The role of serum in interleukin 1 production by human monocytes activated by endotoxins and their polysaccharide moieties

From benzodiazepines to fatty acids and beyond: revisiting the role of ACBP/DBI

Four decades ago Costa and colleagues identified a small, secreted polypeptide in the brain that can displace the benzodiazepine diazepam from the GABA_A receptor, and was thus termed diazepam binding inhibitor (DBI). Shortly after, an identical polypeptide was identified in liver by its ability to induce termination of fatty acid synthesis, and was named acyl-CoA binding protein (ACBP). Since then, ACBP/DBI has been studied in parallel without a clear and integrated understanding of its dual roles. The first genetic loss-of-function models have revived the field, allowing targeted approaches to better understand the physiological roles of ACBP/DBI in vivo. We discuss the roles of ACBP/DBI in central and tissue-specific functions in mammals, with an emphasis on metabolism and mechanisms of action.

Metabolic and psychiatric effects of acyl coenzyme A binding protein (ACBP)/diazepam binding inhibitor (DBI)

Acyl coenzyme A binding protein (ACBP), also known as diazepam binding inhibitor (DBI) is a multifunctional protein with an intracellular action (as ACBP), as well as with an extracellular role (as DBI). The plasma levels of soluble ACBP/DBI are elevated in human obesity and reduced in anorexia nervosa. Accumulating evidence indicates that genetic or antibody-mediated neutralization of ACBP/DBI has anorexigenic effects, thus inhibiting food intake and inducing lipo-catabolic reactions in mice. A number of anorexiants have been withdrawn from clinical development because of their side effects including an increase in depression and suicide. For this reason, we investigated the psychiatric impact of ACBP/DBI in mouse models and patient cohorts. Intravenously (i.v.) injected ACBP/DBI protein conserved its orexigenic function when the protein was mutated to abolish acyl coenzyme A binding, but lost its appetite-stimulatory effect in mice bearing a mutation in the γ2 subunit of the γ-aminobutyric acid (GABA) A receptor (GABA_AR). ACBP/DBI neutralization by intraperitoneal (i.p.) injection of a specific mAb blunted excessive food intake in starved and leptin-deficient mice, but not in ghrelin-treated animals. Neither i.v. nor i.p. injected anti-ACBP/DBI antibody affected the behavior of mice in the dark–light box and open-field test. In contrast, ACBP/DBI increased immobility in the forced swim test, while anti-ACBP/DBI antibody counteracted this sign of depression. In patients diagnosed with therapy-resistant bipolar disorder or schizophrenia, ACBP/DBI similarly correlated with body mass index (BMI), not with the psychiatric diagnosis. Patients with high levels of ACBP/DBI were at risk of dyslipidemia and this effect was independent from BMI, as indicated by multivariate analysis. In summary, it appears that ACBP/DBI neutralization has no negative impact on mood and that human depression is not associated with alterations in ACBP/DBI concentrations.

Lipopolysaccharide inhibits virus-mediated induction of interferon genes by disruption of nuclear transport of interferon regulatory factors 3 and 7

We have studied the effects of lipopolysaccharide (LPS) on the Newcastle disease virus (NDV)-mediated induction of cytokine genes expression. Raw cells treated with LPS before or after virus infection showed down-regulation in the expression of interferon A and, to a lesser extent, interferon B genes. In contrast, induction of the interleukin (IL)-6 gene was enhanced. The effects of LPS were not a result of the suppression of virus replication, because the transcription of viral nucleocapsid gene was not affected. Consistent with these findings, LPS also suppressed the NDV-mediated induction of chloramphenicol acetyltransferase reporter gene driven by murine interferon A4 promoter in a transient transfection assay. Furthermore, LPS inhibited virus-mediated phosphorylation of interferon regulatory factor (IRF)-3 and the consequent translocation of IRF-3 from cytoplasm to nucleus. The LPS-mediated inhibition of IFNA gene expression was much weaker in infected Raw cells that constitutively overexpressed IRF-3. The nuclear translocation of IRF-7 in infected cells was also inhibited by LPS. These data suggest that LPS down-regulates the virus-mediated induction of IFNA genes by post-translationally targeting the IRF-3 and IRF-7 proteins.

