JOURNAL OF BACTERIOLOGY, May 1969, p. 347-350 Vol. 98, No. 2 Copyright @ 1969 American Society for Microbiology Printed in U.S.A.

Susceptibility to Polymyxin B of Penicillin G-induced Proteus mirabilis L Forms and Spheroplasts MICHAEL TEUBER Abteilung Mikrobiologie, Institut fuir Angewandte Botanik der Technischen Hochschule Miinchen, Munich, Germany Received for publication 6 January 1969

A polymyxin B-resistant strain of Proteus mirabilis was converted into L forms and spheroplasts in the presence of penicillin G. This treatment caused a 400-fold increase in polymyxin B susceptibility. The acquired susceptibility was in the range of the natural susceptibility reported for susceptible gram-negative bacteria (-1 ,ug/ml). The high susceptibility to polymyxin B was lost as soon as the spheroplasts and L forms were allowed to reconvert into the bacillary form in penicillin-free media. This behavior is strong evidence that the natural resistance of Proteus strains to polymyxins is due to the impermeability of the outer cell wall structures to these antibiotic substances.

The primary mode of action of the polymyxin MATERIALS AND METHODS group antibiotics against susceptible bacteria Strain. The P. mirabilis strain used throughout seems to be a specific reaction with the cyto- this investigation was a laboratory strain from the plasmic membrane, which finally results in the collection of our department. It was classified. ac- irreversible breakdown of the permeability barrier cording to Bergey's Manual of Determinative Bac- of the cells (13, 14). It was suggested by Newton teriology. It does not have penicillinase activity. (13), on the basis of cell wall analysis of sus- Medium. The medium used contained, per liter of ceptible and resistant Pseudomonas species, that distilled water: tryptic digest of casein (Merck, in resistant bacteria the cell wall structures lo- Darmstadt, Germany), 10 g; glucose, 5 g; yeast ex- cated outside of the cytoplasmic membrane might tract (paste, Zyma Blaes, Munich, Germany) freed from insoluble material by centrifugation, 5 g; and protect this membrane from the action of poly- NaCl, 3 g; the final pH was 7.3 to 7.5. For solid myxin by their impermeability to the anti- media, 2% powdered agar-agar (Serva, Heidelberg, biotic. A direct approach to this question would Germany) was added. be the analysis of susceptible forms derived in Antibiotics. Polymyxin B sulfate (sterile powder) vitro from resistant bacteria. Proteus strains was generously supplied by Pfizer GmbH, Karlsruhe, should be suitable organisms for this kind of Germany. Penicillin G (sodium salt) was a gift from work. These organisms have an unusually high Hochst-Werke, Frankfurt/M., Germany; its specific degree of resistance against polymyxins, in activity was 1,650 units/mg. contrast to most other gram-negative bacteria, Incubation. All experiments were performed at 37 C. Liquid media were aerated by shaking on a which are very susceptible (1). In addition, reciprocal shaker (100 strokes/min). Proteus cells can be converted into spheroplasts Agar-diffusion test. An inoculum of 107 bacteria and L forms and cultivated as such in the pres- from an overnight culture was plated on solid me- ence of penicillin, a procedure altering the dium. Polymyxin B sulfate (50 ,ug in 10 Aliters of morphological and chemical properties of the water) was pipetted onto sterile filter paper discs cell wall to a high degree (2, 5, 6, 8). A detailed (diameter, 9 mm) placed in the middle of the petri analysis of stable Proteus L forms and their dishes. parent bacteria was recently published by Weibull Growth curves. Erlenmeyer flasks (100 ml) con- and co-workers (18). taining 20 ml of broth were inoculated with 0.1-ml amounts of overnight cultures of rod-shaped Proteus In this publication, I report studies which show cells or of spheroplasts. The optical density of the that P. mirabilis could, in fact, be rendered highly cultures was determined with 1-cm cells at 600 nm susceptible to polymyxin B in vitro by conversion (Fig. 1). into penicillin G-induced L forms and sphero- Growth response of spheroplasts to polymyxin