Anti-TNF antibodies: lessons from the past, roadmap for the future.

Tumor necrosis factor alpha (TNF) is an important cell-signaling component of the immune system. Since its discovery over 20 years ago, much has been learned about its functions under normal and disease conditions. Nonclinical studies suggested a role for TNF in chronic immune-mediated inflammatory diseases, such as rheumatoid arthritis, Crohn's disease, and psoriasis, and therefore neutralizing monoclonal antibodies specific to human TNF were developed for clinical evaluation. Treatment with anti-TNF monoclonal antibodies (infliximab, adalimumab, and certolizumab pegol) has been shown to provide substantial benefit to patients through reductions in both localized and systemic expression of markers associated with inflammation. In addition, there are beneficial effects of anti-TNF treatment on markers of bone and cartilage turnover. Further exploration of changes in these markers and their correlation with clinical measures of efficacy will be required to allow accurate prediction of those patients most in need of these treatments. Both the clinical and commercial experience with these anti-TNF antibodies provide a wealth of information regarding their pharmacological effects in humans.

Identification of TNF-alpha binding peptides from a D-amino acid hexapeptide library that specifically inhibit TNF-alpha binding to recombinant p55 receptor.

Tumour necrosis factor alpha (TNF-alpha) is a pro-inflammatory cytokine with pleiotropic activity that binds to two transmembrane receptors. Its role in mediating the inflammatory response to injury or infection has been well documented and it has been shown to be a causative factor in rheumatoid arthritis, inflammatory bowel disease and septic shock. Using synthetic peptide libraries composed exclusively of D-amino acids, two distinct hexapeptide families that block the binding of TNF-alpha to its receptors were identified. In the deconvolution of the library, activity increased from submillimolar to the low micromolar range with the most active compound having an IC50 of 0.33 microM. With the aid of biotinylated constructs of these hexapeptides it was possible to demonstrate that their antagonistic effect is due to specific binding to TNF-alpha and not to its receptor.

The mouse/human chimeric monoclonal antibody cA2 neutralizes TNF in vitro and protects transgenic mice from cachexia and TNF lethality in vivo.

The pleiotropic cytokine tumour necrosis factor-alpha (TNF) is thought to play a central role in infectious, inflammatory and autoimmune diseases. Critical to the understanding and management of TNF-associated pathology is the development of highly specific agents capable of modifying TNF activity. We evaluated the ability of a high affinity mouse/human chimeric anti-TNF monoclonal antibody (cA2) to neutralize the in vitro and in vivo biological effects of TNF. cA2 inhibited TNF-induced mitogenesis and IL-6 secretion by human fibroblasts, TNF-priming of human neutrophils, and the stimulation of human umbilical vein endothelial cells by TNF as measured by the expression of E-selectin, ICAM-1 and procoagulant activity. cA2 also specifically blocked TNF-induced adherence of human neutrophils to an endothelial cell monolayer. Receptor binding studies suggested that neutralization resulted from cA2 blocking of TNF binding to both p55 and p75 TNF receptors on the cells. In vivo, repeated administration of cA2 to transgenic mice that constitutively express human TNF reversed the cachectic phenotype and prevented subsequent mortality. These results demonstrated that cA2 effectively neutralized a broad range of TNF biological activities both in vitro and in vivo.

Reactivity of the human antiendotoxin immunoglobulin M monoclonal antibody HA-1A with lipopolysaccharides from rough and smooth gram-negative organisms.

Clinical data suggest that the human immunoglobulin M antiendotoxin antibody HA-1A reduced mortality in patients diagnosed with gram-negative bacteremia and bacteremia with shock. Previous studies have demonstrated that HA-1A binds to the lipid A domain of lipopolysaccharide (LPS). The present study evaluated the ability of HA-1A to interact with LPs isolated from various strains of gram-negative bacteria by using liquid-phase rate nephelometry and solid-phase immunoblotting assays. HA-1A formed immune complexes in solution with LPSs isolated from both rough and smooth gram-negative organisms. Western blot (immunoblot) analysis of these LPS preparations revealed that HA-1A bound to LPS isolated from rough gram-negative organisms and to a rough LPS-like component present in smooth LPS. HA-1A also bound to LPS-protein complexes found in certain commercial rough LPS preparations. Preincubation of HA-1A with lipid A completely blocked subsequent binding of HA-1A to LPS in both liquid- and solid-phase assay formats, suggesting that the interaction of HA-1A with LPS is through the lipid A domain. Evidence that the binding of HA-1A to LPS was mediated through the antigen-combining (Fv) region of the antibody was provided by the finding that a murine anti-idiotypic antibody to HA-1A inhibited binding. These findings suggested that the broad antiendotoxin reactivity exhibited by HA-1A appeared to be due to the ability of HA-1A to bind to the conserved lipid A moiety of LPSs derived from both smooth- and rough-phenotype gram-negative bacterial strains.