Recognition of endotoxin by cells leading to transmembrane signaling

Bacterial endotoxins act at picomolar to nanomolar concentrations to stimulate a wide variety of cell types including phagocytic and endothelial cells. The major elements identified to date that are crucial for recognition of endotoxin are lipopolysaccharide (LPS)-binding protein, membrane-bound CD14 and, most recently, soluble CD14. Recent results also indicate that membrane-bound CD14 is probably one part of a multi-component LPS receptor. An immediate consequence of engagement of this functional LPS receptor is protein tyrosine phosphorylation and initiation of the multiple intracellular events associated with LPS-induced cell activation.

Alpha 1-acid glycoprotein potentiates lipopolysaccharide-induced secretion of interleukin-1 beta, interleukin-6 and tumor necrosis factor-alpha by human monocytes and alveolar and peritoneal macrophages

Although the physiological role of alpha 1-acid glycoprotein (AGP), an acute-phase protein, is poorly understood, several lines of evidence support a modulatory action on the immune response. In this study, we investigated the effect of AGP on the production of interleukin (IL)-1 beta, IL-6 and tumor necrosis factor (TNF)-alpha by human monocytes, macrophages and the monocytic THP-1 cell line. AGP significantly enhanced (2- to 7-fold) the production of these cytokines in monocytes induced by suboptimal concentrations of lipopolysaccharide [E. coli lipopolysaccharide (LPS): 100 ng/ml] in serum-free conditions, whereas it had little or no effect in the absence of LPS. The potentiating effect of AGP was inhibited by specific antibodies. It was concentration dependent and the greatest enhancement was observed with 250-500 micrograms/ml. Moreover, AGP only potentiated the effect of suboptimal concentrations of LPS. AGP did not alter the time course of LPS-induced IL-1 beta, IL-6 or TNF-alpha secretion. AGP acts as a co-inducer and could also potentiate cytokine secretion triggered by Neisseria meningitidis LPS and muramyl dipeptide. The glycan moiety of AGP did not seem to be involved in its potentiating effect, since both its major glycoforms and asialo-AGP potentiated the effect of LPS to the same extent as native AGP. Possible differences in the effect of AGP according to cell maturation were investigated using isolated human macrophages: AGP potentiated LPS-induced cytokine production by both peritoneal and alveolar macrophages. These data suggest that AGP can modulate monocyte/macrophage functions, thereby contributing to the amplification and regulation of immune and inflammatory responses.

Enzymatically deacylated Neisseria lipopolysaccharide (LPS) inhibits murine splenocyte mitogenesis induced by LPS

Acyloxyacyl hydrolase is a leukocyte enzyme that selectively removes the secondary acyl chains from the lipid A moiety of gram-negative bacterial lipopolysaccharides (LPS). As predicted by the reported contribution of secondary acyl chains to the bioactivities of lipid A analogs, enzymatic deacylation of Salmonella typhimurium Rc LPS substantially reduces its potency in the dermal Shwartzman reaction and in several in vitro assays that measure responses of human endothelial cells and neutrophils, whereas the potency of this LPS for inducing murine splenocyte mitogenesis is affected much less. In the experiments described here, we studied the impact of acyloxyacyl hydrolysis on the bioactivities of several LPS that differ from Salmonella LPS in carbohydrate and lipid A structures. Deacylated LPS from Escherichia coli, Haemophilus influenzae, Neisseria meningitidis, and S. typhimurium were similarly reduced in potency in the Limulus lysate test (30- to 60-fold reduction in potency relative to the corresponding mock-treated LPS), and the ability of all of these deacylated LPS to stimulate neutrophil adherence to human endothelial cells was reduced by a factor of 100 or more. For LPS from E. coli, H. influenzae, and Pseudomonas aeruginosa, the impact of deacylation on spleen cell mitogenesis was also similar to that observed for S. typhimurium LPS: deacylation reduced potency by less than 15-fold. Unexpectedly, the potency of Neisseria LPS in the murine splenocyte mitogenicity test was reduced over 100-fold by deacylation, and deacylated Neisseria LPS could block the mitogenic activity of Neisseria and Salmonella LPS. These studies indicate that the contribution of secondary acyl chains to the bioactivities of a given LPS cannot be predicted with confidence from the reported structure-activity relationships of lipid A or from the behavior of other deacylated LPS.

