The Ice Nucleation Gene from Pseudomonas syringae as a Sensitive ...

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Jan. 1995, p. 273–277 0099-2240/95/$04.0010 Copyright q 1995, American Society for Microbiology

Vol. 61, No. 1

The Ice Nucleation Gene from Pseudomonas syringae as a Sensitive Gene Reporter for Promoter Analysis in Zymomonas mobilis CONSTANTIN DRAINAS,1,2* GEORGIOS VARTHOLOMATOS,1 2 AND NICHOLAS J. PANOPOULOS Sector of Organic Chemistry and Biochemistry, Department of Chemistry, University of Ioannina, 45110 Ioannina, Greece,1 and Department of Environmental Science, Policy and Management, College of Natural Resources, University of California, Berkeley, California 947202 Received 2 August 1994/Accepted 24 October 1994

The expression of the ice nucleation gene inaZ from Pseudomonas syringae in Zymomonas mobilis strains under the control of three different promoters was investigated to establish the utility of the gene as a reporter and examine the possible use of the organism as a source of ice nuclei for biotechnological applications. A promoterless version of the inaZ gene was placed under the control of three different promoters: Ppdc (pyruvate decarboxylase), a homologous strong promoter from Z. mobilis; Pbla (b-lactamase) of plasmid pBR325; and PhrpR, the promoter of hrpR, a regulatory gene from P. syringae pv. phaseolicola. The apparent strengths of all three promoters, measured by quantifying the ice nucleation activity at 29&C, were lower in Z. mobilis than in Escherichia coli. The levels of ice nucleation activity expressed under the Ppdc promoter were significantly higher than those obtained with the two heterologous promoters in Z. mobilis. Plasmid pCG4521 (RK2 replicon) gave much lower levels of ice nucleation activity when propagated in strain uvs-51, a plasmid instability mutant of Z. mobilis, compared with the wild-type strain. The ice nucleation activity in Z. mobilis cultures showed unusual partitioning in that the culture supernatants obtained after low-speed centrifugation contained the majority of ice nuclei. Analysis of the ice nucleation spectra revealed that the cell pellets contained both ‘‘warm’’ and ‘‘cold’’ nuclei, while the culture supernatant contained primarily cold nuclei, suggesting that the cold nucleus activity may be extracellular. However, all nucleation activity was retained by 0.22-mm-pore-size filters.

Zymomonas mobilis is a gram-negative anaerobic bacterium of industrial importance that produces ethanol from simple hexoses in high rates and yields (3). In addition to its tolerance of high concentrations of ethanol (up to 12% [vol/vol]) and glucose (up to 20% [wt/vol]) (19), the bacterium has other unusual adaptations of both environmental and biological interest. For example, it can tolerate moderately high concentrations of certain salts (e.g., MgSO4, up to 5% [wt/vol] [3a]) and low pH (3.5) (13) and is highly resistant to UV radiation and chemical DNA-damaging agents (21), as well as to most antibiotics commonly used as selective agents in routine genetic analysis (17). Plasmid vectors have been developed, mutants can be isolated by various mutagenic treatments, and various genes have been cloned in heterologous hosts such as Escherichia coli (17). However, several other routine methods and tools for molecular genetic investigations, including the use of reporter genes for analysis of gene transcription and promoter function, have not yet been developed in Z. mobilis. We have investigated the expression of an ice nucleation gene, inaZ, originally cloned from Pseudomonas syringae (5), in Z. mobilis for two reasons. (i) We wished to determine its utility as a potential reporter; this gene is a very useful reporter in other gram-negative bacteria because its activity can be measured by a very sensitive assay ($105-fold more sensitive than conventional reporters such as b-galactosidase) and its

