Interactions of Lipid A and Liposome-Associated Lipid A with Limulus polyphemus Amoebocytes

Transcription

1 INFECTION AND IMMUNITY, Mar p /83/ $02.00/0 Copyright 1983, American Society for Microbiology Vol. 39, No. 3 Interactions of Lipid A and Liposome-Associated Lipid A with Limulus polyphemus Amoebocytes EARL C. RICHARDSON,' BENOY BANERJI,' ROBERT C. SEID, JR.,2 JACK LEVIN,4 AND CARL R. ALVINGl* Departments of Membrane Biochemistry' and Bacterial Diseases,2 Walter Reed Army Institute of Research, Washington, D.C ; The Marine Biological Laboratory, Woods Hole, Massachusetts ; and the Departments of Laboratory Medicine and Medicine, University of California School of Medicine, San Francisco, California Received 19 August 1982/Accepted 9 December 1982 Lipid A or lipid A fractions and liposomes containing lipid A were tested for the ability to gel Limulus amoebocyte lysates and for effects on intact Limulus amoebocytes. Liposomes having a relatively low concentration of lipid A did not produce coagulation of lysate and were designated as Limulus-negative, but liposomes having a high concentration of lipid A were Limulus-positive. Limulusnegative liposomes had no effect on intact amoebocytes. Limulus-positive liposomes caused a striking transformation in the appearance of amoebocytes in that the cells sent out long filamentous extensions that formed a tangled network of processes between cells. The filamentous projections were similar to those that have been previously observed in the presence of gram-negative bacteria. We conclude that amoebocytes have the ability to recognize Limulus-positive liposomes, but the lack of activation of Limulus lysate or the absence of amoebocyte recognition does not prove the absence of liposomal lipid A. We also found that individual lipid A fractions were heterogeneous in their ability to gel lysate. Of eight fractions tested, one (fraction 1) had no detectable activity above the background, and the seven others had activity that ranged from 10-fold to 10,000- fold above the background. The heterogeneity of lipid A fractions detected in assays with amoebocyte lysate was consistent with the finding of heterogeneity in other functional assays of lipid A fractions. The blood (hemolymph) of the horseshoe crab, Limulus polyphemus, contains only one type of cell, the amoebocyte. After the injection of gram-negative bacteria into L. polyphemus, the clumping of amoebocytes and intravascular coagulation and thrombosis can occur, resulting in the death of the animal (for a review, see reference 6). The mechanisms involved in the coagulation process have also been studied in vitro. Upon in vitro contact either with gramnegative bacteria or with endotoxin derived from the bacteria, the hemolymph clots (6, 8). Coagulation only occurs in the presence of amoebocyte-derived proteins (8, 9). The gelation (coagulation) that results from the interaction of amoebocyte lysate with endotoxin has been employed as an extremely sensitive in vitro assay for endotoxin (7, 8, 10, 12). The constituent of endotoxin that causes Limulus lysate gelation is a glycophospholipid designated lipid A (11, 16). Lipid A actually is a heterogeneous group of at least eight components, each component having a different longchain fatty acid composition (5). The removal of fatty acid esters from lipid A by mild alkaline hydrolysis substantially reduces the Limulus gelation activity (13, 16). Because of the presence of the fatty acid constituents, lipid A is readily inserted into the lipid bilayer of liposomes (phospholipid vesicles). The inclusion of lipid A in liposomes markedly reduces several functions of lipid A, including Limulus lysate gelation activity (14). The purpose of the present study was to examine the effects of lipid A and lipid A fractions and lipid A and its fractions incorporated into liposomes on both Limulus amoebocyte lysate and intact Limulus amoebocytes. MATERIALS AND METHODS Lipids. Commercial sources for lipids were as follows: dimyristoyl phosphatidylcholine (DMPC), Sigma Chemical Co., St. Louis, Mo. or Calbiochem- Behring Corp., San Diego, Calif.; cholesterol (CHOL), Calbiochem-Behring; dicetyl phosphate (DCP), K & K Laboratories, Plainview, N.Y. Lipid A. Lipid A from Shigellaflexneri lipopolysaccharide, prepared by the Westphal method (lot no. 1385

