Your browser is out-of-date!

Update your browser to view this website correctly. Update my browser now

×

Discipline
Biological
Keywords
Synthetic Biology
Suppressor TRNA
Translation Initiation
Escherichia Coli
Initiator TRNA
Observation Type
Standalone
Nature
Standard Data
Submitted
Apr 3rd, 2019
Published
May 1st, 2019
  • Abstract

    Translation initiation is a sequential process involving interactions between the 30S small ribosomal subunit, initiation factors and initiator tRNA. The Escherichia coli K-12 strain is unique in the Escherichia because it has two different initiator tRNA sequences, tRNAfMet1 encoded by the metZWV genes and tRNAfMet2 encoded by the metY gene. A mutant of the metY gene was previously made where the anticodon sequence, responsible for specifying the start codon where translation initiation begins, was changed so that it bound to the amber stop codon UAG instead of the usual AUG start codon. This amber initiator tRNA has already been shown to be functional in the K-12 strain, but it is unclear whether it would function in other strains normally lacking the tRNAfMet2 variant. In this work, we transformed E. coli K-12, and four other generally regarded as safe (GRAS) laboratory strains, with a plasmid expressing the amber initiator tRNA and evaluated its functionality and growth effects on the bacteria. We performed these tests because, despite these strains all belonging to E. coli phylogenetic group A, it is well known that there is significant variation between even closely related E. coli strains in their metabolism, transcriptional response to exogenous DNA expression and rates of amber stop codon suppression. We found that the amber initiator functions similarly across the five strains, effectively initiating translation at the orthogonal UAG start codon and that it had modest growth-slowing effects in the Crooks, W, and K-12 strains. The five tested E. coli strains in this work (K-12, B, C, W, and Crooks) are important workhorses of academic and industrial research and development. The path is now clear to deploy the amber initiator tRNA into these five strains to precisely control gene expression.

  • Figure
  • Introduction

    Translation initiation in Escherichia coli relies on recruitment of initiation factors 1-3 (IF1-IF3), the 30S ribosomal subunit, and the initiator tRNA (tRNAfMet) to a messenger RNA (mRNA). Several excellent reviews with more detailed information on the process of bacterial translation initiation have been published.

    E. coli encodes two sequence variants of the initiator tRNA, called tRNAfMet1 and tRNAfMet2. The difference between these sequence variants is one nucleotide at position 46 where tRNAfMet1 is a G (modified to 7mG in tRNA) and tRNAfMet2 is an A. Both the tRNAfMet1 (metZWV genes) and tRNAfMet2 (metY gene) variants are present in the E. coli K-12 strains, but other strains of E. coli lack the tRNAfMet2 and instead have a fourth copy of tRNAfMet1 in the metY loci in their genomes (Fig. 1B). tRNAfMet2 has been introduced previously into E. coli B strains with no apparent deleterious effect.

    The amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} (Fig. 1C) was created by Varshney and RajBhandary from a modified E. coli K-12 metY gene encoding a tRNA that can initiate translation at the ‘amber’ stop codon (UAG) instead of the usual AUG start codon. A major activity of biological engineers is the construction of genetic parts and devices that tell the cell when to turn on a gene, resulting in protein production. Predicting whether engineered genetic constructs will work across different organisms is non-trivial and often results in non-functional or incorrectly functioning systems. Differences in basal levels of amber suppression within common K-12 strains is already known. Expressing the amber initiator tRNA shows great promise as a way of programming orthogonal translation initiation at UAG sites in engineered genetic constructs, but it is unknown whether tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} will function outside of the K-12 lineage of E. coli.

  • Objective

    To enable the adoption of this powerful amber initiator tRNA system to drive programmable gene expression, we wanted to know how transportable it is across E. coli strains? In particular, how would the amber initiator functions in laboratory E. coli strains that normally do not contain the tRNAfMet2 sequence variant?

  • Results & Discussion

    We began our investigation by building a paired set of plasmids to both express the amber initiator tRNA as well as measure the translation initiation efficiency at AUG and UAG codons. The amber initiator plasmid features a recoded metY gene (metY(CUA)) that expresses tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} . The amber initiator tRNA features an altered anticodon sequence that is the reverse complement of the desired UAG start codon (Fig. 1C). The plasmid contains an IPTG-inducible tac promoter and the Clo DF13 (medium copy) origin of replication expected to function across diverse E. coli strains (Fig. S3). We designed three reporter plasmids to measure translation initiation events caused by expression of tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} from the amber initiation plasmid. The reporter plasmids use super-folder green fluorescent protein (sfGFP) with three start codon variants: UAG to measure translation initiation efficiency from the amber initiator tRNA, AUG as a positive control, and GCC as a negative control (Fig. S3). The sfGFP reporter is driven by a constitutive T5 promoter and has the ColE1 origin of replication, which is compatible with the Clo DF13 origin, so both plasmids can be simultaneously maintained within one E. coli cell.

    We next transformed the amber initiator and reporter plasmids into five common generally recognized as safe (GRAS) laboratory strains of E. coli from a diverse set: MG1655 (K-12 strain), BL21 (B strain), C122 (C strain), ATCC 9637 (W strain), and ATCC 8739 (Crooks strain). All five E. coli strains are designated risk group 1 but arise from distinct lineages (Fig. 1A).

    After transforming both amber initiator plasmid and reporter plasmids into all strains, we measured bulk fluorescence under repressing and inducing conditions (Fig. 1D). We observed similar trends in all tested strains. As expected, the reporter strain with sfGFP(GCC) start codon did not appreciably change expression due to tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression. In contrast, the sfGFP(UAG) reporter had fluorescence levels similar to sfGFP(GCC) when under tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} repressed conditions, but increased markedly when tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression was induced (Fig. 1D). On average across the strains we saw an 88-fold increase in fluorescence upon tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression with MG1655 showing the smallest increase and C122 showing the largest increase. As noted previously, the combination of lack of expression when repressed, along with a large increase after tRNA induction, could be useful for conditionally expressing toxic proteins due to lack of translational leakage from the UAG stop codon in the absence of the tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} .

