Mutations in the tacF Gene of Clinical Strains and Laboratory Transformants of Streptococcus pneumoniae: Impact on Choline Auxotrophy and Growth Rate

Bacterial strains and growth and transformation conditions.The streptococcal strains included in this study are described in Table 1. The R6 strain used here was purchased from the American Type Culture Collection (ATCC BAA-255). Streptococci were grown in Todd-Hewitt broth supplemented with 0.5% yeast extract (THY), in C medium (21) supplemented with yeast extract (0.8 mg ml⁻¹; Difco Laboratories) (C+Y medium), or in a chemically defined medium (CDM) (37) supplemented with asparagine (50 μg ml⁻¹) and sodium pyruvate (250 μg ml⁻¹). The CDM was purchased from JRH Biosciences (Lenexa, KS). Where indicated, CDM was supplemented with choline chloride (5 to 50 μg ml⁻¹) or ethanolamine (40 μg ml⁻¹). In some experiments, we observed that Cho-ind pneumococci grew faster in a modified C medium lacking choline (C−Cho medium) than in CDM. Cells were incubated at 37°C without shaking, and growth was monitored by measuring A₅₅₀.

Unless otherwise stated, S. pneumoniae was transformed to choline independence by incubating 0.5 ml of competent R6 cells in C medium supplemented with bovine serum albumin (0.8 mg ml⁻¹) (35) with 1 μg chromosomal DNA prepared from strain SK598 or P072 (see below) for 30 min at 30°C. Afterwards, 1 ml of prewarmed C+Y medium was added and the mixture was incubated at 37°C for 90 min. The culture was then centrifuged and washed with 3 ml of CDM, and the pellet was resuspended in 10 ml of the same medium. After ca. 72 h of incubation at 37°C, cells growing as chains at the bottom of the tube could be observed. A portion of these cells was diluted in fresh C−Cho medium and incubated overnight at 37°C. This procedure was repeated at least twice, and single colonies were recovered by plating appropriate dilutions in tryptic soy agar (Difco Laboratories) plates supplemented with 5% defibrinated sheep blood. The Cho-ind phenotype of the transformants was retested by using C−Cho medium.

PCR amplification, cloning, and nucleotide sequencing.Routine DNA procedures were performed essentially as described elsewhere (28). S. pneumoniae chromosomal DNA was prepared as previously described (10). The DNA extraction procedure described by Ezaki et al. (9) was used for all other streptococci. DNA fragments were purified by using a High Pure PCR product purification kit (Roche). For PCR amplification and sequencing of the tacF gene, we used the oligonucleotide primers atg (1151360), 5′-TGAATGAAAAGTATAAAATTAAATGCTCT-3′; b (1151863), 5′-CGATATTGTTGTCTATACACTTGTGATG-3′; e (1151890/c), 5′-CATCACAAGTGTATAGACAACAATATCG-3′; c (1152264), 5′-GTTTTGGACTCATGGTTTTAGGA-3′; a (1152358/c), 5′-AAAAGCGAAGAGAGAGGTCAAGATGC-3′; and stop (1152886/c), 5′-GGCATCTTCAACGGTTAGTTGTTTC-3′. The numbers indicate the position of the first nucleotide of the primer in the genomic sequence of strain R6 (17), and /c means that the sequence corresponds to the complementary strand. The ATG initiation codon of tacF is shown underlined. For PCR amplification of the tacF gene (or gene fragments), high-fidelity Pfu polymerase (Biotools) was used.

The nucleotide sequence was determined by following a PCR cycle sequencing method (BigDye terminator version 3.1 cycle sequencing kit) using an automated ABI Prism 3700 DNA sequencer (Applied Biosystems). All primers for PCR amplification and nucleotide sequencing were purchased from Sigma.

The sequence data for the S. pneumoniae strains Spain^23F-1, INV104B, INV200, and OXC141 were produced by the S. pneumoniae Sequencing Group at the Sanger Institute and can be obtained from http://www.sanger.ac.uk/Projects/S_pneumoniae.

