Control of dinucleoside polyphosphates by the FHIT-homologous HNT2 gene, adenine biosynthesis and heat shock in Saccharomyces cerevisiae

Background

The FHIT gene is lost early in the development of many tumors. Fhit possesses intrinsic ApppA hydrolase activity though ApppA cleavage is not required for tumor suppression. Because a mutant form of Fhit that is functional in tumor suppression and defective in catalysis binds ApppA well, it was hypothesized that Fhit-substrate complexes are the active, signaling form of Fhit. Which substrates are most important for Fhit signaling remain unknown.

Results

Here we demonstrate that dinucleoside polyphosphate levels increase 500-fold to hundreds of micromolar in strains devoid of the Saccharomyces cerevisiae homolog of Fhit, Hnt2. Accumulation of dinucleoside polyphosphates is reversed by re-expression of Hnt2 and is active site-dependent. Dinucleoside polyphosphate levels depend on an intact adenine biosynthetic pathway and time in liquid culture, and are induced by heat shock to greater than 0.1 millimolar even in Hnt2+ cells.

Conclusions

The data indicate that Hnt2 hydrolyzes both ApppN and AppppN in vivo and that, in heat-shocked, adenine prototrophic yeast strains, dinucleoside polyphosphates accumulate to levels in which they may saturate Hnt2.

The human FHIT gene, located at the chromosome 3 fragile site FRA3B, is inactivated early in the development of many tumors [1]. Murine Fhit is also located at a fragile site [2, 3] and mice heterozygous for disruption of Fhit, given low intragastric doses of the mutagen N-nitrosomethylbenzylamine, develop stomach and sebaceous tumors [4] that can be prevented by viral Fhit expression [5]. Fhit, a dimer of 147 amino acid subunits, is a member of the histidine triad (HIT) superfamily of nucleotide hydrolases and transferases [6, 7]. Members of the Hint branch of the HIT superfamily are found in all forms of life [8]. The S. cerevisiae Hint homolog, Hnt1, and rabbit Hint possess adenosine monophosphoramidase activity that functions in yeast to positively regulate function of Kin28, Ccl1 and Tfb3, which constitute the kinase component of general transcription factor TFIIH [9]. A new Hint related protein, Aprataxin, is mutated in individuals with ataxia with oculomotor apraxia [10, 11] and has a yeast homolog termed Hnt3 [9]. Members of the Fhit branch of the HIT superfamily have been found in fungi [12, 13], animals [2, 14, 15] and plants [7] and hydrolyze diadenosine tetraphosphate, diadenosine triphosphate and other 5'-5"'-dinucleoside polyphosphates. The middle histidine of the histidine triad (His96 in human Fhit), which is critical for hydrolysis of ApppA by Fhit [14, 16], is not necessary for tumor suppression [17, 18]. Nonetheless, wild-type and His96Asn forms of Fhit are saturated by ApppA in the low micromolar range and form stable complexes with non-hydrolyzable ApppA in which two ApppA analogs are bound per Fhit dimer and all phosphates cluster on one surface of the protein [16]. These observations suggested that Fhit-substrate complexes may be the active, signaling form of Fhit and that the function of the catalytically essential histidine may be to terminate the lifetime of signaling complexes [16].

Given that neither transcriptional nor post-transcriptional regulation has been reported for Fhit protein, the level of biological activity of Fhit may be controlled by levels of Fhit substrates, inhibitors, and proteins that interact with Fhit-nucleotide complexes. Fhit proteins from humans [19] and worms [15] bind ApppA and AppppA with K_m values of 2 to 3 μM. Human Fhit [14] and the S. cerevisiae Fhit homolog [13], which was called Aph1 but is here termed Hnt2 under nomenclature aproved by the Saccharomyces Genome Database, cleave ApppA more readily while Aph1, the S. pombe homolog, cleaves AppppA more readily [20]. Consistent with the ApppA hydrolase activity of purified Fhit protein, most cancer cell lines that are Fhit negative at the protein level have higher levels of ApppA than cell lines that are Fhit positive [21]. Nonetheless, the actual concentrations of dinucleoside polyphosphates were submicromolar in every cell culture sample [21] and thus, under the reported culture conditions, the measured dinucleoside polyphosphates would not be expected to occupy the Fhit active site substantially [19]. Dinucleoside polyphosphate levels were measured in adenine-requiring S. cerevisiae strains before or after disruption of the Fhit-homologous HNT2 gene [13] and in adenine-requiring S. pombe strains as a function of disruption and overexpression of the Fhit-homologous aph1 gene [22]. Recently, it was observed that diadenosine polyphosphates undergo a divalent cation-dependent conformational change that might mediate their biosynthesis, catabolism or signaling properties [23].

