rights / license: research collection in copyright - non ...6644/eth... · vanessa anissa nathalie...
TRANSCRIPT
Research Collection
Doctoral Thesis
The effect of the URM1 pathway on translation by thiolation ofspecific tRNAs
Author(s): Rezgui, Vanessa Anissa Nathalie
Publication Date: 2012
Permanent Link: https://doi.org/10.3929/ethz-a-007624233
Rights / License: In Copyright - Non-Commercial Use Permitted
This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.
ETH Library
DISS. ETH NO. 20870
The effect of the URM1 pathway on translation
by thiolation of specific tRNAs
A dissertation submitted to
ETH ZURICH
for the degree of
Doctor of Sciences
presented by
Vanessa Anissa Nathalie Rezgui
M.Sc. University of Geneva
born on April, 28th 1981
citizen of Switzerland (Zurich)
accepted on the recommendation of
Prof. Matthias Peter – examiner
Prof. André Gerber – co-examiner
Prof. Paola Picotti – co-examiner
2012
1
TABLE OF CONTENTS SUMMARY ........................................................................................................ 4
RESUMÉ ............................................................................................................ 5
1 Introduction. ............................................................................................... 6
1.1 Translation............................................................................................................................................ 6 1.2 Translation regulation....................................................................................................................... 7 1.3 Regulation of tRNAs ........................................................................................................................ 8 1.3.1 tRNA abundance ...........................................................................................................................9 1.3.2 tRNA aminoacylation levels.................................................................................................. 11 1.3.3 tRNA localization ...................................................................................................................... 11
1.4 tRNA modifications ........................................................................................................................11 1.4.1 Wobble base modifications ................................................................................................... 13 1.4.2 The mcm5s2U34 modification................................................................................................. 14
1.5 The Ubiquitin-related modifier 1 ................................................................................................16 1.6 Phenotypes associated with lack of wobble uridine modification ...................................17 1.7 Aim of the work................................................................................................................................19
2 RESULTS......................................................................................................20
2.1 General translation is unaffected by lack of URM1 .............................................................20 2.2 URM1 is important for efficient expression of AAA, CAA, and GAA-rich proteins
23 2.3 Differential expression of AAA, CAA, and GAA-rich genes is not due to differential
transcription .........................................................................................................................................................28 2.4 The protein expression changes induced by lack of URM1 are small............................30 2.5 URM1 is required for efficient translation of CMS1 and YPL199C................................31 2.6 Thiolation is required for codon-specific translation ...........................................................33 2.6.1 Generation of an inducible dual-fluorescent reporter .............................................. 33 2.6.2 URM1 is required for expression of AAA, CAA, and GAA enriched reporters ... 36 2.6.3 Reporters with the G-ending codons are less affected by lack of thiolation.... 39 2.6.4 The effect of URM1 on translation is dependent on codon frequency ................ 41 2.6.5 Addition of various drugs does not increase the urm1∆ effect on the
translation reporters ..................................................................................................................................... 41 2.7 Thiolation enhances A-site binding and peptide bond formation ....................................44 2.8 URM1-deleted cells are sensitive to paromomycin ..............................................................47
2
2.9 Activation of galactose promoter is defective in urm1∆ cells ..........................................48
3 DISCUSSION................................................................................................51
3.1 Uracil34 modifications stabilizes tRNAs at the A-‐site .....................................................51 3.2 Consequences on translation fidelity....................................................................................51 3.3 Effects of lack of thiolation on expression of AAA, CAA, and GAA rich mRNAs .52 3.4 Effects of lack of thiolation on expression of AAG, CAG, and GAG rich proteins .53 3.5 Codon context ...................................................................................................................................54 3.6 The role of URM1 in rich conditions.........................................................................................55 3.7 URM1 involvement in stress regulation ...................................................................................55 3.7.1 Thiolation and tRNA levels.................................................................................................... 55 3.7.2 Regulation of tRNA modification levels........................................................................... 57
3.8 Influence of tRNA modifications on each other ....................................................................58 3.9 Identification of further regulated proteins.......................................................................59 3.10 Functions of tRNA modifications beyond translation.......................................................60 3.11 Other functions of Urm1 .............................................................................................................61 3.12 Concluding remarks ......................................................................................................................62
4 Materials and Methods...............................................................................64
4.1 [35S] metabolic labeling..................................................................................................................64 4.2 Ribosome extraction .......................................................................................................................64 4.3 Polysome profiles.............................................................................................................................65 4.4 Hot phenol extraction .....................................................................................................................65 4.5 Gene ontology (GO) enrichment analysis ...............................................................................65 4.6 Quantitative RT-PCR .....................................................................................................................66 4.7 Cloning of the dual-fluorescent translation reporter ............................................................66 4.8 Expression of fluorescent reporter .............................................................................................66 4.9 Cycloheximide chase ......................................................................................................................67 4.10 Western blotting .............................................................................................................................67 4.11 Galactose induction.......................................................................................................................67 4.12 Drug sensitivity assay ..................................................................................................................67 4.13 SILAC labeling...............................................................................................................................68 4.14 Protein Extraction and digestion for MS...........................................................................68 4.15 Strong cation exchange (SCX) fractionation....................................................................68 4.16 Peptide purification and iso-‐electric focusing................................................................68 4.17 LC-‐MS/MS.......................................................................................................................................69 4.18 Protein identification and quantitation.............................................................................69
3
4.19 Data normalization and statistical analysis of differential abundance................70 4.20 Random Forrest Analysis ........................................................................................................70 4.21 Total tRNA preparation............................................................................................................70 4.22 Lysine tRNA synthetase cloning and purification.........................................................71 4.23 Aminoacylation and purification of [14C]Lys-‐tK ............................................................71 4.24 Biochemical and kinetic assays.............................................................................................71
5 References..................................................................................................75
6 Abbreviations .............................................................................................82
7 Curriculum vitae .........................................................................................84
8 Acknowledgments ......................................................................................86
Summary
4
SUMMARY
During protein synthesis the genetic information encoded by the cell is read by the
ribosome to produce proteins. Efficiency and fidelity of this process is essential to
promote cell growth and viability. tRNAs are the key molecules that match codons
with correct amino acids. The URM1 pathway specifically thiolates the uridine at the
wobble position of tKUUU, tQUUG, and tEUUC and is required for resistance to various
stresses such as nutrient starvation and oxidative agents. Previous in vitro studies
suggested that uridine thiolation is important to enhance recognition of lysine,
glutamine and glutamic acid codons.
In this study, we used a combination of in vivo and in vitro approaches to analyze
the effect of lack of wobble uridine thiolation of specific tRNAs at the molecular and
cellular level. We show that URM1 is important for efficient translation of genes
enriched for AAA, CAA, and GAA codons. Moreover we show that this is mediated
by increased binding to the ribosomal A-site and peptide bond formation. Further we
found that URM1 is required for resistance to paromomycin and activation of the
galactose metabolism pathway, suggesting a role in translation fidelity and carbon
metabolism.
Together these data show that tRNA modification at the wobble position
modulates the decoding capacity of specific tRNAs to control protein expression. We
suggest a model in which tRNA modifications at the anticodon loop regulate protein
expression in a codon-specific manner when tRNA levels are limiting.
Resumé
5
RESUMÉ Durant la synthèse des protéines, l’information génétique encodée par la cellule est
traduite pour la production de protéines. L’efficacité et la fidélité de ce processus sont
essentielles à promouvoir la croissance et la viabilité des cellules. L’ARNt est la
molécule clé permettant de correspondre un codon avec un acide aminé spécifique.
URM1 et ses protéines associées sont responsables pour la thiolation de l’uridine à la
position wobble de tKUUU, tQUUG, and tEUUC et sont nécessaires à la résistance à des
situations de stress variées, tels que la carence en nutriments et les agents oxydatifs.
De précédentes études in vitro, ont impliquées la thiolation dans amélioration de la
reconnaissance des codons pour la lysine, la glutamine et l’acide glutamique.
Dans cette étude, nous avons utilisé une combinaison d’approches in vivo et in
vitro pour analyser les conséquences de la perte de la thiolation de l’uridine wobble
de certains ARNt au niveau cellulaire et moléculaire. Notre étude démontre qu’URM1
est important pour une traduction efficace de gènes enrichis en codons AAA, CAA et
GAA. En outre, cette étude montre que cet effet est due à une augmentation de la
liaison de l’ARNt au site ribosomal A et à une augmentation de la formation de
dipeptides. De plus, nous avons trouvé qu’URM1 est nécessaire à la résistance à la
paromomycine et à l’activation de la voie du métabolisme au galactose, suggérant un
rôle d’URM1 dans la fidélité du processus de traduction et du métabolisme des
sources de carbone.
Ensemble ces données montrent que les modifications d’ARNt à la position
wobble ajustent la capacité de décodage d’ARNt spécifiques pour contrôler
l’expression de protéines. Nous proposons donc un modèle dans lequel les
modifications d’ARNt dans l’anticodon régulent l’expression protéinique de manière
codon spécifique lorsque les niveaux d’ARNt sont limitants.
Introduction
6
1 INTRODUCTION.
1.1 Translation
Faithful protein synthesis is fundamental to ensure that functional proteins are
produced and therefore guarantees cell viability. During protein synthesis the
ribosome reads the genetic information encoded by the mRNA and translates it into a
polypeptide chain. This tightly regulated process that relies on the recognition of a
unit of three nucleotides, the codon, and its translation into the corresponding amino
acid is done by the ribosome.
The eukaryotic 80S ribosome is constituted of a small, 40S, and a large, 60S,
subunit. The mRNA bound to the small subunit is read by the transfer RNAs (tRNAs)
in the large subunit, where peptide bond formation happens and the growing
polypeptide chain is released through the exit tunnel. tRNAs are the adaptor
molecules matching codons with specific amino-acids (aa). The ribosome contains
three adjacent tRNA binding sites: the Aminoacyl-tRNA site (A-site), the Peptidyl-
tRNA site (P-site), and the Exit site (E-site). A tRNA molecule proceeds from the A
to the P site during amino-acid incorporation and is subsequently released at the E site
(Figure 1).
Figure 1: Translation elongation.
Scheme depicting the ribosome with the three tRNA binding regions: the aminoacyl-tRNA
site (A), the peptidyl-tRNA site (P), and the exit site (A). The right panel shows how
aminoacylated tRNA molecules enter the ribosome at the A site, proceed to the P site where
Introduction
7
transfer of the polypeptidyl chaine takes place and then leave the ribosome at the E-site, after
transfer of the growing amino-acid chain to the next tRNA.
Translation can be divided into three main steps: initiation, elongation and
termination. In eukaryotes, the first step, initiation, is the recognition of the mRNA by
the preinitiation complex (PIC) composed of the small ribosomal subunit, the initiator
Methionine tRNA (fMet in prokaryotes), eEF1A (the homologue of the prokaryotic
EF-Tu), and initiation factors. During this process the PIC scans the mRNA for the
first AUG codon and subsequently recruits the large ribosomal subunit to start
elongation of the polypeptide chain. Elongation, the translation of codons into the
corresponding amino-acids is itself subdivided into several steps: initial selection,
proofreading, accommodation, peptidyl-transfer, and finally translocation. Elongation
starts with entry of the ternary complex (TC) constituted of aminoacylated-tRNA (aa-
tRNA) and eEF1A bound to GTP, in the A site of the ribosome. There the tRNA is
tested for codon-anticodon pairing. The GTP hydrolysis by eEF1A is the proofreading
step ensuring that the correct tRNA is kept and incorrect tRNAs rejected from the
ribosome. Correct codon-anticodon pairing induces conformational changes
triggering GTP hydrolysis by eEF1A. This accommodation step is followed by
peptidy-transfer where the nascent chain is covalently bound to the amino-acid of the
newly accommodated tRNA. In the last step of elongation the elongation factor eEF2
(the homologue of the prokaryotic EF-G) induces GTP hydrolysis and subsequently
the translocation of the peptidyl-tRNA to the P site. The A site is then free for the
next aa-tRNA. Termination is the last step, where the stop codon is recognized by a
release factor instead of a tRNA and triggers the release of the polypeptide chain from
the terminal peptidyl-tRNA and the dissociation of the ribosome from the mRNA.
1.2 Translation regulation A large fraction of the energy of growing cells is devoted to translation. A
eukaryotic cell produces 20 ribosomes per second and contains around 300’000
ribosomes for 60’000 mRNAs (Phizicky and Hopper 2010). A ribosome incorporates
on average 2 amino-acids per seconds and several ribosomes bind simultaneously on
each mRNA. However this is only an average estimate as translation speed is not
uniform along an mRNA. Initiation constitutes the rate-limiting step (Milon,
Introduction
8
Konevega et al. 2008; El Yacoubi, Bailly et al. 2012), then a ramp of low efficiency
translation is found on the first 30-50 codons to avoid ribosomal jamming on the
mRNA (Tuller, Carmi et al. 2010), followed by faster elongation and termination. In
general translation elongation rates are limited by the codon-anticodon recognition.
The ribosome needs to differentiate between a cognate tRNA, which form three
correct base-pairing with the codons, and a near-cognate tRNA containing a single
pair mismatch. This process is dependent on tRNA concentration and affinity of the
tRNA to the ribosome and the codon at the A-site. For example, some proteins
contain natural pausing sites at rare codons, which are often important for protein
folding (Kimchi-Sarfaty, Oh et al. 2007; Zhang, Hubalewska et al. 2009). The
concentration of tRNA recognizing such rare codons is very low and therefore the
ribosome pauses until it finds the appropriate tRNA. tRNA concentration can
therefore regulate general and gene-specific translation levels.
1.3 Regulation of tRNAs To keep up with the high demand for proteins of the cell, tRNA molecules are very
abundant. tRNA genes are mostly found in multiple copies and the nuclear yeast
genome contains 274 tRNA genes. The copy number for each tRNA varies greatly
between different tRNA species. In yeast, gene copy number ranges from 1 to 16
depending on the tRNA and the copy number correlates well with abundance of the
corresponding tRNA (Percudani, Pavesi et al. 1997). tRNA genes encodes for short
RNAs of 73 to 93 nucleotides adopting a cloverleaf secondary structure and a L-
shaped tertiary structure (Figure 2). tRNAs are composed of five main regions. The
acceptor stem composed of the 5’ end and the 3’ CCA end, which is aminoacylated
by specific aminoacyl-synthetases, the D-loop, the T-loop, the variable loop and the
anticodon loop (figure). The anticodon loop contains the anticodon, which recognizes
codons on the mRNA. The anticodon bases 36, 35, and 34 pair with nucleotide 1, 2,
and 3 of the mRNA codon respectively. Position 34, also called the wobble position is
the only position allowing non-Watson Crick pairing and therefore, in most cases,
permits one tRNA to recognize more than one codon. There is a strong correlation
between tRNA abundance and frequency of the corresponding codon, suggesting co-
evolution of tRNA and codon choice. Highly expressed genes such as ribosomal
Introduction
9
genes are enriched for codons recognized by abundant tRNAs, suggesting that tRNA
can influence protein expression. tRNA regulation at different levels influences
protein synthesis: i) tRNA abundance, ii) tRNA aminoacylation levels, iii) tRNA
localization, iv) and tRNA modifications.
Figure 2: Secondary and tertiary structure of tRNAs.
(A) Schematic representation of the secondaty cloverleaf structure of a tRNA molecule. The
main regions are indicated. The circles represent nucleotides and the connecting lines the
hydrogen bonds between adjacent nucleotides. (B) Three dimensional structure of the tRNA
adopting an L-shape. From Wikipedia (tRNA).
1.3.1 tRNA abundance
tRNA abundance is dependent on tRNA transcription and degradation rates and
both process are regulated in response to environmental conditions. Transcription of
tRNA by RNA polymerase III (PolIII) takes place in the nucleolus and is regulated by
Maf1 in response to environmental conditions (Pluta, Lefebvre et al. 2001). Under
rich growth conditions, Maf1 is phosphorylated by protein kinase A (PKA) and the
mammalian target of rapamycin complex 1 (mTORC1)-dependent kinase, Sch9 and
Introduction
10
does not interact with PolIII. However, in response to environmental conditions
inducing slower growth such as shifts from fermentation to respiration and nutrient
deprivation Maf1 is dephosphorylated and inhibits PolIII activity (Huber,
Bodenmiller et al. 2009; Wei and Zheng 2011). tRNA molecules are very stable and
therefore their levels were long thought to depend mainly on transcription rate. The
measured tRNA half-lives are very long ranging from 3 days in avian liver, 50 h in
chicken muscles, to 44h in Eulgena gracilis (Nwagwu and Nana 1980; Kanerva and
Maenpaa 1981; Karnahl and Wasternack 1992). However emerging evidences show
that there are tRNA quality control and degradation mechanisms. Two pathways
monitoring tRNA integrity have been described: the nuclear surveillance pathway and
the rapid tRNA decay (RTD) (Kadaba, Krueger et al. 2004; Alexandrov, Chernyakov
et al. 2006). The nuclear surveillance turnover pathway targets pre-tRNA to 3’-5’
degradation after polyadenylation. The RTD pathway subjects mature tRNA lacking
different combinations of mutations to 5’-3’ degradation. tRNA degradation can also
occur in response to stresses such as oxidative conditions, nutritional deficiency,
growth in stationary phase and heat shock (Thompson, Lu et al. 2008; Yamasaki,
Ivanov et al. 2009; Nawrot, Sochacka et al. 2011). Under oxidative stress an RNase,
Rny1 in yeast or angiogenin in mammalian cells, is released from the vacuole or the
lysosome into the cytoplasm and induces tRNA cleavage at the anticodon loop
(Thompson and Parker 2009; Yamasaki, Ivanov et al. 2009). Deep sequencing
revealed the presence of various tRNA fragments in the cells such as tRNA-derived
stress-induced RNAs, tRNAs-derived fragments, and tRNAs halves (Li, Luo et al.
2008; Li, Ender et al., 2012). These fragments are thought to act as signaling
molecules and have been linked with cell proliferation (Ivanov, Emara et al. 2011 ; Li
and Hu 2011). tRNA cleavage is also used as a pathogenic and a host defense
mechanism. For example, tRNA cleaving toxins, such as PrrC, colicins, onconase,
and the Klyveromyces lactis γ-toxin are produced in response to an infection or by a
competing organism (Nawrot, Sochacka et al. 2011). These toxins target the
anticodon-loop of specific tRNAs and are often modification specific. For exemple
the Kluveromyces lactis γ-toxin specifically targets mcm5s2U34 modified tRNAs.
