MBE Advance Access originally published online on April 9, 2007
Molecular Biology and Evolution 2007 24(7):1480-1491; doi:10.1093/molbev/msm067
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Research Articles |
Stress-Driven In Vivo Selection of a Functional Mini-Gene from a Randomized DNA Library Expressing Combinatorial Peptides in Escherichia coli
Department of Biology and Biochemistry, University of Houston
E-mail: fox{at}uh.edu.
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A plasmid-borne randomized mini-gene library expressing a population of combinatorial 20-mer peptides with no bias toward any biological function was used as an initial source of genetic variance during stress-driven evolution of Escherichia coli. The transformed bacteria were evolved under multiple rounds of selective pressure imposed by nearly lethal concentrations of NiCl2, AgNO3, or K2TeO3. At the final stage, clones conferring resistance to NiCl2 were found to carry identical functional mini-genes, which conferred significant nickel tolerance on the host cells. Expression of the mini-gene markedly improved growth parameters of the evolved clones at subinhibitory concentrations of NiCl2 while being slightly detrimental in the absence of the stress. This substantial increase in resistance, as compared with control cultures adapted in the absence of the mini-gene, is shown to be largely due to coadaptation with changes elsewhere in the E. coli genome. Clones resistant to AgNO3 and K2TeO3 harbored plasmid variants with an inactive mini-gene and with a deleted mini-gene operon, respectively. In those cases, an exploration of the mini-gene sequence space apparently was fruitless, and the developed toxicity tolerance was likely to be exclusively associated with acquired adaptive mutations. Overall, the results demonstrate a very natural outcome in which the mini-genes were expected to be either successfully integrated into the bacterial genetic network or rejected depending on their effect on host fitness. This approach is immediately useful as a laboratory model to study the dynamics of bacterial adaptive evolution at the molecular level and is especially relevant as a rapid method to study cellular response to recently acquired genetic material.
Key Words: adaptive evolution • mini-gene • in vivo selection • coadaptation • metal resistance
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Directed evolution of natural microbial populations has been shown to be an efficient approach to study molecular mechanisms of natural selection, adaptation, and speciation. The short generation time, large population size, simple life cycle, and ease of maintenance and storage make bacterial and viral systems exceedingly suitable for evolution experiments (Domingo and Holland 1997
; Elena et al. 2000; Travisano and Rainey 2000; Souza et al. 2002; Elena and Lenski 2003). Nevertheless, such experiments typically require multiple iterations of the mutation–selection cycle that implies 1) diversification of parental genetic material by spontaneous or induced mutagenesis and 2) selective amplification of successful genotypes through differential reproduction of the microorganisms under defined environmental constraints. At the end of the procedure, the acquired genetic changes can be examined and related to those phenotypic features, which differentiate the evolved cell lineages from the ancestral strain. A simple way to direct the evolution of a microbial population is to make it propagate under an appropriately applied stress. Stress-induced imbalances in cellular metabolism result in reduced fitness of the wild-type lineage. At the same time, some of the emerging mutants may exhibit a substantial tolerance of the harmful factor. During prolonged cultivation under stressful conditions, these resistant phenotypes will gradually substitute the wild type. Accordingly, the population will drift toward higher frequencies of the mutated genes associated with the resistant clones. The original genotype will eventually be replaced with a new one, which confers an improved fitness on the microbes exposed to the hostile environment.
In natural systems, spontaneous mutagenesis is one of the major sources of genetic diversity on which the selection process can operate (Arber 2003). Acquired mutations are likely to be dispersed unpredictably along the whole genome, which makes it difficult to locate and characterize them and to exploit them for specific modification of a defined chromosomal locus. To circumvent the shortcomings of natural systems, the microbial population is frequently engineered to harbor a combinatorial gene library in which significant genetic variance is associated with a short segment of coding DNA sequence to produce a selection target of conveniently small size. This approach has been especially popular with those seeking a practical outcome.
In recent years, such combinatorial libraries have been widely used for the isolation of gene products with desirable properties by various in vivo selection techniques (Cipolla 2004; Neylon 2004; Aharoni et al. 2005; Falciani et al. 2005). The most popular methods are utilized in practical applications where one seeks to select macromolecules, which are able to bind a ligand of interest in a highly specific manner (Benhar 2001; Hoess 2001; Wittrup 2001). In these cases, the selection experiments are organized in such a way that the selective value of the competing phenotypes depends solely on the efficiency of direct interaction between the ligand and the proteins expressed from the gene library. For example, in phage and bacterial surface display, 2-hybrid assays, and related techniques, a small random DNA fragment is fused into the gene of an appropriate carrier protein such that the expressed chimera can interact with the intended target through a strictly defined mechanism (Joung 2001; Adda et al. 2002; Rosander et al. 2002; Szardenings 2003; Jostock and Dubel 2005; Uchiyama et al. 2005). Thus, exploration of sequence space of the random insert proceeds under constraints imposed not so much by environmental factors as by properties of the carrier protein and the ligand to be bound.
The outlines of an alternative approach which is far more meaningful from the perspective of understanding molecular evolution was developed in the course of studies on the mechanism of Escherichia coli resistance to macrolide antibiotics (Tenson et al. 1996; Tenson and Mankin 2001; Vimberg et al. 2004). In this earlier work, plasmid-encoded mini-genes containing a promoter, a Shine–Dalgarno sequence, and a terminator were used to express 5- or 21-codon random open reading frames (ORFs) (Tenson et al. 1997; Tripathi et al. 1998). Using this system, several functional peptides similar to the natural pentapeptide resistance factor, MRMLT, were identified in resistant clones evolved in the presence of subinhibitory concentrations of various macrolide antibiotics. Although the immediate goal was to obtain peptides with specific properties, a secondary by-product was the development of a far more realistic approach. Thus, this work established the potential of random gene libraries to serve as a tool for screening a minimally constrained DNA sequence space in search of genes, which increase fitness of the host organism under stressful conditions. Such a loosely defined selection criterion imposes few restrictions on the acceptable mechanisms of gene activity. Hence, the selection can be conducted such that the response occurs in the context of the entire genome. Thus, other mutations can occur in response to the presence of the mini-gene, and therefore the system is far more realistic. Thus, in addition to finding novel functional genes, the approach might be used to better understand the evolution process itself because it provides a convenient laboratory model to simulate natural selection.