2 1386 RICHARDSON ET AL. INFECT. IMMUN. I- 2 D 10-6 F. lo-4 o F-, = 3A :-. 1o-2 hot-air oven for at least 8 h. Stock solutions containing A) LIPID A (B) LIPID A free lipid A or lipid A incorporated into liposomes N SOLUTION IN LIPOSOMES were sonicated for 3 min in an Ultra-Med II sonic bath. Serial 10-fold dilutions were made with pyrogen-free X7 normal saline. A 0.1-ml sample of each dilution was pipetted into a tube (10 by 75 mm), followed by the X*-w -w - E. coli addition of 0.1 ml of reconstituted Limulus amoebocyte lysate. After gentle mixing, the solutions were allowed to incubate undisturbed for 1 h at 37 C. The test was judged positive when the gel that formed did not collapse upon complete inversion of the tube. Control dilutions of liposomes containing no lipid A, endotoxin alone, and sterile saline alone were assayed at the same time. The gelation endpoints of io the 10-fold dilutions were used to calculate the minimal amount of lipid A that caused a positive Limulus test Video-enhanced contrast DIC microscopy. The ef- LIPID A PHOSPHATE (nmoles) fects of Limulus-positive and Limulus-negative lipoation of FIG. 1. Gelz Lmlssomes on intact amoebocytes were investigated. Diiaton liposome-assoc of Limulus lysate by lipld A and rect observations were obtained by the Allen videoiated lipid A. (A) Lipid A dissolved in enhanced contrast, differential interference contrast 0.5% triethylarmine is compared with 0.5% triethyl- (AVEC-DIC) microscopy method (1). DIC images amine alone. (1B) Liposomes containing the indicated using the Zeiss Axiomat at highest working aperture amounts of lipiid A per 0.1 ml (before dilution) were (1.3) and X/9 bias retardation were invisible to the eye, compared. In each case, before dilution, 0.1 ml of but yielded highly contrasted video images with a liposomes also contained 1,umol of DMPC, 0.75 p.mol Hamamatsu C chalnicon camera with gain and of CHOL, an d 0.11,umol of DCP. The horizontal offset adjustments driven by a computer frame memoa visual aid that indicates the level of ry (Hamamatsu) programmed to subtract mottle (field dashed line is activity of the control liposomes lacking lipid A. inhomogeneities) from the video image (R. D. Allen and N. S. Allen, submitted for publication). Amoebocytes were obtained by the cardiac punc ; Difco I ture of healthy limuli with pyrogen-free needles and duced by acetic saboraores Detr M s p syringes. One or two drops of fresh hemolymph were acid hydrolysis to remove the polysac- placed on an endotoxin-free coverslip (50 by 24 mm). charide chain ( x5); it contained 0.4 pmol of phosphate per mg of lipid One drop of the liposome preparation to be tested was A. The lipid A was rendered chloroform-soluble b~ placed adjacent to the hemolymph so that there was treatment with EDTA and was further contact purified by Bl between the two after a smaller igh-dyer extraction (5). The lipid A by 22 mm) was placed coverslip (22 on top. The cover glass sand- into eight distinct fractions having could be separnated different wich was Rfs by immediately sealed with melted thin-layer chromatography (TLC), and valap (1 part each of the frac petrolatum:1 part lanolin:1 part paraffin) and viewed at ctions was purified by preparative TLC room temperature. The concentration gradient of lipoefore (5). When lipid A or lipid A as described t fractions were somes that was produced by the above maneuver used in aqueous media, they were dried to remove allowed the chlo: identification of appropriate fields for,roform and solubilized in 0.5% triethyl- study. amine. Preparation Images were recorded on videotape with a c)fliposof es. Liposomes were prepared Sony 2610 video cassette recorder. Video-micrographic imof DMPC, CHOL, and DCP in molar ages were recorded with a Nikon 35-mm camera from a mixture ratios of 2/1.5 constituent, it wa0.22. When hepd A lcposome was a equipped with a 55-mm macro lens. Plus X film was was add form in the con iexposed at