Together these studies show that tRNA concentration is highly regulated to adjust
translation levels and ultimately growth to environmental conditions.
Introduction
11
1.3.2 tRNA aminoacylation levels
To match codons with amino-acids, each tRNA molecule needs to be charged with
the correct amino acid by aminoacyl-tRNA synthetases in a process called
aminoacylation. During aminoacylation, the amino-acid is attached to the 3’ CAA
extension the tRNA in an ATP-dependent manner. tRNA synthetases are responsible
for the fidelity of translation and need to accurately recognize the tRNA species and
the corresponding amino-acid. Indeed, during translation, codon-anticodon
recognition is independent from the amino-acid charged on the tRNA. tRNA
synthetases are often able to recognize more than one tRNA coding for the amino-
acid but they are amino-acid specific. The rate of codon translation is dependent on
the concentration of charged tRNAs and tRNA charging is regulated in response to
environmental conditions. Upon starvation, the overall levels of aminoacyl tRNA are
reduced and recent work show that selective charging can take place under certain
growth conditions (Dittmar, Sorensen et al. 2005; Zaborske, Narasimhan et al. 2009;
Zaborske and Pan 2010). Thereby, tRNA charging can regulate both general and
amino-acid specific translation.
1.3.3 tRNA localization Recent studies challenged the long-standing view that tRNA molecules are
processed in the nucleus and unilaterally delivered to the cytoplasm for translation. In
fact, tRNA trafficking is very dynamic and tRNA molecules constantly travel out and
into the nucleus. There is also a retrograde pathway in response to nutrient availability
delivering mature tRNAs from the cytoplasm to the nucleus. Upon starvation tRNA
molecules are sent back to the nucleus where they accumulate, before shuttling back
to cytoplasm upon more favorable growth conditions (Shaheen and Hopper 2005).
1.4 tRNA modifications
tRNAs require simultaneously homogeneity to allow uniform binding to eEF1A, to
the ribosome and usage in common translation mechanism and heterogeneity to allow
specific binding to aminoacyl-tRNA synthetases and cognate codons. With over 90
different posttranscriptional modifications, tRNAs are the most highly modified
Introduction
12
RNAs and every yeast tRNA molecule contains between 7 to 17 modifications
(Phizicky and Alfonzo 2010). Some modifications are common to all tRNAs while
some are specific to certain tRNA species. Modified nucleotides are found throughout
the molecule and influence many aspects of tRNA, such as stability and recognition
by aminoacyl synthetases. Importantly the largest variety of modifications is found in
the anticodon loop at position 34 in the anticodon and position 37 adjacent to the
anticodon (Figure 3). These modifications are thought to be important for efficient
codon recognition. In particular, position 34 is important to increase or restrict
wobbling (Phizicky and Hopper 2010).
Figure 3: tRNA are highly modified molecules.
Schematic representation of a tRNA molecule indicating the modified nucleotides and their
positions. Data are compiled from all tRNAs and all kingdoms. Dots represent nucleotides.
Modifications found at position 34 and 37, marked as black dots, are highlighted in the grey
boxes. Picture from (Grosjean 2000).
Introduction
13
1.4.1 Wobble base modifications
Standard Watson-Crick interactions allow adenine to pair only with thymine or
uracil and guanine only with cytosine. However during codon anticodon interaction
the wobble position permits non Watson-Crick pairs. This allows uracil to recognize
adenine and guanine and allows inosine, a modified adenine nucleotide, to recognize
uracil, adenine and cytosine. Wobble pairing explains how the set of 42 tRNA species
in yeast recognizes all the 61 sense codons and translates them into 20 amino acids
(Table 1). As a consequence, 6 tRNAs contain inosine and 13 uracils at the wobble
position. Uracil-34, at the wobble position of tRNA, is a hot spot for modifications
and is rarely found unmodified. Uracil bases are mostly found as methyl derivatives
in all kingdoms. Furthermore Uracil34 from tRNAs for Lys, Gln and Glu is universally
modified to 5-methyl-2-thio derivatives. Yeast and higher eukaryotes bear the
methoxycarbonylmethyl-2-thiouridine (mcm5s2U) modification at Uracil34 of
cytoplasmic tKUUU, tQUUG, and tEUUC (Figure 4). While the mcm5 modification is found
on five tRNAs, thiolation, s2, is exclusively found on UUX anticodons probably to
compensate for the weaker hydrogen bonds. Indeed uracil only forms two hydrogen
bonds with the complementary nucleotide while guanine and cytosine pair with three
hydrogen bonds.
Figure 4: The 5-methoxy-carbonyl-methyl-2-thio-uridine modification (mcm5s2U34)
Chemical structure of the mcm3s2 (in red) wobble uridine and its location on a schematic
tRNA molecule. Circles represent nucleotides, with the anticodon loop nucleotides in grey.
Introduction
14
Table 1: The genetic code in yeast
Table showing the correspondence between codons, tRNA anticodons and amino acids in
Saccharomyces cerevisiae. The modifications found in the anticodon are indicated. The red
square indicates the split-codon boxes recognized by thiolated (s2) tRNAs. From (Johansson,
Esberg et al. 2008)
1.4.2 The mcm5s2U34 modification The mcm5s2 modification requires over 20 proteins in two distinct enzymatic
pathways. The elongator (ELP) pathway is required for the mcm5 and the URM1
pathway for the s2/thiolation modification (Figure 5). Components of the mcm5
modification pathway include the Elp-complex, constituted of Elp1-Elp6, Kti11-13,
as well as the methyltransferases Trm9 and Trm112. The Elp-complex and the Kti
proteins catalyze early steps of the mcm5 and the ncm5 modification, which found on
eight tRNAs besides tKUUU, tQUUG, and tEUUC. The Elp-complex was originally
discovered by co-immunoprecipitation with active PolII and showed in vitro histone
acetylase activity. The complex was implicated in many processes such as
transcription, endocytosis, and tubulin acetylation (Creppe, Malinouskaya et al.
2009). However several studies suggest that most, if not all, of the elp∆ phenotypes
are mediated by their function in tRNA modification (Esberg, Huang et al. 2006). The
Elp-complex is associated with Kti11 and Kti12 (Fichtner and Schaffrath 2002). The
Kti proteins were identified in a screen for Kluveromyces lactis killer toxin insensitive
mutants (Butler, White et al. 1994). Indeed, Kluveromyces lactis produces the
zymocin toxin, which kills yeast cells by inducing G1 arrest. This arrest is due to
specific cleavage of mcm5s2 Uracil34 of tKUUU, tQUUG, and tEUUC by the toxin leading to
Introduction
15
cell death by depleting these tRNAs (Huang, Johansson et al. 2005; Lu, Huang et al.
2005). Therefore cells lacking either the thio/s2 or the mcm5 modification are resistant
to the toxin. The methyltranferases Trm9 and Trm112 are responsible for the final
methylation of ncm5 to mcm5.
Figure 5: The URM1 and the ELP pathways are responsible for the mcm5s2 modification
of uracil 34
The wobble uracil (U34) of specific tRNAs is modified by the URM1 and the ELP pathways to
form mcm5s2U34. The modified structures are indicated in red.
Thiolation is only found on tKUUU, tQUUG, and tEUUC and depends on a chain of
sulfur transfer from cysteines to tRNAs. Yeast cells have two distinct sulfur relay
pathways for cytoplasmic and mitochondrial tRNAs. Both require first conversion of
cysteine into alanine by the cysteine desulfurase Nfs1, which subsequently forms a
persulfide. The sulfur is then incorporated to the cytoplasmic tRNAs by the URM1
sulfur relay pathway involving Tum1, Urm1, Uba4, the Urm1 activating enzyme, and
the Ncs2/Ncs6 heterocomplex (Leidel, Pedrioli et al. 2009; Noma, Sakaguchi et al.
2009) (Figure 6). While there is a small residual tRNA thiolation in tum1∆ cells,
mutants of all the other pathway components are completely devoid of thiolated
tRNAs. The ELP and the URM1 pathways are independent and loss of one
modification does not abolish the other. In mitochondria, however, tRNA thiolation is
not dependent on the URM1-pathway but on a yet not identified sulfur relay system
involving MTU1.
Introduction
16
Figure 6: The URM1 pathway.
Schematic of the URM1 sulfur relay system responsible for the thiolation of Uridine34 of
specific tRNAs. The sulfur transferred from cysteine to the tRNA is highlighted in red.
1.5 The Ubiquitin-related modifier 1 Urm1, a small protein of 100 amino acid, shows high sequence homology to
bacterial sulfur carriers and high structural homology to ubiquitin-like modifier
(Figure 7). It has the typical ß-grasp fold and double glycine end, hallmark of
ubiquitin-like modifiers and several sulfur carriers. These structural similarities
suggest an evolutionary relationship between bacterial sulfur carrier proteins and
eukaryotic ubiquitin-like modifiers and places Urm1 at the crossroad (Pedrioli, Leidel
et al. 2008; Petroski, Salvesen et al. 2011). Beside its function in tRNA thiolation,
Urm1, like ubiquitin and ubiquitin-related modifier, is covalently conjugated to target
proteins in a process called urmylation (Furukawa, Mizushima et al. 2000; Goehring,
Rivers et al. 2003; Van der Veen, Schorpp et al. 2011). Only a few target proteins
such as Ahp1 and the mammalian homologues of Uba4 and Ncs6, MOCS3 and
ATPBD3, respectively, were identified but the functional relevance of the protein
conjugation is unclear (Van der Veen, Schorpp et al. 2011). The protein conjugation
of Urm1 to the components of the URM1 pathway, MOCS3 and ATPBD3, under
oxidative stress suggests a link between the tRNA thiolation and the urmylation
pathways. Interestingly, both functions of Urm1 require the C-terminal double glycine
motif and activation by Uba4. Activation of Urm1 requires acyl-adenylated by Uba4,
which leads to formation of a thiocarboxylate with Uba4 is formed in an ATP-
Introduction
17
dependent manner (Noma, Sakaguchi et al. 2009). The dual-nature of Urm1 makes it
an interesting protein to understand how ubiquitin-like modification systems evolved
from prokaryotic sulfur carriers.
Figure 7: The ubiquitin-related modifier 1, Urm1.
Structures of Urm1 and ubiquitin showing the typical ß-grasp fold of ubiquitin like modifiers.
From (Pedrioli, Leidel et al. 2008).
1.6 Phenotypes associated with lack of wobble uridine modification In higher eukaryotes homologues of the yeast URM1 and UBA4 (encoded in
mammals by MOCS2A and MOCS3 respectively) are also responsible for thiolation
of wobble uracil of the same set of tRNAs (Nakai, Nakai et al. 2008; Schlieker, Van
der Veen et al. 2008; Leidel, Pedrioli et al. 2009; Noma, Sakaguchi et al. 2009).
Likewise, the ELP pathway and its function in the mcm5 modification of tRNAs is
conserved across species (Mehlgarten, Jablonowski et al. 2010; Leihne, Kirpekar et
al. 2011; van den Born, Vagbo et al. 2011). A plethora of pleiotropic phenotypes are
associated with lack of mcm5s2 in different organisms and these effects are thought to
be the indirect consequences of mRNA transcripts enriched in codons read by
thiolated tRNAs that are not properly translated (Maraia, Blewett et al. 2008).
In yeast, the URM1 gene is not essential but combined deletion of an ELP gene and
URM1, results in an extremely slow growing phenotype. Lack of URM1 alone also
renders the cells sensitive to various drugs such as diamide, rapamycin, and caffeine,
linking thiolation with oxidative stress and nutrient signaling. Components of the ELP
pathway have been linked with telomeric gene silencing and DNA damage response
(Begley, Dyavaiah et al. 2007; Chen, Huang et al. 2011; Patil, Dyavaiah et al. 2012).
Introduction
18
Interestingly overexpression of unmodified tKUUU, tQUUG, and tEUUC is able to
bypass the requirement for tRNA thiolation. Indeed sensitivity to diamide and
synthetic lethality of the double ncs6∆ elp∆ mutants is recued by overexpression of
unmodified tKUUU, while rapamycin and caffeine sensitivities are rescued by
overexpression of unmodified tQUUG (Leidel, Pedrioli et al. 2009). These data support
the idea that only a small subset of genes enriched for Urm1-dependent codons is
affected by lack of thiolation. In Caenorhabditis elegans, lack of the ELP1, ELP3 or
NCS6 homologues, ELPC-1, ELPC-3 and TUC-1 respectively, led to defects in salt
chemotaxis learning. In addition, elpc-1, tuc-1 double mutants showed defects in
neurological function and were lethal at 25°C but viable at 15°C, allowing
temperature shifting experiments at different larval stages. Eggs grown at restrictive
temperature led to arrest in embryogenesis while temperature shifting at later stages
showed abnormal development of vulva and germ cells, leading to small sterile
adults. These data correlated with increased protein expression of ELPC-1 in neuronal
and vulval cells (Chen, Tuck et al. 2009). In human HeLa cells, shRNA directed
against URM1 induced increased cell size and multinucleated cells suggesting a role
of URM1 in cell cycle progression and cytokinesis. Several diseases in humans are
linked with hypomodification of U34 in the cytoplasm or mitochondria such as
Myoclonic Epilepsy with Ragged Red Fibers (MERRF), Mitochondrial
Encephalopathy Lactic Acidosis and Stroke-like episode (MELAS) and the
neurodegenerative familial dysautonomia (Umeda, Suzuki et al. 2005; Yasukawa,
Kirino et al. 2005; Svejstrup 2007). The conservation of the URM1 and the ELP
pathways and their involvement in various cellular and developmental processes
highlights the importance of tRNA modification at the wobble uridine.
Introduction
19
1.7 Aim of the work
General protein translation is a well-studied basic cellular process and numerous
biochemical and structural studies allowed to understand the interactions between
ribosomal components, tRNA and mRNA at the molecular level. Almost 100
nucleotide modifications on tRNAs are known to date and due to the sheer number
and complexity of these modifications, little is known about their effects on
translation.
Previous studies identified the URM1 pathway, which is responsible for thiolation
of the wobble uracil of tKUUU, tEUUG, and tQUUC, and implicated this pathway in
resistance to various stresses. The aim of this work was to characterize how thiolation
of specific tRNAs at the wobble position influences translation in the cell and
understand the underlying molecular mechanism. In particular, we wanted to identify
proteins, whose expression is altered by lack of this modification. From these findings
we hope to understand how wobble base tRNA modifications mediate increased
stress-resistance.
Results
20
2 RESULTS
Manuscript in preparation:
Urm1-dependent thiolation of specific tRNAs is required to efficiently translate a subset of proteins by promoting binding to the ribosomal A-site
Contributions:
Patrick Pedrioli and Kshitiz Tyagi performed the quantitative SILAC-based mass
spectrometry analysis with statistical analysis of the data
Maria Anisimova and Stefan Zoller performed the bioinformatic analysis
Namit Ranjan performed the in vitro A-site binding to the ribosome and dipeptide
bond formation analysis in the laboratory of Marina Rodnina
2.1 General translation is unaffected by lack of URM1
Deletion of URM1 completely abolishes thiolation of the wobble uracile of tKUUU,
tQUUG, tEUUC. This modification has been shown to be important for recognition of the
corresponding cognate codons AAA, CAA, and GAA in vitro. These codons are
amongst the most abundant and are virtually present in all proteins. We therefore
wanted to know whether overall protein expression levels are impaired by the lack of
U34 thiolation. For this we measured [35S]Met and [35S]Cys incorporation into newly
synthesized proteins in wild-type and urm1∆ yeast cells in a pulse-chase experiment.
Equal amount of cells were labeled with [35S]Met and [35S]Cys for 15 min at 30°C
followed by addition of unlabeled Met and Cys for 5 min. After cell lysis and protein
precipitation with 10% TCA, protein pellets were resuspended in urea/SDS buffer and
[35S]-activity was measured by liquid scintillation counting. We found no significant
differences in [35S]-incorporation into proteins from wild-type and urm1∆ cells
(Figure 8), showing that general protein expression is unaffected by lack of thiolated
tRNAs. Control cells treated with cycloheximide before labeling to block translation
showed almost no incorporation and ruled out that the [35S]-labeling arose from
unspecific binding or tRNA aminoacylation of tMet and tCys. In parallel, we ran a
fraction of the extracts on a SDS-PAGE and [35S]-radioactivity was detected on a
Results
21
phosphor screen. Proteins from wild-type and urm1∆ cells showed a similar pattern
(Figure 8), confirming that there is no obvious changes in protein synthesis efficiency.
Figure 8: Overall protein synthesis is not affected in urm1∆ cells.
Wild-type (WT) and urm1∆ cells were pulsed for 15 min with [35S]Met and [35S]Cys in the
presence (+) or absence (-) of cycloheximide (CHX). [35S]-incorporation into proteins was
quantified by liquid scintillation counting. Counts per minute (CPM) were normalized to the
wild-type value. Data show mean ± standard error of the mean of three independent
experiments. In the right panel [35S]-labeled proteins from wild-type and urm1∆ cells were
run on SDS-PAGE and detected by phosphorimager after the pulse-chase experiment.
Ribosomes can be found in the cell in different states: as free 40S and 60S
subunits, ready to start elongation as 80S or as polysomes, which are multiple
ribosomes bound on one mRNA molecule. Significant changes affecting initiation,
elongation, or termination would be reflected by changes in polysome profiles and the
ratio of polysomes to total ribosomes can be used to estimate the fraction of
ribosomes actively involved in translation. We hypothesized that lack of modification
at the wobble uridine might slow down ribosomes at AAA, CAA, and GAA codons
thereby increasing the fraction of polysomes and decreasing the proportion of free
ribosomes. To test this, we performed polysome profile analysis of ribosomes from
wild-type, urm1∆, and uba4∆ cells. After translation block with cycloheximide, the
cells were lysed by bead beating. Proteins were separated by centrifugation on a 6-45
% sucrose gradient and the ribosomes detected with A260. We found that distribution
Results
22
of 40S, 60S and 80S particles as well as polysomes in urm1∆ and uba4∆ cells were
not significantly different from the distribution in wild-type cells (Figure 9), further
supporting that general translation is not notably impaired in cells lacking U34
thiolation.
Figure 9: Lack of URM1 does not affect general translation.
Polysome profiles of wild-type, urm1∆, and uba4∆ protein extracts separated on a 6-45%
sucrose gradient. The last panel shows the quantification of the polysome profiles and the
distribution of 40S, 60S, 80S particles and polysomes as percentage (%) of total ribosomes
from the average of three independent experiments.