Data analysis.Sequence comparisons were undertaken by running the BLAST program (1) using EMBL/UniProtKB databases (http://www.ncbi.nlm.nih.gov/BLAST) (last date accessed, 10 August 2007), as well as available preliminary genomic data for S. pneumoniae 670 and S. mitis^T (The J. Craig Venter Institute website at http://tigrblast.tigr.org/cmr-blast/) and for the pneumococcal strains Spain^23F-1, INV104B, INV200, and OXC141 (The Sanger Institute; http://www.Sanger.ac.uk/Projects/S_pneumoniae) and G54 (7) (http://bioinfo.cnio.es/old/data/Spneumo). Since the published genomic sequence of the tacF_G54 allele has two indeterminate nucleotides, the gene was resequenced. Multiple gene or protein sequence alignments were conducted by using the Clustal W program (33) available at the website of the EMBL-European Bioinformatics Institute (http://www.ebi.ac.uk/clustalw). A consensus prediction of transmembrane helices was obtained by comparing the results rendered by 10 different computer programs (see Table S1 in the supplemental material). DNA and protein sequences were also analyzed by using the Genetics Computer Group (GCG) software package (version 10.0) (6). Pairwise evolutionary distances (PEDs) were determined by using the Distances program.

Nucleotide sequence accession numbers.The nucleotide sequences determined in this study were deposited in the EMBL/GenBank/DDBJ databases under the accession numbers AM901296 to AM901310.

Generating and characterizing the Cho-ind pneumococcal strains.We first observed that, as reported for S. oralis (16), S. mitis strains exhibited a Cho-ind phenotype; that is, they were able to grow in CDM lacking choline or any other choline analog (Fig. 1). However, whereas the type strain of S. mitis grew slowly, forming small chains with some cells being abnormal in shape and size (Fig. 1E) as described for S. oralis, S. mitis strain SK598 formed long chains of cells when incubated in any medium tested, regardless of the presence or absence of choline or choline analogs (Fig. 1C and unpublished observations). Whether SK598 is capable of synthesizing ethanolamine-containing TAs when grown in choline-deprived medium, as it does when grown in choline-containing medium (3), has not been investigated. In contrast, the type strain of Streptococcus pseudopneumoniae (the closest relative of S. pneumoniae known to date) was unable to multiply in choline-free (or any choline analog-free) medium (data not shown). Based on these observations, we then went on to prepare chromosomal DNA from S. mitis SK598 to transform S. pneumoniae R6 and isolated one clone able to grow in choline-free medium (designated strain P023) for further study. Strain P023 showed all the characteristics previously reported for the Cho-ind isolates of S. pneumoniae characterized so far (5, 29, 39). In brief, when incubated in the absence of choline, P023 formed long chains of cells (Fig. 1B) and failed to lyse either in the stationary phase of growth or after the addition of deoxycholate. However, P023 was seen to generate deoxycholate-sensitive diplococci (or short chains) several hours after the addition of at least 5 μg ml⁻¹ choline chloride to the choline-free growth medium. The Cho-ind phenotype of strain P023 could be retransformed to the parental R6 strain (choline dependent) by another round of transformation using P023 DNA as donor material. One of the new Cho-ind R6 transformants, strain P072, was further examined.

• Open in new tab
• Download powerpoint

FIG. 1.

Phase-contrast micrographs of Cho-ind strains grown in CDM. Cultures of S. pneumoniae P023 (A, B), S. mitis SK598 (C), or S. mitis NCTC 12261^T (D, E) were grown in the presence of 50 μg ml⁻¹ choline chloride (A, C, and D) or in the absence of choline or its analogs (B, E). Cultures of R6Chi (F, G) or P501 (H, I) were grown in the presence of 40 μg ml⁻¹ ethanolamine (F, H) or in the absence of choline or its analogs (G, I). Arrows indicate cells of abnormal shape and size. Bar, 10 μm.