    Surprisingly, tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} induction also caused increased fluorescence from the normal sfGFP(AUG) start codon reporter in all strains. We would not have expected the increase because it has been well established that the wild-type tRNAfMet2 initiator does not initiate translation from UAG stop codons. We would therefore not expect the reciprocal codon-anticodon pairing between the AUG start codon and the amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} to result in productive translation initiation. We next expressed wild-type metY from the pULTRA plasmid and found no fluorescence increase from the UAG reporter as expected, and a similar magnitude increase in fluorescence from the AUG reporter as was seen with the amber initiator. We speculate that tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression has caused an increase in the cellular concentration of ribosomes and this has resulted in increased translation from the sfGFP(AUG) reporter. Support for this idea comes from a recent paper showing evidence that expressing initiator tRNA in E. coli increases mature ribosome levels and accelerates rRNA processing and our recent work showing the amber initiator can increase ribosome protein abundance in cells. We can speculate that expressing tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} has a similar effect on ribosomal maturation and rRNA processing in the cell, considering the presence of identical identity elements (aside from the anticodon sequence) in both initiator tRNAs.

    We next compared the effects of expressing the amber initiator on relative increases of sfGFP(AUG) and sfGFP(UAG) reporter expression to determine the suitability of the strains using the amber initiator (Fig. 1E). The results show the strains are all relatively similar, with E. coli C122 being the best and Crooks being the worst for selectively expressing UAG start codons over AUG initiation.

    To determine the fitness effects of expressing the amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} on the different E. coli strains, we measured growth rate and maximal optical density differences when the tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} transcription was induced versus repressed (Fig. 1F). We observed that the Crooks strain had both its growth rate and maximum optical density moderately reduced when the tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} was induced, while the W and K-12 strains had a major reduction in growth rate but not maximum optical density (Fig. 1F). The B and C strain growth characteristics were unaffected by tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} induction. The patterns of fitness changes following tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} induction seemed to mirror the strain’s relative abilities to produce UAG-initiated reporter protein (Fig. 1E). That is, the B and C strains produced the most UAG reporter fluorescence compared to AUG reporter fluorescence, while the Crooks and W strains produced the least (Fig. 1E). The exception to this pattern is the K-12 strain, which produced the third highest amount of UAG reporter per AUG reporter fluorescence despite the growth rate being reduced by the most (~50%) (Fig. 1E and 1F). The reason for the difference in K-12 may be due to interference between the amber tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} and the genomic copy of wild-type tRNAfMet2 in this strain, which is missing in the other strains examined in this study (Fig. 1B). Further work will need to be done to uncover the reasons for these strain-specific fitness differences.

    In this work, we have shown that the amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} can be expressed in a fully functional form in five different strains of generally regarded as safe laboratory E. coli. Furthermore, the genomic absence of the tRNAfMet2 (metY) gene variant does not seem to have a major bearing on the effectiveness of the amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} function within those strains (B, C, W, and Crooks) (Fig. 1B). Lastly, this work points towards possible uses for the amber initiator tRNA to enable expression of toxic proteins in strains of E. coli, such as BL21, that are often used for the production of protein for structural biology, investigation of protein function, and for therapeutics.

  • Methods

    Bacterial strains

    The following strains of E. coli were used in this study: MG1655 (K-12 strain), BL21 (B strain, ATCC #BAA-1025), C122 (C strain, NCTC122, Public Health England), Crooks (ATCC #8739), and W (ATCC #9637). All strains were grown in lysogeny broth Miller (LBM) or on LBM agar, supplemented with carbenicillin (Sigma-Aldrich #C9231) and/or spectinomycin (Sigma-Aldrich #S4014).

    Phylogenetic Tree

    A phylogenetic tree of 16 diverse E. coli strains with Escherichia fergusonii as an outgroup was created using the Phylogeny.fr analysis pipeline. Sequences were collected from NCBI GenBank: BL21 (TaKaRa) (NZ_CP010816.1), Crooks ATCC 8739 (NC_010468.1), K-12 substr. W3110 (NC_007779.1), K-12 substr. MG1655 (NC_000913.3), Nissle 1917 (NZ_CP007799.1), W strain ATCC9637 (NC_017635.1), C strain C122 (NZ_LT906474.1), O6:K15:H31 536 (NC_008253.1), 55989 (NC_011748.1), O127:H6 E2348/69 (NC_011601.1), IAI39 (NC_011750.1), O157:H7 str. Sakai (NC_002695.1), UMN026 (NC_011751.1), CFT073 (NC_004431.1), SMS-3-5 (NC_010498.1), Escherichia fergusonii ATCC 35469 (NC_011740.1). MLST analysis of sequences was performed by MLST2.0 run by the Center for Genomic Epidemiology, DTU (https://cge.cbs.dtu.dk/services/MLST/) using the MLST E. coli#1 configuration (adk, fumC, gyrB, icd, mdh, purA, recA genes) and default settings. MLST allele sequences for each of the 16 strains were downloaded, concatenated, and submitted to results from the Phylogeny.fr Phylogeny Analysis pipeline (). In this pipeline, a MUSCLE multiple sequence alignment of concatenated MLST alleles was performed. The alignment was then processed by Gblocks and phylogeny analysis performed by PhyML with tree drawing by TreeDyn.

