Carrageenan Primes Leukocytes To Enhance Lipopolysaccharide-Induced Tumor Necrosis Factor Alpha Production

Abstract

We have previously reported that pretreatment with carrageenan (CAR) enhances lipopolysaccharide (LPS)-induced tumor necrosis factor alpha (TNF-α) production in and lethality for mice. Whole blood cultured in vitro was used to show that CAR pretreatment results in about a 200-fold increase in LPS-induced TNF-α production. CAR by itself did not induce TNF-α production. However, CAR-treated cultured medium sensitized whole blood to make more LPS-induced TNF than did saline-treated cultured medium in vitro. It was also demonstrated that CAR pretreatment increases TNF-α mRNA levels of both blood cells and peritoneal exudate cells, but not of bone marrow cells. Immunoelectron microscopic analysis revealed that polymorphonuclear leukocytes and macrophages are TNF-α-producing cells in CAR-treated mice. In CAR-treated mice, TNF-α was seen early after LPS injection in leukocytes in hepatic sinusoids and on the surfaces of endothelial cells. TNF-α was also detected late after LPS injection in hepatocytes which become edematous. These results suggest that CAR primes leukocytes to produce TNF-α in response to LPS and that they play an important role in the pathogenesis of liver injury.

Tumor necrosis factor alpha (TNF-α) has a variety of biological activities which affect a number of cells, such as inhibition of cellular growth, production of cytokines, induction of shock, and so on (2, 4, 33). In general, TNF-α is produced in macrophages by stimulation of lipopolysaccharide (LPS) (19). It has been demonstrated that Mycobacterium bovis BCG and Propionibacterium acnes (Corynebacterium parvum) increase the sensitivity of macrophages to LPS (31, 37), and this priming effect on macrophages appears at least 4 days after the administration of macrophage activators (37). d-Galactosamine and actinomycin D have been also used as endotoxin sensitizers (9–12, 32). Unlike BCG and C. parvum, treatment with d-galactosamine increases susceptibility of mice to the lethal effects of LPS several thousand-fold immediately (10, 11). UTP depletion by d-galactosamine is considered to be responsible for the development of sensitization to LPS (11). In addition, d-galactosamine does not affect LPS-induced cytokine gene expression in Kupffer cells (7).

On the other hand, carrageenan (CAR; a high-molecular-weight sulfated polygalactose isolated from marine plants) increases LPS-induced TNF-α production at least 2 h after it is injected (22). Since CAR is known to destroy macrophages (6, 27), it is likely that TNF-α is not produced in macrophages but in other cells. Polymorphonuclear leukocytes (PMNs) are widely accepted as key effector cells in both host defense and tissue destruction. Although PMNs are viewed as terminally differentiated cells which are devoid of RNA and protein synthesis, several lines of evidence have shown that PMNs release various cytokines, including TNF-α (5, 8, 18).

In the present study, we demonstrated that the pretreatment of mice with CAR increases TNF-α mRNA levels in leukocytes. Further, primed leukocytes including PMNs produce a large amount of TNF-α in response to LPS and play a major role in the pathogenesis of liver injury.

MATERIALS AND METHODS

Mice.

Male C3H/HeN mice were purchased from Seiwa Experimental Animal Co. (Oita, Japan). Mice were housed in groups of 10 and allowed food and water ad libitum. All experiments were done with 6- to 8-week-old mice. Protocols in this work were approved by the Institutional Animal Care Committee.

Reagents.

Phenol-extracted Escherichia coli LPS (O127:B8) was purchased from Difco Laboratories, Detroit, Mich. Iota-carrageenan (lot 59C-0328) was purchased from Sigma, St. Louis, Mo. Both LPS and CAR were dissolved in pyrogen-free physiological saline (Otsuka Pharmaceutical Co., Osaka, Japan). CAR solution was autoclaved at 121°C for 15 min before use, and LPS contamination was not detected by the Limulus amoebocyte lysate assay (Endospecy ES-6; Seikagakukougyou Co., Tokyo, Japan).

LPS-induced TNF production in whole blood ex vivo.

