Plasma lipopolysaccharide-binding protein is found associated with a particle containing apolipoprotein A-I, phospholipid, and factor H-related proteins.

Neutrophils exhibit a dramatic enhancement of integrin-mediated cell adhesion in response to lipopolysaccharide (LPS). This response requires CD14 on the neutrophil and plasma proteins in solution. We have purified the factor from plasma that facilitates the adhesive response of neutrophil to LPS by using a combination of affinity and ion-exchange chromatography. Previous work has shown that the activity is associated with apolipoprotein A-I (apoA-I), and here we show that this activity is associated with an apoA-I-bearing complex of protein and phospholipid. Native polyacrylamide gel electrophoresis (PAGE) analysis showed a ladder of bands in the Mr 200,000 region, and electron microscopy revealed round, indented particles of 11.4 ± 0.12 nm in diameter. Characterization of these particles revealed a density of 1.219-1.264 g/ml and ∼10 molecules of lipid phosphate per Mr 200,000 complex. SDS-PAGE showed that each of the bands seen in native PAGE was composed of several polypeptides. These were identified as apoA-I, LPS binding protein (LBP), and factor H-related proteins (FHRPs). Physical association of apoA-I, LBP, and FHRP in these particles was further confirmed using double immunodiffusion, and association of LBP and FHRP in plasma was confirmed by coimmunoprecipitation. FHRPs are the numerically dominant protein components in these particles, and all plasma FHRP-1 appears to be associated with these particles. We suggest that FHRPs may be the defining constituent of this novel “lipoprotein” particle.

Spontaneous binding of LPS to CD14 occurs only slowly but can be dramatically enhanced by LBP (10), a lipid transfer protein present in plasma (1). As a result, PMNs exhibit little if any response to LPS alone, but responses are strong and rapid in the presence of LBP (1,3). LBP acts catalytically to transfer single LPS molecules from LPS micelles to a binding site on CD14 (1). The catalytic action of LBP is consonant with its sequence similarity to cholesterol ester transfer protein and phospholipid transfer protein (11), well characterized plasma lipid transfer proteins. Cholesterol ester transfer protein and phospholipid transfer protein are both found on the surface of high density lipoprotein (HDL) particles (12,13), and recent observations from our laboratory indicate that LBP is associated with apoA-I, the principal protein of HDL (14).
There are many minor subclasses of HDL with distinct physical characteristics and specialized functions. Examples include apolipoprotein J-containing lipoproteins (15), trypanosome lytic factor (16 -18), and very high density lipoprotein (19 -21). Here we have purified the proteins from plasma that facilitate CD14-dependent responses of PMN to LPS. We found that this activity is borne on a very high density particle composed of phospholipids, apoA-I, LBP, and FH-related proteins 1 and 2 (FHRP-1 and FHRP-2). Additional proteins yet to be characterized are also present on the particle. The physical properties of this particle are described herein.

EXPERIMENTAL PROCEDURES
Reagents-Fresh frozen normal human plasma (citrated) was obtained from the New York Blood Center (New York, NY). HiPAC-Aldehyde chromatographic resin was obtained from Chromatochem (Missoula, MT). Fibrinogen (plasminogen-free human plasma fibrinogen) was purchased from Calbiochem (San Diego, CA). Recombinant LBP and rabbit polyclonal anti-recombinant LBP were as described (1). Monoclonal antibody against LBP (17G4) was a generous gift from Dr. David Emanuel (Indianapolis, IN). Polyclonal anti-FH antibody and anti-apoA-I antiserum was obtained from Incstar (Stillwater, MN). 3D11, a mouse monoclonal antibody that reacts with FH and FHRP-1 (22), was a generous gift from Dr. Vesa Koistinen (Helsinki, Finland). ReLPS was obtained from List Biological Laboratories (Campbell, CA). Goat anti-rabbit IgG and rabbit anti-mouse IgG conjugated with alkaline phosphatase were purchased from Bio-Rad, and rabbit anti-goat IgG was purchased from Incstar. Ouchterlony diffusion plates were purchased from The Binding Site (Birmingham, United Kingdom). Purified apoA-I and complement FH were purchased from Incstar and Calbiochem, respectively.