Nucleotide sequences of tacF alleles in choline-dependent and Cho-ind strains.As already mentioned, it has been reported that a tacF gene product containing a single Val²³⁴ to Phe mutation can confer a Cho-ind phenotype to S. pneumoniae (5). To confirm this observation, we PCR amplified the entire tacF gene of strain P072 using the oligonucleotides atg and stop and used this amplicon to transform the R6 strain. Cho-ind transformants could be isolated, confirming that tacF_P072 was indeed required for choline independence. To gain further insight into the molecular characteristic(s) differentiating the pneumococcal tacF gene from that of S. mitis, the tacF genes of strains Rx1, JY2190, P023, P072, S. pseudopneumoniae CCUG 49455^T, and S. mitis SK598 were PCR amplified with oligonucleotides atg and stop and sequenced, and the almost-complete nucleotide sequence of the gene (1,462 out of 1,488 bp, not including the oligonucleotide primers) was compared with those deposited in the EMBL database and preliminary nucleotide sequences available from other sources (Table 1). The S. pneumoniae strains R6, D39, R6x, TIGR4, SP9-BS68, and Rx1 (present study) share the same tacF allele (designated allele 1) (Table 1). The results shown in Fig. 2A and B indicate that the tacF alleles of all the S. pneumoniae strains analyzed were closely related (PEDs, that is, estimated number of substitutions per 100 bases, <0.9), whereas tacF_P072 (allele 14) was clearly divergent (PED ≅ 5.0) (not shown). As expected, the tacF allele of S. pseudopneumoniae CCUG 49455^T (choline dependent) proved to be evolutionarily more related to S. pneumoniae (PED < 2.0) than to S. mitis (PED > 4.7) (Fig. 2B). Naturally, tacF_P072 was very similar (PED = 0.07) to tacF_SK598, differing only in terms of 1 nucleotide at position 700 (codon 234; position 1) (Fig. 2A). This G-to-T transversion was also found in the tacF_P023 allele, which in turn was observed to be identical to tacF_P072, as expected (unpublished results). Thus, the product of this last allele, TacF_P072, would presumably show the substitution of Val²³⁴ by a Phe residue; quite surprisingly, this point mutation was the one described by Damjanovic et al. for tacF_R6Chi (5) (Fig. 3). The divergence between the tacF alleles of the type strain of S. mitis and strain SK598A was estimated at 3.49% (Fig. 2B), possibly reflecting the greater evolutionary divergence among the S. mitis isolates than among the S. pneumoniae strains (19). Interestingly, Cho-ind strain JY2190 (allele 4) has 2 nucleotides that differ from those of tacF allele 1 (Fig. 2A) and both mutations led to the appearance of a new amino acid residue (Pro¹⁰⁴ to Leu and Ala²¹⁴ to Thr) (Fig. 3).

• Open in new tab
• Download powerpoint

FIG. 2.

Sequence variation in the tacF gene of S. pneumoniae and S. mitis strains. (A) Multiple alignment of a 1,462-bp sequence. Only the polymorphic sites (numbered vertically) are depicted. Sites 1, 2, and 3 indicate the first, second, and third nucleotides of the codon, respectively. Black boxes indicate codons in which nucleotide changes cause amino acid substitutions. Colons represent nucleotides identical to those of allele 1. Black and open bars at the top indicate the tacF gene fragments that were PCR amplified by using oligonucleotide mixtures atg and e and a and b, respectively. The overlapping region between those DNA fragments is shadowed in gray. (B) A matrix of PEDs between aligned sequences is shown. Values represent the estimated number of substitutions per 100 bases with no distance correction.

• Open in new tab
• Download powerpoint

FIG. 3.

Sequence variation among the different TacF alleles. Only polymorphic residues are depicted (numbered vertically). Colons represent amino acid residues identical to those of allele 1. Black and open bars at the top indicate the locations of tacF fragments amplified by PCR using oligonucleotide mixtures atg and e and a and b, respectively. + and − indicate TacF alleles present in choline-dependent and Cho-ind strains, respectively. The positions of amino acid residues predicted to be located in the transmembrane loop structure are shadowed in gray.

Identifying the amino acid residues of TacF that confer the Cho-ind phenotype.To establish whether one or more of the amino acid changes observed in the Cho-ind strain P072 were responsible for the Cho-ind phenotype, the entire tacF_P072 allele (or parts of it) was PCR amplified using combinations of different oligonucleotide primers, and the resulting DNA fragments were used to transform the pneumococcal R6 strain (Table 1). First, the entire tacF_P072 allele was amplified with the oligonucleotides atg and stop and used as donor DNA. Then, the tacF genes in two Cho-ind transformants isolated in independent transformation experiments (designated strains P500 and P501, respectively) were sequenced so that the recombination boundaries between donor and recipient (tacF_R6) alleles could be determined (Fig. 2A and 4). The recombination crossover points in transformant P500 (allele 15) were located between nucleotide positions 85 to 93 and 1273 to 1294 (taking as 1 the first nucleotide of the ATG initiation codon). In P501, recombination took place at positions 1 to 69 and 701 to 740 (allele 16). Unexpectedly, two additional mutations, present neither in the donor nor in the recipient allele, were found in P500, i.e., A⁶³⁷- and A¹²³¹-to-G transitions (corresponding to the first position of codons 213 and 411, respectively; Fig. 2A). Whether these mutations were introduced during PCR amplification or might have arisen spontaneously after transformation is not known. Interestingly, the spontaneous Rx1 mutant Cho-ind strain JY2190 also contained two amino acid changes (allele 4) (Fig. 3 and 4). To gain further information, an internal fragment of tacF_P072 was PCR amplified using oligonucleotides a and b and the resulting DNA (496 bp) used to transform the R6 strain. The nucleotide sequence of the tacF gene of a Cho-ind transformant (named strain P502) was then determined, and two mutations not present in the donor tacF_P072 allele were found (two A to G transitions at positions 660 and 706) that caused a change from Asn²²⁰ to Ser and Thr²³⁶ to Ala, respectively (allele 17) (Fig. 3 and 4).