Here we discover a requirement of adenine biosynthesis for high-level dinucleoside polyphosphate accumulation in the absence of the Fhit homolog in S. cerevisiae. By constructing active site mutants of Hnt2 that were expressed in yeast, we demonstrate that ApppN and, to a lesser degree, AppppN levels are controlled by the Hnt2 active site. An added benefit of these constructions is the availability of yeast strains that possess high levels of dinucleoside polyphosphates and at the same time express a mutant Fhit-homologous protein, because these are conditions which have been postulated to constitute the signaling form of Fhit [16]. Finally, using controlled genotypes we revisited conditions that lead to increased accumulation of dinucleoside polyphosphates [24–33]. Recognizing that hnt2 deletion is a pathological condition, we were particular interested in identifying conditions that lead to accumulation of such compounds in cells that contain a functional HNT2 gene, rather than simply identifying conditions that produce diadenosine polyphosphate accumulation in the absence of Hnt2. While cells without a functional HNT2 gene accumulate dinucleoside polyphosphates in excess of 10 μM in a variety of nonstressed and stressed conditions, 46°C heat shock was the only condition that produced dinucleoside polyphosphate accumulation in excess of 10 μM in cells containing a functional HNT2 gene. These conditions did not render the cells conditionally null for Hnt2 because cells expressing HNT2 continued to limit dinucleoside polyphosphate accumulation during hours of heat shock, though at levels of ~0.1 mM.

Recently, discovery that the Hint-homologous HNT1 gene is required for high temperature growth on galactose and observations that alleles of cak1, kin28, ccl1 and tfb3 are hypersensitive to loss of Hnt1 enzyme activity provided evidence that Hnt1 enzyme activity positively regulates Kin28 function, particularly on galactose media [9]. Though phenotypic consequences of hnt2 mutations have yet to be discovered, our observations suggest that synthetic lethal interactions with hnt2 mutations are likely to be found in adenine prototrophic strains undergoing heat shock.

Disruption of HNT2 and tetrad analysis of dinucleoside polyphosphate levels

An earlier report demonstrated that disruption of HNT2 was tolerated by haploid yeast strains without an effect on growth and that ApppN and AppppN accumulate 30 and 3-fold, respectively on account of the hnt2 deletion [13]. Because those data were obtained by random spore analysis, we considered it important to test whether elevated dinucleoside polyphosphate levels co-segregate with hnt2 disruption in all tetrads examined and whether any other commonly used genetic markers affect dinucleoside polyphosphate levels. Diploid strain BY71 (Table 1), created to be heterozygous for MAT, ADE2, HIS3, LEU2, LYS2, MET15, TRP1, URA3 and HNT2, was allowed to sporulate and was then dissected. As shown in Figure 1, tetrads produced four viable colonies, two of which were large and white, and two of which were smaller and pink on YPD medium, which scored as ade- on SDC -ade medium. Markers segregated 2:2 in nearly all cases and possession of a 1976 bp PCR fragment using primers 4726 and 4722 always correlated with geneticin-resistance while possession of a 1200 bp product correlated with geneticin-sensitivity, as expected for segregants containing a nondisrupted HNT2 gene.

Disruption of hnt2. Strain BY16 was crossed with strain BY4717 to generate diploid strain BY71, which was dissected to generate haploid progeny. Markers were scored for mating type, auxotrophic requirements and G418-resistance. The two small segregants per tetrad are ade2Δ mutants. Here, four segregants that had been scored for G418-resistance were scored for size of the HNT2 locus by PCR using diagnostic primers 4722 and 4726. G418-sensitive progeny, BY71-3a and BY71-3c produced a product of 1200 bp while G418-resistant progeny BY71-3b and BY71-3d produced a product of 1976 bp, demonstrating physical linkage of kanMX2 to hnt2 disruption.

