Structural Mimicry of Receptor Interaction by Antagonistic Interleukin-6 (IL-6) Antibodies

Interleukin 6 plays a key role in mediating inflammatory reactions in autoimmune diseases and cancer, where it is also involved in metastasis and tissue invasion. Neutralizing antibodies against IL-6 and its receptor have been approved for therapeutic intervention or are in advanced stages of clinical development. Here we describe the crystal structures of the complexes of IL-6 with two Fabs derived from conventional camelid antibodies that antagonize the interaction between the cytokine and its receptor. The x-ray structures of these complexes provide insights into the mechanism of neutralization by the two antibodies and explain the very high potency of one of the antibodies. It effectively competes for binding to the cytokine with IL-6 receptor (IL-6R) by using side chains of two CDR residues filling the site I cavities of IL-6, thus mimicking the interactions of Phe²²⁹ and Phe²⁷⁹ of IL-6R. In the first antibody, a HCDR3 tryptophan binds similarly to hot spot residue Phe²⁷⁹. Mutation of this HCDR3 Trp residue into any other residue except Tyr or Phe significantly weakens binding of the antibody to IL-6, as was also observed for IL-6R mutants of Phe²⁷⁹. In the second antibody, the side chain of HCDR3 valine ties into site I like IL-6R Phe²⁷⁹, whereas a LCDR1 tyrosine side chain occupies a second cavity within site I and mimics the interactions of IL-6R Phe²²⁹.

Combining somatic mutations present in different in vivo affinity-matured antibodies isolated from immunized Lama glama yields ultra-potent antibody therapeutics

Highly potent human antibodies are required to therapeutically neutralize cytokines such as interleukin-6 (IL-6) that is involved in many inflammatory diseases and malignancies. Although a number of mutagenesis approaches exist to perform antibody affinity maturation, these may cause antibody instability and production issues. Thus, a robust and easy antibody affinity maturation strategy to increase antibody potency remains highly desirable. By immunizing llama, cloning the 'immune' antibody repertoire and using phage display, we selected a diverse set of IL-6 antagonistic Fabs. Heavy chain shuffling was performed on the Fab with lowest off-rate, resulting in a panel of variants with even lower off-rate. Structural analysis of the Fab:IL-6 complex suggests that the increased affinity was partly due to a serine to tyrosine switch in HCDR2. This translated into neutralizing capacity in an in vivo model of IL-6 induced SAA production. Finally, a novel Fab library was designed, encoding all variations found in the natural repertoire of VH genes identified after heavy chain shuffling. High stringency selections resulted in identification of a Fab with 250-fold increased potency when re-formatted into IgG1. Compared with a heavily engineered anti-IL-6 monoclonal antibody currently in clinical development, this IgG was at least equally potent, showing the engineering process to have had led to a highly potent anti-IL-6 antibody.

Opportunities for functional selectivity in GPCR antibodies

Monoclonal antibodies (mAbs) have been used for decades as tools to probe the biology and pharmacology of receptors in cells and tissues. They are also increasingly being developed for clinical purposes against a broad range of targets, albeit to a lesser extent for G-protein-coupled receptors (GPCRs) relative to other therapeutic targets. Recent pharmacological, structural and biophysical data have provided a great deal of new insight into the molecular details, complexity and regulation of GPCR function. Whereas GPCRs used to be viewed as having either "on" or "off" conformational states, it is now recognized that their structures may be finely tuned by ligands and other interacting proteins, leading to the selective activation of specific signaling pathways. This information coupled with new technologies for the selection of mAbs targeting GPCRs will be increasingly deployed for the development of highly selective mAbs that recognize conformational determinants leading to novel therapeutics.

Abstract 1774: Development of potent trimeric DR4 agonist Atrimers with therapeutic potential

TRAIL death receptor DR4 is a promising therapeutic target in oncology with expression in a wide variety of tumors. DR4 agonists, including TRAIL and monoclonal antibodies, can induce p53-independent apoptosis and are currently being evaluated in clinical trials in combination with chemotherapy. We aimed to surpass currently available therapeutics by developing trimeric death receptor agonists with properties expected to significantly exceed that of recombinant human TRAIL. The agonists can mimic the natural trimer-trimer interaction of the native ligand/receptor, but do not cross-react with the decoy receptors.

Potent DR4 agonist AtrimersTM were engineered using human tetranectin, a trimeric human serum protein of 60 kDa, as a scaffold. A panel of unique DR4 binders was selected from novel phage libraries displaying the C-type lectin domain (CTLD) of tetranectin containing randomized loop sequences. DR4 Atrimers have sub-nanomolar affinity to recombinant DR4-Fc and showed no detectable binding to recombinant forms of DR5 or the decoy receptors. In vitro, the DR4 Atrimers efficiently killed DR4-positive cancer cell lines with sub-nanomolar EC50, but did not kill DR4-negative cell lines. DR4 Atrimers induced cell death of DR4-expressing tumor cells through the caspase pathway, but did not kill primary human B cells and hepatocytes. Interestingly, DR4 Atrimers have differential activities on various cell lines, and also vary in their degree of internalization. While some Atrimers show potent killing of Colo-205 and are rapidly internalized, other agonist Atrimers did not show measurable internalization. These unique properties open the possibility of developing potent naked Atrimers with prolonged half-lives due to lack of internalization, as well as leveraging rapidly internalizing DR4 Atrimers for design of Atrimer-drug conjugates. Further characterization of DR4 Atrimers is ongoing in Colo-205 xenograft models.

DR4 agonist AtrimersTM with their superior potency and expected improvement in tumor penetration (vs. antibodies) represent a novel class of targeted cancer therapeutics for efficient induction of apoptosis and provide a promising approach for the treatment of a broad range of cancer types.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 1774. doi:10.1158/1538-7445.AM2011-1774

Immune modulation by melanoma and ovarian tumor cells through expression of the immunosuppressive molecule CD200

BACKGROUND AND OBJECTIVE

Immune escape by tumors can occur by multiple mechanisms, each a significant barrier to immunotherapy. We previously demonstrated that upregulation of the immunosuppressive molecule CD200 on chronic lymphocytic leukemia cells inhibits Th1 cytokine production required for an effective cytotoxic T cell response. CD200 expression on human tumor cells in animal models prevents human lymphocytes from rejecting the tumor; treatment with an antagonistic anti-CD200 antibody restored lymphocyte-mediated tumor growth inhibition. The current study evaluated CD200 expression on solid cancers, and its effect on immune response in vitro.