Human monoclonal antibody HA-1A binds to endotoxin via an epitope in the lipid A domain of lipopolysaccharide.

HA-1A, a human IgM mAb, has been shown to significantly reduce mortality in septic patients with Gram-negative bacteremia, especially those with septic shock, in a controlled clinical trial. To confirm the reported specificity of this antibody for the lipid A domain of endotoxin, several assay systems were developed. These assay systems included an ELISA, which measured the binding of HA-1A to lipid A adsorbed to a solid phase; a rate nephelometry assay, which measured the ability of HA-1A to bind and aggregate lipid A in solution; and a dot-blot immunoassay, which measured the ability of HA-1A to interact with lipid A adsorbed to Immobilon-P. In all three assay systems, HA-1A bound in a dose-dependent manner to lipid A prepared from Salmonella minnesota R595 LPS, whereas negative control human IgM mAb or polyclonal antibodies did not. Several experimental approaches were employed to demonstrate the specificity of HA-1A in these assay systems. Both polymyxin B and murine IgG mAb (8A1) with a specificity for lipid A were able to competitively inhibit HA-1A reactivity with lipid A in a dose-dependent manner. Furthermore, a murine IgG anti-Id mAb (9B5.5) developed against HA-1A was also able to block the binding of HA-1A to lipid A in these assay formats. HA-1A reactivity with synthetic lipid A confirmed that HA-1A binding to the natural lipid A was not the result of contaminants in the latter. Finally, the reactivity of HA-1A against a variety of glucosamine-containing and fatty acid-containing compounds was assessed. Some weak interaction was seen with cardiolipin and chitin, but not with serum proteins, lipoteichoic acid, or DNA. Collectively, these results conclusively establish that HA-1A binds to the lipid A region of LPS by an interaction with the V region of the antibody.

Rapid infarct imaging with a technetium-99m-labeled antimyosin recombinant single-chain Fv: evaluation in a canine model of acute myocardial infarction.

Studies of monoclonal antibody-based imaging agents show that blood clearance is inversely proportional to molecular size, i.e., Fab or Fab' > F(ab')2 > IgG. Indium-111-antimyosin Fab-DTPA is a highly specific and sensitive marker for myocardial necrosis. An improvement on current antibody diagnostic imaging may result from the use of smaller labeled fragments. We report the first in vivo targeting of acute myocardial infarction with a novel recombinant single-chain Fv (sFv) antimyosin protein. The sFv (MW = 27,594) is approximately one-half the size of the Fab and is comprised of the heavy and light chain variable regions from the myosin-specific murine monoclonal antibody R11D10 which were joined by a 15-amino-acid linker and expressed as a fusion protein (sFv) in E. coli. The binding affinity of the sFv for cardiac myosin was similar to the affinity observed for the Fab fragment. Technetium-99m labeling of the sFv was accomplished by the attachment of a cleavable, ester-linked bifunctional chelator (RP-1). Comparative studies in mice showed 99mTc-sFv-RP-1 cleared significantly faster (p < 0.001) than 99mTc-Fab'-RP-1 and 111In-Fab-DTPA antimyosin fragments. Furthermore, measurement of 99mTc-sFv-RP-1 blood clearance in a canine model of acute myocardial infarction gave a mean T1/2 of 0.54 +/- 0.13 hr versus 2.80 +/- 0.57 and 2.58 +/- 0.64 hr for Fab-DTPA and Fab'-RP-1 (p < 0.05), respectively. Despite its comparatively rapid clearance, 99mTc sFv-RP-1 had similar uptake in the infarct compared to the Fab'-RP-1. In addition, infarct visualization was more rapid with the sFv. Thus, these data demonstrate antimyosin sFv possesses characteristics necessary for rapid imaging of myocardial necrosis.

Evaluation of indium-111-labeled anti-fibrin antibody for imaging vascular thrombi.

