ABSTRACT
Mistranslation occurs when an amino acid not specified by the standard genetic code is incorporated during translation. Since the ribosome does not read the amino acid, tRNA variants aminoacylated with a non-cognate amino acid or containing a non-cognate anticodon dramatically increase the frequency of mistranslation. In a systematic genetic analysis, we identified a suppression interaction between tRNASerUGG, G26A, which mistranslates proline codons by inserting serine, and eco1-1, a temperature sensitive allele of the gene encoding an acetyltransferase required for sister chromatid cohesion. The suppression was partial with a tRNA that inserts alanine at proline codons and not apparent for a tRNA that inserts serine at arginine codons. Sequencing of the eco1-1 allele revealed a mutation that would convert the highly conserved serine 213 within β7 of the GCN5-related N-acetyltransferase core to proline. Mutation of P213 in eco1-1 back to the wild-type serine restored function of the enzyme at elevated temperature. Our results indicate the utility of mistranslating tRNA variants to identify functionally relevant mutations and identify eco1 as a reporter for mistranslation. We propose that mistranslation could be used as a tool to treat genetic disease.
INTRODUCTION
Mistranslation occurs when an amino acid that differs from that specified by the standard genetic code is incorporated into nascent proteins during translation. Mistranslation naturally occurs at a frequency of approximately one in ten thousand codons in all cells and increases under specific environmental conditions or upon mutation of the translational machinery (Santos et al. 1999; Bacher et al. 2007; Ling et al. 2007; Kramer and Farabaugh 2007; Drummond and Wilke 2009; Javid et al. 2014). Contrary to Crick’s Frozen Accident Theory (Crick 1968), mistranslation is tolerated at levels approaching 8% (Berg et al. 2019). Accurate translation has two major components. The first is tRNA aminoacylation, catalyzed by aminoacyl-tRNA synthetases (aaRS) that specifically couple an amino acid to the 3’ end of their cognate tRNA(s) [reviewed in Pang et al. (2014)]. The second specificity step is codon decoding at the ribosome, which relies on base pairing between codon and anticodon. Loss of fidelity at either step can lead to mistranslation. Mistranslation is dramatically increased by tRNA variants that are inaccurately aminoacylated or contain mutations within their anticodon (Geslain et al. 2010; Hoffman et al. 2017; Lant et al. 2017; Berg et al. 2017; Zimmerman et al. 2018). Nucleotides in the tRNA that are recognized by a specific aaRS are called identity elements and consist of single nucleotides, nucleotide pairs, and structural motifs (de Duve 1988; Hou and Schimmel 1988; Giegé et al. 1998). Since the anticodon links an amino acid to its codon assignment, it is not surprising that tRNA recognition often involves identity elements within the anticodon. However, for tRNASer and tRNAAla the anticodon plays no role in the specificity of charging in yeast and for tRNALeu the anticodon only plays a minor role (Giegé et al. 1998), making these tRNAs particularly amenable to engineering for increased mistranslation.
Studies demonstrating the ability of tRNA variants to correct genetic errors by replacing a non-functional residue with a functionally competent residue predate the deciphering of the genetic code (Crawford and Yanofsky 1959; Stadler and Yanofsky 1959; Yanofsky and Crawford 1959). Intergenic suppressors that change codon meaning were called informational suppressors, since they alter the information flow from DNA to protein (Gorini and Beckwith 1966). Some of the early studies included a demonstration by Benzer and Champe (1962) of the suppression of nonsense mutations. They reasoned that suppression acts by changing the genetic code to add a new sense codon. Mutations in the Escherichia coli tryptophan synthase (trpA) gene provided a selection to identify mistranslation inducing mutations that rescued growth in tryptophan free media. These studies characterize tRNA variants that led to Gly insertion at Arg or Cys codons (Carbon et al. 1966; Jones et al. 1966; Gupta and Khorana 1966). Other suppressor mistranslating tRNAs have been identified in yeast. For example, Goodman et al. (1977) mapped a tRNATyr with a G to T transversion mutation resulting in a U at the wobble nucleotide position making it a nonsense suppressor.
