Apolipoprotein CI enhances the biological response to LPS via the CD14/TLR4 pathway by LPS-binding elements in both its N- and C-terminal helix.

Timely sensing of lipopolysaccharide (LPS) is critical for the host to fight invading Gram-negative bacteria. We recently showed that apolipoprotein CI (apoCI) (apoCI1-57) avidly binds to LPS, involving an LPS-binding motif (apoCI48-54), and thereby enhances the LPS-induced inflammatory response. Our current aim was to further elucidate the structure and function relationship of apoCI with respect to its LPS-modulating characteristics and to unravel the mechanism by which apoCI enhances the biological activity of LPS. We designed and generated N- and C-terminal apoCI-derived peptides containing varying numbers of alternating cationic/hydrophobic motifs. ApoCI1-38, apoCI1-30, and apoCI35-57 were able to bind LPS, whereas apoCI1-23 and apoCI46-57 did not bind LPS. In line with their LPS-binding characteristics, apoCI1-38, apoCI1-30, and apoCI35-57 prolonged the serum residence of 125I-LPS by reducing its association with the liver. Accordingly, both apoCI1-30 and apoCI35-57 enhanced the LPS-induced TNFalpha response in vitro (RAW 264.7 macrophages) and in vivo (C57Bl/6 mice). Additional in vitro studies showed that the stimulating effect of apoCI on the LPS response resembles that of LPS-binding protein (LBP) and depends on CD14/ Toll-like receptor 4 signaling. We conclude that apoCI contains structural elements in both its N-terminal and C-terminal helix to bind LPS and to enhance the proinflammatory response toward LPS via a mechanism similar to LBP.

ApoCI has been demonstrated to act as a negative acutephase protein after LPS administration and during bacterial sepsis (16)(17)(18), suggestive of a role of apoCI in infl ammation and sepsis. Indeed, we recently showed that apoCI strongly binds to LPS, thereby prolonging the residence time of LPS in the circulation ( 19 ). In addition, apoCI stimulated the LPS-induced production of TNF ␣ by macrophages in vitro and in mice in vivo. By enhancing the biological response toward LPS and Gram-negative bacteria, apoCI improved the antibacterial attack, reduced bacterial outgrowth, and protected against intrapulmonal Klebsiella pneumoniae -induced fatal sepsis ( 19 ). Moreover, by enhancing the systemic infl ammatory state, apoCI aggravated LPS-induced atherosclerosis in hyperlipidemic apoE-defi cient mice ( 20 ). We were able to extrapolate these fi ndings to humans by showing that plasma apoCI Modifi ed from De Haas et al. ( 34 ). a Boldface residues represent basic amino acids; italic residues represent hydrophobic amino acids; underlined amino acids represent a comparable amino acid region.
b The receptor-binding domain of apoE has not yet been described as a LPS-binding domain.
ride (Fluka Chemie, Buchs, Switzerland) in 1 ml of 50 mM borate buffer (pH 8.5) for 18 h. After dialysis against PBS, 0.5 ml of the derivatized product was radiolabeled using 10 µL of 4 mg/ml chloramine T (Merck, Darmstadt, Germany) and 10 µL (0.25 mCi) Na 125 I (Amersham, Little Chalfont, UK). The reaction was stopped with 10 µL of 4 mg/ml NaS 2 O 4 , and the radioiodinated product was dialyzed extensively against PBS (pH 7.4). Prior to experiments, the 125 I-LPS was sonicated three times for 30 s, with 1 min intervals on ice in between, using a Soniprep 150 (MSE Scientifi c Instruments, Crawley, UK) at 10 µm output. The quality of the 125 I-LPS was routinely checked by agarose gel electrophoresis. The specifi c activity of 125 I-LPS was ‫ف‬ 1.0 × 10 3 cpm/ng.

