THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 263, No. 29, Issue of October 15, pp. 15211-15216,1988 Printed in U.S.A.

Nonlinear Relationship between Concentration and Activity of a Bacterial Ice Nucleation Protein*

(Received for publication, February 29, 1988)

Maurice W. Southworth, Paul K. Wolber, and Gareth J. Warren From Advanced Genetic Sciences, Znc., Oakland, California 94608

The expression level of an ice nucleation gene (in&) was varied in Escherichia coli to observe the relation- ship between activity and gene product. The ice nu- cleation activity increased as the 2nd to 3rd power of the membrane concentration of the inaZ gene product, implying that molecules of InaZ protein interact coop- eratively in groups of two to three at the rate-limiting step of ice nucleus assembly. The 2nd to 3rd power relationship was independent of the threshold temper- ature at which ice nucleation was measured and was consistent over a 500-fold range of protein concentra- tion. Such a relationship indicates that the same rate- limiting step must be common to the formation of ice nuclei displaying all the various threshold tempera- tures within a bacterial population. Observations of Pseudomonas syringae, expressing the inaZ gene at various levels, were consistent with a similar relation- ship and hence a similar mechanism of ice nucleus assembly in Pseudomonas.

Many materials can nucleate the formation of ice in super- cooled water, and each nucleation site has a characteristic threshold temperature, at and below which it is active. Certain Gram-negative bacteria, including some strains of Pseudom- onas, contain nucleation sites active even at temperatures above -3 "C (Schnell and Vali, 1972; Maki et al., 1974; Lindow et al., 1978); ice nuclei with such high thresholds are otherwise rare in nature. This phenomenon appears to potentiate the damaging of plants by frost (Lindow, 1983).

The ice nucleation phenotype can be conferred on Esche- richia coli by the presence of a single gene from Pseudomom syringae (Orser et al., 1985). One such gene (in&) has been sequenced (Green and Warren, 1985) and its gene product has been identified (Wolber et al., 1986). Most (70%) of the protein's primary structure is repetitive, with motifs of 8, 16, and 48 amino acids (Green and Warren, 1985). It seems likely that these repeats are responsible for aligning water molecules into an ordered array, and thus acting as a template for the formation of a seed crystal of ice (Warren et al., 1986).

To examine the process by which synthesis of the InaZ protein leads to the formation of ice nuclei, genetic manipu- lations were carried out to vary the rate of synthesis of InaZ protein and thus vary its concentration in uiuo. The concen- tration of InaZ protein was examined in relation to the frequency at which ice nuclei (measured over a range of threshold temperatures) were formed in bacterial populations.

*This work was supported by United States National Science Foundation Grants IS18560178 and ISI8701197. The costs of publi- cation of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertise- ment'' in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The observed relationship appears to be significant in the consideration of ice nucleus assembly and is the subject of this report.

EXPERIMENTAL PROCEDURES

Bacterial Strains-Two E. coli K12 strains were used; JM83, ara, A&c-pro), rpsL, thi, @OdlacZM15 (Vieira and Messing, 1982) for many of the plasmid constructions and JC10291, ara, arg, galK, his, lacy, leuB, mtl, proA, rpsL, supE, thi, thr, tsx, A(srl-recA)303 (Willis et al., 1981) for all the ice nucleation assays and protein determina- tions. RGP36 (Warren et al., 1987), a A i d 9 9 derivative of P. syringae S203 (Green and Warren, 1985), was used in all the Pseu- domonas experiments.

Plasmid Constructions-Vectors for expressing i d in E. coli were constructed as depicted in Fig. 1. A 4.4-kilobase fragment from pRLG12 (Green and Warren, 1985) was cloned, via the intermediate pGJ144, into pUC9 (Vieira and Messing, 1982) to give pMWS1. DNA fragments containing the complete i d gene were cloned into pUC8, pUC9, and pBR322. The different fusion constructs were created by cloning at different restriction sites upstream from the i d gene. Four restriction sites were used, XhoII (position -22 relative to the inaZ initiator codon), Aha111 (-28), PvuI (-106), andHindIII (-800). The vector pKK223-3 (Brosius and Holly, 1984) received i d via an Aha111 fusion, placing i d downstream of the strong tac promoter in plasmid pMWS10.