Chromatography-Four units of fresh frozen plasma were thawed at room temperature, pooled, aliquoted, and frozen at Ϫ70°C until analyzed further. To begin an experiment, aliquots were thawed at 37°C and immediately placed on ice. Subsequent manipulation of the plasma was done strictly either on ice or at low temperature (4°C). Plasma was spun in a Sorvall SS-34 rotor at 17,000 rpm for 20 min and passed through a Sephadex G-25 (Pharmacia Biotech Inc.) column prior to chromatography. HiPAC-Aldehyde resin was packed into a HR 5/20 or 10/30 FPLC column (Pharmacia) and equilibrated with starting buffer (Dulbecco's PBS without calcium and magnesium, 1 mM EDTA, and 0.02% sodium azide). For the scaled up procedure, 50 ml of plasma was loaded into the column at a rate of 2 ml/min and washed with starting buffer until the absorbance (A 280 ) came down to base line. Flow-through fractions were collected and stored for further analysis. Adsorbed material was eluted into ϳ20 ml with 0.5 M ammonium acetate buffer, pH 3.0, immediately subjected to dialysis against 20 mM Tris, pH 8.0 or 8.5 overnight, and loaded onto a Mono Q (10/10, Pharmacia) column at a rate of 2 ml/min. The column was washed with starting buffer until the absorbance (A 280 ) came down to base line and eluted with a gradient of NaCl from 0 to 1 M at 4 ml/min with a fraction size of 5 ml. Each fraction was tested for activity by the PMN adhesion assay described below. In some experiments, the active fractions were chromatographed again on the Mono Q HR 5/5 column to obtain purer and/or a more concentrated sample. Ceramic hydroxyapatite (Bio-Rad) was packed in a FPLC HR5/5 column (Pharmacia), and chromatography was done using 10 mM potassium phosphate buffer, pH 6.8, as starting buffer. The sample was loaded, washed with starting buffer, and eluted with a gradient of potassium phosphate from 10 -400 mM over 40 ml at a rate of 1 ml/min. One-ml fractions were collected.
All the results shown used frozen plasma. Identical choromatographic elution pattern, SDS, and native PAGE profile and Western blotting images were obtained using plasma from freshly drawn blood (never frozen; data not shown).
Response of PMN to LPS-The enhancement of leukocyte integrinmediated cell adhesion by LPS was measured using a two-step assay as described earlier (2). Briefly, freshly isolated PMN were labeled with 3-5 M of carboxyfluorescein diacetate, succinimidyl ester (Molecular Probes, Eugene, OR); PMNs were then incubated with stimuli for 10 min at 37°C. Stimuli usually consisted of a fixed dose of LPS (10 ng/ml) and varying doses of plasma or a fraction of plasma to be assayed. The PMNs were then washed and transferred to Terasaki plates coated with fibrinogen. Adhesion of the PMNs was determined by reading the fluorescence with a fluorescence plate reader (Cytofluor 2300; Millipore, Bedford, MA) before and after washing, and the percentage of PMNs adherent on the plate was calculated. All assays were done as triplicate.
Lipid Analysis-Lipid was extracted from the Mono Q-purified active sample with chloroform and methanol according to the method of Bligh and Dyer (23). Phosphate was determined in the extracted sample according to the method of Ames and Dubin (24). Concentrations of triglycerides and total cholesterol (cholesterol plus cholesterol ester) were determined by an enzymatic method using a commercial kit (Sigma).
Density Determination-The density of the particle was determined by ultracentrifugation (25). Mono Q-purified active samples were mixed with sodium bromide solutions of various densities to yield final densities of 1.006, 1.019, 1.065, 1.119, 1.219, and 1.264 g/ml. The samples were centrifuged in a Beckman Ti 42.2 rotor for 5 h at 4°C at 40,000 rpm, conditions calculated to bring lipoproteins to an equilibrium position (26). At the end of the spin, the top and bottom 30 l of the 230-l sample were analyzed for the presence of protein by BCA protein assay (Pierce). As a control, purified human plasma low density lipoprotein (density Ͻ1.060 g/ml) was analyzed in parallel.