Agarose gel electrophoresis
125 I-LPS (150 ng) was incubated (30 min, 37°C) in the absence or presence of apoCI-derived peptides or apoCI 1-57 at the indicated molar ratios. Aliquots of incubation mixtures were subjected to electrophoresis in 0.75% (w/v) agarose gels at pH 8.8 using 70 mM Tris-HCl, 80 mM hippuric acid, 0.6 mM EDTA, and 260 mM NaOH buffer. Bromophenol blue (Merck) served as a front marker. For detection of 125 I-LPS, gels were dried overnight at 65°C, exposed to a phosphorimaging plate (BAS-MS2040; Fuji Photo Film, Co. Ltd, Tokyo, Japan), and radioactivity was detected on a phosphor imaging analyzer (Fujix BAS-1000; Fuji Photo Film Co. Ltd.).

Serum residence and liver association of 125 I-LPS
Mice were anesthetized by intraperitoneal injection of domitor (0.5 mg/kg; Pfi zer, New York, NY), dormicum (5 mg/kg; Roche Netherlands, Mijdrecht, The Netherlands), and fentanyl (0.05 mg/kg; Janssen-Cilag B.V., Tilburg, The Netherlands) and the abdomens were opened. 125 I-LPS (10 µg/kg), preincubated (30 min 37°C) with or without apoCI-derived peptides, full-length apoCI  , or BSA at the indicated molar ratios, was injected via the vena cava inferior. Thirty minutes after injection, the serum residence and the liver association of 125 I-LPS were determined. Blood samples (<50 µL) were taken from the vena cava inferior and allowed to clot for 30 min. The samples were centrifuged for 10 min at 7,000 rpm, and 20 µL serum samples were counted for 125 I-radioactivity. The total amount of radioactivity in the serum was calculated using the equation: serum volume (ml) = 0.04706 × body weight (g) ( 31 ). At the same time, liver lobules were tied off, excised, weighed, and counted for radioactivity. Mice were killed and the remainder of the liver was excised and weighed. Uptake of 125 I-LPS by the liver was corrected for the radioactivity in the serum assumed to be present in the liver (84.7 µL serum/g wet weight) ( 32 ).

Challenge of macrophages with LPS
RAW 264.7 cells, a murine macrophage cell line, were seeded into 24-well plates (1 × 10 6 cells/well) or 96-well plates (60 × 10 3 cells/well) and cultured overnight at 37°C in DMEM, 10% FBS, 1% pen/strep. Cells were washed with DMEM and incubated with LPS (1 ng/ml) that was preincubated (30 min 37°C) with or without apoCI-derived peptides, apoCI 1-57 , soluble CD14 (sCD14), or LBP (both from Biometec, Greifswald, Germany) in DMEM supplemented with 0.01% human serum albumin (4 h at 37°C). As controls, cells were incubated with apoCI-derived peptides or apoCI 1-57 alone at the highest concentration used in combination with LPS. Where indicated, cells were preincubated with apoCI 1-57 (0.05 and 5 µg/ml), anti-CD14 antibody, or isotype control monoclonal antibody (10 µg/ml; Biometec) for 30 min at 37°C, washed, and subsequently stimulated with LPS. The medium was collected, and TNF ␣ was determined in the medium using the commercially available mouse TNF ␣ -specifi c OptEIA TM proteins, such as BPI, LPS-binding protein (LBP) ( 25 ), lactoferrin (Lf) ( 26 ), apoE ( 27 ), and MD2 ( 3 ) ( Table 1 ). Similar to the previously identifi ed LPS-binding motif of apoCI, these alternating cationic/hydrophobic motifs within apoCI are highly conserved during evolution ( 28 ) and may well be involved in the LPS-binding properties of apoCI. In this study we aimed at analyzing the structure and function relationship of apoCI with respect to binding of LPS and modulating the response to LPS. In addition, we aimed at unraveling the mechanism by which apoCI enhances the LPS-induced infl ammatory response. We designed and generated an array of N-and C-terminal apoCI-derived peptides containing the LPS-binding motif and/or a varying number of highly conserved alternating cationic/hydrophobic motifs. We demonstrate that peptides containing either the full N-terminal helix or the full C-terminal helix can bind LPS and enhance the infl ammatory response toward LPS, albeit with reduced effi ciency as compared with full-length apoCI. Furthermore, we show that apoCI stimulates the LPS-induced proinfl ammatory response via an LBP-like mechanism dependent on functional CD14/TLR4 signaling. We anticipate that in addition to the previously identifi ed LPS-binding motif, the highly conserved alternating cationic/hydrophobic motifs present throughout the apoCI sequence participate in the binding to LPS and enhancement of the biological response to LPS via a mechanism similar to LBP.

