The heat sensitivity of cytokine-inducing effect of lipopolysaccharide

Transcription

1 The heat sensitivity of cytokine-inducing effect of lipopolysaccharide Baochong Gao,*,,1 Yun Wang,*, and Min-Fu Tsan*, *Research Service, VA Medical Center, Washington, DC; Department of Medicine, George Washington University, Washington, DC; and Department of Medicine, Georgetown University, Washington, DC Abstract: Heat inactivation by boiling has been widely used as a criterion to determine whether the observed effects of a protein preparation are a result of lipopolysaccharide (LPS) contamination. However, the heat sensitivity of LPS cytokine-inducing activity has not been characterized. In the current study, we demonstrated that the endotoxin activity, i.e., Limulus amebocyte lysate-gelating activity, and the tumor necrosis factor (TNF- )- inducing activity of LPS (Escherichia coli K-12 JM83, K-12 LCD25, and F583) were sensitive to boiling. Heat treatment by boiling for 15 min was sufficient to inactivate 90% of the LPS TNF- inducing activity. The heat-induced inactivation of LPS activities was not a result of adherence of boiled LPS to the wall of the container, i.e., polypropylene tubes, or aggregation of boiled LPS. In addition, boiled LPS retained its ability to bind polymyxin B. The presence of protein (ovalbumin) in LPS did not affect the heat sensitivity of LPS. Conversely, boiling reduced the size of LPS aggregates as determined by electrophoresis using native polyacrylamide gel. Likewise, the TNF- -inducing activity of diphosphoryl lipid A (DPLA) was also sensitive to boiling. Thin-layer chromatographic analysis of boiled DPLA revealed that the heatinduced inactivation of DPLA TNF- -inducing activity was not a result of its conversion to monophosphoryl lipid A. We conclude that the TNF- inducing activity of LPS and DPLA is sensitive to boiling and suggest that heat sensitivity as an indicator of whether the observed effects of a protein preparation are a result of LPS contamination should be used with caution. J. Leukoc. Biol. 80: ; Key Words: LPS endotoxin activity tumor necrosis factor (TNF- ) macrophages INTRODUCTION Lipopolysaccharide (LPS; endotoxin), a major component of the outer membrane of the Gram-negative bacterial cell wall, is among the most potent modulators of the innate immune system. It is responsible for a host of toxic effects that occur in patients infected with Gram-negative bacteria, including fever, disseminated intravascular coagulation, and hemodynamic changes, which may lead to multiple organ failure characteristics of the septic shock syndrome [1, 2]. Since Richard Pfeifer demonstrated at the end of the 19th century that heat-inactivated Vibrio cholerae were capable of inducing irreversible shock in experimental animals, LPS has become known as the heat-resistant endotoxin [1 3]. In fact, it is believed that the pyrogenic effect of LPS is resistant even to autoclaving and that dry heat at 250ºC for 1 2 h or at 180ºC for 4hisrequired to render a substance pyrogen-free [4]. Thus, it has been recommended that a simple heat inactivation, i.e., boiling for at least 30 min, can be used to determine whether the observed effects of a protein preparation are a result of LPS contamination [5]. This has become the single most widely used standard to rule out LPS contamination as the cause of the observed effects. The concept that LPS is heat-resistant derives from the fact that unlike bacterial exotoxins, which are peptides inactivated readily and completely by boiling, some LPS effects remain, even after boiling for 1 h, as demonstrated by Richard Pfeifer s original observation more than a century ago. Because of its ubiquitous presence and its potent pyrogenic effect, the Food and Drug Administration and the United States Pharmacopeia have set strict standards for the limits of LPS contamination in pharmaceutical products [6, 7]. The most effective way of endotoxin decontamination is by dry heat at temperatures between 170 C and 250 C for a few hours. This can achieve many orders of magnitude of reduction in the LPS endotoxin activity, as measured by the Limulus amebocyte lysate (LAL) assay [8 10] and is part of the current good manufacturing practice used in the pharmaceutical industry [4]. However, even wet heat at 100 C (boiling) has been shown to cause significant reduction in the LPS endotoxin activity [11, 12]. Although heat sensitivity has often been used as a criterion to determine whether the observed cytokine-inducing effect of a protein preparation is a result of LPS contamination [13], the heat sensitivity of LPS cytokine-inducing activity has not been studied carefully. In the current study, we characterized the heat sensitivity of the tumor necrosis factor (TNF- )-inducing effect of LPS. 1 Correspondence: VA Medical Center (10R), 50 Irving Street, N.W., Washington, DC baochong.gao@med.va.gov Received December 16, 2005; accepted April 7, 2006 doi: / jlb /06/ Society for Leukocyte Biology Journal of Leukocyte Biology Volume 80, August