When initially introduced into a growing microbial population, a random gene library in effect imitates a mutational hot spot or laterally transferred genetic material being incorporated from an unrelated organism. Studies on this type of model system can address important questions such as how novel genetic entities come into being, how the environment influences genetic changes in an evolving organism, and how foreign genetic material can be integrated into an existing well-balanced genetic network. In the present work, we make substantial improvements to this earlier technology that makes it more suitable for use in evolutionary studies. In particular, an improved selection scheme that continually challenges the cells with nearly lethal levels of the stress encourages coadaptation of the cell and the mini-gene. In addition, mini-genes of more realistic size are used, and the peptide products synthesis cannot be readily inactivated by point mutations that produce stop codons.
In the experiments described herein, 3 genetically identical E. coli populations, each carrying a replicate of a mini-gene library, were cultivated under stress associated with subinhibitory concentrations of NiCl2, AgNO3, or K2TeO3, which are known to be potent biotoxic agents (Nies 1999). The library design features a 60-bp ORF, which is preceded by a ribosome-binding site and transcribed from an isopropyl-1-thio-ß-D-galactopyranoside (IPTG)–controlled T7 promoter. The mini-gene library encodes a 20-mer peptide, which has a random central segment and exhibits no obvious bias toward any specific function. At the beginning of the evolution experiment, the library is the only source of the variance in an otherwise genetically uniform bacterial population derived from a single E. coli clone. It remains a sole target of selection until the genetic composition of the population is further diversified by spontaneous mutagenesis. The emerging mutations may affect fitness of the competing clonal lineages and therefore interfere with the selection of advantageous mini-gene variants from the library. Beneficial mutations that arise at an early stage of culture growth are particularly hindering because they can promote the selection of a neutral or even deleterious mini-gene by means of genetic hitchhiking. However, their interference with the selection process might be diminished if the irrelevant members of the mini-gene library became extinct very quickly. Therefore, to accelerate the population takeover by the most advantageous variant of the mini-gene, the bacterial culture was propagated in the regime of serial transfer into increasingly stressful conditions to maintain the selective pressure at a maximum throughout the experiment. Using this novel strategy, we sought to determine if a meaningful genetic entity could be found in a limited pool of arbitrary sequences. Herein, we describe the outcome of these experiments and analyze the advantages and limitations of the selection protocol.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Chemicals and antibiotics were from Sigma (Saint Louis, MO) or Fluka (Buchs SG, Switzerland). Enzymes were from New England Biolabs (Beverly, MA) and Promega (Madison, WI). Synthetic deoxyoligonucleotides were purchased from Integrated DNA Technologies (Coralville, IA) and MWG Biotech (High Point, NC). Escherichia coli strain BLR(DE3) (F– ompT hsdSB (rB– mB–) gal dcm
(srl–recA)306::Tn10 (TcR) (DE3)) was obtained from Novagen (Madison, WI). Unless otherwise mentioned, cells were grown at 37 °C with constant shaking in LBMOPS medium (10 g/l bacto-tryptone [Difco], 5 g/l yeast extract [Amresco], 150 mM MOPS-NaOH, pH 7.2) supplemented with 100 µg/ml ampicillin and 50 µg/ml kanamycin. The optical density of the bacterial suspension was measured with a Varian Cary 118 spectrophotometer at 600 nm. Whenever culture medium contained a specific additive (IPTG, NiCl2, AgNO3, K2TeO3, ZnSO4, or CuSO4), the same compound was included in a blank sample during OD600 measurement. Plasmids were isolated according to standard protocols and partially sequenced with the use of a BigDye Terminator v. 3.1 Cycle Sequencing Kit (Applied BioSystems, Foster City, CA) according to the manufacturer's instructions. The following sequencing primers were used: P11, d(GGTATCTTTATAGTCCTGTC); FRWP1, d(TGGCCTTTTGCTCACATG); RVRP1, d(TGAACCATCACCCTAATC); RVRP2, d(CCATGATGGATACTTTCTCG); RVRP3, d(GTCCAGATAGCCCAGTAG); SP1, d(AGCGGATAACAATTTCACACAGGA); SP2, d(GTAAAACGACGGCCAGT). Analysis of the fluorescent products of the sequencing reaction was performed by SeqWright, Inc. (Houston, TX)
Library Construction
The pCR21-T7pt plasmid was derived from the cloning vector pCR2.1 (Invitrogen, Carlsbad, CA) by rearrangement of its multiple cloning site (MCS) sequence and introduction of the T7 late terminator, T
, downstream of the T7 RNA polymerase promoter and MCS. The plasmid was cleaved at XbaI and BamHI sites, treated with calf alkaline phosphatase, and gel purified. A 100-bp-long DNA cassette harboring an ORF with randomized sequence was synthesized in a single reaction by overlap extension polymerase chain reaction (PCR) from 3 deoxyoligonucleotides, 32-mer d (GCTCTAGAAGGAGATATACATATGTCTCACGC), 32-mer d (CGGGATCCTAGGGATGTTATTCATGAGCGGAG), and 61-mer d (CATATGTCTCACGCT (NNC)11SRCTCCGCTCATG), where N is an equimolar mixture of A, T, G, or C, S is an equimolar mixture of G or C, and R is an equimolar mixture of A or G. The PCR product was treated with XbaI and BamHI restriction enzymes, gel purified, and ligated into linearized dephosphorylated pCR21-T7pt plasmid (40 Units/µl of T4 DNA ligase, 0.04 Unit/µl of yeast inorganic pyrophosphatase in 1x NEB T4 DNA ligase buffer (New England Biolabs), 100 h at 8–12 °C). The resulting plasmid library, pCRRL1 (fig. 1), was further purified by phenol–chloroform extraction and ethanol precipitation. The size of the inserted DNA fragment was verified by PCR with a pair of flanking primers, SP1 and SP2 (see above). No cassette oligomerization or unproductive recircularization of pCR21-T7pt was observed.
