarnason I, MacPherson A, Hollander D. Intestinal perme-ability. An overview. Gastroenterology 1995;108:1566–81.

24. Cianchi G, De Gaudio AR, Zecchi Orlandini S, Martini M,Evangelista S. Endotoxin-mediated intestinal damage: protec-tive effect of capsaicin in the rat. Scand J Gastroenterol 1998;33:736–42.

Received 16 June 1998Accepted 9 September 1998

1174 N. E. Osman et al.

Scan

d J

Gas

troe

nter

ol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y SU

NY

Sta

te U

nive

rsity

of

New

Yor

k at

Sto

ny B

rook

on

10/3

1/14

For

pers

onal

use

onl

y.