Duplicated ribosomal protein paralogs promote alternative translation and drug resistance | Panda Anku

Gene conversion of duplicated ribosomal protein genes reveals paralog subfunctionalization

Comparing the expression level of Yeast’s 29 duplicated ribosomal protein gene pairs with significant differences in both regulatory and coding sequence using RNA sequencing and mass spectrometry, we identified five gene pairs: one 40S and four 60S RPG pairs, that are differentially expressed (Fig. 1a–c and Supplementary Fig. 1a, b). Most of these genes exhibit differences in the regulatory sequence controlling their expression (<70% similar) (Supplementary Fig. 1b), which explain why they produce different amounts of RNA and proteins (Fig. 1b, c). In general, paralogs producing most RNA also produced most protein, except in the case of eL37/RPL37 paralogs, which are differentially translated22. In all cases, one of the two paralogous proteins (the major form) dominated over the other (the minor form) by 3 to 4 times, suggesting that most of the ribosomes are produced from one gene copy (Fig. 1c). Deleting one of the two duplicated ribosomal protein gene (dRPG) copies reduced the abundance of the mRNA generated by the pair, except in the case of uL30/RPL7 (B∆ and A∆) and the major copy of eL36/RPL36 (B∆, Fig. 1d).

Fig. 1: Homogenization of yeast ribosomal proteins identifies dose independent paralogs subfunctionalization.
figure 1

a Gene conversion strategy for “homogenization” of ribosomal protein genes. Genes showing paralog-specific response to different conditions were selected for homogenization. Homogenized strains were created by replacing one paralog locus including all regulatory sequences, introns and UTRs, with sequence of the other. Cells carrying one or two copies of a single paralog were examined for growth in liquid media and maximum growth rates were calculated for each strain. Duplicated ribosomal protein gene expression levels were examined by RT-qPCR to identify dose-dependency of growth defects. b RNA abundance in transcripts per million (TPM) of each paralog obtained by RNA sequencing n = 2 biologically independent samples. Universal and conventional names of each protein are shown below. c Abundance of protein isoforms generated by each duplicated ribosomal protein gene was determined by Swath Multiple Reaction Monitoring (MRM) identifying the major and minor isoforms. No suitable peptides to reliably distinguish between L27 paralogs were found, hence their peptide signal intensities were not determined (ND). d Total RNA generated by each gene pair was detected using primers common to both paralogs in RT-qPCR after deletion or homogenization of duplicated ribosomal protein genes and is shown relative to mRNA detected in wild-type strain. B∆ and A∆ indicate RNA detected in strains lacking the B or A paralog, while AA and BB indicate RNA detected in cells containing two copies of the A or B isoforms respectively. e Growth rates of deletion and homogenized strains were determined and shown relative to wild-type. f Growth rates of deletion and homogenized strains were determined in media containing staurosporine (3 µg/ml), hygromycin (100 µg/ml) or NaCl (0.9 M) and the effect on growth are shown relative to wild-type treated similarly. Bars shown in (cf) represent the means of n = 3 biologically independent samples shown as data points, p values from two-tailed unpaired t test are indicated in (f).

To probe the function of the gene pairs we next used a strategy of gene homogenization whereby we replace one of the paralogs with the other in its natural locus. This holistic approach helped us identify all differences between paralog functions, including those stemming from difference in expression pattern or intervening sequences, and allowed us to distinguish between gene dose and gene type requirement. The results indicated that homogenization of duplicated ribosomal protein genes, except el27bb/rpl27bb, does not decrease the overall amount of mRNA produced for each RPG (AA and BB, Fig. 1d). As expected, the deletion of the major paralog, which produces most RPs in the cell, impaired cell growth in rich media (Fig. 1e). This growth defect is restored by the expression of two copies of the minor paralog, except in the case of uL30 and eL27/RP27, where the minor paralog duplication fails to complement the deletion of the major copy (Fig. 1e). This indicates that not all copies of dRPGs could independently support normal growth and that the copy-specific requirement varies between ribosomal proteins.

Since many minor paralogs were dispensable for growth under normal condition, we next examined the effect of gene deletion or homogenization on growth under stress. We exposed our different strains to stresses in the form of exposure to three drugs: Staurosporine, Hygromycin and NaCl, which induce an array of different stresses. As expected, exposure to stress revealed different sets of paralog’s copy number-specific effects that were not observed under normal growth conditions and could not be explained by reduced RPG expression or constitutive drug independent decrease in growth rate. For example, deletion of the eL27 major paralog, which reduced overall eL27 protein expression and inhibited growth under normal conditions, enhanced growth in the presence of staurosporine and hygromycin but not NaCl (Fig. 1f). In contrast, deletion of the minor paralog of uL30, which did not inhibit gene expression or cell growth, reduced cell resistance to staurosporine (Fig. 1f). These data indicate that duplicated ribosomal protein genes are not completely redundant and they identify uL30 as a good model for studying differences between their functions.