|
Selection Experiment
The pCRRL1 plasmid library was electroporated into competent E. coli BLR(DE3) cells derived from a single colony. The size of the obtained clone library was estimated by plating on Luria-Bertani agar with ampicilin and kanamycin. The transformed cells were transferred into LBMOPS medium with antibiotics and grown for 4 h at 37 °C to ensure that several replicates of the library were present. Aliquots were taken from the resulting bacterial suspension in order to inoculate 31 tubes containing 3 ml of fresh medium supplemented with 1 mM IPTG. In order to ensure the expression of the designed ORF, 1 mM IPTG was present in the medium during all steps of the selection experiment. Cells were incubated at standard conditions for 2 h, and then selective pressure was applied by adding increasing concentrations of 1 of 3 toxic agents, NiCl2, AgNO3, or K2TeO3, to each set of 10 tubes. The concentration range of the toxic compounds extended from 17–25% to 250–300% of their minimal inhibitory concentration (MIC) in liquid medium. MIC values were determined in preliminary experiments with E. coli BLR(DE3) carrying pCR21-T7pt plasmid and found to be 4 mM for NiCl2, 0.06 mM for AgNO3, and 2 µM for K2TeO3. A control culture was grown in LBMOPS medium supplemented with antibiotics and IPTG, without any toxic additives. After 22 h of incubation, OD600 of all cultures was measured. From each set, one culture that exhibited the highest toxicity tolerance was chosen as an inoculant for the next step. A new selection round was performed under the same conditions as the previous one except that the concentration range of the toxic agents was shifted upward of the observed maxima of the toxicity tolerance. The cycle was repeated until no further increase in toxicity tolerance was observed. At the last step, an aliquot from the most resistant culture was spread on LB agar plate with 100 µg/ml ampicilin and 50 µg/ml kanamycin. After 14–20 h of growth at 37 °C, several colonies were randomly picked and tested individually for the IPTG dependence of the evolved toxicity tolerance in LBMOPS medium with antibiotics and appropriate toxic agent. To simplify the screening procedure, the bacterial growth was characterized by an OD600 value measured in triplicate after 24 h of cultivation.
Assessment of Toxicity Tolerance in Nickel(II)-Resistant Clones
An overnight culture of a selected strain was grown for precisely 14 h and later used to inoculate 2 sets of tubes containing fresh medium with or without 1 mM IPTG, respectively. After inoculation, cells were allowed to grow for 2 h. Next, increasing amounts of NiCl2 were added to each set. At appropriate times, aliquots were removed, diluted if necessary, and OD600 was measured. Control experiments were performed with a culture that was obtained at the 1st step of the selection experiment after single growth cycle in the absence of any toxic agent. In specificity tests, NiCl2 was substituted by ZnSO4 or CuSO4.
Homology Modeling
A structural model of the 20-mer peptide MSHAYFVCNRCDSSNHSAHE was built with the use of the DeepView/Swiss PdbViewer 3.7 software package (Guex and Peitsch 1997). The Metalloprotein Database at The Scripps Research Institute (http://metallo.scripps.edu) was extensively searched for an appropriate 3-dimensional (3D) template. The search yielded several zinc knuckle motifs as the most meaningful homologs of the 20-mer peptide. Six zinc knuckle domains with known structures were selected for modeling: 2 fragments of the HIV-1 nucleocapsid protein (PDB id. 1MFS), pos. 8–27 and pos. 29–48; 2 fragments of the HIV-1/MN nucleocapsid protein (PDB id. 1AAF), pos. 8–27 and pos. 29–48; a zinc finger–like domain from the HIV-1 GAG protein p55 (PDB id. 2ZNF); and a C-terminal zinc finger–like domain from the MMTV nucleocapsid protein (PDB id. 1DSV). Sequences of the 20-mer peptide and the various templates were aligned manually using the CX2CX4H motif as a consensus. Then, the template structures were superposed, and sequence of the 20-mer peptide was threaded onto them. Residues 1–5 of the peptide were removed from the model because of unsatisfactory fit with the templates. After minor adjustment of side chain geometries, this preliminary model was submitted to Swiss-Model server at http://swissmodel.expasy.org/SWISS-MODEL.html (Schwede et al. 2003) for further refinement. In the returned peptide structure, a tridentate-binding site for tetracoordinated metal ion was formed by side chains of Cys-8, Cys-11, and His-16. The 4th coordination position remained vacant allowing occupation by a solvent molecule. The model was systematically searched for a simple modification, which could bring an additional aminoacyl ligand into the coordination sphere of the metal ion with minimal change of backbone configuration. The best variant was an alteration of 
and 
angles of a single amino acid, Ser-17, from (–96.046o) and 153.146o to (–133.314o) and (–21.086o), respectively. It resulted in the close proximity of the imidazole ring of His-19 to the metal ion, and simultaneously an approach of Glu-20 to Arg-10 such that an ionic pair and H-bond network might occur between their side chains. The structure was further refined by energy minimization in a GROMOS96 force field with an IFP43B1 parameter set. The position of the metal ion was adjusted in agreement with the parameters of Ni–S (Cys) and Ni–N (Imidazole-His) bonds observed in nickel(II)-NikR complex (PDB id. 1Q5Y). The quality of the model was evaluated by PROCHECK, PROVE, and WHAT IF programs at European Molecular Biology Laboratory server (http://biotech.embl-ebi.ac.uk:8400) and found satisfactory.
