When Coupled to Natural Transformation in Acinetobacter sp. Strain ADP1, PCR Mutagenesis Is Made Less Random by Mismatch Repair

Random PCR mutagenesis is a powerful tool for structure-functionanalysis of targeted proteins, especially when coupled withDNA integration through natural transformation followed by selectionfor loss of function. The technique has been applied successfullyto structure-function analysis of transcriptional regulators,enzymes, and transporters in Acinetobacter sp. strain ADP1.However, the mismatch repair system prevents the full spectrumof nucleotide substitutions that may be selected at the levelof protein function from being recovered. This barrier may beovercome by introducing PCR-mutagenized genes into strains inwhich the corresponding genes have been deleted.

INTRODUCTION

Top

Introduction

Random PCR mutagenesis provides an array of mutations, manyof which cause amino acid substitutions that may alter the functionalproperties of proteins. When coupled to natural transformationin Acinetobacter sp. strain ADP1, the procedure can be streamlined,as PCR amplicons of targeted genes are provided directly tothe host and integrated into the chromosome by homologous recombination(15). When followed by a suitable screen or selection, the processcan provide an overview of the amino acid side chains that contributeto function. A merit of the procedure is its objectivity becausemutations are identified on the basis of phenotypic variation.No presumptions are brought to bear on the identification oftargets for mutagenesis, so amino acid contributions that mightnot have been predicted can be characterized (7, 10, 24). Nevertheless,the procedure must be regarded with caution because not allnucleotide substitutions are made with equal probability duringPCR amplification (34). Furthermore, as described here, mismatchrepair (MMR) also contributes to making the process less thanfully random because it reduces the frequencies of transitionmutations that are introduced into the chromosome.

Transformation-facilitated random PCR mutagenesis has been successfullyapplied to proteins involved in the catabolism and regulationof phytochemicals in the nutritionally versatile Acinetobactersp. strain ADP1. These investigations have given valuable insightinto structure-function relationships in these proteins (7,15, 16, 17, 24, 29). However, when we applied the random PCRmutagenesis approach to pcaK, a gene encoding a protocatechuatetransporter belonging to the ubiquitous major facilitator superfamily(28), evidence of bias in the mutant collection was observed.A library of 117 pcaK mutants was analyzed, and in 20 cases,identical base pair substitutions were recovered in independentlyderived mutants (A. Buchan and L. N. Ornston, unpublished data).While saturation was being approached with respect to the fieldof pcaK variants that could be recovered using this technique,absent from the collection were several anticipated mutations,including those abolishing conserved charged residues in transmembrane-spanningregions. Furthermore, the nucleotide substitutions were heavilybiased towards transversions (71%) relative to transitions (29%).This pattern contrasts the distribution of nucleotide substitutionsrecovered from random PCR mutagenesis approaches coupled withcloning and expression of a target gene in Escherichia coli(Fig. 1). As generation of PcaK mutants in Acinetobacter sp.strain ADP1 depends upon replacement of the wild-type pcaK allelewith mutated variants via homologous recombination, a transformation-associatedbias caused by MMR was considered.

View larger version (18K):
[in this window]
[in a new window]
FIG. 1. Distribution of single-nucleotide substitutions generated by a general random PCR mutagenesis approach (41). Natural transformation through which heteroduplex DNA was formed (25, 26) was used to obtain Acinetobacter mutants used in this analysis (15 [n = 76]; A. Buchan and L. N. Ornston, unpublished data [n = 108]). Mutants derived from E. coli-based approaches in which heteroduplex DNA is not formed were obtained from the following references: references 18 (n = 48), 30 (n = 14), 1 (n = 10), 36 (n = 35), 2 (n = 8), 33 (n = 9), 35 (n = 6), 42 (n = 8), 9 (n = 5), 19 (n = 18), 37 (n = 11), 40 (n = 13), 27 (n = 19), 4 (n = 5), and 20 (n = 76). If nucleotide substitutions are not explicitly described in the references, they were deduced from protein substitutions, and ambiguous substitutions were omitted. Error bars represent variations in frequencies among studies conducted with the two organisms. A. baylyi, Acinetobacter baylyi.