The wide host range plasmid pRK767 (derived from pRK747; Jones and Gutterson, 1988), was used in the construction of expression plasmids for experiments involving P. syringae. All the constructs shown in Fig. 2 used the Aha111 inaZ gene fragment, with pMWS24 derived from pMWS9 (HindIIIIEcoRI digest) and pMWS35 derived from pMWS13 (PstIIHindIII digest). pMWS28 was obtained by cloning the inuZ gene from pMWSlO (PstIIHindIII) into pKK233-2 (Amann and Brosius, 1985) and cloning an EcoRI fragment of this intermediate construct into pRK767.

DNA Preparations-DNA preparations were by the method of Holmes and Quigley (1981) and Birnboim and Doly (1979).

Pseudomonas Matings-Triparental matings were carried out with donor JC10291 strains containing mobilizable (pRK767-derived) and helper (pRK2013) plasmids and with RGP36 as recipient, by the method of Ditta et al. (1980). Transconjugants were selected for rifampicin and tetracycline resistance.

Culture Conditions-E. coli cells were cultured in L-broth at 37 "C to an OD,, between 0.4 and 0.5 and then incubated at 23 "C for 1 h. Ice nucleation assays and membrane preparations were then carried out.

Pseudomonas cells were inoculated onto NAG plates (nutrient agar + 2.5% gylcerol) by spreading and incubated at 24 "C for 3 days. Cells were washed off the plates, and ice nucleation assays and membrane preparations were then carried out.

Ice Nucleation Assay-The method was based on the droplet- freezing assay developed by Vali (1971), with modifications as de- scribed by Corotto et al. (1986). The number of droplets/sample was 10 or 20, with at least two different samples tested per foil; this made it possible to compare pairs of constructs directly in each experiment. Most of the samples were tested at least four times to increase the accuracy of the determinations.

Membrane Preparations-For E. coli the method of Ito et al. (1977) modified by Wolber et al. (1986) was used.

A Pseudomonas membrane preparation was developed that isolated all the InaZ protein but not all of the total membrane protein (Deininger et al., 1988). A suspension of 10'' cells in 1 ml of 10 mM

15211

15212 Activity of Ice Nucleation Protein

HE HE

FIG. 1. Construction of plasmids used to vary in& gene expression in E. coli. A, AhuIII; B, BamHI; E, EcoRI; H, HindIII; P , PstI; S, SmaI; X, XhoII; AIS, point of AhnIII-SmaI fusion; XIB, point of XhoII-BamHI fusion. Ap', Tc', genes encoding resistance to ampicillin and tetracycline, respectively. Open bars represent the i d gene; its direction of transcription is toward the pointed end. Cross- hatched bars represent antibiotic resistance genes and their direction of transcription is toward the pointed end. Arrows indicate direction of transcription from the corresponding promoter (tac, lac, or ina). All maps are drawn to scale. kb, kilobases.

sodium phosphate buffer (pH 7.0) was incubated for 15 min at 0 "C in the presence of 1 mM o-phenanthroline, 1 mg/ml lysozyme, 10 mM EDTA, and 40 mM octylthio-P-D-glucopyranoside. MgC12 (50 mM) and bovine pancreatic DNase (0.1 mglml) were then added and the suspension incubated a further 15 min on ice. The membrane material was pelleted and resuspended in 10 mM MgClz.

Both types of membrane preparation were stored at -20 "C until ready for use.

