Substitution, Insertion, Deletion, Suppression, and Altered Substrate Specificity in Functional Protocatechuate 3,4-Dioxygenases

Abstract

Protocatechuate 3,4-dioxygenase is a member of a family of bacterial enzymes that cleave the aromatic rings of their substrates between two adjacent hydroxyl groups, a key reaction in microbial metabolism of varied environmental chemicals. In an appropriate genetic background, it is possible to select for Acinetobacter strains containing spontaneous mutations blocking expression of pcaH or -G, genes encoding the α and β subunits of protocatechuate 3,4-dioxygenase. The crystal structure of the Acinetobacter oxygenase has been determined, and this knowledge affords us the opportunity to understand how mutations alter function in the enzyme. An earlier investigation had shown that a large fraction of spontaneous mutations inactivating Acinetobacter protocatechuate oxygenase are either insertions or large deletions. Therefore, the prior procedure of mutant selection was modified to isolate Acinetobacter strains in which mutations within pcaH or -G cause a heat-sensitive phenotype. These mutations affected residues distributed throughout the linear amino acid sequences of PcaH and PcaG and impaired the dioxygenase to various degrees. Four of 16 mutants had insertions or deletions in the enzyme ranging in size from 1 to 10 amino acid residues, highlighting areas of the protein where large structural changes can be tolerated. To further understand how protein structure influences function, we isolated strains in which the phenotypes of three different deletion mutations in pcaH or -G were suppressed either by a spontaneous mutation or by a PCR-generated random mutation introduced into the Acinetobacter chromosome by natural transformation. The latter procedure was also used to identify a single amino acid substitution in PcaG that conferred activity towards catechol sufficient for growth with benzoate in a strain in which catechol 1,2-dioxygenase was inactivated.

Oxygenases are enzymes that split a molecule of oxygen and introduce one or both of the oxygen atoms into their substrates (18). The enzymes have long been the topic of study because of their ability both to modify a wide variety of stable substrates and to control highly reactive and potentially dangerous oxygen species. These studies have demonstrated the important roles of oxygenases in diverse organisms. In humans, for instance, the aromatic compound aspirin inhibits an oxygenase involved in the synthesis of steroids during inflammation (30). In plants, an enzyme that appears to regulate programmed cell death has significant sequence identity to bacterial oxygenases involved in aromatic catabolism (15). The enzyme protocatechuate 3,4-dioxygenase is a member of a family of bacterial oxygenases (18, 29) that use iron as a cofactor to cleave the aromatic rings of their substrates between two adjacent hydroxyl groups, a key step in mineralization of plant products (5). The crystal structure of protocatechuate 3,4-dioxygenase from a strain of Pseudomonas putida (originally classified as Pseudomonas aeruginosa) has been determined (36, 37), and more recently, the structure of the enzyme from Acinetobacter sp. strain ADP1 was solved (56, 57).

The product of protocatechuate oxygenase is carboxymuconate, and Acinetobacter strains blocked in its metabolism fail to grow with any substrate when they are exposed to either protocatechuate or its metabolic precursors. Thus, it is possible to select for strains in which spontaneous secondary mutations block earlier steps in either transport (6) or catabolism of aromatic compounds (Fig. 1). Strains derived from such selections contained spontaneous mutations affecting residues important for structure or function in various oxygenases: p-hydroxybenzoate hydroxylase, encoded by pobA (8, 21), protocatechuate 3,4-dioxygenase, encoded by pcaH and pcaG (13), and vanillate demethylase, encoded by vanA and vanB (52). For genetic analysis of two regulatory proteins, PobR (governing expression of pobA) and PcaU (governing expression of pcaHG), PCR-generated mutations were introduced into the Acinetobacter chromosome by natural transformation, producing both loss-of-function and gain-of-function mutations (26, 27).

FIG. 1.

Positive selection of strains blocked in protocatechuate catabolism. Strain ADP500 with the engineered ΔpcaBDK1 mutation cannot grow with succinate in the presence of protocatechuate because of the toxic accumulation of β-carboxy-cis,cis-muconate. ADP500 derivatives with spontaneous mutations in pcaH and -G, genes encoding the two subunits of protocatechuate 3,4-dioxygenase, can therefore be selected based on their resistance to protocatechuate. Analogous to the conversion of protocatechuate to carboxymuconate by protocatechuate 3,4-dioxygenase (PcaHG), catechol 1,2-dioxygenase (CatA) converts catechol to muconate in the benzoate branch of the β-ketoadipate pathway.

The present investigation is a genetic analysis of Acinetobacter protocatechuate 3,4-dioxygenase complementing the recently determined crystal structure of the enzyme (56, 57). A refinement in the positive-selection protocol enabled rapid identification of strains with a heat-sensitive mutation in pcaH or -G, and subsequently, strains with spontaneous or PCR-generated second-site suppressors of some of these mutations were selected. Transformation-facilitated PCR mutagenesis was also used to select one strain in which a single amino acid substitution in PcaG allowed protocatechuate 3,4-dioxygenase to functionally replace catechol 1,2-dioxygenase (Fig. 1).

MATERIALS AND METHODS

Strains and culture conditions.

Acinetobacter sp. strain ADP1, originally designated Acinetobacter calcoaceticus BD413 (24), was routinely grown with 10 mM succinate in a mineral medium (13). Unless otherwise indicated, cells were grown at 37°C and the mineral medium was supplemented with 5 mM p-hydroxybenzoate, 5 mM quinate, 3 mM protocatechuate, or 2.5 mM benzoate. Because of the instability of protocatechuate, we used only fresh plates with this carbon source, made with stock solutions (pH 7.0) that had been stored frozen until use. Escherichia coli strains (50), namely, DH5α (purchased as competent cells from Gibco BRL) and TG2 (gift of M. Biggin and T. Williams), were grown with Luria-Bertani medium. Ampicillin was added at 100 μg/ml to select for E. coli cells with plasmids derived from pUC18 or pUC19 (59), and tetracycline was added at 12.5 μg/ml to select for E. coli or Acinetobacter cells with plasmids derived from pRK415 (25).

