A Family of Lipopolysaccharide Binding Proteins Involved in Responses to Gram-negative Sepsis*

The lipopolysaccharides (LPS) of Gram-negative bacteria initiate potentially fatal processes in many host organisms. Recently published amino acid se- quence data suggest that there is a family of LPS binding proteins that may participate in the host response to Gram-negative bacteremia. The first two members of the family to be identified are an LPS binding protein present in serum after an acute phase response in humans, mice, rabbits, and rats and a bac- tericidal/permeability increasing protein present in the primary granules of human and rabbit neutrophils. LPS binding protein and bactericidal/permeability in- creasing protein share an ability to bind to LPS, have homologous NHz-terminal amino acid sequences, and are immunologically cross-reactive. Nevertheless, these two molecules differ in their effects on LPS and Gram-negative bacteria, in their sites of biosynthesis, and localization in vivo. The lipopolysaccharides (LPS)’ of Gram-negative bacteria initiate injurious and potentially fatal processes in animals and man. The observation that organisms as venerable as the horseshoe crab (1) possess protective mechanisms against Gram-negative bacteremia suggests that there has been con-siderable time for anti-LPS mechanisms to evolve. Thus it would not be surprising to discover a family of proteins possessing LPS, and more specifically lipid A, binding sites, but with different protective functions. We report here that lipopolysaccharide binding protein (LBP) (2) and bacteri-cidal/permeability

The lipopolysaccharides (LPS) of Gram-negative bacteria initiate potentially fatal processes in many host organisms. Recently published amino acid sequence data suggest that there is a family of LPS binding proteins that may participate in the host response to Gram-negative bacteremia. The first two members of the family to be identified are an LPS binding protein present in serum after an acute phase response in humans, mice, rabbits, and rats and a bactericidal/permeability increasing protein present in the primary granules of human and rabbit neutrophils. LPS binding protein and bactericidal/permeability increasing protein share an ability to bind to LPS, have homologous NHz-terminal amino acid sequences, and are immunologically cross-reactive. Nevertheless, these two molecules differ in their effects on LPS and Gram-negative bacteria, in their sites of biosynthesis, and localization in vivo.
The lipopolysaccharides (LPS)' of Gram-negative bacteria initiate injurious and potentially fatal processes in animals and man. The observation that organisms as venerable as the horseshoe crab (1) possess protective mechanisms against Gram-negative bacteremia suggests that there has been considerable time for anti-LPS mechanisms to evolve. Thus it would not be surprising to discover a family of proteins possessing LPS, and more specifically lipid A, binding sites, but with different protective functions. We report here that lipopolysaccharide binding protein (LBP) (2) and bactericidal/permeability increasing protein (BPI) (3) are the first two members of such a protein family to be recognized. The two proteins share the ability to bind to the lipid A region of LPS, display a high degree of NHz-terminal sequence homology, and are immunologically cross-reactive. Despite these similarities, the cellular origin, localization, and function of the two proteins are clearly different (2-4, 7, 9-12), although * This work was supported by United States Public Health Service Grants A115136 and GM37696 and Department of the Army Contract DAMD-88-C-8017. This is publication 5250 IMM from the Research Institute of Scripps Clinic. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ TO whom correspondence should be addressed Dept. of Immunology, Research Institute of Scripps Clinic, 10666 North Torrey Pines Rd., La Jolla, CA 92037.
both function in the complex host response to Gram-negative septicemia and endotoxemia.

MATERIALS AND METHODS
Purified LBP (2) and rabbit neutrophil granules (4, 5) were prepared by published procedures; BPI was partially purified by extraction from granules with sodium acetate (3). Electroblotting (6), the bactericidal assay using BPI (3), the protein-LPS binding assay in serum (7), and amino acid sequence similarity searches (8) were performed as described previously. Immunofluorescence studies of rabbit neutrophils were as described (9) using a goat anti-rabbit LBP to stain the elicited rabbit neutrophils.