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Library Design
A mini-gene library (fig. 1) was designed to express a 20-mer peptide with 12 randomized positions in its central part and an invariant 4-amino acid–long N- and C-terminus. The core element of the library was a 100-bp dsDNA cassette, which was developed as a standard module for composite genetic constructs with randomized segments. The cassette contained a 60-bp ORF with a Shine–Dalgarno sequence located 8 bp upstream of the translation start codon, XbaI and BamHI restriction sites at the ends, and several auxiliary restriction sites (BsmAI, FokI, and BspHI) that were intended for cassette modification and not used in this study. Randomized segment of the ORF was presented as 11 contiguous NNC triplets followed by 1 SRC triplet on the noncoding DNA strand, where N corresponds to A, T, G, or C, S is G or C, and R is A or G. The SRC triplet together with downstream invariant triplets encoded a 5-amino acid–long C-terminal tag (H/R/D/G)SAHE, which was expected to increase peptide stability in the E. coli cytoplasm (Bowie and Sauer 1989; Parsell et al. 1990
). The total sequence space of the library theoretically included 7.04 x 1013 DNA variants or 3.45 x 1013 peptide variants. The use of randomized triplets with a fixed C in the 3rd position inhibits the emergence of internal stop codons, equalizes distribution of different amino acids in the peptide sequence, and significantly decreases the coding redundancy of the library. It also diminishes the risk of a bias caused by an occasional digestion of the randomized segment by restriction enzymes in the course of library construction. On the other hand, this design excludes methionine, tryptophan, lysine, glutamine, and glutamic acid from the variable part of the peptide. However, because none of the above-mentioned amino acids possesses a unique functionality that cannot be provided by other amino acids, they were considered to be dispensable in this study. The cassette was inserted unidirectionally into pCR21-T7pt vector between a T7 promoter and a transcription terminator. This resulted in a plasmid library designated as pCRRL1, which carries ampicillin and kanamycin resistance determinants and requires host-provided T7 RNA polymerase to express the randomized ORF. Placing the randomized ORF under strong promoter on high copy number plasmid ensures a very intense intracellular synthesis of the 20-mer peptide and therefore was expected to contribute significantly to the modified E. coli phenotype. It is worthy to note that strong expression of the mini-gene might be a source of metabolic stress, which would combine with the imposed environmental one. However, when concentration of a bacteriotoxic substance in the medium is close to lethal, the environmental stress is likely to prevail and thus to direct the competition between clonal lineages associated with different mini-gene variants.
Selection and Primary Characterization of Ni2+-, Ag+-, and TeO32–-Resistant Clones
The pCRRL1 library was transformed into E. coli BLR(DE3) cells, which carry a T7 RNA polymerase gene under control of IPTG-regulated lacUV5 promoter. The number of individual clones was estimated to be 
105. The clone library was allowed to grow for 4 h in order to obtain several replicates of it and then split into parts intended for parallel evolution experiments using different toxic agents. Expression of the randomized ORF was induced by IPTG, and the multistep selection experiment was started by adding appropriate amounts of NiCl2, AgNO3, or K2TeO3 to the cell cultures (fig. 2). After 22–24 h of growth under stress, bacterial cultures exhibiting the highest toxicity tolerance in each set were used as a seed for the next cycle. A concentration gradient of the toxic agents was chosen to cover a range within which an MIC was expected to be found. Thus, at every step the selective pressure was adjusted depending on the achieved level of the toxicity tolerance to keep the evolving bacterial populations under subinhibitory conditions. Substantial increase in bacterial resistance to Ni2+, Ag+, and TeO32– was observed during 4, 5, and 7 consecutive selection cycles, respectively. After that, no further change in toxicity tolerance was detected. During the selection procedure, the apparent MIC values increased from 4 to 10 mM NiCl2, from 0.03 to 0.90 mM AgNO3, and from 0.7 to 20 µM K2TeO3. Eight clones were isolated from each of the 3 evolved resistant populations by plating. These were tested for the dependence of growth on mini-gene expression by measuring OD600 after 24 h of cultivation in the presence of corresponding toxic agent, with and without IPTG (table 1). Significant IPTG-induced growth improvement was observed for several Ni2+-resistant clones, whereas growth of Ag+- and TeO32–-resistant selectants was not affected by IPTG. Plasmid sequencing revealed absolute sequence homogeneity within each group of the studied clones. All 8 Ni2+-resistant clones harbored the same plasmid (designated as pCRRL1-N94-01) with a mini-gene variant coding for the peptide MSHAYFVCNRCDSSNHSAHE. In addition, a 9-bp deletion was detected in the spacer region between the ORF and the transcription terminator. However, because no important element of the mini-gene operon was located there, the deletion was unlikely to affect the peptide expression. Ag+-resistant clones were found to bear pCRRL1-A226-01 plasmid with mini-gene variant coding for an 18-mer with the sequence MSHATATPASRRRLPLRS. The shortening of the peptide was due to 2 noncontiguous single-base deletions inside the ORF, which resulted in frameshift and premature translation stop. Interestingly, the first nucleotide in the transcribed part of the mini-gene operon was found to be T instead of the original G. This can decrease the transcription efficiency (Imburgio et al. 2000) and disrupt a stability tag at the 5'-terminus of the transcript, thus making it vulnerable to ribonucleolytic degradation (Bouvet and Belasco 1992; Mackie 2000). All TeO32–-resistant clones carried a version of plasmid pCRRL1-T507-01 that has a 225-bp deletion of the whole mini-gene transcriptional unit together with adjacent downstream sequence. This deletion obviously abolishes T7 RNA polymerase activity on the plasmid, making it irresponsive to IPTG induction.