The experiments described above show that overall protein expression is not
affected, however lack of tRNA thiolation at U34 might affect protein expression of
Results
23
specific genes. To address this, we used quantitative proteomics, which can monitor
abundance of multiple specific proteins in a single assay. We therefore compared
protein abundances in wild-type and urm1∆ cells by SILAC-based mass spectrometry
(Patrick Pedrioli and Kshitiz Tyagi) and found that the protein abundance ratio of
wild-type versus urm1∆ cells for most genes was unchanged (Figure 10), confirming
the experiments described above. In contrast, a small subset of proteins showed subtle
protein expression changes in urm1∆ cells (Figure 10).
Together these data suggest that although the average frequency of thiolation-
dependent codons is high (11.5%), it is not sufficient to induce significant changes in
overall translation but rather induces modest changes in a small proteome subset.
Figure 10: Lack of URM1 affects only a small subset of genes.
Density plot of wild-type (WT)/urm1∆ protein abundance ratios obtained using SILAC based
quantitative proteomics..
2.2 URM1 is important for efficient expression of AAA, CAA, and GAA-rich
proteins In order to better understand how cells are affected by lack of thiolation, we
wanted to identify the subset of proteins differentially expressed in urm1∆ cells. To
account for small changed in protein abundances and identify candidates with high
confidence, Patrick Pedrioli and Kshitiz Tyagi devised a quantitative proteomic
workflow able to maximize proteome coverage and statistical significance. The
Results
24
SILAC based measurements were performed in six biological replicates, to increase
confidence, and the samples were subjected to extensive fractionation and searched
using a combination of multiple database search engines to increase the number of
detected and identified proteins. Relative abundance of 3,818 proteins were
quantified, corresponding to 51% of the predicted yeast proteome at a 1% false
discovery rate (FDR) (Figure). 62 proteins were highly significantly (p-value < 0.05)
downregulated and 55 were highly significantly upregulated in urm1∆ cells. To
identify the cellular processes affected under lack of thiolation, we performed Gene
Ontology (GO) annotation analysis on a larger set of 286 upregulated and 267
downregulated proteins (Figure 12). Upregulated proteins were enriched for gene
products involved in catabolic processes and in response to stresses such as oxidation,
heat and unfolded proteins, suggesting that urm1∆ cells are under higher stress under
standard growth conditions than their wild-type counterparts. Interestingly Tum1, a
component of the URM1 pathway, and Trm112, implicated in the mcm5 modification,
were significantly upregulated, indicating that the cells are compensating for the lack
of URM1. Downregulated proteins were enriched for gene products involved in
translation such as translation initiation and ribosome assembly. Importantly,
bioinformatic analysis revealed that the genes dataset with downregulated protein
expression is significantly enriched for the Urm1-dependent codons AAA, CAA, and
GAA while the frequency of these codons in the genes dataset with upregulated
protein expression was similar to the average frequency in the whole genome (Figure
13). Furthermore genes whose protein expression was downregulated were
significantly biased for the A-ending codons, AAA, CAA, and GAA over the G-
ending codons, AAG, CAG, and GAG (Figure 13 and Figure 14). In contrast, genes
whose protein expression was upregulated included more the G-ending codons when
compared to the whole genome (Figure 13).
It was found that overexpression of unmodified tEUUU, tQUUG, and tEUUC in urm1∆
cells rescues urm1∆ drug sensitivities (Bjork, Huang et al. 2007; Leidel, Pedrioli et al.
2009). Thus we wanted to test whether overexpression of tRNAs was likewise able to
restore wild-type protein expression levels in urm1∆ cells. tEUUU, tQUUG, and tEUUC
were overexpressed separately or together in urm1∆ cells and protein abundance
ratios compared to wild-type cells. Figure 11B shows that when all three tRNAs were
overexpressed together, most of the genes that were previously differentially
Results
25
expressed in urm1∆ cells restored wild-type protein expression levels. Overexpression
of the tRNAs separately had intermediate levels of rescue.
Figure 11: Genes differentially expressed in urm1∆ cells
(A) Volcano plot showing the wild-type (WT)/urm1∆ protein abundance ratios from SILAC-
based mass spectrometry (MS) measurements. Associated confidence is expressed as -log10
(adjusted p-value) from six biological replicates. The false-discovery rate (FDR) of 5%
indicated as a dotted red line was chosen as threshold for statistical significance. (B) Heat
map of the WT/urm1∆ protein abundance ratios of the significantly up- and down-regulated
proteins from Figure 1C in cells overexpressing tKUUU, tQUUG, tEUUC individually, or in
combination compared to cells without plasmid.
Experiments by Patrick Pedrioli and Kshitiz Tyagi
Results
26
Figure 12: Processes affected in urm1∆ cells.
(A and B) Gene Ontology enrichment analysis of the genes upregulated (A) and
downregulated (B) in urm1∆ cells. The processes enriched more two-fold more are indicated.
Black bars represent the percentage of genes belonging to the category in the differentially
expressed genes. Grey bars represent the percentage of the genes belonging to the category in
the whole genome.
Results
27
Figure 13: Downregulated genes are enriched for AAA, CAA, and GAA, codons.
Box plots showing the distribution and the median of the absolute frequency (left panel) or
the codon bias (right panel) of AAA, CAA, and GAA codons. The codon bias indicates how
often the A-ending codons, AAA, CAA, and GAA, are preferred over the G-ending codons,
AAG, CAG, and GAG. A value of 1 means that only the A-ending codons are used. The
whole yeast genome values are compared to the set of downregulated and upregulated genes.
The red line represents the value for YPL199C and the dotted red line the value for CMS1,
two genes enriched for AAA, CAA, and GAA codons and significantly downregulated in the
quantitative proteomic analysis. These two genes are further analyzed in Figure xx.
Analysis by Stefan Zoller and Maria Anisimova.
Figure 14: URM1 is required for translation of A-ending codons.
Results
28
Volcano plots show the protein abundance ratios with statistical significance measured by
quantitative proteomics of the top 1% yeast genes with the highest frequency of AAA, CAA,
and GAA codons (left panel) or the 1% genes with the highest frequency of AAG, CAG, and
GAG codons.
Analysis by Patrick Pedrioli and Kshitiz Tyagi
2.3 Differential expression of AAA, CAA, and GAA-rich genes is not due to differential transcription
We hypothesized that the protein downregulation in urm1∆ cells, of genes enriched
for the thiolation-dependent codons AAA, CAA, and GAA, was due to decrease in
translation rates. To verify that the differential protein expression was not due to
decrease in transcription or mRNA stability, we selected 13 genes that were highly
significantly downregulated in urm1∆ cells and enriched for Urm1-dependent codons
(Table 2). and measured their mRNA expression levels by quantitative real-time PCR.
Total mRNA was extracted with hot phenol from wild-type and urm1∆ cells, reverse
transcribed and mRNA levels quantified by SYBR green fluorescence with a real-
time light cycler. All the primers were tested for specificity and primer pair
efficiencies were calculated from dilution curves. After normalization to PGK1,
mRNA levels were calculated with the ∆∆Ct method corrected for the primer
efficiency. mRNA abundance in wild-type cells was set to 100%. All candidates
showed similar expression levels in wild-type and urm1∆ cells (Figure 15),
suggesting that the observed protein level changes are due to changes at the protein
level. ACT1 and URM1 were used as negative and positive controls respectively. We
verified that PGK1 and ACT1 were not enriched for thiolation-dependent codons and
were not differentially expressed in the quantitative mass spectrometry experiment,
qualifying them as good controls. We included as well some genes that were found by
quantitative proteomic to have differential protein expression but were not enriched
for Urm1-dependent codons; TUB1, TUM1, TRM12, KOG1 protein expressions were
upregulated, while MET6 and MET10 protein expressions were downregulated in the
SILAC-based mass spectrometry. Surprisingly those genes did also not show
differential mRNA levels, showing that also these genes are not regulated upstream of
translation.
Results
29
ORF Gene Name Log2(WT/urm1∆)
Log10(adj.P value)
Freq (AAA, CAA, GAA)
YPL199C YPL199C 0.97 0.012 28 %
YLR003C CMS1 0.87 0.020 19.1 %
YIL063C YRB2 0.82 0.022 20.2 %
YDL213C NOP6 0.76 0.013 16.9 %
YDR233C RTN1 0.65 0.021 20.6 %
YCR016W YCR016W 0.61 0.049 16.8 %
YOL109W ZEO1 0.55 0.046 30.9 %
YHR135C YCK1 0.52 0.025 23.4 %
YFR001W LOC1 0.52 0.036 16.6 %
YMR235C RNA1 0.51 0.020 19.1 %
YLR449W FPR4 0.47 0.035 19.9 %
YPR148C YPR148C 0.46 0.023 17.2 %
YKL054C DEF1 0.40 0.043 16.2 %
Table 2: URM1 regulated genes.
List of selected candidates significantly downregulated in urm1∆ cells and enriched for
Urm1-dependent codons showing the measured protein abundance ratios averaged from six
experiments and the associated significance. The last row shows the frequency of AAA,
CAA, and GAA codons in the gene.
Results
30
Figure 15: Transcription in wild-type and urm1∆ cells.
The mRNA levels of various candidate genes were quantified by quantitative PCR in wild-
type (WT) and urm1∆ cells. ACT1 and URM1 were used as negative and positive controls
respectively. Data show the mean ± standard error of the mean of three independent
experiments.
2.4 The protein expression changes induced by lack of URM1 are small We next wanted to validate the list of candidate proteins found to be
downregulated in urm1∆ cells by SILAC-based mass spectrometry (Table 2) using an
alternative methods. To this end, protein expression levels of selected C-terminally
TAP-tagged candidates were tested in wild-type and urm1∆ cells by western blot
analysis. Cells growing exponentially in minimal SD-complete medium were lysed
and proteins extracted with 10% TCA. Protein extracts were tested with PAP and
anti-Pgk1, as a loading control. We found that protein expression levels of most of the
candidates were similar in wild-type and urm1∆ cells (Figure 16), suggesting that
their protein expression is not significantly affected by lack of thiolation. In contrast,
a significant decrease was observed for Cms1 and Ypl199c when expressed in an
urm1∆ background, suggesting that thiolation is important for expression of these two
proteins. Surprisingly, most of the candidates did not show the expected change in
protein expression. One explanation could be the presence of the TAP tag, which can
change the codon composition and therefore frequency of the thiolation-dependent
codons. Another explanation is that western blotting is only semi-quantitative
(Heidebrecht, Heidebrecht et al. 2009) and might not really be suitable to detect such
small expression changes. This might indicate that the quantitative SILAC-based
proteomic workflow used here showed higher sensitivity than the western blot
approach.
Results
31
Figure 16: Protein expression of selected candidates in wild-type and urm1∆ cells.
TAP-tagged candidates in wild-type and urm1∆ backgrounds were tested by western blotting.
Proteins extracted with 10% TCA were tested with PAP or anti-Pgk1 antibodies.
2.5 URM1 is required for efficient translation of CMS1 and YPL199C Out of these 13 candidates we further analyzed Cms1 and Ypl199c, which showed
the strongest changes in expression in the quantitative proteomic approach and by
western blotting. Protein expression levels from TAP-tagged Cms1 and Ypl199c were
analyzed by western blotting and quantified. Protein extracted from wild-type an
urm1∆ cells exponentially grown in SD-complete were tested with PAP and anti-
Pgk1. Quantification was performed from three independent experiments and every
sample was loaded on SDS-PAGE and quantified twice. We observed urm1∆ to wild-
type protein abundance ratios of 0.41 ± 0.2 and 0.33 ± 0.2 for Cms1p and Ypl199cp
respectively (Figure 17A), confirming the range of changes observed by quantitative
mass spectrometry. The quantitative PCR experiment in Figure 15 showed that the
protein expression decrease in urm1∆ cells was not due to decreased mRNA levels
(Figure 17B). To test whether differential degradation in wild-type and urm1∆ cells
was responsible for the protein abundance decrease, we analyzed protein stability of
TAP-tagged Cms1 and Ypl199c in a cycloheximide chase experiment. Exponentially
growing cells were treated with cycloheximide to stop translation. Equal volume of
wild-type and urm1∆ cells were harvested at different time point and lysed by TCA.
Protein stability was followed over two hours, corresponding to one doubling time,
Results
32
and protein levels were analyzed by western blotting (Figure 17C and D). Half-lives
of 60 and 70 min were measured for Cms1 and Ypl199c respectively in both wild-
type and urm1∆ cells (Figure 17E and F). However no significant difference in
stability between wild-type and urm1∆ cells was observed for Cms1 and Ypl199c.
The protein levels of the control Pgk1 were stable and did not increase, confirming
the cycloheximide induced growth arrest. Taken together these data show that lack of
thiolation does not affect mRNA levels or stability of the two candidates and therefore
that differential translation rates likely account for the decreased protein levels in
urm1∆ cells.
Figure 17: URM1 is required for efficient translation of Cms1 and Ypl199c.
(A) Western blot of TAP-tagged Cms1 or Ypl199c from wild-type (WT) and urm1∆ cells
using PAP antibodies or anti-Pgk1 as loading control. The quantification indicates the
Results
33
urm1∆/WT protein abundance ratio averaged from three independent experiments. C.f. also
Figure 16 (B) The mRNA levels of CMS1 and YPL199C and the ACT1 or URM1 controls
were quantified by quantitative PCR in wild-type and urm1∆ cells. Data show the mean ±
standard error of the mean of three independent experiments. Same experiment as Figure 15
but showing only selected genes. (C-F) Protein stability of Cms1 (C) and Ypl199c (D) in
wild-type and urm1∆ cells was followed over time after translation block with cycloheximide
(CHX). Equal volumes of culture were harvested at different time points after cycloheximide
block and protein levels were probed with PAP and anti-Pgk1 (E-F) Quantification of the
cyloheximide chase in C and D showing the degradation curves of Cms1 (E) and Ypl199c (F)
in wild-type and urm1∆ cells. Protein levels over time were compared to the expression level
at time zero. Data show the mean ± SEM of three independent experiments.
2.6 Thiolation is required for codon-specific translation Quantitative proteomic analysis revealed that genes enriched for the Urm1-
dependent codons AAA, CAA, and GAA are preferentially downregulated. However,
some codon-enriched genes did not show the expected downregulation suggesting
that positioning of thiolation-dependent codon might be involved or that some
compensation mechanisms are taking place. Furthermore some downregulated genes
were not enriched for the Urm1-dependent codons, suggesting that they are indirect
targets. To examine the direct link between codon usage and translation regulation in
vivo, we constructed a codon-specific translation reporter. Such a synthetic reporter
allows us to compare the effect of codon-enriched sequences on protein expression in
vivo while excluding compensation mechanisms to regulate protein levels in the cell.
2.6.1 Generation of an inducible dual-fluorescent reporter Most of the commonly used translation reporters are based on renilla and firefly
luciferase activities where one is used as a normalization control and the other as a
codon-specific reporter (Salas-Marco and Bedwell 2005). The sequence of interest is
inserted in the middle of a renilla-firefly fusion protein and its effect measured by
changes in the ratio of renilla and firefly luciferase activity. However, such luciferase-
based reporters in yeast have several disadvantages. They do not allow in vivo
measurements and require cell lysis followed by several enzymatic steps that
Results
34
increases measurement errors. To avoid these drawbacks, we constructed a new
reporter based on fluorescent proteins. To increase sensitivity we used quadruple-CFP
(4 × CFP) and quadruple-venus, an enhanced YFP, (4 × YFP) serving as general
protein expression control and codon-specific translation reporter respectively (Figure
18A). The use of CFP and venus has several advantages. Fluorescent proteins are very
stable, ruling out degradation effects and they are almost identical in sequence,
excluding differential codon composition effects between CFP and YFP. Importantly,
they are easy to detect and quantify by microscopy or by FACS in single cells
enabling cell to cell variations and cell cycle effects to be accounted for. To control
expression levels of the reporter and avoid potential toxic effects of constitutive
fluorophore expression, we placed CFP and venus under inducible GAL1 promoters.
In order to avoid secondary effects of galactose addition, we used a β-estradiol
responsive Gal4 variant that bypasses the requirement for galactose and induces
GAL1 promoters in glucose medium upon addition of ß-estradiol (Louvion, Havaux-
Copf et al. 1993). Codon-enriched sequences, hereafter referred to as codon-trap, are
inserted in the inducible dual-fluorescent reporter by two unique restriction sites two
amino acids after the N-terminus of the quadruple-venus. After reporter induction in
cells, YPF and CFP signal intensities were measured by inverted fluorescence
microscopy in a 30°C incubation chamber. YFP reporter expression was normalized
to CFP levels in single cells with the YeastQuant image processing platform (Pelet,
Dechant et al. 2012) and the mean of the YFP/CFP signal over many cells compared
in wild-type and mutant cells (Figure 18C). This workflow enabled the imaging and
analysis of over 1000 cells in a single experiment, allowing single cell measurements
from a large population size.
Results
35
Figure 18: The codon-specific dual-fluorescent translation reporter.
(A) Schematic representation of the dual-fluorescent codon-specific translation reporter.
Quadruple-venus (4xYFP) or quadruple-CFP (4xCFP) proteins serve as codon-specific
translation reporters and internal translation control, respectively. Expression of both
fluorescent reporters under GAL1 promoters is induced by addition of ß-estradiol. Codon-
traps composed of a run of ten identical codons, (XXX)10, are inserted at the N-terminus of
YFP. The reporter is integrated in wild-type (WT) and mutant cells. (B) Schematic
representation of the induction workflow. At time zero (0) cells were induced with 50 nM ß-
estradiol for 3h, then translation was blocked with cycloheximide (CHX). Cells were
incubated for 1h at 30°C to allow fluorophore maturation and finally imaged. (C)
Representative images showing the expression of the codon-specific reporter in wild-type and
urm1∆ cells. Expression levels of YFP and CFP reporters were measured in wild-type and
urm1∆ cells after 3h induction. YFP/CFP ratio was subsequently analyzed in single cells with
the YeastQuant program.
Results
36
2.6.2 URM1 is required for expression of AAA, CAA, and GAA enriched
reporters Repeats of ten or more consecutive CAA and GAA codons are rare but
physiologically relevant. To test this extreme case, we analyzed the effect of a codon-
trap consisting of ten consecutive CAA codons, (CAA)10, on expression of our
reporter in a time-course experiment. Equal OD600 of wild-type and urm1∆ cells were
grown in exponential phase in SD-complete medium. Expression of CFP and
(CAA)10-YFP reporters was induced at time zero by addition of 50 nM β-estradiol and
YFP and CFP fluorescence followed over time. We found that after 45 min,
corresponding to the maturation time of the expressed fluorophore, both wild-type
and urm1∆ cells started to show increasing reporter signal, however expression in
urm1∆ cells was significantly slower and urm1∆ cells only reached ∼ 60 % of wild-
type expression after 5 hours (Figure 20A).