Enzyme-linked Immunosorbent Assay-A sandwich assay for LBP was done as described previously (14). Briefly, plates were coated with anti-LBP monoclonal antibody 17G4 at 5 g/ml. After blocking with a dry milk solution, samples were then added and incubated for 1 h at room temperature or overnight at 4°C. After washing, bound proteins were detected using rabbit polyclonal antibody against LBP and secondary antibody conjugated with alkaline phosphatase. A fluorescent signal was generated using the fluorogenic substrate, Attophos (JBL Scientific, San Luis Obispo, CA) and measured using a Cytofluor 2300 (Millipore).
Electrophoresis-SDS (nonreducing) and native PAGE were done using either the PhastSystem (Pharmacia) or Novex system (Novex), according to the manufacturer's recommendation. Isoelectric focusing was done using the PhastSystem, according to the manufacturer's recommendation. The sample in Tris buffer was directly loaded on isoelectric focusing media 3-9 (Pharmacia), and pI was determined by comparison with reference standards (Pharmacia). For electroelution experiments, samples were electrophoresed on Novex 4 -20% gradient polyacrylamide gels under nondenaturing condition for 2.5 h at 120 V. Lanes of a gel were cut into eight equal pieces, and the pieces were placed individually in Centrilutor (Amicon, Beverly, MA) tubes and electroeluted according to the manufacturer's recommendations at 120 V for 2 h at 4°C. A parallel lane was stained with Coomassie Blue for reference. Two-dimensional PAGE was done with the PhastSystem. The first dimension was native PAGE with a 4 -15% gradient gel. After electrophoresis, the lane was cut out and boiled with SDS sample buffer for 5 min before being electrophoresed on a 10 -15% gradient gel for the second dimension. A parallel lane was stained with silver (PhastGel silver kit; Pharmacia).
Western Blot Analysis-After SDS-PAGE, samples were transferred to nitrocellulose or Immobilon (Millipore, Bedford, MA) membranes using the PhastSystem, or Novex systems, following the instructions of the manufacturers. After transfer, the membranes were blocked with 10% nonfat dry milk, 0.2% Tween 20 (Bio-Rad), and 0.02% sodium azide in PBS for 1 h at room temperature; washed with PBS containing 0.1% milk, 0.2% Tween 20, and 0.02% sodium azide (western buffer); and incubated with primary antibodies in Western buffer either 1 h at room temperature or 4°C overnight. Membranes were washed again and incubated with alkaline phosphatase-conjugated secondary antibodies diluted 1:1000 in Western buffer for 1 h at room temperature. After washing with distilled water, membranes were stained with nitroblue tetrazolium and bromochloroindolyl phosphate (Bio-Rad) in 100 mM Tris base, 100 mM NaCl, and 5 mM MgCl 2 , pH 9.5.
Amino Acid Sequencing-After SDS-PAGE, polypeptides were transferred to polyvinylidene difluoride membranes, briefly stained with Coomassie Blue, and washed with water. Bands were excised, and proteins were sequenced in a ABI 470A or ABI 477A instrument.
Electron Microscopy of the Particles-Electron microscopy of Mono Q-purified active samples was done by standard negative staining methods (27) with minor modification. Purified samples were concentrated with a Microcon centrifuge concentrator (Amicon, Beverly, MA), and buffer was exchanged at the same time into 0.125 M ammonium acetate and 0.35 mM Na-EDTA, pH 7.4. This sample was mixed 1:1 with 2% phosphotungstic acid neutralized with NaOH. Samples were loaded on Formvar carbon-coated grids, wicked off, and allowed to air dry. Samples were examined with a JEOL 100CX transmission electron microscope.
Double Immunodiffusion and Immunoprecipitation-Double immunodiffusion was done by incubating samples and antibodies in Ouchterlony plates (Binding Site) at 4°C overnight to 24 h. Plates were washed extensively with PBS and stained with Coomassie Blue. Immunoprecipitation was done with antibodies coupled to Affi-Prep Hz support (Bio-Rad). Antibodies (goat anti-FH and rabbit anti-LBP) were coupled to the support according to the manufacturer's instruction. Nonimmune antibodies (goat anti-rabbit immunoglobulin and rabbit anti-goat immunoglobulin; Incstar) were coupled in parallel as controls. Antibodycoupled beads (10 l) were incubated with freshly drawn plasma (500 l, diluted 1:1 with citrated PBS) at 4°C for 1 h with gentle mixing by rotator and washed 5-7 times with 1 ml of PBS; adsorbed proteins were eluted in SDS sample buffer.

