Isolation of Monomeric and Dimeric Secreted MD-2

Potent cell activation by endotoxin requires sequential protein-endotoxin and protein-protein interactions involving lipopolysaccharide-binding protein, CD14, MD-2, and Toll-like receptor 4 (TLR4). MD-2 plays an essential role by bridging endotoxin (E) recognition initiated by lipopolysaccharide-binding protein and CD14 to TLR4 activation by presenting endotoxin as a monomeric E·MD-2 complex that directly and potently activates TLR4. Secreted MD-2 (sMD-2) exists as a mixture of monomers and multimers. Published data suggest that only MD-2 monomer can interact with endotoxin and TLR4 and support cell activation, but the apparent instability of MD-2 has thwarted efforts to more fully separate and characterize the individual species of sMD-2. We have taken advantage of the much greater stability of sMD-2 in insect culture medium to fully separate sMD-2 monomer from dimer by gel sieving chromatography. At low nanomolar concentrations, the sMD-2 monomer, but not dimer, reacted with a monomeric complex of E·sCD14 to form monomeric E·MD-2 and activate HEK293/TLR4 cells. The monomer, but not dimer, also reacted with the ectodomain of TLR4 with an affinity comparable with the picomolar affinity of E·MD-2. These findings demonstrate directly that the monomeric form of sMD-2 is the active species both for reaction with E·CD14 and TLR4, as needed for potent endotoxin-induced TLR4 activation.

of the innate immune system. In many strains and species of GNB, this process depends on host recognition of and response to the unique complex glycolipid, endotoxin (lipooligosaccharide (LOS) or lipopolysaccharide (LPS)), that comprises much of the outer leaflet of the outer membrane of GNB (1,2). Minute amounts of endotoxin (E), presented either as an integral part of the outer membrane of GNB or as large aggregates of extracted and purified endotoxin, can stimulate pro-inflammatory responses (2)(3)(4)(5). This sensitivity depends upon an ordered series of endotoxin-protein and protein-protein interactions that include the host proteins LPS-binding protein (LBP), membrane-bound or soluble (s)CD14, secreted or TLR4-associated MD-2, and TLR4 (6 -8). Engagement of endotoxin-rich membranes or isolated endotoxin aggregates by LBP facilitates the extraction of endotoxin monomers by CD14 to form monomeric E⅐CD14 complexes that are the most efficient substrate for transfer of endotoxin to MD-2 (4,6,9). The monomeric E⅐MD-2 complex is necessary and sufficient to induce TLR4dependent cell activation by endotoxin (4,6,10). MD-2 associates noncovalently with the N-terminal ectodomain of TLR4 and plays a pivotal role not only in TLR4 activation but also in trafficking of TLR4 to the cell surface (7,(11)(12)(13). MD-2 also likely interacts transiently with CD14 facilitating transfer of endotoxin monomer from CD14 to MD-2 and, at high molar excess of CD14, reverse transfer of endotoxin from MD-2 to CD14 (14). Simultaneous engagement by MD-2 of both endotoxin and TLR4 is required for activation of TLR4 by endotoxin (4,15).
MD-2 belongs to the ML (MD-2-like lipid recognition) domain family of proteins (16). A structural hallmark of these proteins is the presence of a deep hydrophobic pocket into which specific ligands (e.g. specific glycolipids) bind. The recently reported x-ray structure of MD-2 has confirmed the presence in MD-2 of a ␤-barrel immunoglobin-fold structure (17). In contrast to the very stable and potently bioactive properties of the monomeric E⅐MD-2 complex (4,18), recombinant MD-2 expressed and secreted from mammalian cell cultures, such as human embryonic kidney (HEK) 293 cells, is recovered mainly as inactive multimers with little monomer detected (19 -22). The recovery of active secreted MD-2 (sMD-2) from mammalian cell cultures is further compromised by the instability of sMD-2 at 37°C in serum-free culture medium (9,23,24). In contrast to MD-2 secreted from mammalian cell cultures, nearly 50% of MD-2, produced by insect cells after infection with baculovirus containing MD-2 cDNA, is monomeric and remains active during storage at 4°C for more than 1 year and at 37°C for more than 6 h (9). 4 That monomeric MD-2 is the active form of MD-2 necessary for TLR4-dependent activation has been previously proposed (19,25,26). However, this conclusion has been largely based on studies in which the physical state of sMD-2, either alone or in complex with LPS or the TLR4 ectodomain (TLR4 ECD ), has been deduced from the electrophoretic properties of sMD-2 during SDS-PAGE under nonreducing conditions. Efforts to compare more directly the functional properties of monomeric and multimeric forms of sMD-2 been hampered, to date, by the instability of MD-2.