DNA manipulations.

Crude cell lysates of Acinetobacter strains for use in transformation reactions were prepared by incubation of pelleted cells (from a 5-ml culture) in 0.5 ml of saline-citrate buffer with 0.05% sodium dodecyl sulfate at 60°C for 1 h (24). Plasmids were isolated from 5-ml E. coli cultures with a Wizard miniprep kit (Promega) and from 5-ml Acinetobacter cultures by the alkaline lysis miniprep procedure (50) with two extraction steps, the first with phenol-chloroform and the second with chloroform. To isolate template DNA for PCR, pelleted Acinetobacter cells from a 5-ml culture were washed and treated with InstaGene Matrix as recommended by the supplier (Bio-Rad); 5 μl of the resulting solution was then used in 50-μl PCR mixtures including 0.5 U of Taq polymerase (Boehringer Mannheim), 10 pmol of each primer, and 10 nmol of each deoxynucleoside triphosphate. Standard PCR conditions were used: 30 cycles of denaturing at 94°C for 45 s, annealing at 56°C for 45 s, and elongation at 72°C for 1 min 30 s. PCR primers were synthesized by the Keck Biotechnology Resource Lab (Yale University).

Selection and characterization of strains with a spontaneous heat-sensitive mutation in pcaH or -G.

Strain ADP500 (13, 20) contains the engineered ΔcatD101::Km^r and ΔpcaBDK1 mutations. Single colonies of succinate-grown ADP500 were transferred to patches on freshly prepared plates with 10 mM succinate and 3 mM protocatechuate. Cells with spontaneous secondary mutations blocking protocatechuate catabolism do not accumulate toxic levels of β-carboxy cis,cis-muconate and are able to grow (Fig. 1). To prevent analysis of siblings, only one mutant derivative was picked per single colony of ADP500.

After incubation for 2 days at 37°C, colonies of ADP500 derivatives with secondary mutations were transferred to each of three plates: one plate contained the original selective medium and two plates contained 10 mM succinate and 5 mM p-hydroxybenzoate. The latter two plates were incubated for 2 days, one at 37°C and one at 22°C. Strains with a conditional resistance to p-hydroxybenzoate (with significantly better growth at 37°C than at 22°C) were picked from the plate with protocatechuate and purified by two further rounds of growth on plates in the presence of protocatechuate. The ΔpcaBDK1 deletion was restored to wild type by transformation with plasmid pZR3 (13, 20) under nonselective conditions, followed by selection for growth with benzoate. Expression of PcaD in recombinants is sufficient for growth with benzoate and complements the engineered ΔcatD101::Km^r mutation inactivating the isofunctional enzyme in the benzoate branch of the β-ketoadipate pathway (13, 20).

Each of the resulting strains (ADP1102 to -1117) contained a spontaneous mutation blocking protocatechuate catabolism. Single colonies of succinate-grown mutant cells were then tested for growth, at 37 and 22°C, on plates with p-hydroxybenzoate as the sole carbon and energy source. The heat-sensitive mutation in either pcaH or -G was localized by marker rescue as described previously (13) except that quinate was used as the selective growth substrate (Fig. 1). DNA transformations for marker rescue or linkage experiments were performed by growing recipient Acinetobacter strains with 10 mM succinate overnight in 5-ml cultures. After addition of 10 μl of 1 M succinate, these cultures were then incubated (to induce competence) at 37°C for 30 min in a gyratory shaker before being spread onto selective plates onto which donor DNA was then spotted.

Recovery of Acinetobacter chromosomal DNA by gap repair.

DNA fragments containing mutant pcaH and -G genes were recovered from the chromosome by gap repair with plasmid pZR1007 as previously described (2, 16). To reduce the chance that the recovered DNA would reintegrate into the chromosome, the duration of steps involving Acinetobacter cells was minimized. The 2.4-kb HindIII restriction fragment containing pcaK′CHGquiB′ (13, 19) from each pZR1007 derivative was purified by preparative gel electrophoresis and ligated with HindIII-digested pUC18, and the resulting plasmids were amplified in E. coli DH5α. Cells with plasmids in which the insert was oriented so as to be expressed from the vector promoter formed distinctively small colonies on Luria-Bertani medium plates with ampicillin. Minipreps of such cells, however, gave low plasmid yields, so the plasmids were transferred to E. coli TG2 cells (in which lacI^q decreases the potentially toxic uninduced expression of the cloned genes).

Selection and characterization of a strain in which protocatechuate 3,4-dioxygenase functionally replaced catechol 1,2-dioxygenase.

To select strains in which a gain-of-function mutation in protocatechuate 3,4-dioxygenase conferred activity towards catechol, pcaH and -G were first inserted into the gene for catechol 1,2-dioxygenase (catA) so that their expression would be regulated by benzoate (4). Plasmid pIB1344 (35), which contains catA, was cut with SalI and recircularized, generating pZR7559 with a unique SphI site within catA. A DNA fragment containing pcaC′HG (in which the Ω element from pHP45Ω [44] had been inserted into the BamHI site in pcaH) was amplified with Pfu polymerase (Stratagene) with plasmid pZR200Ω (gift of M. Stein) as the template and primers HG1 and HG6 (see below). The PCR DNA and SphI-digested pZR7559 were blunted with the DNA polymerase Klenow fragment and ligated together, and transformants of E. coli DH5α that were resistant to streptomycin, spectinomycin, and ampicillin and contained plasmids in which pcaHG and the disrupted catA were in the same orientation were selected.