METHODS AND RESULTS

CD200 protein was expressed on the surface of 5/8 ovarian cancer, 2/4 melanoma, 2/2 neuroblastoma and 2/3 renal carcinoma cell lines tested, but CD200 was absent on prostate, lung, breast, astrocytoma, or glioblastoma cell lines. Evaluation of patient samples by immunohistochemistry showed strong, membrane-associated CD200 staining on malignant cells of melanoma (4/4), ovarian cancer (3/3) and clear cell renal cell carcinoma (ccRCC) (2/3), but also on normal ovary and kidney. CD200 expression on melanoma metastases was determined by RT-QPCR, and was found to be significantly higher in jejunum metastases (2/2) and lung metastases (2/6) than in normal samples. Addition of CD200-expressing, but not CD200-negative solid tumor cell lines to mixed lymphocyte reactions downregulated the production of Th1 cytokines. Inclusion of antagonistic anti-CD200 antibody restored Th1 cytokine responses.

CONCLUSION

These data suggest that melanoma, ccRCC and ovarian tumor cells can express CD200, thereby potentially suppressing anti-tumor immune responses. CD200 blockade with an antagonistic antibody may permit an effective anti-tumor immune response in these solid tumor types.

Blockade of CD200 in the presence or absence of antibody effector function: implications for anti-CD200 therapy

CD200 is an immunosuppressive molecule overexpressed in multiple hematologic malignancies such as B cell chronic lymphocytic leukemia, multiple myeloma, and acute myeloid leukemia. We previously demonstrated that up-regulation of CD200 on tumor cells suppresses antitumor immune responses and that antagonistic anti-human CD200 mAbs enabled human PBMC-mediated tumor growth inhibition in xenograft NOD/SCID human (hu)-mouse models. Ab variants with effector function (IgG1 constant region (G1)) or without effector function (IgG2/G4 fusion constant region (G2G4)) exhibited high antitumor activity in a human tumor xenograft model in which CD200 was expressed. In this report, we seek to select the best candidate to move forward into the clinic and begin to decipher the mechanisms of tumor cell killing by comparing anti-CD200-G1 vs anti-CD200-G2G4 in two related animal models. In a CD200-expressing xenograft NOD/SCID hu-mouse model where CD200 ligand/receptor interactions are already established before initiating treatment, we find that anti-CD200-G1 is a less effective Ab compared with anti-CD200-G2G4. Separately, in a model that evaluates the effect of the Abs on the immune cell component of the xenograft NOD/SCID hu-mouse model distinctly from the effects of binding to CD200 on tumor cells, we find that the administration of anti-CD200-G1 Abs completely abolished human PBMC-mediated tumor growth inhibition. Along with supporting in vitro studies, our data indicate that anti-CD200-G1 Abs efficiently mediate Ab-dependent cellular cytotoxicity of activated T cells, critical cells involved in immune-mediated killing. These studies suggest important implications regarding the selection of the constant region in anti-CD200 immunotherapy of cancer patients.

Rationale for anti-CD200 immunotherapy in B-CLL and other hematologic malignancies: new concepts in blocking immune suppression

Immune evasion in cancer is increasingly recognized as a contributing factor in the failure of a natural host antitumor immune response as well as in the failure of cancer vaccine trials. Immune evasion may be the result of a number of factors, including expansion of regulatory T cells, production of immunosuppressive cytokines, downregulation of HLA class I and tumor-associated antigens and upregulation of immunosuppressive molecules on the surface of tumor cells. CD200, a cell surface ligand that plays a role in regulating the immune system, has been shown to be upregulated on the surface of some hematologic and solid tumor malignancies. This review characterizes the role of CD200 in immune suppression, and describes strategies to target this molecule in the oncology setting, thus directly modulating immune regulation and potentially altering tolerance to tumor antigens.

In vivo targeting of antigens to human dendritic cells through DC-SIGN elicits stimulatory immune responses and inhibits tumor growth in grafted mouse models

Multiple cancer vaccine trials have been carried out using ex vivo generated autologous dendritic cells (DCs) loaded with tumor antigen before readministration into patients. Though promising, overall immunologic potency and clinical efficacy might be improved with more efficient DC-based therapies that avoid ex vivo manipulations, but are instead based on in vivo targeting of DCs. For initial in vivo proof of concept studies, we evaluated targeting of proteins or peptides to DCs through DC-specific intercellular adhesion molecule 3-grabbing nonintegrin (DC-SIGN). Because the biology of DC-SIGN is different between mice and humans, we assess human DC-SIGN targeting in the setting of elements of a human immune system in a mouse model. Administration of anti-DC-SIGN antibodies carrying either tetanus toxoid peptides or keyhole limpet hemocyanin (KLH) to Rag2gammaC mice reconstituted with human immune cells raised stimulatory human T-cell responses to the respective antigen without additional adjuvant requirements. Furthermore, administration of anti-DC-SIGN antibody-KLH conjugate enhanced the adjuvant properties of KLH resulting in inhibition of RAJI (Human Burkitt's Lymphoma Cell Line) cell tumor growth in Nonobese Diabetic/Severe Combined Immunodeficient mice transplanted with human immune cells. Thus, mouse models reconstituted with human immune cells seem to be suitable for evaluating DC-targeted vaccines, and furthermore, targeting to DCs in situ via DC-SIGN may provide a promising vaccine platform for inducing strong immune responses against cancer and infectious disease agents.