SDS'-Polyacrylamide Gel Electrophoresis and Immunoassay-The discontinuous buffer system of Laemmli (1970) was used; membrane samples were boiled for 10 min in SDS and solubilizer and loaded onto a 1-mm-thick acrylamide gel (5% separation and 3% stacking layers). Proteins were electroeluted from the gel onto nitrocellulose according to Towbin et al. (1979) and immersed in antibody (1:lOO dilution in buffer) specific to the InaZ protein for 1 h. After washing off the antibody solution, %3-protein A (1:2000 in distilled HzO) was added. After 1 h the 35S-protein A was washed off and an autoradi- ograph was taken. The protocol followed the manufacturer's instruc- tions for the Immuno-blot protein A-horseradish peroxidase assay

' The abbreviation used is: SDS, sodium dodecyl sulfate.

pMWS24 14.2 kb

pMws35 14.2 kb

pRLG25

HE

FIG. 2. Plasmids used to vary in& gene expression in P. sy- ringae. Q, PuuI; other conventions are as in Fig. 1; kb, kilobases.

200-

97-

68-

43- 26-

PURIFIED INAZ PROTEIN pMWS19 pMWS14 oMWS20

' 1 2 3 4 5 " 6 7 8 9 " 1 0 1 1 1 2 1 3 " 1 4 1 5 n -

FIG. 3. Autoradiogram of a Western blot probed with anti- InaZ antibody and 35S-protein A. Four series of dilutions were run, including that of purified InaZ protein as a standard. Each consisted of a series of 2-fold dilution steps, with dilution increasing from left to right. The markings at left indicate the migration positions of protein standards with molecular masses 200, 97, 68, 43, and 26 kDa.

(Bio-Rad), with 35S replacing the horseradish peroxidase label. After development, the autoradiograph was used as a template to indicate where on the nitrocellulose the InaZ protein had banded. Each lane was then cut out and the amount of 35S determined. Alternatively, the autoradiograph was scanned with a densitometer and the amount of protein determined by measuring the area under the peaks. (Den- sitometry was used only for estimating the lower quantities of protein, since it provides better sensitivity but a less linear response.)

method of Smith et al. (1985) and used the BCA (bicinchoninic acid) Total Protein Assay-All total protein measurements were by the

protein assay reagent, as supplied by Pierce Chemical Co.

RESULTS

Variation of I m Z Protein Concentrations in E. coli-The in& gene, along with various amounts of sequence upstream from its initiation codon, was inserted into different expres- sion vectors; this varied both the promoter strength and the distance between promoter and gene. The resulting plasmids were transformed into E. coli, giving bacterial strains that

Activity of Ice Nucleation Protein 15213

TABLE I Concentrations of membrane-associated ZnaZ protein from strains of E. coli and P. syringae

Plasmid Promoter Fusion site Total protein In& protein Extraction efficiency Molecules lnaZ protein/cell

w l m l d m 1 % pMWSlO tat AhnIII 6.05 720 64 5.4 X 104 pMWSl lac,ina HindIII 7.05 155 75 1.1 X 10' pMWS2O lac XhOII 7.15 155 76 1.1 X 104 pMWS9 lac AhnIII 8.55 155 91 8.7 X 103 pRLG12 puc,ina HindIII 8.90 56 95 3.0 X 103 pMWS14 Puc XhOII 7.70 36 82 2.3 X 103 pMWS19 Puc AhaIII 8.15 12.8 87 7.6 X 10' pMWS17 l n a HindIII 8.45 12.5 90 7.0 X lo2 pRLG25 None PVUI 8.45 1.8 90 1.0 x 10' pMWS21 None XhoII 8.65 0.4 92 2.3 X 10' pMWS13 None AhaIII 9.40 <0.3 100 4 . 7 X 10' pMWS28" trc AhaIII 2.65 325 100 9.2 X 104 pMWS24" lac AhaIII 2.30 29.6 87 9.6 X 103 pMWS35" unk AhnIII 1.80 0.5 68 2.1 x 10'

Results determined in P. syringae.

Temperature, "C FIG. 4. Cumulative ice nucleation frequency plots for E. coli

strains expressing in&. The plots are labeled by the names of the expression plasmids contained in the examined strains. The limit of detectability was estimated at 4 X lo-' ice nuclei/cell. Points are plotted at this limit where nucleation was undetectable; dashed lines join these limit points to actual data points.

varied in their levels of expression of in&. Ice nuclei are present in the membrane fraction of cells that express inaZ (Wolber et al., 1986). Therefore, membrane extracts were prepared in order to determine the concentrations of ice nucleation protein in the location where it is active.