Prior studies of MMR systems of divergent bacterial lineagesprovide a framework for how cellular machinery might block integrationof transition-containing PCR products in Acinetobacter. MMRplays a well-recognized role in repair of replication-basederrors, but it is also active in preventing recombination betweendivergent sequences (31, 32, 38, 39). Heteroduplex DNA containingeven single-nucleotide mismatches formed during homologous recombinationcan be targets of MMR-mediated heteroduplex rejection, alsoknown as antirecombination (5, 23). MutS, the mismatch bindingprotein and an essential component of MMR, preferentially bindsmismatches that arise from transition mutations over those arisingfrom transversion substitutions (3), thereby affecting the frequencywith which mutant alleles are integrated into the chromosomeduring recombination (11). Furthermore, recent studies of E.coli have demonstrated that MMR-mediated sequence-specific biasesin genetic engineering approaches employing bacteriophage ${lambda}$ arerelieved with MutS-defective cells (6). Therefore, an interpretationof the conflicting substitution profiles presented in Fig. 1is that the Acinetobacter MMR system reduces the integrationefficiency of transition mutation-bearing PCR products duringthe transformation step. To test this hypothesis, PCR mutagenesisof pcaK was conducted with strains containing or lacking functionalMutS, and the resulting nucleotide substitutions in the mutantgenes were compared.

Acinetobacter strain ADP8502 contains the ${Delta}$ vanK7602 and ${Delta}$ pcaBDmutations that facilitate selection of PcaK-deficient strainson medium containing protocatechuate and a nonselecting growthsubstrate (7). The basis of the selection is twofold. First,as both VanK and PcaK transport protocatechuate (8), selectionfor strains defective in PcaK activity is carried out in a ${Delta}$ vanKbackground. Second, PcaB-deficient strains that take up andmetabolize protocatechuate accumulate a toxic metabolite thatinhibits growth on the selective medium (12). Inactivation ofMutS in strain ADP8504 was achieved by an ${Omega}$ (SpR/StR) insertionby the following procedure: strain ADP8502 was transformed withthe linearized plasmid pZR7008 (38) followed by selection forresistance to streptomycin and spectinomycin at respective concentrationsof 10 and 40 µg/ml. PCR mutagenesis was performed usingcrude cell lysates of Acinetobacter sp. strain ADP1 with theprimers PK1 and PK2 (7) in 100-µl volumes. Twenty-fivemicroliters from each reaction mixture was added in parallelto 500-µl competent cultures of ADP8502 and ADP8504. Following3 h of incubation at 37°C with shaking, the transformationreaction mixtures were plated on minimal medium (31) containing5 mM fumarate and 1 mM protocatechuate and incubated in thedark at 22°C for 28 h. The ${Delta}$ pcaBD deletion was correctedby transformation of purified strains with a plasmid carryingwild-type pcaBD (pZR8550). The pcaK genes of mutant strainswere sequenced in their entirety by using PK1, PK2, and an internalprimer, PK3 (5' CCGATTTCTAACAGGTATCGGATTG 3'). pcaK was sequencedfrom representative mutants derived from ADP8502 (n = 13) andADP8504 (n = 18) transformed with Taq-amplified pcaK DNA. Substitutionsare presented relative to the pcaK coding strand and as "nucleotide/nucleotide"(e.g., T/C). The arrangement of the nucleotide designationsdoes not necessarily reflect the original substitution (i.e.,T/C represents either a T-to-C or C-to-T substitution).

Comparison of PCR-generated mutations in strains containing and lacking functional MutS.

Top

Comparison of pcr-generated...

Consistent with the mutator phenotype typical of MutS-deficientstrains (5), the frequency of spontaneous mutations giving riseto protocatechuate-resistant colonies was sevenfold higher instrain ADP8504 than that in strain ADP8502. Transformation ofADP8504 with Taq-amplified pcaK DNA resulted in 30-fold moreprotocatechuate-resistant colonies than did that of the samestrain receiving no PCR product (see Fig. S1 in the supplementalmaterial). This is similar to the 24-fold increase seen whenamplified products are provided to the wild-type mutS strainADP8502 and suggests that more than 95% of the protocatechuate-resistantcolonies derived from either strain transformed with pcaK PCRproducts possess Taq-introduced mutations.