    Plasmid construction

    Reporter plasmids were designed to contain superfolder GFP (sfGFP) with different start codons. A 1971bp gBlock (Integrated DNA Technologies, Inc) was designed and synthesized that contained the metY gene from MG1655 (GenBank: U00096.3) with an anticodon mutation (CAU > CUA). A pULTRA-CNF vector (Addgene #48215) with a Clo DF13 origin of replication and a spectinomycin-resistance cassette was used as the plasmid backbone. The amber initiator plasmid was assembled via in vitro homologous assembly. Cultures for plasmid construction were grown in LBM supplemented with spectinomycin. Plasmids were isolated from transformants, sequence verified, and stored in elution buffer (10 mM Tris-Cl, pH 8.5).

    Culture growth conditions for assay measurements

    Transformants from glycerol stocks were streaked on LBM plates with appropriate antibiotics: spectinomycin (100 µg/mL) for amber initiator plasmid and/or carbenicillin (100 µg/mL) for reporter plasmids and grown overnight at 37°C. Individual colonies were inoculated in 2 mL of LBM containing appropriate antibiotics and grown overnight at 37°C shaking at 200 rpm. After overnight growth, each culture was diluted 1:100 into 400 µL LBM in a 96-well deep well plate containing either 1 mM IPTG to induce, or 2% glucose (w/v) to repress, metY(CUA) expression.

    Fluorescence measurements

    Measurements of fluorescence intensity from the amber initiator plasmid system was performed as before with the following modifications: Absorbance was measured at 600 nm (OD600) to estimate culture density, followed by fluorescence (excitation = 485 nm, emission = 520 nm, bandwidth = 9 nm) measured at a single gain setting (high gain sensitivity = 111), on a PHERAstar FSX (BMG Labtech) plate reader.

    Fitness analysis

    Strains were grown overnight as described above. Cell densities were measured and cells were passaged to a starting OD600 of 0.1 into fresh 200 µL LBM with appropriate antibiotics and either 1 mM IPTG or 2% glucose (w/v). Cultures were grown in a flat-bottom 96-well plate sealed with gas-permeable seal (Sigma-Aldrich # Z763624) at 37°C (300 rpm) while culture light scattering (OD600) over time was measured on a SPECTROstar NANO (BMG Labtech) plate reader at 5 min intervals for 10 h. Our analysis method was adapted from previous work and used to compare ratios of growth rate and maximal cell density for individual strains.

    The growth rate and maximal cell density (max OD600) were calculated using the R package Growthcurver v.0.2.1. Ratios of growth rates and maximal cell densities were determined for strains with the amber initiation plasmid under induced growth condition (1 mM IPTG) versus repressed growth condition (2% glucose (w/v)).

    Whole Genome Alignment

    The following nucleotide sequences from GenBank were analyzed using Mauve whole genome alignment performed with default parameters: K-12 substr. MG1655 (NC_000913.3), BL21 (NZ_CP010816.1), C strain C122 (NZ_LT906474.1), W strain ATCC9637 (NC_017635.1), and Crooks ATCC 8739 (NC_010468.1).

    metY Multiple Sequence Alignment

    Gene sequences were extracted from GenBank genome sequences and aligned using MUSCLE algorithm using default parameters.

  • Funding statement

    PRJ was supported by the Molecular Sciences Department, Faculty of Science & Engineering, and the Deputy Vice-Chancellor (Research) of Macquarie University. RMV is a recipient of the Macquarie University Research Excellence PhD scholarship (MQRES). PFY was supported by a summer research scholarship from the Molecular Sciences Department of Macquarie University.

  • Ethics statement

    This study was completed in accordance with Macquarie University biosafety committee approval, application ID 5201600567.

  • References
  • 1
    Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum

    Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum ipsum

    Lorem ipsum Lorem ipsum Lorem ipsum
    2
    Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum

    Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum ipsum

    Lorem ipsum Lorem ipsum Lorem ipsum
    3
    Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum

    Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum ipsum

    Lorem ipsum Lorem ipsum Lorem ipsum
    4
    Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum

    Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum ipsum

    Lorem ipsum Lorem ipsum Lorem ipsum
    5
    Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum

    Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum ipsum Lorem ipsum Lorem ipsum Lorem ipsum Lorem ipsum ipsum

    Lorem ipsum Lorem ipsum Lorem ipsum
    Matters9.5/20

    An orthogonal amber initiator tRNA functions similarly across diverse Escherichia coli laboratory strains

    Affiliation listing not available.
    Abstractlink

    Translation initiation is a sequential process involving interactions between the 30S small ribosomal subunit, initiation factors and initiator tRNA. The Escherichia coli K-12 strain is unique in the Escherichia because it has two different initiator tRNA sequences, tRNAfMet1 encoded by the metZWV genes and tRNAfMet2 encoded by the metY gene. A mutant of the metY gene was previously made where the anticodon sequence, responsible for specifying the start codon where translation initiation begins, was changed so that it bound to the amber stop codon UAG instead of the usual AUG start codon[1]. This amber initiator tRNA has already been shown to be functional in the K-12 strain[1][2], but it is unclear whether it would function in other strains normally lacking the tRNAfMet2 variant. In this work, we transformed E. coli K-12, and four other generally regarded as safe (GRAS) laboratory strains, with a plasmid expressing the amber initiator tRNA and evaluated its functionality and growth effects on the bacteria. We performed these tests because, despite these strains all belonging to E. coli phylogenetic group A, it is well known that there is significant variation between even closely related E. coli strains in their metabolism[3][4][5], transcriptional response to exogenous DNA expression[6] and rates of amber stop codon suppression[7]. We found that the amber initiator functions similarly across the five strains, effectively initiating translation at the orthogonal UAG start codon and that it had modest growth-slowing effects in the Crooks, W, and K-12 strains. The five tested E. coli strains in this work (K-12, B, C, W, and Crooks) are important workhorses of academic and industrial research and development. The path is now clear to deploy the amber initiator tRNA into these five strains to precisely control gene expression.