Each membrane extract was titrated by running a series of %fold dilutions through an SDS-polyacrylamide gel, and com- paring band intensities with those of a standard. The standard consisted of purified InaZ protein (Wolber et al., 1986) which had been calibrated by measuring its protein concentration

according to Smith et al. (1985). Each gel received dilution series from two to three samples along with a dilution series of the standard. After electrophoresis, proteins were blotted to nitrocellulose and probed with anti-InaW and 3SS-protein A. Anti-InaW is an antibody raised against InaW protein (Wolber et al., 1986) which cross-reacts strongly with InaZ (Deininger et al., 1988). Bands were visualized by autoradi- ography (an example is shown in Fig. 3); the main band in each lane corresponds to intact InaZ protein, while the minor bands represent degradation products.

To estimate the band intensity of InaZ protein in a gel track, the quantity of bound %-protein A was measured after cutting the nitrocellulose into strips (using the corresponding autoradiogram as a template). The counts found on each strip were multiplied by the corresponding dilution factor to esti- mate the original concentration of InaZ protein, and a mean of such estimates was taken from each dilution series. This method was not suitable for the low level expression plasmids because of nonspecific binding of the 35S-protein A to the nitrocellulose; it was not possible to cut a strip without including a significantly contaminated border. Therefore, densitometry was performed on the autoradiogram to deter- mine the protein concentrations of the low level expression plasmids.

To correct for the varying efficiency of the membrane extraction protocol, the total protein concentration of each extract was measured. Assuming that the highest yield of total protein represented 100% recovery, we corrected the estimates of InaZ concentration by multiplying first estimates by the inverse of extraction efficiencies. The corrected figures were expressed as copy numbers of membrane-associated InaZ molecules/cell (Table I). For each calculation the bacterial concentration was estimated to be 5.6 x lo8 colony forming units/ml from the ODm value (0.7) and the conversion factor of 8 X lo8 (Maniatis et al.. 1982).

The InaZ protein concentrations in the membrane (Table I) correlated well with the strengths of the corresponding promoters. The strongest promoter was tac (De Boer et al., 19821, which was used in the construction of pMWS10. Clon- ing the inuZ gene into pUC8 or pUC9 placed the gene under control of either the lac promoter or an apparently weaker promoter reading in the opposite direction, the "puc" pro- moter. The final series of constructs involved the insertion of the inaZ gene between the PstI and EcoRI sites of pBR322, orienting the gene to read toward the EcoRI site. In this location, transcription congruent with the orientation of inaZ

15214 Activity of Ice Nucleation Protein r

0.5

0.4

0.3

0.2

0.1

0

Molecules lnaZ per cell

r

100

10-1

10-2

10-3

10-4

10-6

10-6

10-7

(linear scale x ;O,OOO)

FIG. 5. Alternative methods of plotting the relationship between ice nucleation frequency (ice nuclei present/cell) and InaZ protein concentration (molecules of InaZ protein/cell) in E. coli, using logarith- mic (log scale) and linear scales for each. The third dimension ( z axis) represents the degree of supercooling at which activity was measured. plotted on a logarithmic scale in each case. Where nucleation was undetectable, . - limit points were plotted.

-

is at a very low level. Cells containing such constructs on average contained fewer than 17 membrane-associated mole- cules of InaZ protein.

A consistent feature of the results is the influence of the site of transcriptional fusion. Fusions 800 base pairs upstream of the translational initiation codon (Hind111 fusions) gave more InaZ protein than proximal fusions to the same exoge- nous promoter. This might be due to the presence of a promoter within the 800 base pairs preceding the coding sequence (the “ina” promoter) whose strength in E. coli would be similar to that of the puc promoter. This is apparent when pMWS17 (having only the inu promoter) and pMWS19 (hav- ing only the puc promoter) are compared (7.0 x lo2 and 7.6 x 10’ copies/cell, respectively).