Animals
C57Bl/6 mice (own breeding) and TLR4-mutant (C3H/HeJ) and wild-type TLR4-expressing (C3H/HeN) mice from Jackson Laboratories (Bar Harbor, ME) were housed at the breeding facility of TNO-Quality of Life in a temperature-and humidity-controlled environment and were fed ad libitum with regular chow (Ssniff, Soest, Germany). All experiments were approved by the animal ethics committee of TNO or the Leiden University Medical Center. Experiments were conducted in male mice at 10-12 weeks of age.

Synthesis of apoCI-derived peptides
The synthesis of human apoCI-derived peptides was carried out by the Peptide Synthesis Facility of the Department of Immunohematology and Blood Transfusion at the Leiden University Medical Center (Leiden, The Netherlands) by solid phase peptide synthesis on a TentagelS-AC (Rap, Tübingen, Germany) using 9-fl uorenylmethoxycarbonyl/t-Bu chemistry, benzotriazole-1-yl-oxytris-pyrrolidino-phosphonium hexafl uorophosphate/ N -methylmorpholine for activation, and 20% piperidine in N -methylpyrrolidone for fl uorenylmethoxycarbonyl removal ( 29 ). The peptides were cleaved from the resin, deprotected with trifl uoroacetic acid/water, and purifi ed on Vydac C18. The purifi ed peptides were analyzed by reversed phase -HPLC and their molecular masses were confi rmed by MALDI-time-of-fl ight mass spectrometry (purity >95%). Synthesized full-length human apoCI (apoCI 1-57 ; purity >95%) was obtained from Protein Chemistry Technology Center (UT Southwestern Medical Center, Dallas, TX).

Effect of apoCI-derived peptides on the electrophoretic mobility of 125 I-LPS
To investigate the LPS-deaggregating properties of the apoCI-derived peptides in vitro, we incubated the peptides with 125 I-LPS and examined the electrophoretic mobility of the resulting radioactive complexes on agarose gel ( Fig.  2 ). Whereas micellar 125 I-LPS alone did not migrate (R f = 0), incubation of 125 I-LPS with full-length apoCI (apo-CI 1-57 ) at a 1:1 molar ratio resulted in a shift of all 125 I-LPS toward the front of the gel (R f = 0.95), confi rming our previous fi ndings ( 19 ). ApoCI 1-38 , apoCI 1-30 , and apoCI  showed reduced effi ciency to deaggregate 125 I-LPS compared with apoCI 1-57 . These peptides did not deaggregate 125 I-LPS at a 1:1 molar ratio but were effective at a 1:20 molar ratio. ApoCI 1-23 and apoCI 46-57 did not deaggregate 125 I-LPS at a 1:20 molar ratio ( Fig. 2 ) nor at a 1:60 molar ratio (not shown).

Effect of apoCI-derived peptides on the serum residence and liver association of 125 I-LPS
To examine whether the LPS-deaggregating characteristics of the various peptides would be refl ected in their ability to modulate the kinetics of LPS in vivo, 125 I-LPS was incubated with the peptides and intravenously injected into C57Bl/6 mice ( Fig. 3 ). 125 I-LPS alone was rapidly cleared from serum and ‫ف‬ 90% of the injected dose associated with the liver. ApoCI 1-57 almost completely prevented ELISA (BD Biosciences Pharmingen) according to the manufacturer's instructions. Likewise, thioglycollate-elicited TLR4mutant (C3H/HeJ) and wild-type TLR4-expressing (C3H/HeN) peritoneal macrophages from 12-week-old mice were seeded into 24-well plates (0.6 × 10 6 cells/well) and cultured as described above. Cells were washed, incubated with LPS preincubated with or without apoCI 1-57 (10-fold molar excess), and TNF ␣ was determined in the medium.