In this study, we have taken advantage of the stability of MD-2 produced and secreted by insect cells to separate monomeric and dimeric forms of sMD-2 using size exclusion chromatography under nondenaturing and nonreducing conditions. This separation has made it possible to investigate directly the functional properties of the monomer and dimer forms of sMD-2. The studies presented here demonstrate directly that only monomeric MD-2 has the ability to act as an acceptor for transfer of endotoxin from E⅐sCD14 as shown by generation of E⅐MD-2 and TLR4 activation in the presence of E⅐sCD14 and to interact with high affinity with the TLR4 ectodomain. Together, these properties explain the importance of the monomeric form of sMD-2 in activation of cells expressing TLR4 by endotoxin.

Materials-[ 3 H]LOS or [ 14 C
]LOS (25,000cpm/pmol; 3000 cpm/pmol, respectively) was isolated from an acetate auxotroph of Neisseria meningitidis serogroup B metabolically labeled and isolated as described (3). LBP and sCD14 were gifts from Xoma (Berkley, CA) and Amgen Corp. (Thousand Oaks, CA), respectively. Human serum albumin (HSA) was obtained as an endotoxin-free, 25% stock solution (Baxter Health Care, Glendale, CA). Chromatography matrices (Sephacryl HR S200 and S300) were purchased from GE Healthcare. Express Five TM medium was purchased from Invitrogen and supplemented with 2 mM glutamine per the manufacturer's instructions.
Production of Recombinant Proteins: Insect Cell-derived Recombinant Polyhistidine-tagged MD-2-cDNA encoding human MD-2 was inserted into pBAC11 (Novagen) using XhoI-and NotI-sensitive restriction sites as described (4). Sf9 cells were used for transfection and amplification of baculovirus, whereas High Five TM cells in serum-free medium (Express Five TM ) were used for protein production (4). Conditioned medium containing secreted human sMD-2-His 6 was prepared in collaboration with Biovest International/National Cell Culture Center (Minneapolis, MN) and stored at 4°C. The conditioned medium containing human sMD-2-His 6 was used either directly or after concentration 10 -12-fold using Amicon Ultra Centrifugal devices (Millipore) (molecular weight cut-off, 10,000) followed by sterile filtration.
Gel Filtration Chromatography of Recombinant MD-2-His 6 Secreted from High Five TM Insect Cells-For preparative runs, 2 ml of 10 -12ϫ concentrated conditioned insect medium containing recombinant human sMD-2-His 6 was applied to 1.6ϫ ϳ70-cm column of Sephacryl S200 pre-equilibrated either in Dulbecco's phosphate-buffered saline (PBS) Ϯ 0.1% HSA or in Express Five TM insect medium supplemented with 2 mM glutamine and eluted in the same medium at a flow-rate 0.5 ml/min with the collection of 1-ml fractions. Small differences in elution of sMD-2-His 6 dimer (see Figs. 2 and 3) were a result of small differences in column length. The presence of MD-2-His 6 in individual fractions was determined by immunoblot as described below. Peak fractions corresponding to sMD-2 monomer or dimer were pooled, sterile-filtered, and stored at 4°C. Pools of sMD-2 monomer and dimer were used within 2 days of collection for functional assays (see below). Sephacryl S200 was calibrated with the following proteins using Bio-Rad standards for gel filtration: thyroglobulin (650,000, V o ), IgG (158,000), HSA (66,000), ovalbumin (44,500), myoglobin (17,500), and vitamin B12 (1200, V i ).