In vivo targeting of DC-SIGN-positive antigen-presenting cells in a nonhuman primate model

In vivo targeting of antigen-presenting cells (APCs) with antigens coupled to antibodies directed against APC-specific endocytic receptors is a simple and a promising approach to induce or modulate immune responses against those antigens. In a recent in vitro study, we have shown that targeting of APCs with an antigen coupled to an antibody directed against the endocytic receptor DC-SIGN effectively induces a specific immune response against that antigen. The aim of the present study was to determine the ability of the murine antihuman DC-SIGN antibody AZN-D1 to target APCs in a cynomolgus macaque model after its administration in vivo. Immunohistochemical analysis demonstrated that macaques injected intravenously with AZN-D1 have AZN-D1-targeted APCs in all lymph nodes (LNs) tested and in the liver. DC-SIGN-positive cells were mainly located in the medullary sinuses of the LNs and in the hepatic sinusoids in the liver. No unlabeled DC-SIGN molecules were found in the LN of AZN-D1-injected macaques. Morphologic criteria and staining of sequential LN sections with a panel of antibodies indicated that the DC-SIGN-targeted cells belong to the myeloid lineage of APCs. In conclusion, this is the first study that shows specific targeting of APCs in vivo by using antibodies directed against DC-SIGN.

CD200 expression on tumor cells suppresses antitumor immunity: new approaches to cancer immunotherapy

Although the immune system is capable of mounting a response against many cancers, that response is insufficient for tumor eradication in most patients due to factors in the tumor microenvironment that defeat tumor immunity. We previously identified the immune-suppressive molecule CD200 as up-regulated on primary B cell chronic lymphocytic leukemia (B-CLL) cells and demonstrated negative immune regulation by B-CLL and other tumor cells overexpressing CD200 in vitro. In this study we developed a novel animal model that incorporates human immune cells and human tumor cells to address the effects of CD200 overexpression on tumor cells in vivo and to assess the effect of targeting Abs in the presence of human immune cells. Although human mononuclear cells prevented tumor growth when tumor cells did not express CD200, tumor-expressed CD200 inhibited the ability of lymphocytes to eradicate tumor cells. Anti-CD200 Ab administration to mice bearing CD200-expressing tumors resulted in nearly complete tumor growth inhibition even in the context of established receptor-ligand interactions. Evaluation of an anti-CD200 Ab with abrogated effector function provided evidence that blocking of the receptor-ligand interaction was sufficient for control of CD200-mediated immune modulation and tumor growth inhibition in this model. Our data indicate that CD200 expression by tumor cells suppresses antitumor responses and suggest that anti-CD200 treatment might be therapeutically beneficial for treating CD200-expressing cancers.

Immune evasion by CD200: New approaches to targeted therapies for chronic lymphocytic leukemia (CLL)

2519

Background: Although the human immune system is capable of raising an immune response against many cancer types, that response is insufficient to eradicate the cancer in most patients, possibly due to immune evasion through negative regulation of the immune system by the tumor. We identified the immune-suppressive molecule CD200 to be upregulated 1.5–5.4-fold on CLL cells in all 80 patients examined. Interaction of CD200 with its receptor alters cytokine profiles from Th1 to Th2 in mixed lymphocyte reactions, and results in the induction of regulatory T cells, which are thought to hamper tumor-specific effector T cell immunity. We addressed whether CD200 expression on tumor cells plays a role in immune evasion, thereby preventing elimination of tumor cells by the immune system in a xenograft hu/SCID mouse model, and whether treatment with an antagonistic anti-CD200 antibody affects tumor growth. Methods: The human non-Hodgkins lymphoma cell lines RAJI and Namalwa were transduced with human CD200 and injected subcutaneously together with human peripheral blood lymphocytes (PBL) into NOD/SCID mice. Tumor growth over time was compared among mice that either received CD200-expressing tumor cells or received tumor cells lacking CD200 expression. In subsequent experiments, mice were treated with chimeric or humanized anti-CD200 antibodies (doses ranged from 1 to 20 mg/kg) by intravenous injection. Treatment was either started immediately or 7 days after tumor cell injection. Results: As expected, PBLs reduced CD200-negative RAJI or Namalwa tumor growth by up to 75%. In contrast, growth of RAJI or Namalwa tumors expressing CD200 at levels comparable to that of CLL was not reduced by PBLs. Administration of anti-CD200 antibodies at 5 mg/kg resulted in nearly complete tumor growth inhibition (1/10 mice developed a small tumor) over the course of the study even when treatment was started 7 days after tumor cell injection. Conclusions: CD200 expression on tumor cells inhibits the ability of human lymphocytes to eradicate tumor cells. Treatment of CD200-expressing tumors with antagonistic anti-CD200 antibodies inhibits tumor growth, indicating the potential for anti-CD200 therapy as a promising approach for CLL.

No significant financial relationships to disclose.

Immune evasion by melanoma and ovarian tumor cells through upregulation of the immunosuppressive molecule CD200

2545

Background: Immune escape by tumors can occur by multiple mechanisms, each a significant barrier to immunotherapy. Upregulation of the immunosuppressive molecule CD200 on chronic lymphocytic leukemia cells inhibits Th1 cytokine production required for an effective cytotoxic T cell response. CD200 expression on human tumor cells in animal models prevents human lymphocytes from rejecting the tumor; treatment with an antagonistic anti-CD200 antibody restored lymphocyte-mediated tumor growth inhibition. This study evaluated CD200 expression on other cancers, and its effect on immune response. Methods: CD200 levels in ovarian adenocarcinoma and metastatic melanoma samples were evaluated by RT-QPCR and immunohistochemistry. Cell-surface CD200 on melanoma and ovarian cancer cell lines was assessed by flow cytometry. The effect of CD200 on cytokine production in mixed lymphocyte reactions (MLR) was assessed by adding the cells to cultures containing human monocyte-derived dendritic cells and allogeneic T cells. Th1 and Th2 cytokines in culture supernatants were detected by ELISA. Results: RT-QPCR showed CD200 expression levels upregulated in serous ovarian adenocarcinoma compared to normal samples. In malignant melanoma, CD200 expression in jejunum metastases was significantly higher than in normal samples, and 2 of 6 lung metastases showed CD200 upregulation. IHC showed strong, membrane-associated CD200 staining on malignant cells of two melanoma patients. Three ovarian cancer patients showed varying levels of CD200 tumor staining; all showed strong stromal staining. CD200 was highly expressed on the cell surface of SK-MEL-24 and SK-MEL-28 melanoma and OV-CAR-3 ovarian cancer cell lines and moderately expressed on the melanoma cell line SK-MEL-5. Addition of these cell lines to MLRs downregulated the production of Th1 cytokines; addition of CD200-negative cell lines did not. Inclusion of an antagonistic anti-CD200 antibody during the culture restored Th1 cytokine responses. Conclusion: Melanoma and ovarian tumor cells can upregulate CD200, thereby potentially suppressing anti-tumor immune responses. Therapy with an antagonistic anti-CD200 antibody may permit an effective cytotoxic immune response against the tumor cells.