The insert from one such plasmid was introduced into the chromosome of Acinetobacter strain ISA25 (ΔcatBCIJF) by selecting for transformants in which inactivation of catA prevented the toxic accumulation of muconate generated during growth with succinate in the presence of benzoate (12, 58). PCR amplification across catA in one Sm^r Spc^r Amp^s transformant with primers catA1 and catA2 (see below) and an extension time of 1 min 30 s generated a fragment of 3.4 kb, consistent with the absence of the 2-kb Ω element within pcaH. This PCR DNA was used, as before, to generate an ISA25 transformant strain resistant to benzoate, and the catBCIJF deletion in one such strain was corrected by transformation with pPAN4 (35, 53) and selection for growth with muconate, generating ADP7559. Slow growth with muconate of ADP7559 suggested the existence of a spontaneous mutation affecting muconate transport (58), as has been observed for ISA25 derivatives in two separate studies (3, 58).

PCR mutagenesis of ADP7559 generated ADP7615. The pcaG7615 gain-of-function mutation in the latter strain was recovered from the chromosome by PCR amplification with the primers catAK (5′-GGGGGTACCCTGATTCTACATGGCACG-3′) and catAR (5′-GGGGGAATTCATCGGTAATAATACTACGGCG-3′) whose 5′ ends include KpnI and EcoRI restriction sites (underlined), respectively. After digestion with KpnI and EcoRI, this PCR fragment was ligated with KpnI- and EcoRI-digested pUC19, generating pZR7615 in which pcaH and -G are expressed from the plasmid promoter. pZR7615 was transferred from E. coli DH5α cells to TG2 cells, and the sequences of the cloned genes containing pcaG7615 were reconfirmed.

Transformation-facilitated PCR mutagenesis.

For mutagenesis (26, 27), standard PCR was performed as described above except that the number of cycles was increased to 35. To generate DNA with suppressors of a heat-sensitive pcaH or -G deletion, primers HG1 (5′-CTACATTGTTCACTTTATGCAGGC-3′) and HG6 (5′-GATATACGGCCCGTTCCATAGTC-3′) were used, amplifying a 1.5-kb PCR fragment. To mutagenize the transposed pcaH and -G genes in ADP7559, primers catA1 (5′-GGTATAGAAACGACTATCG-3′) and catA2 (5′-CAAGTGTATGTCGTAACGC-3′) were used, amplifying a 3.4-kb fragment. PCR-amplified fragments, generated with template DNA from the planned recipient strain, were used directly in transformation reaction mixtures as follows. Two hundred microliters of a 5-ml culture of the recipient strain, grown overnight with 10 mM succinate, was transferred to a fresh 5-ml culture, and after growth in a gyratory shaker for 2 h (to induce competence), 500 μl of the culture was transferred to Falcon 15-ml polypropylene tubes together with 20 μl of the PCR mixture. After overnight shaking incubation, transformation reaction mixtures were spread onto selective plates.

DNA sequence analysis.

To isolate template DNA for sequencing, PCR mixtures prepared under standard conditions were purified with 8 μl of GeneClean Glassmilk according to the recommendations of the supplier (Bio 101, Inc.) and resuspended in 25 μl of water, and 8 μl was used for ABI PRISM Dye Terminator Cycle Sequencing with AmpliTaq DNA polymerase FS (Perkin-Elmer). Cycle sequence reaction mixtures were processed as previously described (26).

RESULTS

Characterization of strains with a spontaneous heat-sensitive mutation in pcaH or -G.

By positive selection for derivatives of Acinetobacter strain ADP500 that have a spontaneous mutation blocking the β-ketoadipate pathway, 94 strains (13) that had a mutation affecting protocatechuate 3,4-dioxygenase, the enzyme which cleaves the protocatechuate aromatic ring generating carboxymuconate (Fig. 1), were characterized. Only 11 of the 94 strains, however, had a missense mutation in pcaH or -G, a gene encoding the β or α subunit of the dioxygenase, respectively. Nearly one-quarter of the mutants had a deletion extending into DNA outside of these two genes (13). Six of the missense mutations caused a heat-sensitive phenotype, blocking growth with protocatechuate at 37°C but not at 22°C (13). It therefore seemed feasible to use this phenotype as a way to identify, early in the selection protocol, those ADP500 derivatives most likely to have a spontaneous missense mutation in pcaH or -G. This identification process would allow efforts to be focused on strains which could contribute to understanding how structure influences function in the dioxygenase enzyme and eliminate time spent identifying mutants with large deletions, frameshifts, or nonsense mutations.

To identify heat-sensitive pcaH or -G mutants in this study, ADP500 derivatives still containing the ΔpcaBDK1 deletion and able to grow in the presence of protocatechuate at 37°C were tested for resistance to p-hydroxybenzoate during growth with succinate at both 37 and 22°C (Fig. 1). Screening was done with p-hydroxybenzoate because cells were found to grow poorly upon successive transfers onto plates containing protocatechuate, a compound that has inherent toxicity (41). Consistently, approximately 5% of the spontaneous mutants analyzed were resistant to p-hydroxybenzoate at 37°C but not at 22°C. After correction of the ΔpcaBDK1 deletion in 17 such strains, the expected heat-sensitive phenotype was revealed: growth with p-hydroxybenzoate was completely blocked at 37°C but allowed at 22°C. The spontaneous mutation in each of these 17 strains, with one exception, was mapped by marker rescue to pcaH and pcaG and sequenced. The one unique strain had a point mutation in the gene for PcaU, the transcriptional activator governing pca operon expression, and its characterization is part of a separate study (51).

The pcaH and -G mutations and their associated amino acid changes in the heat-sensitive mutants isolated in this study or previously (13) are listed in Table 1. Figure 2 shows the positions of the mutations in the primary sequence of Acinetobacter PcaH and -G. The pcaG13 and pcaH1112 point mutations were each found in two independently isolated strains, and the pcaG13 and pcaG1114 mutations changed the same nucleotide but caused different amino acid substitutions (Table 1). Most of the remaining mutations affected residues distributed throughout the linear amino acid sequence of PcaH and -G (Fig. 2). Although all the mutants grew with p-hydroxybenzoate at 22°C but not at 37°C, protocatechuate 3,4-dioxygenase appeared to be impaired to various degrees: the most severely affected mutants (ADP6116 and ADP1116) grew with p-hydroxybenzoate extremely slowly even at 22°C.

TABLE 1.