2 MATERIALS AND METHODS Materials Protein-free LPS (from Escherichia coli K-12 JM83, rough strain) was a generous gift of Dr. John E. Somerville (Bristol-Myers Squibb Pharmaceutical Research Institute, Princeton, NJ). It had an exdotoxin activity of endotoxin unit (EU)/ng (n 11), as determined using the LAL assay (see below). This LPS preparation is free of bacterial lipoprotein contamination, as we have previously demonstrated that at concentrations as high as 1000 ng/ml, it did not induce TNF- production by macrophages from Toll-like receptor 4 (TLR4)-deficient mice (C57BL/10ScCr) [14] and that its TNF- -inducing activity in RAW murine macrophages was completely inhibitable by polymycin B [15]. 3 H-LPS (from E. coli K-12 LCD25, rough strain, specific radioactivity: 0.94 Ci/ g) was purchased from List Biological Laboratories, Inc. (Campbell, CA) and contained an endotoxin activity of EU/ng (n 11). The 3 H label was confined to the fatty acyl chains located in the lipid A moiety of LPS. Monophosphoryl lipid A (MPLA), diphosphoryl lipid A (DPLA), and LPS from E. coli F583 were purchased from Sigma Chemical Co. (St. Louis, MO). The endotoxin activity of MPLA and DPLA was determined to be EU/ng (n 8) and EU/ng (n 9), respectively. LPS F583 had an endotoxin activity of EU/ng (n 8). Before use, LPS and 3 H-LPS were dissolved in sterile, pyrogen-free water, and lipid A was dissolved in 2% triethylamine. They were then sonicated on ice for 30 s with a Sonic Dismembrator (Fisher Scientific, Houston, TX) and diluted with phosphate-buffered saline (PBS; Gibco/Invitrogen Corp., Carlsbad, CA). Ovalbumin (OVA) was obtained from Sigma Chemical Co., and polymyxin B-agarose (Detoxi Gel) was obtained from Pierce (Rockford, IL). Pro-Q Emerald 300 LPS stain kits were purchased from Molecular Probes (Eugene, OR). Cell culture RAW murine macrophages (from American Type Culture Collection, Manassas, VA) were cultured in Dulbecco s modified Eagle s medium (DMEM; Gibco/Invitrogen Corp.), supplemented with 10% fetal bovine serum (FBS; Gibco/Invitrogen Corp.) and antibiotics, as described previously [15, 16]. Subcultures of macrophages were prepared every 2 3 days by scraping cells into fresh medium. Measurement of endotoxin activity The endotoxin activities of LPS, 3 H-LPS, MPLA, DPLA, and OVA preparations were determined as described previously [15, 16] using the LAL assay kit (BioWhittaker, Walkersville, MD) according to the manufacturer s recommendation. Heating of LPS Heating of LPS and 3 H-LPS at 20 ng/ml 200 g/ml or lipid A at 200 g/ml in 1.5 ml sterile polypropylene tubes (Sarstedt Microtubes with o-ring screw cap, Fisher Scientific, Pittsburgh, PA) was carried out in a boiling water bath for 15, 30, or 60 min. In some experiments, OVA (0.1 or 1 mg/ml) was added to the LPS solution prior to heating. Determination of TNF- release by murine macrophages Murine macrophages were seeded in 24-well plates at cells/well on the day before the experiment. After washing three times with the medium, cells were treated with or without control or boiled LPS ( ng/ml), 3 H-LPS (0.2 ng/ml), or MPLA or DPLA (2 ng/ml) in 250 l DMEM containing 10% FBS for 4 h at 37ºC. At the end of treatment, media were collected and clarified by centrifugation at 10,000 revolutions per minute for 5 min in a microcentrifuge (Hermle-Labortechnik, Wehingm, Germany). TNF- content of the media was then determined by a quantitative sandwich enzyme-linked immunosorbant assay using the Quantikine M mouse TNF- immnoassay kit (R&D Systems, Minneapolis, MN) according to the manufacturer s recommendation. All experiments were carried out with duplicate samples. Binding of LPS to polymyxin B agarose The ability of 3 H-LPS to bind polymyxin B was determined using polymyxin B agarose columns as described previously [15, 16]. Briefly, aliquots of 0.5 ml polymyxin B agarose were poured into Poly-Prep disposable columns (Bio- Rad, Hercules, CA), which were washed with 5 vol 1% sodium