Monoclonal antibody 59D8 developed by Hui et al., binds to fibrin but not fibrinogen. An 111In-labeled Fab fragment of 59D8 was studied in vitro and in animal models to evaluate its potential for imaging thrombi and emboli in man. Rabbits and dogs were used as models for studying thrombus uptake in vivo. Thrombi and emboli up to 4 days old were successfully visualized at 4-24 hr postinjection in five of eight rabbits. In dogs, 0.5-hr-old and 24-hr-old thrombi were successfully imaged at 24 hr in six of eight animals, and 3/6 of these were positive at 3-4 hr postinjection. Thrombus-to-blood ratios in the dogs averaged 7.1 +/- 1.3. The findings suggest this antibody may be useful for imaging thrombi in man.

Radiosensitizer conjugation to the carcinoma 19-9 monoclonal antibody.

Misonidazole was covalently conjugated (3-68 mol drug/mol antibody) to 19-9 monoclonal antibody directed against a colorectal carcinoma tumor-associated antigen as a method for targeting radiosensitizing agents. This attachment was accomplished by the mixed anhydride method using the hemisuccinate derivative of misonidazole. Evaluation of conjugates in vitro shows a loss of antibody binding activity with increasing loading levels; however, significant binding activity is retained even at relatively high sensitizer/antibody ratios. This observation was consistent in three binding assays: a competitive radioimmunoassay; an enzyme immunoassay; and an affinity column assay. From these studies, it was concluded that the optimal loading factor for misonidazole-antibody conjugates, when it is used for immunochemotherapy lies between 8 and 15. In vitro release studies indicated that conjugates are hydrolytically stable (t1/2 = 4 days) under physiological conditions.

Evidence that viral transforming gene products and epidermal growth factor stimulate phosphorylation of the same cellular protein with similar specificity.

Treatment of A-431 human epidermoid carcinoma cells with epidermal growth factor (EGF) was shown to enhance the phosphorylation of a Mr = 34,000 protein. Because the phosphorylation of an analogous protein is enhanced in various cell lines transformed by Rous sarcoma virus (RSV) (Erikson, E., and Erikson, R. L. (1980) Cell 21, 829-836), we characterized the phosphorylation of the A-431 Mr = 34,000 protein under these two conditions in order to determine whether there are common pathways between viral transformation and EGF stimulation. The results of tryptic phosphopeptide mapping and phosphoamino acid analysis showed that the Mr = 34,000 protein was phosphorylated in an identical manner by the EGF-stimulated protein kinase activity and by the protein kinase activity of the RSV transformation-specific protein or of its normal cell homolog. Although the specific protein kinase that phosphorylates the Mr = 34,000 protein under conditions of EGF-stimulation is not yet identified, these studies demonstrate that at least one consequence of EGF stimulation is identical with one of the consequences of viral transformation.

Novel p19-related protein in Rous-associated virus type 61: implications for avian gag gene order.

Virions of Rous-associated virus type 61 contain a previously unrecognized p19-related protein, called p19f, which comigrates with gag protein p12 during electrophoresis in sodium dodecyl sulfate-polyacrylamide gels but can be separated by gel filtration chromatography in 6 M guanidine hydrochloride. It is shown that the existence of p19f accounts for the earlier inability to order p27 and p12 by the pactamycin mapping procedure. Remapping with pactamycin by using methods which take this new protein into account yielded a gag gene order of NH2-p219-p27-p12-p15-COOH. It also confirmed earlier positions for the env and pol genes and placed unclassified protein p10 near a translational initiation site. The pactamycin-derived mapping position of p12 differs from reports based on tryptic analysis. An analysis of procedural shortcomings emphasizes the need for more definitive determinations of the avian gag gene order.

Proteins of Rous-associated virus 61, an avian retrovirus: common precursor for glycoproteins gp85 and gp35 and use of pactamycin to map translational order of proteins in the gag, pol, and env genes.

Cells infected by Rous-associated virus 61 (RAV-61) contained a precursor-like protein, pr90, that was specifically precipitated by antiserum directed against envelope glycoproteins, gp85 and gp35. Tryptic peptide mapping showed that pr90 contained tryptic sequences of both gp85 and gp35. Pactamycin mapping experiments indicated that the two glycoproteins are translated from the env-mRNA in the order (5') gp85--gp35. The pactamycin mapping experiments also indicated a translational order of p10--(p27, p12)--p15 for the gag proteins; this agreement with the order previously reported from tryptic mapping studies on precursor pr76 of avian myeloblastosis virus implied that the stoichiometry of the core proteins was unchanged when virions were assembled in the presence of pactamycin. The reverse transcriptase proteins, unlike those of the env and gag genes, fell on the right side of the pactamycin map. This result is in accord with the idea that most, if not all, of the reverse transcriptase protein is translated by read-through of the gag(pol) message rather than by translation of a hypothetical pol-mRNA devoted solely to synthesis of that protein.