Protein degradation and dynamic tRNA thiolation fine-tune translation at elevated temperatures

Maintenance of protein quality control has implications in various processes such as neurodegeneration and ageing. To investigate how environmental insults affect this process, we analysed the proteome of yeast continuously exposed to mild heat stress. In agreement with previous transcriptomics studies, amongst the most marked changes, we found up-regulation of cytoprotective factors; a shift from oxidative phosphorylation to fermentation; and down-regulation of translation. Importantly, we also identified a novel, post-translationally controlled, component of the heat shock response. The abundance of Ncs2p and Ncs6p, two members of the URM1 pathway responsible for the thiolation of wobble uridines in cytoplasmic tRNAs tKUUU, tQUUG and tEUUC, is down-regulated in a proteasomal dependent fashion. Using random forests we show that this results in differential translation of transcripts with a biased content for the corresponding codons. We propose that the role of this pathway in promoting catabolic and inhibiting anabolic processes, affords cells with additional time and resources needed to attain proper protein folding under periods of stress.

For the mass spectrometry analysis, the nucleosides were resuspended in water and separated by liquid chromatography over an Hypercarb graphite column (35005-100065, Thermo Scientific) connected to an UltiMate 3000 uHPLC system (Dionex) running a 1.0 μl/min gradient of buffer B (0.1% (v/v) HCOOH, 90% (v/v) CH 3 CN) in buffer A (0.1% (v/v) HCOOH, 2% (v/v) CH 3 CN) ranging from 1% to 37% over 45 min. The column-oven temperature was maintained at 65 °C throughout the run. A fused silica emitter (PicoFrit columns, PF360-75-10-N-5, New Objective) was used to spray eluting nucleosides into an LTQ-Orbitrap Velos Pro mass spectrometer (Thermo Finnigan) operating at a source voltage of 1.8 kV. MS1 scans were acquired in profile mode at an FT-MS resolution of 100,000. The MS was running in data-independent acquisition mode (DIA) using collision-induced fragmentation (CID) with an isolation width set at 2.0 m/z and normalised collision energy of 35.0. MS/MS spectra were acquired in centroid mode with the IonTrap Normal scan rate. Raw MS files were converted to the mzXML format (3) and analyzed by an in-house developed software to identify and quantify the nucleosides. The identifications were validated by manually comparing the fragmentation spectra with previously reported ones (9, 10).
APM supplemented denaturing PAGE 0.5 to 1.0 μg of bulk tRNA was mixed with an equal volume of loading buffer (AM8547, LifeTechnologies) and electrophoresed through a 10% acrylamide gel containing 7 M urea, 0.5X TBE and 50 μg/ml [(N-acryloylamino)phenyl] mercuric chloride (APM) (synthesised in house according to the procedure described in (11) at 200 V for 90 min. The gels were prerun for 15 minutes at 200 V and each well was washed by pipetting electrophoresis buffer multiple times. After electrophoresis, the gels were stained with SYBR Gold (Invitrogen) diluted 1:10,000 in 0.5X TBE for 5 min before imaging.