Measurement of Ice Nucleation Activities in E. coli-For each strain of interest, we determined the population fre- quency of ice nuclei active at a series of successively lower temperature points. The results were expressed as cumulative nucleation spectra (Vali, 1971) and formed the basis for comparing the ice nucleating phenotypes of the constructs.

There was a clear qualitative correlation between promoter strength and ice nucleation activity. This was seen most distinctly in the case of the strong tac promoter construct

(pMWS10): the first measurable threshold temperature (-2.8 “C) was the highest among all constructs, and at every temperature more ice nuclei were active than for any other construct (Fig. 4). In general, the order of activity at all temperatures followed the order of protein concentrations. The assays appeared to be sensitive enough to detect a con- sistent difference between XhoII and AhaIII fusion con- structs: the XhoII fusion constructs gave “better” ice nuclea- tion spectra than the corresponding AhaIII fusion constructs, whether fusion was to the puc promoter (comparing pMWS14 and pMWS19) or to the very weak promoter in pBR322 (comparing pMWS21 and pMWS13). The consistent quali- tative correlation between expression and phenotype led us to examine the nature of this relationship.

Quantitative Correlation of Protein Concentration and Ac- tivity in E. coli-The relationship between ice nucleation protein concentration in the membrane and its activity was examined quantitatively by various methods of plotting activ- ity (at threshold temperatures ranging from -3 to -10 “C) against concentration. (For these purposes, the data sets for pMWS1 and pMWS2O were combined by averaging, since they were very similar; likewise, the data sets for pMWS17 and pMWS19 were combined.) Whereas linear-linear and

Activity of Ice Nucleation Protein 15215

Molecules lnaZ per cell (log scale)

FIG. 6. A plot of the relationship between ice nucleation frequency and InaZ protein concentration in P. syringae, using logarithmic scales for each. The plot is comparable to that at lower right in Fig. 5.

logarithmic-linear plots curved sharply, logarithmic-logarith- mic plots more closely approximated straight lines over almost 3 orders of magnitude of protein concentration (Fig. 5). Thus, activity appears to vary as a power of protein concentration; a line of slope n on the logarithmic-logarithmic plot would indicate an nth power relationship. The relationship appeared similar at each temperature of measurement, as evidenced by the roughly parallel plots for different temperatures; there- fore, it must govern the formation of ice nuclei of any thresh- old temperature.

Below an InaZ protein content of 2.5 X lo3 copies/cell, the slope of approximately 2 indicates that the ice nucleation frequency increases as the square of the InaZ protein concen- tration. This range of concentrations covers all the pBR322- based constructs and the two puc promoter constructs with fusions at the AMI1 and X M I sites. Between protein con- tents of 2.5 X lo3 and 1.1 X lo4 molecules/cell, the slope of the plot increases, indicating that activity is proportional to the InaZ protein concentration raised to the 3rd to 4th power. Above 1.1 X lo4 copies/cell of InaZ protein, the relationship is approximately linear. However, the available methods can- not measure more than one nucleus/cell, and hence curves approaching this asymptote do not represent accurately the total number of ice nuclei present. This applies to the plots above 1.1 x lo4 copies/cell.

InuZ Protein Concentration and Activity in P . syringae- Plasmid constructs able to express in& to various levels in P. syringae were derived from the wide host range vector pRK767. The inaZ gene was placed under control of the trc and lac promoters in plasmids pMWS28 and pMWS24, re- spectively. The orientation of the inaZ gene is reversed, relative to pMWS24, in plasmid pMWS35: this places i d under control of unknown promoter(s) (which for convenience we designate the unk promoter).

The plasmid constructs were introduced into P. syringae RGP36 (Warren et al., 1987) by conjugation from E. coli. Strain RGP36 was chosen because it is derived from the strain which was the source of the inaZ gene, and because it contains a chromosomal deletion which prevents it from synthesizing InaZ protein or expressing nucleation activity. In experiments analogous to those in the preceding sections, derivative strains containing pMWS24, pMWS28, or pMWS35 were assayed for nucleation activity and also assayed for the content of InaZ protein in membrane extracts (Table I). The respective

constructs gave InaZ protein copy numbers of 9.6 x lo3, 9.2 X IO4, and 2.1 X 10' molecules/cell, thus covering a 450-fold range of protein concentration.