Full size image

Haploid segregants from three complete tetrads were cultured for 24, 48 and 72 hours in SDC medium, lysed, and levels of ApppN and AppppN were determined. As shown in Table 2, strains containing an intact HNT2 gene were never observed to have calculated intracellular ApppN levels above 3 μM and typically were found to have ApppN levels below 1 μM. Strains disrupted for hnt2 had ApppN levels of approximately 6 to 43 μM after one day of culture, rising to approximately 30 to 300 μM and 50 to 350 μM after two and three days of culture, respectively. The ade2 mutation was partially epistatic to the effect of hnt2 disruption on ApppN accumulation. hnt2Δ strains containing ade2 mutations were several fold lower in ApppN accumulation than hnt2ΔADE2 isolates. Thus, deletion of hnt2 afforded a 48-fold increase in ApppN in ade2 mutants, consistent with an earlier report of a 31-fold effect [13], but a 211-fold increase in ADE2 strains.

Earlier, hnt2 deletion was reported to increase AppppN levels only 2.5-fold but the study was performed in ade2 mutants [13]. Consistent with that report, the three hnt2 ade2 strains showed only a 2-fold higher AppppN level than the three HNT2 ade2 strains, when nucleotide levels were averaged across the three time points. In contrast, hnt2ΔADE2 strains achieved a 3.7-fold higher level of AppppN than HNT2 ADE2 strains. Larger increases in AppppN concentrations have been seen with disruption of Apa1 and Apa2, the diadenosine tetraphosphate phosphorylases in S. cerevisiae[34, 35], indicating that they have a more significant role in controlling AppppN levels than does Hnt2. In the case of disruption of the Fhit and Hnt2-homologous aph1 gene in S. pombe, which encodes an enzyme relatively specific for a AppppA [20], a 290-fold increase in AppppA concentration was observed [22]. Our data indicate that Hnt2 hydrolyzes ApppN and AppppN in vivo in budding yeast and that an intact adenine biosynthetic pathway is required for high-level synthesis and accumulation of adenylylated dinucleoside polyphosphates.

Hnt2 active site-dependence of dinucleoside polyphosphate accumulation

Catalytic activity of the Fhit ApppA hydrolase depends on His96 [14, 16]. To test whether the active site of Hnt2 is necessary to control dinucleoside polyphosphate accumulation in vivo, wild-type and mutant alleles of HNT2 that differ at His109, corresponding to human Fhit His96, were expressed from the HNT2 promoter on plasmids (Table 3) in hnt2ΔADE2 strain BY71-6c. As shown in Table 4, reintroduction of wild-type HNT2 produced a 40 to 125-fold reduction in intracellular concentrations of ApppN and a two to seven-fold reduction in levels of AppppN. This reduction was active-site dependent: adding back multicopy HNT2 with the nucleophilic histidine replaced by alanine or aspartate reduced dinucleoside polyphosphate levels less than two-fold.

Dinucleoside polyphosphate levels may not be limited by the levels of lysyl tRNA synthetase

AppppA is induced by heat shock in bacteria [36] and the induction of AppppA was thought to be a function of the heat-shock inducible LysU lysyl tRNA synthetase. However, deletion of lysU had no effect on heat-shock inducible AppppA accumulation [37]. The KRS1 gene [38] encoding cytosolic lysyl tRNA synthetase was cloned into multicopy plasmid pRS423 [39] to generate plasmid pM1. Strains BY71-16d (ADE2 HNT2) and BY71-6c (ADE2 hnt2) were transformed with pM1 and the pRS423 control plasmid, and cultures were harvested at 24, 48 and 72 hours. Determination of ApppN and AppppN concentrations revealed that ApppN levels are substantially higher in hnt2 mutants than in isogenic wild-types at all time points and that plasmids conferring multiple copies of KRS1 did not increase ApppN or AppppN levels at any culture time point (Table 5). To address whether plasmid pM1 indeed increased lysyl tRNA synthetase activity, lysates from pRS423 and pM1-transformed BY71-6c were assayed for incorporation of ³H lysine into yeast tRNA. As shown in Figure 2, tRNA-dependent lysine incorporation was increased 2.1-fold by expression of KRS1 from a multicopy plasmid.

Moderate overexpression of lysyl tRNA synthetase activity via multicopy plasmid pM1 Lysates from strain BY71-6c transformants with control plasmid pRS423 and multicopy KRS1 plasmid pM1 were assayed for incorporation of ³H lysine in the absence and presence of added yeast tRNA.