B. plasts. Spheroplast cultures were grown in broth to an 347 348 TEUBER J. BACTERIOL. optical density of 0.3 at 600 nm. At this point, poly- TABLE 1. Susceptibility of Proteus mirabilis Lforms myxin B sulfate was added, and the optical density was to 50 jug ofpolymyxin B sulfate, as tested by the measured after an additional 2 hr of incubation (Fig. agar-diffusion test 2). The appearance of the cells during growth was checked with a Zeiss phase-contrast microscope. Penicillin G Zone of no growth Nomenclature. Growth in form of large, irregular- (units/ml of (mm) Shape of organisms shaped bodies is called L form, irrespective of the medium) (m)Saeoornis stability of the cells possessing this property. Regular- shaped, round bodies are called spheroplasts. 0 0 rods 1 1 elongated, swollen RESULTS rods 2.5 15 i 2 L forms Conditions for the formation of penicillin- 5.0 15 i 2 L forms induced L forms and spheroplasts. On solid medium, growth of L-form colonies was ob- tained at a penicillin G concentration of 2.5 penicillin levels. Microscopic examination of units/ml. In a typical experiment, from 107 these colonies revealed that all of them were of rod-shaped bacteria plated, about 103 L-phase the intermediate growth type. When 20 selected colonies developed at 2.5 units of penicillin G/ml L-phase colonies grown on penicillin-containing and about 102 at 5 units/ml. Higher concentra- plates (2.5 units/ml) were transferred to a penicil- tions of penicillin were less favorable. The colo- lin-free medium, they regained a bacillary appear- nies were 1 to 2 mm in diameter after 48 hr of ance and their polymyxin resistance ( > 50,gg/ml), incubation. Usually, no rod-shaped bacteria whereas transfer to a penicillin-containing medium grew under these conditions. On the other hand, preserved the high polymyxin susceptibility to- 1 unit of penicillin/ml was not sufficient to induce gether with L-form growth. From these results, it L forms. Microscopic examination showed is clear that the penicillin-induced changes in the elongated rods swollen at the region of cell divi- cell wall have to be severe enough to allow L- sion. The L forms could be converted back to a form growth in order to produce full suscepti- bacillary type of growth by a single transfer to bility to polymyxin B. Since a quantitative evalua- the penicillin-free medium, as checked with 20 tion with polymyxin is difficult to do by the agar randomly selected L-phase colonies. plate diffusion test (because of the low diffusion In liquid culture at a penicillin G concentra- rate of polymyxin), further experiments were tion of 2.5 units/ml, L forms isolated from performed with liquid cultures. penicillin-containing plates could be cultivated in As judged turbidimetrically (Fig. 1), addition the form of spheroplasts with a doubling time of polymyxin B sulfate to logarithmically growing of 78 min, as compared with 42 min for the penicillin spheroplasts of P. mirabilis in broth parent bacillary form in penicillin-free medium. resulted in complete and immediate cessation of Doubling time was defined as the time needed to growth. The parent bacillary form, however, was allow doubling of the optical density during the only slightly affected by the same doses of poly- linear part of the growth curve (see Fig. 1). myxin B sulfate (5 and 50 ,ug/ml). The drop in the Continuous microscopic examination during the optical density of the spheroplast culture ob- growth cycle revealed only spheroplasts, but no served after polymyxin application was caused by rods. a macroscopically and microscopically visible The unusually low penicillin concentration aggregation of the spheroplasts to clumps con- necessary to induce L forms and spheroplasts is taining up to several hundred organisms. This ob- a particular property of the P. mirabilis strain servation is in accordance with the report by used. Latterrade and Macheboeuf (7) that rough forms Susceptibility to polymyxin B sulfate of L of gram-negative bacteria are agglutinated by forms and spheroplasts. From the data presented polymyxin. Extensive lysis could be excluded be- in Table 1, it appears that