ASA 100 and developed for 6 min in a 1:3 centrations indicated in each individual dilution of Microdol-X. experiment. Th ie lipid A was dried along with the other lipids in pearfollowed by 1 1h under high vacuum in a desiccator. A Woods Hole, Mass. These observations were made in collaboration with shaped flasks on a rotary evaporator, Robert D. Allen at the Marine Biological Laboratory, small number of 0.5-mm glass beads was added, and the dried lipids were shaken with the aid of a Vortex mixer in a volume of 0.15 M NaCl such that the final aqueous dispersion of DMPC was 10 mm. Limulus amoebocyte lysate assay. Lyophilized Limulus amoebocyte lysate was purchased from Associates of Cape Cod (Woods Hole, Mass.). Endotoxin (Escherichia coli EC-2) was provided by the Bureau of Biologics, Food and Drug Administration, and was used as a positive reference control. Pyrogen-free water and sterile saline (0.9%) were obtained from Cutter Laboratories (Berkeley, Calif.). All glassware was made endotoxin free by treatment at 180 C in a RESULTS Effects of lipid A and liposomal lipid A on Limulus lysate activity. Lipid A was extremely effective in producing gelation of Limulus lysate (Fig. 1). Activity was detectable at a concentration of 0.25 x 10-6 nmol of lipid A phosphate per ml. Upon incorporation of lipid A into liposomes, the Limulus lysate gelation activity was markedly reduced (Fig. 1B). Each of the bars (Fig. IB)

3 VOL. 39, 1983 LIPID A, LIPOSOMES, AND LIMULUS AMOEBOCYTES 1387 CD, C-6 CONTROL I LIPID A FRACTION NUMBER FIG. 2. Limulus amoebocyte lysate gelation activities of individual lipid A fractions. In each case, the undiluted solution of the lipid A fraction contained 4 nmol of phosphate per 0.1 ml. The control consisted of a sham extract of gel from a TLC plate developed under conditions identical to those used for obtaining the individual lipid A fractions. The gel used approximated, but was not identical to, the amount of gel obtained during the purification of each fraction. represents a different liposome preparation having a density of lipid A that progressively increased from 0.25 to 4 nmol of lipid A phosphate per,umol of DMPC. The liposomes themselves (i.e., lacking lipid A) had a minor amount of activity that was detected at a dilution of At the lowest concentration of liposomal lipid A tested (0.25 nmol of lipid A), activity was detectable only at a dilution of Thus, lipid A in solution (detectable at a 10-6 dilution; Fig. 1A) had at least 100,000-fold higher Limulus lysate gelation activity than did the equivalent amount of lipid A present in the lowest density in liposomes (detectable only at a 10-1 dilution, Fig. 1B). In fact, liposomal lipid A at the lowest lipid A density had no greater activity than the control liposomes lacking lipid A (Fig. IB). Because of this, liposomes containing the lowest concentration of lipid A (0.25 nmol of phosphate) are referred to below as Limulus-negative liposomes. When the liposomal lipid A density was increased, the Limulus lysate gelation activity increased (Fig. 1B). Liposomes having the highest concentration of lipid A (4 nmol) had 1,000- fold-greater activity than did the control liposomes lacking lipid A (Fig. 1B). Because of this, liposomes containing 4 nmol of lipid A are referred to below as Limulus-positive liposomes. Activities of lipid A fractions. Each of eight purified fractions of lipid A was assayed for Limulus lysate gelation activity (Fig. 2). The activities were compared with that of a control in which an extraction was performed on the silica gel of a preparative TLC plate processed in a manner identical to that used to isolate the eight fractions. The control had a moderate level of background activity that was positive to a 10-4 dilution. Although there is no direct supporting evidence, it is possible that the silicic acid scraped from the blank plate as a control contained a small amount of endotoxin, and this may have accounted for the higher background. Seven of the eight fractions (fraction 1 was the exception) had activities greater than that of the control. Positive activities ranged from 10-fold (fraction 