RESULTS
Incubation of PMN with 10 ng/ml LPS alone results in no enhancement of integrin-mediated adhesion. However, the addition of normal human plasma (NHP) enables a strong LPSand CD14-dependent adhesion response (1)(2)(3)(4), and the dose dependence of this effect of plasma is shown in Fig. 1B. We sought the plasma factors that mediate this response to LPS and have found a chromatographic resin (HiPAC-Aldehyde) that can purify the activity from NHP. NHP (5 ml) was loaded onto the HiPAC-Aldehyde column (HR 5/20), and the flowthrough fraction was assayed for the ability to promote the adhesion of PMN to fibrinogen by LPS. Although it contained 99% of plasma proteins, it was completely devoid of activity. The column was eluted with low pH buffer, and more than 90% of the original activity was recovered in the eluate ( Fig. 1 and Table I). When this process was scaled up to utilize 50 ml of NHP on a HR 10/30 column somewhat less activity (65%) was recovered. The eluate of the HiPAC-Aldehyde column was further fractionated by ion-exchange chromatography on Mono Q. Activity was recovered as a single peak, at about 150 mM NaCl (Fig. 2). Active fractions were pooled and analyzed. We confirmed that this purified fraction stimulated cells in an LPSdependent fashion and that the activity was completely blocked by monoclonal antibody against CD14 (Fig. 2C). Table I shows that about 700-fold enrichment was achieved using two steps of chromatography.
Further chromatography on ceramic hydroxyapatite yielded a column profile with a single broad peak of protein and biological activity, and these peaks coincided closely (Fig. 3). SDS-PAGE gel profiles of the material in this peak appeared identical to those of the starting material (data not shown). This observation and additional analytical data described below suggest that the preparation obtained from the Mono Q column is relatively homogeneous.
On native PAGE gradient gels, the active purified fraction from plasma resolved into four or more discrete, evenly spaced bands in the interval from M r 150,000 -250,000 (Fig. 4B). A small and variable amount of protein was also observed in the M r ϳ60,000 region. To verify that the species represented in the M r 150,000 -250,000 region of native gels are active in enabling responses of PMN to LPS, a native gel was run and cut into eight pieces; then electroelution of each gel piece was done as described under "Experimental Procedures." Assay of each eluted sample showed that all of the activity fell in the M r 200,000 region (Fig. 5), confirming that activity was associated with species in this region. Isoelectric focusing analysis of the purified fraction showed a tight ladder of six to seven bands with pI of 6.5-7.3 (Fig. 4B). The simplicity of its isoelectric focusing and native PAGE profiles suggests that our preparation is relatively homogeneous. Electron microscopy of negatively stained samples revealed particles of discoid shape with an indentation in the center, some forming rouleaux (Fig. 4A). Measurement of 100 particles revealed an average diameter of 11.4 Ϯ 0.12 nm. This size is consistent with the molecular weight estimation from native gels.