penicillin-induced L cause only very few ghosts were detected under forms of P. mirabilis are susceptible to concentra- the microscope. In Fig. 2, results are presented tions of polymyxin B sulfate to which the rod- which indicate that as little as 0.1 ,ug of polymyxin shaped parent form is totally resistant. The in- B sulfate per ml had a significantly inhibitory termediate form (elongated, swollen rods) growing effect on growing penicillin spheroplasts of P. at the low penicillin G concentration (1 unit/ml) mirabilis. The minimal concentration required to seems to be almost as resistant as the true rod- halt growth completely under the conditions em- shaped form. Occasionally, and mainly after pro- ployed was calculated from these data to be 1.6 longed incubation, a few colonies developed ,ug of polymyxin B sulfate/ml. For a 50% reduc- within the zone of inhibition at the two higher tion of the growth rate, 0.2 ,ug/ml was sufficient. VOL. 98, 1969 POLYMYXIN-SUSCEPTIBLE SPHEROPLASTS OF PROTEUS 349 tibility of the P. mirabilis spheroplasts was de- termined to be 400-fold. In conclusion, it can be 5 0 °50 stated that P. mirabilis spheroplasts are at least as - 2.0 ROD susceptible to polymyxin B as most of the other E 'L gram-negative bacteria which are sensitive a 0 0 priori (for a comparison, see 14). co 1.0 L IW 1-. I,/I DISCUSSION ,'( Since polypeptide antibiotics of the polymyxin >. 0.5 L .-- type act on bacterial membranes because of their zn ,, cationic detergent character (13, 14), two differ- z K ent explanations of how resistant bacteria are 3^SPHEROPLASTS protected are conceivable: (i) the antibiotic is 0-J chemically or enzymatically inactivated before it ik _0 _. _0 can reach the membrane, or (ii) the cell wall 0 0.1 structures located outside of the cytoplasmic a- membrane cannot be penetrated by the antibiotic. 0 pg/mi I - The first alternative may be excluded in the case of 0l) -0 50 PX 0 I polymyxin B since no degradation products could l be detected in our laboratory by either electro- 0 2 4 6 8 10 12 phoretic or chromatographic methods (unpub- TIME (hr) lished data). The second hypothesis was already stated by Newton (13) for susceptible and resist- FIG. 1. Susceptibility to polymyxin B sulfate (PX) of penicillin G-induced Proteus mirabilis spheroplasts ant Pseudomonas strains. In addition to being as seen in liquid cultures. supported by my results, it is supported by the observation of Taubeneck (16) that a stable L form of P. mirabilis was highly susceptible to macrolide antibiotics (erythromycin, oleando- 0.8 mycin, and others) whereas the parent bacillary form was resistant. Similarly, Tulasne and Minck E 4 m (17) reported that L forms derived from P. vul- 0 0 garis and P. morgani demonstrated increased c 0.6 susceptibility to tyrothricin (a mixture of tyro- and A report on I- cydine gramicidin). gram-positive bacteria which fits this hypothesis was published by Shockman and Lampen (15), who observed a 5-fold increase in the susceptibility of Strepto- U) z coccus faecalis spheroplasts to polymyxin B, to- w 0 gether with a 10-fold higher susceptibility to the macrolide oleandomycin. In addition, methicillin- -J0.2 induced protoplasts of several Staphylococcus aureus strains were shown to possess a 10- to

a- 50-fold enhanced polymyxin susceptibility (4). 0 At the present time, however, the possibility 0 cannot be excluded that the membrane itself is 0 0.1 1.0 10 100 made more sensitive to polymyxin when P. POLYMYXIN B SULFATE mirabilis and other bacteria are converted into log pg/mi penicillin-induced spheroplasts. This possibility FIG. 2. Growth inhibition of penicillin spheroplasts was discussed by Montgomerie et al. (9). This of Proteus mirabilis after 2 hr of incubation with low interpretation would require one to postulate doses of polymyxin B. The initial optical density of that the Proteus membrane is more rigid than the the spheroplast cultures used was 0.3. membrane of other gram-negative bacteria, which seems very unlikely if one takes into account the To get the same reduction with the parent rod- work of Neu. He demonstrated that osmotic shaped bacteria, a concentration of 80 Ag/ml had shock caused the release of the nucleotide pool to be applied in a control experiment. From these from all investigated gram-negative bacteria, in- results, the absolute increase in polymyxin suscep- cluding Proteus (12). On the other hand, none of 350 TEUBER J. BACrERIOL. the lytic enzymes (ribonuclease I, 5'-nucleotidase, Sensitivity of coccal and L forms of Staphylococcus aureus cyclic alkaline phosphatase, to five antibiotics. J. Bacteriol. 