Heat-sensitive mutations in pcaH and -G

Straina	Genotype	Mutation
		Nucleotide(s)b	Amino acid(s)c
ADP6116	pcaG1	C47T	T12I
ADP1102	pcaG1102	Δ(C237–C266)d	Δ(N76–T85)
ADP1117	pcaG1117	C391A	Q122K
ADP2507	pcaG11	C395T	A123V
ADP1107	pcaG1107	C449T	T141I
ADP1109	pcaG13	C451T	R142C
ADP6208	pcaG13	C451T	R142C
ADP1114	pcaG1114	C451A	R142S
ADP6510	pcaG19	T460G	F145V
ADP1103	pcaG1103	2×(A481–A486)d	2×(N152–A153)
ADP1110	pcaG1110	T536G	L170R
ADP1104	pcaH1104	G134T	S344I
ADP1106	pcaH1106	C143T	T347I
ADP1105	pcaH1105	C221T	P373L
ADP2509	pcaH11	G227C	G375A
ADP1108	pcaH1108	G286C	A395P
ADP1113	pcaH1113	G302C	W400S
ADP1111	pcaH1111	2×(T370–C372)d	2×(F423)
ADP1116	pcaH1116	Δ(T412–A414)d	Δ(Y437)
ADP1112	pcaH1112	C476T	P458L
ADP1115	pcaH1112	C476T	P458L
ADP6222	pcaH17	C503T	A467V

^aStrains with numbers lower than 1200 were isolated in this investigation; the remainder were described previously (13). 

^bNucleotides are numbered starting from the A residue in the ATG start codon of each gene. 

^cTo facilitate comparison to protocatechuate 3,4-dioxygenase in P. putida (37), PcaG amino acids are numbered 1 to 200 while PcaH amino acids are numbered 300 to 540 (Fig. 1). 

^dInsertions are assumed to be tandem duplications (2×), and deletions (Δ) are assumed to remove the second of two direct DNA repeats. 

FIG. 2.

Spontaneous and PCR-generated mutations in protocatechuate 3,4-dioxygenase from Acinetobacter sp. strain ADP1. Protocatechuate 3,4-dioxygenase is an oligomer (29) of heterodimers composed of PcaG (the α subunit) and PcaH (the β subunit). The two subunits have significant amino acid identity, and it has been suggested that they are derived from duplication and divergence of the gene for the homodimer subunit of the ancestral enzyme (36, 37). Acinetobacter protocatechuate 3,4-dioxygenase subunits (GenBank accession no. L05770) are presented so that homologous amino acid sequences occupy corresponding positions in PcaG (top) and PcaH (bottom). In order to achieve this alignment, gaps indicated by dashes were introduced into the sequences. With the exception of pcaG7615, G425A, nucleotide sequence changes causing the amino acid alterations shown here are presented in Tables 1 and 2. Alleles with designations smaller than 1000 were identified in a previous investigation (13). To facilitate comparison with structural data from previous studies (37, 56), the same numbering system has been used for the primary structure with PcaG amino acids numbered 1 to 200 and PcaH amino acids numbered 300 to 540 (the five Acinetobacter PcaG residues that align with a gap between residue 88 and 89 in P. putida PcaG are designated 88a to -e). Amino acid substitutions are indicated by arrows pointing to the mutant amino acid. Duplicated residues are underlined, and the mutant repetitions are shown above the primary sequence. Shaded boxes superimposed on the primary sequence indicate residues missing in deletion mutations. Amino acids in shaded boxes separated by arrows from the primary sequence indicate suppressor mutations, and the borders of the boxes (solid, dotted, or not bordered) match those of the shaded boxes indicating the mutations that are suppressed. The double-dotted border indicates a mutation that confers the ability to grow at 37°C to a strain containing both ΔpcaG1102 and pcaG7554. The R133H substitution caused by pcaG7615 and conferring activity towards catechol is indicated by a box enclosing the structure of the compound. The glycyl residue at position 60 in PcaG appears to be a natural variant and is found in all the mutants described in this study (including ADP7615), whereas the wild-type strain ADP1 has serine (indicated in parentheses in the figure) at the corresponding position.

The majority of strains had a missense mutation, but unexpectedly, protocatechuate 3,4-dioxygenase was more severely altered in four newly isolated strains. Deletions in pcaH1116 and pcaG1102 caused deletions of 1 and 10 amino acid residues, respectively, and mutations in pcaH1111 and pcaG1103 caused insertions of 1 and 2 amino acids, respectively (Table 1; Fig. 2). Both deletions may have arisen by recombination between directly repeated DNA sequences, resulting in loss of one of the repeats, TAT in the ΔpcaH1116 strain and GATAC in the ΔpcaG1102 strain. The 3-bp pcaH1116 deletion appears to lie within a local region of DNA instability, given its proximity to the previously described overlapping null mutations pcaH21 and pcaH13, duplications of 3 and 7 bp, respectively, (13). The cause at the DNA level of the duplication of TTC in pcaH1111 and AAATGC in pcaG1103 is likely to have been mediated by slippage at the tandem DNA repeats immediately flanking each of these mutations, GGTGGT and GAAGCAGAAGCA, respectively. As expected, phenotypic revertants were readily obtained from strains carrying either of these duplications.

Suppression by intragenic and extragenic point mutations.

Strains in which a pcaH or -G deletion did not cause a null phenotype were particularly attractive candidates to use to identify second-site suppressor mutations since the encoded enzyme was still partly functional and a background of direct reversion would not be a concern. Suppression was first tried with a strain from a previous study, ADP6338, containing the ΔpcaH7 leaky 4-amino-acid deletion which causes slow growth with p-hydroxybenzoate at both 37 and 22°C (13) (Fig. 2). Liquid cultures of ADP6338 grown with 10 mM succinate and 5 mM p-hydroxybenzoate were serially transferred, and after the third transfer, one culture failed to develop the purple color indicative of protocatechuate accumulation due to deletion of pcaH7. Partial suppression of the ΔpcaH7 leaky phenotype in a purified isolate from this culture was confirmed by comparing the growth on a plate with p-hydroxybenzoate of the derived strain, designated ADP7552, with that of ADP6338.