Our previous studies showed that the factor in plasma necessary for enabling responses of PMN to LPS is associated with FIG. 1. HiPAC-Aldehyde purification of the plasma factor that mediates PMN adhesion in response to LPS. NHP (8 ml) was applied to a HiPAC-Aldehyde HR 5/20 FPLC column and eluted as described under "Experimental Procedures." Fractions were tested for the ability to mediate PMN adhesion in response to LPS. In A, a fixed dose of 2% NHP (P), HiPAC-Aldehyde column eluate (E), or flow-through (F) was assayed in the presence or absence of 10 ng/ml LPS. Background adhesion with buffer alone (B) is also shown. In B, the dose dependence of the response mediated by 10 ng/ml LPS and starting NHP (f), eluate (q), and flow-through (å) was measured after equivalent dilutions. Each data point is the average of three samples; bars, S.D. This result was seen in more than five experiments.  2. Mono Q purification of the plasma factor that mediates PMN adhesion in response to LPS. In A, the eluate from a HiPAC-Aldehyde column was loaded onto a Mono Q column after buffer exchange (into 20 mM Tris, pH 8.0). Elution was done using a NaCl gradient up to 1 M. In B, the histogram shows the results of a PMN adhesion assay performed with pooled pairs of fractions. Very little protein and no activity appeared in flow-through fractions. More than 90% of the activity applied was recovered from fractions 17-20. In the assay shown, the maximum adhesion with 2% NHP was 23.9 Ϯ 2.3% and with buffer alone was 2.0 Ϯ 0.6%. In C, active fractions from Mono Q were pooled and analyzed on a PMN adhesion assay in the presence or absence of monoclonal anti-CD14 antibody, 3C10. Each column represents: 1, no antibody; 2, 3C10, 3, 3G8, monoclonal anti-Fc␥III antibody; and 4, without active Mono Q fractions. Anti-CD14 antibody, 3C10, completely blocked the activity of active Mono Q fractions. apoA-I (14), and we thus sought evidence that our preparation contains lipids. Lipids were extracted, and phosphate analysis of the extract showed 13.8 nmol of phosphate per 318 g of protein. Measurement by enzymatic assay revealed, to our surprise, no detectable cholesterol or cholesterol ester (n ϭ 3). Our assays would have detected as little as 1.7% cholesterol of total protein weight. Similarly, triglycerides were either not detected (n ϭ 2) or were at the limit of detection (n ϭ 1), suggesting that the particles contain less than 2.7% triglycerides of total protein weight. The density of our particles was determined as between 1.219 and 1.264 g/ml. This high density and high protein:lipid ratio of our particle is observed in very high density lipoprotein particles.
SDS-PAGE showed that our preparation contained several protein bands (Fig. 4B). These include a triplet of bands near M r 85,000 (tp85), bands of M r 60,000 and M r 50,000 (p60, p50), a M r 38,000/35,000 doublet (p38/35), a M r 30,000/28,000 doublet (p30/28), and a M r 27,000 band (p27). The M r 30,000/ 28,000 bands were seen in variable amounts, depending on the source of NHP. All of the other bands were reproducibly obtained in more than 10 separate preparations, although some minor variation was noted in relative intensity of each band. Electroeluted samples from the M r ϳ200,000 area of a native gel gave rise to an identical SDS gel pattern, except for the absence of p50 (Fig. 5, inset), indicating that each of the bands seen in the SDS-PAGE derive from M r 200,000 particles except p50. Additionally, two-dimensional gel analysis, with native PAGE in the first dimension and SDS-PAGE in the second, showed that each of the native gel species in the M r 150,000 -250,000 range gave rise to an almost identical pattern of polypeptides in the second dimension (Fig. 6).
Thus, we have isolated a complex of lipid and protein that mediates responses of PMN to LPS. The properties of the complex are most similar to very high density lipoprotein. We have identified most of the polypeptides by a combination of immunochemistry and protein sequencing.
LBP-LBP is known to enable PMN to respond to LPS (3), and the presence of LBP in purified fractions was explored by Western blot analysis with antibodies raised against recombinant LBP. A M r 60,000 band was detected in a position identical with that of purified recombinant LBP (data not shown) and with p60 of our sample (Fig. 7). An enzyme-linked immunosorbent assay for LBP showed that there were 12 g of LBP in a 150 g/ml preparation of the particle. This finding indicates that 8% of the protein in our preparation is LBP and suggests that LBP comprises the M r 60,000 band in our preparation. Enzyme-linked immunosorbent assay analysis further showed that LBP was only found in the active fractions from the HiPAC-Aldehyde and Mono Q columns.
Apolipoprotein A-I-Recent experiments from our laboratory have shown that LBP and its biological activity are quantitatively retained on columns of anti-apoA-I (14). We, therefore, used Western blot analysis to determine if apoA-I was present in our purified fraction. A M r 27,000 band (p27), which corresponds to the size of apoA-I, was recognized by anti-apoA-I, both in the fractions purified from the Mono Q column (Fig. 7) and in the samples electroeluted from the high molecular weight area of native gels (data not shown). These results suggest that p27 in our preparation is apoA-I.