As before, we attempted to correlate nucleation activities with protein concentrations by means of logarithmic-logarith- mic plots (Fig. 6). The data set which we were able to collect from the three constructs was too limited to indicate the relationship unambiguously. However, the results were con- sistent with the power relationship observed in E. coli: lines joining data points had slopes of between 3.0 and 4.0 except where one of the data points was near to the asymptote, when the slopes were reduced.

DISCUSSION

We have observed a simple but markedly nonlinear rela- tionship between the concentration of InaZ protein in the cell membrane and the ice nucleation activity of the bacterial strain. The activity, expressed in terms of the abundance of ice nuclei formed, increases as the 2nd to 3rd power of the InaZ protein concentration. Our favored interpretation is that ice nuclei are assembled by aggregation of InaZ protein and that the 2nd to 3rd power dependence on concentration de- rives, by the law of mass action, from a bimolecular to tri- molecular assembly reaction. In general, any co-operative event could explain the observed relationship, and aggregation is merely a specific example. The co-operative interpretation seems straightforward if it is accepted that concentration governs activity. It seems very unlikely that the reverse is true: that the presence of ice nuclei governs the amount of protein. Experimental artifacts also seem implausible as sources of the observed relationship; neither the estimation of protein concentration nor the assay for nucleation activity are known to be subject to gross errors, and the relationship was seen over 3 orders of magnitude of protein concentration.

Different ice nuclei in a bacterial population are active at (or below) different threshold temperatures; the present study was intended to investigate these differences. It was surprising that the slope of the concentration-activity relationship did not differ significantly, whatever temperature threshold was used to define activity. Both theoretical considerations (Fletcher, 1958) and experimental evidence (Govindarajan and Lindow, 1988) have indicated that larger ice-like tem- plates will possess warmer threshold temperatures. (A size difference greater than 100-fold was inferred between nuclei active at -12 and -2 "C.) Because InaZ protein is believed to be the template for ice crystallization (Green and Warren, 1985), it had been expected that warmer threshold ice nuclei would be larger by virtue of a greater content of InaZ protein. A difference in InaZ content would in turn be expected to cause activity at different thresholds to be reduced dispropor- tionately when the supply of InaZ protein was limited. In- stead, it seems evident that formation of all types of nuclei is rate-limited by the same event. How, therefore, does the cell generate variability in threshold temperatures? One way would be by further aggregation of the nuclei initially formed; this process might not be limited by the availability of InaZ protein in the observed range of concentrations. Aggregation might instead be limited by stochastic chain-terminating events in the growing ice nucleus; thus the distribution of nucleus sizes would be independent of protein availability. Another way to generate variability would be by modification of an InaZ-containing particle of fixed size; such modification could cause deactivation or enhancement. The latter scenario does not conflict with either theoretical or experimental ob- servations on nucleus size, since the effective ice nucleus may

15216 Activity of Ice Nucleation Protein

be only a section of the InaZ-containing particle (Warren and Wolber, 1987).

The slope of concentration-activity relationship appears to be consistently lower in one part of the concentration range (less than 2.5 X lo3 copies/cell) than another (between 2.5 x lo3 and 1.1 x IO4 copies/cell). This effect could be an artifact of relatively minor errors in the estimation of protein concen- trations, which would not substantially affect our earlier conclusions.

Results obtained from P. syringae indicate that the concen- tration-activity relationship may be similar to that in E. coli. However, the nucleation spectra from P. syringae reproducibly begin at slightly warmer temperatures than those from E. coli. This difference might be due to the different lipid composi- tions of the membranes of the two species (Nikaido and Hancock, 1986).

Acknowledgments-We thank W. Tucker and P. Lund for criticism of the manuscript.