expression can be quantified by very simple methods, such as a droplet-freezing assay (6, 22, 9). (ii) We wanted to evaluate the use of Z. mobilis as a potential source of biological ice nuclei for industrial applications. Biological ice nuclei are produced primarily by certain gram-negative bacteria, most of which are either plant pathogens or nonpathogenic plant epiphytes. The ubiquitous presence of many of these species in plants and other natural habitats makes bacterial ice nucleation a common phenomenon in nature and a major factor contributing to frost damage to agronomic crops (reference and references therein). Biological ice nuclei of bacterial origin are currently used in artificial snowmaking (Snowmax, a freeze-dried preparation of P. syringae cells) and have potential applications in frozen food industries, rainfall enhancement, and hail control, in addition to other applications (8). The use of bacterial ice nuclei in environmental and, particularly, food industries faces various safety, performance, toxicity, and regulatory constraints. Some of these constraints may be overcome by producing ice nuclei in nonpathogenic, consumer-safe microorganisms. Z. mobilis is a microbe that is generally regarded as safe (GRAS), as it is used traditionally in the tropics for the production of beverages (13, 19). Therefore, ice-nucleating Z. mobilis cells provide an acceptable alternative for a wide range of biotechnological applications. The present study concerns the transfer of the ice nucleation gene inaZ in Z. mobilis and the quantification of its expression under the control of three different promoters. The results clearly show that the inaZ gene can serve as an easily assayable reporter gene in this bacterium and that a significant proportion of ice nuclei active at temperatures of above 278C may be released into the culture medium.

* Corresponding author. Mailing address: Sector of Organic Chemistry and Biochemistry, Department of Chemistry, University of Ioannina, 45110 Ioannina, Greece. Phone: 30-651-98372. Fax: 30-65147832. 273

274

DRAINAS ET AL.

APPL. ENVIRON. MICROBIOL.

TABLE 1. Bacterial strains and plasmids Species

E. coli

Strain

HB101 DH5a DH5a DH5a DH5a DH5a DH5a Z. mobilis CU1Rif2 uvs-51

Plasmid

pRK2013 pDS3154 pPTZ3 pCG4521 pUZ119 pPTZ3-inaZ pDS3154-inaZ pZMO1,2,3,5,6 pZMO1,3,5,6

Relevant plasmid markersa

Reference(s) for plasmid

Knr Tra1 Apr Cmr Mob1 Apr Tcr Mob1 Tcr Ice1 Mob1 Apr Ice1 Apr Tcr Ice1 Mob1 Cmr Ice1 Mob1 ND ND

4 15 12 5a, 11 2 Present work Present work 1 21

a Ap, ampicillin; Cm, chloramphenicol; Kn, kanamycin; Tc, tetracycline; Mob1, mobilization functions; Tra1, conjugal transfer functions; ND, not determined.

MATERIALS AND METHODS Strains, plasmids, and growth conditions. The strains and plasmids used are listed in Table 1. E. coli cells were grown in Luria broth by standard methodology. E. coli DH5a (Bethesda Research Laboratories, Gaithersburg, Md.) was used as the host for subcloning and for maintenance of recombinant plasmids. Z. mobilis CU1Rif2, a rifampin-resistant derivative of CU1 (1), and uvs-51, a UVmethyl methane sulfonate-sensitive strain with impaired plasmid stability (21), were the recipients for inaZ gene constructs. The Z. mobilis strains were grown on liquid or agar complete medium at 308C as described previously (1). Filtersterilized antibiotic solutions were added to these media as required for genetic selections or plasmid maintenance at the following concentrations: ampicillin (30 mg ml21, a marker for E. coli only), chloramphenicol (20 mg ml21 for E. coli; 100 mg ml21 for Z. mobilis), kanamycin (20 mg ml21 for E. coli), rifampin (20 mg ml21 as counterselection against E. coli), and tetracycline (40 mg ml21). Recombinant DNA techniques. Plasmid DNA was extracted from E. coli strains by the alkaline lysis method described by Sambrook et al. (14) and from Z. mobilis strains by a previously described method (15). Restrictions, ligations,