Factor H-related Proteins-p38/35 and p30/28 were characterized by N-terminal amino acid sequencing (Table II). The N-terminal sequence of both bands at p38/35 showed identity to the N terminus of FHRP-1, and the N-terminal sequence of p30/28 showed identity to that of FHRP-2. FHRPs are a family of abundant plasma proteins with unknown function, purified and cloned based on homology to complement FH (28). Both FHRP-1 and FHRP-2 run on SDS gels as doublets, and the doublets are thought to represent glycoforms (29,30). Several additional results confirm that p38/35 and p30/28 in our preparation are identical to FHRP-1 and FHRP-2, respectively. (a) They show gel behavior identical to that reported for FHRP-1 and FHRP-2 (29). (b) Polyclonal anti-FH, which is known to cross-react with FHRP-1 and FHRP-2, recognized p38/35 and p32/30 in Western blot (data not shown). (c) A monoclonal antibody, 3D11, directed against the C-terminal region of FH, has been shown to cross-react with FHRP-1 (22,31). We have FIG. 3. Hydroxyapatite purification of the plasma fractions that mediate PMN adhesion in response to LPS. Purified active samples from Mono Q were loaded onto ceramic hydroxyapatite packed in a FPLC (HR 5/5) column and equilibrated in potassium phosphate buffer (10 mM, pH 6.8). Neither protein nor activity appeared in the flow-through fractions (data not shown). Elution was performed with a linear gradient of phosphate from 10 mM up to 400 mM. Total protein (q) was measured from pooled pairs of fractions by the BCA protein assay (Pierce). All the protein (99%) was eluted at 70 -210 mM phosphate. Each pooled fraction was diluted 1:64 in phosphate-buffered saline and tested in the PMN adhesion assay (Ⅺ). In this assay, buffer alone yielded 4.8 Ϯ 1.2% adhesion.
Additional observations support the conclusion that FHRP-1 and FHRP-2 are associated with the particles described here. Western blots of plasma before and after chromatography on HiPAC-Aldehyde column showed that all of the FHRP-1 in plasma was retained in the column and was eluted with ammonium acetate, pH 3.0 (Fig. 8A). Western blot analysis of the entire Mono Q profile probed with anti-FH antibody showed bands consistent with FHRP-1 only in the fractions with biological activity (Fig. 8B). This finding suggests that FHRP-1 is quantitatively associated with the active species described here and may be a marker for this particle. The published purification of FHRP from NHP (29) used sequential chromatography and lead, at the penultimate step, to a preparation with an SDS profile identical to that shown in Fig. 4C. This result suggests that FHRP-1 is stably associated with a distinct set of additional proteins in plasma and is consistent with the hypothesis that FHRPs and the other proteins in Fig. 4C are part of a complex or particle.
To confirm the physical association of LBP, apoA-I, and FHRPs, we performed double immunodiffusion (Ouchterlony) analysis. Polyclonal antibodies against apoA-I, FH, and LBP were allowed to diffuse against particles purified by HiPAC-Aldehyde chromatography and Mono Q (Fig. 9A). A single immunoprecipitin line was observed with anti-LBP (well a) and that line fused with the anti-apoA-I line (well c). The anti-LBP line did not spur across the anti-apoA-I line, indicating that LBP is associated with apoA-I. The anti-apoA-I line did spur across the anti-LBP line in some gels, suggesting that some of the apoA-I is not associated with LBP. Similarly, the LBP line did not spur across the FHRP immunoprecipitin line (well b), indicating that LBP is associated with FHRPs. The FHRP line did spur across both the LBP line and the apoA-I line, suggesting that some of the FHRPs are not associated with LBP and some are not associated with apoA-I. This finding is in keeping with the observation that FHRP-1 is more abundant than LBP or apoA-I in the preparations. Additional studies with a mixture of purified apoA-I, LBP, and FH showed that spurs in all combinations were readily seen (data not shown), confirming the sensitivity of the procedure. These studies thus confirm the physical connection of LBP, apoA-I, and FHRPs in our particles. The association of LBP and FHRP in plasma was further confirmed by co-immunoprecipitation of proteins from fresh plasma. Fig. 9B shows that anti-FH precipitated not only FHRP-1 but also LBP from fresh plasma.