Previously, we engineered serine tRNA variants that mis-incorporate serine at proline codons by replacing the UGA anticodon with UGG (Berg et al. 2017, 2019). These tRNAs contain secondary mutations to reduce tRNA functionality and modulate mistranslation levels since plasmids expressing a tRNA with this anticodon change alone can not be transformed into yeast. One variant, tRNASerUGG, G26A, contains a G26A secondary mutation and when expressed from a centromeric plasmid results in a frequency of serine incorporation at proline codons of ~ 5% as determined by mass spectrometry (Berg et al. 2019). Zimmerman et al. (2018) and Geslain et al. (2010) have also demonstrated the possibility of mistranslating serine for a number of amino acids in yeast and mammalian cells.
In this report we use a mistranslating tRNA variant to identify the causative mutation in the Saccharomyces cerevisiae eco1-1 allele as a serine to proline missense mutation. Eco1 is an acetyltransferase required for sister chromatid cohesion during DNA replication (Tóth et al. 1999; Unal et al. 2007; Ben-shahar et al. 2008). Mutations in the human homolog of ECO1 (ESCO2) cause Roberts syndrome (Vega et al. 2005), a rare genetic disorder characterized by limb reduction and craniofacial abnormalities. We demonstrate the utility of mistranslation as a tool to identify causative mutations, demonstrate eco1 as a selectable reporter to monitor mistranslation and provide support for the possibility of using mistranslation as a tool to cure genetic disease.
MATERIALS AND METHODS
Yeast strains and DNA constructs
The SGA starter strain, Y7092 (MATa can1Δ::STE2pr-SpHIS5 lyp1Δ his3Δ1 leu2Δ0 ura3Δ0 met15Δ0), was a kind gift from Dr. Brenda Andrews (University of Toronto). Strains from the temperature sensitive collection are derived from the wild-type MATa haploid yeast strain BY4741 (MATa his3Δ0 leu2Δ0 met15Δ0 ura3Δ0; Winzeler and Davis 1997) and described in Costanzo et al. (2016). CY8613 (MATα HO::natNT2-SUP17UGG, G26A can1Δ::STE2pr-SpHIS5 lyp1Δ) contains a gene encoding tRNASerUGG, G26A integrated into Y7092 at the HO locus as described below. The isogenic control strain lacking the mistranslating tRNA encoding gene is CY8611 (MATα HO::natNT2 can1Δ::STE2pr-SpHIS5 lyp1Δ).
Yeast strains were grown in yeast peptone media containing 2% glucose (YPD) or synthetic media supplemented with nitrogenous bases and amino acids (SD) at the temperature indicated. Plates lacking uracil (-URA) were made in SD media supplemented with 0.6% (g/vol) casamino acids, 0.25% adenine and 0.5% tryptophan. The temperature sensitive collection was maintained in 1536-format on YPD plates containing 200 mg/L geneticin (G418; Invitrogen). The SGA query strain was maintained on YPD plates containing 100 mg/L nourseothricin-dihydrogen sulfate (NAT; Werner BioAgents). Double mutants containing both the SGA query and temperature sensitive mutation were maintained on synthetic dropout plates lacking histidine, arginine and lysine with monosodium glutamate (1 g/L) as the nitrogen source, canavanine (50 mg/L), thialysine (50 mg/L), G418 (200 mg/L) and NAT (100 mg/L).
Centromeric plasmids expressing tRNASer (pCB3076), tRNASerUGG, G26A (pCB4023), tRNAProUGG G3:U70 (pCB2948) and tRNASerUCU, G26A (pCB4301) are described in Berg et al. (2017), Hoffman et al. (2017) and Berg et al. (2019).
Construction of SGA Query and Control Strains
The protocol for constructing the SGA query strains was adapted from the PCR-mediated gene deletion method of Tong et al. (2001). A DNA fragment containing 200 bp of upstream HO flanking region and 200 bp of the HO gene was synthesized by Life Technologies and cloned into pGEM®-T Easy (Promega Corp.) as a NotI fragment (pCB4386). The natNT2 marker from pFA6a-natNT2 was PCR amplified using primers UK9789/UK9790 (Table 1) and cloned into pCB4386 as an EcoRI fragment to generate the control SGA integrating vector (pCB4394). The gene encoding tRNASerUGG, G26A was PCR amplified from pCB4023 (Berg et al. 2017) using primers UG5953/VB2609 and inserted as a HindIII fragment into pCB4394 to generate pCB4397. pCB4394 and pCB4397 were digested with NotI, transformed into Y7092 and transformants selected on 100 mg/L NAT to generate the SGA strains CY8611 and CY8613, respectively. Integration of the fragments was verified by PCR.