88:630-632. phosphodiesterase, 5. Kandler, O., and G. Kandler, 1960. Die L-Phase der Bakterien. and others) was released from Proteus by this Ergeb. Mikrobiol. Immunitaetsforsch. 33:97-127. treatment, whereas all other gram-negative organ- 6. Klieneberger-Nobel, E. 1960. L-forns of bacteria, p. 361-386. isms excreted 50 to 100% of the corresponding In I. C. Gunsalus and R. Y. Stanier (ed.), The bacteria, activities into the medium. In view of the accumu- vol. 1. Academic Press Inc., New York. 7. Latterrade, C., and M. Macheboeuf. 1950. Recherches bio- lating evidence that these enzymes are located be- chimiques sur le mode d'action de la polymyxine. Ann. Inst. tween the cytoplasmic membrane and the cell wall Pasteur 78:753-758. (3), and with attention directed to the striking 8. McQuillen, K. 1960. Bacterial protopiasts, p. 249-359. In I. C. similarity of the enzymes purified to date from a Gunsalus and R. Y. Stanier (ed.), The bacteria, vol. 1. Academic Press Inc., New York. variety of gram-negative bacteria including 9. Montgomerie, J. Z., G. M. Kalmanson, and L. B. Guze. 1966. Proteus (10, 11), it is not unreasonable to suggest The effects of antibiotics on the protoplast and bacterial that Proteus strains indeed may have a very rigid forms of Streptococcus faecalls. J. Lab. Clin. Med. 68:543- to molecules 551. outer cell wail impermeable large 10. Neu, H. C. 1968. The 5'-nucleotidases (uridine diphosphate such as the above-mentioned antibiotics and sugar hydrolases) of the Enterobacteriaceae. Biochemistry enzymes. Under this assumption, the polymyxin 7:3766-3773. resistance of Proteus may well be correlated with 11. Neu, H. C. 1968. The cyclic phosphodiesterases (3'-nucleo- wall structure. tidases) of the Enterobacteriaceae. Biochemistry 7:3774- this postulated cell 3780. Work is in progress in our laboratory to de- 12. Neu, H. C., and J. Chou. 1967. Release of surface enzymes in termine the nature and chemistry of the cell wall Enterobacteriaceae by osmotic shock. J. Bacteriol. 94:1934- structure(s) responsible for the polymyxin resist- 1945. ance of Proteus. 13. Newton, B. A. 1956. The properties and mode of action of the polymyxins. Bacteriol. Rev. 20:14-27. 14. Sebek, 0. K. 1967. Polymyxins and circulin, p. 142-152. In ACKNOWLEDGMENTS D. Gottlieb and P. D. Shaw (ed.), Antibiotics, vol. 1, Mecha- nism ofaction. Springer Verlag, Berlin. I thank 0. Kandler for helpful comments and suggestions. The 15. Shockman, G. D., and J. 0. Lampen. 1962. Inhibition by anti- valuable technical assistance of R. Keck is gratefully acknowl- biotics of the growth of bacterial and yeast protoplasts. J. edged. Bacteriol. 84:508-512. 16. Taubeneck, U. 1962. Susceptibility of Proteus mirabilis and LriRATURE CITED its stable L forms to erythromycin and other macrolides. Nature 196:195-196. 1. Courtieu, A. L., J. J. Monnier, P. de Lajudie, and F. N. Guil- 17. Tulasne, R., and R. Minck. 1952. Sensibilitd comparde des lermet. 1961. Spectre antibactdrienne de la colistine vis-k-vis formes normales et des formes L de deux souches de Proteus 1200 souches. Ann. Inst. Pasteur Suppl. No. 4, p. 14-31. vis-&-vis de quelques antibiotiques. Compt. Rend. Soc. Biol. 2. Guze, L. B. (ed.). 1968. Microbial protoplasts, spheroplasts 146:778-780. and L-forms. The Williams & Wilkins Co., Baltimore. 18. Weibull, C., W. D. Bickel, W. T. Haskins, K. C. Milner, and 3. Heppel, L. A. 1967. Selective release of enzymes from bac- E. Ribi. 1967. Chemical, biological, and structural proper- teria. Science 156:1451-1455. ties of stable Proteus L forms and their parent bacteria. J. 4. Kagan, B. M., S. Zolla, R. Busser, and S. Liepnieks. 1964. Bacteriol. 93:1143-1159.