deoxycholate followed by 20 vol PBS. 3 H-LPS was boiled at 2 g/ml for 60 min and diluted to 400 ng/ml in PBS. Control or boiled 3 H-LPS (250 l) was loaded onto each 0.5 ml polymyxin B agarose column and incubated at room temperature for 60 min after collecting a 200- l void volume. Columns were then eluted with PBS in 250 l fractions. The radioactivity present in the eluted fractions and polymyxin B agarose were quantified by liquid scintillation counting. Analysis of LPS by gel electrophoresis Control and boiled LPS samples were loaded at 2 g/lane onto sodium dodecyl sulfate (SDS) gels with 10 20% polyacrylamide gradient (Novex, Invitrogen Corp.) or native gels with 4 12% polyacrylamide gradient. After electrophoresis, LPS was stained with Pro-Q Emerald 300 LPS stain kits, according to the manufacturer s recommendation, visualized, and photographed using the FluorChem 8000 digital imaging system (Alpha Innotech Corp., San Leandro, CA). Analysis of lipid A by thin-layer chromatography (TLC) TLC analysis of control and boiled lipid A (10 g/lane) was carried out on Silica Gel 60 Matrix Merck TLC plates (Sigma Chemical Co.) using the solvent system chloroform/methanol/water/triethylamine (300:120:20:1). Lipid A was detected by spraying with 15% sulfuric acid in ethanol and heating at 130 C for 30 min. The plates were photographed using the FluorChem 8000 digital imaging system (Alpha Innotech Corp.). Statistical analysis Results were expressed as mean SD. Levels of significance were determined using a two-tailed Student s t-test [17], and a confidence level of greater than 95% (P 0.05) was used to establish statistical significance. RESULTS Effect of durations of boiling on the cytokineinducing activity of LPS To assess the heat sensitivity of LPS cytokine-inducing activity, investigators have heated LPS in a boiling water bath for various intervals ranging from 15 min to 60 min and tested the cytokine-inducing activities of LPS at concentrations ranging from 1 ng/ml to 1 g/ml with various conclusions [13, 18]. Therefore, we first heated LPS (K-12 JM83, 200 ng/ml) in polypropylene tubes by boiling for 15, 30, or 60 min and determined the ability of control (nonheated) or heated LPS at concentrations ranging from 0.1 to 10 ng/ml to induce TNF- release by murine macrophages. As shown in Figure 1, the amounts of TNF- released by macrophages reached a plateau at a concentration of 1 ng/ml control LPS. Boiling of LPS for min resulted in a clear reduction of LPS-induced TNF- release by macrophages, and the highest reduction was observed after 60 min of boiling. The concentration of control LPS, which induced 50% of the maximal TNF- release, was 0.2 ng/ml, whereas that of LPS, boiled for 15 or 30 min, was 2 ng/ml (i.e., a 90% reduction in the TNF- -inducing activity), and it was 10 ng/ml after boiling for 60 min. These results also demonstrated that the amounts of TNF- released were similar at concentrations of 5 ng/ml for control LPS or LPS boiled for 15 or 30 min, giving the impression that LPS was heat-resistant, when LPS was tested only at concentrations greater than 5 ng/ml. 360 Journal of Leukocyte Biology Volume 80, August

3 myxin B agarose columns and incubated at room temperature for 1 h. The agarose columns were then eluted with PBS, and the radioactivity of eluted fractions as well as polymyxin B agarose was determined. As shown in Figure 3, A and B, the percentage of control LPS bound to polymyxin B agarose was %, and that of boiled LPS was % (n 3, P 0.10). Thus, heat treatment had no significant effect on the ability of 3 H-LPS to bind polymyxin B, suggesting that 3 H was still associated with the lipid A moiety of LPS. Effect of boiling on LPS in the presence of proteins Fig. 1. Effect of duration of boiling on the TNF- -inducing activity of LPS (K-12 JM83, 200 ng/ml), which was heated by boiling for 15, 30, or 60 min. The ability of control and boiled LPS at concentrations ranging from 0.1 to 10 ng/ml to induce TNF- release from murine macrophages was then determined. Effect of boiling on the adherence of LPS to the container wall It is possible that the loss of LPS activity after boiling was a result of adherence of LPS to the container wall, rendering LPS unavailable for testing. To address this possibility, 3 H-LPS (K-12 LCD25, 20 ng/ml, 200 ng/ml, or 2 g/ml, in polypropylene tubes) was heated by boiling for 15, 30, or 60 min. The endotoxin activity (LAL-gelating activity), the TNF- -inducing activity, and the radioactivity of control