Immunoblotting-To detect polyhistidine labeled MD-2, an anti-polyhistidine antibody (Tetra-His antibody; Qiagen) was used. The samples were heated at 100°C for 10 min in Laemmli sample buffer Ϯ 200 mM dithiothreitol ϩ 6 M urea and electrophoresed through a 4 -20% gradient acrylamide gel (Pierce) using Tris/HEPES/SDS buffer and transferred to nitrocellulose. The nitrocellulose was washed with 20 mM Tris, 0.5 M NaCl, pH 7.5, containing 0.05% Tween 20 and 0.2% Triton X-100 (TBSTT), blocked to reduce nonspecific background with 3% bovine serum albumin in 20 mM Tris, 0.5 M NaCl, pH 7.5 for 1 h at 25°C, and incubated with the anti-His 4 antibody in the same buffer overnight. After washing with TBSTT, the blot was incubated with donkey anti-mouse IgG conjugated to horseradish peroxidase (Jackson Immunologicals) for 1 h at 25°C in 10 mM Tris, 150 mM NaCl, pH 7.5, 0.1% Tween 20 containing 3% goat serum and washed with TBSTT exhaustively. The blots were developed using the Pierce SuperSignal substrate system. The levels of sMD-2-His 6 in experimental samples were quantified by densitometric analysis of immunoblots, using known amounts of E⅐MD-2 as standards. To facilitate quantitative comparisons of sMD-2-His 6 , the samples were pretreated with dithiothreitol so that all sMD-2 was monomeric. insect cell medium containing sMD-2-His 6 or the indicated amount of isolated monomeric or dimeric sMD-2-His 6 . Products of the reaction were resolved by S200 chromatography in PBS as previously described (4, 6) unless otherwise indicated. Fractions containing complex were combined, sterile-filtered, and stored at 4°C.
HEK293 Cell Activation Assay-HEK293/TLR4 cells were cultured as has been previously described (28). For cell activation assays, the cells were grown to confluency and then washed twice with warm PBS, pH 7.4, and incubated overnight at 37°C, 5% CO 2 , and 95% humidity in 96-well plates containing Dulbecco's modified Eagle's medium, 0.1% HSA with [ 14 C]LOS⅐sCD14 (2 nM or 60 pM as indicated) or LOS aggregates (20 nM) Ϯ the indicated source of sMD-2. Activation of HEK293/ TLR4 cells was assessed by measuring accumulation of extracellular IL-8 by ELISA (BD Clontech, Inc., Palo Alto, CA).

Greater Yield and Stability of Bioactive Recombinant Human sMD-2 in Conditioned Insect Cell versus Mammalian Cell
Medium-We have previously shown that harvested conditioned medium of insect cells infected with baculovirus encoding human MD-2-His 6 contains significant amounts of bioactive sMD-2 that can readily react with monomeric E⅐sCD14 complex to generate monomeric E⅐MD-2 (4,9). E⅐MD-2, but not E⅐sCD14 or purified endotoxin aggregates, can bind to TLR4 ECD with picomolar affinity and activate cells expressing TLR4 without MD-2 (e.g. HEK/TLR4 cells) at picomolar concentrations (4,9,14). Addition of as little as 1 l of this conditioned medium (representing ϳ1-2 ng of sMD-2; ϳ500 pM final concentration of sMD-2) together with 200 pM LOS⅐sCD14 (1 ng LOS/ml) produced maximum activation of HEK/TLR4 cells (Fig. 1A), whereas parental HEK293 cells lacking TLR4 were not affected (4) (data not shown). Cell activation required co-incubation of the conditioned medium containing sMD-2 with LOS⅐sCD14 (i.e. generation of LOS⅐MD-2); neither LOS⅐sCD14 ( Fig. 1A) nor the conditioned medium alone (4) induced activation of the HEK/TLR4 cells.