|
|
Assessment of Mini-Gene-Dependent Toxicity Tolerance in the Selected Resistant Clones
The existence of a positive correlation between the level of mini-gene expression and the extent of toxicity tolerance would strongly argue in favor of a role for the mini-gene in the mechanism of stress resistance of the evolved clones. Therefore, IPTG dependence of the toxicity tolerance was explored to establish the importance of the selected mini-gene variants for the improved bacterial performance under stress. Comparison of the bacterial growth parameters measured in the presence and absence of IPTG made it possible to distinguish the specific effect of mini-gene expression on culture survival and propagation from the basal resistance to the hostile environment acquired through adaptive genomic mutations. In the case of TeO32–-resistant clones, the observed lack of IPTG influence on their growth coincides with the deletion of the mini-gene operon from the plasmid. Thus, it is evident that the evolved TeO32– tolerance must be wholly attributed to mutations in the bacterial chromosome. The situation with the Ag+-resistant selectants is less certain. The presence of a seemingly functional mini-gene operon in the plasmid was nevertheless not accompanied by an IPTG-induced stimulation of bacterial growth in Ag+-containing medium. For 3 of the selected clones, the effect of IPTG on the growth curves was studied more thoroughly at 0.02–0.8 mM AgNO3 to confirm the results of the initial tests. Within the accuracy of measurement, no difference was found between cultures grown in the presence and absence of IPTG (data not shown). Transformation of the original E. coli BLR(DE3) strain with plasmids isolated from the resistant clones did not result in a phenotype with increased silver tolerance (data not shown). Therefore, it was concluded that either the peptide expressed from the mini-gene does not contribute by itself to the evolved silver tolerance or the expression proceeds with very low efficiency. Among 8 tested Ni2+-resistant selectants, 4 exhibited 2- to 4-fold IPTG-induced increase in OD600 measured after 24 h of cultivation in 9 mM NiCl2-containing medium. For the rest of the clones, only marginal growth stimulation by IPTG was observed under the same conditions. To elucidate such a diverse behavior, 2 clones representing the extremities of bacterial response to IPTG, N9405 and N9408, were taken for a detailed study of IPTG influence on their growth under nickel stress (fig. 3). It was found that IPTG significantly stimulates the growth of both clones when the concentration of NiCl2 in the medium was 4 mM and above. At lower nickel concentrations, IPTG provided little stimulation or even inhibited bacterial growth. Despite generally similar reaction to the IPTG induction, the clones apparently differed in their growth parameters, especially in the duration of the lag phase. It was concluded that the mini-gene is active in both selectants, but its actual contribution to the evolved nickel tolerance is likely to be modulated by differences in genetic background and may not be obvious under some experimental conditions. In contrast to the resistant clones, the nonadapted control culture was able to grow only when NiCl2 concentration in the medium was below 4 mM and was always insensitive to IPTG induction. Because the basal nickel tolerance of N9405 and N9408 clones was significantly higher than that of the control culture, it is clear that they acquired adaptive genomic mutations during the selection procedure.
|
Evaluating Growth Parameters of the Ni2+-Resistant N9405 Selectant
The influence of mini-gene expression on growth parameters of the bacterial host was studied in more detail in the N9405 selectant for which significant dependence of Ni2+ tolerance on IPTG was reliably established during preliminary tests. Bacterial growth curves were interpreted in terms of the Verhulst (eq. 1) and Gompertz (eq. 2) 3-parameter models (Tsoularis and Wallace 2002
):
 | (1) |
 | (2) |
, and k is an intrinsic growth rate, which represents reproduction rate for an isolated microorganism. The maximum specific growth rate µmax was calculated according to equation (3) for Verhulst model and equation (4) for Gompertz model using estimated values of N0, A, and k:
 | (3) |
 | (4) |
30% decrease of k value, whereas changes in A and N0 parameters were insignificant. The corresponding µmax value was 25% lower in the presence of IPTG than in its absence. Thus, although the expression of the mini-gene provided much better resistance to nickel intoxication, it was detrimental to growth in a stress-free environment.
|
In comparison with the nonadapted control culture, the N9405 clone yields noticeably lower cell density in stationary phase under normal growth conditions. This cannot be attributed to the activity of the mini-gene operon because the plateau of the growth curves shows little dependence on the IPTG induction. Instead, these decreased cell yields may result from the acquired genomic mutations, which provide a competitive advantage at (sub)inhibitory concentrations of NiCl2 but are deleterious in the stress-free environment.
Assay of Zn2+ and Cu2+ Tolerance of the N9405 Selectant
To obtain insight into the specificity of nickel tolerance provided by the selected mini-gene, we evaluated the viability of the N9405 clone in the presence of Zn2+ and Cu2+ (fig. 4). In both cases, the IPTG induction did not significantly change the resistance profiles, which means that the mini-gene expression does not contribute to the culture survival under stress induced by these metal ions. Also, the basal metal tolerance of the N9405 clone was only marginally higher (toward Zn2+) or even lower (toward Cu2+) than that of the control culture. These observations imply that the evolved mechanisms of Ni2+-resistance cannot protect the N9405 cells against zinc and copper intoxication. This demonstrates that both the selected mini-gene variant and the acquired mutations outside the mini-gene operon have considerable specificity to nickel ions.
|
Molecular Modeling of the Peptide Expressed from the Mini-Gene Operon in the N9405 Selectant
It seems likely that a mechanism by which the mini-gene might confer nickel resistance to the E. coli cells requires a direct interaction between nickel ions and the 20-mer peptide expressed from the mini-gene operon. We therefore examined the potential of the 20-mer to form a nickel-binding site. A structure of the putative peptide complex with nickel(II) (fig. 5) was generated with the homology-modeling module of DeepView/Swiss-Pdb Viewer (Guex and Peitsch 1997
). The known structure of zinc knuckle domain was used as a template because the major part of its signature motif was present in the N9405 20-mer peptide. In the computed 3D model of the complex, nickel(II) is coordinated by sulfur atoms of Cys-8 and Cys-11 and by N-donor imidazole rings of His-16 and His-19. Cys-8 and Cys-11 are located on the opposite sides of a ß-turn, which is the only regular secondary structure element in the model. The ß-turn is followed by a loop wrapped around the nickel ion and closed by the interaction between Arg-10 and Glu-20 side chains. The loop harbors His-16 and His-19 thus can participate in the formation of a tetrahedral metal coordination site. The model clearly demonstrates an ability of the sequence MSHAYFVCNRCDSSNHSAHE to fold into a realistic structure with potentially high affinity toward nickel(II).
|
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In the present study, we used a plasmid-borne library of expressed random ORFs (fig. 1) as a source of highly diversified genetic material for in vivo selection of functional genes that could facilitate resistance of the host cells toward toxic substances. The mini-gene used here expresses an mRNA of 148 residues compared with the 97- and 145-residue mRNAs used earlier (Tenson et al. 1997; Tripathi et al. 1998
). Once transformed into E. coli, the mini-gene library became a target for selection during bacterial growth under restrictive conditions. The method was expected to yield mini-genes that improve the bacterial growth in stressful environment if useful primary sequences were available in the sequence space encompassed by the library. The approach proved successful when the library-bearing E. coli culture was propagated under Ni2+-induced stress. A steady increase in nickel resistance of the bacterial population was observed throughout 4 consecutive selection cycles (fig. 2). By its conclusion, the selection process converged on a functional mini-gene, which markedly improved the performance of the evolved strains at elevated concentrations of NiCl2. The increased nickel tolerance of the isolated resistant clones was attributable to acquired adaptive mutations as well as the presence of the selected mini-gene. These effects were easy to recognize and separate because experimental system allows the expression of the mini-gene and consequently its contribution to bacterial stress resistance to be regulated with IPTG. In the absence of IPTG, the MIC of NiCl2 for the resistant clones carrying the mini-gene was in the 9–10 mM range as compared with 4 mM for the nonadapted control culture. Induction of mini-gene expression did not practically extend the range of tolerable NiCl2 concentrations but considerably improved the bacterial growth parameters near its upper limit. At the same time, the performance of the nickel-resistant clones under stress-free conditions was somewhat lower than that of the control strain, which can be regarded as a cost of adaptation to a hostile environment.