Induction of IL-1 during hemodialysis: transmembrane passage of intact endotoxins (LPS)

Circulating monocytes of patients undergoing chronic hemodialysis are triggered to produce interleukin-1 (IL-1) in vivo. Intradialytic induction of IL-1 is associated with complement activation in patients dialyzed with first-use cellulose membranes. Chronic stimulation of IL-1 production occurs because of an yet unidentified mechanism in patients dialyzed with high permeability membranes. The present study demonstrates that intact bacterial lipopolysaccharide (LPS) molecules may cross cuprophan, AN69 and polysulfone membranes under in vitro conditions simulating in vivo hemodialysis. The experiments used purified LPS from Neisseria meningitidis and LPS from Pseudomonas testosteroni, a bacterial strain grown out from a clinically used dialysate. LPS were purified to homogeneity and radiolabeled. Transmembrane passage of 3H-labeled LPS was observed within the first five minutes of dialysis. A total of 0.1 to 1% of 3H-labeled LPS were recovered in the dialysate compartment after one hour of dialysis. High amounts of LPS, representing 40 to 70% of the amount originally present in the dialysate, were absorbed onto high permeability membranes. Low amounts of LPS were absorbed onto cuprophan membranes. The amount of LPS absorbed decreased with the concentration of LPS in the dialysate. LPS recovered from the blood compartment exhibited the same molecular weight as that used to contaminate the dialysate. Biochemically detectable transmembrane passage of LPS was not associated with that of material detectable using the limulus amebocyte lysate (LAL) assay. An IL-1-inducing activity was, however, detected in the blood compartment upon dialysis with high permeability membranes, as previously found by others with cuprophan membranes.(ABSTRACT TRUNCATED AT 250 WORDS)

Recombinant C5a enhances interleukin 1 and tumor necrosis factor release by lipopolysaccharide-stimulated monocytes and macrophages

Lipopolysaccharide (LPS) is a potent inducer of interleukin 1 (IL 1) synthesis and release, and of tumor necrosis factor (TNF) secretion. Many signals can enhance the LPS-induced production of these cytokines. We have previously observed that addition of low amounts of normal human serum to the culture medium enhances IL 1 production. Among serum factors, anaphylatoxins C3a and C5a and/or their desArg derivatives have been shown to enhance LPS-induced IL 1 and TNF production. However, the capacity of natural anaphylatoxins to induce by themselves the production of cytokines remains a controversial issue. We have investigated the capacity of human recombinant C5a (hrC5a) to induce IL 1 and TNF production. Despite its lack of direct triggering, hrC5a was able to act synergistically with LPS, leading to higher IL 1 and TNF release by human monocytes and mouse peritoneal macrophages. As assessed by the comitogenic assay, hrC5a increased IL 1 release, whereas cell-associated IL 1 activity was not significantly modified. Measurement by enzyme-linked immunosorbent assay of human IL 1 beta led to similar conclusions, whereas measurement of IL 1 alpha by radioimmunoassay indicated, in addition, an increase in intracellular IL 1 alpha.

Production and purification of bovine monocyte-derived interleukin 1

Few studies have addressed the biological and molecular nature of bovine interleukin 1 (IL-1). In an effort to increase our understanding of the role of bovine IL-1 in bovine immunology, we investigated various parameters of its production by LPS-stimulated monocytes in vitro. Bovine monocytes isolated by our methods constitutively released IL-1 activity, as measured by the murine thymocyte IL-1 assay. Monocyte release of IL-1 activity was further augmented when the cells were incubated with 0.005-10 micrograms per ml of Escherichia coli lipopolysaccharide (LPS). The presence of 1, 5, or 10 percent heat-inactivated fetal bovine serum (FBS) enhanced LPS-stimulated bovine monocyte release of IL-1 activity as compared with monocytes cultured under serum-free conditions. We used a combination of size-exclusion and reverse-phase high-performance liquid chromatography (HPLC) to purify bovine IL-1 from serum-free monocyte culture supernatants. Size-exclusion HPLC resulted in a single peak of biological activity with an approximate molecular weight of 18,000 daltons. Further purification by reverse-phase HPLC demonstrated at least three major molecular species with IL-1 activity. Besides providing information about production of IL-1 by bovine monocytes in vitro, this study also describes a protocol to purify bovine IL-1 for future studies addressing its biological functions.