In contrast to the conditioned insect cell medium containing sMD-2, harvested conditioned medium from HEK293 cells expressing and secreting MD-2 was much less potent in activating HEK/TLR4 cells when incubated together with LOS⅐sCD14 ( Fig. 1A) or in generating monomeric LOS⅐MD-2 (9,24). Comparison of the two conditioned media by SDS-PAGE/immunoblot under nonreducing conditions demonstrated a significant difference in the amount and physical state of sMD-2 in the two media under these conditions. As seen in   Fig. 1B). These findings indicate that the harvested conditioned medium from infected insect High Five TM cells contain much more active sMD-2 than that present in conditioned medium of transiently transfected HEK293 cells. The yields of bioactive sMD-2 shown were the optimum we could obtain under our culture conditions and were not increased by either higher titers of infection, plasmid DNA, or time of incubation before harvesting.

Separation of sMD-2 Monomer and Dimer by Sephacryl S200
Chromatography in Insect Culture Medium-The findings above strongly suggested that conditioned medium from infected High Five TM insect cells provided the more favorable starting material for separation and functional characterization of sMD-2 monomer and multimers. An important limitation of virtually all of the previous studies on the physical state of sMD-2, including our own (Fig. 1B), has been the reliance on SDS-PAGE (i.e. presence of SDS) to resolve sMD-2 monomer and multimers. To better define the physical state of bioactive sMD-2 in the conditioned medium (i.e. under nondenaturing as well as nonreducing conditions), conditioned insect medium containing sMD-2-His 6 was concentrated 10 -12-fold and subjected to Sephacryl S200 size exclusion chromatography using PBS as eluant. Individual collected fractions were analyzed for MD-2-His 6 content by SDS-PAGE under nonreducing conditions followed by immunoblot using an anti-polyhistidine antibody. Fig. 2A shows that, under these chromatographic conditions, fractions containing sMD-2 species eluting with an apparent M r ϳ40,000 (fractions 74 -78), corresponding to the expected elution of dimeric sMD-2, contained almost exclusively dimeric MD-2 as judged by SDS-PAGE/immunoblot. Earlier fractions along the upslope of the sMD-2 dimer peak were enriched on immunoblots with an additional, more slowly migrating, sMD-2 species that by behavior on Sephacryl S200 and SDS-PAGE appeared to be a trimer of sMD-2. Comparison of immunoblots of the recovered fractions with the concentrated conditioned medium applied to the column indicated recovery of sMD-2 dimer (and trimer) of Ͼ50%, whereas recovery of monomeric sMD-2 was substantially less ( Fig. 2A), suggesting a selective loss of sMD-2 monomer. The small amount of recovered immunoreactive material consistent with the size of monomeric sMD-2-His 6 eluted over a surprisingly broad range of fractions, including those (after fraction 95) in which molecules much smaller than the monomer of sMD-2 (i.e. M r , Ͻ20,000) would be expected to elute. The poor recovery and broad and late elution of monomeric sMD-2-His 6 suggested aberrant behavior of the sMD-2-His 6 monomer in the gel filtration matrix in PBS. Supplementing PBS with 0.1% HSA in the column and elution buffer did not significantly increase recovery or change the elution pattern of the sMD-2 monomer (data not shown).
Because bioactive MD-2-His 6 is stable for more than 1 year in the conditioned insect cell medium, we repeated size exclusion chromatography using the insect cell medium (Express Five TM ; Invitrogen) as the equilibration and elution buffer. In contrast to chromatography in PBS, chromatography in Express Five TM medium resulted in markedly improved recovery of the sMD-2 monomer (Fig. 2B). However, even under these conditions, sMD-2 monomer eluted significantly later than that expected for a protein of molecular mass of ϳ20,000 daltons. The aberrant elution of monomeric sMD-2 during Sephacryl S200 chromatography was fortuitous in that it resulted in virtually complete separation of sMD-2 dimer and monomer (Fig. 2B). As a result, fractions containing almost exclusively the monomer or dimer forms of sMD-2 could be pooled (Fig. 2 legend) and used for functional assays (see below). The apparent homogeneity of these sMD-2 pools was confirmed by nonreducing SDS-PAGE/immunoblot analysis (Fig.  2C). Under reducing conditions, all of the immunoreactive material migrated as M r ϳ 20,000, i.e. as monomeric MD-2 (Fig. 2C).