Mice were injected intraperitoneally (i.p.) with 5 mg of CAR in saline or with saline as a control. After 16 h, blood was drawn into a heparinized syringe and was immediately diluted with 5 volumes of endotoxin-free saline. The diluted blood (990 μl) was then placed in a 24-well plate (Becton Dickinson, Paramus, N.J.). After incubation at 37°C for 0 to 4 h in the presence of LPS (10 μl) at 100 ng/ml, the culture medium was centrifuged at 100 × g for 10 min to remove cell debris and then stored at −80°C for assaying TNF.

Effect of CAR-treated cultured medium on LPS-induced TNF production in vitro.

Blood was drawn into a heparinized syringe by cardiac puncture from the mice and immediately diluted with 5 volumes of endotoxin-free RPMI 1640. The diluted blood was then placed into a 24-well plate. Twenty microliters of saline or CAR (5 mg/ml) was added to each well and then incubated at 37°C for the determined time. After incubation, the culture medium was centrifuged at 100 × g for 10 min to remove cell debris. Another 6 ml of diluted blood was centrifuged and the supernatant was removed. Then cell pellets were resuspended in 6 ml of the cultured medium with saline or CAR. Each cell suspension was then placed into a 24-well plate at 980 μl and incubated at 37°C for 1 h. Then each well was stimulated with LPS (10 μl) at a final concentration of 100 ng/ml and incubated for 4 h. The cultured medium was centrifuged and stored at −80°C for assaying TNF.

Preparation of PECs.

Peritoneal exudate cells (PECs) were obtained 16 h after injecting mice i.p. with 5 mg of CAR in saline or with saline as a control as previously reported (22). PECs were resuspended and adjusted to a concentration of 2 × 10⁶ cells/ml with RPMI 1640 medium containing benzylpenicillin potassium (100 IU/ml) and streptomycin (100 μg/ml).

TNF assay.

TNF-α activity in the supernatants was determined by colorimetric measurement of the cytotoxicity of L929 cells as described previously (22). TNF-α activity was expressed in units per milliliter, which is the reciprocal of the dilution necessary for the lysis of 50% of the cells. One unit per milliliter is equivalent to 0.63 pg of recombinant murine TNF-α (Genzyme, Cambridge, Mass.) in our assay (21).

Northern blot analysis of total RNA.

Total cellular RNA was prepared with 4 M guanidine isothiocyanate as described by Ullrich et al. (34). Poly(A) RNA was isolated from bone marrow by using oligo(dT) latex (Takara Shuzo, Tokyo, Japan). About 10 μg of total RNA and 5 μg of poly(A) RNA were electrophoresed on a 1% agarose gel containing 16% formaldehyde and then transferred to a nylon membrane (Hybond-N; Amersham). Prehybridization of the filter was carried out at 43°C for 4 h in 6× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) containing 50% formamide, 0.5% sodium dodecyl sulfate, 0.2% Ficoll, and 0.2% polyvinylpyrrolidone. Hybridization was carried out in the same solution containing about 10⁵ cpm of ³²P-labeled mouse TNF-α cDNA per cm². After incubation at 43°C for 18 h, the filter was washed four times with 2× SSC–0.2% sodium dodecyl sulfate at 60°C for 30 min. The same filter was also hybridized with ³²P-labeled β-actin cDNA to normalize the amount of RNA used. The amount of TNF-α mRNA was determined with a BAS 2000 bioimage analyzer (Fujix, Tokyo, Japan).

Immunocytochemistry.

About 10⁶ whole-blood cells or PECs isolated as described above were plated in an eight-well Lab-Tek chamber slide (Nunc Inc., Naperville, Ill.). After incubation for 60 min in the presence of LPS (100 ng/ml), the cells were fixed by immersing the slide in phosphate-buffered saline containing 4% paraformaldehyde and 0.5% glutaraldehyde. An immune reaction was carried out to detect TNF-α as described previously (20). Briefly, the slide was incubated overnight at 4°C with a rabbit polyclonal antibody for mouse TNF-α (1:100 dilution; Genzyme Co.) or a rabbit immunoglobulin G (IgG) as a control and then incubated again for 1 h at 37°C with a biotinylated secondary anti-rabbit goat IgG (1:200 dilution). The immune complexes were detected with a streptavidin-peroxidase complex (1:600 dilution; Histofine SAB-PO kits). The slides were thoroughly washed in 0.05 M Tris-HCl, postfixed in 1% osmium tetroxide in the buffer for 1 h at 4°C, dehydrated in an ascending alcohol series, embedded in Quetol 812, and examined in a JEM 1200 EX electron microscope.