Challenge of mice with LPS
Mice were injected intravenously with LPS (25 µg/kg) preincubated (30 min at 37°C) without or with apoCI-derived peptides or apoCI 1-57 , at a 60-fold and 5-fold molar excess of peptide, respectively, in saline containing 0.1% (w/v) BSA. Just after injection (t 0 ) and at the indicated times, blood samples were taken from the tail vein and stored on ice. Samples were centrifuged (4 min at 14,000 rpm), and TNF ␣ levels were determined in the plasma using the commercially available mouse TNF ␣ module set BMS607MST (Bender MedSystems, San Bruno, CA).

Statistical analysis
Data were analyzed using nonparametric Mann-Whitney U tests. P -values < 0.05 were considered signifi cant.

Design of apoCI-derived peptides
To determine the structure and function relationship of apoCI with respect to binding of LPS and modulation of the in vivo behavior of LPS, we designed and generated fi ve peptides derived from human apoCI. Three peptides contain the full N-terminal ␣ -helix (i.e., apoCI 1-38 , apo-CI 1-30 ) or a part thereof (apoCI 1-23 ), and two peptides contain the full C-terminal ␣ -helix (i.e., apoCI  ) or a part thereof (apoCI 46-57 ) ( Fig. 1B ). In addition to length, the 125 I-LPS was incubated with apoCI 1-57 at a 1:1 molar ratio. Aliquots of the incubation mixtures (approximately 1 × 10 5 cpm) were subjected to electrophoresis in an 0.75% (w/v) agarose gel at pH 8.8. The resulting gel was dried and assayed for radioactivity by autoradiography. molar excess, apoCI  was still approximately 3-fold less active than apoCI   ( Fig. 4B ).
Taken together, it appears that the effects of these various apoCI-derived peptides on the in vivo kinetics of 125 I-LPS indeed refl ect their relative LPS-deaggregating properties ( Fig. 2 ). Importantly, both the N-and C-terminal helix contains structural components to deaggregate LPS and modulate the kinetic behavior of LPS in vivo.

Effect of apoCI-derived peptides on the LPS-induced TNF ␣ response in vitro
We previously showed that full-length apoCI 1-57 increases the LPS-induced TNF ␣ response both in vitro and in vivo ( 19 ). Therefore, we fi rst determined the ability of the peptides to increase the LPS-induced TNF ␣ response in murine RAW 264.7 macrophages compared with apoCI 1-57 ( Fig. 5 ). Incubation of macrophages with the peptides only did not result in detectable TNF ␣ secretion in the medium (not shown). the serum clearance and liver association of 125 I-LPS at a 1:1 molar ratio, which is in line with our previous observations ( 19 ). The N-terminal peptides apoCI 1-30 and apo-CI 1-38 also caused a substantial increase in the serum residence of 125 I-LPS and a concomitant decrease in the liver association at a 20-fold molar excess, whereas apo-CI 1-23 was ineffective ( Fig. 3A ). Both C-terminal peptides contain the previously identifi ed LPS-binding motif but were less effi cient with respect to modulating the kinetics of 125 I-LPS compared with the N-terminal peptides ( Fig.  3B ). At a 20-fold molar excess, apoCI  showed a modest increase of the serum residence and decrease of the liver association of 125 I-LPS, whereas apoCI 46-57 was ineffective.
The dose dependency of the effect of apoCI 1-30 (containing the full N-terminal helix) and apoCI  (containing the full C-terminal helix) to modulate the kinetics of LPS was investigated in more detail ( Fig. 4 ). ApoCI 1-30 had approximately the same effect as full-length apoCI 1-57 at a 60-fold higher ratio ( Fig. 4A ). In contrast, at this 60-fold  studies, both apoCI 1-30 and apoCI 35-57 enhanced the LPS-induced TNF ␣ response, albeit with reduced efficiency compared with apoCI 1-57 . At a 60-fold molar excess, the LPS-induced TNF ␣ response was increased by apoCI 1-30 (2.3-fold) and tended to be increased by apoCI  (1.8-fold).
Collectively, our fi ndings indicate that both the N-and C-terminal helix contain structural components able to deaggregate LPS, modulate the in vivo behavior of LPS, and enhance the LPS-induced TNF ␣ response in vitro and in vivo.