Efficient Production of Monomeric E⅐MD-2 Requires Monomeric sMD-2 and Presentation of Endotoxin as E⅐sCD14-Key functional properties of sMD-2 include: 1) reactivity with E⅐sCD14 to form monomeric E⅐MD-2; 2) binding to the TLR4 ectodomain; and 3) activation of TLR4 by bound E⅐MD-2. Thus, we first compared the ability of sMD-2 monomers and dimers to react with [ 14 C]LOS⅐sCD14 (M r ϳ 60,000) to generate [ 14 C]LOS⅐MD-2 (M r , ϳ 25,000) as assessed by size exclusion chromatography. Fig. 3A shows that incubation of [ 14 C]LOS⅐sCD14 with sMD-2 monomers caused substantial conversion of [ 14 C]LOS⅐sCD14 to monomeric [ 14 C]LOS⅐MD-2. In contrast, the addition of even 10-fold more sMD-2 dimer produced very little monomeric [ 14 C]LOS⅐MD-2, indicating strong reactivity of sMD-2 monomer but little or no reactivity of sMD-2 dimer with LOS⅐sCD14. This conclusion was further supported by monitoring the consumption of sMD-2 monomer versus dimer during incubation of [ 14 C]LOS⅐sCD14 with the sMD-2-containing conditioned insect cell culture medium (Fig. 3B). Formation of monomeric [ 14 C]LOS⅐MD-2 ( Fig. 3B) resulted in 1) the appearance, by immunoblot, of sMD-2 monomers in fractions containing monomeric [ 14 C]LOS⅐MD-2 (peak is fraction 77); 2) the disappearance of free sMD-2 monomer (note absence of sMD-2 by immunoblot after fraction 84); and 3) no apparent loss of free sMD-2 dimer. These findings directly demonstrate the selective reactivity of monomeric LOS⅐sCD14 with monomeric sMD-2.
Earlier studies have demonstrated that sMD-2 can also react with purified endotoxin (12,25). We therefore repeated the same experiment using [ 14 C]LOS aggregates instead of monomeric LOS⅐sCD14. Fig. 3C shows that, under these experimental conditions, there was neither conversion of LOS aggregates to monomeric LOS⅐MD-2 nor association of sMD-2 monomer, dimer, or higher order multimers with the LOS aggregates that eluted much earlier in the void volume during Sephacryl S200 chromatography. Taken together, these findings demonstrate that efficient sMD-2-endotoxin interactions require presentation of sMD-2 as a monomer and presentation of endotoxin as a monomeric E⅐sCD14 complex.

Efficient Interaction of sMD-2 with TLR4 Ectodomain
Requires Monomeric sMD-2-Activation of TLR4 by endotoxin requires simultaneous binding of MD-2 to both endotoxin and the TLR4 ectodomain (TLR4 ECD ). This can occur either by  AUGUST 8, 2008 • VOLUME 283 • NUMBER 32 direct binding of monomeric E⅐MD-2 to TLR4 or by transfer of endotoxin from monomeric E⅐CD14 to a preformed MD-2/ TLR4 heterodimer (4,9,12,14,25). Previous studies have demonstrated direct interactions of sMD-2 with the TLR4 ECD but estimated an affinity that was nearly 50-fold lower than E:MD-2-TLR4 ECD binding (12 nM) (26) versus 200 -300 pM) (9). To determine whether sMD-2 binds to the TLR4 ECD with an affinity comparable with that of E⅐MD-2, we took advantage of an  Fig. 4 (A and  B), the monomeric form of sMD-2 caused dose-dependent inhibition of formation of ([ 3 H]LOS⅐MD-2/ TLR4 ECD ) 2 , with half-maximal inhibition produced by ϳ1-2 nM sMD-2 monomer (Fig. 4B). Purified sMD-2 monomer also produced dose-dependent inhibition of cell (HEK/TLR4) activation by LOS⅐MD-2 as measured by the accumulation of extracellular IL-8 (Fig. 4C). In contrast, the dimeric form of sMD-2, even at 10 -30ϫ higher concentration, produced little or no inhibition of formation of ([ 3 H]LOS⅐MD-2/TLR4 ECD ) 2 (Fig. 4A) or of HEK/TLR4 cell activation by LOS⅐MD-2 (Fig. 4C). Thus, the monomer form of sMD-2, but not the dimer, binds with high affinity to the TLR4 ectodomain.