tRNA northern blot analysis
Electrophoretically separated tRNAs were transferred to a positively charged nylon membrane (Hybond-N+, GE Healthcare) using a semi-dry blotter (Fastblot, biometra) at a constant current of 400 mA for 40 min in 0.5X TBE as the transfer buffer. Transferred tRNAs were cross linked using a UV crosslinker (C-1000 UVP) at an energy setting of 1200 J for 30 s, while the membrane was still moist. Subsequently, the membrane was prehybridised using 5ml of PerfectHyb Plus (Sigma) buffer with 1x ProtectRNA RNAse inhibitor and 0.1 µg/µl ssDNA for 2 h in a 50 ml centrifuge tube. 10 pmol of oligonucleotide probe was labeled with 10U of T4 Polynucleotide Kinase (New England Biolabs) and 5 µl of 3000 Ci/mmol 10 uCi/µl ϒ 32 P ATP (PerkinElmer) at 37 °C for 60 min. Labelling was quenched by 1 µl of 0.5 M EDTA at 65 °C for 20 min. The labeled probes were cleaned using the illustra MicroSpin G-25 Columns (GE Healthcare). After pre-hybridisation the labeled probe was added directly to the tubes and incubated overnight at 55 °C in a hybridisation oven. The membrane was washed twice for 10 min each and again twice for 30 min each at 55 °C using 10 ml wash buffer (3x SSC, 25 mM NaH 2 PO 4 pH 7.5, 5% SDS, 10x Denhardt's reagent) pre-warmed to 37 °C. Finally, the membrane was washed for 8 min in high stringency buffer (1x SSC, 10% SDS) pre-warmed to 37 °C and exposed to the X-Ray Film.
The probes used were; 5'-CTCCTCATAGGGGGCTCGAACCC-3' for Sc-tK UUU , 5'-AGGTCCTACCCGGATTCGAACCGG-3' for Sc-tQ UUG , 5'-CGCCCAAACAGGGACTTGAACCC-3' for Hs-tK UUU , 5'-GGTCCCACCGAGATTTGAACTCGG-3' for Hs-tQ UUG , 5'-CGACTCTGGTGGGATTCGAACCC-3' for Hs-tR UCU and 5'-TGCGTTGGCCGGGAATCGAACCCG-3' for Hs-tG UCC . consistently quantified across the label switch conditions were removed. (A) and (B) represent the distributions of peptide ratios before and after the label switch filter in the form of a scatter plot of binned hexagonal tiles. The tiles colour indicates the number of peptides in each bin. (C) MA plots for the three biological replicates. M: log 2 (peptide ratio); A: (log 2 (light peptide area) + log 2 (heavy peptide area)) / 2. (D) and (E) Box-and-Whisker plots of the protein ratios from the three biological replicates before and after median normalisation respectively. The box represents the inter quartile range (50% of the data) with the solid-line in the centre of the box representing the median of the distribution of protein ratios. (F) Dynamic range of protein expression after stringent filtering. The histogram illustrates the number of quantified proteins passing the filtering criteria in relation to their copy per cell number (14). The red and blue dashed lines represent the median and mean in the linear scale, respectively. Figure S2: 30 ℃ vs 37 ℃ proteomics data quality check and normalisation. (A) Scatter plot of the statistically significant protein ratios in common in the present study and the heat shock response study by Nagaraj et al. (15). (B) Heat-map comparison of statistically significant changes in protein abundances from this study and in mRNA abundances estimated from Gasch et al. (16). Micro-array analyses ct1 29 ℃/37 ℃ along with two repeats of ct2 37 ℃ vs. 29 ℃ were treated as replicates and a Bayes moderated t-test was used to determine significant changes. Only significant changes in common are plotted. (C -D) Word clouds showing the most significantly overrepresented (p-value < 0.001) GO terms for Biological Processes from (C) up-or (D) down-regulated proteins; analysed by FunSpec (17). GO terms in common with those significantly over-represented in the wild-type vs. urm1∆ analysis (presented in (8)) are shown in red.  Table 2 and  Supplementary Table S2 for the mean Ncs2p protein abundance ratio and the associated statistical significance. Figure S5: Manual validation of Ncs6p identification and quantification. Shown are the annotated MS/MS spectra, obtained from the Lorikeet Spectrum Viewer for Comet, and the extracted ion chromatograms (XICs), obtained from XPRESS, for the indicated peptides of Ncs6p that were identified and quantified in the WT 30 °C/37 °C proteomics analysis. Refer to Table 2 and  Supplementary Table S2 for the mean Ncs6p protein abundance ratio and the associated statistical significance. Figure S6: tRNA hypothiolation and crosstalk between the URM1 and ELP pathways. (A) WT cells, growing exponentially at 30 °C, were diluted into prewarmed YPD medium and incubated at 37 °C. Samples were taken at the indicated time points, processed to extract bulk tRNA and analysed by [p-(N-acrylamino)-phenyl]mercuric chloride (APM) supplemented denaturing PAGE. ON: over-night. (B) Electropherogram of bulk tRNA isolated from wild-type, urm1∆ or elp1∆ cells and analysed by APM-dPAGE. APM interacts and retards the thiolated tRNAs that appear as a slow moving band at the top of the gel. (C) tE UUC runs as a single band in a PAGE after purification.