and Southern blotting were carried out by standard methodology (14). Hybridizations were performed under high-stringency conditions with a nonradioactive labelled probe (digoxigenin-11-dUTP), using the Boehringer Mannheim DNA labelling and detection kit (nonradioactive; catalog no. 1093 657) according to the instructions of the manufacturer. Bacterial conjugations. Conjugal transfer of recombinant plasmids in Z. mobilis cells was performed essentially as described before (1), using double-donor filter mating. The two E. coli donor strains, one containing the helper plasmid pRK2013 (4) and the other containing the mobilizable recombinant plasmid carrying inaZ, were mixed (108 cells per ml each in 1 ml of Luria broth) about 1 h prior to the final mixing with a suspension of 2 ml of 3 3 108 Z. mobilis recipient cells. Selection of transconjugant colonies was made on complete medium plates containing 100 mg of chloramphenicol or 40 mg of tetracycline per ml, including 20 mg of rifampin per ml for counterselection of E. coli. Ice nucleation tests. Cultures were grown at 248C for 24 h. Except as otherwise stated, antibiotic selection was routinely employed to ensure plasmid maintenance. Cells were separated from the culture medium (1 ml) by centrifugation in an Eppendorf microcentrifuge (12,000 3 g, 10 min), washed three times, and resuspended in the original volume of distilled water. Nucleation activity was quantified by droplet-freezing assays, essentially as described previously (6, 20). Whole cultures, washed cell suspensions, and culture supernatant were serially (10-fold) diluted, and 10-ml droplets from each dilution were placed on the surface of an aluminum foil sheet (spray coated with a 2% solution of paraffin in xylene and heat dried to remove the solvent) floating on an ethanol bath set at 298C (or as otherwise specified). Ice nucleation activity was calculated by the equation of Vali (20), using a software program (6a), and was expressed as the logarithm of ice nuclei per CFU of each fraction. Cell pellets and supernatants contained about 109 and 102 CFU/ml of culture, respectively. Differences in expression levels were calculated by taking the square root of the differences in nucleation activity levels (in two other bacteria that have been examined, the ice nucleation activity levels measured at 298C increase as the second power of the amount of InaZ protein in the cells [6, 16]).

RESULTS Subcloning and transfer of inaZ in Z. mobilis. Plasmid pUZ119 (2) (Fig. 1A) carries the inaZ gene on a 3.7-kb EcoRI fragment from Tn3-Spice (6) inserted in the EcoRI site of

FIG. 1. Partial physical maps of inaZ reporter plasmids: (A) pUZ119 (6.90 kb); (B) pPTZ3-inaZ (10.50 kb); (C) pDS3154-inaZ (11.24 kb); (D) pCG4521 (29.80 kb). REPpUC119, REP322, REP325, and REPpZMO3, replicons of pUC119, pBR322, pBR325, and the Z. mobilis natural plasmid pZMO3, respectively; pZMO2, the entire sequence of Z. mobilis natural plasmid pZMO2. ä and ã relevant promoters; ., direction of replication; 3, orientation of the inaZ reading frame. The inaZ fragment containing the ice nucleation gene is 3.7 kb; the remaining portions of the plasmids are not drawn to scale. E, EcoRI; H, HindIII; P, PstI.