and boiled LPS were then determined. The endotoxin activity and the TNF- -inducing activity were measured at LPS concentrations of 0.5 and 0.2 ng/ml, respectively, as preliminary experiments had shown that these concentrations were within the linear range of the assays [15, 16]. As shown in Figure 2A, the endotoxin activity of boiled 3 H-LPS, as compared with that of control 3 H-LPS, was reduced markedly by 80 96% [P values (boiled LPS vs. control LPS) 0.05]. Likewise, the TNF- -inducing activity of boiled 3 H-LPS was reduced markedly by 81 93% [P values (boiled LPS vs. control LPS) 0.05]. In contrast, the radioactivity of boiled 3 H-LPS was only reduced by 14 35%. Thus, adherence of boiled LPS to polypropylene tubes, if any, could not account for the loss of LPS endotoxin activity or TNF- -inducing activity. Effect of boiling on LPS binding to polymyxin B It is possible that heating of 3 H-LPS releases 3 H from LPS. Thus, measurement of radioactivity of heated 3 H-LPS may not accurately reflect the amount of 3 H-LPS present in the solution. To assess whether 3 H was still associated with LPS after heating, we took advantage of the binding capacity of LPS to polymyxin B, a cationic antibiotic that binds lipid A [19]. 3 H-LPS (K-12 LCD25, 2 g/ml) was heated by boiling for 60 min. Control and heated 3 H-LPS were then loaded onto poly- Heat inactivation is widely used to determine whether the observed effects of a protein solution are a result of LPS contamination [5]. Therefore, we determined whether the presence of protein affected the heat sensitivity of LPS. 3 H-LPS (K-12 LCD25, 200 ng/ml) was boiled for 60 min in the presence or absence of OVA at 0.1 or 1 mg/ml. The endotoxin activities of control and boiled 3 H-LPS were then determined. The OVA preparation had an endotoxin activity of 200 EU/mg, which was equivalent to 30 ng 3 H-LPS/mg OVA. As shown in Figure 4, heat treatment markedly reduced the endotoxin activity of 3 H-LPS, and the presence of OVA at 0.1 or 1 mg/ml did not affect the heat sensitivity of 3 H-LPS. Effect of boiling on aggregation of LPS LPS is an amphiphilic molecule that forms aggregates in aqueous solutions at concentrations higher than its critical micellar concentration [20, 21]. It is possible that heat treatment of LPS may cause further aggregation of LPS. Heat-induced LPS aggregation may account for the loss of activity, as aggregated LPS may have lower activities [22, 23]. We took two approaches to address this possibility. First, we determined the endotoxin activity and the TNF- inducing activity of control and heated LPS (K-12 JM83) before and after sonication for 30 s to disperse aggregates. As shown in Figure 5, sonication of control and boiled LPS slightly increased their endotoxin activities as well as their TNF- inducing activity. However, it had no effect on the heatinduced inactivation of LPS. Second, we determined the sizes of control and boiled LPS aggregates using polyacrylamide gel electrophoresis (PAGE) under denaturing and nondenaturing (native) conditions. The low-staining sensitivity of LPS on PAGE required the use of LPS at a relatively high concentration. LPS (K-12 JM83, 200 g/ml) was boiled for 15, 30, or 60 min. The endotoxin activity and the TNF- -inducing activity of control and boiled LPS were determined. In addition, 2 g/lane control or boiled LPS was loaded onto a SDS gel of 10 20% or native gel of 4 12% polyacrylamide gradient for electrophoretic analysis. As shown in Figure 6, A and B, heat treatment markedly reduced the endotoxin activity and the TNF- -inducing activity of LPS, even when LPS was boiled at a concentration of 200 g/ml. Under denaturing condition, control and boiled LPS showed similar mobility, reaching the front of the SDS-PAGE (Fig. 6C). However, under the native condition (Fig. 6D), most of control LPS barely entered the gel, suggesting that LPS was mainly present as aggregates of high molecular weights. In Gao et al. Heat sensitivity of LPS 361