RESULTS AND DISCUSSION
We have recently described the isolation and characterization of LBP from acute phase rabbit serum (2). The 60-kDa glycoprotein is estimated to be present in normal rabbit serum at 5100 ng/ml, while the concentration of LBP in acute phase serum rises to 10-50 pg/ml 24 h after the induction of an acute phase response (2). LBP is also present in the acute phase sera of humans, rats, and mice (2). Although LBP was discovered as a result of its ability to bind to LPS from the Re595 rough mutant of Salmonella minnesota, subsequent work (10) has established that LBP binds to many types of rough or smooth forms of LPS as well as to lipid A with association constants of at least los M-'. The work of Elsbach and co-workers (11) has also resulted in the recognition of another LPS-binding protein known as BPI. This is a protein of either 50 kDa (11, rabbit) or 60 kDa (3, human) and has been isolated from rabbit and human neutrophils where it is tightly bound in the membranes of myeloperoxidase containing primary granules (9). Isolated BPI binds avidly to LPS in the outer membrane of Gram-negative bacteria and rapidly kills these organisms (12).
With the publication of NHz-terminal sequences for both LBP (2) and BPI (3), the relationship between the two proteins became apparent. The 20 amino-terminal residues of human BPI (3) and the first 20 residues of rabbit LBP (2) are shown in Fig. 1. Eleven of the 20 residues are identical, with five others being chemically conservative substitutions that could be accomplished by single base changes in the DNA. Given this degree of homology, we asked whether the two proteins might be immunologically cross-reactive. To answer this question, sodium dodecyl sulfate-polyacrylamide gel electrophoresis gels of whole rabbit neutrophil granules (4,5) and purified LBP (2) were electroblotted to nitrocellulose membranes and probed with an anti-LBP antiserum (6). The results in Fig. 2 show that the granules have two components reactive with anti-LBP serum. The electrophoretic mobility of one is near that expected for rabbit BPI (50 kDa, 11); this component is most likely BPI. Ooi et al.
(3) have also described a 25-30-kDa amino-terminal fragment of rabbit BPI that retains BPI activity. The second component of rabbit neutrophil granules reactive with anti-LBP serum has the electrophoretic mobility expected for this fragment. Applying im-  munofluorescence techniques with permeabilized whole rabbit neutrophils and anti-LBP the cross-reactive material appeared to be intracellular and punctate (Fig. 3) as reported for human BPI (9). Despite the sequence and immunologic similarities, the two proteins originate in different cells. BPI is a specific product of the neutrophil that is synthesized at or around the promyelocyte stage of differentiation (9). On the other hand, LBP synthesis can be detected in rabbit liver and explanted rabbit hepatocytes after an acute phase stimulus.2 Given the structural and immunologic similarity of LBP and BPI, we next compared the functional properties of the two molecules. BPI has two activities, killing and permeabilizing Gram-negative bacteria (12). We tested the ability of LBP to kill two strains of Gram-negative bacteria as described by Elsbach et al. (11). As shown in Table I, purified LBP had no effect on the viability of these organisms at concentrations where purified BPI is very effective. In other experiments with as much as 240 pg/ml LBP, no bactericidal effects were seen. It may be that the carbohydrate present on LBP prevents expression of antibacterial activity. Experiments to prepare deglycosylated LBP for testing the role of the carbohydrate moiety are underway.

HumanBPI
A well characterized activity of LBP is to bind to LPS in acute phase serum and, with the LPS isolated from S. minnesotu Re595, to slow the binding of LPS to high density lipoprotein (7). To determine if this property of LBP is shared with BPI, an extract prepared from rabbit neutrophil granules was added to normal rabbit serum, and the kinetics of binding to high density lipoprotein of added 3H-Re595 LPS were tested as described previously (7). As seen in Fig. 4, the addition of granule extract to normal rabbit serum altered the buoyant density of Re595 LPS from 1.33 (7)  in contrast to LBP.LPS complexes the BPI.LPS complex appeared to be stable in the presence of high density lipoprotein, since there was no progressive transfer of LPS to a form of density ~1 . 2 g/ml as shown in Fig. 4. Thus the functional similarity of the two proteins appears to be limited to the ability to bind LPS.
There are a number of other proteins for which a lipid A binding site has either been demonstrated or can reasonably be inferred. These are listed in Table 11. With the exception of one of the limulus proteins (13) and the @-subunit of CR3, LFA-1, and p150,95 (14), which show no similarity to the available partial sequences of BPR and CBP, amino acid sequence data for these proteins are not yet available. Thus evaluation of homologies must be deferred until the complete sequences of LBP, BPI, and the other proteins listed in Table  I1 are fully delineated. Nevertheless, the amino acid sequence homology, immunologic cross-reactivity, and most importantly the presence of an LPS binding site in LBP and BPI demonstrate a unique relationship not previously appreciated. The biological role of LBP is unknown at present. Nevertheless, it appears that binding of LPS to LBP alters the endotoxic properties of LPS in ways that impact the host response to Gram-negative infection and endotoxemia. We have recently shown that LPS.LBP complexes prime the oxidative burst response of human neutrophils to formylated peptides much more rapidly than LPS alone (15). One can imagine several scenarios in which the effects of LBP and BPI would at least be additive in a concerted response to Gram-negative bacteremia. For example, bacteria in a developing septic focus could activate macrophages (16) to induce acute phase reactant synthesis in hepatocytes. The LBP . LPS complexes could enhance the priming of neutrophils to kill bacteria by both BPI and oxidative means.
Both LBP and BPI are inducible proteins; the former in hepatocytes* and the latter in the promyelocytic HL-60 cell line differentiated to the neutrophil lineage (9). The delineation of the signals that regulate the expression of these proteins, the characterization of the cis-acting elements that control transcription, and the determination of the DNA sequence of the genes for both LBP and BPI will provide new information that may aid in the identification of other previously unrecognized members of this family of proteins. Finally, the ability to study proteins that have a lipid A binding site may provide new insights into how bacterial lipopolysaccharides interact with mammalian cells and proteins to initiate changes that lead to fatal endotoxic shock.