In addition, occasionally a liquid culture shaking at 37°C would not grow when it was provided with 5 mM p-hydroxybenzoate as the carbon source and inoculated with a single colony of succinate-grown ADP6338. In one case, the culture suddenly became turbid after 18 days of incubation. A purified isolate, designated ADP7553, grew slightly faster than the parental strain (but not as well as ADP7552) on a plate with p-hydroxybenzoate at 37°C. The cause of the varied behavior of ADP6338 is unknown but may reflect different physiological states conferring upon cells a range of sensitivities to the sudden intracellular accumulation of protocatechuate generated during growth with p-hydroxybenzoate.

Evidence that a mutation genetically linked to the pcaH7 deletion was the cause of suppression in both ADP6338 derivatives came from results of a DNA transformation experiment using a recipient strain with the 440-bp pcaCH1 deletion (13) encompassing the pcaH7 deletion locus: recombinants that grew at 37°C with p-hydroxybenzoate like ADP7552 or ADP7553 were readily obtained with donor DNA from the corresponding strain either in crude cell lysate or as a 2.4-kb HindIII restriction fragment containing pcaCHG recovered from the chromosome by gap repair (2, 16). Confirming this linkage, sequencing of pcaH and -G in each strain revealed that protocatechuate 3,4-dioxygenase contained, in addition to the deletion of PcaH residues 319 to 322 due to the pcaH7 deletion, a G13S change in PcaG in ADP7552 and a P317L change in PcaH in ADP7553 (Fig. 2; Table 2).

TABLE 2.

Suppressor mutations in pcaH and -G

Strain	Primary mutation		Suppressor mutation			Growth with p-hydroxybenzoatea at:
	Designation of deletion	Amino acid(s) deleted	Designation(s)	Nucleotide change(s)	Amino acidb change(s)	22°C	30°C	37°C
ADP6338	pcaH7	A319–P322				Slow	Slow	Slow
ADP7552	pcaH7	A319–P322	pcaG7552	G49A	G13S	+	+	+
ADP7553	pcaH7	A319–P322	pcaH7553	C53T	P317L	+	+	+
ADP1102	pcaG1102	N76–T85				+	−	−
ADP7554	pcaG1102	N76–T85	pcaG7554	A279C	Q(88a)H	+	+	−
ADP7617	pcaG1102	N76–T85	pcaG7617	A279T	Q(88a)H	+	+	−
ADP7555	pcaG1102	N76–T85	pcaG7555	G117T	L35F	+	+	−
ADP7556	pcaG1102	N76–T85	pcaG7554, pcaG7556	A279C, G232A	Q(88a)H, D74N	+	+	+
ADP1116	pcaH1116	Y437				+	−	−
ADP7557	pcaH1116	Y437	pcaH7557	T496G	L465V	+	+	+

^a+ indicates growth at a rate indistinguishable from that of the wild type, and − indicates the absence of growth. 

^bThe five PcaG residues designated 88a to -e in Acinetobacter strain ADP1 align with a gap between residues 88 and 89 in P. putida PcaG. 

During this investigation but as part of a separate study, PCR mutagenesis was combined with natural transformation to obtain loss-of-function as well as gain-of-function mutations in the PobR regulatory protein (26, 27). To show in that study that the technique was broadly applicable for genetic analysis of chromosomal genes in Acinetobacter, we identified PCR-generated mutations that suppressed the heat-sensitive pcaG1102 deletion, causing loss of PcaG amino acid residues 76 to 85 (26) (Fig. 2). Each suppressor mutation caused an amino acid substitution near the deletion in the PcaG linear amino acid sequence and restored the ability to grow with p-hydroxybenzoate at 30°C but not at 37°C (26) (Fig. 2; Table 2). However, during a later experiment using cells with the pcaG7554 suppressor, a single colony that could grow with p-hydroxybenzoate at 37°C appeared, suggesting that further spontaneous suppression of the heat-sensitive phenotype had been selected. Subsequent sequencing of pcaH and -G in this strain revealed an additional point mutation causing a D74N substitution, again near the original mutation in the PcaG primary sequence (Fig. 2; Table 2).

Of the heat-sensitive pcaH or -G mutations in Table 1, deletion of pcaH1116 caused one of the strongest phenotypes: strains with this mutation grew with p-hydroxybenzoate extremely slowly even at 22°C. Nevertheless, PCR mutagenesis also successfully suppressed deletion of pcaH1116, generating a strain in which an L465V substitution in PcaH in combination with the deletion of Y437 caused by the pcaH1116 deletion restored growth with p-hydroxybenzoate even at 37°C (Fig. 2; Table 2). PCR amplification across both mutations in the suppressed strain created DNA that gave rise to over 1,000 times the number of transformants in the initial selection, consistent with the sequenced second-site mutation being the cause of suppression.

An alternate form of suppression.

PCR-generated suppressor deletions of pcaG1102 were found in colonies that appeared on a plate with p-hydroxybenzoate after overnight incubation at 30°C. At this time there was no growth on control plates of ADP1102 cells that had not been transformed with PCR DNA. However, after several days’ further incubation, the control plates usually contained several colonies with exceptional properties. These strains maintained the ability to grow at 30°C with p-hydroxybenzoate even after the selective pressure was relieved during successive rounds of growth with succinate, suggesting that the strains had acquired a heritable suppressor mutation. Yet, DNA sequencing of pcaH and -G in two such strains revealed only the original ΔpcaG1102 mutation and DNA in crude cell lysates of these strains could not transform the new phenotype into the parental strain.