Full size image

Heat shock is the most effective stress for elevation of dinucleoside polyphosphates

In the wild and in the laboratory, yeast are exposed to stresses such as hypo-osmotic or hyperosmotic conditions, toxic cations, heat shock and cell-cycle disruptive reagents. To test whether such conditions induce dinucleoside polyphosphates in hnt2- or Hnt2+ cells, we incubated ADE2 hnt2 and ADE2 HNT2 cells in water, 1 M sorbitol, 2 mM CdCl₂, 46 °C heat shock, 10 mM caffeine, or in rich media for two hours and determined dinucleoside polyphosphate levels. Additionally, to test whether moderate overexpression of the lysyl tRNA synthetase gene affected accumulation, we compared control transformants to multicopy KRS1 transformants of the two strains. As shown in Table 6, the hnt2 samples had substantially higher ApppN levels than HNT2 samples under all conditions. Among the hnt2 samples, only heat shocked samples showed evidence of ApppN levels higher than the levels in nonstressed hnt2 cells. Similarly, among the HNT2 samples, the heat shocked samples showed increased ApppN levels compared with control-treated cells while CdCl₂ and other treated samples showed no significant changes. KRS1 on a multicopy plasmid showed no significant alteration of ApppN levels in any sample. As with other experiments, AppppN levels were lower than ApppN levels in all cases. Heat shock was the best inducer of AppppN. Hypotonic, hypertonic and caffeine treated media produced no increase in AppppN (not shown).

To further investigate the kinetics of heat shock and cadmium-induction of ApppN and AppppN levels, we transformed hnt2ΔADE2 strain BY71-6c with multicopy plasmids containing no HNT2 gene, the wild-type HNT2 gene, or the HNT2-His109Ala or HNT2-His109Asp alleles of HNT2. Cultures were exposed to either 2 mM CdCl₂ or 46°C heat shock and intracellular concentrations of ApppN and AppppN were determined at 30-minute timepoints.

As shown in Figure 3, ApppN and AppppN levels are higher in hnt2 strains than in cells with a functional HNT2 gene, and were not significantly elevated by 2 mM CdCl₂. However, when cells were heat shocked, as shown in Figure 4, ApppN and AppppN levels increased substantially in cultures with every HNT2 genotype (absence, presence or active site mutation). Increases in dinucleoside polyphosphates in Hnt2+ cultures cannot be attributed to thermal inactivation of Hnt2 because the presence of HNT2 plasmids continues to reduce incremental increases in dinucleoside polyphosphates even in the fourth hour of the heat shock. Furthermore, the active-site mutant alleles of HNT2, HNT2-His109Ala and HNT2-His109Asp, provided on plasmids pB32 and pB86, demonstrated intermediate abilities to control dinucleoside polyphosphate levels, suggesting that elevated temperature reduces the catalytic defects of these mutant enzymes. The high levels of ApppN (~100 μM) and AppppN (~10 μM) and the fact that Hnt2-containing samples continue to reduce the rate of increase in dinucleoside polyphosphates without reducing their concentrations demonstrate that heat shock induces dinucleoside polyphosphate synthesis and that Hnt2 is saturated under such conditions. In Xenopus oocytes, however, some work has suggested that heat shock-dependent accumulation of AppppN is largely due to inactivation of degradative enzymes [30].

Cadmium is a poor inducer of dinucleoside polyphosphates Calculated intracellular concentrations of ApppN and AppppN of transformants of strain BY71-6c as a function of time in 2 mM CdCl₂.

Full size image

Heat shock induces dinucleoside polyphosphates Calculated intracellular concentrations of ApppN and AppppN of transformants of strain BY71-6c as a function of time at 46°C.

Full size image

It had been reported that Hnt2 controls ApppN levels in vivo, with a minor effect on AppppN [13]. Here we show that the ade2 mutation present in earlier experiments prevents accumulation of ApppN and AppppN and reduced the magnitude of the Hnt2 effect. In ADE2 strains examined herein, deletion of HNT2 increased levels of ApppN and AppppN by factors of approximately 200 and 4, respectively. Mutagenesis [14], X-ray crystallography [16, 40], and stereochemical analysis [41] indicate that His96 is the nucleophile that attacks the α-phosphate of Fhit substrates. Our analysis shows that the corresponding residue, His109 of Hnt2, is required for hydrolysis of ApppN and AppppN substrates in vivo.