Although the precise role of the selected mini-gene in nickel detoxification has not yet been established, one might speculate that the mechanism of the evolved resistance relies on direct binding of nickel ions to 20-mer peptide encoded by the mini-gene. Indeed, the selected peptide sequence contains a CX2C motif, which has been found to be involved in the coordination of Zn2+, Cu2+, Ni2+, Co2+, Fe2+, or Hg2+ ions through the thiol groups of cystines in a wide variety of metalloproteins and metal transporters (Dykema et al. 1999; Xing and DeRose 2001; DeSilva et al. 2002; Opella et al. 2002). The remainder of the binding site can be formed by the side chains of histidine residues in a manner similar to that observed in some types of zinc fingers (fig. 5) (Krishna et al. 2003). It nevertheless seems unlikely that the peptide would specifically bind only Ni2+ ions because analogous cystine- and histidine-containing metal-binding sites in previously studied proteins cannot efficiently discriminate between the above-mentioned cations (Hartwig 2001). Therefore, the peptide might be expected to chelate Zn2+ and Cu2+ as well as Ni2+. However, the induction of the mini-gene expression did not increase zinc and copper tolerance of the nickel-resistant N9405 clone (fig. 4). Thus, the mechanism of mini-gene-dependent nickel detoxification is unlikely to be restricted to simple sequestration of Ni2+ ions by the peptide. Instead, the improved performance of the selected clones at subinhibitory Ni2+ concentrations may be determined by cooperative functioning of the 20-mer peptide and some host cellular factors, which are responsible for the observed nickel specificity of the mini-gene protective effect. If this is the case, the selected peptide may be regarded as a key element of newly arisen metabolic subnet capable of countering Ni2+-induced stress.
In contrast to the Ni2+-driven selection, the parallel experiments with Ag+ and TeO32– yielded the resistant phenotypes that were either associated with inactive variant of the mini-gene (Ag+) or carried a plasmid in which the whole mini-gene operon was deleted (TeO32–) (table 1). The high toxicity tolerance of the evolved silver- and tellurite-resistant clones was apparently provided by the adaptive genomic mutations, whereas the accompanying plasmid variants were likely selected by means of genetic hitchhiking. These results suggest a lack of useful genetic information in the explored part of the mini-gene sequence space. If no advantageous peptide could be expressed from the library, it seems natural that the bacterial population would eventually be taken over by the clones that carry plasmid with a deleted or inert mini-gene as the next best variant. On the other hand, the same outcome might be expected if some genomic mutations with strong beneficial effect on the bacterial resistance were readily available at the early stages of the selection process. In such a circumstance, even if the mini-gene library actually contained useful sequences, it might be eliminated depending on which mini-gene version happened to accompany the highly beneficial mutation.
The observed rapid and dramatic improvement in silver and tellurite tolerance of the evolving E. coli populations, which resulted in 30-fold and 28-fold overall increase in MIC of Ag+ and TeO32–, respectively (fig. 2), is rather consistent with an emergence of a single adaptive mutation with high selective value than with sequential accumulation of multiple weakly advantageous genetic changes. Although beneficial mutations with strong positive effect on fitness are believed to be exceptionally rare (de Visser and Rozen 2005), there are examples where they can be acquired with relative ease (Tenaillon et al. 2004; Wright 2004). For example, any recombination event that places a cryptic E. coli operon tehAB behind a strong promoter would likely result in a substantial increase in tellurite tolerance of the mutant clone (Taylor 1999). Such a resistant mutant might have swept the E. coli population that was cultivated in the presence of TeO32–, thus promoting the spread of an accompanying mini-gene irrespectively to its selective value. If this was the case, a deleterious mini-gene might be accidentally selected and dominate the population for a short time, only to be later eliminated via deletion of the whole mini-gene operon. Evolution of the E. coli population under Ag+-induced stress might have followed a similar scenario. However, our results do not exclude the possibility that the mini-gene found in the Ag+-resistant clones contributed to the increased silver tolerance at the early stages of the selection procedure. An initial positive effect on fitness might have later been obliterated due to the development of alternative and more efficient mechanisms of toxicity tolerance during progressive adaptation to the stress. Known silver-resistant E. coli mutants have been reported to withstand high environmental concentrations of Ag+ ions mainly due to intensified efflux and decreased permeability of the outer membrane (Li et al. 1997). Rapid development of such potent concurrent mechanisms of Ag+ detoxification could greatly devaluate the initial contribution of the mini-gene to silver tolerance, if there was any. After being rendered useless for cell survival, the mini-gene could be inactivated by the observed postpromoter mutation.
Unlike Ag+ and TeO32–, nickel is an essential E. coli micronutrient because it is a key component of the metalloenzymes hydrogenases I–III and glyoxylase I (Mulrooney and Hausinger 2003). Under normal growth conditions, intracellular concentration of Ni2+ ions is tightly regulated by several influx and efflux systems to remain within a range, which is compatible with E. coli viability (Nies 1999; Mulrooney and Hausinger 2003; Rodrigue et al. 2005; Iwig et al. 2006). Therefore, a substantial improvement in E. coli resistance to high Ni2+ levels might require a longer time for acquisition of beneficial genetic changes than in the case of Ag+ or TeO32– due to more stringent constraints on an acceptable Ni2+ detoxification mechanism. This process likely was accelerated and enhanced by the presence of the mini-gene.