Based on this experiment, for further analysis we analyzed expression differences
in wild-type and urm1∆ cells after 3h induction. After 3h induction protein synthesis
was blocked by addition of cycloheximide and the cells were incubated for 1h to
allow complete fluorophore maturation before imaging (Figure 18B). YFP/CFP
intensities were then compared in wild-type and mutant cells. After 3h induction
expression of the (CAA)10 reporter in urm1∆ cells was about 40 % lower than the
expression in wild-type cells (Figure 20B), confirming the results of the time-course
experiment. To verify that reporter expression reflects the translational state of the
cell, we compared the CFP and YFP fluorescence in single cells. CFP and YFP
fluorescence correlated in both wild-type and urm1∆ cells (Figure 19), confirming
that cells expressing CFP well also expressed (CAA)10-YFP well. As expected the
range of CFP expression was comparable in wild-type and urm1∆ cells, while YFP
signal intensity was significantly reduced in urm1∆ cells. We further analyzed the
behavior of both reporters, we looked at the distribution of the CFP and YFP signal
separately in wild-type and urm1∆ cells. Distribution of the CFP signal in wild-type
and urm1∆ cells was comparable with a peak of cells expressing at 500 (Figure 19).
In contrast the distribution of the YFP signal showed a shift to lower intensities of
urm1∆ cells (Figure 19), confirming that only the YFP reporter containing the codon-
trap is affected by lack of thiolation. Reporter expression peaks were lower in urm1∆
cells because the overall number of urm1∆ cells was slightly reduced. Indeed, due to
Results
37
slower growth of urm1∆ cells in minimal medium, the density of urm1∆ cells was
slightly reduced compared to wild-type cells after the 3h induction.
Next, we wanted to determine whether AAA and GAA codons are also affected by
lack of thiolation. We therefore analyzed the effect of codon-traps with ten AAA
repeats, (AAA)10, and ten GAA repeats, (GAA)10. We observed a similar 40%
expression decrease with the (AAA)10 and (GAA)10 reporters in urm1∆ cells relative
to wild-type cells (Figure 20B). Although the absolute levels of expression of the
reporter with different codon-traps varied in wild-type cells, the relative decrease
between urm1∆ and wild-type cells was the similar, suggesting that thiolation affects
all three tRNAs in a similar manner. To confirm that the reporter expression decrease
was due to lack of URM1, we introduced back URM1 on a centromeric plasmid. We
observed that ectopic expression of URM1 restored wild-type levels of reporter
expression when compared with cells with the empty vector (Figure 20C), excluding
the possibility that secondary mutations are accounting for the translation decrease.
Cells lacking UBA4, the Urm1 activating enzyme, phenocopy urm1∆ cells. Thus
we also tested the effect of the (CAA)10 reporter in wild-type and uba4∆ cells. We
observed a 40% decrease in reporter expression in uba4∆ cells compared to wild-type
(Figure 20D), suggesting that this effect is dependent on activation of Urm1 by Uba4.
Since thiolated Uracil34 additionally carries the mcm5, which require the Elp-complex,
we analyzed whether loss of mcm5 affects translation in a similar manner. To this end,
we tested the effect of lack of ELP3, which abolishes the mcm5 modification, on our
translation reporter. A 30% decrease in reporter expression was observed in elp3∆
cells relative to wild-type cells (Figure 20D). These data suggest that both the mcm5
and the s2 modifications are important for efficient expression of CAA enriched
sequences.
Results
38
Figure 19: YFP and CFP expression analysis of the (CAA)10 translation reporter in
single cells.
(A) Dot plots showing the correlation of YFP and CFP expression in wild-type (left panel)
and urm1∆ cells (right panel). Each dot represents a single cell and shows the measured YFP
and CFP signal intensity after 3h induction. The regression line is indicated. (B) Histogram
showing the distribution of control CFP intensity signal (top panel) and YFP intensity signal
(bottom panel) in wild-type and urm1∆ cells expressing the (CAA)10-YFP reporter.
Results
39
Figure 20: URM1 enhances translation of reporters enriched for AAA, CAA, and GAA
codons
(A) Time-course of translation reporter expression with a (CAA)10 codon-trap induced at time
zero in wild-type (WT) and urm1∆ cells. Data show the mean YFP/CFP ratio ± standard error
of the mean from at least 100 cells plotted as percentage (%) of maximum expression. (B)
Expression of the translation reporter after 3h induction with (GAA)10, (AAA)10, or (CAA)10
codon-traps in wild-type or urm1∆ cells. (C) Expression of the translation reporter after 3h
induction with (CAA)10-codon trap was compared in urm1∆ cell with an empty vector or a
vector expressing URM1. (D) Expression of the translation reporter after 3h induction with
(CAA)10 codon-trap in urm1∆, uba4∆, or elp3∆ cells. Data in (B-D) show the mean YFP/CFP
ratio ± standard error of the mean from at least 1000 cells plotted as percentage (%) of wild-
type control.
2.6.3 Reporters with the G-ending codons are less affected by lack of thiolation All the thiolation-dependent codons are part of a so-called split codon box, in
which U and C ending codons code for a different amino acid than A and G-ending
codons. Different codons that code for the same amino acid are called synonymous
codons as the translated proteins bear the same amino acid sequence and all the
tRNAs that recognize such synonymous codons are named isoacceptor tRNAs. In
Results
40
budding yeast, while the AAA, CAA, and GAA codons are recognized by the
thiolated tKUUU, tQUUG, and tEUUC respectively, the synonymous AAG, CAG, and
GAG codons are recognized by the non-thiolated isoacceptors tKCUU, tQCUG, and tECUC
respectively. To address whether the effect of lack of thiolation on protein synthesis
was codon-specific, we tested the effect of the synonymous codon recognized by the
non-thiolated tRNAs. We compared inclusion of (AAG)10, (CAG)10, and (GAG)10
codon-traps on translation in wild-type and urm1∆ cells. Importantly, with inclusion
of AAG codons, we did not observe differences in expression of the fluorescent
reporter between wild-type and urm1∆ cells (Figure 21). This is in contrast with the
results we had observed the synonymous (AAA)10 codon-trap (Figure 20D), showing
that the effect on translation is codon specific and not due to amino-acid composition
effects. However expression of reporters with (CAG)10, and (GAG)10 codon-traps was
decreased by about 20% in urm1∆ cells, suggesting that these codons are also
dependent on URM1 although to a less extent than the synonymous A-ending codons.
Figure 21: URM1 affects translation of GAG and CAG but not AAA codons.
Expression of the fluorescent translation reporter after 3h induction with (GAG)10, (AAG)10,
and (CAG)10 codons traps in wild-type (WT) and urm1∆ cells. These codons are recognized
by the non-thiolated tQUUG, tKUUU, tEUUC respectively. Data show the mean YFP/CFP ration ±
standard error of the mean of at least 1000 cells plotted as percentage (%) of wild-type.
Results
41
2.6.4 The effect of URM1 on translation is dependent on codon frequency
We next wanted to test how shorter runs of codons affect translation. We therefore
inserted (CAA)5 or (GAA)5 codon-traps in the translation reporter and compared
expression after 3h induction in wild-type and urm1∆ cells. We observed a 20%
reduction of translation in urm1∆ cells with the (CAA)5 and (GAA)5 codon-traps
(Figure 22). This corresponds to half of the effect seen with runs of ten codons and
suggests that the effect on translation is dependent on the frequency of the codons.
Figure 22: The effect of URM1 on translation is dependent on codon frequency.
Expression of the translation reporter after 3h induction with (GAA)5 and (CAA)5 codon-traps
in wild-type (WT) and urm1∆ cells. Data show the mean YFP/CFP ration ± standard error of
the mean from at least 600 cells plotted as percentage (%) of wild-type.
2.6.5 Addition of various drugs does not increase the urm1∆ effect on the
translation reporters Cells lacking URM1 are sensitive to paromomycin (2.8) and we wanted to check
whether addition of paromoycin or other translation inhibitors can aggravate the effect
of urm1∆ cells on translation. To this end we pre-treated wild-type and urm1∆ cells
with paromomycin, cycloheximide or hygromycin B for 30 min prior to induction of
the (AAA)10 reporter. Then expression of the reporter was then induced for 3h. We
observed a 40% reduction in (AAA)10 reporter expression in urm1∆ cells compared to
wild-type cells pretreated with paromomycin or hygromycin B (Figure 23A). We
observed a stronger 55% reduction when cells were pretreated with cycloheximide
(Figure 23A), however, the absolute expression levels were very low and the standard
Results
42
deviation was really high in urm1∆ cells. At such low expression levels the
fluorescent signal is very close to the background autofluorescence and therefore
looses sensitivity. We found that the relative expression of wild-type and urm1∆ cells
was similar between untreated cells and cells treated with paromomycin and
hygromycin B (Table 3), suggesting that the effect of URM1 on translation of the
reporter is by a different mechanism from the drugs. Addition of cycloheximide and
hygromycin B significantly decreased the expression levels of the reporter in wild-
type and urm1∆ cells (Figure 23B), consistent with their inhibitory role in translation
elongation. In contrast, addition of paromomycin did not affect overall expression
levels of the reporter (Figure 23B). This is in agreement with the effect of
paromomycin on translation fidelity. Paromomycin induces misincorporation and
should therefore not directly affect elongation rates.
It has been reported that cells exposed to oxidative stress have reduced thiolation
levels (Nawrot, Sochacka et al. 2011) and urm1∆ cells are sensitive the oxidative
agent diamide. In order to check whether expression of Urm1-dependent codons is
reduced under diamide-induced oxidative stress with tested expression of the
(AAA)10, (CAA)10, (GAA)10 reporters pre-treated with diamide for 30 min. We
observed around 40% decrease of reporter expression in urm1∆ cells compared to
wild-type cells (Figure 23C), showing no additional effect on relative wild-type to
urm1∆ expression when compared to the untreated cells (Table 3). These results show
that diamide-induced oxidative stress does not enhance the effect of lack of thiolation
under the tested conditions.
Although urm1∆ cells are sensitive to paromomycin and diamide, we did not
observe any effect of these drugs on expression of our codon-enriched reporter in
urm1∆ cells, suggesting that these sensitivities are not caused by aggravated effect on
translation. However we cannot rule out that the concentration and treatment time
with the drug were too low to induce detectable changes.
Results
43
Figure 23: Expression of the translation reporter under different drugs.
(A) Expression of the translation reporter after 3h induction with (AAA)10 codon-trap in wild-
type (WT) and urm1∆ cells. Cells were untreated or pre-treated with the translation inhibitors
paromomycin (Paro), cycloheximide (CHX), or hygromycin B (HygroB) for 30 min. Data
show the mean YFP/CFP ration ± standard error of the mean from at least 500 cells plotted as
percentage (%) of wild-type. (B) Same data as in (A) but expressed as percentage (%) of the
untreated wild-type control. (C) Expression of the translation reporter after 3h induction with
(GAA)10, (AAA)10, and (CAA)10 codon-traps in wild-type and urm1∆ cells. Cells were pre-
treated with diamide for 30 min. Data show the mean YFP/CFP ration ± standard error of the
mean from at least 500 cells plotted as percentage (%) of wild-type.
Results
44
Strain Drug treatment Inserted codon-trap % of wild-type
urm1∆ none (AAA)10 60 urm1∆ none (CAA)10 60 urm1∆ none (GAA)10 60 uba4∆ none (CAA)10 60 elp3∆ none (CAA)10 70 urm1∆ none (AAG)10 100 urm1∆ none (CAG)10 75 urm1∆ none (GAG)10 80 urm1∆ none (CAA)5 80 urm1∆ none (GAA)5 75 urm1∆ paromomycin (AAA)10 60 urm1∆ cycloheximide (AAA)10 45 urm1∆ hygromycin B (AAA)10 60 urm1∆ diamide (AAA)10 60 urm1∆ diamide (CAA)10 55 urm1∆ diamide (GAA)10 60
Table 3: Expression of the different translation reporters in different conditions
Table recapitulating the different codon-traps that were used in different mutants and under
different stresses. The average reporter expression relative to the wild-type is indicated in
percentage (%).
2.7 Thiolation enhances A-site binding and peptide bond formation Overexpression of unmodified tRNAs rescues the urm1∆ cells drug sensitivities
(Leidel, Pedrioli et al. 2009), suggesting that thiolation is required to enhance the
affinity of tRNAs to cognate codons. To understand how thiolation affects translation
at the molecular level, Namit Ranjan in a collaborative effort with the lab of Marina
Rodnina in Göttingen, compared the effect of thiolated and unthiolated tRNA on
ribosomal A-site binding and peptide bond formation in vitro. The first step of
decoding requires pairing of the tRNA anticodon with the codon at the ribosomal A-
site. To measure the affinity of thiolated and non-thiolated tKUUU to the A-site we
isolated full length native tRNAs from wild-type and urm1∆ yeast cells. Total tRNAs
isolated from wild-type or urm1∆ yeast cells were incubated with purified lysyl tRNA
synthetase and [14C]-labeled lysine to aminoacylate lysine tRNAs. Then EF-TuGTP,
that specifically binds aminoacylated tRNAs, was added to form the EF-
TuGTP[14C]Lys-tK ternary complex. This complex was then purified from the other
non-aminoacylated tRNAs by size exclusion chromatography. The purified aa-[14C]-
tK ternary complex was mixed with 70S initiation complex purified from E. coli and
Results
45
loaded with f[3H]Met-tM and mRNA. The synthetic mRNA contained an initiator
AUG followed by an AAA codon recognized by f[3H]Met-tM and [14C]-tK
respectively. Binding of [14C]Lys-tK to the AAA codon in the A-site was measured
(Figure 24A). It should be noted that the tKUUU and tKUUC isoacceptors could not be
separated, however tKUUC cannot pair efficiently with the AAA codon. After 10
minutes incubation of ternary complex with initiation complex, ribosomes were
precipitated and unbound tRNAs were washed off. Then [14C]-tK binding to the
ribosome was assessed by liquid scintillation counting. Interestingly, A-site binding
of aa-tK from urm1Δ was decreased by 60% compared to wild-type controls (Figure
24B), demonstrating an important role of tRNA thiolation to enhance cognate codon
binding. A similar effect was seen when using tRNAs extracted from elp3∆ cells that
lack the mcm5 modification (Figure 24B), showing that both modifications are
enhancing A-site binding. To determine the kinetic parameters of this binding
reaction, ribosomal complexes were purified with high [MgCl2], and pept-tK
dissociation induced with low [MgCl2]. The equilibrium dissociation constant, Kd, and
the rate constant of tRNA dissociation (koff) and association (kon) the A-site were
calculated from the dissociation rates. Strikingly, the kon for pept-tK from urm1Δ cells
(0.0052 s-1μM-1) was about three times slower than for wild-type control cells (0.015
s-1μM-1) (Figure 24C).
The final step in decoding results in formation of a peptide bond, we therefore
measured kpep, the rate of ribosome-catalyzed formation of f[H3]Met[14C]Lys-tK, using
quench-flow analysis with rapid mixing of an excess of initiation complex with the
ternary complex (EF-TuGTP[14C]Lys-tK). Dipeptide formation was strongly
reduced with tRNAs from urm1∆ cells with apparent rate constants, kpep, of 1.03 s-1
and 0.223 s-1 using aa-tK from wild-type and urm1Δ, respectively (Figure 24D).
Together, these in vitro experiments demonstrate that thio-modification at the wobble
position stabilizes cognate codon-anticodon interactions at the ribosome, and thereby
enhances the efficiency of translation.
Results
46
Figure 24: tRNA modifications of U34 promote A-site binding and dipeptide formation in
vitro.
(A) Schematic illustration of the decoding and peptide bond formation processes. (B)
Ribosomal A-site binding of [14C]Lys-tK isolated from wild-type (WT), urm1∆, or elp3∆
cells containing tKUUU with mcm5s2U34, mcm5U34, or s2U34 respectively was measured after
incubation of initation complex with ternary complex. Data show the mean [14C] signal ±
standard error of the mean from three experiments plotted as percentage (%) of wild-type. (C)
The rate constants of tRNA dissociation (koff) and tRNA association (kon) during peptidyl-
tRNA formation was measured for [14C]Lys-tK isolated from wild-type or urm1∆ cells. Error
bars represent standard error of the mean. The equilibrium dissociation constant (Kd) is shown
below ± standard error of the mean. (D) The rate of dipeptide formation (Kpep) using tK
isolated from wild-type or urm1∆ cells. Error bars represent standard error of the mean.
These experiments were performed by Namit Ranjan.
Results
47
2.8 URM1-deleted cells are sensitive to paromomycin
Modifications at the wobble position were proposed to restrict the flexibility of this
position and therefore increase discrimination between cognate and near-cognate
substrates. Paromomycin, an aminoglycoside antibiotic, was shown to increase
incorporation of near-cognate codons by structural rearrangements in the ribosome
(Pape, Wintermeyer et al. 2000; Ogle, Carter et al. 2003; Salas-Marco and Bedwell
2005). Similarly lack of thiolation might decrease selectivity toward the cognate
codon and induce misincorporation. To check whether lack of URM1 and the effect of
paromomycin are additive, we tested paromomycin sensitivity of urm1∆ cells.
Deletions of components of the URM1 pathway (urm1∆, uba4∆, ncs2) or components
of the ELP pathway (elp1∆) were spotted in serial dilutions on YPD plates
supplemented with paromoycin, cycloheximide or hygromycin B and incubated at
30°C for 2-3 days. uri1∆ cells were used as positive control. As a side note all the
gene deletions were made by replacement with the nourseothricin resistance gene
(NAT cassette) because the kanamycin resistance cassette (KAN) also confers
resistence to paromomycin. All the mutants were sensitive to paromomycin (Figure
25), suggesting that both the s2 and the mcm5 modifications are enhancing translation
fidelity. In contrast, cells lacking thiolation or mcm5 at Uracil34 were only weakly
sensitive to cycloheximide and hygromycin B, two translation inhibitors affecting
translation elongation (Figure 25). Furthermore, overexpression of unmodified tRNA
tKUUU, tQUUG, tEUUC was able to rescue the paromomycin sensitivity (Figure 25 lower
panel), further supporting the notion that the mcm5s2 modification is enhancing
cognate codon binding.