Supplementary
Supplementary Figure S7: MS and MS/MS spectra for mcm 5 U and mcm 5 s 2 U. (A) MS spectra of mcm 5 s 2 U and mcm 5 U nucleoside and nucleobase are shown. (B) MS/MS spectra of mcm 5 s 2 U and mcm 5 U nucleobases obtained after CID based fragmentation. Spectra were annotated by matching against the published fragmentation products of the modified uridines (9,10).

Supplementary Figures S8, S9
and S10: Extracted ion chromatograms for mcm 5 U and mcm 5 s 2 U. tRNA tQ UUG (S8), tE UUC (S9) and tR UCU (S10) were purified from wild-type yeast cells grown in either rich medium at 30 °C, or rich medium at 37 °C, or in absence of sulfur amino acids at 30 °C, and were digested and dephosphorylated to nucleosides and analysed by LC-MS-MS/MS. Shown here are the extracted ion chromatograms (XICs) for the m/z values corresponding to protonated nucleoside (MH + ) and nucleobase (BH + ) of A, mcm 5 s 2 U and mcm 5 U from either tQ UUG , tE UUC or tR UCU . The area under the XIC was used to estimate the abundance of the corresponding nucleoside. Note that the overlapping nucleoside-nucleobase elution profiles are generated by post-source fragmentation of the N-glycosidic bond. This overlap is used as a criterion for the identification of a nucleoside in addition to its MS/MS spectra (9). Figure S11: Volcano plots for codon biases. Each volcano plot in the figure shows the abundance ratio and statistical significance of the proteins whose corresponding genes are among the top 2% of yeast genes with the highest frequency of the indicated codon. The dotted red line (y=2) indicates 1% FDR. The black dotted line (x=0) indicates a protein ratio of 1.

Supplementary
Supplementary Figure S12: The tRNA recognising AGA is thiolated in higher eukaryotes. APM-dPAGE and northern blot analyses of the bulk tRNAs isolated from HEK-293 cells. Figure S13: An NCS2 mutant deregulates tRNA thiolation. Electropherogram from APM-dPAGE analysis of the bulk tRNA isolated from wild-type (WT) or ncs2∆ cells transformed with either an empty plasmid (pEV), a plasmid with wild-type NCS2 (pNCS2), or one with ncs2_A212T (pncs2_A212T) and grown under the indicated conditions.

Supplementary Figure S14: Over-expression of tRNAs does not affect their thiolation.
Electropherogram from APM-dPAGE analysis of bulk tRNA isolated from wild-type (WT) cells transformed with either a high-copy empty plasmid (pEV) or the same plasmid with tRNA genes for tK UUU , tQ UUG and tE UUC (ptKQE) and grown under the indicated conditions.
Supplementary Figure S15: Filtering and quality control of the 30 ℃ vs 37 ℃ + ptKQE analysis. SILAC labelling design for replicate number 2, was switched (denoted by label design -1). The means of log 2 of ratios of each peptide from the two designs were compared and the peptides that were not consistently quantified across the label switch conditions were removed. (A) and (B) represent the distributions of peptide ratios before and after the label switch filter in the form of a scatter plot of binned hexagonal tiles. The tiles colour indicates the number of peptides in each bin.
(C) Peptide level MA plots. M is equal to the log 2 of peptide abundance ratios (30 ℃/37 ℃ + ptKQE) and A is equal to 1/2 * log 2 of the product of peptide abundances from the two conditions. Systematic inter-sample variances were checked and removed using median normalisation. (D) and (E) Box-and-Whisker plots of the protein ratios from the three biological replicates before and after median normalisation respectively. The box represents the inter quartile range (50% of the data) with the solid-line in the centre of the box representing the median of the distribution of protein ratios.
Peptide ratios before filtering Peptide ratios after filtering label design 1 log 2 (peptide ratio) label design -1 -log 2 (peptide ratio)       XICs for nucleoside and nucleobases from tR UCU  GAG  GAT  GCA  GCC  GCG  GCT  GGA  GGC  GGG   GGT  GTA  GTC  GTG  GTT  TAA  TAC  TAG  TAT   TCA  TCC  TCG  TCT  TGA  TGC  TGG  TGT  TTA   TTC