It has been noted that during suppression analysis of bacteria, often the most readily isolated cells are those in which amplification on the chromosome of the partly dysfunctional gene generates sufficient mutant protein activity to allow growth under the selective conditions (49). Such a phenomenon may account for the properties of the strains in which the pcaG1102 deletion appeared to have been spontaneously suppressed. DNA amplification has also been shown to be an adaptive response of wild-type bacterial cells (33, 46–49). In Acinetobacter, the ability to coamplify (49) all the genes necessary for catabolism of protocatechuate, quinate, and p-hydroxybenzoate to citric acid cycle intermediates is a selective benefit that may have acted continuously over evolutionary time to favor the current supraoperonic clustering (2, 14, 22) of these genes.

A single amino acid substitution in PcaG (R133H) confers catechol 1,2-dioxygenase activity.

Isolation of strains with a PCR-generated suppressor mutation as described above suggested that the technique might be a powerful method of isolating gain-of-function pcaH or -G mutations. In the benzoate branch of the β-ketoadipate pathway, catechol 1,2-dioxygenase occupies a position equivalent to that of protocatechuate 3,4-dioxygenase, and the respective substrates of these two enzymes differ by only a carboxyl group (Fig. 1). Comparison of the amino acid sequences of the two oxygenases indicates common ancestry (19, 34). An intriguing question, therefore, is to what extent the two enzymes have specialized to perform their respective tasks since their divergence from a common ancestor (55).

To address this problem, we constructed strain ADP7559, in which a copy of pcaH and -G was inserted near the middle of catA (34), the gene encoding the subunits of the catechol 1,2-dioxygenase homodimer (42). This insertion prevented growth with benzoate but presumably allowed benzoate to induce expression of the transposed genes (4). PCR amplification across pcaH and -G in this strain generated DNA which gave rise to transformants of ADP7559 that could grow with benzoate, although the brown staining produced during growth of the transformants indicated some accumulation of catechol. No background of spontaneous mutants was detected. Sequencing of the transposed pcaH and -G in 13 transformants, each independently generated by different PCRs, revealed in every case the same G425A point mutation (pcaG7615) causing an R133H substitution in PcaG (Fig. 2). The pcaH and -G genes from within catA in one of the transformants (ADP7615) were cloned by PCR into plasmid pZR7615, and after confirmation of the DNA sequence, transformation with this DNA was shown to confer the ability to grow with benzoate upon the parental strain.

DISCUSSION

Genetic analysis of protocatechuate 3,4-dioxygenase structure and function.

In a previous investigation using a parental strain with the pcaBDK1 deletion, derivatives in which a conditional mutation in pcaH or -G provided resistance to the toxic intracellular accumulation of carboxymuconate produced from protocatechuate in the medium were isolated (13) (Fig. 1). These mutations were identified after pcaBDK DNA was restored to wild type. In contrast, this study describes a protocol for the identification of strains with a conditional pcaH or -G mutation before any genetic manipulation, thereby facilitating the generation of a large collection of such strains useful for an analysis of the effects on protocatechuate 3,4-dioxygenase function of relatively subtle alterations in the structure of the protein.

The pcaH and -G mutations isolated in this study and previously (13) have a range of heat sensitivity phenotypes. Unexpectedly, one-quarter of the mutants produced in this study had small deletions or insertions in pcaH or -G. The largest deletion (ΔpcaG1102) removed 10 amino acids from a surface loop in the α subunit that contacts the β subunit of the enzyme (37, 57). The sequences of this loop are highly varied among the organisms for which the sequences are known, with the Acinetobacter strain ADP1 and Rhodococcus opacus 1CP enzymes having five additional residues and the Burkholderia cepacia and Pseudomonas marginata enzymes having two additional residues relative to the number of residues in the corresponding loop in the P. putida enzyme (10, 11, 19, 43, 60). This finding illustrates how the positive-selection protocol used here reveals not only general structural determinants of protein stability (1, 31, 32, 40) but also regions of the protein where large alterations can be tolerated.

Thermodynamic experiments have shown that the free energy of folding of proteins is relatively small, approximately 5 to 15 kcal/mol (45). This small value favoring protein folding is due to a delicate balance between two large numbers, the free energy of the noncovalent interactions in the folded state and that in the unfolded state. Interactions between hydrophobic residues are the principal thermodynamic force in the stabilization of the folded state, while solvation energies are the principal destabilizing force. Computational analyses of models of protein folding (7) have shown that mutations can lead to large changes in the denaturation temperature of a protein with little change to the overall structure of the protein. Thus, the protein can be active at reduced temperatures but inactive at higher temperatures.

In a method similar to the plating selection used to obtain mutants of pcaH and -G, bacteriophage T4 lysozyme mutants in which the enzyme was phenotypically inactive at 42°C were obtained (17, 23). When purified, these enzymes were found to have reductions in thermal stability that could be rationalized upon examination of the structure. The recently determined crystal structure of protocatechuate 3,4-dioxygenase from Acinetobacter sp. strain ADP1 (57) similarly facilitates analysis of the mutations in this study for their possible effects. Figure 3 shows the positions of all temperature-sensitive, deletion, and suppressor mutations superimposed on a drawing of the Cα trace of the Acinetobacter enzyme. Table 3 describes the structural consequences of the mutations. Typically, the mutations create van der Waals clashes, cavities in hydrophobic regions, or isolated buried charges destabilizing the native structure of an individual subunit, of an intersubunit interface, or of the interface between molecules in the dodecameric aggregate. For three of the mutations, residues in the active site are involved so that the reduction in activity may not arise from a lack of stability. T12I abolishes hydrogen bonds which stabilize the main chain around Pro15. Since the side chain of Pro15 forms part of the narrow waist of the substrate binding site, changes in its position may allow protocatechuate to bind to the iron in less productive orientations. W400S eliminates a residue that stacks against His462, an iron ligand. The substitution also removes the Nɛ that may form a hydrogen bond with molecular oxygen during the reaction cycle. P458L may move the end of Arg457, which has been proposed to stabilize the development of a negative charge on C-4 of protocatechuate, allowing an electrophilic attack by molecular oxygen (38).

FIG. 3.