Inhibition of endotoxin-induced activation of human monocytes by human lipoproteins

Toxicity of lipopolysaccharide (LPS) (endotoxin) is, to a large extent, mediated by the activation of monocytes/macrophages and subsequent release of monokines, such as interleukin-1 (IL-1) and tumor necrosis factor alpha (TNF-alpha). It is known that LPS binds readily to serum lipoproteins and that LPS-lipoprotein complexes are less toxic than unbound LPS. Here we present data analyzing the impact of the LPS-serum interaction at the cellular level. By measuring IL-1 TNF-alpha, and IL-6, the interaction of different LPSs or lipid A with human serum could be shown to prevent the activation of human monocytes. The amounts of LPS inactivated by normal human serum did not exceed 10 ng/ml. The LPS-inactivating capacity of serum was shown to be a function of the lipoproteins. Other serum components, such as naturally occurring anti-LPS immunoglobulin G, complement, or nutritive lipids, had no significant influence in our system. Our experiments suggest that serum lipoproteins control endotoxin-induced monocyte activation and monokine release.

Interleukin-1 induction by lipopolysaccharides: structural requirements of the 3-deoxy-D-manno-2-octulosonic acid (KDO)

We previously showed the importance of the 3-deoxy-D-manno-2-octulosonic acid (KDO) residue in endotoxins (lipopolysaccharides, LPS) for the induction of the synthesis and release of interleukin-1 (IL-1) by human monocytes. We further investigated the effect of some structural variations within the KDO molecule on IL-1 production induced by LPS. Deamination of Bordetella pertussis LPS, followed by mild anhydrous acidic methanolysis released a hexasaccharide (fragment B'), which had a terminal methyl ketoside KDO residue with a methyl-esterified carboxyl group. This fragment was unable to induce IL-1 production by human monocytes. Fragment B' could be converted into an active hexasaccharide by de-esterification (fragment B-OMe), but not by reduction of the methyl ester group. The KDO residues in the LPS of some bacterial species have been shown to be phosphorylated and we observed that these LPS were weak IL-1 inducers. Phosphorylated KDO present in Vibrio cholerae and B. pertussis LPS respond poorly in the thiobarbiturate assay (specific for KDO). However, if these LPS were dephosphorylated with aqueous hydrofluoric acid (HF) their KDO response in this assay was increased 5.4- to 2.6-fold, respectively. In parallel, the HF-treated LPS were more potent IL-1 inducers than untreated endotoxins. These data confirm that the KDO residue(s) present in all endotoxins play(s) a major role in the signal(s) leading to IL-1 production by human monocytes, and show that IL-1 induction by LPS (1) requires a free carboxyl group in the KDO and (2) is correlated with the degree of substitution of the KDO.

Endotoxin release from neisseria meningitidis. Relationship between key bacterial characteristics and meningococcal disease

A method was established in order to measure total and filtrable (CF) endotoxin in cultures of Neisseria meningitidis strains. The Limulus lysate (LAL) test gave results which paralleled those of the standard rabbit pyrogen test and of gas chromatography (GLC), concerning detection of different CF endotoxin levels. Meningococci varied in their ability to produce CF endotoxin. Cultures of similar bacterial densities, whether with a high (E+) or a low (E-) release of endotoxin in CF, had both high yields of endotoxin in sonicated culture suspensions determined by the LAL test. GLC demonstrated only small differences in total LPS contents between E+ and E- strains. This suggests that strains with similar cell wall endotoxin contents may vary in CF endotoxin. Electron microscopy revealed that E+ strains presented a high number of free, outer cell wall fragments (blebs, tubuli, membranes and aggregates of such structures) in surroundings. Few such free, small structures were found around E- strains. The amount of CF endotoxin of E+ strains was in part a function of the number of colony forming units (CFU/ml), and generally followed the growth curve. Because of its moment of appearance, and also based on electron microscopy findings, CF endotoxin appeared mainly to be released from living bacteria. The CF level of endotoxin was low or not detected at all in cultures of E- strains although their cultures reached higher mean CFU-levels than the niveaus required for the detection of CF endotoxin in the E+ strains. The E+ property was strain dependent. Meningococci isolated from CSF or blood had a significantly higher proportion of E+ strains (88.2%) and a higher CF endotoxin titre (greater than or equal to 10(3); 34.5%), than isolates from carriers (32.3% and 10.8%, respectively) (p less than 0.001 and p less than 0.001, respectively). A high mean CFU/ml in cultures seemed to be more often associated with isolates from patients than from carriers, more often with the presence than abscence of capsular polysaccharide (p less than 0.05), and more often with the presence than absence of the E+ property (p = 0.002). E+ strains were mostly serogroupable (i.e. encapsulated), regardless of source of the isolate (99% case and 80% carrier isolates). In contrast, serogroupable bacteria were not necessarily E+ when isolated from carriers (54.8%). The serogroup most apt to cause disease tended to have the highest proportion of E+ strains and the highest level of CF endotoxin.(ABSTRACT TRUNCATED AT 400 WORDS)