4 Fig. 2. Effect of boiling on the adherence of LPS to the container wall. 3 H-LPS (K-12 LCD25, 20 ng/ml, 200 ng/ml, or 2 g/ml, in polypropylene tubes) was heated by boiling for 15, 30, or 60 min. The endotoxin activity of control and boiled LPS at 0.5 ng/ml and the ability of control and boiled LPS at 0.2 ng/ml to induce TNF- release from murine macrophages were then determined. The radioactivity of control and boiled LPS was also determined by scintillation counting. (A) Endotoxin activity; (B) TNF- release; and (C) radioactivity. Results were the mean SD of three experiments; each experiment was done with duplicate samples. (A and B) P values (boiled LPS vs. control LPS) CPM, Counts per minute. contrast, boiled LPS showed a time-dependent decrease in the size of the aggregates. LPS, after boiling for 60 min, was present, mainly as aggregates of 17 kda in size. These results suggest that heat treatment actually reduced the size of LPS aggregates, instead of causing further aggregation of LPS. Effect of boiling on lipid A Lipid A is the endotoxic principal of LPS and consists of a dimmer of N-acetylglucosamine, which is phosphorylated at Positions 1 and 4, i.e., DPLA. It has been shown that removal of the acid labile phosphate from Position 1 of DPLA results in a marked reduction in the cytokine-inducing activity [20]. To determine whether the above, observed boiling-induced reduction in LPS TNF- -inducing activity was a result of conversion of its lipid A moiety from DPLA to MPLA, we studied the effect of boiling on DPLA and MPLA. We first determined the heat sensitivity of cytokine-inducing activities of MPLA, DPLA, and LPS (F583), from which MPLA and DPLA were derived. As shown in Figure 7A, MPLA (2 ng/ml) had no TNF- -inducing activity in RAW murine macrophages. The TNF- -inducing activity of DPLA (2 ng/ml) was about half that of LPS (F583, 0.5 ng/ml). However, the TNF- -inducing activities of DPLA and LPS were sensitive to heat-inactivation by boiling. TLC was then used to determine whether the above, observed heat inactivation of DPLA was a result of its conversion to MPLA. As shown in Figure 7B, control MPLA (lane 1) and DPLA (lane 2) showed multiple bands (species) on TLC, as 362 Journal of Leukocyte Biology Volume 80, August

5 Fig. 3. Effect of boiling on LPS binding to polymyxin B. 3 H-LPS (K-12 LCD25, 2 g/ml) was heated by boiling for 60 min. Control or boiled LPS (250 l) at 400 ng/ml was loaded onto a polymyxin B agarose column and incubated at room temperature for 1 h after collecting a 200- l void volume. Columns were eluted with four fractions of 250 l PBS. The radioactivity of all eluted fractions and polymyxin B agarose from each column were then determined by scintillation counting (A), and the percentage of control and boiled LPS bound to polymyxin B agarose was calculated (B). Values presented were the means SD of three experiments. reported in the literature [24]. However, MPLA and DPLA could be distinguished readily from the TLC. Boiling of DPLA for 60 min (lane 3) did not alter the TLC pattern from that of control DPLA, suggesting that boiling did not cause a conversion of DPLA to MPLA. DISCUSSION The results presented in the current study demonstrated that the LPS endotoxin activity, i.e., the LAL-gelating activity, and the TNF- -inducing activity were sensitive to boiling. Heat treatment by boiling for 15 min was sufficient to inactivate 90% of the LPS TNF- -inducing activity. The heat-induced inactivation of LPS was not a result of adherence of boiled LPS to the wall of polypropylene tubes or aggregation of boiled LPS. Fig. 4. Effect of boiling on LPS in the present of OVA. 3 H-LPS (K-12 LCD25, 200 ng/ml) was heated by boiling for 60 min in the presence or absence of OVA at 1 or 0.1 mg/ml. The endotoxin activity of control and boiled LPS at 0.5 ng/ml was then determined. Results were the mean SD of three experiments, and each experiment was done with duplicate samples. P values (boiled LPS vs. control LPS) In addition, boiled LPS retained its ability to bind polymyxin B. The presence of protein (OVA) did not affect the heat sensitivity of LPS. Conversely, boiling reduced the size of LPS aggregates. Likewise, the TNF- -inducing activity of DPLA was also shown to be heat-sensitive, and the heatinduced inactivation of DPLA was not a result of its conversion to MPLA. It has been shown that commercially available LPS preparations might be contaminated with bacterial lipoproteins [25, 26]. Thus, it is possible that the observed LPS heat sensitivity was a result of contamination by lipoproteins. This is unlikely for the following reasons. First, we have demonstrated previously that the LPS preparation (E. coli K-12 JM83) used in the current study was not contaminated with bacterial lipoproteins, as even at a concentration of 1000 ng/ml, it failed to cause the induction of TNF- mrna and the production of TNF- by macrophages from TLR4-deficient mice (C57BL/10ScCr) [14]. In addition, its TNF- -inducing activity in RAW murine macrophages was