VOL. 61, 1995

pUC119 and with inaZ in the opposite orientation relative to the lac promoter of the vector. The inaZ gene was excised from this plasmid as a 3.7-kb PstI fragment and was subcloned in the PstI site of the pRK2013-mobilizable vectors pPTZ3 (12) and pDS3154 (15) downstream from the promoters of the pyruvate decarboxylase gene of Z. mobilis (Ppdc) and the b-lactamase gene of the E. coli plasmid pBR325 (Pbla), respectively (Fig. 1B and C, respectively). The E. coli DH5a transformants were initially screened for ice nucleation activity, and the plasmids contained in them were phenotypically characterized by restriction enzyme digestion and by hybridization with the digoxygenin-11-dUTP-labelled inaZ fragment as a probe. Only constructs carrying the inaZ gene in the appropriate orientation relative to Ppdc and Pbla promoters expressed ice nucleation activity. These constructs were transferred from E. coli to Z. mobilis CU1Rif2 by pRK2013-assisted conjugation, and Z. mobilis transconjugants expressing ice nucleation activity were obtained. These transconjugants were designated Ice1. Plasmid pCG4521 (Fig. 1D) is a derivative of the pRK2013mobilizable cosmid vector pLAFR3 (18) containing the inaZ gene under the control of the promoter of hrpR (PhrpR), a gene that regulates the expression of other hrp genes in P. syringae pv. phaseolicola (5a, 11). This plasmid was transferred to Z. mobilis CU1Rif2 as described above, and Ice1 transconjugants were isolated. Transconjugants harboring the above plasmids were analyzed for the presence and structural integrity of the plasmids by back-transformation of plasmid DNA in E. coli, restriction digestion, and Southern hybridization. All plasmids analyzed from Z. mobilis transconjugants and E. coli back-transformants showed the same restriction profiles as those of the respective original constructs when hybridized under stringent conditions with the inaZ probe. These results demonstrate that Z. mobilis can be readily converted to an ice-nucleating organism and that the two heterologous promoters examined, Pbla and PhrpR, are expressed in Z. mobilis. Relative strengths of the Ppdc, Pbla, and PhrpR promoters in Z. mobilis. It has been reported earlier that promoter strengths and levels of gene expression in various bacteria can be measured by using the inaZ gene as a reporter (6, 16). In this study we compared the strengths of the Ppdc, Pbla, and PhrpR promoters in Z. mobilis and E. coli transconjugants carrying plasmids pPTZ3-inaZ, pDS3154-inaZ, and pCG4521, respectively, by assaying ice nucleation activity at 298C. The results are summarized in Fig. 2 (for reasons discussed below, the values presented here were determined after pelleting and resuspending cells in water). In Z. mobilis CU1Rif2, the homologous Ppdc-inaZ fusion gave the highest ice nucleation activity levels among the three fusions studied and the PhrpR-inaZ fusion gave the lowest. Whereas the reporter activity levels measured in these experiments do not solely reflect promoter strength (since the copy number of the different plasmid replicons is unknown and, in the case of pCG4521, a significant fraction of the population fails to maintain the plasmid even under selective conditions [see below]), Ppdc appears to be the strongest of the three promoters in Z. mobilis and PhrpR appears to be the weakest. The levels of ice nucleation activity expressed in E. coli were roughly equal for all three fusions (the differences did not exceed ca. threefold). Given the likely low copy number of pCG4521 (RK2 replicon) compared with those of the other two plasmids (ColE1-like), the hrpR promoter appears to be particularly strong in this host. Assuming that the three replicons maintain the same relative copy numbers in the two bacterial hosts, the apparent strengths of the Pbla and PhrpR promoters are considerably weaker (about 10- and 35-fold, respectively) in Z. mobilis compared with E. coli.

ICE NUCLEATION GENE EXPRESSION IN Z. MOBILIS

275

FIG. 2. Ice nucleation activity in Z. mobilis (lanes 1, 2, 3, 4, and 5) and E. coli (lanes 6, 7, and 8) strains harboring inaZ reporter plasmids as follows: 1, CU1Rif2/pPTZ3-inaZ (Ppdc); 2, CU1Rif2/pDS3154-inaZ (Pbla); 3, UVS51/ pDS3154-inaZ; 4, CU1Rif2/pCG4521 (PhrpR); 5, UVS51/pCG4521; 6, DH5a/ pPTZ3-inaZ; 7, DH5a/pDS3154-inaZ; 8, DH5a/pCG4521.