8) to 10,000-fold (fraction 5) greater than that of the control (Fig. 2). Upon incorporation in a high density into liposomes, the activity of each fraction, including the silica gel control, was diminished (Fig. 3). Fraction 1 still lacked activity, and fraction 8 (the least active of the seven active lipid A fractions) now also lacked activity compared with the control. Fractions 2 through 7 still showed 10-fold- to 1,000-fold-greater activity in the Limulus test as compared with the control (Fig. 3). Effects on intact amoebocytes. The amoebocytes of L. polyphemus normally contain a considerable number of large granules, as visualized by video-assisted phase-contrast microscopy (Fig. 4A). The granules packed the cytoplasm, but did not obscure the nucleus (Fig. 4A). When the amoebocytes were incubated with endotoxin (or with soluble lipid A), rapid degranulation occurred, leaving large patches of relatively clear cytoplasm (Fig. 4B). This degranulation, which also occurs more slowly by contact activation in the absence of endotoxin, apparently is not lethal to the cell, and degranulated amoebocytes possess residual amoeboid motility, as described elsewhere (4). Because of our ability to produce liposomes that were either unreactive (Limulus-negative) or reactive (Limulus-positive) in the amoebocyte lysate assay (Fig. 1B), we examined the effects of liposome-associated lipid A on intact amoebocytes. The liposomes were easily identified by video-assisted phase-contrast microscopy as distinctive crater-like structures (Fig. SA). The careful examination of numerous cells resulted in only one clear-cut example of the possible

4 1388 RICHARDSON ET AL. CDL 10-4 ~11 C-, 2Q- 100 LIO SOM LIPO SOME CONTROL LIPID A FRACTION NUMBER FIG. 3. Limulus lysate activities of liposomes containing individual lipid A fractions. In each case, before dilution, 0.1 ml of liposomes contained 1,imol of DMPC, 0.75,umol of CHOL, 0.11,umol of DCP, and 4 nmol of lipid A fraction phosphate. The control consisted of an appropriate volume of extract from the sham experiment described in the legend to Fig. 2. phagocytosis of a liposome by an amoebocyte (Fig. 5B, arrow). The incubation of Limulus-negative liposomes with amoebocytes did not result in the rapid degranulation usually observed with endotoxin or free lipid A. This was a subjective conclusion that was based on continuous observation. Small increases in the rate of normal contactactivated degranulation might not have been detected under the observation conditions. In contrast to the above, the incubation of Limulus-positive liposomes with amoebocytes resulted in a rapid and marked transformation in the appearance of the cells (Fig. 6). Long spiny needle-like projections or filaments, which were uniformly distributed, extended rapidly from the surface of every cell, and this morphological change was accompanied by rapid degranulation (Fig. 6). The projections covering the surfaces of the amoebocytes initially contained amoebocyte cytoplasm at their bases. This could be seen in some projections that had blunt cytoplasmic bases that tapered continuously into thin needlelike tips (Fig. 6). The spiculated appearance of Limulus amoe- INFECT. IMMUN. bocytes that is illustrated in Fig. 6 was observed only when the cells were incubated with Limu- Ius-positive liposomes containing lipid A. Liposomes alone or lipid A or whole endotoxin alone did not induce this distinctive cellular morphology, but the changes in the cellular appearance were similar to those commonly observed in the presence of gram-negative bacteria (6, 15). Although almost every amoebocyte underwent the change to a spiculated appearance, we did not detect even a single instance of the apparent ingestion of Limulus-positive liposomes such as that illustrated in Fig. SB. However, clumps of liposomes often were observed associated with the tips of spiny projections (Fig. 6, arrow). DISCUSSION Previously, it was found that the incorporation of lipid A into liposomes strongly abrogated the ability of lipid A to produce the gelation of Limulus amoebocyte lysate (14). This observation is extended in the