Potent Activation of Cells Expressing TLR4 without MD-2 Requires Presence of Monomeric Form of sMD-2 and Presentation of Endotoxin as Monomeric E⅐CD14-We have previously
shown that insect cell culture medium containing sMD-2 promotes activation by E⅐sCD14 of cells expressing TLR4 without MD-2 (4,6). The selective reactivity of the monomeric form of sMD-2 with monomeric LOS⅐sCD14 shown in Fig. 3 suggests that the activity of the conditioned insect cell medium is due specifically to the presence of monomeric sMD-2. To test this hypothesis, we compared the effect of increasing amounts of unfractionated conditioned insect cell medium containing sMD-2-His 6 or of isolated dimeric or monomeric sMD-2-His 6 with LOS⅐sCD14 on activation of HEK/TLR4 cells, as measured by the accumulation of extracellular IL-8. Fig. 5A shows closely similar dose-dependent effects of the conditioned medium containing sMD-2 and of isolated monomeric sMD-2 in promoting activation of HEK/TLR4 cells when co-incubated with LOS⅐sCD14. In contrast, dimeric MD-2 displayed negligible activity (Fig. 5A). These data indicate that activation of cells expressing TLR4 without MD-2 by LOS⅐sCD14 requires the presence of monomeric sMD-2.
Potent activation of HEK/TLR4 cells by endotoxin required not only presentation of sMD-2 as a monomer but also presentation of endotoxin as a monomeric complex with CD14. Thus, only limited cell activation was observed when endotoxin was added as LOS aggregates, even when Ͼ100-fold greater amounts of endotoxin and of sMD-2 were added (Fig. 5B). These differences in the potency of cell activation parallel the differences observed in formation of monomeric LOS⅐MD-2 when LOS⅐sCD14 versus LOS aggregates were incubated with sMD-2-containing culture medium (Fig. 3, compare B and C).
Time-dependent Decay of Bioactive sMD-2 at 37°C and Protection by Insect Culture Medium-Loss of bioactive sMD-2 has been demonstrated during incubation at 37°C in serum-free medium (9,23,24). In an effort to better understand the apparent protective effects of the insect culture medium on the recovery and preservation of bioactive sMD-2, we compared the stability of insect cell-derived bioactive sMD-2 at 37°C after dilution in either insect culture (Express Five TM ) medium or PBS with or without 0.1% albumin. Using inhibition of [ 3 H]LOS⅐MD-2 binding to TLR4 ECD as a functional assay, we observed a significant difference in stability of bioactive sMD-2 at 37°C after dilution in PBS with (Fig. 6A) or without (data not shown) 0.1% albumin versus dilution in Express Five TM medium (Fig. 6A). As previously seen in serum-free mammalian cell culture medium (9,23,24), there was a substantial loss of bioactive sMD-2 during 24 h of incubation in PBS Ϯ 0.1% albumin but no loss of sMD-2 activity during similar incubation in the Express Five TM medium (Fig. 6A). These findings show clearly that the Express Five TM medium provides a favorable environment for preservation of bioactive sMD-2.