Synthetic genetic array analysis and validation
The SGA assay was performed as described by Tong et al. (2001). The SGA control and query strains (CY8611 and CY8613) were mated to a temperature-sensitive collection (Ben-Aroya et al. 2008; Li et al. 2011; Kofoed et al. 2015; Costanzo et al. 2016) arrayed in quadruplicate 1536-array format on YPD plates using a BioMatrix robot (S&P Robotics Inc.). Mated strains were grown overnight then pinned onto YPD + NAT/G418 plates to select for diploids. Haploids were generated by pinning the diploid strains onto enriched sporulation plates and incubating for 1 week at 22°C. The haploids then underwent three rounds of selection for double mutants that had both the tRNA mutation and temperature-sensitive allele. First, strains were pinned on SD – His/Arg/Lys + canavanine/thialysine plates to select for MATa haploids. Next, colonies were pinned twice onto (SD/MSG) - His/Arg/Lys + canavanine/thialysine/G418/NAT to select double mutants. Colonies were incubated for two days between pinnings at room temperature. The double mutants were grown at 30°C for 5 days. Images from day 3 were analyzed and scored using SGAtools (Wagih et al. 2013). An SGA score ≥ 0.5 and a p-value ≤ 0.05 was used to identify alleles that were potentially suppressed by tRNAProUGG, G26A. Suppression was validated by transforming URA3 centromeric plasmids expressing tRNAProUGG, G26A (pCB4023) or wild-type tRNASer (pCB3076; Berg et al. 2017) into each strain and comparing growth on plates lacking uracil.
Isolation of eco1-1 and mutagenesis
Genomic DNA was isolated from the temperature sensitive eco1-1 strain as described by Hoffman and Winston (1987). eco1-1 was PCR amplified using primers YA9871/YA9872. The PCR included Q5® High-Fidelity DNA Polymerase (New England Biolabs) to minimize incorporation errors. The PCR product was sequenced with primer YA9872 and cloned into pGEM®-T Easy (pCB4639) where it was re-sequenced with M13 forward and reverse primers. Pro213 was mutagenized to Ser213 by two-step PCR mutagenesis. Primer pairs YA9871/YB1043 and YA9872/YB1042 were used in the first round followed by outside primers YA9871 and YA9872. The PCR product was subcloned into pGEM®-T Easy and cloned as an EcoRI fragment into YCplac33 to give pCB4673. Similarly, eco1-1 was subcloned into YCplac33 to give pCB4662.
Modeling
The protein sequence for Saccharomyces cerevisiae Eco1 (SGD ID: S000001923) was aligned to Homo sapiens ESCO1 (Uniprot: Q5FWF5-1) and ESCO2 (Uniprot: Q56NI9-1), Mus musculus Esco1 (Uniprot: Q69Z69) and Esco2 (Uniprot: Q8CIB9), Danio rerio esco1 (Uniprot: X1WEK0) and esco2 (Uniprot: Q5SPR8) and Drosophila melanogaster eco (Uniprot: Q9VS50) using Clustal Omega (Madeira et al. 2019). The S213P mutation, corresponding to S770, was modelled on the human ESCO1 structure (PDB: 4MXE; Kouznetsova et al. 2016) using Missense3D (Ittisoponpisan et al. 2019).
RESULTS
We performed an SGA screen to identify genes demonstrating genetic interactions with tRNASerUGG, G26A, a tRNA variant that mistranslates serine at proline codons, using a temperature sensitive collection containing 1016 temperature sensitive alleles (Kofoed et al. 2015; Costanzo et al. 2016). The tRNA encoding gene was integrated at the HOanalysis demonstrates the utility of mistranslation to locus and selected for by NAT resistance. The control strain contained the natNT2 marker integrated at HO, but no tRNA. Screens were performed at 30°C and analyzed using SGAtools (Wagih et al. 2013). The negative genetic interactions identified will be described elsewhere. In this screen, the eco1-1 strain had a genetic interaction score of 0.52 (P = 1.0 x 10-5) suggesting it grew better than expected in the presence of tRNASerUGG, G26A. To validate the positive genetic interaction, centromeric plasmids carrying the gene encoding tRNASerUGG, G26A (pCB4023) or wild-type tRNASer (pCB3076) were transformed into the temperature sensitive strain and its growth compared on plates lacking uracil at 24°C, 30°C and 37°C. As shown in Figure 1A, tRNASerUGG, G26A improved growth of the eco1-1 strain at 30°C and 37°C. Note that the partial toxicity of the mistranslating tRNA (tRNASerUGG, G26A) is apparent when the cells are grown at 24°C.