Other than effects on the concentrations of intracellular nucleotides, neither deletion of HNT2 nor mutation of His109 of HNT2 had phenotypic consequences. In the case of the Hint-homologous HNT1 gene of S. cerevisiae, the phenotype of the single mutant was mild and the biological pathway, positive regulation of Kin28, the yeast homolog of Cdk7, was revealed by synthetic lethal interactions between hnt1 and hypomorphic alleles of cak1, kin28, ccl1 and tfb3[9]. Because backup systems to limit problems with hnt2 mutant cells apparently exist, synthetic lethal interactions may be critical to identify the Hnt2 biological pathway. Thus, if the major function of HNT2 is simply control of dinucleoside polyphosphates lest their accumulation inhibit other enzymes (i.e., McLennan's "foe" hypothesis [42]), then synthetic lethal phenotypes ought to be sought with hnt2 mutant strains that are wild-type for the adenine biosynthetic pathway, potentially under heat shock conditions. On the other hand, if the major function of Hnt2 depends on formation of an enzyme-substrate complex [16], then synthetic lethal interactions ought to be sought in heat shocked ADE2 HNT2-His109Ala strains. In fact, because ADE2 HNT2-His109Ala strains have high levels of dinucleoside polyphosphates and an Hnt2 polypeptide, such a strain may be sensitized to secondary mutations, whether dinucleoside polyphosphates are friend or foe [42].

General molecular biology

Yeast media and procedures were as described [43, 44]. S. cerevisiae transformations were carried out by the lithium acetate method [45]. E. coli strain XL-1 Blue was used for bacterial cloning and plasmid amplification. Bacterial media and molecular biology techniques were as described [44].

Disruption of hnt2

A 1570 bp DNA fragment containing an hnt2Δ::kanMX2 disruption cassette was generated as described [46]. Primers 4716 (5'GAAGCTCCATTGATCTATCTTGGGCTCAGAATGATCTTAAGCAAAACAAAGCTTCGTACGCTGCAG) and 4717 (5'CGTAAGTATGAATCTATTATTTATTGAACTATAGTGTTAAACCAGGGCCACTAGTGGATCTGA) were used to amplify the yeast expressible geneticin-resistance gene from pFA6a-kanMX2[46] with 50 bp DNA ends corresponding to sequences upstream and downstream of HNT2. The resulting fragment was transformed into haploid S. cerevisiae strain BY4727, and transformants were selected on YPD with 400 μg/ml geneticin. To ensure that geneticin-resistance was linked to hnt2 disruption, genomic DNA from several geneticin-resistant colonies was analyzed by diagnostic PCR primers 4726 (5'TCGCTGATTTGGTAGTCTC) and 4722 (5'GAGTCTCCTCGAGGAAAG). A transformant shown to contain the 1976 bp hnt2Δ::kanMX2 product in place of the wild-type 1200 bp HNT2 product was named strain BY16. Strain BY16 was crossed with strain BY4717 [47] to generate diploid strain BY71. Strain BY71 was allowed to sporulate and subjected to tetrad dissection to generate haploid strains BY71-1a through BY71-16d. Genotypes of all yeast strains are provided in Table 1.

Plasmid constructions

The HNT2 gene was amplified from genomic DNA of yeast strain SEY6210 [48] with primers PB1 (5'GCAGCGGATCCTTGGGAT) that spanned a Bam HI site upstream of the promoter and PB2 (5'GAGTCTCCTCGAGGAAAG) that spanned a Xho I site downstream of the terminator. The 1316 bp Bam HI-Xho I fragment containing HNT2 was ligated to Bam HI and Xho I-cleaved plasmid pRS423 [39] to generate plasmid pB05. Plasmids pB32 and pB86, carrying H109A and H109D alleles of HNT2, were constructed by site-directed mutagenesis [49] of plasmid pB05 using primers PB3 (5'ATAATGTGTGTAGCCAAGTGGGGT) and PB4 (5'TAATGTGTGTATCCAAGTGGGGTAC). The S. cerevisiae gene encoding lysyl tRNA synthetase (KRS1) was amplified as a 2.9 kbp genomic fragment from strain SEY6210 using primers MR20 (5'CGAGCTCGGTTGGA TGACTTTAAAATGACTAAGTTTGTAGTATCCTCTTTGC ATACTC) and MR21 (5'TCCCCCGGGGGAGCTCCTTTAGGGCTACCGAACATAAACAAATTTAGGTAA TGAGTTTTC). This product, cloned into the Sma I restriction site of plasmid pRS423, generated plasmid pM1 in which KRS1 is oriented anti to HIS3. Plasmids are summarized in Table 3.

