INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Quantitative results obtained by in vitro antimicrobial susceptibility testing are very important for proper management with antimicrobial chemotherapy for patients with serious infectious diseases (29). At present, a variety of laboratory methods are used for susceptibility testing. It is well known that the results of susceptibility tests are influenced by inoculum size, composition of the medium, concentrations of divalent cations, calcium, and magnesium, pH, and incubation conditions. When the final results of susceptibility tests are influenced by methodology, interpretation errors with serious clinical implications may occur. To avoid such risks, a series of procedures must be standardized to ensure accurate and reproducible results. Such efforts to develop standardized procedures for susceptibility tests have been reported (25). In the United States, the National Committee for Clinical Laboratory Standards develops standardized procedures, which are published in approved or proposed form, for susceptibility testing (20, 21).

The broth and agar dilution methods are tests commonly used to quantitatively measure the in vitro activities of antimicrobial agents against a given bacterial isolate. Recently, we determined the MICs of expanded-spectrum cephems against clinical isolates of Proteus vulgaris by using both the broth microdilution method and the agar dilution method (A. Ohno, M. Datz, Y. Ishii, L. Ma, and K. Yamaguchi, Abstr. 38th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 101, 1998). Surprisingly, for several oximino-type expanded-spectrum cephems, the majority of MICs determined by the broth microdilution method were over 10 dilutions higher than those obtained by the agar dilution method. Inevitably, MIC results produced by the broth dilution method were considered resistant, but MIC results obtained by the agar dilution method were interpreted as susceptible. This phenomenon was tentatively designated the MIC gap phenomenon (Ohno et al., 38th ICAAC).

P. vulgaris produces class A -lactamases (CumA). The cumA gene exists on the chromosome (24). CumA is very different from TEM- or SHV-type class A -lactamases in that oximino-type expanded-spectrum cephems are preferred substrates for CumA (2, 12, 27). Therefore, the MIC gap phenomenon may be due to the activity of CumA. On the other hand, CumA is inducible, as are class C -lactamases (AmpC) produced by most gram-negative bacteria. The induction of AmpC is closely linked to the recycling system for muramyl peptides released from the bacterial peptidoglycan (15). Factors involved in the peptidoglycan recycling system are AmpG, a transmembrane protein which acts as a permease for GlcNAcanhMurNAc-tripeptide (18) or -pentapeptide (5) degraded enzymatically from the peptidoglycan, and AmpD, a cytosolic N-acetylmuramyl-L-alanine amidase (once inside, these muramyl peptides are cleaved by AmpD, and tripeptide or pentapeptide is released and recycled as the precursor of peptidoglycan [14]). -Lactams result in increased degradation of peptidoglycan, and GlcNAc-anhMurNAc-tripeptide or -pentapeptide or anhMurNAc-tripeptide or -pentapeptide accumulates in the cytoplasm. Excess muramyl peptide binds to the transcriptional regulator protein AmpR and converts AmpR into an activator of -lactamases (1).

Datz et al. (4) demonstrated that induction of CumA utilizes a pathway identical to that observed for the control of AmpC synthesis in gram-negative bacteria. CumG and CumD in P. vulgaris correspond to AmpG and AmpD, and the functions are shared with those of AmpG and AmpD. Like AmpR, CumR is also a transcriptional regulator protein; however, the function is not shared with that of AmpR (4).

In this study, we defined the association between the MIC gap phenomenon and the peptidoglycan recycling and -lactamase induction systems of P. vulgaris.

(Part of this study was presented at the 38th Interscience Conference on Antimicrobial Agents and Chemotherapy [Ohno et al., 38th ICAAC].)

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Strains. A total of 42 isolates of P. vulgaris were collected from five medical centers across Japan during 1996. These isolates were unselected or sequential isolates. We also used P. vulgaris B317 and isogenic mutants with mutations in the peptidoglycan recycling and CumA induction systems (B317D, B317G, and B317R) and strains transcomplemented by the cloned ampDE, ampG, ampR, and cumR genes (B317D/pMD201, B317G/pMD301, B317R/pMD101, B317R/pMD401, and B317R/pMD501) (4), which were kindly provided by M. Datz, Laboratory d'Enzymologie, Centre d'Ingénièrie des Protéines, Institut de Chimie, Université de Liège, Sart Tilman, Belgium. These strains were stored at 80°C in 15% glycerol until used. The characteristics of each strain and plasmid were as follows. B317 is a clinical isolate. B317D carries a defect in CumD (a cytosolic N-acetylmuramyl-L-alanine amidase) and overproduces CumA. B317G carries a defect in CumG (a permease for GlcNAc-anhMurNAc-tripeptide or -pentapeptide) and must be induced for CumA. B317R carries a defect in CumR (transcriptional regulator protein for CumA; corresponds to AmpR for AmpC) and must be induced for CumA. pMD101 is a plasmid harboring ampR and ampC from Citrobacter freundii. pMD201 is a plasmid harboring the cloned ampDE gene from Escherichia coli. pMD301 is a plasmid harboring the cloned ampG gene from E. coli. pMD401 is a plasmid harboring the cloned ampR gene from C. freundii. pMD501 is a plasmid harboring the cloned cumR gene from P. vulgaris.

Antimicrobial agents. Cefotaxime (CTX; Hoechst Japan Co., Osaka, Japan), cefteram (CEM; Toyama Chemical Co., Tokyo, Japan), cefpodoxime (CPD; Sankyo Co., Tokyo, Japan), ceftizoxime (ZOX; Hujisawa Yakuhin Co., Osaka, Japan), moxalactam (MOX; Shionogi & Co., Osaka, Japan), ceftibuten (CTB; Shionogi), cephaloridine (CER; Shionogi), and ampicillin (AMP; Meiji Seika Kaisha, Tokyo, Japan) were used in this study. Each antibiotic was in a powder form of known potency.

Susceptibility testing. All isolates were subjected to in vitro antimicrobial susceptibility tests by broth microdilution and agar dilution according to the guidelines of the NCCLS (20). Antimicrobial agents used for the determination of MICs against clinical isolates of P. vulgaris were CTX, ZOX, MOX, CEM, CPD, and CTB. Furthermore, CTX, CER, and AMP were used for P. vulgaris B317 and the isogenic mutants. Dehydrated Mueller-Hinton broth (MHB; Difco Laboratories, Detroit, Mich.) adjusted to the correct cation concentrations was used for broth microdilution testing. A 0.1-ml quantity of each antimicrobial agent dilution was dispensed into each well of a 96-well microdilution tray by using a dispensing device. For the agar dilution method, dehydrated Mueller-Hinton agar (Difco) was used. In this test, 18 ml of molten test agar was poured into petri plates containing 2 ml of the appropriate dilution of antimicrobial agent solution prepared at 10 times the desired final concentration and rapidly mixed. Prepared trays and plates were used on the same day. Cultures were adjusted to a 0.5 McFarland standard and diluted 1:10 in sterile saline. Inoculum suspensions were used simultaneously for both methods. The microdilution trays were inoculated using an automatic MIC-2000 inoculator (Dynatech Laboratories, Inc., Alexandria, Va.) so that the final inoculum was approximately 10⁵ CFU/well. The agar plates were inoculated with an inoculum-replicating apparatus, Micro-Planter (Sakuma Seisakusho, Tokyo, Japan), at 10⁴ CFU/spot. The trays and plates were incubated for 48 h in ambient air at 35°C. The MIC represented the lowest concentration of antibiotic that completely inhibited visible bacterial growth and was read at 16 to 18 h and at 48 h after incubation.

Induction of -lactamase and extraction of crude enzyme. Overnight cultures of P. vulgaris B317 and isogenic strains in MHB were diluted 20-fold into 10 ml of fresh medium and incubated with continuous shaking at 35°C. After 2 h of incubation, CTX was added at a final concentration of 8 µg/ml. After a further 3 h of incubation, cells were harvested, washed three times with 0.1 M phosphate buffer (pH 7.0), and disrupted by ultrasonication. Broken cells were centrifuged at 100,000 × g for 30 min at 4°C, and the supernatant was used as the crude enzyme. The concentration of protein was determined by the method of Lowry et al. (19). -Lactamase activity was determined by a spectrophotometric method measuring the decrease in absorbance at 265 nm of CER (100 µM) at 30°C.

Determination of enzyme kinetic parameters. The values of V_max, K_m, and K_cat of purified P. vulgaris -lactamase (Wako Pure Chemicals Industries, Ltd., Osaka, Japan) were determined by computerized spectrophotometry using various absorbances at appropriate wavelengths for CER, CTX, CEM, CPD, ZOX, MOX, and CTB. The wavelength used for the photometric assay was that which yielded the maximum difference spectrum when an unhydrolyzed substrate was scanned against a hydrolyzed substrate. The wavelengths determined for the agents were as follows: CER, see above; CTX, 264 nm; CEM, 262 nm; ZOX, 250 nm; CPD, 265 nm; MOX, 275 nm; and CTB, 260 nm.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Susceptibility tests for clinical isolates. The susceptibility test results are summarized in Table 1 as the MICs at which 50 and 90% of the isolates were inhibited (MIC₅₀ and MIC₉₀, respectively). The MIC gap phenomenon, i.e., MICs obtained by the broth microdilution method that were markedly higher than those obtained by the agar dilution method, was observed for most oximino-type cephems but not for other types of cephems. However, the MIC gap phenomenon was not evident at the MIC₅₀ of ZOX despite the fact that this drug is an oximino type. On the other hand, the susceptibility test results obtained for 42 isolates by the E test and disk diffusion methods were similar to those obtained by the agar dilution method and also showed the MIC gap phenomenon with respect to the broth microdilution method (data not shown). The MICs were also higher following incubation for 48 h. Furthermore, the MIC gap phenomenon was not observed for CEM and was only weak for CPD at 48 h in the MIC₉₀ evaluation.

Susceptibilities of P. vulgaris B317, isogenic mutants with mutations in the peptidoglycan recycling and -lactamase induction systems, and transcomplemented strains to CTX, CER, and AMP. The MIC gap phenomenon was observed clearly for P. vulgaris B317, B317D, B317D/pMD201, B317G/pMD301, and B317R/pMD501 and slightly for P. vulgaris B317R/pMD101 but not for B317G, B317R, and B317R/pMD401 (Table 2). Furthermore, the phenomenon appeared with CTX but not with CER and AMP. A typical skip-growth effect was also expressed in B317R/pMD501 (Fig. 1).

View larger version (76K): [in this window] [in a new window]   FIG. 1.   Results of determination of MICs of CTX (a to c) and AMP (d to f) against P. vulgaris B317R/pMD501 by the broth microdilution method. Reading of the MIC was done after 48 h of incubation.

Effects of inoculum size and agar concentration on the MIC gap phenomenon. In the agar dilution method, MICs markedly increased when 10⁷ CFU/spot was inoculated. Furthermore, the MICs significantly increased with decreases in agar concentrations (Table 3).

Induction of -lactamase. -Lactamase activity was not detected in crude extracts from P. vulgaris B317G, B317R, and B317R/pMD401. P. vulgaris B317, B317D/pMD201, B317G/pMD301, B317R/pMD101, and B317R/pMD501 produced -lactamase only in the presence of inducing conditions. P. vulgaris B317D produced -lactamase in the presence or absence of inducing conditions (Table 4).