The spectrum of PCR-generated mutations identified in the MutS-deficientstrain (ADP8504) receiving pcaK PCR products was distinct fromthat found with the MutS-positive strain (ADP8502). Transitionsaccounted for 84% of all pcaK nucleotide substitutions derivedfrom the MutS-deficient strain compared to 42% of substitutionsobtained from ADP8502 provided with the same PCR product pools(see Table S2 in the supplemental material). The substitutionprofiles for the 12 pcaK variants obtained in the functionalMutS background parallel those found in a previous study ofPCR-generated mutations in pcaK (A. Buchan and L. N. Ornston,unpublished data) and were folded into the larger data set forcomparisons. This substitution pattern was also remarkably consistentwith that of another large mutant library generated in Acinetobactersp. strain ADP1 (PobR) (see Table S1 in the supplemental material)(15). Relative to substitutions recovered in functional MutSstrains, the frequencies of both A/G and T/C transitions wereelevated about three- and twofold, respectively, in the MutS-deficientstrain. A second feature characteristic of the PcaK mutantsderived from the MutS-deficient strain was that most (73%) containedmultiple (ranging from two to five) mutations within pcaK (seeTable S2 in the supplemental material). In contrast, only 15%of the mutants derived from the functional MutS strain receivingthe same PCR product pools had more than one mutation in pcaK,a finding consistent with previous studies of PcaK and PobR(see Table S1 in the supplemental material). However, significantlyhigher frequencies of multiple mutations at a target locus havebeen reported using this approach (24). As MMR systems are subjectto saturation (13, 21), the prevalence of multiple mutationsin a target gene is likely a reflection of the capacity of theMMR system and the composition of the PCR pool provided as donorDNA.

Circumventing the transformation-associated bias.

Top

Circumventing the transformation...

Application of the transformation-facilitated random PCR mutagenesisapproach to a heterologous gene yielded a different patternof nucleotide substitutions. Kok et al. (15) designed a dockingsite in the Acinetobacter sp. strain ADP1 chromosome that wouldallow integration and expression of foreign genes and PCR-mutagenizedvariants of that gene. In this system, the recipient straindoes not have DNA corresponding to the gene targeted for PCRmutagenesis. Of the 25 nucleotide substitutions identified in17 mutants, 92% were transitions. This pattern is analogousto that of E. coli-based libraries (see Table S1 in the supplementalmaterial) and consistent with our interpretation of the roleof MMR. If the gene targeted for PCR mutagenesis is not presentin the chromosome of the Acinetobacter recipient strain, heteroduplexesare avoided and the MMR-mediated biases associated with thetransformation- facilitated random PCR mutagenesis approachare circumvented. Therefore, through chromosomal deletion ofthe gene of interest and the use of a counterselectable marker,such as sacB (14, 22), to facilitate integration of mutagenizedvariants, it is likely that a different spectrum would be recovered.It would be beneficial to take this approach in the presenceof a functional MutS because, as noted above, deficiencies inMutS give rise to a high frequency of strains with multiplemutations in the target gene.

ACKNOWLEDGMENTS

We are grateful to David Young and Graham Walker for valuablediscussions.

This research was supported by NIH grant GM063628. A.B. wassupported by NSF Postdoctoral Research Fellowship DBI-0200164.

FOOTNOTES

* Corresponding author. Mailing address: Department of Molecular, Cellular, and Developmental Biology, Yale University, P. O. Box 208103, New Haven, CT 06520-8103. Phone: (203) 432-3498. Fax: (203) 432-3350. E-mail: nicholas.ornston{at}yale.edu.

Supplemental material for this article may be found at http://aem.asm.org/.

Present address: Department of Microbiology, M409 Walters LifeSciences, University of Tennessee, Knoxville, TN 37996.

REFERENCES

Top