    Figure 1. An orthogonal amber initiator tRNA functions similarly across diverse Escherichia coli laboratory strains.

    (A) Phylogenetic tree of 16 diverse E. coli strains. The tree has drawn from MUSCLE multiple sequence alignment of concatenated multisequence locus typing (MSLT) gene sequences (adk, fumC, gyrB, icd, mdh, purA, recA). Alignment processed by Gblocks and phylogeny analysis by PhyML. The tree was drawn by TreeDyn. Bootstrapping values (red) displayed on tree generated by MrBayes v3.2[8] and represent the confidence level of the displayed branching topology, with 1 being the highest level of confidence. The scale bar represents the number of nucleotide substitutions per site. Phylogeny.fr used for analysis pipeline. Five E. coli laboratory strains analyzed in this work shown by bold green text[9][10][11][12][13][14]. See figure S1 for Mauve whole genome alignment between E. coli strains in green[15][16].

    (B) metZWV and metY loci in E. coli strains. Initiator tRNAfMet1 and tRNAfMet2 differ in sequence by a single nucleotide at position 46 in the variable loop, where tRNAfMet1 has 7mG and tRNAfMet2 has an A[17]. The dashed lines represent similar genomic loci in each strain while boxes schematically represent the variation of tRNAfMet1 and tRNAfMet2 placement within each locus. See figure S2 for metY multiple sequence alignment.

    (C) Initiator tRNA structure showing anticodon pairing with the mRNA start codon. Left inset, wild-type initiator tRNAfMet2 pairing with canonical AUG start codon. Right inset, amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} pairing with amber stop codon UAG.

    (D) Amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} can initiate translation from UAG start codons in five common laboratory E. coli lineages. Normalized expression levels from sfGFP reporter beginning with one of three start codons (AUG, UAG, or GCC) with wild-type tRNAfMet2 or amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression either repressed or induced. Each bar displays the average of three biological replicate measurements. Error bars represent one standard deviation. See figure S3 for amber initiator expression plasmid and fluorescent reporter plasmid maps.

    (E) tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression results in higher UAG versus AUG start codon initiation in K-12, C and B strains versus W and Crooks strains. Each data point displays the average of three biological replicate measurements. Error bars represent one standard deviation. A diagonal line indicates the equal amount of normalized fluorescence from UAG and AUG-initiating reporters.

    (F) tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression results in decreased fitness in Crooks, W, and K-12 strains. All strains harbor the pULTRA::tac-metY(CUA) plasmid. Ratios on the Y- and X- axis were calculated by dividing growth rate and max OD600 from cultures induced with 1 mM IPTG to those repressed with 2% glucose. Each data point is the average of three biological replicates. Error bars represent one standard deviation.

    Introductionlink

    Translation initiation in Escherichia coli relies on recruitment of initiation factors 1-3 (IF1-IF3), the 30S ribosomal subunit, and the initiator tRNA (tRNAfMet) to a messenger RNA (mRNA). Several excellent reviews with more detailed information on the process of bacterial translation initiation have been published[18][19][20].

    E. coli encodes two sequence variants of the initiator tRNA, called tRNAfMet1 and tRNAfMet2. The difference between these sequence variants is one nucleotide at position 46 where tRNAfMet1 is a G (modified to 7mG in tRNA) and tRNAfMet2 is an A[17]. Both the tRNAfMet1 (metZWV genes) and tRNAfMet2 (metY gene) variants are present in the E. coli K-12 strains, but other strains of E. coli lack the tRNAfMet2 and instead have a fourth copy of tRNAfMet1 in the metY loci in their genomes (Fig. 1B). tRNAfMet2 has been introduced previously into E. coli B strains with no apparent deleterious effect[21].

    The amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} (Fig. 1C) was created by Varshney and RajBhandary from a modified E. coli K-12 metY gene encoding a tRNA that can initiate translation at the ‘amber’ stop codon (UAG) instead of the usual AUG start codon[1]. A major activity of biological engineers is the construction of genetic parts and devices that tell the cell when to turn on a gene, resulting in protein production. Predicting whether engineered genetic constructs will work across different organisms is non-trivial and often results in non-functional or incorrectly functioning systems[22]. Differences in basal levels of amber suppression within common K-12 strains is already known[7]. Expressing the amber initiator tRNA shows great promise as a way of programming orthogonal translation initiation at UAG sites in engineered genetic constructs[2], but it is unknown whether tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} will function outside of the K-12 lineage of E. coli.

    Objectivelink

    To enable the adoption of this powerful amber initiator tRNA system to drive programmable gene expression, we wanted to know how transportable it is across E. coli strains? In particular, how would the amber initiator functions in laboratory E. coli strains that normally do not contain the tRNAfMet2 sequence variant?

    Results & Discussionlink

    We began our investigation by building a paired set of plasmids to both express the amber initiator tRNA as well as measure the translation initiation efficiency at AUG and UAG codons. The amber initiator plasmid features a recoded metY gene (metY(CUA)) that expresses tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} . The amber initiator tRNA features an altered anticodon sequence that is the reverse complement of the desired UAG start codon (Fig. 1C). The plasmid contains an IPTG-inducible tac promoter and the Clo DF13 (medium copy) origin of replication expected to function across diverse E. coli strains (Fig. S3)[23]. We designed three reporter plasmids to measure translation initiation events caused by expression of tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} from the amber initiation plasmid. The reporter plasmids use super-folder green fluorescent protein (sfGFP) with three start codon variants: UAG to measure translation initiation efficiency from the amber initiator tRNA, AUG as a positive control, and GCC as a negative control (Fig. S3). The sfGFP reporter is driven by a constitutive T5 promoter and has the ColE1 origin of replication[24], which is compatible with the Clo DF13 origin, so both plasmids can be simultaneously maintained within one E. coli cell.