Dissociation of cell-associated interleukin-1 (IL-1) and IL-1 release induced by lipopolysaccharide and lipid A

The capacities of lipopolysaccharide (LPS) and lipid A to trigger mouse BALB/c peritoneal macrophages and to induce the production of cell-associated interleukin-1 (IL-1) and membrane-associated IL-1 and IL-1 release have been compared. Bordetella pertussis lipid A was 1,000 to 10,000 times less efficient than the native LPS to induce IL-1 release by freshly isolated elicited macrophages. When resident macrophages were studied, lipid A, at high concentrations (greater than 2 micrograms/ml), induced significant levels of cell-associated IL-1 but little or no IL-1 release. With synthetic lipid A built up with the Escherichia coli lipid A structure (compound 506), IL-1 activity was present in the supernatants of elicited peritoneal macrophages and to a lesser extent in those of resident macrophages. However, the release of IL-1 induced by synthetic lipid A 506 remained much lower than those induced by rough LPS. Membrane-associated IL-1 could be induced on BALB/c macrophages with LPS and natural or synthetic lipid A, the LPS being the most active. In C3H/HeJ mice, neither natural nor synthetic lipid A could induce detectable cell-associated IL-1, whereas LPS could induce cell-associated and membrane IL-1 activity but no IL-1 release. Our results indicate that fragments of endotoxins may induce the production of IL-1 but the entire structure of the LPS molecule is the most effective to induce intracellular IL-1 production, expression of membrane IL-1, and release of IL-1.

Therapy of suspected septicemia in neonatal foals using plasma-containing antibodies to core lipopolysaccharide (LPS)

Equine antiserum to core lipopolysaccharide (LPS) was evaluated in a double-blind prospective study for therapeutic benefit in suspected septicemia in neonatal foals. Forty foals younger than 7 days of age were included in the study by satisfaction of clinical and laboratory criteria, suggestive of gram-negative septicemia. Twenty-two foals were treated with core LPS antiserum (plasma produced from horses which were hyperimmunized with rough gram-negative mutant bacterin) and 18 foals received "nonimmune" plasma (from horses prior to immunization against core LPS). All foals received antimicrobials, fluids, and other supportive care measures, depending on clinical signs and according to accepted current practice. The clinical and laboratory data of each foal were monitored and recorded daily for 14 days after plasma treatment or until death. The overall survival rate of these 40 foals with septicemia was 52.5%. The most prevalent diagnoses in addition to septicemia were enteritis and pneumonia. Of 30 positive bacterial cultures, 93% were due to gram-negative organisms. There was no statistically significant increase in survival rate in the 22 foals given core LPS antiserum (P greater than 0.05).

Inhibition by gangliosides of the specific binding of lipopolysaccharide (LPS) to human monocytes prevents LPS-induced interleukin-1 production

In a previous work we have reported that gangliosides inhibit interleukin 1 (IL-1) release by human monocytes stimulated with lipopolysaccharides (LPS). In the present study we extend this work to IL-1 production and we correlate these observations with the capacity of gangliosides to inhibit the binding of radiolabeled LPS to its specific receptor on human monocytes. Preincubation of 3H-LPS with crude bovine brain gangliosides, as well as purified human brain mono, di, and trisialogangliosides (GM1, GD1a, and GT1b, respectively), led to an inhibition of the specific binding of LPS to the cell surface. Neither ceramide nor N-acetyl neuraminic acid, two constituents of gangliosides, was able by itself to inhibit the specific binding. A strict parallelism was observed with respect to inhibition on LPS-induced IL-1 production and release. Asialoganglioside (asialo-GM1) was inactive in both assays, suggesting that the N-acetyl neuraminic acid plays a role within the ganglioside molecule, with respect to inhibitory activity. We conclude that LPS-induced production and release by human monocytes is not due to a signal triggered by nonspecific absorption and/or intercalation of LPS into cell membrane which occur through hydrophobic interaction mediated by the lipid A region. Addition of exogenous sialogangliosides which blocked LPS-induced IL-1 production and release, did not modify significantly the nonspecific binding of 3H-LPS, whereas it did inhibit the specific binding which is mediated by the polysaccharide moiety of the LPS molecule. These results establish a relationship between the specific endotoxin receptor on monocytes and a LPS-induced cellular function.