Role of TLR4 in the stimulating effect of apoCI on the LPS-induced TNF ␣ response in vitro and in vivo
We evaluated whether apoCI and the apoCI-derived peptides augment the LPS-induced TNF ␣ response via activation of TLR4, the signaling receptor of LPS. We stimulated peritoneal macrophages isolated from TLR4mutant (C3H/HeJ) and wild-type TLR4-expressing (C3H/HeN) mice with LPS alone or in the presence of apoCI   ( Fig. 7A ). In wild-type macrophages, a 10-fold molar excess of apoCI 1-57 enhanced the LPS-induced TNF ␣ response approximately 2-fold compared with LPS alone, which is comparable to the results obtained ApoCI 1-57 enhanced the LPS-induced TNF ␣ response approximately 3-fold and 24-fold at a 10-fold and 100fold molar excess, respectively ( Fig. 5A ), in line with our previous observations ( 19 ). Preincubation of macrophages with 0.05 and 5 µg/ml apoCI 1-57 (comparable with a 100-fold and 10,000-fold molar excess, respectively) had no effect on the LPS-induced TNF ␣ response (not shown). Both apoCI   ( Fig. 5B ) and apoCI   ( Fig. 5C ) also increased the LPS-induced TNF ␣ response. However, approximately 10-fold more peptide was required to achieve the level of stimulation as observed with apoCI 1-57 . ApoCI 1-30 was more effective than apoCI  , in line with its more pronounced effect on deaggregation of LPS and modulation of the in vivo behavior of LPS.

Effect of apoCI-derived peptides on the LPS-induced TNF ␣ response in vivo
Subsequently, we determined the ability of the peptides to increase the TNF ␣ response after intravenous injection of LPS in C57Bl/6 mice in vivo. At a 5-fold molar excess, apoCI 1-57 enhanced the LPS-induced plasma TNF ␣ levels 3.8-fold at 1 h after injection ( Fig.  6A ). Similarly to the in vitro macrophage stimulation Fig. 5. Effect of peptides containing the full N-terminal helix (apoCI  ) and full C-terminal helix (apoCI  ) on the LPS-induced TNF ␣ response in vitro. RAW 264.7 cells were incubated (4 h at 37°C) in DMEM supplemented with 0.01% human serum albumin with LPS (1 ng/ml) that was preincubated (30 min at 37°C) without or with apoCI 1-57 (A), apoCI 1-30 (B), or apoCI  (C) at the indicated molar ratios. TNF ␣ was determined in the medium by ELISA. Data are expressed as mean TNF ␣ concentration ± SD (n = 3-4). * P < 0.05 compared with LPS alone. Fig. 6. Effect of peptides containing the full N-terminal helix (apoCI  ) and full C-terminal helix (apoCI  ) on the LPS-induced TNF ␣ response in vivo. LPS (25 µg/kg) incubated (30 min at 37°C) without (white circles) or with (black circles) a 5-fold molar excess of apoCI 1-57 (A), or a 60-fold molar excess of apoCI 1-30 (black squares) or apoCI  (white squares) (B), were injected intravenously into C57Bl/6 mice. At the indicated time points, blood samples were taken and TNF ␣ levels were determined in plasma by ELISA. Values are means ± SEM (n = 6). Statistical differences were assessed compared with LPS alone. * P < 0.05, ** P < 0.01.