Results
48
Figure 25: urm1∆ cells are sensitive to paromomycin.
(A) Cell viability of wild-type (WT) cells and cells lacking components of URM1 and the
ELP pathway was assessed by dilution spottings on plates supplemented with different
translation inhibitors. URI1 is involved in translation initiation and was used as a positive
control. (B) Cell viability of wild-type and urm1∆ cells with empty vector or with a vector
containing tKUUU, tQUUG, tEUUC on a 2µ overexpression plasmid.
2.9 Activation of galactose promoter is defective in urm1∆ cells We showed that protein expression of CMS1 and YPL199C is reduced by about
60% in urm1∆ cells. However many of the other candidates tested did not show
significant decrease in expression levels in urm1∆ cells. The quantitative mass
proteomic analysis showed that the protein expression changes are lower than two
fold (Figure 10 and Table 2). We hypothesized that overexpression of the candidate
genes might exacerbate the differences in protein expression. To this end we used
plasmids from the galactose inducible GST-tagged collection. This collection
comprises GST-tagged genes expressed under galactose (GAL) upstream activation
sequence (UAS) on a 2µ plasmid allowing overexpression of the genes upon
galactose addition. Expression of GST-tagged genes was induced with galactose in
Results
49
wild-type and urm1∆ cells and protein expression analyzed by western blot analysis.
Interestingly, we found that protein expression of GST-Rtn1 and GST-Zeo1, two
candidates from the proteomic list of downregulated genes (Table 2), was drastically
reduced in urm1∆ cells compared to wild-type cells (Figure 26). As controls we also
tested GST-Act1 and GST-Ygl082w and we observed a similar decrease of
expression in urm1∆ cells of these genes compared to wild-type cells (Figure 26B),
suggesting that the downregulation is due to the GAL UAS or the GST tag. We also
assessed expression GST-tagged Ygl082w in mutants of other components of the
thiolation pathway. We observed similar decrease in protein expression with GST-
Ygl082w in uba4∆, ncs2∆ and ncs6∆ cells (Figure 26), suggesting that this effect is
due to lack of tRNA thiolation. We then tested whether we could rescue the defect of
galactose-dependent activation by using a ß-estradiol responsive Gal4 variant. This
modified Gal4 variant bypasses the requirement for galactose and activate genes upon
ß-estradiol addition. Induction of the GAL GST-Cms1 and GST-Ypl199c constructs
with ß-estradiol restored wild-type levels of induction in urm1∆ cells (Figure 26).
This shows on one the hand that URM1 is required for efficient induction of galactose
genes and one the other hand that overexpression does not aggravate the differential
expression between wild-type and urm1∆ cells.
Taken together these data imply a role of tRNA modifying enzymes for efficient
activation of GAL genes. Further investigations are required to find the genes
mediating this decrease in galactose gene induction and whether this is due to
differentially translated proteins in urm1∆ cells.
Results
50
Figure 26: URM1 is required for activation of galactose gene induction.
(A) Expression of GST tagged Zeo1 and Act1 after galactose induction was assessed in wild-
type and urm1∆ cells by western blotting. (B) Expression of GST-Rtn1 in wild-type and
urm1∆ cells induced by galactose of by ß-estradiol in a strain containing a modified ß-
estradiol-inducible Gal4 (AD-GEV). (C) Expression of GST-tagged Ygl082w after galactose
induction was assessed in wild-type, urm1∆, uba4∆, ncs2∆, or ncs6∆ cells by western
blotting. (A-C) Pgk1 is used as a loading control.
Discussion
51
3 DISCUSSION
3.1 Uracil34 modifications stabilizes tRNAs at the A-‐site
During elongation, incoming tRNAs bind to the A-site for initial selection and
stably associated tRNAs are then translocated to the P-site for peptide bond
formation. Our in vitro experiments show that URM1 and ELP mediated
modifications at the wobble position stabilize cognate tRNAs at the ribosomal A-site
and promote peptide bond formation. In S. cerevisae, loss of thiolation does not
abolish the presence of the mcm5 modification, and cells lacking either mcm5 or s2 are
viable. However simultaneous loss of mcm5 and s2 is detrimental as elp3∆ urm1∆
double mutant cells are extremely slow growing, suggesting that these U34
modifications cooperatively promote translation elongation. Indeed, previous in vitro
studies showed that a synthesized anticodon stem loop fragment of human LYS3
tRNA, hASLLys3UUU, lacking the mcm5s2, t6 modifications was unable to bind to AAA
and AAG codons at the ribosomal A-site (Ashraf, Sochacka et al. 1999; Yarian,
Marszalek et al. 2000; Vendeix, Murphy et al. 2012).
3.2 Consequences on translation fidelity
Efficient codon anticodon interaction at the ribosomal A-site is the key ensuring
efficiency and fidelity of translation. Altered codon recognition can lead to decreased
proteins synthesis rate, misincorporation and/or frameshift (Pape, Wintermeyer et al.
1999; Baranov, Gesteland et al. 2004). Fidelity of translation relies on the distinction
between cognate and non-cognate tRNAs (Urbonavicius, Qian et al. 2001). The
difference in energy caused by the loss of a single base pair mismatch is not sufficient
to maintain the low error frequency in the cells. Therefore aminoacyl-tRNAs (aa-
tRNA) undergo two selection steps dependent on GTP hydrolysis: initial selection and
proofreading (Gromadski, Schummer et al. 2007). Both steps monitor the half-life of
the tRNA in the decoding site A-site, which depend on the codon-anticodon pairing.
Lack of thiolation decreases stability of the cognate tRNA at the A-site and thereby
decreases the discrimination capacity of the proofreading mechanisms between
cognate and near-cognate tRNAs, potentially leading to misincorporation.
Paromomycin, an aminoglycoside antibiotic, stimulates misreading of near-cognate
tRNAs. The paromomycin sensitivity data and recent studies (Dewez, Bauer et al.
Discussion
52
2008; Maynard, Macklin et al. 2012; Patil, Chan et al. 2012) suggest that thiolation is
indeed important for correct amino-acid incorporation and maintenance of reading
frame. To study how thiolation affects fidelity of translation, further in vitro studies
using a mixture of competing tRNAs or comparison of mRNAs with different codons
are ongoing in our laboratory.
Fidelity of translation can also be impaired by frameshifting. Especially repetitive
sequences such as clusters of AAA codons are known to induce slippage whereby the
whole reading frame is perturbed. In vitro studies using mRNA with slippery
sequences will also allow us to address whether thiolation is required for reading
frame maintenance. The ribosome binding data and the translation reporter results
show that thiolation is required for efficiency of protein synthesis in vitro and in vivo
but they do not allow us to discriminate whether this is solely leading to reduced
translation rate or also frameshifting and misincorporation. Reduced translation rates
of AAA, CAA, GAA, codons could lead in the cell to decreased protein expression
levels of mRNAs containing theses codons.
3.3 Effects of lack of thiolation on expression of AAA, CAA, and GAA rich mRNAs
In the yeast genome, AAA, CAA, and GAA codons that recognize by thiolated
tRNAs are amongst the most abundant codons. Together they constitute 11.5 % of all
codons as measured by the codon usage database (http://www.kazusa.or.jp/codon/cgi-
bin/showcodon.cgi?species=4932). This value is more than two fold higher than the
expected percentage if all codons would be expressed equally (4.7%). Nevertheless
our metabolic labeling experiments, polysome profiles and quantitative proteomics
clearly show that overall protein synthesis levels are unaffected by lack of thiolated
tRNAs under standard growth conditions. The dynamics of translation gives us some
explanation for this. In vivo under non-limiting conditions initiation constitutes the
rate-limiting step of translation, while elongation is very fast (Milon, Konevega et al.
2008; Kudla, Murray et al. 2009). Furthermore Kudla et al tested, in a systematic
study, the effect of synonymous mutations in the sequence of GFP on expression
levels (Kudla, Murray et al. 2009). They found that GFP expression levels were not
associated with codon bias but with mRNA folding, suggesting codon-anticodon
Discussion
53
interactions are not the rate-limiting factor under the tested conditions. Several studies
including ours show that a strong enrichment of codons is required to induce
significant changes in expression (Begley, Dyavaiah et al. 2007; Bauer, Matsuyama et
al. 2012; Chan, Pang et al. 2012). CMS1 and YPL199C containing 17.2% and 16.2%
of thiolation-dependent codons respectively, showed over 50% decrease in protein
levels in urm1∆ cells. We demonstrated that this effect is not dependent on
transcription nor stability and therefore most likely due to differential translation. It
would be interesting to replace the Urm1-dependent codons in CMS1 and YPL199C
into the synonymous G-ending codons and test whether we can restore wild-type
expression levels in urm1∆ cells. This would however imply that URM1 has no effect
on the synonymous G-ending codons, AAG, CAG, and GAG.
3.4 Effects of lack of thiolation on expression of AAG, CAG, and GAG rich proteins
Previous in vitro studies had also implicated thiolation in efficient recognition of
the synonymous G-ending codons, AAG, CAG, and GAG (Yarian, 2002). However
in vivo we did not observe impaired translation of genes enriched for these codons in
urm1∆ cells, suggesting recognition of these codons is not strictly dependent on
URM1. Indeed, in vivo the G-ending codons are recognized by the non-thiolated
tEUUC, tQGUC, tECUC. Furthermore our translation reporter experiments show that a
reporter with an AAG codon-trap is not differentially expressed in urm1∆ cells.
Interestingly translation reporters with CAG and GAG codon-traps were still affected
by lack of thiolation although to a lower extend than their A-ending counterparts.
Efficiency of codon recognition is not only dependent on codon-anticodon binding
affinity but also on tRNA concentration. In vivo AAG codons can be recognized by
both tKUUU and tKUUC, leading to a competition of both iso-acceptors. If we consider
their respective gene copy number, as an estimate for abundance, tKUUC is twice as
abundant as tKUUU (14 versus 7 copies respectively). AAG codons are hardly affected
by lack of thiolation because they are more frequently translated using the non-
thiolated tKUUC. For CAG and GAG the situation is different as the isoacceptor tRNA
are much less abundant. tQUUG and tQCUG have 9 and 1 copy respectively and tEUUC
and tECUC 14 and 2 copies respectively. This explains the translation reporter results
Discussion
54
obtained with the CAG and GAG codon-traps where deletion of URM1 still affects
expression of the reporter compared to wild-type cells, albeit less pronounced than
with the CAA and GAA codon-traps (Table 3).
3.5 Codon context While the quantitative proteomic experiment show a clear enrichment of AAA,
CAA, and GAA-rich genes in the set of downregulated proteins, it did not reveal a
significant correlation between codon frequency and expression levels. In addition
some strongly codon enriched genes did not show the expected protein expression
decrease in urm1∆ cells, suggesting that increased codon frequency alone is not
sufficient. Increasing studies highlight the importance of codon context on translation
efficiency however the exact mechanism is still poorly understood (Cannarozzi,
Schraudolph et al. 2010; Moura, Pinheiro et al. 2011). A study in bacteria revealed a
ribosomal quality control mechanism that detects a wrongly accommodated tRNA
and induces increased probability of a second misincorporaton event (Zaher and
Green 2011). This shows that every codon-anticodon interaction can be influenced by
the previous codon recognition event. Indeed, during elongation the ribosome
contains simultaneously two or three tRNA molecules and each tRNA-ribosome
interaction induces structural rearrangements to the ribosome. It is then likely that two
genes with the same codon frequency but one with clusters of codons and one with
spread codons have different translation rates. There are several measures of condon
context such as the tRNA pairing index and the “codon pair bias”, which aims to
understand how codon pairs are selected (Tats, Tenson et al. 2008). The codon
specific reporter described in this work provides a useful quantitative tool to study
codon context using a synthetic gene but in an in vivo situation. For example, the
codon-traps can be inserted at different positions within the reporter or gaps of
unrelated codons can be inserted between the codons of interest. This would allow to
systematically test sequence patterns. For example, we could test whether clusters of
the codons are affecting translation more than overall codon frequency. Furthermore
as translation is not uniform along mRNAs, we could test whether position of the
codon-traps in the mRNA influences expression levels in cells lacking thiolation.
Discussion
55
3.6 The role of URM1 in rich conditions
Expression levels of all the proteins are regulated at multiple levels and
misregulation are often associated with reduced cell function. The changes in protein
expression induced by lack of URM1 are up to two fold, implying a subtle effect of
thiolation on specific protein synthesis. Analysis of haploinsufficiency in yeast
showed that only about 3% of the genes showed some growth defect when present in
only one copy under standard growth conditions (Deutschbauer, Jaramillo et al.
2005), suggesting that two fold decrease in expression are generally not deleterious to
the cells. This is consistent with urm1∆ cells being viable under standard growth
conditions. However URM1 lacking cells are slightly slow growing on minimal
medium, suggesting that the small reduction of several proteins can affect the cells
under certain conditions. Several proteins are also slightly (up to twofold) upregulated
in urm1∆ cells. These genes were not enriched for specific codons and are therefore
likely indirect consequences of the downregulated genes or compensatory
mechanisms. Under rich growth conditions, we found Cms1 and Ypl199c to be
downregulated of about 50% in urm1∆ cells and quantitative proteomic as well as
western blot analysis of tagged candidates showed that they were amongst the most
strongly downregulated proteins. Both proteins are non-essential and functionally not
characterized, making it difficult to asses the conditions were these genes are required
and analyze the effect of this 50% reduction in urm1∆ cells. However a recent screen
for quinine-sensitive mutants in yeast, revealed that URM1 and ELP pathway
components are required for quinine resistance (Dos Santos and Sa-Correia 2011).
They also found that deletion of CMS1 and other AAA, GAA, CAA enriched genes
are sensitive to quinine, suggesting that thiolation might specifically regulate a set of
genes required for quinine resistance. Further studies will be required to confirm these
data and test for example whether overexpression of Cms1 can rescue the sensitivity
of urm1∆ cells to quinine.
3.7 URM1 involvement in stress regulation
3.7.1 Thiolation and tRNA levels Although urm1∆ cells are viable under standard growth conditions, they are
sensitive to a wide range of drugs, showing that thiolation becomes important under
Discussion
56
stress conditions (Leidel, Pedrioli et al. 2009). Interestingly overexpression of the
unmodified tKUUU, tQUUG, and tEUUC is able to rescue all the phenotypes (Bjork, Huang
et al. 2007), suggesting that the modification is necessary to enhance codon
recognition when tRNA levels are limiting. This implies that thiolation is important to
enhance efficiency of translation and expression of codon-enriched genes under stress
conditions. Under rich conditions tRNA levels are high and most of the tRNAs are
thiolated, however under specific conditions levels of available aminoacylated-tRNA
molecules can drop and under such tRNA limiting conditions it is important that the
remaining available tRNAs are recognizing codons efficiently (Chan, Dyavaiah et al.
2010).
Starvation for example has been shown to reduce aminoacylation of tRNA and
induce trapping of tRNAs in the nucleus (Shaheen and Hopper 2005; Zaborske,
Narasimhan et al. 2009) and interestingly URM1 lacking cells are sensitive to
rapamycin and caffeine, two drugs which induce a starvation response by inhibition of
mTORC1 (Leidel, Pedrioli et al. 2009). Moreover tRNA cleavage has been shown to
decrease tRNA levels under stress conditions such as oxidative stress (Thompson, Lu
et al. 2008) and urm1∆ cells are also sensitive to diamide, an oxidative agent (Leidel,
Pedrioli et al. 2009). Reduction of the concentration of available tRNAs is a control
mechanism of the cell that reduces translation rates and therefore cell growth.
Combined loss of URM1-dependent tRNA thiolation and decreased tRNA levels
could further decrease translation of codon-enriched transcripts in urm1∆ cells.
Deregulation of genes important for specific stress resistance, induced by lack of
thiolation, can lead to drug sensitivity in urm1∆ cells. Indeed overexpression of
hypomodified tK alone is sufficient to rescue sensitivity to diamide, while
overexpression of tE rescues sensitivity to rapamycin and caffeine, suggesting that
misregulation of different subset of genes is responsible for different sensitivities.
Combined, tRNA thiolation and thiolation-dependent codon frequency offer a fast
way of regulating genes to allow cells to adapt to stress conditions such as oxidative
stress and starvation. Thiolation is important for translation of Urm1-dependent
codons and we suggest that this enhancement gets specifically important when tRNA
levels are limiting and translation of specific genes is required. Therefore it allows
fine-tuning of translation under stress conditions when translation of specific genes is
Discussion
57
necessary. Further studies will be required to identify the genes important for specific
stress resistance.
3.7.2 Regulation of tRNA modification levels
While codon-composition of genes is set in a specific species and therefore does
not offer much variability, the proportion of modified versus unmodified tRNA can be
regulated (Figure 27). Recent work showed that the levels of some tRNA
modification are indeed regulated upon different stress conditions (Chan, Dyavaiah et
al. 2010). Furthermore in higher eukaryotes, levels of tRNAs and tRNA modifications
in the anticodon loop seem to be tissue-specifically regulated (Dittmar, Goodenbour
et al. 2006; Brandmayr, Wagner et al. 2012). However, very little known about the
regulation of tRNA modifications and further work quantifying ratio of
thiolated/unthiolated tRNAs under different stress conditions will be necessary. The
observation that ELPC-1 is preferentially expressed in some tissues such as neuronal
cells in Caenorhabditis elegans is in support of such a mechanism. Whether Urm1
expression and thereby tRNA thiolation are also regulated and offers a post-
transcriptional mechanism to differentially express genes in different cell types, needs
further investigations. For example oxidative stress is thought to oxidize tRNAs and
thereby decrease U34 thiolation (Chan, Dyavaiah et al. 2010 ; Nawrot, Sochacka et al.
2011). Interestingly urm1∆ cells are sensitive to the oxidative agent diamide, raising
the possibility that this drug decreases the level of thiolated tRNAs. To test this
hypothesis we could check whether the protein expression levels of Cms1 and
Ypl199c in urm1∆ cells are further decreased after exposure to diamide. However we
did not observe any effect of diamide with our translation reporter, suggesting that
sensitivity to diamide is rather mediated by mistranslated proteins.
Discussion
58
Figure 27: tRNA modifications in stress responses
Scheme depicting how tRNA modification, in red, could influence levels of available tRNAs
in response to stresses and thereby influence translation efficiency.