Stereoview Cα trace of protocatechuate 3,4-dioxygenase from Acinetobacter sp. strain ADP1 showing the positions of spontaneous and PCR-generated mutations. Each protomer of protocatechuate 3,4-dioxygenase is composed of an α subunit (dashed line), a β subunit (solid line) and one nonheme iron (black sphere). Each mutation is labeled with a sphere and a letter which corresponds to a letter in Table 3. Point mutations and insertions are in white spheres. These mutations are scattered throughout the structure but usually lead to disruption of interfaces between secondary or tertiary structural elements. The deletion mutations X, Y, and Z are depicted in green, blue, and red spheres, respectively. Rescue mutations are shown in lighter shades of the corresponding color.

TABLE 3.

Mutations of Acinetobacter protocatechuate 3,4-dioxygenase and structural consequences

Designation	Amino acid change(s)a	Description of change(s)
A	T12I	Residue in the α subunit along one wall of the active site. Thr12 Oγ1 forms hydrogen bonds with A132O, I135O and G14N. These residues help orient P15 to allow its side chain to interact with the aromatic ring of the substrate.
B	Q122K	Buried side chain forms hydrogen bonds with side chains of R142 and T342. Mutation places two buried positively charged residues in the αβ interface.
C	A123V	Side chain buried under the short helix. Mutation creates van der Waals clashes, which destabilizes the structure of the loop.
D	T141I	Side chain in a group of hydrophobic residues (V143, L170, F185, I187, F197, and F198) in the α subunit. Mutation abolishes hydrogen bonds made by T141 Oγ1 with V196O and R142N as well as creating van der Waals clashes with H140, R142, and/or F197, destabilizing the α subunit.
E	R142C, R142S	R142 makes a buried salt link with D345 and hydrogen bonds with Q122. Mutations bury an unpaired charged side chain (D345) at the αβ interface, destabilizing it.
F	F145V	Side chain buried in a group of hydrophobic residues (L51, P111, I126, P124, V173, A172, and Y183). Cavity produced from mutation destabilizes the α subunit.
G	2×(N152–A153)	Insertion of 2 residues at the N terminus of the short helix near the β subunit may disrupt the helix, destabilizing the α subunit, or may extend the helix, destabilizing the αβ interface.
H	L170R	Side chain buried in a group of hydrophobic residues (V157, L 158, I161, V196, F198, I337, and I339) in the αβ interface. Mutation buries an unpaired charged side chain, destabilizing the interface.
I	S344I	Side chain in a group of hydrophilic residues (Q122, R142, D155, T342, and E345) in the αβ interface. S344 Oγ1 makes hydrogen bonds with Y79 OH and H125 Nɛ2. Mutation abolishes these hydrogen bonds while creating van der Waals clashes.
J	T347I	T347 Oγ1 makes hydrogen bonds with W71 Nɛ1 and L343O in the αβ interface. Mutation abolishes these hydrogen bonds while creating van der Waals clashes.
K	P373L	Side chain in a group of hydrophobic residues (I364, L365, L372, F423, and I442). Replacement with larger side chain creates several van der Waals clashes, destabilizing the β subunit near the local twofold axis.
L	G375	Main chain (φ and Ψ) angles are not allowed for residues other than glycine. Mutation causes refolding of this segment near local twofold axes, destabilizing the β subunit and dodecameric aggregate.
M	A395P	Side chain in a group of hydrophobic residues (V384, V392, V397, I525, F463, and L465) in the β subunit. Mutation of this residue at the N terminus of a β strand produces van der Waals clashes, with the main chain of V392 and K393 destabilizing the β subunit.
N	W400S	Side chain is in a largely hydrophobic environment (V17, L21, L67, F131, and I475) and stacks against the side chain of H462 (iron ligand). W400 Nɛ1 makes a hydrogen bond with a conserved solvent molecule in the postulated oxygen interaction pocket.
O	2×(F423)	F423 is in the loop joining two β strands. Insertion of an additional residue forces reorganization of the loop, destabilizing the β subunit.
P	P458L	Mutation is in a segment near the iron ligands H460 and H462 and adjacent to R457, which is proposed to stabilize the development of a negative charge on the substrate. Mutation creates van der Waals clashes with the main chain of D483-I486, destabilizing the β subunit.
Q	A467V	Mutation creates van der Waals clashes with A365, E468, and/or Q472, destabilizing the β subunit.
X	Δ(N76–T85)	Deletion of residues in a surface loop between β strands. Several contacts with residues (T347–F352) in an adjacent loop. The Y79 side chain is buried in an αβ interface near the site of the S344I and Q122K mutations.
X1	Δ(N76–T85) + L35F	20 Å away from the deletion. Mutation fills a cavity in a group of hydrophobic residues (L90, W92, F351, L396, and F356), stabilizing the αβ interface.
X2	Δ(N76–T85) + N(88a)H	In the same loop as the deletion. Mutation allows more stable loop conformation.
X3	Δ(N76–T85) + N(88a)H + D74N	Mutation is right before deletion. Mutation allows more stable loop conformation.
Y	Δ(A319–P322)	Deletion of residues in a loop that is part of the interaction with other proteins in a dodecameric aggregate. Adjacent to Y324, which makes hydrogen bonds with the carboxyl group of the substrate. Deletion might alter substrate binding.
Y1	Δ(A319–P322) + G13S	Serine side chain makes a hydrogen bond with A132O. Mutation may allow R133 to reposition to form an interaction with the carboxyl group of the substrate.
Y2	Δ(A319–P322) + P317L	On same loop as the deletion. Mutation reduces conformational rigidity of the imino ring, allowing a more stable structure.
Z	Δ(Y437)	At the beginning of a β strand in the β subunit. Aromatic ring interacts with a group of hydrophobic residues (V384, V397, F439, F463, L465, L474, F478, F523, I525, and L527). Preceding Y436 can replace deleted Y437, while region immediately before deletion refolds.
Z1	Δ(Y437) + L465V	Mutation is on a different β strand in the β subunit but is in the group of hydrophobic residues affected by the original mutation and is inserted into the same cavity that includes Y436. Mutation allows a more stable refolding of the loop.