present investigation. We have now found that the degree of activity in the Limulus test is dependent on the concentration or density of lipid A in the liposomes. Lowdensity lipid A, at a concentration that had strong activity outside liposomes, lacked activity after incorporation within liposomes. In contrast, lipid A that was present at a high density in liposomes retained a considerable amount of activity. The ability to produce Limulus-negative and Limulus-positive liposomes thus gave us a novel opportunity to examine the role of the phospholipid bilayer in controlling the activities of lipid A in other biological systems. The interaction of gram-negative bacteria with intact Limulus amoebocytes caused distinctive morphological changes in the cell that were first observed more than 20 years ago (6). These changes, which are part of the initial clotting process of L. polyphemus hemolymph, are characterized by the loss of granular content and the formation of long connecting extensions or processes by amoebocytes (6, 15). We have found in this study that the same distinctive spiculated appearance of amoebocytes that is caused by bacteria is also produced by the incubation of amoebocytes with Limulus-positive liposomes (Fig. 6). In our experience, the characteristic long filamentous processes that are induced by Limulus-positive liposomes or by gram-negative bacteria are not produced by the incubation of the cells either with soluble endotoxin, lipid A, or Limulus-negative liposomes. Our data suggest that Limulus-positive particles, whether the particles are bacteria or liposomes containing lipid A, may be recognized by amoebocytes that then respond by producing long extensions which form a netlike tangle of filaments. Limulus-negative liposomes have a quantity of lipid A

5 vx VOL. 39, 1983 A LIPID A, LIPOSOMES, AND LIMULUS AMOEBOCYTES 1389 FIG. 4. Video-assisted phase-contrast microscopy of intact Limulus amoebocytes. (A) Before degranulation. (B) After partial degranulation. Abbreviations: N, nucleus; G, granule; C, cytoplasm exposed as a clear area after degranulation. that, if it were present outside of liposomes, would rapidly gel Limulus lysate (Fig. 1). The inability of the lipid A of Limulus-negative liposomes to induce either the coagulation of lysate (Fig. 1B) or the characteristic filamentous processes in intact cells (Fig. 6) suggests that the active region of the lipid A can be completely embedded in the lipid bilayer of the liposome. This observation implies that the lipid A fatty acids, which would be expected to be associated with the hydrophobic regions of the phospholipid bilayers, might be responsible both for its ability to gel Limulus amoebocyte lysate and for the recognition of lipid A by intact amoebocytes. The concept of an active role for the fatty acids is supported by the disappearance of lysate activity after the removal of lipid A fatty acids (13, 16). A This study also demonstrates that lipid A is heterogeneous with respect to Limulus lysate gelation activity. The eight fractions of lipid A that were employed in this study were numbered according to increasing Rfs on TLC (5). There was a positive correlation between the Rf of the fraction and the quantity of esterified fatty acids (5). Individual lipid A fractions differed by as much as 10,000-fold in their Limulus gelation activities, but there was no correlation between coagulant activity and the number of esterified fatty acids. Fraction 1, which had the least number of fatty acids (5), had no detectable coagulant activity above the control (Fig. 2). In contrast, fraction 8, which has a very high quantity of esterified fatty acids, was the next weakest in the Limulus lysate assay (Fig. 2 and 3). This observation suggests that qualitative B - A. Pl. r -A.BW & FIG. 5. (A) Video-assisted phase-contrast microscopy of a clump of Limulus-negative liposomes. (B) Limulus amoebocytes after slow degranulation in the presence of Limulus-negative liposomes. The arrow indicates a single liposome that might have been ingested by the cell. The other circular structures in the clear cytoplasm are mitochondria.