The pDS3154-inaZ and pCG4521 plasmids were also introduced in a previously described Z. mobilis mutant, uvs-51, that is impaired for segregational plasmid stability (21). The pPTZ3-inaZ plasmid could not be introduced in this mutant because the pZMO2 constituent replicon, which is the one functioning in wild-type Z. mobilis cells, is extremely unstable in the uvs-51 background even though the strain lacks the pZMO2 plasmid found in the parental CU1Rif2 strain (Table 1). The presence of the above two inaZ fusion plasmids in uvs-51 transconjugants was confirmed by back-transformation in E. coli and restriction-hybridization as described above, and the segregational stability was examined under both selective and nonselective growth conditions. The ice nucleation activity levels expressed by these plasmids in strain uvs-51 showed the same qualitative relationship as those obtained in strain CU1Rif2. Plasmid pDS3154-inaZ gave essentially the same levels of ice nucleation activity in strains uvs-51 and CU1Rif2 (Fig. 2, lanes 2 and 3). The stability of this plasmid was essentially the same in these two host strains under selective and nonselective conditions, as described before for plasmid pDS3154 (21). On the other hand, pCG4521 expressed significantly lower ice nucleation activity in uvs-51 than in CU1Rif2 (Fig. 2, lanes 4 and 5). Although the plasmid was quite unstable in both strains, it exhibited greater segregational loss in the mutant. In cultures grown under selective conditions, only about 15% of the uvs-51 cells retained the plasmid after 10 cell divisions compared with 50% for CU1Rif2; the plasmid was effectively lost under nonselective conditions after about 30 cell divisions in uvs-51 cultures and after about 60 divisions in CU1Rif2 cultures. Plasmidless cells did not retain any of the ice nucleation activity. These results suggest that plasmid segregational stability may influence ice nucleation activity. However, plasmid copy number and protein decay experiments are necessary to clarify this point. Characteristics of ice nucleation activity in Z. mobilis. An unusual observation made in the course of this study was that the ice nucleation activity levels measured in whole cultures often exceeded one nucleus per CFU. Values of greater than

276

DRAINAS ET AL.

APPL. ENVIRON. MICROBIOL.

viable cells found in the supernatant but may be present in a cell-free form. However, no detectable ice nucleation activity was recovered when the spent growth medium was passed through 0.22-mm-pore-size Millipore filters, suggesting that supernatant ice nucleation could be associated with nonfilterable particles or nonviable cells.

DISCUSSION

FIG. 3. Ice nucleation spectra of Z. mobilis harboring pPTZ3-inaZ. (A) Cell suspension (109 CFU/ml); (B) culture medium (102 CFU/ml). The numbers of ice nuclei were 108 and 105 per ml, respectively. Ice nucleation activities are expressed on the basis of the number of CFU present in each fraction.

1.0 nucleus per cell are not expected if the ice nuclei are cell bound (22). This is the case with most bacteria studied to date except Erwinia herbicola, which sheds cell-free ice nuclei into the growth medium (10). To determine if a similar phenomenon occurred in Z. mobilis, CU1Rif2 transconjugants expressing inaZ under the control of Ppdc were grown at 248C for 24 h and centrifuged at 12,000 3 g for 10 min, and the pellets were washed and resuspended in the original volume of sterile distilled water. The frequency distribution of ice nuclei active at different temperatures (ice nucleation spectrum) of the resuspended cell pellet and that of the supernatant were determined separately and were expressed on the basis of CFU present in each fraction. The cumulative ice nucleation activity plots (Fig. 3) revealed the following: (i) the threshold nucleation temperature (warmest temperature in which droplet freezing was observed) was 248C and was associated with the cell pellet; (ii) the activity log (ice nuclei per cell) found in the supernatant was .0 for the most part; and (iii) the supernatant contained almost exclusively ice nuclei active below 278C and was totally devoid of nuclei active at warmer temperatures. These findings suggest that ice nuclei may not be associated with the few