[Table: see text]

Internalizing antibodies to the C-type lectins, L-SIGN and DC-SIGN, inhibit viral glycoprotein binding and deliver antigen to human dendritic cells for the induction of T cell responses

The C-type lectin L-SIGN is expressed on liver and lymph node endothelial cells, where it serves as a receptor for a variety of carbohydrate ligands, including ICAM-3, Ebola, and HIV. To consider targeting liver/lymph node-specific ICAM-3-grabbing nonintegrin (L-SIGN) for therapeutic purposes in autoimmunity and infectious disease, we isolated and characterized Fabs that bind strongly to L-SIGN, but to a lesser degree or not at all to dendritic cell-specific ICAM-grabbing nonintegrin (DC-SIGN). Six Fabs with distinct relative affinities and epitope specificities were characterized. The Fabs and those selected for conversion to IgG were tested for their ability to block ligand (HIV gp120, Ebola gp, and ICAM-3) binding. Receptor internalization upon Fab binding was evaluated on primary human liver sinusoidal endothelial cells by flow cytometry and confirmed by confocal microscopy. Although all six Fabs internalized, three Fabs that showed the most complete blocking of HIVgp120 and ICAM-3 binding to L-SIGN also internalized most efficiently. Differences among the Fab panel in the ability to efficiently block Ebola gp compared with HIVgp120 suggested distinct binding sites. As a first step to consider the potential of these Abs for Ab-mediated Ag delivery, we evaluated specific peptide delivery to human dendritic cells. A durable human T cell response was induced when a tetanus toxide epitope embedded into a L-SIGN/DC-SIGN-cross-reactive Ab was targeted to dendritic cells. We believe that the isolated Abs may be useful for selective delivery of Ags to DC-SIGN- or L-SIGN-bearing APCs for the modulation of immune responses and for blocking viral infections.

Antibodies selected from combinatorial libraries block a tumor antigen that plays a key role in immunomodulation

We searched for cell-surface-associated proteins overexpressed on B cell chronic lymphocytic leukemia (CLL) to use as therapeutic antibody targets. Antibodies binding the immunosuppressive molecule CD200 were identified by cell panning of an antibody phage display library derived from rabbits immunized with primary CLL cells. B cells from 87 CLL patients exhibited 1.6- to 5.4-fold cell-surface up-regulation of CD200 relative to normal B cells. An effect of increased CD200 expression by CLL cells on the immune system was evaluated in mixed lymphocyte reactions. Addition of primary CLL but not normal B cells to macrophages and T cells downregulated the Th1 response, as seen by a 50-95% reduction in secreted IL-2 and IFN-gamma. Antibodies to CD200 prevented downregulation of the Th1 response in most B cell CLL samples evaluated, indicating abrogation of the CD200/CD200R interaction can be sufficient to restore the Th1 response. A disease-progression-associated shift of the immune response from Th1 to Th2 has been observed in numerous cancers. Because this cytokine shift is also believed to promote the induction of regulatory T cells, reverting the immune response to Th1 through direct targeting of the cancer cells may provide therapeutic benefits in CLL by encouraging a cytotoxic T cell response.

A nondeletional mechanism for central T-cell tolerance

To be positively selected, immature thymocytes must receive signaling through their T-cell receptor (TCR), and engagement of relatively low-affinity self-peptides permits further T-cell maturation. However, mature T cells no longer overtly respond to such low-affinity antigens, indicating that T cells acquire a higher threshold for activation during thymopoiesis. We wondered whether partial interference in positive selection could produce T cells that respond to the selecting self-peptide. This possibility was tested by injecting procainamide-hydroxylamine (PAHA), a lupus-inducing drug, into the thymus of adult normal mice. Three weeks after the second injection, IgG antichromatin antibodies appeared in the circulation and remained for several months. The murine antichromatin antibodies reacted with the (H2A-H2B)-DNA subnucleosome complex, the predominant specificity in patients with procainamide-induced lupus. In thymus organ and reaggregate cultures, PAHA had no effect on negative selection of T cells with high affinity for a co-present antigen, but acted on CD4+ CD8+ immature T cells as they underwent positive selection. TCR transgenic T cells specific to cytochrome c peptide 88-104 acquired the capacity to respond to the low-affinity analogue at position 99 (lys-->ala) if PAHA was present during their development. PAHA also blocked the capacity of a T-cell line to become anergic after anti-CD3 treatment, suggesting that PAHA prevents the production of negative regulators that accumulate in response to partial signaling through the TCR. These results are consistent with the view that T cells acquire self-tolerance during positive selection, and disruption of this process can result in autoreactive T cells and systemic autoimmunity.

Early cellular events in systemic autoimmunity driven by chromatin-reactive T cells

In vivo exposure of the thymus of normal mice to procainamide-hydroxylamine, a lupus-inducing drug, causes development of chromatin-reactive T cells. Autoantibodies subsequently appear, but their origin and significance are unknown. The current studies were undertaken to determine the specificities of B cells that respond to chromatin-reactive T cells at the initiation of this autoimmune process. Three days after adoptive transfer of 6 x 10(6) chromatin-reactive T cells, B cells with the capacity to secrete IgM anti-chromatin antibodies were detected in 1/10(6) splenocytes, and these became 10- to 50-fold more numerous if either the donor T cells or the recipient had defective Fas due to the lpr allele. Five days later these mice developed IgG anti-chromatin-secreting B cells at a precursor frequency of 3-6 x 10(-5). B cells with dDNA-binding activity isolated from mice primed in vivo to a complex of methylated pigeon cytochrome c and dDNA could stimulate naive, cytochrome c-reactive T cells in vitro, demonstrating that B cells can internalize dDNA-bound proteins through their dDNA immunoblobulin receptor and can functionally present a T cell epitope. However, no capacity of chromatin for binding anti-dDNA antibodies was detected, and IgM dDNA-specific B cells did not expand when challenged with chromatin-reactive T cells in vivo. The rapid and robust expansion of anti-chromatin-secreting B cells indicates that the normal immune repertoire includes nontolerant autoreactive B cells that respond to strong T cell drive and are readily manifested if Fas-mediated activation-induced cell death is inhibited.

