Novel non-labile covalent binding of sulfamethoxazole reactive metabolites to cultured human lymphoid cells

Abstract

Sulfamethoxazole (SMX) causes rare hypersensitivity syndrome reactions characterized by fever and multi-organ toxicity. Covalent binding of SMX reactive metabolites to cellular proteins has been demonstrated but the link between cytotoxicity and targets of covalent binding has not been explored. We therefore investigated the relationship between covalent binding of the reactive SMX-hydroxylamine (SMX-HA) metabolite, and its cytotoxicity to a hystiocytic lymphoma (U937) cell line. Incubation of U937 cells with 0–1 mM SMX-HA for 3 h resulted in dose-dependent cytotoxicity, as assessed by tetrazolium dye conversion at 24 h. SMX-HA caused dose-dependent covalent binding to cellular proteins as assessed by immunoblotting with SMX antisera at 3 and 24 h. Covalent binding was predominantly to proteins of approximately 45, 59 and 75 kDa, but other targets were also observed. The relative extent of binding to proteins was significantly different from the relative cytotoxicity at 24 h. Further, cells surviving at 24 h also had extensive covalent binding. Covalent binding was observed under reducing (β-mercaptoethanol) and non-reducing conditions to plasma membrane and microsomal but not cytosolic proteins. This non-labile covalent binding has not been previously reported. These observations suggest that extensive covalent binding does not necessarily lead to cell death, allowing the accumulation of potentially immunogenic drug-protein conjugates. These observations in whole cells may be relevant to the immunopathogenesis of SMX hypersensitivity syndrome reactions.

Introduction

The clinical use of the antimicrobial sulfamethoxazole (SMX) may result in rare, unpredictable idiosyncratic hypersensitivity syndrome reactions (HSR) characterized by fever, skin rash, lymphadenopathy, hepatotoxicity and hematological disorders [1]. The formation of reactive metabolites of the parent drug is thought to be a crucial step in determining an adverse outcome in many idiosyncratic drug reactions [2], [3]. Reactive metabolites may be directly cytotoxic or may covalently bind to cellular macromolecules to form neoantigens that initiate a pathological immune response. SMX undergoes cytochrome P450-mediated hydroxylation (CYP2C6 and 2C9 in the rat and humans, respectively) to form a hydroxylamine (SMX-HA) [4]. In vitro studies have determined that SMX-HA is toxic to lymphocytes and is also known to covalently bind to human liver microsomal proteins [1], [5]. However, SMX-HA is further oxidized to a nitroso-metabolite and it is this reactive intermediate which is thought to be responsible for the toxicity and covalent binding associated with SMX-HA¹[5], [6], [7]. Patients with idiosyncratic reactions to SMX have antibodies that recognize specific endoplasmic reticulum proteins [8] and antibodies against SMX have been identified in some patients [8], [9], [10], [11]. These results suggest that sulfonamide idiosyncratic HSR are immune-mediated.

It has been proposed that covalent binding to cellular proteins is required to initiate an idiosyncratic HSR. Covalent binding of other drugs associated with idiosyncratic toxicities, including halothane [12], diclofenac [13] and other non-steroidal anti-inflammatory drugs, dihydralazine [14], [15], and carbamazepine [16] have been studied in vivo in rats and in other animals. In these studies, extensive covalent binding was observed, without evidence of significant cellular damage in vivo and the patterns of covalent binding observed in vivo were similar to those observed after in vitro generation of reactive species. Further, antibody responses have been observed in patients that recognize proteins that are targets of covalent binding [8], [12], [14]. This has suggested a link between covalent binding to proteins and the clinical disease.

In the case of SMX, attempts to demonstrate covalent binding to liver proteins after in vivo dosing in rats were unsuccessful [5]. When rat liver slices were exposed to high concentrations of SMX or SMX-HA, a small amount of covalent binding was observed, predominantly to a band of approximately 130 kDa, however this was not associated with cytotoxicity [5]. Covalent binding of SMX to rat and human liver microsomal proteins has been demonstrated in vitro in microsomal incubations and the reactive metabolite SMX-HA has been shown to spontaneously bind to microsomal protein in the absence of further bioactivation [5]. Covalent binding to the cell surface of lymphocytes, splenocytes and keratinocytes has been demonstrated in vivo in male Wistar rats after administration of the nitroso-metabolite, although covalent binding of SMX-HA only occurred if the rats were depleted of glutathione [17]. In human studies, covalent binding to a serum protein has been observed, even in patients that do not experience HSR [18] but the role of this binding in the clinical syndrome is unclear [19]. Studies in patients have identified occasional antibodies against the drug epitope, with most patients having antibodies against the same microsomal proteins that were modified in in vitro incubations with SMX or SMX-HA [1], [8], [9]. These studies support a link between covalent binding and an immunological response. Two of the major targets of covalent binding and the antibody response have been identified as protein disulfide isomerase (PDI) and glucose regulated protein (GRP)78, both endoplasmic reticulum stress proteins [8].

While these studies support an immunological basis for HSR and an important role for covalent binding, they do not provide an explanation for individual susceptibility. Individual susceptibility to HSR has been assessed using the toxicity of sulfonamides and their reactive metabolites to isolated mononuclear leukocytes (MNL) from humans, a convenient, easily accessible cell type [20], [21], [22]. MNL from patients who have suffered sulfonamide HSR are significantly more sensitive to the cytotoxicity of SMX-HA than are MNL from controls [20], [22]. Despite extensive research to characterize the nature of these apparent defects in SMX susceptible individuals, the basis for the differential toxicity has not been determined and the mechanistic link to HSR and the immune response observed in patients remains unclear. We are therefore trying to understand the link between covalent binding in general or to specific target proteins and cytotoxicity in order to identify possible pathways for further investigation.

In summary, the steps linking covalent binding of SMX-HA to either a cytotoxic response and/or to the observed immune response [5] are not known, and how the interactions between these events may contribute to clinical disease have not been established. We therefore wished to explore in more detail the relationship between covalent binding of SMX-HA and its cytotoxicity to a lymphoid cell line as a representative of MNL, and to determine if the targets of covalent binding in intact cells were in fact similar to that observed after incubations with subcellular fractions. A human histiocytic lymphoma cell line (U937) was selected as a model for human MNL because it is susceptible to the toxicity of SMX-HA and the use of a continuous cell line will eliminate differences due to variability in primary cultures of MNL from different individuals.