Mice were pretreated with CAR or saline as described above. At 0.5, 2, and 5 h after LPS injection intravenously, mice were perfused with saline and then with phosphate-buffered saline containing 4% paraformaldehyde and 0.5% glutaraldehyde. Liver samples were taken and cut into approximately 2-mm blocks, and each block was fixed in the same solution for 1 h. Sections approximately 30 to 40 μm in thickness were prepared and inserted into a sample mesh pack. After blocking endogenous peroxidase activity in periodic acid solution for 45 s, each section was used for the detection of TNF-α as described above.

RESULTS

TNF-α production in whole blood ex vivo.

We have previously demonstrated that LPS-induced TNF-α production is enhanced in the serum of CAR-pretreated mice (22). To understand the mechanism of the effect of CAR, we first analyzed LPS-induced TNF-α activities in whole blood cultured ex vivo. This allowed us to see the effects of CAR and LPS separately. As shown in Fig. 1, whole blood cells produced TNF-α in the presence of LPS. The extent of the induction was about 200 times higher in whole blood from CAR-treated mice than in blood from control mice. When cultured in LPS-free medium, no TNF-α was detected in either CAR-pretreated or saline-pretreated (control) blood. This result indicates that CAR by itself does not induce TNF-α production; rather, it activates the blood cells to be more sensitive to LPS. Figure 2 shows the time course of TNF production induced by LPS stimulation in CAR-pretreated or control blood. We could not find any TNF in CAR-pretreated blood immediately after LPS stimulation. However, CAR-pretreated blood started to produce TNF after 1 h of stimulation with LPS.

FIG. 1.

LPS-induced TNF-α production in whole blood. Whole blood from control or CAR-pretreated mice was stimulated in vitro with saline or LPS at 100 ng/ml for 4 h.

FIG. 2.

Time course of LPS-induced TNF-α production in whole blood. Whole blood from control or CAR-pretreated mice was stimulated in vitro with LPS (100 ng/ml) for 0 to 4 h.

Effect of cultured medium of CAR on LPS-induced TNF production in whole blood cells in vitro.

According to our ex vivo data, CAR primes whole blood to produce TNF. To understand the mechanism by which CAR enhances LPS-induced TNF production, we investigated whether the culture medium containing CAR enhances LPS-induced TNF production in vitro. As shown in Fig. 3, neither CAR-treated medium nor saline-treated medium induced TNF production without LPS. However, cultured medium containing CAR enhanced LPS-induced TNF production, in contrast to saline-treated medium. This data suggested that CAR produced some factors that activated blood cells for TNF production during the 4-h incubation. It has been reported that gamma interferon has a priming effect that enhances LPS-induced TNF production. However, we could not detect gamma interferon in the medium with an enzyme-linked immunoassay (data not shown).

FIG. 3.

Effect of cultured medium of CAR on LPS-induced TNF production. Whole blood with saline or CAR (final concentration of 100 μg/ml) was incubated for 4 h. CAR0hr represents TNF production in the supernatants of medium in whole blood immediately centrifuged after CAR treatment. Each medium was transferred to another aliquot of whole blood and incubated at 37°C for 1 h. Then, each well was stimulated with LPS (LPS+) or saline (LPS−) at a final concentration of 1 μg/ml and incubated for 4 h.

Effect of CAR on TNF-α mRNA.

To investigate how CAR increases the sensitivity of whole blood to LPS, we carried out Northern blot hybridization to analyze levels of TNF-α mRNA. Figure 4A shows that CAR pretreatment increased TNF-α mRNA levels about 2- and 15-fold in blood and PECs, respectively. In contrast, no increase of TNF-α mRNA was observed in bone marrow (Fig. 4A, compare lanes 7 and 8). The basal levels of TNF-α mRNA expression in the control blood were about three times higher than in the bone marrow. This might have resulted from an activation of whole-blood leukocytes by a low concentration of circulating cytokines. Interestingly, CAR pretreatment increased TNF mRNA about threefold over LPS treatment. In addition, LPS stimulation increased the expression of TNF mRNA level about twofold in both CAR-pretreated and control whole blood (Fig. 4B). This data demonstrates that CAR increases expression of TNF-α mRNA in whole-blood leukocytes severalfold and that CAR is a strong primer of TNF.