Phagocyte-mediated oxidation in idiosyncratic adverse drug reactions

The drug-metabolizing capacity of the liver is well known but cannot account for most idiosyncratic adverse drug reactions. Of the extrahepatic sources of reactive drug metabolites, the neutrophil has received the most attention because of its vast numbers and robust oxidizing machinery. Many drugs associated with autoimmunity are susceptible to oxidative transformation by the enzymatic action of myeloperoxidase, a protein released into the extracellular environment when neutrophils are activated. Production of the resulting drug metabolites within lymphoid organs maximizes their immune-perturbing effects. Mechanisms proposed for the initiation of drug-induced blood dyscrasias, hypersensitivity reactions, or lupus-like symptoms center around three views: (1) presentation of the implicated compound in the major histocompatibility complex of antigen-presenting cells via direct binding or after processing as a hapten bound to self-macromolecules, (2) direct cytotoxicity, or (3) interference in the development of T-cell tolerance in the thymus. How participation of reactive drug metabolites in these processes might lead to symptomatic disease is discussed.

Disruption of positive selection of thymocytes causes autoimmunity

To differentiate into T cells, immature thymocytes must engage, through their antigen-specific T-cell receptor, peptides derived from self proteins presented by cortical epithelial cells in the thymus, a process called positive selection. Despite this requirement for self-recognition during development, mature T cells do not normally show autoreactivity. Mice injected in the thymus with procainamide-hydroxylamine, a metabolite of procainamide, develop autoimmune features resembling drug-induced lupus. Here, we show that when thymocytes undergo positive selection in the presence of procainamide-hydroxylamine, they fail to establish unresponsiveness to low affinity selecting self antigens, resulting in systemic autoimmunity.

Initiation of autoimmunity by a reactive metabolite of a lupus-inducing drug in the thymus

Drug-induced lupus is a side effect of deliberate ingestion of various medications, but its etiology, underlying mechanisms, and pathogenesis are puzzling. In vivo metabolic transformation of lupus-inducing drugs to reactive products explains how a heterogeneous set of drugs can mediate the same disease syndrome. Evidence has accumulated that drugs are transformed by extracellular oxidation from reactive oxygen species and myeloperoxidase produced when neutrophils are activated, maximizing the in situ accumulation of reactive drug metabolites within lymphoid compartments. The metabolite of procainamide, procainamide hydroxylamine, displays diverse biologic properties, but no apparent autoimmune effect has been observed. However, when procainamide hydroxylamine was introduced into the thymus of young adult normal mice, a delayed but robust autoimmune response developed. Disruption of central T-cell tolerance by intrathymic procainamide hydroxylamine resulted in the production of chromatin-reactive T cells that apparently drove the autoantibody response in the periphery. Drug-induced autoantibodies in this mouse model were remarkably similar to those in patients with procainamide-induced lupus. Therefore, this system has considerable promise to provide insight into the initiating events in drug-induced lupus and may provide a paradigm for how other xenobiotics might induce systemic autoimmunity.

Linkage of immune self-tolerance with the positive selection of T cells

Development and maturation of antigen-specific T cells take place in the thymus in a process dependent on recognition by the T cell antigen receptor (TCR) of endogenous self-peptides presented by several types of specialized stromal cells. Paradoxically, emerging T cells are not self-reactive, and it is commonly believed that deletion of high avidity autoreactive T cells is the principal mechanism for establishing self-tolerance. However, there is increasing evidence that the positive selection of T cells on self-peptides presented by thymic cortical epithelial cells must be linked with a process that prevents their subsequent activation when the same self-peptides are encountered in the periphery. Consequently, a higher activation threshold is established that can be overcome only with ligands of higher affinity, which would normally be foreign peptides. The molecular basis for this increase in activation threshold is unknown, but observations on differential signaling by peptide analogs, on increased TCR expression during T cell maturation and on energy induction in the absence of costimulation provide promising leads. Linkage of self-tolerance with positive selection is a simple and evolutionary sound explanation for self/non-self discrimination and offers a framework for understanding systemic autoimmunity.

Persistence of autoreactive T cell drive is required to elicit anti-chromatin antibodies in a murine model of drug-induced lupus

Long-term treatment with procainamide and numerous other medications is occasionally associated with the development of drug-induced lupus. We recently established a murine model for this syndrome by disrupting central T cell tolerance. Two intrathymic injections of procainamide-hydroxylamine (PAHA), a reactive metabolite of procainamide, into (C57BL/6 x DBA/2)F1 mice resulted in the appearance of chromatin-reactive T cells and anti-chromatin autoantibodies. The current study explores in this model the role of autoreactive T cells in autoantibody production and examines why autoantibodies after a single intrathymic drug injection were much more limited in isotype and specificity. Injection of as few as 5000 chromatin-reactive T cells into naive, syngeneic mice induced a rapid IgM anti-denatured DNA response, while injection of at least 100-fold greater number of activated T cells was required for induction of IgG anti-chromatin Abs, suggesting that small numbers of autoreactive T cells can be homeostatically controlled. Mice subjected to a single intrathymic PAHA injection after receiving splenic B cells from an intrathymic PAHA-injected syngeneic donor also developed anti-chromatin Abs, but adoptive transfer of similarly primed T cells or of B cells without intrathymic PAHA injection of the recipient failed to produce an anti-chromatin response. However, anti-chromatin Abs developed after a single intrathymic PAHA injection in Fas-deficient C57BL/6-lpr/lpr mice, suggesting that activation-induced cell death limited autoimmunity in normal mice. Taken together, these results imply that chromatin-reactive T cells produced by intrathymic PAHA created a B cell population primed to somatically mutate and Ig class switch when subjected to a heavy load or second wave of autoreactive T cells.

