To the editor—Calleja et al. reported that the antibiotic rifampicin was capable of binding to the glucocorticoid receptor (Kd 9.9 nM) and transactivating it1, and it has been suggested that rifampicin's immunosuppressive properties may be mediated by the glucocorticoid receptor1,2.

Glucocorticoid action on pituitary corticotroph cells is well-defined, and suppression of secreted adrenocorticotrophin (ACTH) is a marker of glucocorticoid action on an endogenous gene. This effect is mediated by the binding of activated glucocorticoid receptors to two 5' promoter-negative glucocorticoid response elements on the ACTH precursor gene pro-opiomelanocortin. If rifampicin has glucocorticoid-like activity it would be expected to suppress pituitary ACTH secretion and endogenous glucocorticoid production, resulting in a compensated state. However patients on stable glucocorticoid replacement therapy treated with rifampicin do not have this confounding effect but usually require higher doses of hydrocortisone, as rifampicin is a potent inducer of hepatic glucocorticoid metabolizing enzymes.

The apparent ability of rifampicin, a molecule with no steroid homology, to activate the glucocorticoid receptor has important implications for understanding receptor physiology. Therefore, we examined the action of rifampicin (Sigma R3501) on a pituitary corticotroph cell line, AtT20. The synthetic glucocorticoid dexamethasone at a concentration of 100 nM suppressed ACTH secretion from 2651 (375 ng/l to 1024 ± 401 ng/l, whereas 100 nM rifampicin was without effect (2336 ± 142 ng/l) (Fig. 1a). Because rifampicin did not trans-repress ACTH expression, we examined whether it was able to have any effect on a well-characterized positive glucocorticoid reporter gene (MMTV-luc) in AtT20 cells. After 16 hours of incubation, 100 nM dexamethasone caused a 6.4-fold induction in reporter activity, but rifampicin (1000 nM) had no effect. The lack of effect on AtT20 cells could not be explained by metabolism of rifampicin. Although its final concentration in conditioned medium dropped from 100 nM (calculated) to 18 ± 4 nM (measured by HPLC and ultraviolet absorbance at 254 nM) after 48 hours of incubation, this is well above the previously reported Kd for receptor binding (9.9 nM).

Figure 1: a, AtT20 cells were incubated with vehicle alone, dexamethasone or rifampicin.
figure 1

ACTH concentration in conditioned medium was measured by two-site IRMA after 24 hours. b, COS 7 cells were transfected with both MMTV-luc reporter and hGR expression vectors before treatment with dexamethasone (100nM) or rifampicin (100nM) for 24 hours.

Full size image

The possibility remained that rifampicin action was dependent on cell-type specific factors in the COS 7 cells used by Calleja et al. Accordingly, the action of rifampicin was measured in COS 7 cells co-transfected with a human glucocorticoid receptor expression vector and MMTV-luc reporter (Fig. 1b). Dexamethasone at a concentration of 100 nM caused a 47-fold induction in reporter activity (from 629 ± 65 to 29742 ± 5479 light units) but rifampicin (100 nM) had no effect (624 ± 108 light units).

These data indicate that rifampicin is not a biologically significant ligand for the glucocorticoid receptor in the pituitary corticotroph cell. The earlier report by Calleja et al. may reflect the action of cell-type or promoter-type specific rifampicin metabolites or, possibly, contamination of rifampicin by other compounds.