FIG. 4.

Expression of TNF-α mRNA. (A) Northern hybridization of total RNA (lanes 1 to 6) and poly(A) RNA (lanes 7 and 8) was carried out with a TNF-α cDNA probe. The same blot was also hybridized with a β-actin cDNA probe. Lanes represent RNA from bone marrow (lanes 1, 2, 7, and 8), whole blood (lanes 3 and 4), and PECs (lanes 5 and 6). Odd-numbered lanes, control mice; even-numbered lanes, CAR-pretreated mice. (B) Northern hybridization of total RNA was carried out with a TNF-α cDNA probe. The same blot was also hybridized with a β-actin cDNA probe. Lanes contain RNA from whole blood without LPS (lanes 1 and 2) and with LPS (100 ng/ml) for 1 h (lanes 3 and 4). Odd-numbered lanes, control mice; even-numbered lanes, CAR-pretreated mice.

TNF-α production in PMNs of CAR-pretreated mice.

To identify the TNF-α-producing cells in blood, we carried out an immunoelectron microscopic analysis with an anti-TNF-α antibody (Fig. 5). Consistent with the results shown above, no TNF-α was seen in any blood cells in the absence of LPS (Fig. 5a). In contrast, large amounts of TNF-α were detected in granules in PMNs of the CAR-pretreated mouse blood. This TNF-α was induced by low doses of LPS, which did not induce TNF in the blood of the control mice (Fig. 5b and c). Further, destroyed macrophages were seen in the blood of the pretreated mice (Fig. 5d). We could detect only one destroyed macrophage that produced TNF-α from stimulation by LPS in our preparations (Fig. 5e). These observations strongly suggest that CAR primes leukocytes, which mainly produce LPS-induced TNF-α.

FIG. 5.

TNF-α production in PMNs of CAR-pretreated mice. Electron micrographs show a leukocyte of a CAR-pretreated mouse (a), a control leukocyte stimulated with 100 ng of LPS per ml for 1 h (b), a leukocyte of a CAR-pretreated mouse stimulated similarly with LPS (c), a macrophage of a CAR-pretreated mouse (d), and a destroyed macrophage of a CAR-pretreated mouse stimulated with 100 ng of LPS per ml for 1 h (e). Bars, 500 nm (a to c) and 1 μm (d and e).

TNF-α in liver injury.

As reported previously, CAR in combination with LPS, but not by itself, induces liver injury (15). Given the results described above, this might be the consequence of increased levels of TNF-α. To determine the relationship between TNF-α level and liver injury, the kinetics of the appearance of TNF-α in liver was investigated. Neither CAR nor LPS alone induced detectable TNF-α or a significant histological change (Fig. 6a and b). However, 30 min after LPS injections in CAR-pretreated mice, PMNs were seen in the hepatic sinusoids, and TNF-α was localized in the granules of these cells, as was the case for blood PMNs (Fig. 2 and 6c). After 2 h, TNF-α was found both on the apical surfaces and in the lysosomes of sinusoidal endothelial cells (Fig. 6d). In addition to these areas, TNF-α was also detected in lysosomes of the hepatocytes (Fig. 6a, b, and e). After 5 h, the hepatocytes became edematous and TNF-α was still found in the lysosomes (Fig. 6f). Thus, accumulation of TNF-α in hepatocytes appears to associate closely with LPS-induced liver injury in CAR-pretreated mice.

FIG. 6.

Localization of TNF-α in liver. Electron micrographs show a hepatocyte from a control mouse at 5 h after LPS (50 μg) injection (a), a hepatocyte from a CAR-pretreated mouse without LPS injection (b), a leukocyte in the hepatic sinusoid of a CAR-pretreated mouse at 30 min after LPS injection (c), hepatic sinusoid endothelium of a CAR-pretreated mouse at 2 h after LPS injection (d), a hepatocyte of a CAR-pretreated mouse at 2 h after LPS injection (e), and a hepatocyte of a CAR-pretreated mouse at 5 h after LPS injection (f). Bars, 200 nm (a and b), 500 nm (d to f), and 1 μm (c).