completed inhibitable by polymycin B, a specific LPS inhibitor, which has no effect on lipoproteins [15]. Second, the LPS endotoxin activity was equally sensitive to heat inactivation. Bacterial lipoproteins do not have endotoxin activity, i.e., LAL-gelating activity. The fact that the biological activities of LPS are sensitive to boiling has been reported previously by a number of investigators. Neter et al. [27] reported that heating LPS in a boiling water bath for 60 min reduced the LPS immunogenicity in rabbits by 95%, and it had no effect on the LPS antigenecity. Likewise, Piotrowicz and McCartney [11] and Fujii et al. [12] showed that boiling E. coli LPS markedly reduced its endotoxin activity, and Vikstrom [28] reported that the immunosuppressive activity of Shigella sonnei LPS was heat-sensitive. As demonstrated in the current study, boiling LPS, ranging from 20 ng/ml to 200 g/ml for min, markedly reduced the endotoxin activity as well as the TNF- -inducing activity of E. coli LPS. It is thus important to ask why heat sensitivity continues to be used widely as an indicator of whether the observed effects of a protein preparation are a result of LPS contamination. The results presented in the current study provide a rational explanation for this discrepancy. Gao et al. Heat sensitivity of LPS 363

6 Fig. 5. Effect of sonication on endotoxin activity and TNF- -inducing activity of LPS (K-12 JM83, 200 ng/ml), which was heated by boiling for 15, 30, or 60 min. The endotoxin activity of control and boiled LPS at 0.5 ng/ml and the ability of control and boiled LPS at 0.2 ng/ml to induce TNF- release from murine macrophages before and after sonication for 30 s at 4ºC were then determined. (A) Endotoxin activity; (B) TNF- release. As shown in Figure 1, a critical factor is the amount of LPS used to test the heat sensitivity of LPS. As LPS is a potent modulator of the innate immune system, LPS in pg/ml concentrations is sufficient to induce the inflammatory cytokine release from macrophages. In fact, as demonstrated in Figure 1, the amount of TNF- released by murine macrophages reached a plateau at a LPS concentration of 1 ng/ml. With a LPS concentration of less than 1 ng/ml, the effect of heat inactivation can be detected easily, even after boiling for only 15 min. Conversely, if a LPS concentration greater than 5 ng/ml were used, no difference would be seen in the TNF- -inducing activity between control and boiled LPS, even after boiling for 30 min. This is because there was sufficient residual LPS activity present in the boiled LPS to induce the maximal release of TNF- from macrophages. Thus, it can be concluded that LPS is heat-sensitive or heat-resistant, depending on the concentrations of LPS used to test the heat sensitivity. As most studies used a concentration of LPS ranging from 10 ng/ml to 1 g/ml to test the heat sensitivity [13, 18], it could be concluded that LPS was heat-resistant. Conversely, the contaminating concentration of LPS in the test samples may be less than 1 ng/ml, which will be readily shown to be heatsensitive. When using LPS heat sensitivity as a criterion to determine whether the observed effect is a result of contami- Fig. 6. Effect of boiling on the size of LPS aggregates. LPS (K-12 JM83, 200 g/ml) was heated by boiling for 15, 30, or 60 min. The endotoxin activity (A) of control and boiled LPS at 0.5 ng/ml and the ability of control and boiled LPS at 0.2 ng/ml to induce TNF- release from murine macrophages (B) were determined. Control or boiled LPS (2 g/lane) was loaded onto a SDS gel of 10 20% (C) or native gel of 4 12% (D) polyacrylamide gradient. After electrophoresis, LPS was visualized by staining with Pro-Q Emerald 300 LPS stain kits and photographed. (C) Lane 1, Control LPS; lane 2, LPS boiled for 60 min. (D) Lane 1, Control LPS, lanes 2 4, LPS boiled for 15, 30, and 60 min, respectively. (A and B) The results presented are means SD of three experiments, and each experiment was done with duplicate samples. P values (A and B, control LPS vs. boiled LPS) (C and D) The results from a representative experiment. Two additional experiments were carried out and showed similar results. 364 Journal of Leukocyte Biology Volume 80, August