    We next transformed the amber initiator and reporter plasmids into five common generally recognized as safe (GRAS) laboratory strains of E. coli from a diverse set: MG1655 (K-12 strain), BL21 (B strain), C122 (C strain), ATCC 9637 (W strain), and ATCC 8739 (Crooks strain). All five E. coli strains are designated risk group 1[25] but arise from distinct lineages (Fig. 1A).

    After transforming both amber initiator plasmid and reporter plasmids into all strains, we measured bulk fluorescence under repressing and inducing conditions (Fig. 1D). We observed similar trends in all tested strains. As expected, the reporter strain with sfGFP(GCC) start codon did not appreciably change expression due to tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression. In contrast, the sfGFP(UAG) reporter had fluorescence levels similar to sfGFP(GCC) when under tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} repressed conditions, but increased markedly when tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression was induced (Fig. 1D). On average across the strains we saw an 88-fold increase in fluorescence upon tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression with MG1655 showing the smallest increase and C122 showing the largest increase. As noted previously[1], the combination of lack of expression when repressed, along with a large increase after tRNA induction, could be useful for conditionally expressing toxic proteins due to lack of translational leakage from the UAG stop codon in the absence of the tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} .

    Surprisingly, tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} induction also caused increased fluorescence from the normal sfGFP(AUG) start codon reporter in all strains. We would not have expected the increase because it has been well established that the wild-type tRNAfMet2 initiator does not initiate translation from UAG stop codons[26][27]. We would therefore not expect the reciprocal codon-anticodon pairing between the AUG start codon and the amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} to result in productive translation initiation. We next expressed wild-type metY from the pULTRA plasmid and found no fluorescence increase from the UAG reporter as expected, and a similar magnitude increase in fluorescence from the AUG reporter as was seen with the amber initiator. We speculate that tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} expression has caused an increase in the cellular concentration of ribosomes and this has resulted in increased translation from the sfGFP(AUG) reporter. Support for this idea comes from a recent paper showing evidence that expressing initiator tRNA in E. coli increases mature ribosome levels and accelerates rRNA processing[28] and our recent work showing the amber initiator can increase ribosome protein abundance in cells[2]. We can speculate that expressing tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} has a similar effect on ribosomal maturation and rRNA processing in the cell, considering the presence of identical identity elements (aside from the anticodon sequence) in both initiator tRNAs.

    We next compared the effects of expressing the amber initiator on relative increases of sfGFP(AUG) and sfGFP(UAG) reporter expression to determine the suitability of the strains using the amber initiator (Fig. 1E). The results show the strains are all relatively similar, with E. coli C122 being the best and Crooks being the worst for selectively expressing UAG start codons over AUG initiation.

    To determine the fitness effects of expressing the amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} on the different E. coli strains, we measured growth rate and maximal optical density differences when the tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} transcription was induced versus repressed (Fig. 1F). We observed that the Crooks strain had both its growth rate and maximum optical density moderately reduced when the tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} was induced, while the W and K-12 strains had a major reduction in growth rate but not maximum optical density (Fig. 1F). The B and C strain growth characteristics were unaffected by tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} induction. The patterns of fitness changes following tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} induction seemed to mirror the strain’s relative abilities to produce UAG-initiated reporter protein (Fig. 1E). That is, the B and C strains produced the most UAG reporter fluorescence compared to AUG reporter fluorescence, while the Crooks and W strains produced the least (Fig. 1E). The exception to this pattern is the K-12 strain, which produced the third highest amount of UAG reporter per AUG reporter fluorescence despite the growth rate being reduced by the most (~50%) (Fig. 1E and 1F). The reason for the difference in K-12 may be due to interference between the amber tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} and the genomic copy of wild-type tRNAfMet2 in this strain, which is missing in the other strains examined in this study (Fig. 1B). Further work will need to be done to uncover the reasons for these strain-specific fitness differences.

    In this work, we have shown that the amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} can be expressed in a fully functional form in five different strains of generally regarded as safe laboratory E. coli. Furthermore, the genomic absence of the tRNAfMet2 (metY) gene variant does not seem to have a major bearing on the effectiveness of the amber initiator tRNAfMet2CUA \textrm{tRNA}{fMet2 \atop CUA} function within those strains (B, C, W, and Crooks) (Fig. 1B). Lastly, this work points towards possible uses for the amber initiator tRNA to enable expression of toxic proteins in strains of E. coli, such as BL21, that are often used for the production of protein for structural biology, investigation of protein function, and for therapeutics.

    Methodslink

    Bacterial strains

    The following strains of E. coli were used in this study: MG1655 (K-12 strain), BL21 (B strain, ATCC #BAA-1025)[29][30], C122 (C strain, NCTC122, Public Health England), Crooks (ATCC #8739), and W (ATCC #9637)[31]. All strains were grown in lysogeny broth Miller (LBM) or on LBM agar, supplemented with carbenicillin (Sigma-Aldrich #C9231) and/or spectinomycin (Sigma-Aldrich #S4014).