• Open in new tab
• Download powerpoint

FIG. 4.

Schematic representation of recombinant tacF genes generated by genetic transformation. The genomic coordinates of the R6 allele appear at the top. The locations of the different oligonucleotides used for PCR amplification and sequencing are also shown. Stars indicate the location of amino acid residues that are different from those of the tacF_R6 allele. Amino acid changes that appeared de novo after one round of transformation (in the one-letter code) are underlined and depicted as open stars. The shading indicates the approximate positions where recombination took place. Since the tacF alleles of strains R6 and JY2190 differ by only two nucleotides, both gene schematics have been equally shadowed.

In sharp contrast with the above results, the transformation of the R6 strain with a 623-bp fragment of tacF_P072 generated by using oligonucleotides c and stop did not give rise to Cho-ind transformants, at least after 96 h of incubation. Identical results were obtained when the PCR-amplified tacF_R6 gene was used as donor DNA. Nevertheless, when the culture transformed with tacF_R6 was incubated in C−Cho medium for up to 8 days (instead of the usual 72-h incubation), a Cho-ind pneumococcal strain (designated P550; allele 19) could be isolated. Surprisingly, two previously undescribed mutations (Phe¹⁰⁷ to Ser and Phe²⁰⁹ to Leu) were detected in TacF_P550 (Fig. 3 and 4). Additional transformation experiments were performed using R6 as the recipient strain. tacF_P501 allele 16 was PCR amplified by using oligonucleotides a and b (496 bp) to generate the P511 transformant. Thus, the PCR-generated gene fragment should contain one nonsilent mutation present in the tacF_P501 allele; that is, G⁷⁰⁰ to T (coding for Phe²³⁴) (Fig. 4). Table 1 describes the results of further experiments (carried out by using different PCR-amplified fragments of tacF from the DNA of strains P550, JY2190, or SK598 as donor material) in which we were able to isolate several new Cho-ind pneumococcal transformants, namely, strains P551, P575, and P600, respectively. Comparison of the different deduced amino acid sequences (Fig. 2A, 3, and 4) along with microscopic observations (see below) might suggest that a minimum of two amino acid changes was required to confer improved fitness on a pneumococcal Cho-ind strain when growing in choline-free medium.

Improved growth of Cho-ind strains in aminoalcohol-free medium requires at least two tacF mutations.When strain R6Chi or P501 was incubated at 37°C in CDM containing ethanolamine, the cells were always very homogeneous in shape and size (Fig. 1F and H, respectively). However, R6Chi cells grown in CDM lacking choline (or any other choline analog) were of abnormal shape and size (Fig. 1G), which contrasted with the normal morphology exhibited by the P501 strain (Fig. 1I). The growth characteristics exhibited by strain JY2190 were indistinguishable from those of P501. When R6Chi, JY2190, and P501 were incubated in CDM containing ethanolamine, all of them showed a similar growth rate (Fig. 5A). In contrast, when incubated in CDM lacking any aminoalcohol, it could be observed that P501 and JY2190 grew more rapidly and with a shorter lag period than R6Chi (Fig. 1B). This observation was fully confirmed by the results of cocultivation experiments (Fig. 5C and D). Strains JY2190 Opt, R6Chi Cm, and P501 Str were constructed by transformation with DNA prepared from the multiresistant strain M222 (Table 1), mixed in different proportions, and incubated at 37°C in CDM lacking choline (or any choline analog). The viability of the different strains was monitored by plating appropriate dilutions in tryptic soy blood agar plates supplemented with the corresponding antibiotic. In agreement with previous observations (see above) the strains harboring two tacF mutations, i.e., JY2190 Opt and P501 Str, showed similar growth rates (Fig. 5C). R6Chi Cm displayed a very long lag period such that, even when this strain was cocultivated with P501 Str in a 200:1 proportion, the latter strain was capable of prevailing after 36 h of incubation (Fig. 5D).