Measurement of dinucleoside polyphosphate levels

Twelve haploid segregants, pregrown in liquid SDC medium, were inoculated into 250 ml of SDC at starting density of 10⁴ cells per ml. At 24, 48 and 72 hours of growth, 50 ml of cells were harvested, cells counted microscopically, lysed, and levels of AppppN and ApppN were determined as described [13]. Intracellular concentrations of ApppN and AppppN were calculated using 7 × 10^-14 l as the volume of a haploid cell [50]. Cultures of BY71-6c were transformed with plasmids pRS423, pB05, pB32 and pB86 and transformants were selected on SDC-his media. Intracellular concentrations of ApppN and AppppN were determined for transformants from cultures in SDC-his media as above. To determine whether a multicopy lysyl tRNA synthetase plasmid affected accumulation of dinucleoside polyphosphates, we transformed HNT2 ADE2 strain BY71-16d and hnt2ΔADE2 strain BY71-6c with control plasmid pRS423 and with plasmid pM1. Transformants were grown for 24, 48 and 72 hours in SDC-his media and intracellular dinucleoside polyphosphate concentrations were determined as above. To survey dinucleoside polyphosphate induction as a function of potential stress conditions, strain BY71-6c was transformed with either pRS423 or pM1 (effectively hnt2Δ and hnt2Δ YEpKRS1, respectively) and strain BY71-16d was transformed with the same plasmids (effectively HNT2Δ and HNT2Δ YEpKRS1, respectively). After 48 hours of culture, cells were pelleted and resuspended in either YPD media, water, SDC-his media, or the same media supplemented with 2 mM CdCl₂, 10 mM caffeine or 1 M sorbitol. The SDC-his sample was incubated for 2 hours at 46°C while other samples were incubated for 2 hours at room temperature prior to extraction for determination of dinucleoside polyphosphate concentrations. To determine the time course of ApppN and AppppN levels as a function of stresses, we used strains BY71-16d and BY71-6c transformed with pRS423, pB05, pB32, or pB86. Transformants were cultured for 48 hours, treated with 2 mM CdCl₂ or 46°C heat shock, and then harvested for nucleotide quantitation at 30 minute intervals. Experiments presented in Tables 2, 4, 5 and 6 were performed two, three, five and three times, respectively. Experiments presented in Figures 3 and 4 were performed four times each. Because diadenosine polyphosphate levels vary with time in culture, generating a higher or lower range of values in independently conducted experiments, we did not average values obtained in separate experiments. For the data presented in Table 6, triplicate cultures were prepared and the ApppN levels for identically treated samples are provided as averages ± standard deviations.

Lysyl tRNA synthetase activity assay

Strain BY71-6c was transformed with plasmids pRS423 and pM1 (multicopy KRS1) and cultures were grown as for measurement of dinucleoside polyphosphate levels. Lysates were prepared by glass bead disruption in 50 mM Tris Cl pH 7.5, 1 mM DTT, 40% glycerol. Incorporation of tritiated lysine into tRNA was measured by modification of the protocol of Hou [51]. Reactions contained 10 micrograms of total protein in 20 mM KCl, 10 mM MgCl₂, 4 mM dithiothreitol, 2 mM ATP, 50 mM Na HEPES pH 7.5, 20 μM lysine (10 μCi ³H lysine), and were performed with or without 15 μg yeast tRNA (Sigma) in a total volume of 60 μl. At one, two, three, five and twelve minute time points, 10 μl aliquots were spotted on filter paper and placed in 5% wt/vol trichloroacetic acid, washed in 5% trichloroacetic acid twice, washed twice in 95% ethanol, rinsed in ether and scintillation-counted.

Authors' note

Tetrads shown in figure 1 were dissected and PCR analyzed to show the procedure followed to obtain strains. They do not correspond to tetrads mentioned in table 1.

This work was supported by a grant from the National Cancer Institute (CA75954). We thank Dr. Ya-Ming Hou and her laboratory for guidance in measuring lysyl tRNA synthetase activity and an anonymous reviewer for helpful suggestions.

Affiliations

1. Structural Biology & Bioinformatics Program, Kimmel Cancer Center, Philadelphia, Pennsylvania, USA

   • Marta Rubio-Texeira
   • , James M Varnum
   • , Pawel Bieganowski
   •  & Charles Brenner

2. Department of Biology, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA

   • Marta Rubio-Texeira

Corresponding author

Correspondence to Charles Brenner.

Authors' contributions

Dr. Bieganowski constructed most of the initial plasmids and strains for this study. Drs. Rubio-Texeira and Dr. Varnum performed the experiments described in this work in consultation with Dr. Brenner, who proposed the experimental questions. Drs. Rubio-Texeira, Varnum and Brenner wrote the manuscript, which was read and approved by each author.

Reprints and Permissions