7 Fig. 7. Effect of boiling on the TNF- -inducing activity and phosphorylation of lipid A. MPLA, DPLA, and LPS, all from E. coli F583, were boiled at 200 g/ml for 60 min. The TNF- -inducing activity (A) was determined using lipid A at 2 ng/ml and LPS at 0.5 ng/ml. Results were the means SD of three experiments, each done with duplicate samples. *, P values (control vs. boiled) The control MPLA and DPLA and boiled DPLA were analyzed using TLC (B). Ten micrograms were loaded per lane. Lane 1, MPLA; lane 2, control DPLA; lane 3, boiled DPLA. The TLC plate is a representative of three experiments with similar results. nating LPS, it is, therefore, important to compare the heat sensitivity of LPS at the same concentration as what is present in the sample being tested [15, 16]. In the current study, we determined the heat sensitivity of LPS using two different parameters, i.e., the endotoxin activity and the TNF- -inducing activity. Although the endotoxin activity, as measured by the LAL assay, is the most widely used method to quantify the amount of LPS, it does not predict the cytokine-inducing activity of LPS. A mutant E. coli LPS lacking myristoyl fatty acid at the 3 R-3-hydroxymyristate position of the lipid A moiety, i.e., nonmyristoyl LPS, retains its endotoxin activity, but it fails to induce the release of TNF- by human macrophages [29, 30]. The 3 H-LPS used in the current study was from E. coli K12 LCD25, and the nonlabeled LPS was from E. coli K-12 JM83. The 3 H-LPS (K-12 LCD25) had a higher endotoxin activity than that of the LPS (K-12 JM83), i.e., EU/ng versus EU/ng. In contrast, at the same concentration (i.e., 0.2 ng/ml), the LPS (K-12 JM83) induced a much higher TNF- release from macrophages than did the 3 H-LPS [K-12 LCD25; ng/ml (n 9) vs ng/ml (n 9)]. However, 3 H-LPS (K-12 LCD25) and LPS (K-12 JM83) were equally sensitive to heat inactivation, whether it was assessed by the endotoxin activity or the TNF- -inducing activity. LPS is an amphiphilic molecule, which forms aggregates in aqueous solutions. There has been controversy regarding whether the aggregated LPS is more active or less active than the monomeric LPS in inducing TNF- release from macrophages. Earlier studies [22, 23] have suggested that the aggregated LPS is less active than the monomeric LPS. However, more recently, Mueller et al. [31] reported that aggregated LPS was more active than the monomeric LPS. In the current study, we demonstrated that boiling LPS caused a time-dependent decrease in the size of LPS aggregates. The physical and chemical bases for these heat-induced changes in LPS aggregate sizes and LPS activities are not clear. However, using DPLA, we have shown that the heat-induced loss of TNF- inducing activity was not a result of conversion to MPLA. Further studies are necessary to identify the physical and chemical changes induced by boiling LPS responsible for the changes in LPS aggregate sizes and LPS activities. With the ready availability of recombinant DNA products, there has been an intense interest in recent years in the extracellular functions of these molecules. Likewise, there has been a plethora of reports suggesting that a number of endogenous molecules may be potent activators of the innate immune system via TLRs, i.e., endogenous ligands of TLRs [18, 32]. The results presented in the current study provide clear evidence that the endotoxin activity and the TNF- -inducing activity of LPS are sensitive to heat by boiling, even for 15 min. We, therefore, suggest that heat sensitivity as an indicator of the presence of LPS contamination in protein solutions should be used with caution. ACKNOWLEDGMENT This material is based on work supported by the Office of Research and Development, Department of Veterans Affairs. REFERENCES 1. Morrison, D. C., Ryan, J. L. (1979) Bacterial endotoxins and host immune responses. Adv. Immunol. 28, Rietschel, E. T., Brade, H. (1992) Bacterial endotoxin. Sci. Am. 267, Rietschel, E. T., Kirikae, T., Schade, F. U., Ulmer, A. J., Holst, O., Brade, H., Schmidt, G., Mamat, U., Grimmecke, H. D., Kusumoto, S., Zahringer, U. (1993) The chemical structure of bacterial endotoxin in relation to bioactivity. Immunobiology 187, Sharma, S. K. (1986) Endotoxin detection and elimination in biotechnology. Biotechnol. Appl. Biochem. 8, Majde, J. A. (1993) Microbial cell-wall contaminants in peptides: a potential source of physiological artifacts. Peptides 14, Food and Drug Administration Health and Human Services (1987) Guideline on Validation of the Limulus Amebocyte Lysate Test As An End- Product Endotoxin Test for Human and Animal Parenteral Drugs, Biological Products, and Medical Devices, 7. Pharmacopeial Forum (1982) The New USP Reference Standard Endotoxin A Collaborative Project, Tsuji, K., Harrison, S. J. (1978) Dry-heat destruction of lipopolysaccharide: dry-heat destruction kinetics. Appl. Environ. Microbiol. 