Altogether, our results indicate that evolution of the experimental E. coli populations was determined by a competitive dynamics of beneficial genetic innovations originating from different sources of diversity. Being initially associated only with the randomized segment of a plasmid-borne mini-gene, the genetic variance of the E. coli populations later included contributions from mutations arising in the genomic DNA as well as by genetic changes in the plasmid, including those that involve the mini-gene operon. The design of the experimental system allows the focus of the selection to be shifted away from the mini-gene library if the selective value of an emerged mutation was substantially higher than that of any member of the library. However, if independent adaptive mutations are sufficiently rare, the selection on the mini-gene library may be complete before any adaptive mutations are established in the population. Thus, the successful selection of a functional mini-gene in the library bearing an E. coli population proliferating under nickel-induced stress might have been due to limited intrinsic ability of E. coli to adapt quickly to excess levels of Ni2+ ions. In the future, the methodology can likely be made more realistic by placing the selection system in the genome rather than on a plasmid.
As discussed in Introduction, current technologies have used molecular evolution primarily as a tool to modify biological activity of natural macromolecules aiming mainly at their stability, catalytic properties, or affinity toward specific ligands (Bull and Wichman 2001). To this end, in vivo and in vitro selection protocols rely on structurally constrained combinatorial libraries obtained through local randomization of the sequences of appropriate proteins or nucleic acids (Kumar and Ellington 1995; Bittker et al. 2002; Jestin and Kaminski 2004; Williams et al. 2004). At the same time, little attention has been paid to the selection of functional macromolecules from extensively randomized libraries exhibiting no a priori bias toward any specific biological activity. This could explain why the frequency of meaningful variants in a random collection of protein or nucleic acid sequences is considered to be exceptionally low (Keefe and Szostak 2001). In the present study, however, a relevant gene of very modest size emerged from a small pool (105) of arbitrary sequences. This demonstrates the applicability of the described in vivo selection method for the discovery of novel functional macromolecules in unconstrained libraries of random biopolymers. Although the interplay between genetic variances associated with spontaneous mutations and the mini-gene library adds complexity to interpretation of results, it makes the described model system suitable for studying cooperative phenomena in molecular evolution.
In conclusion, the approach described here is a valuable adjunct to real-time studies of molecular evolution in bacteria. For example, the present study can be viewed as an experimental model for the contribution horizontal gene transfer can make to adaptation under conditions of stress when the horizontally transferred gene is already potentially active. However, instead of examining the effect that a particular gene might have, 105 variants were examined in a single experiment. If instead the promoter were eliminated or damaged, one could alternatively examine the cell's ability to activate a known to be useful mini-gene–encoded sequence such as the one selected here for nickel tolerance. More generally, the approach can be used to elucidate coordinated mechanisms of bacterial adaptation by examining global genetic and phenotypic changes using transcriptional analysis or resequencing approaches.
|
Footnotes |
|---|
Dan Graur, Associate Editor
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Adda CG, Anders RF, Tilley L, Foley M. Random sequence libraries displayed on phage: identification of biologically important molecules. Comb Chem High Throughput Screen (2002) 5:1–14.[ISI][Medline]
Aharoni A, Griffiths AD, Tawfik DS. High-throughput screens and selections of enzyme-encoding genes. Curr Opin Chem Biol (2005) 9:210–216.[CrossRef][ISI][Medline]
Arber W. Elements for a theory of molecular evolution. Gene (2003) 317:3–11.[CrossRef][ISI][Medline]
Benhar I. Biotechnology applications of phage and cell surface display. Biotechnol Adv (2001) 19:1–33.[CrossRef][ISI][Medline]
Bittker JA, Phillips KJ, Liu DR. Recent advances in the in vitro evolution of nucleic acids. Curr Opin Chem Biol (2002) 6:367–374.[CrossRef][ISI][Medline]
Bouvet P, Belasco JG. Control of RNase E-mediated RNA degradation by 5'-terminal base pairing in E. coli. Nature (1992) 360:488–491.[CrossRef][Medline]
Bowie JU, Sauer RT. Identification of C-terminal extensions that protect proteins from intracellular proteolysis. J Biol Chem (1989) 264:7596–7602.
Bull JJ, Wichman HA. Applied evolution. Annu Rev Ecol Syst (2001) 32:183–217.[CrossRef][ISI]
Cipolla L. Combinatorial libraries of biocatalysts: application and screening. Comb Chem High Throughput Screen (2004) 7:101–114.[ISI][Medline]
DeSilva TM, Veglia G, Porcelli F, Prantner AM, Opella SJ. Selectivity in heavy metal-binding to peptides and proteins. Biopolymers (2002) 64:189–197.[CrossRef][ISI][Medline]
de Visser JA, Rozen DE. Limits to adaptation in asexual populations. J Evol Biol (2005) 18:779–788.[CrossRef][ISI][Medline]
Domingo E, Holland JJ. RNA virus mutations and fitness for survival. Annu Rev Microbiol (1997) 51:151–178.[CrossRef][ISI][Medline]
Dykema PE, Sipes PR, Marie A, Biermann BJ, Crowell DN, Randall SK. A new class of proteins capable of binding transition metals. Plant Mol Biol (1999) 41:139–150.[CrossRef][ISI][Medline]
Elena SF, Lenski RE. Evolution experiments with microorganisms: the dynamics and genetic bases of adaptation. Nat Rev Genet (2003) 4:457–469.[ISI][Medline]
Elena SF, Miralles R, Cuevas JM, Turner PE, Moya A. The two faces of mutation: extinction and adaptation in RNA viruses. IUBMB Life (2000) 49:5–9.[CrossRef][ISI][Medline]
Falciani C, Lozzi L, Pini A, Bracci L. Bioactive peptides from libraries. Chem Biol (2005) 12:417–426.[CrossRef][ISI][Medline]
Guex N, Peitsch MC. SWISS-MODEL and the Swiss-PdbViewer: an environment for comparative protein modelling. Electrophoresis (1997) 18:2714–2723.[CrossRef][ISI][Medline]
Hartwig A. Zinc finger proteins as potential targets for toxic metal ions: differential effects on structure and function. Antioxid Redox Signal (2001) 3:625–634.[CrossRef][ISI][Medline]
Hoess RH. Protein design and phage display. Chem Rev (2001) 101:3205–3218.[CrossRef][ISI][Medline]
Imburgio D, Rong M, Ma K, McAllister WT. Studies of promoter recognition and start site selection by T7 RNA polymerase using a comprehensive collection of promoter variants. Biochemistry (2000) 39:10419–10430.[CrossRef][Medline]
Iwig JS, Rowe JL, Chivers PT. Nickel homeostasis in Escherichia coli—the rcnR-rcnA efflux pathway and its linkage to NikR function. Mol Microbiol (2006) 62:252–262.[CrossRef][ISI][Medline]
Jestin JL, Kaminski PA. Directed enzyme evolution and selections for catalysis based on product formation. J Biotechnol (2004) 113:85–103.[CrossRef][ISI][Medline]
Jostock T, Dubel S. Screening of molecular repertoires by microbial surface display. Comb Chem High Throughput Screen (2005) 8:127–133.[CrossRef][ISI][Medline]
Joung JK. Identifying and modifying protein-DNA and protein-protein interactions using a bacterial two-hybrid selection system. J Cell Biochem (2001) 37(Suppl):53–57.