DISCUSSION
FHRPs were discovered as proteins with sequence homology and antigenic cross-reactivity with complement FH (28). At the RNA level, at least six distinct transcripts with homology to FH can be detected by Northern blot analysis of human liver RNA (32). Thus far, three (FHRP-1, FHRP-2, and FHRP-3) have been cloned (30,(32)(33)(34). At the protein level, anti-FH antibody detects at least 10 bands in addition to FH in Western blots of plasma (28). Thus far, message has been linked to protein bands only for FHRP-1 and FHRP-2. No function has been described for FHRPs. FHRP-1 consists of five tandem repeats of a 60-amino acid motif known as the short consensus repeat. This motif is also found in complement regulatory proteins such as FH, complement receptor 1, and C4b-binding protein, in the adhesion proteins known as selectins, in the LPS-binding protein of horseshoe crab known as factor C (35), as well as in several other proteins (36). Published data have not described an association of FHRPs with lipoprotein, but a related protein composed of six short consensus repeats known as ␤2-glycoprotein I (also called apolipoprotein H) is known to associate both with HDL particles and with phospholipids (37,38). Here we show that FHRP-1 and FHRP-2 are associated with a complex of phospholipid and other proteins in plasma and that this complex mediates responses of cells to LPS. Our findings both indicate the form of circulating FHRPs and suggest a role for FHRP in carrying and/or regulating the function of LBP.
FHRP-1 is the dominant protein component of the particle described here. FHRP-1 appears severalfold more abundant than either apoA-I or LBP, both in silver-stained (Fig. 4C) or Coomassie-stained (data not shown) gels. The relative abundance of FHRPs over LBP can also be inferred from the finding that essentially all plasma LBP (data not shown) and FHRPs (Fig. 8) are found in these particles, but the reported plasma level for FHRP-1 (40 g/ml; Ref. 29) is 8-fold greater than that reported for LBP (ϳ5 g/ml; Ref. 39). Additionally, Ouchterlony double diffusion analysis suggests that although all LBPs and all apoA-I in the purified particles are associated with FHRPs, not all of the FHRPs are associated with either LBP or apoA-I (Fig. 9). This observation is most compatible with the hypothesis that LBP and apoA-I are present on a subpopulation of particles that share FHRPs as a common constituent. Since FHRPs are the dominant species in these particles and since FHRPs are found only in these particles, we will refer to them as "FHRP-associated lipoprotein particles" (FALPs).
FALPs represent a very small proportion of plasma lipopro- FIG. 4. Electron microscopy and electrophoretic analysis of purified particles. A, Mono Q-purified active samples were prepared for negative staining in electron microscopy as described under "Experimental Procedures." Uniform, round, indented particles were seen (white arrow). Particles were seen to stack and form rouleaux (at 3 o'clock, black arrows). Bar, 75.7 nm. Inset, enlarged images of single particles. In B, the Mono Q-purified fraction was analyzed by native PAGE, isoelectric focusing, and SDS (nonreducing)-PAGE as described under "Experimental Procedures." Silver-stained gels are shown.
tein. By assuming a 30% yield in purification, we calculate that these particles contain no more than 0.9% of plasma phospholipid and 0.7% of plasma apoA-I. They are thus a minor subpopulation of lipoproteins and unlikely to play a large role in the transport of bulk nutrients. FALPs also exhibit considerable structural heterogeneity. This is best seen in native PAGE separations, which reveal several distinct species (Fig. 4B). The structural basis of this heterogeneity is not clear at this time. It is important to point out that we have not identified the triplet of bands at M r 85,000, and these relatively abundant proteins may play an important role in the structure and heterogeneity of FALPs.