3.8 Influence of tRNA modifications on each other
On average every tRNA molecule contains seven modified nucleotides. For
example, tKUUU of Saccharomyces cerevisiae is modified on 11 bases. Due to the huge
variety and complexity of modifications and poor experimental accessibility very
little is known about how the different modifications influence each other. While lack
of most modifications is viable, the combined lack of two modifications is often
lethal, showing that they can work cooperatively. Synthetic genetic array analysis
using urm1∆ as bait revealed other tRNA modifying enzymes as genetic interactors
(Leidel, Pedrioli et al. 2009). It would be interesting to study whether some
modifications act as primer or enhancer for other modifications. Some data suggest
that lack of the mcm5 modification on U34 reduces the level of thiolation at the same
position (Esberg, Huang et al. 2006). Interestingly we found that several tRNA
modifying enzymes are differentially expressed upon lack of thiolation. Pus1 and
Trm8 are downregulated while Tum1, Trm112, Pus7, and Smm1 are upregulated,
suggesting that tRNA modifications influence each other. Pus1 and Pus7 are
pseudouridine synthetases, Smm1 is a dihydrouridine synthetase, Tum1 is involved in
Discussion
59
thiolation while Trm8 and Trm112 are part of methyltransferase complexes. They act
at different positions on different tRNAs. Thus it is conceivable that urm1∆ cells
remodel the modification status of other tRNA to compensate for the lack of
thiolation.
3.9 Identification of further regulated proteins
Our quantitative proteomic analysis reflects steady-state protein expression levels
and therefore reflects the situation after some compensation mechanisms have taken
place. Indeed the differentially expressed proteins are enriched for tRNA modifying
and processing enzymes as well as ribosomal components and suggest that the cell
compensate for the lack of thiolation. In addition, it does not allow discrimination
between direct and indirect consequences of lack of thiolation. Generation of a strain
with inducible shut down of URM1 expression, would allow us to analyze direct
effects of lack of thiolation at different time points before compensatory mechanisms
are initiated. While proteomic methods do not differentiate between differentially
transcribed, translated or degraded proteins, other techniques are needed to asses
translation effects specifically. For example, ribosome profiling, a technique
combining RNase treatment of pulled-down ribosomes, with next generation RNA
sequencing examines ribosome footprints, which are the mRNA sequences protected
from RNase digestion by association with the translating ribosome (Ingolia,
Ghaemmaghami et al. 2009). With this technique the specific mRNAs with altered
translation and the ribosome pausing sites can be identified at the codon level. Several
laboratories are currently investigating the ribosomal footprints of thiolation-deficient
yeast cells (posters at conferences and personal communications). Alternatively,
translation of full-length proteins could be investigated in vitro using tRNAs extracted
from wild-type and mutant cells (Burakovsky, Prokhorova et al. 2012). Such in vitro
translation experiments of specific proteins allow determination of the translation
kinetics with modified and unmodified tRNAs.
The effects of lack of thiolation are pleiotropic and many different phenotypes are
observed. It has so far been difficult to link a phenotype with the downregulation of
one specific gene and it is unclear whether downregulation of one single gene is
sufficient or whether tRNA modifications rather control expression of an entire subset
Discussion
60
of genes, which collectively give rise to a phenotype. Alternatively, protein
expression of the affected gene(s) might only be required under specific conditions.
Recent reports suggest that both might be the case (Bauer, Matsuyama et al. 2012;
Chan, Pang et al. 2012). Quantitative proteomic analysis of the whole proteome of
wild-type and urm1∆ cells under different stress conditions, such as diamide and
rapamycin could help address this question. Overexpression of unmodified tKUUU,
tQUUG, and tEUUC rescue all the known urm1∆ phenotypes. Likewise, overexpression
of mRNAs enriched for Urm1-dependent codons could compensate for the impaired
translation. A screen for suppression of urm1∆ drug sensitivity using an
overexpression library could thus be an interesting option to identify URM1-
dependent genes.
Misincorporation of amino acids might also be the source of some of the urm1∆
phenotypes. Decreased translation fidelity might be more deleterious to the cell than
the urm1∆ induced decrease in protein expression. Especially mis-translation of lysine
residues that are often important catalytic or regulatory sites would then impair
protein activity without even affecting protein levels. Likewise, glutamine stretches
inducing aggregation are often found in prion proteins. Aggregation of such prion
proteins is important for stress resistance (Halfmann and Lindquist 2010; Halfmann,
Alberti et al. 2011) and misincorporation events in a poly-Q region could affect their
aggregation properties and lead to increased stress sensitivities.
3.10 Functions of tRNA modifications beyond translation Recently new functions of tRNAs beyond its adaptor role in translation has
emerged. tRNA have been implicated in many different processes such as
immunostimulatory activity, formation of stress granules, mRNA silencing, cell wall
biosynthsis (Emara, Ivanov et al. 2010; Burroughs, Ando et al. 2011; Dare and Ibba
2012; Gehrig, Eberle et al. 2012) and these functions are often modulated by
modifications. Interestingly in Leishmania tarantolae all mitochondrial tRNAs are
imported from the cytoplasm and U34 thiolation is thought to be a negative regulator
for mitochondrial transport (Kaneko, Suzuki et al. 2003). Yeast mitochondrial DNA
encodes for most of the mitochondrial tRNAs and have a distinct thiolation pathway,
involving MTU1, however it is not clear whether some trafficking from the cytoplasm
Discussion
61
to the mitochondria is still taking place. Current work in our laboratory is analyzing
the mitochondrial thiolation pathway and possible interactions with the cytoplasmic
URM1 pathway. Preliminary data indicate that urm1∆ mtu1∆ cells are synthetic sick
(data not shown), implying a connection between both pathways.
3.11 Other functions of Urm1
URM1 was originally characterized as an ubiquitin-like protein modifier that is
conjugated to target protein via a process called urmylation (Furukawa, Mizushima et
al. 2000; Goehring, Rivers et al. 2003; Van der Veen, Schorpp et al. 2011). The effect
on translation investigated in this work is unlikely to be caused by urmylation. First
we can largely exclude that the differential expression of Cms1 and Ypl199c in
urm1∆ cells is mediated by urmylation as we excluded that degradation causes the
changes in protein expression levels, if urmylation was to be involved in degradation,
and we did not observe appearance of high molecular weight conjugates by western
blotting. Secondly our codon-specific translation reporter is a synthetic protein
unlikely to be a target of urmylation. Finally in the in vitro assay we can also exclude
that urmylation of components of the translation machinery is involved as we used
wild-type bacterial ribosomes. Furthermore, we did not identify any higher molecular
weight bands containing Urm1 conjugates in pull-down of Urm1 under several tested
conditions (data not shown). However, we cannot rule out that urmylation is
influencing the levels of thiolation. For instance, conjugation of Urm1 to target
protein might “sequester” Urm1 molecules, thereby decreasing tRNA thiolation levels
and negatively regulating thiolation. Alternatively urmylation might positively
influence translation as some components of the thiolation machinery were proposed
to be urmylated and could keep components of the thiolation pathway in close
proximity.
Urm1 was shown to be a sulfur carrier in the tRNA thiolation pathway and it is
tempting to speculate that Urm1 might work as such in other sulfuration reactions.
Indeed, nucleotide modifications are not limited to tRNAs and are widespread. Some
RNA modifying enzymes targets different RNA substrates (Massenet, Motorin et al.
1999; Benitez-Paez, Villarroya et al. 2012; Sardana and Johnson 2012). It is not
known whether Urm1 is acting as a sulfur carrier for other nucleotide modifications.
Discussion
62
As the Ncs2-Ncs6 heterocomplex is thought to account for specificity to tRNA in the
URM1 pathway, Urm1 might be implicated in other processes. Advances in RNA
mass spectrometry may in the future identify an exhaustive list of all the RNA
modifications and identify the enzymes involved. The results of this study are
,however, unlikely due to other RNA thiolation events. First overexpression of the
three unmodified tKUUU, tQUUG, tEUUC were able to restore wild-type protein expression
levels for most of the differentially expressed proteins. Furthermore all the urm1∆
phenotypes characterized so far are rescued by overexpression of unmodified tEUUU,
tQUUG, and tEUUC, suggesting that the main function of Urm1 is to thiolate these three
tRNAs. Secondly the in vitro translation experiments involved several purification
steps to enrich for tK and exclude other RNA.
3.12 Concluding remarks In this work we investigated the role of tRNA thiolation at the wobble uridine on
translation. We found that thiolation is important for the efficient recognition of
cognate codons and that this is important to enhance the protein expression of a subset
of AAA, CAA, and GAA-rich genes. This work and previous studies showed that
thiolation is required for resistance to a wide-range of stresses and expression of
galactose-induced genes. Available evidence suggests that tRNA modifications at the
wobble position more generally modulate protein expression in a codon-specific
manner, enhancing cellular function under specific stresses. Different wobble
modifications can thereby fine-tune translation of different sets of genes and might
provide new means of genetic control.
We suggest a model in which under rich conditions when tRNA levels are high and
non-limiting thiolation is dispensable, however under stress conditions when tRNA
levels drop and become limiting, thiolation is important to increase the efficiency of
the remaining molecules (Figure 28).
tRNA thiolation at the wobble Uracil34 is conserved from bacteria to humans
(Schlieker, Van der Veen et al. 2008; Leidel, Pedrioli et al. 2009), suggesting an
important function in translation. Interestingly as consequence of higher GC content
of their genome, higher eukaryotes have a different codon bias. Thus thiolation might
Discussion
63
regulate different sets of genes in different organisms. With over 90 different
modifications, tRNAs offer virtually unlimited possibilities to fine-tune protein
expression in a cell-type specific and stress-dependent manner.
Figure 28: Regulation of translation of specific genes by tRNA modifications
Model showing how under rich conditions tRNA concentration is very high and protein
expression levels (blue ellipses and black squares) are independent of tRNA levels. However
under stress conditions tRNA concentration is limiting and tRNA modification, red dot, is
necessary to enhance synthesis of specific proteins.
Materials and Methods
64
4 MATERIALS AND METHODS
4.1 [35S] metabolic labeling Exponentially growing yeast cells in SD-MET were labeled with [35S] methionine
and cysteine (Hartmann analytics) for 15 min., followed by a 5 min. chase with cold
methionine and cysteine at 30°C. Control cells were treated with 100 µM
cycloheximide just before the labeling. Cells were fixed with 10% TCA on ice for 10
min. Protein pellet was washed with acetone and resuspended in Urea/SDS buffer
(8M urea, 5% SDS, 0.8 M β-mercaptoethanol, 10% glycerol) and then diluted 100
times. Counts per minutes (CPM) were measured in a liquid scintillation counter. In
parallel, protein extracts were run on a 12% SDS-PAGE and fixed in fixation solution
(10% methanol, 10% acetic acid) for 1h. Signal was enhanced by incubation with
en3hance (Perkin Elmer) for 1h and subsequently washed with 70% PEG (MW 8000)
solution. Gel was dried on a Whatmann paper with a vacuum gel dryer (Hoefer) at
80°C for 1h, exposed on a phosphor imager overnight and scanned with a STORM
phosphorimager.
4.2 Ribosome extraction
Ribosomes were extracted from 200 ml exponentially growing cells. Translation
was stopped with 20 µg cyclohexamide and incubation on ice for 10 min. while
regulary mixing. Cells were washed twice with cold lysis buffer (10 mM HCl pH 7.5,
100 mM NaCl, 30 mM MgCl2, 100 µg/ml cyclohexamide, 200 µg/ml Heparin, 1:700
DEPC, freshly prepared) and resuspent in 800 µl lysis buffer. Yeast cells were broken
by vortexing with 300 µl small acid washed beads (0.45 µM ø) 8 times for 30 sec.
with 30 sec. breaks on ice in between. Lysate was transferred to a new tube and beads
washed with 200 µl lysis buffer to collect all the residual lysate. Cell debris were
removed by pelleting at 16’000 g for 10 min. at 4°C. RNA content was assessed by
OD 260. For storage at -80°C lysate was supplemented with 100 µl 70 % glycerol,
frozen in liquid nitrogen.
Materials and Methods
65
4.3 Polysome profiles
Polysomes were separated on a 6 to 45% sucrose gradient. 8 µl OD260 of extracted
ribosomes were added on top of the sucrose gradient in polyallomer Beckman tube
(cat. 331372, 14 × 89 mm) and ultra-centrifuged at 39’000 rpm for 2:45 hours at 4°C
with SW41 swing bucket rotor. 60 % sucrose was pumped into the bottom of the tube
with a needle. Ribosome content was measured and recorded with a teledyne ISCO
UA-6 detector. Polysome profiles were scanned and the surface under the curves
analyzed with imageJ.
4.4 Hot phenol extraction Total RNA was extracted from 10-50 ml exponentially growing cells. After
washing with water, cells were resuspent in 800 µl 1:1 TES phenol solution and
vigorously vortexed for 10 sec. mRNA extraction was performed at 65°C for 1h with
shaking. Then samples were put on ice for 5 min. before centrifugation for 5 min. at
16’000 g at 4°C. The aqueous phase was transferred in a new tube and mixed by
vigorously vortexing for 10 sec. with 400 µl phenol. After centrifugation as above, the
aqueous phase was transferred in a new tube. The extraction was repeat as above with
400 µl chloroform. 40 µl 3M sodium acetate pH 5.3 and 1 ml cold ethanol was added
to precipitate RNA and pelleted by centrifugation at 4°C 16’000 g for 5 min. RNA
pellet was subsequently washed with 1 ml 70% cold ethanol. After drying the pellet
was resuspent in 100 µl RNase free water and the RNA concentration measured with
the OD260.
4.5 Gene ontology (GO) enrichment analysis
The set of upregulated and downregulated genes were analyzed using the SGD
Gene Ontology Slim Mapper. The processes enriched more than two fold compared to
the frequency of genes belonging the category in the whole genome were selected.
Materials and Methods
66
4.6 Quantitative RT-PCR
Total RNA was prepared using the hot phenol method and then treated with
deoxyribonuclease I (Qiagen,) on a RNeasy column. It was then reverse-transcribed to
synthesize the first- strand cDNA using Superscript (Qiagen). Quantitative RT–PCR
was performed using the LightCycler FastStart DNA Master SYBR Green I kit
(Roche, Basel, Switzerland) and the data were analysed using the LightCycler Quick
System 350S (Roche). Quantification was performed using the relative standard curve
method, and the transcriptional level of PGK1 mRNA was used for the normalization
process. The results were expressed as the relative transcriptional level, and the values
obtained from wild-type cells were set to 100%. The primers used are listed in Table
xx. Each gene was amplified by PCR, combining two primers labeled as F and R, e.g.
the PGK1 cDNA was amplified with PGK1-F and PGK1-R.
4.7 Cloning of the dual-fluorescent translation reporter
Quadruple CFP under GAL promoter and with a CYC1 terminator was cloned into
pRS305 with EcoRI and PmeI. Quadruple venus was cloned with SalI NotI restriction
site and SphI and SpeI sites were added after the starting ATG to allow insertion of
codon-traps. Codon-traps insertion generated loss of the SpeI site.
4.8 Expression of fluorescent reporter
Expression of the translation reporter was induced with 50 nM β-estradiol (sigma)
for 3h or 6h for the time-course and translation was blocked with 100 µg/ml
cycloheximide. Cells were imaged after 1h to allow maturation of the fluorophores.
DIC, CFP and YFP images were taken at every position with an inverted epi-
fluorescence microscope (Ti-Eclipse, Nikon) controlled by micromanager with a 40×
oil objective in an incubation chamber at 30°C. YFP exposure time was adjusted
dependent on the codon-trap. (AAA)10, (GAA)10, (CAA)10, and (AAG)10 traps were
imaged with 50 msec, 200 msec, and 60 msec exposure, respectively, in WT and
urm1∆ cells. CFP images were all taken with 400 msec exposure. Single cells were
segmented and analyzed with CellQuant program (Pelet, Dechant et al. 2012).
YFP/CFP ratios were measured in single cells and the mean over a cell population
Materials and Methods
67
calculated with MATLAB and compared to wild-type expression levels. Mean
YFP/CFP ratio in wild-type was set to 100%.
4.9 Cycloheximide chase
TAP-tagged strains grown in logarithmic phase were treated with 100 µM
cycloheximide and samples of equal volume taken at different times after the
translation block. Cells were incubated in 10% TCA for 10 min. on ice, washed with
acetone and resuspended in urea loading buffer.
4.10 Western blotting
Whole cell extracts loaded on 12% acrylamide SDS-PAGE were run at 220V for
45 min and then transferred on nitrocellulose for 1h15 at 120V in a semi-dry
apparatus. After blocking with 5% milk, nitrocellulose membranes were probed with
PAP (Sigma) at a concentration of 1:2500, anti-Urm1 (Abcam) at a concentration of
1:200 or anti-GST (Sigma) a concentration of 1:2500 in 5% milk and washed with
Phosphate buffer saline tween solution (PBST). Loading was controlled using anti-
Pgk1 antibody (invitrogen) at a concentration of 1:2500 in 5% milk and washed with
PBST. For quantification exposed photographic films were scanned and analyzed
with ImageJ.
4.11 Galactose induction
Wild-type and urm1∆ cells with GST-tagged proteins were grown in exponential
phase in Sraffinose-URA and induced by addition of 2% galactose. Cells were
harvested after 3h induction and proteins extracted with 10% TCA.
4.12 Drug sensitivity assay Exponentially growing cells were diluted to OD600 0.6. Six serial 5 fold dilutions
were spotted on YPD plates with or without drug. Plates were incubated for 3-4 days
at 30°C and imaged. The drug concentration used on plates were 1 mg/ml
Materials and Methods
68
paromomycin (Sigma), 0.7 µg/ml cycloheximide (Sigma) or, 0.08 mM Hygromycin B
(Sigma).
4.13 SILAC labeling
SILAC strains were exponentially grown in SD-K-R supplemented with unlabelled
K, R or 13C6,15N2-K (Sigma) and 13C6,15N4-R (Sigma). Equal ODs of light and heavy
culture were mixed and subsequently processed together.
4.14 Protein Extraction and digestion for MS
Cells were lysed with 1.85 M NaOH, 7.6% β -mercaptoethanol, then proteins
precipitated with 50% TCA and washed twice with acetone. Pellets were resuspended
in urea-buffer (8 M urea, 50 mM NH4HCO3, 0.5% RapiGest (Waters)) or a SDS-
buffer (5% SDS, 50 mM NH4HCO3, 10 mM DTT) at 56 °C for 30 min. Then proteins
were treated with 25 mM iodoacetamide at RT for 30 min. Proteins in urea-buffer
were diluted ten times and digested overnight with Trypsin (Promega). Proteins in
SDS-buffer were digested using the FASP method (Jacek R Wiśniewski, nature
methods 2009). Peptides were purified on C18 spin columns (The Nest Group).