6 1390 RICHARDSON ET AL. INFECT. IMMUN. FIG. 6. Appearance of Limulus amoebocytes after exposure to Limulus-positive liposomes. The arrow indicates a clump of liposomes. Such clumps of liposomes were commonly observed associated with the filamentous structures. differences may be more important than quantitative differences in the fatty acid content of lipid A fractions. Functional heterogeneity among lipid A fractions is consistent with previous observations. We have found heterogeneity of activity in every biological system in which individual lipid A fractions have been examined (2, 3, 5). On the basis of all of the above observations, it appears that intact amoebocytes have the ability to recognize liposomal lipid A, but only when the liposomal lipid A is at concentrations adequate to activate amoebocyte lysate. The cells respond to this recognition by producing long filamentous extensions that form a network between cells. The recognition process of amoebocytes for lipid A can be modulated by reducing the density of lipid A so that the liposome lacks lysate gelation activity. We have concluded that there is a good correlation between the cellular recognition of liposomal lipid A and the ability of the liposomes to gel amoebocyte lysate. However, the lack of the production of gelation of Limulus amoebocyte lysate and the absence of recognition by intact amoebocytes do not establish that lipid A is chemically absent from the liposomes. ACKNOWLEDGMENTS We thank Robert D. Allen, Department of Biological Sciences, Dartmouth College, Hanover, N.H., for making our microscopic studies possible. This work was supported in part by research grant HL from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md. LITERATURE CITED 1. Allen, R. D., N. S. Allen, and J. L. Travis Videoenhanced contrast differential interference contrast (AVEC-DIC) microscopy: a new method capable of analyzing microtubule-related motility in the reticulopodial network of Allogromia laticollaris. Cell Motility 1: Alving, C. R., B. Banerji, J. D. Clements, and R. L. Richards Adjuvanticity of lipid A and lipid A fractions in liposomes, p In B. H. Tom and H. R. Six (ed.), Liposomes and immunobiology. Elsevier/ North-Holland Publishing Co., New York. 3. Alving, C. R., B. Banerji, T. Shiba, S. Kotani, J. D. Clements, and R. L. Richards Liposomes as vehicles for vaccines. Prog. Clin. Biol. Res. 47: Armstrong, P. B Adhesion and spreading of Limulus blood cells on artificial surfaces. J. Cell. Sci. 44: Banerji, B., and C. R. Alving Lipid A from endotoxin: antigenic activities of purified fractions in liposomes. J. Immunol. 123: Bang, F. B A bacterial disease of Limulus polyphemus. Bull. Johns Hopkins Hosp. 98: Harris, N. S., and R. Feinstein A new Limulus assay for the detection of endotoxin. J. Trauma 17: Levin, J., and F. B. Bang The role of endotoxin in the extracellular coagulation of Limulus blood. Bull. Johns Hopkins Hosp. 115: Levin, J., and F. B. Bang Clottable protein in Limulus: its localization and kinetics of its coagulation by endotoxin. Thromb. Diath. Haemorrh. 19: Levin, J., T. E. Poore, N. S. Young, S. Margolis, N. P. Zauber, A. S. Townes, and W. R. Bell Gramnegative sepsis: detection of endotoxemia with the Limulus test. Ann. Int. Med. 76: Luderitz, O., C. Galanos, V. Lehmann, M. Nurminen, E. T. Rietschel, G. Rosenfelder, M. Simon, and 0. Westphal Lipid A: chemical structure and biological activity. J. Infect. Dis. 128(Suppl.):S17-S Nandan, R., and D. R. Brown An improved in vitro

7 VOL. 39, 1983 LIPID A, LIPOSOMES, AND LIMULUS AMOEBOCYTES 1391 pyrogen test: to detect picograms of endotoxin contamination in intravenous fluids using Limulus amoebocyte lysate. J. Lab. Clin. Med. 89: Niwa, M., T. Hiramatsu, and 0. Waguri The gelation reaction between horseshoe crab amoebocytes and endotoxins studied by the quantitative clot protein method. Jpn. J. Med. Sci. Biol. 27: Ramsey, R. B., M. B. Hamner, B. M. Alving, J. S. Finlayson, C. R. Alving, and B. L. Evatt Effects of lipid A and liposomes containing lipid A on platelet and fibrinogen production in rabbits. Blood 56: Shirodkar, M. V., A. Warwick, and F. B. Bang The in vitro reaction of Limulus amebocytes to bacteria. Biol. Bull. 118: Yin, E. T., C. Galanos, S. Kinsky, R. A. Bradshaw, S. Wessler, 0. Luderitz, and M. E. Sarmiento Picogram-sensitive assay for endotoxin: gelation of Limulus polvphemus blood cell lysate induced by purified lipopolysaccharides and lipid A from gram-negative bacteria. Biochim. Biophys. Acta 261: