Beta-Cyclodextrin Confers Penicillin Resistance in an Alkaliphile
Paul de Figueiredo1*, Becky Terra1*, Jasbir Kaur1,2, Ana Lenskiy1, Martin Sadilek3, Sen-Itoroh Hakomori2 and Gene Nester1#

*  With Equal Contribution
#  To whom correspondence should be sent.  gnester@u.washington.edu

1.  Department of Microbiology, Box 357242, University of Washington, Seattle, WA 98195
2.  Pacific Northwest Research Institute, 720 Broadway, Seattle, WA 98122
3.  Department of Chemistry, University of Washington, Box 351700, Seattle, WA, 98195-1700
4.  Department of Biochemistry, Teikyo University School of Medicine, Kaga 2-11-1, Itabashi-ku, Tokyo 173-8605



 
 We recently demonstrated that beta-cyclodextrin (bCD) glycozyme activity mediates the hydrolysis of penicillin antibiotics in a model biological system containing an alkaliphilic Bacillus species (ATCC 21594) [1].  Here, we show that cyclodextrin participates in mediating penicillin resistance in this organism. When growing at its optimum pH (10.2) under nutritional conditions where it synthesizes cylcodextrin, the bacterium is resistant to penicillin.  Under conditions where cyclodextrin synthesis is inhibited, the bacterium becomes sensitive to this antibiotic.  A mutant strain with defects in cyclodextrin production is also penicillin-sensitive. These data therefore demonstrate that bCD glycozyme activity plays a physiological role in mediating antibiotic resistance in this model system.

The cyclodextrins are cyclomaltooligosaccharides containing six (alpha), seven (beta), or eight (gamma) a 1-4 linked glucose residues [2] and are a carbon and energy source for a variety of bacteria.  In addition, bCD possesses the remarkable capacity to catalyze assorted chemical reactions in non-biological in vitro systems [3]. We recently extended these studies by demonstrating that bCD can catalyze the hydrolysis of penicillin in vivo in a model biological system containing an alkaliphilic Bacillus species (ATCC 21594). The chosen Bacillus strain thrives in alkaline conditions and produces large amounts of extracellular cyclodextrins [4].  In addition, it grows in the absence or presence of starch.  bCD is produced under the former, but not the latter, conditions [4].  Therefore, starch provides a convenient and natural switch whereby bCD-mediated glycozyme activity may be manipulated in vivo.   In previous studies, we exploited this fact to show that bCD directly catalyzes the hydrolysis of penicillin in vivo [1].  However, the role that bCD glycozyme activity plays in conferring penicillin resistance to this Bacillus strain remained unexplored. Here, we address this issue by demonstrating that cyclodextrin mediates penicillin resistance in this alkaliphile.

We employed two methods to assay ABPC resistance in this model system.  First, we added various amounts of the antibiotic to early log phase cultures and then assessed bacterial growth by measuring the optical density of cultures at various times after antibiotic addition.  This assay allowed us to determine the concentration of antibiotic that inhibited growth by 50% (IC50).  Second, we measured halos around antibiotic disks placed on culture plates harboring Bacillus cells.  By measuring the distance from the edge of a disk to the point at which bacterial growth is first seen provided us with a reliable method for assessing antibiotic resistance [5].  

When cultures were grown in the presence of rich or minimal liquid media containing starch, ABPC resistance (IC50 = ~400mg/ml and ~450mg/ml) was observed (Fig. 1A).  Similar resistance was observed on solid media (clearance = 3.3mm) (Table I). However, when cultures were grown in the absence of starch, this resistance was not seen (IC50 = 7mg/ml; clearance = 8.1mm) (Fig. 1A and Table I). Importantly, starch did not induce resistance to non-b-lactam containing antibiotics, including erythromycin, streptomycin, and rifampin (IC50 = 10mg/ml; clearance = 2 mm) (Fig. 1B and data not shown), and cultures grown in the presence of galactose, lactose, raffinose, and arabinose did not display an antibiotic resistance phenotype (Fig. 1C). Moreover, the degree of resistance was directly correlated with the amount of starch (and hence bCD) present in the culture media (Fig. 1D).  Finally, starch-dependent antibiotic resistance was seen in both defined minimal media and rich media (Fig. 1A). Taken together, these observations were consistent with the idea that starch (but not other carbon sources) (Fig. 1C) stimulated production of a specific and dose-dependent resistance to ABPC.  

We asked whether the activity conferring ABPC resistance was soluble and secreted into the bacterial culture media. To address this question, we examined Bacillus conditioned media derived from cells cultured in the presence or absence of starch.  When Bacillus cells were cultured in control starch-free conditioned media, no significant ABPC resistance was observed (IC50 = 7mg/mL) (Fig. 1E).  However, when conditioned media derived from starch cultures was employed, a dramatic ABPC resistance phenotype was seen (IC50 =  400 mg/mL) (Fig. 1E).  We found that the activity conferring antibiotic resistance was not destroyed when conditioned media derived from starch cultures was boiled for 30 min, suggesting that a protein was not responsible for the activity.  Previous work demonstrated that the activity was not detected in whole cell extracts, or supernatants derived from cells cultured in the absence of starch, using standard chromogenic assays for b-lactamase activity [1].  Moreover, previous PCR experiments using Bacillus genetic material as a template failed to detect the b-lactamase genes present in common laboratory bacterial strains [1].  Finally, the Bacillus strain was sensitive to ABPC (IC50 < 7 mg/ml) when grown in the absence of starch (Fig 1A).  These observations were consistent with the hypothesis that a heat-resistant factor present in conditioned media derived from cells grown in starch (but not glucose alone) was responsible for conferring the ABPC-resistance phenotype.

We next added pure bCD (5mM) to Bacillus cultures grown in the absence of starch and found that bCD-supplemented cultures were significantly more resistant to ABPC (IC50 = 450mg/mL; clearance = 1.7mm) than controls lacking the added oligosaccharide (IC50 = 7mg/ml) (Fig. 1A, Table I).  These data suggested that bCD synthesized in the presence of starch was responsible for mediating the observed antibiotic resistance.  The data did not, however, indicate whether bCD was mediating the effect directly, or inducing the synthesis of the material responsible for the resistance.  Subsequent experiments were therefore aimed at obtaining genetic evidence that would clarify its role.

We examined the antibiotic resistance phenotype of strains harboring defects in bCD production and employed the powerful chemical mutagen ethyl methanesulfonate (EMS) to generate Bacillus mutants with defects in bCD production.  Colonies were screened on bCD-indicator plates harboring phenolphthalein and methyl orange dyes [4, 6].  Promising candidates were retested on bCD indicator plates (Figs. 2A), and their culture supernatants were also tested for bCD content by HPTLC and reverse phase HPLC (data not shown).  One strain, BT13, was found to consistently display deficiencies in bCD accumulation (Fig. 2).  An analysis of BT13’s ABPC resistance phenotype revealed that it was significantly more sensitive to ABPC than WT controls when grown in the presence of starch (WT IC50 = ~400mg/mL; BT13 IC50 = 9mg/mL) (Fig. 1F).  Importantly, the addition of pure bCD (but not glucose) to glucose- or starch-grown BT13 cultures conferred ABPC-resistance (IC50 = ~450 mg/ml; Fig. 1F). Taken together, these data indicated that bCD was directly mediating the ABPC resistance phenotype.

To further explore this possibility, we examined ABPC resistance in a heterologous and well-characterized Bacillus subtilis system. B. subtilis provided a convenient gain-of-function system for these studies because the organism is known to be sensitive to b-lactam containing antibiotics, and can grow under mildly alkaline conditions (pH 9.1).  In addition, bCD-producing (ALKO 2279) and control (ALKO 2013) strains are available [20].  We therefore compared the ABPC resistance phenotypes of bCD-producing and non-producing strains. We found that ABPC resistance was conferred to non-bCD producing strains grown on alkaline solid media (clearance = 21.3 mm) but not its non-alkaline counterpart (clearance = 12.8 mm) (Table IB). In addition, bCD-producing strains grown in liquid media harboring starch were more resistant to ABPC (IC50 = 345 mg/ml) than controls grown in the absence of this carbon source (IC50 = 75 mg/ml).  Importantly, the addition of pure bCD to non-producing cultures conferred an ABPC resistance phenotype (Table IB), consistent with the idea that bCD was responsible for this activity.  Finally, our HPTLC analysis confirmed previous findings [20] demonstrating that bCD accumulates in large quantities (1.8 mM) in ALKO 2279 cultures grown in the presence of starch.  Therefore, these “gain-of-function” data supported the idea that bCD was directly participating in mediating the observed antibiotic resistance.

Our results suggest a novel role for bCD glycozyme in antibiotic hydrolysis and resistance in an alkaliphile.  In addition, because cyclodextrin-producing alkaliphiles may encounter antibiotics containing b-lactam rings in their natural environment [7], our results indicate that bCD-mediated g___o____ activity may constitute a physiologically relevant antibiotic resistance mechanism. Finally, it will be of interest to examine other alkaliphiles for bCD production to see whether this property correlates with ABPC resistance

Figure Legends

Figure 1.  bCD mediates antibiotic resistance. A.  Bacillus (ATCC 21594) was grown in rich or minimal medias in the presence or absence of starch and bCD. The log phase doubling times for untreated cultures, and IC50s for ABPC were then determined as described.  B. The IC50s for ABPC (AMP), rifampin (RIF), erythromycin (ERY), and streptomycin (STR) were determined for WT and BT13 cultures grown in the presence of starch.  RIF, ERY, and STR (but not AMP) lack b-lactam rings.  C. The IC50s for ABPC were determined for WT Bacillus cultures grown in minimal media in the presence of the following carbon sources:  starch, lactose, galactose, raffinose, and arabinose.  D. Increasing the starch:glucose ratio increases the IC50. Cells were grown in media containing several starch:glucose ratios.  The IC50s were calculated from 16-hour readings.  E.  WT Bacillus cells were grown in conditioned media mixed 1:1 with unconditioned media.  AMP was added and the IC50 was determined as described. F. Addition of pure bCD to glucose-containing non-bCD-producing BT13 cultures rescues the BT13 AMP sensitivity phenotype.  All data represent the average of at least 6 replicates.

Figure 2.  bCD phenotypes of WT and BT13 cells.  A.  Untreated bCD indicator plates are red, but turn yellow when incubated in the presence of bCD.   WT Bacillus cells were streaked onto indicator plates containing glucose (and lacking starch) (1) or containing starch alone (2).  BT13 cells were streaked onto an indicator plate containing starch (3).  Pure bCD powder was sprinkled onto an indicator plate containing glucose and lacking starch (4).  All plates were photographed after 48 hours of incubation at 37∞C.  B.  Protein extracts were obtained from (1) ~106 TOP10 E. coli cells harboring a plasmid (pBAD-TOPO 2.1) containing a protein b-lactamase gene, and (2) 109 Bacillus (ATCC 21594) cells.  These extracts were spotted to the center of LB plates containing 10mg/mL AMP.  Plates were then streaked with AMP-sensitive B. subtilis cells (ALKO 2013), incubated at 37°C and photographed after 24 hours.
 
Table I.  bCD confers AMP resistance to the alkaliphilic Bacillus strain (ATCC 21594) (A), and B. subtilis (ALKO 2013) (B) on solid media at containing glucose or starch.  Antibiotic discs containing AMP (10 mg/ml), ERY (10mg/ml), STR (10mg/ml), or RIF (10 mg/ml) were added to plates streaked with Bacillus (ATCC 21594) or B. subtilis.  Zones of antibiotic clearance were measured from the outside of the disc to the closest colony.  The Bacillus sp. experiments (A) were performed at pH 10.2, the B. subtilis experiments were performed at pH 7.0 or 9.1 as indicated. All data represent the average of at least 6 replicates.

Supporting Online Material
Materials and Methods.
References.

References and Notes
1.  1st paper (de Figueiredo, Kaur, Terra, et al., XXX, 2004)
2. Szetjtli, J. 1998. Chem. Rev. 98: 1743-1753.
3. For review, see Bender and Komiyama. Cyclodextrin Chemistry.  Springer-Verlag. 1978. NY.
4.  Park, CS et al. 1989.  Agric Biol Chem.  53:  1167-1169.
5. Bauer et al. Am J Clin Pathol. 1966 45:493-6.
6. Taguchi. K. J. Am. Chem. Soc.  1986.  108:  2705-2709.
7.  Kato, C., et al. Arch. Microbiol. 1989.  151: 91-94.
8.  The authors would like to thank Drs. Milton Gordon, Jim Staley, D. Ellis Monks, Marion Brodhagen, and members of the Nester, Hakomori, and Gordon laboratories for their comments and remarks, and Dr. Carlos Semino for stimulating discussion during the early stages of this work.  Emily Lee and Adonis Acuario provided expert technical assistance.  This work was supported by a National Science Foundation Small Grant for Exploratory Research (MCB-0135592) and a National Institutes of Health Research Grant (GM 32618) to  G. N., an X grant (  )  to S.-I. Hakomori, a Y grant (  ) to T. H, and an American Cancer Society postdoctoral fellowship to P. d. F.