The development of new and more effective ice-nucleating agents is an extremely interesting and active subject today at the scientific and industrial levels. Various inorganic and organic compounds have been used as ice-nucleating agents for artificial snowmaking and cloud seeding. Generally, these substances have very low (cold) threshold nucleation temperatures and, consequently, a narrow utility range (for a review, see reference 8). Bacterial ice nuclei have much higher (warmer) temperature thresholds and can be more effective than inorganic or organic ice-nucleating factors (8, 22). Although Ice1 bacteria can have great potential for many applications, their use is limited because of their possible deleterious effects on plants and their unpredictable effects on humans and other animals. The creation of Ice1 nonepiphytic, nonpathogenic GRAS bacteria will help reduce such potential risks. Expression of the P. syringae inaZ gene has been reported in many gram-negative bacteria, including E. coli (7), fluorescent Pseudomonas spp., and Agrobacterium and Rhizobium spp. (6, 9), as well as in plants (2). However, this is the first report of the expression of an ice nucleation gene in a nonpathogenic GRAS bacterium. Moreover, it was shown that the heterologous promoters Pbla and PhrpR can be expressed in Z. mobilis. The differential expression of inaZ of P. syringae under the control of three different promoters revealed that ice nucleation activity can be used to monitor levels of gene expression and apparent promoter strength in Z. mobilis. Besides promoter strength, the ice nucleation activity levels were influenced by plasmid vector stability. This was concluded by comparing ice nucleation activity conferred by the plasmids pDS3154-inaZ and pCG4521 in strains CU1Rif2 and uvs-51. pDS3154 is a very stable plasmid in both strains because of the Z. mobilis replicon that it contains (15). Ice nucleation activity conferred by pDS3154-inaZ was the same in CU1Rif2 and uvs-51. On the other hand, pCG4521 was found to be less stable in uvs-51 cells. Similarly, ice nucleation activity was found to be lower in uvs-51 cells harboring pCG4521. It is suggested that lower ice nucleation activity in Z. mobilis uvs-51 is not due to lower promoter strength but reflects plasmid loss and/or lower copy number (gene dosage) of inaZ per cell. On the basis of these observations, Ice reporter genes may also be useful for measurements of plasmid stability. A potentially interesting finding was that the culture medium of Z. mobilis Ice1 cells retained an unusually high ice nucleation activity compared with the number of cells still remaining in the supernatant after centrifugation. Complete loss of the activity following filtration of the supernatant through 0.2-mm-pore-size membranes indicates that ice nucleation protein aggregates might be associated with nonfilterable particles. One possibility is that very large membrane vesicles may be released into the medium during growth of Z. mobilis cells. Alternatively, ice nucleation in the supernatant may also be associated with nonculturable cells. The fact that Z. mobilis ice nuclei were both abundant and very active at relatively warm temperatures makes Z. mobilis Ice1 cells very attractive for a variety of biotechnological applications.

ICE NUCLEATION GENE EXPRESSION IN Z. MOBILIS

VOL. 61, 1995 ACKNOWLEDGMENTS We thank S. E. Lindow (University of California, Berkeley) for helpful discussions and provision of facilities and G. A. Sprenger (Forschungszentrum KFA, Ju ¨lich, Federal Republic of Germany) for the kind offer of the pPTZ3 vector. Part of this work was supported by the University of Ioannina (sabbatical leave to C. Drainas) and by the Greek government and EU (STRIDE-HELLAS 33). REFERENCES 1. Afendra, A. S., and C. Drainas. 1987. Expression and stability of a recombinant plasmid in Zymomonas mobilis and Escherichia coli. J. Gen. Microbiol. 133:127–134. 2. Baertlein, D. A., S. E. Lindow, N. J. Panopoulos, S. P. Lee, M. N. Mindrinos, and T. H. H. Chen. 1992. Expression of a bacterial ice nucleation gene in plants. Plant Physiol. 100:1730–1736. 3. Doelle, H. M., L. Kirk, R. Crittenden, H. Toh, and M. B. Doelle. 1993. Zymomonas mobilis—science and industrial application. Crit. Rev. Biotechnol. 13:57–98. 3a.Drainas, C. Unpublished data. 4. Figurski, D. H., and D. R. Helinski. 1979. Replication of an origin-containing derivative of plasmid RK2 dependent on a plasmid function provided in trans. Proc. Natl. Acad. Sci. USA 76:1648–1652. 5. Green, R. L., and G. J. Warren. 1985. Physical and functional repetition in a bacterial ice nucleation gene. Nature (London) 317:645–648. 5a.Grim, C., and N. J. Panopoulos. Unpublished data. 6. Lindgren, P., R. Frederick, A. G. Govindarajan, N. J. Panopoulos, B. J. Staskawicz, and S. E. Lindow. 1989. An ice nucleation reporter gene system: identification of inducible pathogenicity genes in Pseudomonas syringae pv. phaseolicola. EMBO J. 8:1291–1301. 6a.Lindow, S. E. Personal communication. 7. Lindow, S. E., E. Lahue, A. G. Govindarajan, N. J. Panopoulos, and D. Gies. 1989. Localization of ice nucleation activity and the iceC gene product in Pseudomonas syringae and Escherichia coli. Mol. Plant Microbe Interact. 2:262–272. 8. Margaritis, A., and A. Singh Bassi. 1991. Principles and biotechnological