DISCUSSION

In the present study, we demonstrated that PMNs produced TNF-α in response to LPS in mice injected with CAR. CAR pretreatment alone resulted in an increase in the levels of TNF-α mRNA but not of the secreted protein.

Since TNF expression in response to LPS is regulated at both transcriptional and translational levels (13), it is reasonable to speculate that LPS influences posttranscriptional events that produce TNF-α. Therefore, while CAR appears to increase only TNF-α mRNA in whole-blood leukocytes, a large amount of TNF-α is then secreted in response to relatively low doses of LPS. The differential responses of various tissues to CAR might be due to their exposure to different concentrations of CAR. This conclusion is supported by the previous finding that CAR injected i.p. is not detected in bone marrow (27).

Our data demonstrates that CAR primes whole-blood leukocytes to produce TNF ex vivo and in vitro. CAR-treated cultured medium sensitized whole blood to make more LPS-induced TNF than did control medium that was preincubated with saline instead of CAR (Fig. 3). It is not clear at present, however, whether CAR directly or indirectly primes leukocytes after TNF production. In comparison with the whole-blood ex vivo experiments (Fig. 1), enhancement of TNF production was weak in the vitro experiment (Fig. 3). This suggests that inflammatory mediators induced in vivo by CAR pretreatment play an important role in priming for TNF production. It is known that inflammatory mediators, including platelet-activating factor (PAF), amplify the secretory potential of PMNs (1, 30). Further LPS-induced TNF-α production is suppressed in CAR-pretreated mice by a PAF receptor antagonist (21). The priming effect of CAR might thus be mediated through inflammatory mediators such as PAF. It has been also reported that phagocytosis of opsonized yeast can generate the signal necessary for the induction of TNF-α in PMNs (3). However, the actions of CAR on leukocytes are probably different from those of opsonized yeast because CAR pretreatment by itself did not induce TNF-α activity (Fig. 1). As seen in Fig. 5d, CAR also induces the destruction of macrophages. It is possible that some components or mediators from the destroyed macrophages might activate leukocytes. Further experiments will be required to understand the mechanisms for the priming effect of CAR on leukocytes.

We have previously demonstrated that TNF-α is present in secreted granules in macrophages (20). The present study showed that TNF-α also exists in granules in PMNs. Biochemical analysis has shown that TNF-α is produced as a 26-kDa membrane-bound precursor which is cleaved into an active 17-kDa protein (16, 26). We assume that TNF-α is stored in granules and then transferred to the plasma membrane.

It has been reported that endothelial cell necrosis occurs by a functional interaction among PMNs, LPS, and TNF (25, 28, 29, 35). As described here, TNF-α was localized on the apical surfaces of endothelial cells and in lysosomes of hepatocytes at an early stage after injection of LPS. The TNF-α detected in these areas appears not to have been derived from the endothelial cells and hepatocytes, since TNF-α mRNA was not detected in the livers of CAR-pretreated mice (data not shown). The TNF-α would therefore probably have originated in leukocytes found in the liver. TNF-α has been shown to release interleukin-8 from a variety of tissue cells, enhance adhesion of leukocytes to endothelium, and induce leukocyte degranulation as well as oxygen radical release, which causes endothelial cell necrosis (23). It has been also reported that the depletion of circulating leukocytes prevents LPS hepatotoxicity (14, 36). Our results demonstrate that activated leukocytes produce a large amount of TNF-α and that this plays an important role in the pathogenesis of liver injury in the CAR-pretreated endotoxin shock model.

Finally, there are several differences in the biological effects of CAR and d-galactosamine. First, mice pretreated with d-galactosamine produce TNF-α in macrophages, but not in PMNs (12). Second, d-galactosamine does not enhance LPS-induced TNF-α production (24). Third, d-galactosamine increases the sensitivity of host cells to TNF-α several thousand-fold (17), whereas CAR does not show such an effect (data not shown). We conclude, therefore, that the type of priming agents used in endotoxin-sensitized animals should be taken into account when the mechanisms of the pathophysiology of endotoxin shock and organ failure are studied.