Role of CD14 in the stimulating effect of apoCI on the LPS-induced TNF ␣ response in vitro
Finally, we studied the role of CD14 in the apoCI-mediated enhanced LPS-induced infl ammatory response by blocking cell surface-bound CD14 using a CD14-specifi c antibody and by addition of sCD14. We preincubated RAW 264.7 macrophages with anti-CD14 antibody, followed by stimulation with LPS alone or in the presence of apoCI 1-57 ( Fig. 8A ). Blocking CD14 almost completely inhibited the stimulating effect of apoCI on the LPS-induced TNF ␣ response (up to 93%), while an isotype control antibody did not block the enhancing effect of apoCI  . In line with with RAW 264.7 macrophages ( Fig. 5A ). However, LPS without or with apoCI 1-57 did not induce a detectable TNF ␣ response in the TLR4-mutant macrophages. Similar fi ndings were observed with apoCI 1-30 and apoCI  (not shown). Likewise, a 5-fold molar excess of apoCI  augmented the LPS-induced TNF ␣ response approximately 2-fold in wild-type mice, confi rming our fi ndings in C57Bl/6 mice ( Fig. 6A ), but LPS without or with apoCI 1-57 failed to induce a TNF ␣ response in TLR4mutant mice ( Fig. 7B ). Taken together, these fi ndings indicate that the stimulating effect of apoCI on the LPSinduced TNF ␣ response depends on TLR4 signaling. Fig. 7. Role of TLR4 in the stimulating effect of apoCI on the LPS-induced TNF ␣ response in vitro and in vivo. A: Peritoneal macrophages of TLR4mutant (C3H/HeJ) and wild-type TLR4-expressing (C3H/HeN) mice were isolated and incubated in DMEM supplemented with 0.01% human serum albumin (4 h at 37°C) with LPS (1 ng/ml) that was preincubated (30 min 37°C) without (white bars) or with (black bars) a 10-fold molar excess of apoCI  . TNF ␣ was determined in the medium by ELISA. B: LPS (25 µg/kg) incubated (30 min at 37°C) without (white bars) or with (black bars) a 5-fold molar excess of apoCI 1-57 was injected intravenously into TLR4mutant (C3H/HeJ) and wild-type TLR4-expressing (C3H/HeN) mice. After 60 min, blood samples were taken and TNF ␣ levels were determined in plasma by ELISA. Values are means ± SD (n = 4) (A) and means ± SEM (n = 3) (B). Statistical differences were assessed compared with LPS alone. * P < 0.05. Fig. 8. Role of CD14 and LBP in the stimulating effect of apoCI on the LPS-induced TNF ␣ response in vitro. A: RAW 264.7 cells were preincubated (30 min at 37°C) in DMEM supplemented with 0.01% human serum albumin and vehicle (white bars), isotype control antibody (gray bars), or anti-CD14 antibody (black bars), washed, and subsequently incubated (4 h at 37°C), in DMEM supplemented with 0.01% human serum albumin, with LPS (1 ng/ml) that was preincubated (30 min at 37°C) without or with a 100-fold molar excess of apoCI  . TNF ␣ was determined in the medium by ELISA. B, C: RAW 264.7 cells were incubated (4 h at 37°C), in DMEM supplemented with 0.01% human serum albumin, with LPS (1 ng/ml) that was preincubated (30 min at 37°C) without or with a 100-fold molar excess of apoCI  in the presence of soluble CD14 (sCD14) (B) or LBP (C). TNF ␣ was determined in the medium by ELISA. Data are expressed as mean TNF ␣ concentration ± SD (n = 3-4). * P < 0.05 compared with vehicle (A) or LPS alone (B, C); # P < 0.05 compared with control antibody.
vented the uptake of LPS by the liver, prolonged the residence time of LPS in the serum, and enhanced the LPSinduced TNF ␣ response. These fi ndings indicate that both the N-and C-terminal helix contain additional structural elements that can deaggregate LPS and enhance the infl ammatory response toward LPS.