Decay of Functional Activity of Isolated sMD-2 Monomer during Storage at 4°C-Both sMD-2 in the unfractionated conditioned insect culture medium and purified E⅐MD-2 are stable at 4°C for at least one year with no apparent loss in potency toward HEK/TLR4 cells or reactivity of the sMD-2-containing conditioned medium with TLR4 ECD or with E⅐sCD14 to form E⅐MD-2 (data not shown). However, the pooled purified sMD-2 monomer gradually lost activity during storage for several weeks at 4°C, as manifest by a reduced ability to react with [ 3 H]LOS⅐sCD14 to form [ 3 H]LOS⅐MD-2 (Fig. 6B). Remarkably, this functional decay was not accompanied by conversion of sMD-2 monomer to multimeric form, at least as judged by SDS-PAGE/immunoblot under nonreducing conditions (Fig. 6C). Loss of sMD-2 activity could be prevented by storage of sMD-2 monomer at Ϫ80°C (data not shown).

DISCUSSION
We have demonstrated directly, for the first time, that the ability of MD-2 secreted without TLR4 (sMD-2) to contribute to host cell responsiveness to endotoxin depends on the presentation of sMD-2 as a monomer. This was accomplished by the use of conditioned medium from insect cells expressing and secreting recombinant human MD-2 and use of the insect cell medium as the column and elution buffer for separation of sMD-2 monomer and multimers by size exclusion chromatography (Fig. 2). The selective reactivity of the recovered sMD-2 monomer was demonstrated by several different functional parameters relevant to the participation of sMD-2 in host cell responses to endotoxin. These include: reactivity with monomeric E⅐CD14 to form monomeric E:MD-2 (Fig. 3), binding to the ectodomain of TLR4 (Fig. 4), activation of HEK/TLR4 cells when added together with LOS⅐sCD14 (Fig. 5), or inhibition of activation of these cells by LOS⅐MD-2 (Fig. 4C). Marked differences in reactivity of sMD-2 monomer and dimer with LOS⅐sCD14 were observed both in the unfractionated conditioned medium (Fig. 3B) and after isolation (Fig. 3A), indicating that these differences reflect intrinsic differences in the functional properties of sMD-2 monomer and dimer. Our findings confirm earlier conclusions by other investigators (19,21,22,26) who relied mainly on the behavior of sMD-2 in SDS-PAGE under nonreducing conditions to discern MD-2 monomer and multimers and to demonstrate that the product of interactions of sMD-2 with endotoxin and with TLR4 ECD includes, selectively, monomeric MD-2. Comparison of the behavior of sMD-2 in SDS-PAGE under nonreducing conditions with its behavior during size exclusion chromatography under nondenaturing/non reducing conditions (Fig. 2B) revealed a generally good correlation of these two analytical parameters. Thus, at least in conditioned insect cell culture medium and freshly derived protein fractions, the relative levels of monomeric sMD-2 can be assessed by SDS-PAGE/immunoblot under nonreducing conditions (19). However, to demonstrate that monomeric sMD-2 was the selective reactant with E⅐CD14 and TLR4 ECD required the separation of sMD-2 monomer and multimers that we describe in this study.
Our findings suggest that the use of baculovirus-infected High Five TM insect cells to express and secrete potently bioactive sMD-2 monomer was advantageous by increasing the fraction of sMD-2 that was secreted and remained as a bioactive monomer during long periods of storage (Fig. 1). The difference in relative abundance of sMD-2 monomer in conditioned insect cell medium versus conditioned HEK293T cell medium was not due to differences in overall sMD-2 concentration in the two media. Overall sMD-2 content in the two different conditioned media was comparable (Fig. 1B), and concentration of the conditioned insect cell medium did not increase sMD-2 oligomerization (data not shown). Many factors could contribute to the increased secretion and/or stability of insect cell-derived sMD-2 monomer, including differences in glycosylation (29,30), the lower temperature of culture of insect (27°C) versus mammalian (37°C) cells, and components of the culture medium itself. We have no direct evidence that differences in the metabolic properties of insect versus mammalian cells or the different culture temperatures are important, although time-dependent loss of sMD-2 functional activity at 37°C in serum-free medium has been shown (9,23,24). We also observed time-dependent decay of bioactive sMD-2 derived from insect cells but only when sMD-2 was diluted in PBS Ϯ 0.1% albumin and not when diluted in insect culture medium (Fig. 6A). Together with the markedly enhanced recovery of sMD-2 monomer by use of the Express Five TM culture medium in size exclusion chromatography (Fig. 2B), these findings suggest a potentially important role of this culture medium in preserving/maintaining sMD-2 as a bioactive monomer. MD-2 belongs to the ML domain family of proteins, which includes mite allergen proteins (e.g. Der p2, p1, f 2), GM2-activator protein, Npc2, and their orthologs (16). The recent publication of the crystal structure of recombinant MD-2 (17) confirms earlier structural models of MD-2 that were based on the solved structures of other ML domain proteins (31)(32)(33)(34)(35)(36)(37). A hallmark of MD-2, as well as other ML domain proteins, is the presence of a deep, hydrophobic cavity that can expand to accommodate specific (glyco)lipid ligands. Studies with MD-2 have been hampered by its instability and propensity to form an array of multimers. MD-2 multimers typically contain intermolecular disulfide(s), as manifest by the complete conversion of multimers to monomers of MD-2 when reducing agents are included in the SDS-PAGE buffer (19 -22) (Figs. 1B and 2C). The remarkable stability of E⅐MD-2 as a monomeric complex and its water solubility suggest that occupation of the hydrophobic cavity of MD-2 by the acyl chains of endotoxin can significantly increase the stability and solubility of monomeric MD-2. Other hydrophobic compounds, including certain free fatty acids, can occupy the hydrophobic cavity of MD-2 (17) and, at high enough concentrations, may have similar stabilizing effects on sMD-2 monomer. The Express Five TM medium contains several detergents, surfactants, and other lipids that could be responsible for the greater recovery of sMD-2 monomer following size exclusion chromatography (Fig. 2). The delayed elution of the sMD-2 monomer, in comparison with LOS⅐MD-2 (compare Figs. 2B and 3B), manifest in both PBS Ϯ 0.1% albumin and Express Five TM medium, could reflect weak interactions of sMD-2 monomer with the gel matrix that retard elution or a less spherical shape of MD-2 under these experimental conditions.
Our findings also confirm and extend earlier observations indicating that endotoxin must be presented as a monomeric complex with CD14 to react efficiently with MD-2 (4,14). Neither MD-2 monomer nor dimer showed measurable interaction with endotoxin aggregates under the same reaction conditions in which there was quantitative conversion of sMD-2 monomer to E⅐MD-2 by incubation with E⅐CD14 (Figs. 3, B and C). These findings are consistent with the very large differences in apparent affinity that have been previously reported for reaction of MD-2 with immobilized LPS (presumably representing LPS in aggregated form; K d ϭ ϳ65 nM) (25) versus that of MD-2 with E⅐sCD14 (K d ϭ ϳ100 -200 pM) (4). The low levels of HEK/ TLR4 cell activation observed during incubation of HEK/TLR4 cells with endotoxin aggregates in the presence of sMD-2 ( Fig.  5B) could be explained by a yield of monomeric LOS⅐MD-2 that was Յ0.1% of that resulting from reaction of LOS⅐sCD14 with sMD-2, below even the very sensitive limits of detection provided by size exclusion analysis of [ 3 H]LOS of very high specific radioactivity.
Lastly, our studies have not revealed an independent functional role for sMD-2 dimer. The apparent lack of reactivity of sMD-2 dimer with either E⅐CD14, endotoxin aggregates, or TLR4 ECD is inconsistent with a role of sMD-2 dimer as a "decoy" (19) in which interactions could lead to formation of nonproductive E⅐MD-2 (dimer) or TLR4-MD-2 (dimer) complexes that blunt host cell responsiveness to endotoxin. It remains possible that there are other, as yet unidentified substrates for sMD-2 dimer or other, higher order, sMD-2 multimers. It should be noted, however, that the extent to which sMD-2 exists in biological fluids in monomeric versus multimeric form or in association with other host extracellular proteins is not known. The ability to monitor total sMD-2 by ELISA (38) and sMD-2 monomer by reaction with [ 3 H]LOS⅐sCD14 and/or TLR4 ECD may now make it possible to address these important questions.