4.15 Strong cation exchange (SCX) fractionation
SCX fractionation was performed on microspin PolySULFOETHYL Aspartamide
columns (The Nest Group). Peptides were eluted with a six-step NaCl fractionation
(50 mM, 100 mM, 150 mM, 200 mM, 400 mM and 800 mM). Flow through and
fractions were cleaned-up on C18 MicroSpin columns (The Nest Group) and dried in
a vacuum dryer.
4.16 Peptide purification and iso-‐electric focusing
Purified peptides were fractionated by iso-electric focusing on the OffGel
Fractionator (Agilent, G3100AA) according to the manual of the High Res Kit, pH 3–
10 (Agilent, 5188-6424), except the strips were exchanged by either Immobiline
DryStrip pH 3-11 NL, 24 cm (GE Healthcare, 17-6003-77) or pH 3-11 NL, 13 cm
(GE Healthcare 17-6003-75), and ampholytes were substituted by IPG Buffer pH 3-
Materials and Methods
69
11 NL (GE Healthcare, 17-6004-40) used 2%. Peptides were focused into 24 fractions
for 50 kVh at a maximum current of 50 μA, maximum voltage of 8000 V and
maximum runtime of 100 h. Each fraction was acidified with 1% (v/v) CF3COOH,
purified on C18 MicroSpin columns (The Nest Group) and dried in a vacuum dryer.
4.17 LC-‐MS/MS
Dried peptides were resuspended in 0.1% CF3COOH for the LC-MS/MS analysis.
Split-free Easy nLC chromatography system (Proxeon) was used for the online
reverse phase (C18 silica) liquid chromatography. Fused silica columns of 20 cm
length (PicoFrit columns, PF-360-75-10-N-5, New Objective) were packed with C18
silica beads (Magic C18, 200 Å, 3 μm, Michrom Bioresources). A 250 ηl/min
gradient of buffer B (0.08 % (v/v) HCOOH, 90 % (v/v) CH3CN): buffer A (0.1 %
(v/v) HCOOH, 2 % (v/v) CH3CN) ranging from 2% to 35% over 170 min was used to
resolve peptides. The chromatography set up was directly coupled to the mass
spectrometer (LTQ-Orbitrap Velos, Thermo Finnigan) configured for the top-15 data
dependent acquisition (DDA) by collision-induced fragmentation (CID) or top-8
DDA for the higher-energy collisional dissociation (HCD). FT-MS resolution was set
at 60,000.
4.18 Protein identification and quantitation
RAW data files were converted to the mzXML format (Pedrioli, Eng et al. 2004)
and searched against the Saccharomyces Genome Database protein database using
X!Tandem (Craig and Beavis 2004) with the K-score plug-in (MacLean, Eng et al.
2006), OMSSA (Geer, Markey et al. 2004) and Mascot (Matrix Science) and
SEQUEST (University of Washington, license#). Search parameters used were
carboxyamidomethylation (57.022 Da) of Cys as static modification, 13C6,15N2-Lys
(8.01419892 Da), 13C6,15N4-Arg (10.008252778) and oxidation of Met (15.99491463
Da) as variable modifications, semi tryptic digestion with a maximum of two missed
cleavages, 25 ppm and 0.4 Da error tolerances for MS/MS and MS, respectively.
Peptide probabilities were evaluated with PeptideProphet (Nesvizhskii, Keller et al.
2003), iProphet was used to integrate the results from the four peptide search engines
Materials and Methods
70
(Shtynberg submitted) and ProteinProphet (Keller, Nesvizhskii et al. 2002) was used
to estimate protein probabilities. Protein abundance ratios were computed as L/H
(light/heavy) using XPRESS (Han, Eng et al. 2001). Finally proteins were filtered for
1% FDR.
4.19 Data normalization and statistical analysis of differential abundance
Results were stored into an in-house developed database (manuscript in
preparation). From there, protein abundance ratios were imported into R (version no.
2.11.0) (Team 2010). Proteins quantified in only one biological replicate and less than
two peptides were filtered out. log2 of ratios were median normalized using the
preprocessCore library (version 1.8) from the Bioconductor project (version 2.5)
(Bolstad, Irizarry et al. 2003). Statistical analysis of the differential abundance of
proteins was done with the Bayes’ moderated t-test using the LIMMA package
(version 3.8.2) of the Bioconductor project (version 2.8) (Smyth 2004). Proteins ratios
were filtered at the FDR threshold of 5% (or adjusted p-value = 0.05).
4.20 Random Forrest Analysis
Significantly changing proteins were split into two classes and the abundance of
codons that best predicted class membership was extracted by machine learning in R
using the random forest implementation of the party package (version 0.9-99992).
4.21 Total tRNA preparation
Total tRNA was extracted from 18 L exponentially growing cells with 35% acid
phenol (Roti-Aqua-Phenol) for 3 h RT. The aqueous phase was collected by
centrifugation at 4000 rpm for 20 min. The extraction was repeated by adding equal
amount of water. RNA was precipitated with 2% potassium acetate (pH 5.0) and cold-
ethanol for 2 h at -20°C. Pellet was resuspended in 1 M NaCl and mixed vigorously
for 2.5 h at 4°C. The supernatant was collected, and extraction was repeated with 1 M
NaCl. Crude tRNA was precipitated with ethanol. The pellet was dissolved in 0.3 M
sodium acetate (pH 7.0) and stirred vigorously at 4°C. 0.4 volume of isopropanol
Materials and Methods
71
were added drop-wise and the temperature of the mixture was gradually increased at
RT. Supernatant was collected by centrifugation at 4000 rpm for 20 min, and
extraction was repeated with 0.2 volume of isopropanol. To the supernatant 0.4
volume of isopronanol was added and incubated for 2 h at -20°C. The tRNA pellet
was dissolved in water and purified with anion exchange column (DE52, Whatman)
with a 0 to 1 M NaCl gradient. The fractions were pooled and precipitated with
ethanol. The final tRNA pellet was dissolved in water.
4.22 Lysine tRNA synthetase cloning and purification
The Saccharomyces cerevisiae gene for Lysine tRNA synthetase was cloned with
NdeI/BamHI into a modified version of pPROEx vector (Invitrogen), in which EheI
site is replaced by NdeI site. Synthetase expression was induced with IPTG in E. coli
BL21 at 20°C. The protein purified by Ni2+ affinity chromatography and cleaved by
TEV protease (Invitrogen). Purified synthetase was stored in 20 mM HEPES pH7.5,
50 mM NaCl, 1 mM EDTA, 1 mM DTT, 10% Glycerol at -80°C.
4.23 Aminoacylation and purification of [14C]Lys-‐tK
53 A260 units total tRNAs were aminoacylated with 3% purified lysine tRNA
synthetase (LysRS) in aminoacylation buffer (50 mM HEPES pH 7.5, 70 mM NH4Cl,
30 mM KCl, 11 mM MgCl2, 3 mM ATP, 2 mM β-mercaptoethanol) supplemented
with 20 μM [14C]Lys. After incubation for 45 min at 37°C, His6-EF-Tu·GTP was
added for 1 min at RT. EF-Tu·GTP·[14C]Lys-tK was purified using Ni2+ column
(Protino Ni-IDA), proteins were removed with phenol, and aa-tRNA was precipitated
with 2% potassium acetate (pH 5.0) and cold ethanol.
4.24 Biochemical and kinetic assays
For 70S initiation complexes, ribosomes (1.6 μM) were incubated with a fourfold
excess of mRNA with AAA codon following the AUG initiation in presence of 2.4
μM initiation factors IF1, IF2, IF3, 3.2 μM f[3H]Met-tRNAfMet, and 1 mM GTP in
buffer A (50 mM Tris-HCl pH 7.5, 70 mM NH4Cl, 30 mM KCl, 7 mM MgCl2) for 30
Materials and Methods
72
min at 37°C. Ternary complex, EF-Tu·GTP·[14C]Lys-tK, was prepared by incubating
3 μM EF-Tu with 1 mM GTP, 3 mM phosphoenol pyruvate, 0.1 mg/mL pyruvate
kinase for 15 min at 37°C, followed by addition of 1 μM [14C]Lys-tK for 1 min at RT.
Ternary complex was added to the initiation complex and incubated for 10 min at RT.
The amount of [14C]Lys and [3H]Met bound to ribosomes was determined by
nitrocellulose filtration.
For 70S initiation complexes, ribosomes (1.6 μM) were incubated with a fourfold
excess of mRNA (GGCAAGGAGGUAAAUA AUG AAA UUC GUU AC) in
presence of 2.4 μM initiation factors IF1, IF2, IF3, 3.2 μM f[3H]Met-tRNAfMet, and 1
mM GTP in buffer A (50 mM Tris-HCl pH 7.5, 70 mM NH4Cl, 30 mM KCl, 7 mM
MgCl2) for 30 min at 37°C. Ternary complex, EF-Tu·GTP·[14C]Lys-tK, was prepared
by incubating 3 μM EF-Tu with 1 mM GTP, 3 mM phosphoenol pyruvate, 0.1 mg/mL
pyruvate kinase for 15 min at 37°C, followed by addition of 1 μM [14C]Lys-tK for 1
min at room temperature (RT). Ternary complex was added to the initiation complex
and incubated for 10 min at RT. The amount of [14C]Lys and [3H]Met bound to
ribosomes was determined by nitrocellulose filtration.
For dissociation of fMetLys-tK from the A site, 4 μM ribosomes, 6 μM initiation
factors (IF1, IF2, IF3), and 12 μM [14C]Lys-tK was used for initiation complex and
ternary complex, respectively. Ternary complex was incubated with initiation
complex for 1 min at RT to form pretranslocation complex. Then, the Mg2+
concentration of the pretranslocation complex was adjusted to 21 mM to prevent
premature drop-off of fMetLys-tK from the A-site and the complex was kept on ice.
Pretranslocation complexes were purified by size exclusion chromatography
(BioSuite 450 HR, Waters) in buffer A. fMetLys-tK was dissociated from the A-site
with 7 mM Mg2+ at 37°C, and the amount of pept-tRNA bound to the A site at
different time points was determined by nitrocellulose filtration.
To measure the time courses of peptide bond formation, quench–flow assays were
performed at 24°C in a KinTek RQF-3 apparatus. Initiation complex (2 μM) was
rapidly mixed with ternary complex (0.6 μM). After different incubation times,
reactions were stopped with KOH (0.8 M), incubated for 30 min at 37°C, neutralized,
and dipeptides analyzed by RP-HPLC (Katunin, Muth et al. 2002)
Materials and Methods
73
Table 4: strains used in this study
Strain Genotype Derived from
Source yVR1 WT BY4741 Euroscarf
yVR2 urm1::KAN BY4741 This study
yVR213 CMS1-TAP HIS3 BY4741 Euroscarf
yVR227 CMS1-TAP HIS3; urm1::NAT BY4741 This study
yVR210 YPL199C-TAP HIS3 BY4741 Euroscarf
yVR224 YPL199C-TAP HIS3; urm1::NAT BY4741 This study
yVR52 AD-GEV TRP; (AAA)10 reporter URA W303 This study
yVR59 AD-GEV TRP; (AAA)10 reporter URA, urm1::G418 W303 This study
yVR101 AD-GEV TRP; (AAA)10 reporter URA, uba4::G418 W303 This study
yVR161 AD-GEV TRP; (AAA)10 reporter URA, elp3::G418 W303 This study
yVR50 AD-GEV TRP; (CAA)10 reporter URA W303 This study
yVR57 AD-GEV TRP; (CAA)10 reporter URA, urm1::G418 W303 This study
yVR54 AD-GEV TRP; (GAA)10 reporter URA W303 This study
yVR75 AD-GEV TRP; (GAA)10 reporter URA, urm1::G418 W303 This study
yVR74 AD-GEV TRP; (AAG)10 reporter URA W303 This study
yVR78 AD-GEV TRP; (AAG)10 reporter URA, urm1::G418 W303 This study
yVR162 MET15+, urm1::NAT BY4741 This study
yVR163 MET15+ BY4741 This study
yVR37 urm1::NAT BY4741 This study
yVR36 uba41::NAT BY4741 Sebastian Leidel
yVR elp1::NAT BY4741 Sebastian Leidel
yVR ncs2::NAT BY4741 Sebastian Leidel
yVR8 uri1::NAT BY4741 Anna Deplazes
yKT1 WT silac SILAC strain Gustav Ammerer
yKT2 urm1::NAT silac SILAC strain Kshitiz Tygi
BY4741 Genotype: MATa, his3Δ1, leu2Δ0, LYS2, met15Δ0, ura3Δ0
W303-1A Genotype: MATa, leu2-3,112, trp1-1, can1-100, ura3-1, ade2-1, his3-11,15
SILAC strain Genotype: MATa, trp-, CAN1+, lys1::KAN, lys2::KAN, arg4::KAN
Materials and Methods
74
Table 5: Plasmids used in this study
Plasmid Integration Derived from Source pVR20 pRS315-URM1 pRS315 This study
pVR22 AD-GEV (Louvion, Havaux-Copf et al. 1993)
pVR53 (AAA)10 Fluorescent reporter pVR This study
pVR54 (CAA)10 Fluorescent reporter pRS306 This study
pVR52 (GAA)10 Fluorescent reporter pRS306 This study
pVR62 (AAG)10 Fluorescent reporter pRS306 This study
pSZ63 tKUUU pRS425 (Leidel, Pedrioli et al. 2009)
pSL61 tQUUG pRS425 (Leidel, Pedrioli et al. 2009)
pSL62 tEUUC pRS425 (Leidel, Pedrioli et al. 2009)
pSL67 tKUUU , tQUUG , tEUUC pRS425 (Leidel, Pedrioli et al. 2009)
pVR76 GAL GST-RTN1 URA 2µ vector Euroscarf
pVR77 GAL GST- YGL082W URA 2µ vector Euroscarf
Table 6: Primers used for quantitative PCR
Primer name Primer sequence 5’-3’
CMS1-F AGATGATGGACTCGCCTATGA
CMS1-R CATCCTTGGCGTCAAAAATTA
YPL199C-F GCAGATGAAGCGTATAAGAAAAGA
YPL199C-R TTGATAAGCGGTTTGCGATT
URM1-F GAAGATCCTGTCACAGTGGGCGA
URM1-R CGAGCTCCCAATCGGTGTCGT
ACT1-F TCCGTCTGGATTGGTGGT
ACT-R TGAGATCCACATTTGTTGGAAG
PGK1-F ATCAACGATGCCTTCGGTA
PGK1-R CAAGTCGAAACCGACCATAGA
References
75
5 REFERENCES
Alexandrov, A., I. Chernyakov, et al. (2006). "Rapid tRNA decay can result from lack of nonessential modifications." Mol Cell 21(1): 87-‐96.
Ashraf, S. S., E. Sochacka, et al. (1999). "Single atom modification (O-‐-‐>S) of tRNA confers ribosome binding." RNA 5(2): 188-‐94.
Baranov, P. V., R. F. Gesteland, et al. (2004). "P-‐site tRNA is a crucial initiator of ribosomal frameshifting." RNA 10(2): 221-‐30.
Bauer, F., A. Matsuyama, et al. (2012). "Translational Control of Cell Division by Elongator." Cell Rep 1(5): 424-‐433.
Begley, U., M. Dyavaiah, et al. (2007). "Trm9-‐catalyzed tRNA modifications link translation to the DNA damage response." Mol Cell 28(5): 860-‐70.
Benitez-‐Paez, A., M. Villarroya, et al. (2012). "The Escherichia coli RlmN methyltransferase is a dual-‐specificity enzyme that modifies both rRNA and tRNA and controls translational accuracy." RNA 18(10): 1783-‐95.
Bjork, G. R., B. Huang, et al. (2007). "A conserved modified wobble nucleoside (mcm5s2U) in lysyl-‐tRNA is required for viability in yeast." RNA 13(8): 1245-‐55.
Bolstad, B. M., R. A. Irizarry, et al. (2003). "A comparison of normalization methods for high density oligonucleotide array data based on variance and bias." Bioinformatics 19(2): 185-‐93.
Brandmayr, C., M. Wagner, et al. (2012). "Isotope-‐Based Analysis of Modified tRNA Nucleosides Correlates Modification Density with Translational Efficiency." Angew Chem Int Ed Engl.
Burakovsky, D. E., I. V. Prokhorova, et al. (2012). "Impact of methylations of m2G966/m5C967 in 16S rRNA on bacterial fitness and translation initiation." Nucleic Acids Res 40(16): 7885-‐95.
Burroughs, A. M., Y. Ando, et al. (2011). "Deep-‐sequencing of human Argonaute-‐associated small RNAs provides insight into miRNA sorting and reveals Argonaute association with RNA fragments of diverse origin." RNA Biol 8(1): 158-‐77.
Butler, A. R., J. H. White, et al. (1994). "Two Saccharomyces cerevisiae genes which control sensitivity to G1 arrest induced by Kluyveromyces lactis toxin." Mol Cell Biol 14(9): 6306-‐16.
Cannarozzi, G., N. N. Schraudolph, et al. (2010). "A role for codon order in translation dynamics." Cell 141(2): 355-‐67.
Chan, C. T., M. Dyavaiah, et al. (2010). "A quantitative systems approach reveals dynamic control of tRNA modifications during cellular stress." PLoS Genet 6(12): e1001247.
Chan, C. T., Y. L. Pang, et al. (2012). "Reprogramming of tRNA modifications controls the oxidative stress response by codon-‐biased translation of proteins." Nat Commun 3: 937.
Chen, C., B. Huang, et al. (2011). "Elongator complex influences telomeric gene silencing and DNA damage response by its role in wobble uridine tRNA modification." PLoS Genet 7(9): e1002258.
References
76
Chen, C., S. Tuck, et al. (2009). "Defects in tRNA modification associated with neurological and developmental dysfunctions in Caenorhabditis elegans elongator mutants." PLoS Genet 5(7): e1000561.
Craig, R. and R. C. Beavis (2004). "TANDEM: matching proteins with tandem mass spectra." Bioinformatics 20(9): 1466-‐7.
Creppe, C., L. Malinouskaya, et al. (2009). "Elongator controls the migration and differentiation of cortical neurons through acetylation of alpha-‐tubulin." Cell 136(3): 551-‐64.