Persistence of Autoreactive T Cell Drive Is Required to Elicit Anti-Chromatin Antibodies in a Murine Model of Drug-Induced Lupus

Long-term treatment with procainamide and numerous other medications is occasionally associated with the development of drug-induced lupus. We recently established a murine model for this syndrome by disrupting central T cell tolerance. Two intrathymic injections of procainamide-hydroxylamine (PAHA), a reactive metabolite of procainamide, into (C57BL/6 × DBA/2)F1 mice resulted in the appearance of chromatin-reactive T cells and anti-chromatin autoantibodies. The current study explores in this model the role of autoreactive T cells in autoantibody production and examines why autoantibodies after a single intrathymic drug injection were much more limited in isotype and specificity. Injection of as few as 5000 chromatin-reactive T cells into naive, syngeneic mice induced a rapid IgM anti-denatured DNA response, while injection of at least 100-fold greater number of activated T cells was required for induction of IgG anti-chromatin Abs, suggesting that small numbers of autoreactive T cells can be homeostatically controlled. Mice subjected to a single intrathymic PAHA injection after receiving splenic B cells from an intrathymic PAHA-injected syngeneic donor also developed anti-chromatin Abs, but adoptive transfer of similarly primed T cells or of B cells without intrathymic PAHA injection of the recipient failed to produce an anti-chromatin response. However, anti-chromatin Abs developed after a single intrathymic PAHA injection in Fas-deficient C57BL/6-lpr/lpr mice, suggesting that activation-induced cell death limited autoimmunity in normal mice. Taken together, these results imply that chromatin-reactive T cells produced by intrathymic PAHA created a B cell population primed to somatically mutate and Ig class switch when subjected to a heavy load or second wave of autoreactive T cells.

Selective protein adduct formation of diclofenac glucuronide is critically dependent on the rat canalicular conjugate export pump (Mrp2)

Previous work demonstrates that the reactive acyl glucuronide of the nonsteroidal antiinflammatory drug diclofenac forms selective protein adducts in the liver, which may play a causal role in the pathogenesis of diclofenac-associated liver toxicity. Because glucuronide conjugates can be exported into the bile, we explored the role of diclofenac glucuronide hepatobiliary transport in the formation of site-specific protein adducts. Specifically, to analyze intracellular (hepatocytes) versus extracellular (biliary tree) targeting of proteins, we have compared the pattern of diclofenac binding in normal Wistar rats with that in mutant transport-deficient (TR-) rats which lack the functional canalicular isoform of the conjugate export pump, Mrp2. In bile duct-cannulated normal rats, >50% of an iv injected dose of [14C]diclofenac appeared in the bile over a 90-min period. In contrast, in TR- rats virtually no hepatobiliary excretion of diclofenac glucuronide was found. After administration of diclofenac (30 mg/kg/day, ip for 3 days) to rats of both genotypes, a major protein adduct of an apparent molecular mass of 118 kDa was selectively detected by immunoblotting in isolated canalicular, but not in basolateral, membrane subfractions of wild-type rats, whereas no plasma membrane adducts could be found in the livers of TR- rats. Furthermore, immunohistochemical analysis using an anti-diclofenac antibody revealed the presence of distinct diclofenac-modified proteins on canalicular membranes of liver sections from diclofenac-treated normal rats, whereas no adducts could be identified in livers of TR- rats. In Western blots, the major diclofenac-modified canalicular membrane protein did not comigrate with Mrp2, indicating that the glucuronide carrier itself was unlikely to be a target. Collectively, the results demonstrate that the reactive diclofenac glucuronide is selectively transported into bile via Mrp2 and that hepatobiliary transport is critical for diclofenac covalent binding to proteins in the biliary tree.

Selective protein adduct formation of diclofenac glucuronide is critically dependent on the rat canalicular conjugate export pump (Mrp2)

Previous work demonstrates that the reactive acyl glucuronide of the nonsteroidal antiinflammatory drug diclofenac forms selective protein adducts in the liver, which may play a causal role in the pathogenesis of diclofenac-associated liver toxicity. Because glucuronide conjugates can be exported into the bile, we explored the role of diclofenac glucuronide hepatobiliary transport in the formation of site-specific protein adducts. Specifically, to analyze intracellular (hepatocytes) versus extracellular (biliary tree) targeting of proteins, we have compared the pattern of diclofenac binding in normal Wistar rats with that in mutant transport-deficient (TR-) rats which lack the functional canalicular isoform of the conjugate export pump, Mrp2. In bile duct-cannulated normal rats, >50% of an iv injected dose of [14C]diclofenac appeared in the bile over a 90-min period. In contrast, in TR- rats virtually no hepatobiliary excretion of diclofenac glucuronide was found. After administration of diclofenac (30 mg/kg/day, ip for 3 days) to rats of both genotypes, a major protein adduct of an apparent molecular mass of 118 kDa was selectively detected by immunoblotting in isolated canalicular, but not in basolateral, membrane subfractions of wild-type rats, whereas no plasma membrane adducts could be found in the livers of TR- rats. Furthermore, immunohistochemical analysis using an anti-diclofenac antibody revealed the presence of distinct diclofenac-modified proteins on canalicular membranes of liver sections from diclofenac-treated normal rats, whereas no adducts could be identified in livers of TR- rats. In Western blots, the major diclofenac-modified canalicular membrane protein did not comigrate with Mrp2, indicating that the glucuronide carrier itself was unlikely to be a target. Collectively, the results demonstrate that the reactive diclofenac glucuronide is selectively transported into bile via Mrp2 and that hepatobiliary transport is critical for diclofenac covalent binding to proteins in the biliary tree.

A metabolite of the lupus-inducing drug procainamide prevents anergy induction in T cell clones

Therapeutic treatment with procainamide is occasionally associated with the development of drug-induced lupus. This syndrome has become the prototype for an aseptic systemic autoimmune disease caused by a known environmental agent, but the underlying mechanisms remain puzzling. We explored the possibility that lupus-inducing drugs affect processes involved in T cell tolerance to self-Ag. An in vitro model of anergy using established T cell clones was used to determine whether procainamide or one of its metabolites could prevent development of T cell nonresponsiveness to cognate Ag. Addition of procainamide-hydroxylamine, but not procainamide or its further oxidation products during anergy induction by CD3 engagement, caused a dose-dependent recovery of the capacity of T cells to proliferate and secrete IFN-gamma upon subsequent Ag challenge. Resistance to anergy induction required 2 h of exposure to procainamide-hydroxylamine, and this state remained for 8 h, suggesting that uptake of the drug caused a reversible interference in signaling pathways involved in establishing anergy. We suggest that prevention of anergy induction by procainamide-hydroxylamine may also take place in vivo during establishment of T cell tolerance to self-Ag, thereby allowing the production of autoreactive T cells.