To evaluate whether these apoCI-derived peptides were able to bind LPS, we determined the ability of the peptides to alter the electrophoretic mobility of 125 I-LPS. Although we clearly showed that apoCI 1-30 , apoCI 1-38 , and apoCI  deaggregated LPS, we were unable to show colocalization of the peptides with 125 I-LPS, because the binding affi nity of the detecting anti-apoCI antibody was abrogated upon any truncation of full-length apoCI. However, because we previously showed that full-length apoCI fi rmly bound LPS and colocalized with LPS ( 19 ), it is very likely that the LPSdeaggregation potential of the various apoCI-derived peptides refl ect their LPS-binding potential. These fi ndings are in line with the hypothesis of Frecer et al. ( 33 ), who suggested, by using molecular modeling methods, that any peptide containing cationic/hydrophobic motifs within their sequence will recognize and bind LPS with high affi nity. Indeed, such motifs are also involved in the interaction of, for instance, Lf and BPI with LPS ( 34 ). Our observations that apoCI-derived peptides containing the N-terminal helix without the previously established LPSbinding motif (i.e . , apoCI 1-30 and apoCI  ) alter the in vivo behavior of LPS and increase the LPS-induced TNF ␣ response are, therefore, likely explained by the presence of such amphipathic motifs.
The effi ciencies of the apoCI-derived peptides to bind LPS and modulate the infl ammatory response toward LPS were lower compared with full-length apoCI  . It is likely that the structural elements in both helices, represented by the alternating cationic/hydrophobic peptide motifs, act synergistically in the binding of LPS. A similar synergistic interaction with LPS has previously been demonstrated for Lf ( 35 ) and sheep myeloid antimicrobial peptide-29 ( 36 ), a sheep myeloid antimicrobial peptide, which both have two LPS-binding domains, and BPI ( 37 ) and serum amyloid P ( 34 ), which both have three regions that contribute to binding to LPS.
It is intriguing to speculate how apoCI interacts with LPS and enhances the LPS-induced proinfl ammatory response. LPS consists of four different moieties: 1) lipid A, the toxic moiety, 2) the inner core, 3) the outer core, and 4) the O-antigen. We previously reported that apoCI interacts with different forms of LPS from Salmonella minnesota [i.e., full-length wild-type LPS, and the truncated Re595 LPS, containing the lipid A moiety and some KDO sugars ( 19 )], but more recently we also found that apoCI interacts with different types of LPS from Escherichia coli [i.e . , O55:B5 LPS ( 20 ) and J5 LPS (Berbée and Rensen, unpublished observations)]. ApoCI thus interacts with both fulllength, wild-type LPS and truncated forms of LPS, which indicates that the lipid A/KDO moiety of the LPS molecule contains the crucial elements for interaction with apoCI. Because apoCI is highly positively charged, we hypothesize that the amphipathic motifs within apoCI interact with the lipid A moiety via the negatively charged this fi nding, the addition of increasing concentrations of sCD14 to a fi xed concentration of apoCI (100-fold molar excess over LPS) strongly augmented the apoCI-mediated enhanced LPS-induced TNF ␣ response ( Fig. 8B ). In fact, whereas addition of apoCI 1-57 alone induced LPS-induced TNF ␣ levels up to 16.8 ± 4.4 ng/ml and sCD14 alone up to 60.2 ± 3.2 ng/ml, the combination showed a synergistic increase up to 129.5 ± 1.9 ng/ml. Collectively, these fi ndings indicate that the stimulating effect of apoCI on the LPS-induced TNF ␣ response thus depends on functional CD14.
Because LBP is known to bind LPS and subsequently present it to CD14, we studied the interaction between apoCI and LBP in relation to the LPS-induced infl ammatory response ( Fig. 8C ). As expected, LBP alone dose dependently increased the LPS-induced TNF ␣ response up to 16.5 ± 1.2 ng/ml, which is comparable to apoCI 1-57 (20.2 ± 1.2 ng/ml). However, the combination of apoCI  with LBP at most additively stimulated the LPS-induced TNF ␣ response up to only 30.8 ± 1.3 ng/ml. Together, these fi ndings indicate that apoCI does not interact with LBP but suggest that apoCI augments the LPS-induced infl ammatory response via a similar mechanism as LBP does, by presenting the LPS to CD14, which ultimately transfers the LPS to the MD2/TLR4 complex.