Polymyxin-B inhibition of LPS-induced interleukin-1 secretion by human monocytes is dependent upon the LPS origin

Polymyxin-B (PMB) is an antibiotic known to inhibit various biological activities induced by lipopolysaccharides (LPS). We have investigated the ability of PMB to inhibit LPS-induced interleukin-1 (IL-1) secretion by human monocytes in vitro. Interleukin-1 was assayed by the conventional comitogenic assay using mice thymocytes. Our data demonstrate that PMB (1-2 micrograms/assay)-mediated inhibition of LPS-induced IL-1 secretion depends on the origin of the LPS. Interleukin-1 secretions induced by Escherichia coli and Acinetobacter calcoaceticus LPSs, when used at 1 microgram/assay were completely inhibited by PMB, whereas those induced by Neisseria gonorrheae, Neisseria meningitidis, Bordetella pertussis, and Salmonella enteritidis LPSs were unaffected. Neisseria meningitidis, the most potent IL-1 inducer tested could be inhibited by PMB only at concns below 5 ng/assay; when the assay was performed in the presence of serum (0.2%) PMB could not completely inhibit Neisseria meningitidis LPS-induced IL-1 secretion at LPS doses as low as 100 pg/assay. Polymyxin B itself, at doses greater than 50 micrograms/assay, stimulated IL-1 secretion and acted synergistically with LPS to induce IL-1 secretion when used at 10 micrograms/assay. Potential relevance of Lipid A-mediated IL-1 secretion and the use of PMB to detect endotoxin contamination is discussed.

Inhibition of lipopolysaccharide-induced monocyte interleukin 1 secretion by gangliosides

Gangliosides known to be potent immunosuppressors are shown to inhibit the secretion of interleukin 1 (IL 1) by human monocytes stimulated with lipopolysaccharide (LPS). The inhibitory activity was observed either with mono-, di- or trisialogangliosides. The inhibition of LPS-induced IL 1 secretion was obtained in the presence of indomethacin, indicating that the ganglioside inhibitory capacity was not due to prostaglandin induction. Indeed the inhibitory activity seems to be mediated by an interaction with LPS molecules, preventing their ability to deliver a signal to monocytes for IL 1 secretion.

Molecular requirement for interleukin 1 induction by lipopolysaccharide-stimulated human monocytes: involvement of the heptosyl-2-keto-3-deoxyoctulosonate region

Experiments were undertaken to localize in the lipopolysaccharide (LPS) the minimal structural determinants sufficient to initiate the signal leading to interleukin 1 (IL 1) secretion by human monocytes. Our results clearly demonstrated that this signal is triggered by structures present in the so-called inner-core region which chemically consists of 2-keto-3-deoxy-D-manno-octulosonic acid (KDO) and heptose in many LPS of gram-negative bacteria. Thus, the isolated polysaccharide region of Bordetella pertussis endotoxin as well as fragments derived therefrom containing the reducing KDO unit were able to induce similar levels of IL1 induction as the native LPS. Similarly, the trisaccharide alpha-D-manno-heptopyranosyl-(1-3)-alpha-D-manno-heptopyranosyl -(1-5)-3 -deoxy-D-manno-octulosonic acid (hep-hep-KDO), representative for the inner-core region of a large number of enterobacterial LPS, was a very potent IL 1 inducer. Neither KDO monosaccharide, nor the alpha-(2-4)-linked 3-deoxy-D-manno-octulosonic acid disaccharide isolated from Salmonella rough-form LPS promoted the signal indicating that the minimal structure of endotoxin able to induce IL 1 secretion resides in the hep (1-5)-KDO disaccharide.