A metabolite of the lupus-inducing drug procainamide prevents anergy induction in T cell clones

Therapeutic treatment with procainamide is occasionally associated with the development of drug-induced lupus. This syndrome has become the prototype for an aseptic systemic autoimmune disease caused by a known environmental agent, but the underlying mechanisms remain puzzling. We explored the possibility that lupus-inducing drugs affect processes involved in T cell tolerance to self-Ag. An in vitro model of anergy using established T cell clones was used to determine whether procainamide or one of its metabolites could prevent development of T cell nonresponsiveness to cognate Ag. Addition of procainamide-hydroxylamine, but not procainamide or its further oxidation products during anergy induction by CD3 engagement, caused a dose-dependent recovery of the capacity of T cells to proliferate and secrete IFN-gamma upon subsequent Ag challenge. Resistance to anergy induction required 2 h of exposure to procainamide-hydroxylamine, and this state remained for 8 h, suggesting that uptake of the drug caused a reversible interference in signaling pathways involved in establishing anergy. We suggest that prevention of anergy induction by procainamide-hydroxylamine may also take place in vivo during establishment of T cell tolerance to self-Ag, thereby allowing the production of autoreactive T cells.

Autoimmunity caused by disruption of central T cell tolerance. A murine model of drug-induced lupus

A side effect of therapy with procainamide and numerous other medications is a lupus-like syndrome characterized by autoantibodies directed against denatured DNA and the (H2A-H2B)-DNA subunit of chromatin. We tested the possibility that an effect of lupus-inducing drugs on central T cell tolerance underlies these phenomena. Two intrathymic injections of procainamide-hydroxylamine (PAHA), a reactive metabolite of procainamide, resulted in prompt production of IgM antidenatured DNA antibodies in C57BL/6xDBA/2 F1 mice. Subsequently, IgG antichromatin antibodies began to appear in the serum 3 wk after the second injection and were sustained for several months. Specificity, inhibition and blocking studies demonstrated that the PAHA-induced antibodies showed remarkable specificity to the (H2A-H2B)-DNA complex. No evidence for polyclonal B cell activation could be detected based on enumeration of Ig-secreting B cells and serum Ig levels, suggesting that a clonally restricted autoimmune response was induced by intrathymic PAHA. The IgG isotype of the antichromatin antibodies indicated involvement of T cell help, and proliferative responses of splenocytes to oligonucleosomes increased up to 100-fold. As little as 5 microM PAHA led to a 10-fold T cell proliferative response to chromatin in short term organ culture of neonatal thymi. We suggest that PAHA interferes with self-tolerance mechanisms accompanying T cell maturation in the thymus, resulting in the emergence of chromatin-reactive T cells followed by humoral autoimmunity.

Cytotoxic activity of T cells and non-T cells from diclofenac-immunized mice against cultured syngeneic hepatocytes exposed to diclofenac

To evaluate whether hepatocellular protein adducts of the nonsteroidal antiinflammatory drug diclofenac could elicit a specific cell-mediated or antibody-dependent immune response that eventually results in liver cell destruction, we developed a murine ex vivo/in vitro mixed lymphocyte hepatocyte culture (MLHC) model. C57BL/6 mice were immunized either with diclofenac conjugated to keyhole limpet hemocyanin (KLH) or with KLH alone. Splenocytes from mice exhibiting hgih antidiclofenac antibody titers were isolated and co-cultured at an effector/target cell ratio of 100:1 with syngeneic murine hepatocytes preexposed to diclofenac. By 48 and 72 hours, extracellular alanine transaminase (ALT) activity had increased 6.4- and 7.6-fold, respectively, versus the 24-hour value. In contrast, there was no significant cytotoxic response after either drug treatment alone or immunization alone. Furthermore, those cellular populations capable of inducing ALT release also showed lymphocyte stimulation as determined by interleukin-2 (IL-2) receptor expression and lymphocyte proliferation analysis. The extent of cell injury was highest in the presence of lymphocytes highly enriched in T cells and was reduced by 40% in the presence of anti-MHC I antibodies. Similarly, albeit to a lesser extent, non-T cell-enriched lymphocyte fractions also induced hepatocyte injury. The addition of co-culture supernatants to hepatocytes had no effect, thus ruling out the possibility that soluble factors alone mediated the cell injury. However, supernatants from diclofenac-stimulated lymphocytes, combined with nonstimulated splenocytes, triggered an immediate (< 1 hour) cytotoxic response, suggesting antibody-dependent cell-mediated mechanisms of target cell injury. These results indicate that diclofenac-treated hepatocytes carried antigenic determinants that were recognized by T cells and non-T cells derived from diclofenac/KLH-immunized mice, resulting in cell-mediated destruction of the target hepatocytes.

Idiosyncratic liver toxicity of nonsteroidal antiinflammatory drugs: molecular mechanisms and pathology

This review explores the clinical hepatic pathology associated with the use of nonsteroidal antiinflammatory drugs (NSAIDs), possible cellular and molecular mechanisms of injury, and future challenges. NSAIDs comprise a group of widely used compounds that have been associated with rare adverse reactions in the liver, including fulminant hepatitis and cholestasis. These reactions are idiosyncratic, mostly independent of the dose administered, and host-dependent. The mechanisms responsible for the initiation and perpetuation of NSAID-induced hepatotoxicity remain poorly understood and have been largely inferred from clinical manifestation. A mounting body of evidence, however, indicates that many acidic NSAIDs are metabolized to reactive acyl glucuronides that can form covalent adducts with plasma proteins and hepatocellular proteins. In hepatocytes cocultured with lymphocytes, these NSAID-altered proteins can become antigenic. Thus, long-lived, drug-altered proteins may act as immunogens and produce cytotoxic T-cell-mediated or antibody-dependent, cell-mediated toxicity in susceptible patients. Alternatively, individual abnormalities in metabolism or disposition of some NSAIDs may lead to the formation or accumulation of toxic metabolites. Additional work with transgenic animal models is needed to permit better understanding of the general and specific risk factors involved in the pathogenesis of the idiosyncratic liver injuries related to NSAIDs and other drugs.