36, Nakata, T. (1993) Destruction of typical endotoxins by dry heat as determined using LAL assay and pyrogen assay. J. Parenter. Sci. Technol. 47, Avis, K. E., Jewell, R. C., Ludwig, J. D. (1987) Studies on the thermal destruction of Escherichia coli endotoxin. J. Parenter. Sci. Technol. 41, Piotrowicz, B. I., McCartney, A. C. (1986) Effect of heat on endotoxin in plasma and in pyrogen-free water, as measured in the Limulus amoebocyte lysate assay. Can. J. Microbiol. 32, Fujii, S., Takai, M., Maki, T. (2002) Wet heat inactivation of lipopolysaccharide from E. coli serotype O55:B5. PDA J. Pharm. Sci. Technol. 56, Tsan, M-F., Gao, B. (2004) Cytokine function of heat shock proteins. Am. J. Physiol. Cell Physiol. 286, C739 C744. Gao et al. Heat sensitivity of LPS 365

8 14. Tsan, M-F., Clark, R. N., Goyert, S. M., White, J. E. (2001) Inducation of TNF- and MnSOD by endotoxin: role of membrane CD14 and Toll-like receptor-4. Am. J. Physiol. Cell Physiol. 280, C1422 C Gao, B., Tsan, M-F. (2003) Recombinant human heat shock protein 60 does not induce the release of tumor necrosis factor from murine macrophages. J. Biol. Chem. 278, Gao, B., Tsan, M-F. (2003) Endotoxin contamination in recombinant human heat shock protein 70 (Hsp70) preparation is responsible for the induction of tumor necrosis factor release by murine macrophages. J. Biol. Chem. 278, Fisher, R. A., Yates, F. (1963) Statistical Tables for Biological, Agricultural and Medical Research, 6th ed., New York, NY, Hafner. 18. Tsan, M-F., Gao, B. (2004) Endogenous ligands of Toll-like receptors. J. Leukoc. Biol. 76, Morrison, D. C., Jacobs, D. M. (1976) Binding of polymyxin B to the lipid A portion of bacterial lipopolysaccharide. Immunochemistry 13, Brandenburg, K., Wiese, A. (2004) Endotoxin: relationships between structure, function, and activity. Curr. Top. Med. Chem. 4, Sweadner, K. J., Forte, M., Nelsen, L. L. (1977) Filtration removal of endotoxin (pyrogens) in solution in different states of aggregation. Appl. Environ. Microbiol. 34, Takayama, K., Din, Z. Z., Mukerjee, P., Cooke, P. H., Kirkland, T. N. (1990) Physicochemical properties of the lipopolysaccharide unit that activates B lymphocytes. J. Biol. Chem. 265, Takayama, K., Mitchell, D. H., Din, Z. Z., Mukerjee, P., Li, C., Coleman, D. L. (1994) Monomeric Re lipopolysaccharide from Escherichia coli is more active than the aggregated form in the Limulus amebocyte lysate assay and in inducing Egr-1 mrna in murine peritoneal macrophages. J. Biol. Chem. 269, Qureshi, N., Takayama, K. (1982) Puritification and structural determination of nontoxic lipid A obtained from the lipopolysaccharide of Salmonella typgimurim. J. Biol. Chem. 257, Hirschfeld, M., Ma, Y., Weis, J. H., Vogel, S. N., Weis, J. J. (2000) Cutting edge: repuritification of lipopolysaccharide eliminates signaling through both human and murine Toll-like receptor 2. J. Immunol. 165, Lee, H. K., Lee, J., Tobias, P. S. (2002) Two lipoproteins extracted from Escherichia coli K-12 LCD25 lipopolysaccharide are the major components responsible for Toll-like receptor 2-mediated signaling. J. Immunol. 168, Neter, E., Whang, H. Y., Mayer, H. (1972) Immunogenecity and antigenecity of endotoxic lipopolysaccharides: reversible effects of temperature on immunogenecity. In Bacterial Lipopolysaccharides, the Chemistry, Biology and Clinical Significance of Endotoxins (E. H. Kass, S. M. Wolff, eds.), Chicago, IL, The University of Chicago Press, Vikstrom, E. V. (2002) The immunosuppressive activity of chemically modified lipopolysaccharide of Shigella sonnei. Immunol. Lett. 80, Somerville Jr., J. E., Cassiano, L., Bainbridge, B., Cunningham, M. D., Darveau, R. P. (1996) A novel Escherichia coli lipid A mutant that produces an anti-inflammatory lipopolysaccharide. J. Clin. Invest. 97, Tian, L., White, J. E., Lin, H. Y., Haran, V. S., Sacco, J., Chikkappa, G., Davis, F. B., Davis, P. J., Tsan, M-F. (1998) Induction of MnSOD in human monocytes without inflammatory cytokine production by a mutant endotoxin. Am. J. Physiol. 275, C740 C Mueller, M., Linder, B., Kusumoto, S., Fukase, K., Schromn, A. B., Seydel, U. (2004) Aggregates are the biologically active units of endotoxin. J. Biol. Chem. 279, Beg, A. A. (2002) Endogenous ligands of Toll-like receptors: implications for regulating inflammatory and immune responses. Trends Immunol. 23, Journal of Leukocyte Biology Volume 80, August