    Phylogenetic Tree

    A phylogenetic tree of 16 diverse E. coli strains with Escherichia fergusonii as an outgroup was created using the Phylogeny.fr[10] analysis pipeline. Sequences were collected from NCBI GenBank: BL21 (TaKaRa) (NZ_CP010816.1), Crooks ATCC 8739 (NC_010468.1), K-12 substr. W3110 (NC_007779.1), K-12 substr. MG1655 (NC_000913.3), Nissle 1917 (NZ_CP007799.1), W strain ATCC9637 (NC_017635.1), C strain C122 (NZ_LT906474.1), O6:K15:H31 536 (NC_008253.1), 55989 (NC_011748.1), O127:H6 E2348/69 (NC_011601.1), IAI39 (NC_011750.1), O157:H7 str. Sakai (NC_002695.1), UMN026 (NC_011751.1), CFT073 (NC_004431.1), SMS-3-5 (NC_010498.1), Escherichia fergusonii ATCC 35469 (NC_011740.1). MLST analysis of sequences was performed by MLST2.0 run by the Center for Genomic Epidemiology, DTU (https://cge.cbs.dtu.dk/services/MLST/)[32] using the MLST E. coli#1 configuration (adk, fumC, gyrB, icd, mdh, purA, recA genes) and default settings. MLST allele sequences for each of the 16 strains were downloaded, concatenated, and submitted to results from the Phylogeny.fr Phylogeny Analysis pipeline (http://www.phylogeny.fr/). In this pipeline, a MUSCLE multiple sequence alignment of concatenated MLST alleles was performed. The alignment was then processed by Gblocks and phylogeny analysis performed by PhyML with tree drawing by TreeDyn[9][11][11][12][13].

    Plasmid construction

    Reporter plasmids were designed to contain superfolder GFP (sfGFP) with different start codons[33]. A 1971bp gBlock (Integrated DNA Technologies, Inc) was designed and synthesized that contained the metY gene from MG1655 (GenBank: U00096.3) with an anticodon mutation (CAU > CUA). A pULTRA-CNF vector (Addgene #48215)[34] with a Clo DF13 origin of replication and a spectinomycin-resistance cassette was used as the plasmid backbone. The amber initiator plasmid was assembled via in vitro homologous assembly[35]. Cultures for plasmid construction were grown in LBM supplemented with spectinomycin. Plasmids were isolated from transformants, sequence verified, and stored in elution buffer (10 mM Tris-Cl, pH 8.5).

    Culture growth conditions for assay measurements

    Transformants from glycerol stocks were streaked on LBM plates with appropriate antibiotics: spectinomycin (100 µg/mL) for amber initiator plasmid and/or carbenicillin (100 µg/mL) for reporter plasmids and grown overnight at 37°C. Individual colonies were inoculated in 2 mL of LBM containing appropriate antibiotics and grown overnight at 37°C shaking at 200 rpm. After overnight growth, each culture was diluted 1:100 into 400 µL LBM in a 96-well deep well plate containing either 1 mM IPTG to induce, or 2% glucose (w/v) to repress, metY(CUA) expression.

    Fluorescence measurements

    Measurements of fluorescence intensity from the amber initiator plasmid system was performed as before[26] with the following modifications: Absorbance was measured at 600 nm (OD600) to estimate culture density, followed by fluorescence (excitation = 485 nm, emission = 520 nm, bandwidth = 9 nm) measured at a single gain setting (high gain sensitivity = 111), on a PHERAstar FSX (BMG Labtech) plate reader.

    Fitness analysis

    Strains were grown overnight as described above. Cell densities were measured and cells were passaged to a starting OD600 of 0.1 into fresh 200 µL LBM with appropriate antibiotics and either 1 mM IPTG or 2% glucose (w/v). Cultures were grown in a flat-bottom 96-well plate sealed with gas-permeable seal (Sigma-Aldrich # Z763624) at 37°C (300 rpm) while culture light scattering (OD600) over time was measured on a SPECTROstar NANO (BMG Labtech) plate reader at 5 min intervals for 10 h. Our analysis method was adapted from previous work[36] and used to compare ratios of growth rate and maximal cell density for individual strains.

    The growth rate and maximal cell density (max OD600) were calculated using the R package Growthcurver v.0.2.1[37]. Ratios of growth rates and maximal cell densities were determined for strains with the amber initiation plasmid under induced growth condition (1 mM IPTG) versus repressed growth condition (2% glucose (w/v)).

    Whole Genome Alignment

    The following nucleotide sequences from GenBank were analyzed using Mauve whole genome alignment[15][16] performed with default parameters: K-12 substr. MG1655 (NC_000913.3), BL21 (NZ_CP010816.1), C strain C122 (NZ_LT906474.1), W strain ATCC9637 (NC_017635.1), and Crooks ATCC 8739 (NC_010468.1).

    metY Multiple Sequence Alignment

    Gene sequences were extracted from GenBank genome sequences and aligned using MUSCLE algorithm[11] using default parameters.

    Funding Statementlink

    PRJ was supported by the Molecular Sciences Department, Faculty of Science & Engineering, and the Deputy Vice-Chancellor (Research) of Macquarie University. RMV is a recipient of the Macquarie University Research Excellence PhD scholarship (MQRES). PFY was supported by a summer research scholarship from the Molecular Sciences Department of Macquarie University.

    Conflict of interestlink

    The authors declare no conflicts of interest.

    Ethics Statementlink

    This study was completed in accordance with Macquarie University biosafety committee approval, application ID 5201600567.

    No fraudulence is committed in performing these experiments or during processing of the data. We understand that in the case of fraudulence, the study can be retracted by ScienceMatters.