Dare, K. and M. Ibba (2012). "Roles of tRNA in cell wall biosynthesis." Wiley Interdiscip Rev RNA 3(2): 247-‐64.
Deutschbauer, A. M., D. F. Jaramillo, et al. (2005). "Mechanisms of haploinsufficiency revealed by genome-‐wide profiling in yeast." Genetics 169(4): 1915-‐25.
Dewez, M., F. Bauer, et al. (2008). "The conserved Wobble uridine tRNA thiolase Ctu1-‐Ctu2 is required to maintain genome integrity." Proc Natl Acad Sci U S A 105(14): 5459-‐64.
Dittmar, K. A., J. M. Goodenbour, et al. (2006). "Tissue-‐specific differences in human transfer RNA expression." PLoS Genet 2(12): e221.
Dittmar, K. A., M. A. Sorensen, et al. (2005). "Selective charging of tRNA isoacceptors induced by amino-‐acid starvation." EMBO Rep 6(2): 151-‐7.
Dos Santos, S. C. and I. Sa-‐Correia (2011). "A genome-‐wide screen identifies yeast genes required for protection against or enhanced cytotoxicity of the antimalarial drug quinine." Mol Genet Genomics 286(5-‐6): 333-‐46.
El Yacoubi, B., M. Bailly, et al. (2012). "Biosynthesis and Function of Posttranscriptional Modifications of Transfer RNAs." Annu Rev Genet.
Emara, M. M., P. Ivanov, et al. (2010). "Angiogenin-‐induced tRNA-‐derived stress-‐induced RNAs promote stress-‐induced stress granule assembly." J Biol Chem 285(14): 10959-‐68.
Esberg, A., B. Huang, et al. (2006). "Elevated levels of two tRNA species bypass the requirement for elongator complex in transcription and exocytosis." Mol Cell 24(1): 139-‐48.
Fichtner, L. and R. Schaffrath (2002). "KTI11 and KTI13, Saccharomyces cerevisiae genes controlling sensitivity to G1 arrest induced by Kluyveromyces lactis zymocin." Mol Microbiol 44(3): 865-‐75.
Furukawa, K., N. Mizushima, et al. (2000). "A protein conjugation system in yeast with homology to biosynthetic enzyme reaction of prokaryotes." J Biol Chem 275(11): 7462-‐5.
Geer, L. Y., S. P. Markey, et al. (2004). "Open mass spectrometry search algorithm." J Proteome Res 3(5): 958-‐64.
Gehrig, S., M. E. Eberle, et al. (2012). "Identification of modifications in microbial, native tRNA that suppress immunostimulatory activity." J Exp Med 209(2): 225-‐33.
Goehring, A. S., D. M. Rivers, et al. (2003). "Attachment of the ubiquitin-‐related protein Urm1p to the antioxidant protein Ahp1p." Eukaryot Cell 2(5): 930-‐6.
Gromadski, K. B., T. Schummer, et al. (2007). "Kinetics of the interactions between yeast elongation factors 1A and 1Balpha, guanine nucleotides, and aminoacyl-‐tRNA." J Biol Chem 282(49): 35629-‐37.
References
77
Grosjean, H. (2000). "Nucleic acids are not boring long polymers of only four types of nucleotides: A guided tour."
Halfmann, R., S. Alberti, et al. (2011). "Opposing effects of glutamine and asparagine govern prion formation by intrinsically disordered proteins." Mol Cell 43(1): 72-‐84.
Halfmann, R. and S. Lindquist (2010). "Epigenetics in the extreme: prions and the inheritance of environmentally acquired traits." Science 330(6004): 629-‐32.
Han, D. K., J. Eng, et al. (2001). "Quantitative profiling of differentiation-‐induced microsomal proteins using isotope-‐coded affinity tags and mass spectrometry." Nat Biotechnol 19(10): 946-‐51.
Heidebrecht, F., A. Heidebrecht, et al. (2009). "Improved semiquantitative Western blot technique with increased quantification range." J Immunol Methods 345(1-‐2): 40-‐8.
Huang, B., M. J. Johansson, et al. (2005). "An early step in wobble uridine tRNA modification requires the Elongator complex." RNA 11(4): 424-‐36.
Huber, A., B. Bodenmiller, et al. (2009). "Characterization of the rapamycin-‐sensitive phosphoproteome reveals that Sch9 is a central coordinator of protein synthesis." Genes Dev 23(16): 1929-‐43.
Ingolia, N. T., S. Ghaemmaghami, et al. (2009). "Genome-‐wide analysis in vivo of translation with nucleotide resolution using ribosome profiling." Science 324(5924): 218-‐23.
Ivanov, P., M. M. Emara, et al. (2011). "Angiogenin-‐induced tRNA fragments inhibit translation initiation." Mol Cell 43(4): 613-‐23.
Johansson, M. J., A. Esberg, et al. (2008). "Eukaryotic wobble uridine modifications promote a functionally redundant decoding system." Mol Cell Biol 28(10): 3301-‐12.
Kadaba, S., A. Krueger, et al. (2004). "Nuclear surveillance and degradation of hypomodified initiator tRNAMet in S. cerevisiae." Genes Dev 18(11): 1227-‐40.
Kaneko, T., T. Suzuki, et al. (2003). "Wobble modification differences and subcellular localization of tRNAs in Leishmania tarentolae: implication for tRNA sorting mechanism." EMBO J 22(3): 657-‐67.
Kanerva, P. A. and P. H. Maenpaa (1981). "Codon-‐specific serine transfer ribonucleic acid degradation in avian liver during vitellogenin induction." Acta Chem Scand B 35(5): 379-‐85.
Karnahl, U. and C. Wasternack (1992). "Half-‐life of cytoplasmic rRNA and tRNA, of plastid rRNA and of uridine nucleotides in heterotrophically and photoorganotrophically grown cells of Euglena gracilis and its apoplastic mutant W3BUL." Int J Biochem 24(3): 493-‐7.
Katunin, V. I., G. W. Muth, et al. (2002). "Important contribution to catalysis of peptide bond formation by a single ionizing group within the ribosome." Mol Cell 10(2): 339-‐46.
Keller, A., A. I. Nesvizhskii, et al. (2002). "Empirical statistical model to estimate the accuracy of peptide identifications made by MS/MS and database search." Anal Chem 74(20): 5383-‐92.
Kimchi-‐Sarfaty, C., J. M. Oh, et al. (2007). "A "silent" polymorphism in the MDR1 gene changes substrate specificity." Science 315(5811): 525-‐8.
References
78
Kudla, G., A. W. Murray, et al. (2009). "Coding-‐sequence determinants of gene expression in Escherichia coli." Science 324(5924): 255-‐8.
Leidel, S., P. G. Pedrioli, et al. (2009). "Ubiquitin-‐related modifier Urm1 acts as a sulphur carrier in thiolation of eukaryotic transfer RNA." Nature 458(7235): 228-‐32.
Leihne, V., F. Kirpekar, et al. (2011). "Roles of Trm9-‐ and ALKBH8-‐like proteins in the formation of modified wobble uridines in Arabidopsis tRNA." Nucleic Acids Res 39(17): 7688-‐701.
Li, S. and G. F. Hu (2011). "Emerging role of angiogenin in stress response and cell survival under adverse conditions." J Cell Physiol 227(7): 2822-‐6.
Li, Y., J. Luo, et al. (2008). "Stress-‐induced tRNA-‐derived RNAs: a novel class of small RNAs in the primitive eukaryote Giardia lamblia." Nucleic Acids Res 36(19): 6048-‐55.
Li, Z., C. Ender, et al. (2012). "Extensive terminal and asymmetric processing of small RNAs from rRNAs, snoRNAs, snRNAs, and tRNAs." Nucleic Acids Res 40(14): 6787-‐99.
Louvion, J. F., B. Havaux-‐Copf, et al. (1993). "Fusion of GAL4-‐VP16 to a steroid-‐binding domain provides a tool for gratuitous induction of galactose-‐responsive genes in yeast." Gene 131(1): 129-‐34.
Lu, J., B. Huang, et al. (2005). "The Kluyveromyces lactis gamma-‐toxin targets tRNA anticodons." RNA 11(11): 1648-‐54.
MacLean, B., J. K. Eng, et al. (2006). "General framework for developing and evaluating database scoring algorithms using the TANDEM search engine." Bioinformatics 22(22): 2830-‐2.
Maraia, R. J., N. H. Blewett, et al. (2008). "It's a mod mod tRNA world." Nat Chem Biol 4(3): 162-‐4.
Massenet, S., Y. Motorin, et al. (1999). "Pseudouridine mapping in the Saccharomyces cerevisiae spliceosomal U small nuclear RNAs (snRNAs) reveals that pseudouridine synthase pus1p exhibits a dual substrate specificity for U2 snRNA and tRNA." Mol Cell Biol 19(3): 2142-‐54.
Maynard, N. D., D. N. Macklin, et al. (2012). "Competing pathways control host resistance to virus via tRNA modification and programmed ribosomal frameshifting." Mol Syst Biol 8: 567.
Mehlgarten, C., D. Jablonowski, et al. (2010). "Elongator function in tRNA wobble uridine modification is conserved between yeast and plants." Mol Microbiol 76(5): 1082-‐94.
Milon, P., A. L. Konevega, et al. (2008). "Kinetic checkpoint at a late step in translation initiation." Mol Cell 30(6): 712-‐20.
Moura, G. R., M. Pinheiro, et al. (2011). "Species-‐specific codon context rules unveil non-‐neutrality effects of synonymous mutations." PLoS One 6(10): e26817.
Nakai, Y., M. Nakai, et al. (2008). "Thio-‐modification of yeast cytosolic tRNA requires a ubiquitin-‐related system that resembles bacterial sulfur transfer systems." J Biol Chem 283(41): 27469-‐76.
Nawrot, B., E. Sochacka, et al. (2011). "tRNA structural and functional changes induced by oxidative stress." Cell Mol Life Sci 68(24): 4023-‐32.
Nesvizhskii, A. I., A. Keller, et al. (2003). "A statistical model for identifying proteins by tandem mass spectrometry." Anal Chem 75(17): 4646-‐58.
References
79
Noma, A., Y. Sakaguchi, et al. (2009). "Mechanistic characterization of the sulfur-‐relay system for eukaryotic 2-‐thiouridine biogenesis at tRNA wobble positions." Nucleic Acids Res 37(4): 1335-‐52.
Nwagwu, M. and M. Nana (1980). "Ribonucleic acid synthesis in embryonic chick muscle, rates of synthesis and half-‐lives of transfer and ribosomal RNA species." J Embryol Exp Morphol 56: 253-‐67.
Ogle, J. M., A. P. Carter, et al. (2003). "Insights into the decoding mechanism from recent ribosome structures." Trends Biochem Sci 28(5): 259-‐66.
Pape, T., W. Wintermeyer, et al. (1999). "Induced fit in initial selection and proofreading of aminoacyl-‐tRNA on the ribosome." EMBO J 18(13): 3800-‐7.
Pape, T., W. Wintermeyer, et al. (2000). "Conformational switch in the decoding region of 16S rRNA during aminoacyl-‐tRNA selection on the ribosome." Nat Struct Biol 7(2): 104-‐7.
Patil, A., C. T. Chan, et al. (2012). "Translational infidelity-‐induced protein stress results from a deficiency in Trm9-‐catalyzed tRNA modifications." RNA Biol 9(7).
Patil, A., M. Dyavaiah, et al. (2012). "Increased tRNA modification and gene-‐specific codon usage regulate cell cycle progression during the DNA damage response." Cell Cycle 11(19).
Pedrioli, P. G., J. K. Eng, et al. (2004). "A common open representation of mass spectrometry data and its application to proteomics research." Nat Biotechnol 22(11): 1459-‐66.
Pedrioli, P. G., S. Leidel, et al. (2008). "Urm1 at the crossroad of modifications. 'Protein Modifications: Beyond the Usual Suspects' Review Series." EMBO Rep 9(12): 1196-‐202.
Pelet, S., R. Dechant, et al. (2012). "An integrated image analysis platform to quantify signal transduction in single cells." Integr Biol (Camb).
Percudani, R., A. Pavesi, et al. (1997). "Transfer RNA gene redundancy and translational selection in Saccharomyces cerevisiae." J Mol Biol 268(2): 322-‐30.
Petroski, M. D., G. S. Salvesen, et al. (2011). "Urm1 couples sulfur transfer to ubiquitin-‐like protein function in oxidative stress." Proc Natl Acad Sci U S A 108(5): 1749-‐50.
Phizicky, E. M. and J. D. Alfonzo (2010). "Do all modifications benefit all tRNAs?" FEBS Lett 584(2): 265-‐71.
Phizicky, E. M. and A. K. Hopper (2010). "tRNA biology charges to the front." Genes Dev 24(17): 1832-‐60.
Pluta, K., O. Lefebvre, et al. (2001). "Maf1p, a negative effector of RNA polymerase III in Saccharomyces cerevisiae." Mol Cell Biol 21(15): 5031-‐40.
Salas-‐Marco, J. and D. M. Bedwell (2005). "Discrimination between defects in elongation fidelity and termination efficiency provides mechanistic insights into translational readthrough." J Mol Biol 348(4): 801-‐15.
Sardana, R. and A. W. Johnson (2012). "The methyltransferase adaptor protein Trm112 is involved in biogenesis of both ribosomal subunits." Mol Biol Cell.
References
80
Schlieker, C. D., A. G. Van der Veen, et al. (2008). "A functional proteomics approach links the ubiquitin-‐related modifier Urm1 to a tRNA modification pathway." Proc Natl Acad Sci U S A 105(47): 18255-‐60.
Shaheen, H. H. and A. K. Hopper (2005). "Retrograde movement of tRNAs from the cytoplasm to the nucleus in Saccharomyces cerevisiae." Proc Natl Acad Sci U S A 102(32): 11290-‐5.
Shtynberg, D. (submitted). "iProphet: Improved statistical validation of peptide identification in shotgun proteomics." Mol Cell Proteomics.
Smyth, G. K. (2004). "Linear models and empirical bayes methods for assessing differential expression in microarray experiments." Stat Appl Genet Mol Biol 3: Article3.
Svejstrup, J. Q. (2007). "Elongator complex: how many roles does it play?" Curr Opin Cell Biol 19(3): 331-‐6.
Tats, A., T. Tenson, et al. (2008). "Preferred and avoided codon pairs in three domains of life." BMC Genomics 9: 463.
Team, T. R. D. C. (2010). R: A Language and Environment for Statistical Computing, R Foundation for Statistical Computing.
Thompson, D. M., C. Lu, et al. (2008). "tRNA cleavage is a conserved response to oxidative stress in eukaryotes." RNA 14(10): 2095-‐103.
Thompson, D. M. and R. Parker (2009). "The RNase Rny1p cleaves tRNAs and promotes cell death during oxidative stress in Saccharomyces cerevisiae." J Cell Biol 185(1): 43-‐50.
Tuller, T., A. Carmi, et al. (2010). "An evolutionarily conserved mechanism for controlling the efficiency of protein translation." Cell 141(2): 344-‐54.
Umeda, N., T. Suzuki, et al. (2005). "Mitochondria-‐specific RNA-‐modifying enzymes responsible for the biosynthesis of the wobble base in mitochondrial tRNAs. Implications for the molecular pathogenesis of human mitochondrial diseases." J Biol Chem 280(2): 1613-‐24.
Urbonavicius, J., Q. Qian, et al. (2001). "Improvement of reading frame maintenance is a common function for several tRNA modifications." EMBO J 20(17): 4863-‐73.
van den Born, E., C. B. Vagbo, et al. (2011). "ALKBH8-‐mediated formation of a novel diastereomeric pair of wobble nucleosides in mammalian tRNA." Nat Commun 2: 172.
Van der Veen, A. G., K. Schorpp, et al. (2011). "Role of the ubiquitin-‐like protein Urm1 as a noncanonical lysine-‐directed protein modifier." Proc Natl Acad Sci U S A 108(5): 1763-‐70.
Vendeix, F. A., F. V. t. Murphy, et al. (2012). "Human tRNA(Lys3)(UUU) is pre-‐structured by natural modifications for cognate and wobble codon binding through keto-‐enol tautomerism." J Mol Biol 416(4): 467-‐85.
Wei, Y. and X. S. Zheng (2011). "Maf1 regulation: a model of signal transduction inside the nucleus." Nucleus 1(2): 162-‐5.
Yamasaki, S., P. Ivanov, et al. (2009). "Angiogenin cleaves tRNA and promotes stress-‐induced translational repression." J Cell Biol 185(1): 35-‐42.
Yarian, C., M. Marszalek, et al. (2000). "Modified nucleoside dependent Watson-‐Crick and wobble codon binding by tRNALysUUU species." Biochemistry 39(44): 13390-‐5.
References
81
Yasukawa, T., Y. Kirino, et al. (2005). "Wobble modification deficiency in mutant tRNAs in patients with mitochondrial diseases." FEBS Lett 579(13): 2948-‐52.
Zaborske, J. and T. Pan (2010). "Genome-‐wide analysis of aminoacylation (charging) levels of tRNA using microarrays." J Vis Exp(40).
Zaborske, J. M., J. Narasimhan, et al. (2009). "Genome-‐wide analysis of tRNA charging and activation of the eIF2 kinase Gcn2p." J Biol Chem 284(37): 25254-‐67.
Zaher, H. S. and R. Green (2011). "A primary role for release factor 3 in quality control during translation elongation in Escherichia coli." Cell 147(2): 396-‐408.
Zhang, G., M. Hubalewska, et al. (2009). "Transient ribosomal attenuation coordinates protein synthesis and co-‐translational folding." Nat Struct Mol Biol 16(3): 274-‐80.
Abbreviations
82
6 ABBREVIATIONS
A-site aminoacyl-site
aa amino-acid
aa-tRNA aminoacyl-tRNA
ASL anticodon stem loop
CFP cyan fluorescent protein
CHX cycloheximide
CPM counts per minutes
E-site exit-site
ELP elongator protein
FDR false discovery rate
GAL galactose
GO gene ontology
GST glutathione S-transferase
GTP guanidine triphosphate
HygroB hygromycin B
KAN kanamycin resistance cassette
mcm5 5-methoxy-carbonyl-methyl
mcm5s2 5-methoxy-carbonyl-methyl-2-thio
MERRF Myoclonic Epilepsy with Ragged Red Fibers
MELAS Mitochondrial Encephalopathy Lactic Acidosis with Stroke-like
episode