The suppression could arise from a mistranslation event, effectively reverting the mutation or from a positive interaction between eco1-1 and mistranslation in general. To distinguish between these possibilities, we transformed the eco1-1 strain with plasmids expressing tRNASer, tRNASerUGG, G26A, tRNAProG3:U70 (which inserts alanine at proline codons) and tRNASerUCU, G26A (which inserts serine at arginine codons) (Figure 1B). Partial suppression was seen with tRNAProG3:U70. No suppression was seen with tRNASerUCU, G26A or with wild-type tRNASer. We conclude that the tRNA suppresses the allele by mistranslation at a proline codon, rather than through a more general genetic interaction between eco1-1 allele and mistranslation and that misincorporation of serine at proline codons results in stronger suppression of eco1-1 than does misincorporation of alanine.
Based on the suppression by tRNASerUGG, G26A (substitutes serine at proline codons) and to a lesser extent tRNAProUGG, G3:U70 (substitutes alanine at proline codons) we predicted that the eco1-1 strain in the collection contains a mutation resulting in the conversion of a serine residue to proline that was contributing to the temperature sensitive phenotype. We isolated the eco1-1 gene, including up- and down-stream flanking sequence by PCR and cloned the product. The clone was sequenced through the gene. We identified four missense mutations, G184D, S213P, K260R and G273D, and one synonymous mutation (Table 2). The missense mutation altering S213 to proline converts the UCG codon for serine to CCG for proline. We have previously shown that CCG is mistranslated to serine by tRNASerUGG, G26A (Berg et al. 2019). As an indication that the serine to proline mutation was not a clonal artifact, the PCR product was directly sequenced with the 3’ oligonucleotide, which confirmed the presence of the T to C mutation at nucleotide 637 of eco1-1. We note that Tóth et al. (1999) characterized a temperature sensitive allele of eco1 with a mutation converting glycine 211 to aspartic acid that they named eco1-1. It is questionable whether this is the allele in the collection since their study was performed with a W303 strain background.
To analyze which of the four missense mutations (G184D, S213P, K260R or G273D) resulted in changes to important regions of the protein we performed an alignment of Eco1 homologs from yeast, human, mouse, fruit fly and zebrafish. Of these only S213 is in a highly conserved region of the protein (Figure 2A). Furthermore, analysis by SIFT (Sim et al. 2012) suggests that any change of S213 would be detrimental to function. Eco1 encodes a histone acetyltransferase required for chromatid cohesion during DNA replication. Mutations in the human gene (ESCO2) cause Roberts syndrome (Vega et al. 2005). The structure of ESCO2 has been determined (Kouznetsova et al. 2016; Rivera-Colón et al. 2016). S770, the equivalent of yeast S213, is found in β7 strand of the GCN5-related N-acetyltransferase core (Figure 2B). Though not essential for function, mutation of S770 (yeast S213) to alanine reduces catalytic efficiency ~8-fold in vitro, with a minimal effect on thermal stability (Rivera-Colón et al. 2016). Modeling of a proline at this position 213 suggests that this would distort the β7 strand perhaps altering the stability of the protein (Figure 2C).
To determine if the S213P mutation resulted in the temperature sensitive nature of eco1-1, we constructed the allele where only P213 was converted back to TCG for serine. eco1-1 and eco1-1P213S alleles were inserted into a URA3 containing centromeric plasmid and transformed into the eco1-1 strain. The ability of this eco1-1P213S allele, the original eco1-1 allele (carrying four mutations) and the plasmid alone to complement the temperature sensitive nature of the eco1-1 strain was tested by analyzing growth at 24°C and 37°C (Figure 3). At 24°C the eco1-1 strain grows well with or without an additional copy of the eco1 allele. At 37°C an additional eco1-1 allele (with P213) allowed for slightly better growth than empty plasmid (Figure 3). In contrast addition of the wild-type-like S213 eco1-1 allele (eco1-1P213S) restored robust growth of the strain at 37°C. This result confirms that mutation of wild-type serine at 213 to proline is the cause of the temperature sensitive nature of the eco1-1 allele and that mistranslation of the proline codon in eco1-1 to insert serine would be sufficient to allow growth at elevated temperature.