    Referenceslink
    1. U Varshney, U L Rajbhandary
      Initiation of protein synthesis from a termination codon.
      Proceedings of the National Academy of Sciences, 87/1990, pages 1586-1590 DOI: 10.1073/pnas.87.4.1586chrome_reader_mode
    2. Russel M. Vincent, Bradley W. Wright, Paul R. Jaschke
      Measuring Amber Initiator tRNA Orthogonality in a Genomically Recoded Organism
      ACS Synthetic Biology, 8/2019, pages 675-685 DOI: 10.1021/acssynbio.9b00021chrome_reader_mode
    3. Jonathan M. Monk, Anna Koza, Miguel A. Campodonico, Daniel Machado, Jose Miguel Seoane, Bernhard O. Palsson, Markus J. Herrgård, Adam M. Feist
      Multi-omics Quantification of Species Variation of Escherichia coli Links Molecular Features with Strain Phenotypes
      Cell Systems, 3/2016, pages 238-251.e12 DOI: 10.1016/j.cels.2016.08.013chrome_reader_mode
    4. Karoline Marisch, Karl Bayer, Monika Cserjan-Puschmann, Markus Luchner, Gerald Striedner
      Evaluation of three industrial Escherichia coli strains in fed-batch cultivations during high-level SOD protein production
      Microbial Cell Factories, 12/2013, page 58 DOI: 10.1186/1475-2859-12-58chrome_reader_mode
    5. Douglas McCloskey, Julia Xu, Lars Schrübbers,more_horiz, Markus J. Herrgård
      RapidRIP quantifies the intracellular metabolome of 7 industrial strains of E. coli
      Metabolic Engineering, 47/2018, pages 383-392 DOI: 10.1016/j.ymben.2018.04.009chrome_reader_mode
    6. Francesca Ceroni, Alice Boo, Simone Furini, Thomas E Gorochowski, Olivier Borkowski, Yaseen N Ladak, Ali R Awan, Charlie Gilbert, Guy-Bart Stan, Tom Ellis
      Burden-driven feedback control of gene expression
      Nature Methods, 15/2018, pages 387-393 DOI: 10.1038/nmeth.4635chrome_reader_mode
    7. Chang C Liu, Susan E Cellitti, Bernhard H Geierstanger, Peter G Schultz
      Efficient expression of tyrosine-sulfated proteins in E. coli using an expanded genetic code
      Nature Protocols, 4/2009, pages 1784-1789 DOI: 10.1038/nprot.2009.188chrome_reader_mode
    8. Fredrik Ronquist, John P. Huelsenbeck
      MrBayes 3: Bayesian phylogenetic inference under mixed models
      Bioinformatics, 19/2003, pages 1572-1574 DOI: 10.1093/bioinformatics/btg180chrome_reader_mode
    9. Alexis Dereeper, Stephane Audic, Jean-Michel Claverie, Guillaume Blanc
      BLAST-EXPLORER helps you building datasets for phylogenetic analysis
      BMC Evolutionary Biology, 10/2010, page 8 DOI: 10.1186/1471-2148-10-8chrome_reader_mode
    10. A. Dereeper, V. Guignon, G. Blanc, S. Audic, S. Buffet, F. Chevenet, J.-F. Dufayard, S. Guindon, V. Lefort, M. Lescot, J.-M. Claverie, O. Gascuel
      Phylogeny.fr: robust phylogenetic analysis for the non-specialist
      Nucleic Acids Research, 36/2008, pages W465-W469 DOI: 10.1093/nar/gkn180chrome_reader_mode
    11. Robert C. Edgar
      MUSCLE: multiple sequence alignment with high accuracy and high throughput
      Nucleic Acids Research, 32/2004, pages 1792-1797 DOI: 10.1093/nar/gkh340chrome_reader_mode
    12. J. Castresana
      Selection of Conserved Blocks from Multiple Alignments for Their Use in Phylogenetic Analysis
      Molecular Biology and Evolution, 17/2000, pages 540-552 DOI: 10.1093/oxfordjournals.molbev.a026334chrome_reader_mode
    13. Stéphane Guindon, Olivier Gascuel
      A Simple, Fast, and Accurate Algorithm to Estimate Large Phylogenies by Maximum Likelihood
      Systematic Biology, 52/2003, pages 696-704 DOI: 10.1080/10635150390235520chrome_reader_mode
    14. Maria Anisimova, Olivier Gascuel
      Approximate Likelihood-Ratio Test for Branches: A Fast, Accurate, and Powerful Alternative
      Systematic Biology, 55/2006, pages 539-552 DOI: 10.1080/10635150600755453chrome_reader_mode
    15. Aaron C.E. Darling, Bob Mau, Frederick R. Blattner, Nicole T. Perna
      Mauve: Multiple Alignment of Conserved Genomic Sequence With Rearrangements
      Genome Research, 14/2004, pages 1394-1403 DOI: 10.1101/gr.2289704chrome_reader_mode
    16. David J Edwards, Kathryn E Holt
      Beginner’s guide to comparative bacterial genome analysis using next-generation sequence data
      Microbial Informatics and Experimentation, 3/2013, page 2 DOI: 10.1186/2042-5783-3-2chrome_reader_mode
    17. Gautam Das, T. K. Dineshkumar, Swapna Thanedar, Umesh Varshney
      Acquisition of a stable mutation in metY allows efficient initiation from an amber codon in Escherichia coli
      Microbiology, 151/2005, pages 1741-1750 DOI: 10.1099/mic.0.27915-0chrome_reader_mode
    18. Brian Søgaard Laursen, Hans Peter Sørensen, Kim Kusk Mortensen, Hans Uffe Sperling-Petersen
      Initiation of Protein Synthesis in Bacteria
      Microbiology and Molecular Biology Reviews, 69/2005, pages 101-123 DOI: 10.1128/mmbr.69.1.101-123.2005chrome_reader_mode
    19. Jennifer Shepherd, Michael Ibba
      Bacterial transfer RNAs
      FEMS Microbiology Reviews, 39/2015, pages 280-300 DOI: 10.1093/femsre/fuv004chrome_reader_mode
    20. Pohl Milón, Marina V. Rodnina