^a2×, tandem duplications; 88a, PcaG residue in Acinetobacter strain ADP1 that aligns with a gap between residues 88 and 89 in P. putida PcaG. 

The three pcaH or -G deletion mutations for which second-site suppressors were identified in this study (Fig. 2 and 3; Table 2) appear to affect protocatechuate 3,4-dioxygenase in distinct ways: the pcaH7 deletion causes the loss of 4 amino acid residues in a loop containing Tyr324, which forms a hydrogen bond with the substrate in the native enzyme (38, 57); the pcaH1116 deletion causes the loss of one residue at the amino terminus of a β strand in the flattened β barrel of the β subunit; and deletion of pcaG1102 causes the loss of 10 residues in a loop that forms a portion of the interface between the α and β subunits (37, 57). The fact that spontaneous or PCR-generated suppressor mutations were readily identified for each of these deletions, with the pcaH7 deletion being suppressed by either an intragenic or an extragenic mutation (Fig. 2; Table 2), demonstrates the usefulness of combining positive selection and natural transformation for genetic analysis of protocatechuate 3,4-dioxygenase in Acinetobacter strain ADP1. Furthermore, it was discovered, unfortunately in a separate investigation after this study, that the combination of a mutation inactivating the regulatory gene pcaU with the leaky ΔpcaH7 structural gene mutation completely blocked growth with p-hydroxybenzoate or protocatechuate although either mutation alone did not (27). This double mutant was used to identify PCR-generated gain-of-function mutations in pobR which could confer PcaU activity (27). However, the same strategy should also greatly facilitate the identification of PCR-generated mutations in pcaH or -G that suppress the pcaH7 deletion or other leaky mutations that subtly alter the enzyme, since strains with such suppressors would appear against a background of no growth. Further biochemical and structural analysis of these suppressor mutations might shed light on the long slow process by which proteins evolve over time by accumulating beneficial mutations in combinations of two or more.

It has long been noted that the two parallel branches of the β-ketoadipate pathway (Fig. 1), the p-hydroxybenzoate branch encoded by genes in the pca-qui-pob supraoperonic cluster (14, 22) and the benzoate branch encoded by the ben-cat supraoperonic cluster (14, 22), represent an attractive system for studying how the demands for a set of chemical reactions in a single cell line were met twice over evolutionary time to create each of the two branches of the pathway (55). At the two extremes of a continuum, the coenzyme A transferases encoded by pcaIJ and catIJ are over 99% identical (28) whereas the lactonizing enzymes encoded by pcaB and catB, although they act on closely related substrates, come from distinct protein families and do not have significant sequence identity (28). Between these extremes, the two dioxygenase subunits encoded by pcaH and -G have 26 to 36% amino acid identity over 127 to 213 aligned residues with the homodimer subunit encoded by catA, suggesting that the two enzymes have a common ancestor. Nevertheless, it was unexpected that it would take one and only one amino acid substitution in protocatechuate 3,4-dioxygenase to confer activity towards catechol sufficient to allow growth with benzoate in a strain in which catA was inactivated.

The ease with which directed evolution may generate a protocatechuate 3,4-dioxygenase capable of functionally replacing catechol 1,2-dioxygenase is intriguing given that there appears to be one trait which consistently distinguishes the two bacterial dioxygenases: the oligomeric complexity of the enzymes. Protocatechuate 3,4-dioxygenase forms oligomers composed of different numbers of αβ-subunit heterodimers in different bacteria (29): 12 in Acinetobacter (56) and P. putida (37) and 6 in Brevibacterium fuscum (9). Catechol 1,2-dioxygenases are not known to oligomerize beyond homo- and heterodimers (18). Oligomerization has been suggested to enhance thermal stability (37), perhaps indicating a trade-off between stability and catalytic efficiency (54). The active site in Acinetobacter protocatechuate 3,4-dioxygenase is at the αβ interface near a neighboring protomer related by the local threefold-symmetry axis (57). Near the corresponding symmetry axis in the P. putida enzyme there is a second site where the substrate and inhibitors can bind (38, 39), being connected to the active site by a solvent-filled channel. These observations fueled speculation that the oligomeric state of these intradiol dioxygenases plays an important role in function. The finding of the R133H gain-of-function mutation shows that the oligomeric state may play only a small role in substrate specificity and that this can be overcome by a single mutation.

The opening to the active site of Acinetobacter protocatechuate 3,4-dioxygenase is at the bottom of a 15-Å-deep cavity surrounded by a preponderance of basic residues. Calculation of the electrostatic potential around P. putida protocatechuate 3,4-dioxygenase shows that a large sphere of positive electrostatic potential originates around the iron and extends all the way to the outer edge of the active-site cavity (39). It has been proposed that the enzyme uses this excess positive potential to attract negatively charged substrates like protocatechuate into the active site. Arg133 is one of the positively charged residues in the midsection of the active-site cavity. It makes an intrasubunit salt link to Asp65 and an interprotomer salt link to Glu162. Arg133 and other positively charged residues near the entrance to the active-site cavity are not conserved in the catechol 1,2-dioxygenases. Modeling of the R133H mutation indicates that a histidine at this position would not be able to form a direct interaction with Glu162 of the neighboring protomer. Since both residues are exposed, the free-energy penalty of disrupting this interaction should be on the order of 1 kcal/mol or less. Choosing a common rotamer of the histidine would place the ring of the side chain within hydrogen-bonding distance of Tyr324 and Thr326. In the structure of the oxygenase with bound protocatechuate (38, 57), Tyr324 interacts with the carboxyl of the substrate. If the histidine rotated into this area it might interact directly with catechol, facilitating more productive modes of binding during catalysis with catechol.

Yet to be determined is the extent to which the R133H mutation deprives protocatechuate 3,4-dioxygenase of its capacity to support growth with protocatechuate in the wild-type genetic and physiological context of pcaH and -G. In short, did the gain-of-function mutation make the enzyme a generalist, capable of acting effectively upon both catechol and protocatechuate, or a shifted specialist, with effective activity changed from protocatechuate to catechol?