The precise function of FALPs is not clear at this time, but the association with LBPs suggest that FALP proteins could affect LBP action. Interaction of LBP with plasma proteins and modulation of LBP action by plasma proteins are suggested by previous work. We have shown that plasma enables responses to LPS with characteristics different from those of purified LBP. Under defined conditions, an LPS-dependent response mediated by certain chromatographic fractions of plasma can be strongly enhanced by the addition of other chromatographic fractions, suggesting that multiple components may partici-pate in presentation of LPS to cells in plasma (40). Moreover, addition of the protease inhibitor Pefabloc SC strongly blocks LPS activation of PMNs mediated by plasma, whereas activation mediated by purified LBP is not affected by this compound (40). These observations prompted us to suggest that in whole plasma, LPS is presented to cell surface CD14 by the combined action of several proteins and proposed the name "septin" to describe this activity. The results described here support the close interaction of LBP with other proteins and suggest that components of FALPs are candidates that may serve to regulate its activity. This suggestion is strengthened by preliminary studies indicating that Pefabloc SC blocks the ability of FALPs to enable a response of PMNs to LPS (data not shown). On the other hand, it is now clear that the activity we termed septin may result from the combined activity of soluble CD14, lipoprotein particles, and FALPs. Since these individual contributors have pre-existing names and functions not uniquely related to   TP38  EATFXDFPKINHGIL  TP35  EATFXDFPKINHGIL  FHRP-2  EAMFCDFPKINHGIL  TP30  EAMFXDFPKINHGIL  TP28 EAMFXDFPKINHGIL a X represents an undetermined amino acid.
FIG. 5. Protein and biological activity comigrate in native PAGE. Mono Q-purified active samples were separated by native PAGE (8 -16% gradient gel). At the end of the run, the gel was cut into slices, electroeluted, and tested for activity in the PMN adhesion assay. Left panel, a Coomassie Blue-stained parallel lane. Right panel, the assay results from corresponding slices of the gel. In this experiment, 2% NHP yielded 18.4 Ϯ 1.9%, and buffer yielded 5 Ϯ 1.8% adhesion. Bars, S.D. SDS-PAGE of fraction 2 (inset) showed a pattern identical to the starting sample except p50 is missing (see Fig.  4B).
FIG. 6. Two-dimensional gel analysis of purified particles shows that the particles are composed of multiple polypeptides. Purified particles from Mono Q were subjected to two-dimensional gel analysis as described under "Experimental Procedures." The first dimension was native PAGE, and the second dimension was SDS (nonreducing)-PAGE. Gels were stained with silver. The positions of molecular weight markers (in thousands) are shown. A parallel lane in the first dimension native gel was aligned at the top of the second dimension gel. Each band at high molecular weight (M r ϳ200,000) in the native gel was separated into multiple bands in the SDS gel. The low molecular weight (M r ϳ65,000) band in the native gel ran at the position of p50 in the SDS gel. endotoxin and sepsis, we believe term "septin" is unnecessary and have discontinued its use.
In summary, we have purified a novel complex of protein and phospholipid that bears LBPs and FHRPs. The function of this particle is currently under study. FIG. 8. FHRP-1 fractionates with active particles. In A, 10% NHP (P), HiPAC-Aldehyde flow-through (F), and HiPAC-Aldehyde eluate (E) were electrophoresed, blotted, and probed with monoclonal antibody 3D11. Heavy staining in the M r 150,000 region represents plasma FH. In B, the eluate of the HiPAC-Aldehyde column was fractionated on Mono Q, and fractions were probed with polyclonal anti-FH antibody in Western blots. Numbers on the top reflect the odd fraction number of the two fractions pooled. Activity in the PMN adhesion assay is found in fractions 17-20, as described in Fig. 2.   FIG. 9. Physical association of LBP and FHRP or apoA-I. A, Ouchterlony analysis of active particles. Polyclonal antibody against recombinant LBP (purified IgG, 2 mg/ml) (a), FHRP (purified IgG, 2 mg/ml) (b); and apoA-I (antiserum) (c) were loaded on an agarose gel; purified active particles by HiPAC-Aldehyde and Mono Q chromatography were loaded in the center well, and the plate was incubated for 24 h at 4°C. The gel was washed with PBS and stained with Coomassie Blue. In B, fresh plasma was incubated with beads coupled to anti-FHRP (anti-FH) along with controls, as described under "Experimental Procedures." The result of the SDS-PAGE (nonreducing) of immunoprecipitates was transferred and probed with anti-FHRP (left panel) and anti-LBP (right panel). Lanes 1 and 3, samples from anti-FHRP; lanes 2 and 4, samples from control goat immunoglobulin. Lane 5 is purified plasma LBP used as a positive control for Western blot.