Keefe AD, Szostak JW. Functional proteins from a random-sequence library. Nature (2001) 410:715–718.[CrossRef][Medline]
Krishna SS, Majumdar I, Grishin NV. Structural classification of zinc fingers: survey and summary. Nucleic Acids Res (2003) 31:532–550.
Kumar PKR, Ellington AD. Artificial evolution and natural ribozymes. FASEB J (1995) 9:1183–1195.[Abstract]
Li XZ, Nikaido H, Williams KE. Silver-resistant mutants of Escherichia coli display active efflux of Ag+ and are deficient in porins. J Bacteriol (1997) 179:6127–6132.
Mackie GA. Stabilization of circular rpsT mRNA demonstrates the 5'-end dependence of RNase E action in vivo. J Biol Chem (2000) 275:25069–25072.
Mulrooney SB, Hausinger RP. Nickel uptake and utilization by microorganisms. FEMS Microbiol Rev (2003) 27:239–261.[CrossRef][ISI][Medline]
Neylon C. Chemical and biochemical strategies for the randomization of protein encoding DNA sequences: library construction methods for directed evolution. Nucleic Acids Res (2004) 32:1448–1459.
Nies DH. Microbial heavy-metal resistance. Appl Microbiol Biotechnol (1999) 51:730–750.[CrossRef][ISI][Medline]
Opella SJ, DeSilva TM, Veglia G. Structural biology of metal-binding sequences. Curr Opin Chem Biol (2002) 6:217–223.[CrossRef][ISI][Medline]
Parsell DA, Silber KR, Sauer RT. Carboxy-terminal determinants of intracellular protein degradation. Genes Dev (1990) 4:277–286.
Rodrigue A, Effantin G, Mandrand-Berthelot MA. Identification of rcnA (yohM), a nickel and cobalt resistance gene in Escherichia coli. J Bacteriol (2005) 187:2912–2916.
Rosander A, Bjerketorp J, Frykberg L, Jacobsson K. Phage display as a novel screening method to identify extracellular proteins. J Microbiol Methods (2002) 51:43–55.[CrossRef][ISI][Medline]
Schwede T, Kopp J, Guex N, Peitsch MC. SWISS-MODEL: an automated protein homology-modeling server. Nucleic Acids Res (2003) 31:3381–3385.
Souza V, Travisano M, Turner PE, Eguiarte LE. Does experimental evolution reflect patterns in natural populations? E. coli strains from long-term studies compared with wild isolates. Antonie Van Leeuwenhoek (2002) 81:143–153.[CrossRef][ISI][Medline]
Szardenings M. Phage display of random peptide libraries: applications, limits, and potential. J Recept Signal Transduct Res (2003) 23:307–349.[CrossRef][Medline]
Taylor DE. Bacterial tellurite resistance. Trends Microbiol (1999) 7:111–115.[CrossRef][ISI][Medline]
Tenaillon O, Denamur E, Matic I. Evolutionary significance of stress-induced mutagenesis in bacteria. Trends Microbiol (2004) 12:264–270.[CrossRef][ISI][Medline]
Tenson T, DeBlasio A, Mankin AS. A functional peptide encoded in the Escherichia coli 23S rRNA. Proc Natl Acad Sci USA (1996) 93:5641–5646.
Tenson T, Mankin AS. Short peptides conferring resistance to macrolide antibiotics. Peptides (2001) 22:1661–1668.[CrossRef][ISI][Medline]
Tenson T, Xiong L, Kloss P, Mankin AS. Erythromycin resistance peptides selected from random peptide libraries. J Biol Chem (1997) 272:17425–17430.
Travisano M, Rainey PB. Studies of adaptive radiation using model microbial systems. Am Nat (2000) 156:S35–S44.[CrossRef][ISI]
Tripathi S, Kloss PS, Mankin AS. Ketolide resistance conferred by short peptides. J Biol Chem (1998) 273:20073–20077.
Tsoularis A, Wallace J. Analysis of logistic growth models. Math Biosci (2002) 179:21–55.[CrossRef][ISI][Medline]
Uchiyama F, Tanaka Y, Minari Y, Tokui N. Designing scaffolds of peptides for phage display libraries. J Biosci Bioeng (2005) 99:448–456.[CrossRef][ISI][Medline]
Vimberg V, Xiong L, Bailey M, Tenson T, Mankin A. Peptide-mediated macrolide resistance reveals possible specific interactions in the nascent peptide exit tunnel. Mol Microbiol (2004) 54:376–385.[CrossRef][ISI][Medline]
Williams GJ, Nelson AS, Berry A. Directed evolution of enzymes for biocatalysis and the life sciences. Cell Mol Life Sci (2004) 61:3034–3046.[CrossRef][ISI][Medline]
Wittrup KD. Protein engineering by cell surface display. Curr Opin Biotechnol (2001) 12:395–399.[CrossRef][ISI][Medline]
Wright BE. Stress-directed adaptive mutations and evolution. Mol Microbiol (2004) 52:643–650.[CrossRef][ISI][Medline]
Xing G, DeRose VJ. Designing metal-peptide models for protein structure and function. Curr Opin Chem Biol (2001) 5:196–200.[CrossRef][ISI][Medline]