REFERENCES Amann, E., and Brosius, J. (1985) Gene (Amst.) 40,183-190 Birnboim, H. C., and Doly, J. (1979) Nucleic Acids Res. 7,1513-1525 Brosius, J., and Holly, A. (1984) Proc. Natl. Acad. Sci. U. S. A. 8 1 ,

Corotto, L. V., Wolber, P. K., and Warren, G. J. (1986) EMBO J. 5 ,

De Boer, H. A,, Comstock, L. J., and Vasser, M. (1982) Proc. Natl.

Deininger, C. A., Mueller, G. M., and Wolber, P. K. (1988) J. Bacteriol.

Ditta, G., Stanfield, S., Corbin, D., and Helsinki, D. R. (1980) Proc.

Fletcher, N. H. (1958) J. Chem. Phys. 29,572-576

6929-6933

231-236

Acad. Sci. U. S. A. 80, 21-25

170,669-675

Natl. Acad. Sci. U. S. A. 2 7 , 7347-7351

Govindarajan, A. G., and Lindow, S. E. (1988) Proc. Natl. Acad. Sci.

Green, R. L., and Warren, G. J. (1985) Nature 317,645-648 Holmes, D. S., and Quigley, M. (1981) Anal. Biochem. 114, 193-197 Ito, K., Sato, T., and Yura, T. (1977) Cell 11,551-559 Jones, J. D. G., and Gutterson, N. (1988) Gene (Amst.) 61,299-306 Laemmli, U. K. (1970) Nature 227,680-685 Lindow, S. E. (1983) Annu. Reu. Phytopathol. 21,363-384 Lindow, S. E., Amy, D. C., and Upper, C. D. (1978) Phytopathology

Maki, L. R., Galyon, E. L., Chang-Chien, M., and Caldwell, D. R. (1974) Appl. Microbiol. 28,456-460

Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) Molecular Cloniw: A Laboraton, Manual. I). 61. Cold Swing Harbor Labora-

U. S. A, 85, 1334-1338

68,523-527

tory, CLld Spring Ha-rbor, NY , - , "

Nikaido. H.. and Hancock. R. E. W. (1986) in The Bacteria: The Biology ofPseudomonas (Sokatch, J. R., ed) Vol. X, pp. 145-193, Academic Press, Orlando, FL

Orser, C., Staskawicz, B. J., Panopoulos, N. J., Dahlbeck, D., and Lindow, S. E. (1985) J. Bacteriol. 164,359-366

Schnell, R. C., and Vali, G. (1972) Nature 236, 163-165 Smith, P. K., Krohn, R. I., Hermanson, G. T., Mallia, A. K., Gartner,

F. H., Provenzano, M. D., Fujimoto, E. K., Goeke, N. M., Olson, B. J., and Klenk, D. C. (1985) Anal. Biochem. 1 5 0 , 76-85

Towbin, J., Staehelin, T., and Gordon, J. (1979) Proc. Natl. Acad. Sci. U. S. A. 76,4305-4309

Vali, G. (1971) J. Atmos. Sci. 28,402-409 Vieira, J., and Messing, J. (1982) Gene (Amst.) 19 , 259-268 Warren, G. J., and Wolber, P. K. (1987) Cry0 Lett. 8,204-215 Warren, G., Corotto, L., and Wolber, P. (1986) Nucleic Acids Res.

Warren, G. J., Lindemann, J., Suslow, T. V., and Green, R. L. (1987) in Biotechnology in Agricultural Chemistry (LeBaron, H., Mumma, R. O., Honeycutt, R. C., and Duesing, J. H., eds) pp. 215-227,

Willis, D. K., Uhlin, B. E., Amini, K. S., and Clark, A. J. (1981) Mol. American Chemical Society, Washington, D. C.

Gen. Genet. 183,497-504 Wolber, P. K., Deininger, C. A., Southworth, M. W., Vandekerckhove,

J., Van Montagu, M., and Warren, G. J. (1986) Proc. Natl. Acad. Sci. U. S. A. 83,7256-7260

14,8047-8060