• Open in new tab
• Download powerpoint

FIG. 5.

Growth and viability of pneumococcal cultures. R6Chi (•), P501 (♦), or JY2190 (★) cells were grown in C+Y medium until mid-exponential phase and diluted into CDM with ethanolamine (A) or without any aminoalcohol (B, C, and D). In the experiments whose results are shown in panels A and B, about 100 CFU ml⁻¹ of each bacterial strain was initially added. (C) Coculture of P501 (⋄) and JY2190 (☆) cells, at an initial ratio of 1:1. (D) Coculture of P501 (⋄) and R6Chi (○) cells, at an initial ratio of 1:200. Cultures were monitored for growth, and at different times aliquots were removed for determination of viability.

To completely discard any unexpected influence of some (unknown) difference(s) between the genome of our R6 recipient strain and that of the R6Chi strain used by Damjanovic et al. (5), we constructed strain P700 (Table 1). To do that, the tacF_R6Chi allele (allele 13) was PCR amplified using oligonucleotides atg and stop, and this amplicon was used to transform the R6 strain. The transformed culture was diluted and plated in C+Y medium. Up to 50 clones were tested for choline independence by incubation in CDM lacking choline or choline analogs. One Cho-ind R6 transformant containing the tacF_R6Chi allele was designated strain P700 (Table 1), and its morphology and growth characteristics were studied. As reported above for strain R6Chi, P700 also showed abnormal cell morphology and a long lag period when incubated in CDM lacking any aminoalcohol (data not shown).

Molecular modeling of the TacF flippase.TacF is predicted to be an integral α-helix bundle membrane protein with 14 transmembrane helices (5). Despite the existence of many computer programs to predict the secondary structure of α-helix bundle proteins, a correct topology prediction does not mean that the predicted starts and ends of the transmembrane α-helices can be trusted; only the number of transmembrane helices and their approximate positions are reliable (8). In spite of this limitation, we undertook a comparative prediction of the structure of TacF_R6 using 10 different programs (see Table S1 in the supplemental material) and obtained a consensus structure (Fig. 6A). Figure 6B shows the proposed localization of the mutated amino acid residues in selected Cho-ind pneumococcal strains. With the exception of Phe¹⁰⁶, the other amino acid positions could be predicted with reasonable agreement (≥60%) among the different programs. It can be seen that with the exception of TacF_JY2190 (allele 4), in which both mutations reside in different α-helices, and of TacF_P501 (allele 16), which possesses two mutations apparently located outside the membrane, the mutations appear to occur at various combinations of helices 3, 7, and 9 and amino acid loops 2 and 4 outside the membrane. Moreover, it could be observed in a three-dimensional model of TacF_R6 (see Fig. S1 in the supplemental material) that the 11 amino acid residues involved in the choline independence of S. pneumoniae formed three separate clusters: (i) Ile¹⁰⁰, Pro¹⁰⁴, Phe¹⁰⁶, Phe¹⁰⁷, Phe²⁹⁶, and Ile²⁹⁸; (ii) Val³², Phe²⁰⁹, and Ala²¹⁴; and (iii) Val²³⁴ and Ile²⁴⁷. When this model was compared with the mutations found in the different TacF alleles (Fig. 6B), it emerged that with the exception of alleles 20 and 21 that contain residues belonging only to cluster 1, the amino acid changes observed involved different combinations of residues from two different clusters.

• Open in new tab
• Download powerpoint

FIG. 6.

Localization of mutated amino acid residues in the putative TacF flippase of selected Cho-ind pneumococcus strain alleles. (A) A consensus secondary structure model of TacF_R6 (495 amino acid residues) produced by comparison of the results obtained using 10 different computer programs (see Table S1 in the supplemental material) is shown. The positions of several residues are indicated to the left of the corresponding amino acid both on the cytoplasm side (inside) and outside the membrane. Transmembrane α-helices are represented as green rectangles. White, yellow, red, and blue circles correspond, respectively, to hydrophobic, polar, negatively charged, and positively charged residues. (B) Proposed localization of mutated residues in several TacF alleles conferring a Cho-ind phenotype. These are shown enlarged in panel A.