      Kinetic control of translation initiation in bacteria
      Critical Reviews in Biochemistry and Molecular Biology, 47/2012, pages 334-348 DOI: 10.3109/10409238.2012.678284chrome_reader_mode
    21. N Mandal, U L Rajbhandary
      Escherichia coli B lacks one of the two initiator tRNA species present in E. coli K-12.
      Journal of Bacteriology, 174/1992, pages 7827-7830 DOI: 10.1128/jb.174.23.7827-7830.1992chrome_reader_mode
    22. Felix Moser, Nicolette J. Broers, Sybe Hartmans, Alvin Tamsir, Richard Kerkman, Johannes A. Roubos, Roel Bovenberg, Christopher A. Voigt
      Genetic Circuit Performance under Conditions Relevant for Industrial Bioreactors
      ACS Synthetic Biology, 1/2012, pages 555-564 DOI: 10.1021/sb3000832chrome_reader_mode
    23. A. R. Stuitje, E. Veltkamp, H. J. J. Nijkamp, P. J. Weijers
      Origin and direction of replication of the bacteriocinogenic plasmid Clo DF13
      Nucleic Acids Research, 6/1979, pages 71-80 DOI: 10.1093/nar/6.1.71chrome_reader_mode
    24. Eduard Veltkamp, Antoine R. Stuitje
      Replication and structure of the bacteriocinogenic plasmids Clo DF13 and Col E1
    25. Gregory E. Sims, Sung-Hou Kim
      Whole-genome phylogeny of Escherichia coli/Shigella group by feature frequency profiles (FFPs)
      Proceedings of the National Academy of Sciences, 108/2011, pages 8329-8334 DOI: 10.1073/pnas.1105168108chrome_reader_mode
    26. Ariel Hecht, Jeff Glasgow, Paul R. Jaschke, Lukmaan A. Bawazer, Matthew S. Munson, Jennifer R. Cochran, Drew Endy, Marc Salit
      Measurements of translation initiation from all 64 codons in E. coli
      Nucleic Acids Research, 45/2017, pages 3615-3626 DOI: 10.1093/nar/gkx070chrome_reader_mode
    27. Elad Firnberg, Jason W. Labonte, Jeffrey J. Gray, Marc Ostermeier
      A Comprehensive, High-Resolution Map of a Gene’s Fitness Landscape
      Molecular Biology and Evolution, 31/2014, pages 1581-1592 DOI: 10.1093/molbev/msu081chrome_reader_mode
    28. Sunil Shetty, Umesh Varshney
      An evolutionarily conserved element in initiator tRNAs prompts ultimate steps in ribosome maturation
      Proceedings of the National Academy of Sciences, 113/2016, pages E6126-E6134 DOI: 10.1073/pnas.1609550113chrome_reader_mode
    29. Haeyoung Jeong, Hyun Ju Kim, Sang Jun Lee
      Complete Genome Sequence of Escherichia coli Strain BL21
      Genome Announcements, 3/2015, pages e00134-15 DOI: 10.1128/genomea.00134-15chrome_reader_mode
    30. William B. Wood
      Host specificity of DNA produced by Escherichia coli: Bacterial mutations affecting the restriction and modification of DNA
      Journal of Molecular Biology, 16/1966, pages 118-133, IN3 DOI: 10.1016/s0022-2836(66)80267-xchrome_reader_mode
    31. Colin T Archer, Jihyun F Kim, Haeyoung Jeong, Jin Hwan Park, Claudia E Vickers, Sang Yup Lee, Lars K Nielsen
      The genome sequence of E. coli W (ATCC 9637): comparative genome analysis and an improved genome-scale reconstruction of E. coli
      BMC Genomics, 12/2011, page 9 DOI: 10.1186/1471-2164-12-9chrome_reader_mode
    32. Mette V. Larsen, Salvatore Cosentino, Simon Rasmussen, Carsten Friis, Henrik Hasman, Rasmus Lykke Marvig, Lars Jelsbak, Thomas Sicheritz-Pontén, David W. Ussery, Frank M. Aarestrup, Ole Lund
      Multilocus Sequence Typing of Total-Genome-Sequenced Bacteria
      Journal of Clinical Microbiology, 50/2012, pages 1355-1361 DOI: 10.1128/jcm.06094-11chrome_reader_mode
    33. Jean-Denis Pédelacq, Stéphanie Cabantous, Timothy Tran, Thomas C Terwilliger, Geoffrey S Waldo
      Engineering and characterization of a superfolder green fluorescent protein
      Nature Biotechnology, 24/2005, pages 79-88 DOI: 10.1038/nbt1172chrome_reader_mode
    34. Kathryn C. Schultz, Lubica Supekova, Youngha Ryu, Jianming Xie, Roshan Perera, Peter G. Schultz
      A Genetically Encoded Infrared Probe
      Journal of the American Chemical Society, 128/2006, pages 13984-13985 DOI: 10.1021/ja0636690chrome_reader_mode
    35. Daniel G Gibson, Lei Young, Ray-Yuan Chuang, J Craig Venter, Clyde A Hutchison Iii, Hamilton O Smith
      Enzymatic assembly of DNA molecules up to several hundred kilobases
      Nature Methods, 6/2009, pages 343-345 DOI: 10.1038/nmeth.1318chrome_reader_mode
    36. Marc J. Lajoie, Alexis J. Rovner, Daniel B. Goodman, Hans-Rudolf Aerni, Adrian D. Haimovich, Gleb Kuznetsov, Jaron A. Mercer, Harris H. Wang, Peter A. Carr, Joshua A. Mosberg, Nadin Rohland, Peter G. Schultz, Joseph M. Jacobson, Jesse Rinehart, George M. Church, Farren J. Isaacs
      Genomically Recoded Organisms Expand Biological Functions
      Science, 342/2013, pages 357-360 DOI: 10.1126/science.1241459chrome_reader_mode
    37. Kathleen Sprouffske, Andreas Wagner
      Growthcurver: an R package for obtaining interpretable metrics from microbial growth curves
      BMC Bioinformatics, 17/2016, page 172 DOI: 10.1186/s12859-016-1016-7chrome_reader_mode
    Commentslink

    Create a Matters account to leave a comment.