View this table: [in this window] [in a new window]   TABLE 4.   Expression of -lactamase activity in P. vulgaris B317 and other strains under noninducing and inducing conditions

Enzyme kinetic parameters. The V_max values of purified P. vulgaris -lactamase for CER, CTX, CEM, and CPD were markedly higher than those for ZOX (Table 5). The K_cat/K_m values of CER, CTX, CEM, CPD were 90-, 20-, 7-, and 4-fold higher, respectively, than those of ZOX (Table 5). No -lactamase activity was detected for MOX and CTB.

View this table: [in this window] [in a new window]   TABLE 5.   Kinetic parameters for -lactams of the P. vulgaris CumA -lactamasea

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

It is well known that the results of susceptibility tests are influenced by various methodological factors, such as the type of medium, inoculum size, pH, temperature, and incubation time (29); however, major differences in MICs (the MIC gap phenomenon) between the broth microdilution method and the agar dilution method in susceptibility testing of P. vulgaris against several -lactams, such as noted here, have not been reported. At present, the underlying mechanisms that cause such differences in MICs are not well understood. However, it is possible that the MIC gap phenomenon is related to the chromosomal class A -lactamase (CumA) of P. vulgaris because the phenomenon (i) is limited to oximino-type expanded-spectrum cephems, which are among the preferred substrates of CumA; (ii) is not observed with MOX and CTB, which are not degraded by CumA; and (iii) is observed only slightly with ZOX, which is relatively stable in the presence of CumA despite being an oximino-type agent. Furthermore, the observation that one clinical isolate that does not produce CumA did not show the MIC gap phenomenon also seems to support this possibility. When examined by PCR, this isolate possessed a 206-bp DNA segment containing the 165-bp cumR-cumA intercistronic region (4); however, the 1.26-kb DNA fragment containing cumA (27) was not amplified (data not shown). The deletion or insertion might occur in the cumA region of isolate T4. The MIC gap phenomenon was not observed in AmpC -lactamase-producing bacteria, such as C. freundii, Enterobacter cloacae, Serratia marcescens, and Pseudomonas aeruginosa, or in the chromosomal CdiA -lactamase of Citrobacter diversus, which belongs to Ambler's class A and is very similar to CumA (data not shown). On the other hand, the MIC gap phenomenon disappeared in P. vulgaris B317G and B317R but reappeared in B317G/pMD301 and B317R/pMD501. The findings strongly suggested that the MIC gap phenomenon was specific for P. vulgaris and was due to the production of CumA -lactamase in sufficient quantities through the peptidoglycan recycling and -lactamase induction systems expressed within the broth microdilution environment. P. vulgaris B317G is a mutant with defective CumG, a transporter of anh-MurNAc-tripeptide or pentapeptide which functions as an activator of the cumA transcriptional regulator, CumR; P. vulgaris B317R is a mutant with a defective cumR gene. Furthermore, B317G/pMD301 and B317R/pMD501 are the strains transcomplemented by cloned cumG gene and cumR genes, respectively.

The skip-growth effect may also support the relationship between the MIC gap phenomenon and CumA. Laboratorians often encounter a similar skip-growth effect in routine susceptibility testing, and several reports have described similar findings (8, 17, 22). The underlying mechanisms of the skip-growth effect are diverse although not yet sufficiently defined. However, the finding that the typical skip was noted in P. vulgaris B317R/501 (carrying the cloned cumR gene) but not in B317R/pMD401 (carrying the cloned C. freundii ampR gene) indicates that the skip-growth effect is likely to be seen in organisms in which cumA has been induced.

On the other hand, the reason for the significantly low MICs noted with the agar dilution method remains undetermined. P. vulgaris B317D is a mutant with defective function of CumD, a negative modulator of CumA, and constitutively produces a large quantity of CumA. The MIC of CTX for B317D in the agar dilution method was higher than that for the parent strain B317, suggesting that CumA is amply produced on agar plates because of the constitutive productivity of CumA in B317D. However, the MIC for B317D in the broth microdilution method was also higher than that in the agar dilution method, and the degree of the MIC gap phenomenon with both methods was nearly identical to that seen with B317 or B317D/201 (carrying the ampDE genes). Because the production of CumA is suppressed on 1.5% agar plates by an unknown mechanism, even in B317D, which shows a high level of production of CumA, the MIC obtained by agar dilution was not as high as we would have predicted. Although there is no definite evidence, the possible suppression of CumA induction in agar may also be supported by the result that MICs increased in proportion to the decrease in agar concentrations. Moreover, the observation that the MIC gap phenomenon was not observed for CER and AMP may be due to the easy hydrolysis of these agents even by a small amount of CumA produced on agar plates. This conclusion is based on the finding that the narrow-spectrum cephems and penicillin serve as better substrates of CumA than do oximino-type expanded-spectrum cephems (24). However, oximino-type expanded-spectrum cephems showed markedly high MICs against a few clinical isolates in the agar dilution method (data not shown); therefore, another mechanism may also explain the MIC gap phenomenon.

A few studies have examined the influence of agar on antimicrobial activity (9, 13). For example, antibiotics such as aminoglycosides and polymyxins are bound to negatively charged groups on the agar molecule (6). Therefore, the MICs of such antibiotics shift to high values. Furthermore, the divalent cation present in agar also reduces the antibacterial activity of tetracyclines (26). On the other hand, Ward et al. (28) reported that the addition of agar to susceptibility testing media lowered the MICs of amoxicillin-clavulanate against gram-negative bacilli. Regrettably, they did not discuss the reason for this finding.

We examined and identified the effect of inoculum size on results obtained with the agar dilution method. The MICs of CTX against P. vulgaris at an inoculum size of 10⁷ CFU/spot were markedly higher than those at 10⁴, 10⁵, and 10⁶ CFU/spot. It is not clear at this stage whether the increase in MIC due to the effect of inoculum size is related to the production of large amounts of CumA, although a few reports have shown a relationship between the inoculum size effect and -lactamases (7, 30).

The duration of incubation was also noted to influence the results (29); MICs of oximino-type cephems were higher following 48 h of incubation for more than half of the clinical isolates, B317, B317D/pMD201, B317G/pMD301, and B317R/pMD501. Such an effect of incubation time is often recognized in relation to slow growth in fastidious bacteria (10, 16). However, P. vulgaris is typically a nonfastidious bacterium; therefore, the effect of incubation time does not seem to be due to the influence of growth rate.

On the other hand, in vitro susceptibility tests do not always reflect the in vivo situation (23), and in vitro versus in vivo discrepancies may also occur because of problems with varying pHs and antibiotic tissue concentrations at the site of infection (3). The present study does not identify the method (agar dilution or broth microdilution) that may reflect the clinical response when oximino-type expanded-spectrum cephems are used to treat infectious diseases caused by P. vulgaris. Ikeda et al. (11), using a murine experimental model of infection with P. vulgaris, reported that the therapeutic effect of cefmenoxime (oximino-type expanded-spectrum cephem) in mice treated with a high dose was lower than that in those treated with a low dose and that -lactamase activity in the peritoneal cavity increased at higher cefmenoxime doses. Their study perhaps suggests that oximino-type cephems induce -lactamases of P. vulgaris in tissues at high concentrations of antibiotics during therapy.

The most important aspect of any susceptibility test is the accurate detection of resistance, because resistance carries a strong probability of therapeutic failure. The MIC gap phenomenon has been shown to also affect the more commonly used E test and disk diffusion susceptibility tests. We think that the susceptible results obtained with these methods may be false. Therefore, if the clinical laboratory reports to the clinician the antimicrobial susceptibility results for oximino-type expanded-spectrum cephems against P. vulgaris as susceptible by agar dilution, the E test, or disk diffusion methods, the patient may be exposed to ineffective antibiotics with side effects, including the modification of isolates of normal flora.

In summary, the present study indicates that appropriate consideration of proper standardization should be enforced in susceptibility testing with P. vulgaris.