277

applications of bacterial ice nucleation. Crit. Rev. Biotechnol. 11:277–295. 9. Panopoulos, N. J. Ice nucleation genes as reporters. In R. Lee, L. Gusta, and G. Warren (ed.), Principles and applications of biological ice nucleation, in press. American Phytopathology Society Press, St. Paul, Minn. 10. Phelps, P., T. H. Giddings, M. Prochoda, and R. Fall. 1986. Release of cell-free ice nuclei by Erwinia herbicola. J. Bacteriol. 167:496–502. 11. Rahme, L. G., M. N. Mindrinos, and N. J. Panopoulos. 1991. Genetic and transcriptional organization of the hrp cluster of Pseudomonas syringae pv. phaseolicola. J. Bacteriol. 173:575–586. 12. Reynen, M., I. Reipen, H. Sahm, and G. A. Sprenger. 1990. Construction of expression vectors for the gram-negative bacterium Zymomonas mobilis. Mol. Gen. Genet. 223:335–341. 13. Sahm, H., S. Bringer-Meyer, and G. A. Sprenger. 1992. The genus Zymomonas, p. 2287–2301. In A. Balows et al. (ed.), The procaryotes, 2nd ed. Springer-Verlag, Berlin. 14. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y. 15. Scordaki, A., and C. Drainas. 1990. Analysis of plasmid pZMO3 of Zymomonas mobilis ATCC 10988. Plasmid 23:56–66. 16. Southworth, M. W., P. K. Wolber, and G. J. Warren. 1988. Non-linear relationship between concentration and activity of a bacterial ice nucleation protein. J. Biol. Chem. 263:15211–15216. 17. Sprenger, G. A., M. A. Typas, and C. Drainas. 1993. Genetics and genetic engineering of Zymomonas mobilis. World J. Microbiol. Biotechnol. 9:17–24. 18. Staskawicz, B., D. Dahlbeck, N. Keen, and C. Napoli. 1987. Molecular characterization of cloned avirulence genes from race 0 and race 1 of Pseudomonas syringae pv. glycinea. J. Bacteriol. 169:5789–5794. 19. Swings, J., and J. DeLay. 1977. The biology of Zymomonas. Bacteriol. Rev. 41:1–46. 20. Vali, G. 1971. Quantitative evaluation of experimental results on the heterologous freezing nucleation of supercooled liquids. J. Atmos. Sci. 28:402– 409. 21. Vartholomatos, G., M. A. Typas, and C. Drainas. 1993. An ultravioletsensitive mutant of Zymomonas mobilis affecting the stability of its natural plasmid pZMO2. Plasmid 29:10–18. 22. Wolber, P. K. 1992. Bacterial ice nucleation. Adv. Microb. Physiol. 31:203– 237.