DISCUSSION
Our analysis demonstrates the utility of mistranslation to identify the nature of deleterious mutations. Suppression of the eco1-1 temperature sensitive phenotype with tRNASerUGG, G26A, which causes serine mistranslation at proline codons (Berg et al. 2017), and to a lesser extent tRNAProG3:U70, which causes alanine mistranslation at proline codons (Hoffman et al. 2017), suggested that a missense mutation creating a proline codon was involved. Upon sequencing of the eco1-1 allele from the temperature sensitive collection, we found a missense mutation converting S213 to proline. This mutation is in a highly conserved region of the acetyltransferase domain. The contribution of the S213P mutation to the temperature sensitivity of eco1-1 was demonstrated by converting it back to a serine codon. In the context of the other mutations in the eco1-1 allele, the reversion was sufficient to complement the temperature sensitive phenotype.
Aminoacylation of the tRNAs for serine and alanine and to a lesser extent leucine do not depend on the anticodon in yeast. This allows the construction of tRNAs that insert these amino acids at non-cognate codons. Many of these tRNAs have been engineered (for example see Geslain et al. 2010; Zimmerman et al. 2018; Berg et al. 2019). Through analysis similar to that performed here, these mistranslating tRNAs will allow the identification of functionally significant missense mutations in alanine, serine and leucine codons. The one caveat to the method is that mistranslation frequency has a threshold of approximately 8% in yeast due to introducing proteotoxic stress. The protein in question must therefore show significant function when present at relatively low levels. Other examples of proteins that function at such reduced levels are the proline isomerase Ess1 (Gemmill et al. 2005) and the cochaperone Tti2 (Hoffman et al. 2016, 2017).
If one assumes that all mutations have equally likelihood of creating a temperature sensitive allele, the maximum number of temperature sensitive strains in a collection that contain a causative serine to proline mutation can be estimated from the frequency of UCA (0.89%) and UCG (0.85%) serine codons. With a single base change, each of these codons becomes efficiently decoded by the UGG anticodon. The combined usage frequency of UCA and UCG predicts that ~18 of the 1016 strains, contain a serine to proline mutation. As we identified one strain suppressed by tRNASerUGG, G26A in the collection, it suggests that for ~ 6% (1/18) of essential yeast proteins, 5% of their native expression (the extent of proline to serine mistranslation) is sufficient to support viability.
Functional assays that result in phenotypic reversion are one of the simplest methods to screen for mistranslation events. We previously used a leucine to proline mutation in TTI2 to identify mistranslating tRNA variants (Hoffman et al. 2017; Berg et al. 2017). The utility of using tti2 to detect different varieties of mistranslation caused by tRNA variants is somewhat limited because our screening has only revealed leucine to proline mutations to have phenotypic consequences when found in isolation. eco1-1 may provide a more versatile reporter in that the structure is known and altering the acetyltransferase domain would be expected to impact structure or function in a way that results in phenotypic change. We note in particular the conservation of the residues flanking S213 (Figure 2A), making those codons candidates for further reporter engineering.
Many diseases result from missense or nonsense mutations that alter the structure, function and/or stability of a gene product. Our findings complement existing reports documenting the ability of tRNA variants to restore protein function in yeast and bacteria through their ability to mistranslate a deleterious codon (for examples, see reviews by Celis and Piper 1981 and Murgola 1985). Effectively, phenotypic suppression by mutant tRNAs is evidence that mistranslation is able to cure genetic disease. The mistranslation could be achieved through tRNA variants, reducing specificity of aminoacyl-tRNA-synthetases, or decreasing proofreading functions. In contrast to the concerns addressed by Crick in his “Frozen Accident Theory” (Crick 1968), mistranslation is not catastrophic for cell viability. At levels in the 3-5% range mistranslation has minimal affect on yeast cell viability (Hoffman et al. 2017; Lant et al. 2017; Berg et al. 2019). For some key cellular proteins, 3-5% mistranslation is sufficient to restore viability.
FUNDING
This work was supported from the Natural Sciences and Engineering Research Council of Canada [RGPIN-2015-04394 to C.J.B.], the Canadian Institutes of Health Research [FDN-159913 to G.W.B.] and generous donations from Graham Wright and James Robertson to M.D.B. M.D.B. holds an NSERC Alexander Graham Bell Canada Graduate Scholarship (CGS-D).
ACKNOWLEDGMENTS
We thank Julie Genereaux for technical assistance and for comments on the manuscript.