Two bacterial species were examined for colony formation on fluoroquinolone-containing agar. One organism, Mycobacterium bovis BCG, has DNA gyrase as its primary fluoroquinolone target (3), while the other, S. aureus, has DNA topoisomerase IV as its primary target (1, 5, 6, 10). M. bovis BCG was cultured with Middlebrook 7H9 liquid medium enriched with 10% albumin-dextrose complex and 0.05% Tween 80; Middlebrook 7H10 agar plates were used for single-colony isolation (7). S. aureus was cultured in CY liquid medium and grown into colonies on GL agar plates (11). Both organisms were grown to stationary phase in liquid medium, concentrated by centrifugation (5,000 × g for 25 min), and then resuspended in fresh medium. Aliquots containing up to 10¹¹ cells were applied to agar plates containing various fluoroquinolone concentrations; plates were incubated at 37°C for 30 to 42 days for M. bovis BCG and 2 days for S. aureus. All colonies regrew on agar containing the selecting fluoroquinolone. Fluoroquinolones were obtained from Sigma Biochemical Corp. (norfloxacin), Miles Laboratories (ciprofloxacin), and Parke-Davis Pharmaceutical Co. (PD161148 and PD160793). Fluoroquinolone structures are shown in Fig. ​Fig.1.1.

As the fluoroquinolone concentration in agar plates increased, two sharp declines were observed in the fractions of CFU recovered when wild-type cells were applied to plates (Fig. ​(Fig.2).2). The first decline, which occurred around the MIC for 99% of input cells (MIC₉₉; Fig. ​Fig.2),2), we attributed to the inhibition of wild-type cell growth. Above the MIC, a concentration range existed in which the recovery of CFU declined more gradually. The concentration range spanned by this part of the response, the plateau region, varied from one fluoroquinolone to another. For example, the region occurred over a much narrower concentration range for a C-8-methoxy compound (PD161148) than for its C-8-H derivative (PD160793) or for ciprofloxacin (Fig. ​(Fig.2).2). Since the C-8-methoxy compound is more active than its C-8-H derivative or ciprofloxacin against first-step mutants (3, 14, 15), we attributed the plateau region of the curve to the presence of resistant mutants. As shown in Table ​Table1,1, a nucleotide sequence analysis of DNA from colonies recovered from ciprofloxacin-containing agar revealed alterations in the quinolone resistance-determining region of GyrA (M. bovis BCG) and ParC (GrlA; S. aureus). In the case of S. aureus, some colonies that were recovered at low concentrations of ciprofloxacin lacked an expected mutation in the quinolone resistance-determining region of ParC. Genes associated with resistance in these colonies have not yet been identified.

The second sharp drop in mutant recovery (Fig. ​(Fig.2)2) occurred once the fluoroquinolone concentration was sufficient to block the growth of first-step mutants. For example, the MIC for first-step mutants of M. bovis BCG and S. aureus occurred at about the same fluoroquinolone concentration as the second drop in mutant recovery when wild-type cells were challenged (Fig. ​(Fig.2).2). The plateau region seen with M. bovis BCG and S. aureus differed in two ways. First, the plateau extended over a greater range of fluoroquinolone concentration for M. bovis BCG. This observation is consistent with the greater difference in quinolone sensitivity between wild-type cells and first-step mutants for M. bovis BCG (13- to 90-fold [3]) than for S. aureus (about 2-fold [14]). Second, the number of mutants recovered in the plateau region was 4 orders of magnitude higher for S. aureus (Fig. ​(Fig.2),2), indicating that the S. aureus population contained many more first-step mutants. Differences in growth rates and in the target compositions of the two species may contribute to these findings: S. aureus contains both topoisomerase IV and gyrase, while M. bovis BCG probably contains only the latter (an examination of the genomic nucleotide sequence of M. tuberculosis, a closely related organism, failed to identify genes likely to encode topoisomerase IV [2]).

The addition of a C-8-methoxy group to an N-1-cyclopropyl fluoroquinolone lowered the concentration at which the second drop occurred and shortened the plateau region. The effect of adding the methoxy group was about a 10-fold decrease in the case of M. bovis BCG and about a 2.5-fold decrease for S. aureus (Fig. ​(Fig.2).2). These data suggest that the methoxy group lowers the concentration required to prevent mutants from being recovered, a parameter we call the mutant prevention concentration (MPC). The MPC is estimated by determining the minimal antibiotic concentration that results in recovery of no mutants when large numbers of cells are applied to antibiotic-containing agar plates (the use of large numbers of cells, on the order of 10¹⁰ for M. bovis BCG, ensures that the restrictive antibiotic concentration blocks the growth of first-step mutants). The MPC depends on the number of cells applied; consequently, a subscript is added to indicate the number of cells tested (e.g., MPC_10¹⁰ for 10¹⁰ cells applied to plates). This qualification allows data from different organisms, antibiotics, and laboratories to be compared. As with MIC determination, fluoroquinolone concentrations in agar plates can be standardized by having successive concentrations differ by twofold. When smaller increments are used, the MPC can be expressed as a range between the highest concentration at which mutants are recovered and the lowest at which they are not. For example, we performed five independent experiments with S. aureus and obtained the following values for the MPC_10¹⁰ of ciprofloxacin, expressed as the range defined above: 0.7 to 0.8, 0.8 to 0.9, 0.6 to 0.7, 0.8 to 0.9, and 0.8 to 0.9 μg/ml. When only the upper number in each range is considered, the standard deviation was about 10%. Fluoroquinolones that display superior bacteriostatic activity (low MIC) are often quite effective at preventing the selection of resistant mutants (low MPC). However, the relationship is not proportional (Table ​(Table2).2).