Mechanism of covalent adduct formation of diclofenac to rat hepatic microsomal proteins. Retention of the glucuronic acid moiety in the adduct

The nonsteroidal antiinflammatory drug diclofenac can bind irreversibly to hepatocellular proteins via its acyl glucuronide metabolite. In view of a possible involvement of these protein adducts in the pathogenesis of diclofenac-associated liver damage, we investigated the mechanism of adduct formation in rat hepatic microsomes. [14C]Diclofenac covalently bound to hepatic microsomal proteins as a function of exposure time and the concentration of the cofactor, UDP-glucuronic acid (UDPGA). The rate and extent of protein binding were significantly increased in the presence of the imine-trapping agent, sodium cyanide. Moreover, hepatic microsomes incubated with [14C]UDPGA and nonradiolabeled diclofenac resulted in similar covalent binding of the radiolabeled compound to microsomal proteins. Covalent binding of [14C]UDPGA was significantly decreased in the presence of 7,7,7-triphenylheptyl-UDP, a specific inhibitor of UDP-glucuronosyltransferase. Finally, the protein adducts formed after incubation with both the radiolabeled aglycone and radiolabeled glucuronic acid were resolved by sodium dodecyl sulfate gel electrophoresis. Under both conditions, a radiolabeled protein band of apparent M(r) 60 kDa was found by fluorographic analysis. These results indicate that diclofenac acyl glucuronide covalently binds to hepatic microsomal proteins by two apparent mechanisms. Besides nucleophilic displacement of the glucuronic acid, the open-chain glucuronic acid can form an imine bond with a nucleophilic site of the target protein and is thus retained within the adduct.

Selective protein adducts to membrane proteins in cultured rat hepatocytes exposed to diclofenac: radiochemical and immunochemical analysis

The nonsteroidal anti-inflammatory drug diclofenac can be bioactivated to the reactive acyl glucuronide, which covalently binds to hepatocellular proteins in rat hepatocytes. Short term cultured rat hepatocytes were used to further study the formation and nature of protein adducts after exposure to diclofenac. Incubation of cells with [14C]diclofenac (30 microM) for up to 24 hr was associated with a time-dependent increase in radioactivity bound to proteins. Upon subcellular fractionation of hepatocytes exposed to diclofenac for 2 hr, the majority of the radiolabel appeared in the microsomal fraction. By 24 hr, the specific binding had decreased by 50% in this cell compartment. In contrast, the hepatocellular plasma membrane fraction, which also was associated with high specific binding of diclofenac-derived radioactivity by 2 hr, exhibited a approximately 3-fold increase in adduct formation by 24 hr. Lesser amounts of radioactivity were associated with cytosolic proteins. After resolution of the proteins by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and fluorography, the radioactivity was associated with a major protein band with an apparent molecular mass of 60 kDa that was present in both microsomes and plasma membranes. Further, we developed an antidiclofenac antibody against diclofenac-protein adducts by Protein A chromatography of a polyclonal antiserum raised in rabbits against a diclofenac-keyhole limpet hemocyanin adduct. The antidiclofenac antibody did recognize diclofenac-protein adducts on Western blots of homogenates of cultured rat hepatocytes exposed to diclofenac. The major detected adducts included the 60-kDa protein, which was present at all diclofenac concentrations used. In addition, the antibody recognized proteins with apparent molecular masses of 50, 80, and 126 kDa that were not evident in the radiochemical assay. There were no detectable cross-reactive epitopes of proteins recognized by the antibody on Western blots of cultured hepatocytes not treated with diclofenac. Moreover, immunoblots of liver homogenates from rats treated with diclofenac (30 mg/kg/day, intraperitoneally, for 4 days) also exhibited adducts with the 60- and 80-kDa proteins. Collectively, these results suggest that binding of diclofenac to rat hepatocyte proteins is selective and that a 60-kDa microsomal membrane protein (or protein subunit) that accumulates in the plasma membrane fraction appears to be the major target for alkylation both in cultured hepatocytes exposed to diclofenac and in vivo.

Diclofenac covalent protein binding is dependent on acyl glucuronide formation and is inversely related to P450-mediated acute cell injury in cultured rat hepatocytes

In a few patients diclofenac produces mild increases in serum aminotransferase activity and in rare cases may be associated with the occurrence of fulminant hepatic necrosis. Both direct toxic effects of a diclofenac metabolite and hypersensitivity reactions have been suggested as possible molecular mechanisms of liver injury. We investigated the pathways of bioactivation and cytotoxicity of diclofenac, which undergoes both aromatic hydroxylation and acyl glucuronidation, in short-term cultured rat hepatocytes. LDH release was first evident after 4 hr of incubation with diclofenac (> 500 microM). In addition, time- and concentration-dependent covalent binding of [14C]diclofenac to hepatocellular proteins occurred, indicating the presence of a reactive intermediate. To specifically explore the role of the acyl glucuronidation pathway in the induction of cytotoxicity and covalent drug-protein adducts, we used two inhibitors of the UDP-glucuronosyltransferase (UDPGT), borneol and 7,7,7-triphenylheptyl-UDP. LDH release was markedly increased in the presence of either UDPGT inhibitor. Alternatively, covalent binding to hepatocellular proteins was greatly reduced when the glucuronide formation was selectively blocked. Furthermore, in vitro inhibition of P450-dependent oxidative biotransformation with the selective inhibitor of the CYP2C subfamily sulfaphenazole or with cimetidine markedly reduced the extent of cytotoxicity, whereas the degree of covalent adduct formation remained unchanged. Similarly, pretreatment of the rats with phenobarbital (80 mg/kg/day for 4 days) delayed the onset and reduced the extent of diclofenac-induced LDH release. Collectively, these results indicate that the formation of a toxic diclofenac metabolite(s) catalyzed by P4502C in hepatocytes leads to acute lethal cell injury, whereas diclofenac acyl glucuronide formation is associated with covalent binding of a reactive metabolite to hepatocellular proteins that is not related to the acute cytotoxicity. The protein adduct formation and its modulation by UDPGT may, however, be toxicologically relevant for the expression of diclofenac hepatitis.