DISCUSSION
We recently identifi ed apoCI as an LPS-binding protein, with an apparent LPS-binding motif in its C-terminal helix, that acts as a biological enhancer of the proinfl ammatory response toward LPS ( 19 ). Because apoCI contains several highly conserved alternating cationic/hydrophobic motifs throughout its structure that may cooperate in LPS binding, we further investigated the structure and function relationship of apoCI with respect to its ability to bind LPS, to modulate the in vivo behavior of LPS, and to stimulate the proinfl ammatory response to LPS. In addition, we studied the mechanism by which apoCI enhances the LPS-induced infl ammatory response. We designed and generated an array of N-and C-terminal apoCI-derived peptides containing the apparent LPS-binding motif and/or varying numbers of the cationic/hydrophobic motifs. We demonstrate that apoCI contains additional structural elements, in addition to the LPS-binding motif, that enable it to bind LPS and enhance the CD14/TLR4-dependent infl ammatory response toward LPS via a mechanism similar to LBP.
Our previous studies showed that the binding of apoCI to LPS was largely mediated by a Lys-rich LPS-binding motif in the C terminus of apoCI ( KVKEKLK ; residues 48-54), which is highly homologous to the LPS-binding motif of LALF ( 23 ) and CAP-18 ( 24 ). Replacement of the Lys residues by Ala within this motif did decrease the ability of apoCI to bind LPS ( 19 ), but the binding to LPS was not abrogated completely. The mutant peptide was still able to modulate the in vivo kinetics of LPS to some extent ( 19 ), which indicated that apoCI should contain additional elements that cooperate in LPS binding. We now demonstrated that apo-CI 1-30 , apoCI  , and apoCI  all deaggregated LPS, pre-phosphate groups. Indeed, we found that apoCI interacts with the lipid A moiety. Similar to the effect of apoCI on the serum residence and liver uptake of 125 I-LPS, apoCI markedly enhanced the serum residence of 125 I-lipid A (29.7 ± 0.3% compared with 2.2 ± 0.4% for 125 I-lipid A alone) and inhibited the liver uptake (36.3 ± 0.6% vs. 66.5 ± 6.0%) at a 1:1 molar ratio (not shown). Because lipid A is the common determinant of LPS molecules from all bacterial species, apoCI is likely to bind a wide array of wildtype and mutant LPS molecules.
ApoCI is able to enhance the proinfl ammatory response toward LPS via a direct interaction with LPS in vitro and in vivo. Thus far, other proteins that have been reported to bind LPS (e.g., serum amyloid P, apoE, Lf, and LALF) attenuate rather than stimulate the LPS-induced infl ammatory response, except for LBP and CD14, which are required for LPS signaling ( 38,39 ). We conclusively demonstrate by using TLR4-mutant mice that apoCI augments the LPS response by enhancing TLR4-dependent signaling. Although the LDL receptor ( 40,41 ) and scavenger receptors, such as scavenger receptor class A (42)(43)(44) and BI (45)(46)(47), modulate infl ammatory responses to LPS, these receptors apparently are not involved in the stimulating effect of apoCI on the LPS response. Next to TLR4, we show that the stimulating effect of apoCI is also dependent on functional CD14. An anti-CD14 antibody completely blocked the augmenting effect of apoCI, whereas co-incubation with sCD14 synergistically enhanced the effect of apoCI on the LPS-induced infl ammatory response. Interestingly, co-incubation experiments of apoCI with LBP indicate that apoCI augments the LPS-induced infl ammatory response via a similar mechanism as LBP does, by presenting the LPS toward CD14. Based on these fi ndings, we now propose the following mechanism. Plasma apoCI, either bound to lipoproteins or moving freely in plasma, interacts with the lipid A part of LPS, presumably via the negatively charged phosphate groups, and subsequently transfers the LPS to either membrane-bound CD14 or sCD14, which ultimately presents the LPS to the MD2/TLR4 complex, thereby activating the TLR4dependent infl ammatory response.
In summary, we have demonstrated that apoCI contains structural elements in both its N-terminal and C-terminal helix to bind LPS, to modulate the in vivo behavior of LPS, and to enhance the TLR4-dependent proinfl ammatory response toward LPS in vitro and in vivo via an LBP-like mechanism. We anticipate that, in addition to the previously identifi ed LPS-binding motif, the highly conserved alternating cationic/hydrophobic motifs present throughout the apoCI sequence participate in the binding to LPS and enhance the biological response to LPS.