uL30 paralogs differentially modulate ribosome biogenesis and cell growth

The proteins generated from the duplicated uL30 ribosomal protein paralogs differ by five amino acids, 4 of which are clustered in the N-terminus, leading to a major difference in the predicted secondary structure of the N-terminal domain (Supplementary Fig. 2b). The protein featuring the more structured N-terminal domain (uL30A) is more abundant, forming the bulk of the ribosome in the cell, making it more likely to be the housekeeping version of the pair (Supplementary Fig. 2c). Consistently, the expression of this major paralog (uL30A) is required for normal growth (Fig. 2a). As expected, the absence of the major paralog (ul30a∆ and ul30bb) reduced the synthesis of the 60S subunit leading to subunit imbalance, reduction in ribosome abundance and the accumulation of 40S subunit awaiting the 60S subunit, or “half-mers” (Fig. 2c and Supplementary Fig. 2). The subunit imbalance was also confirmed by the decrease in 25S/18S rRNA ratio and the overall decrease in the 25S rRNA and the 60S ribosomal proteins (Supplementary Fig. 2f, g). However, we did not see an overall decrease in the abundance of the mRNA coding for the 60S ribosomal proteins or specific decrease in the uL30 mRNA or protein relative to other 60S proteins (Supplementary Fig. 2e, f). Indeed, despite the defect in the 60S synthesis the total amount of uL30 mRNA and protein produced in ul30bb cells was similar or even slightly greater than that detected in ul30aa cells. This indicates that the amount of uL30 proteins is not rate limiting and that the effect observed in ul30a∆ and ul30bb cells is not caused by decreased production of uL30 proteins in these cells. These data indicate that the major paralog is required for the biogenesis of the 60S subunits, and that the duplication of the chromosomal copy of the minor paralog cannot compensate for the function of the major form even when expressed at similar levels.

Fig. 2: uL30 paralogs differentially alter cell growth, ribosome biogenesis and staurosporine resistance.
figure 2

a, b Growth curves of wild-type (WT), deletion strains (ul30b∆ and ul30a∆) and homogenized strains (ul30aa and ul30bb) grown in complete synthetic media without (a) or in the presence of staurosporine (b). c, d Polysome profiles were obtained from WT, cells lacking one paralog or expressing two copies of the same paralog from cultures grown in complete synthetic media without (c) or in the presence of staurosporine (d). Position of 40S and 60S ribosomal subunits and 80S ribosomes are indicated. Arrows indicate position of half-mers or 40S awaiting the 60S subunit. Curves shown in (ad) are representative examples from n = 3 biologically independent biological samples. e Translation index (mRNA associated with polyribosomes/mRNA associated with monosome and subunits) was determined using RNA sequencing for each uL30 paralog in absence (−) or presence (+) of staurosporine and the ratio for uL30A over uL30B is reported for n = 2 biologically independent biological samples. f Protein abundance of uL30 was determined by MRM in absence or presence of staurosporine and the ratio of uL30A over uL30B is reported for n = 2 biologically independent biological samples. g The amount of uL30 proteins incorporated into ribosomes was determined using MRM in absence or presence of staurosporine and the ratio of uL30A over uL30B is reported. Graphs in (e, f) show points from n = 2, while (g) show points for n = 2 (− staurosporine) and n = 3 (+ staurosporine) biologically independent biological samples.

uL30 minor paralog is required for cell resistance to staurosporine

Surprisingly, deletion of the minor paralog, which does not affect cell growth or ribosome biogenesis, rendered cells sensitive to staurosporine (Figs. 1f, 2b and Supplementary Fig. 3). Cells expressing one or two copies of the A version (ul30b∆ and ul30aa, respectively) were more sensitive to staurosporine and produced shorter polyribosomes than wild-type cells, but the cells were otherwise normal (Fig. 2b, c, and Supplementary Fig. 2f, g). In contrast, cells expressing the B version (ul30a∆ and ul30bb) were more resistant than the wild-type to staurosporine but they grew slower under normal growth conditions (Fig. 2a, b and Supplementary Fig. 3). Exposure to staurosporine significantly reduced the size of the polyribosomes in cells expressing uL30A but it had no effect on those expressing uL30B (Fig. 2c, d and Supplementary Fig. 3b). This suggests that uL30 paralogs have opposite effects on cell tolerance to staurosporine resulting in increased sensitivity to the drug when the protein is produced from the major copy and increased resistance when it is expressed from the minor counterpart.

The minor paralog-dependent resistance of staurosporine is not linked to reduced growth rate since cells growing at rates similar to uL30bb did not exhibit resistance to staurosporine (Supplementary Fig. 4a). Staurosporine resistance also was not reproduced by the deletion of the translation initiation factors eIF2A or TIF4632, which are known to reduce overall translation (Supplementary Fig. 4b). However, resistance to staurosporine was observed after the deletion of the gene coding for the translation elongation factor Tef4 suggesting that the paralog-specific effect of uL30 genes may stem from differences in their capacity to support translation elongation. The requirement of the minor paralog for staurosporine resistance is also supported by its increased expression in the presence of staurosporine. Exposing cells to staurosporine altered the ratio of uL30 proteins by favouring the translation of the minor paralog and its incorporation into active ribosomes (Fig. 2e–g and Supplementary Fig. 5). This suggests that the translation of uL30 paralogs is differentially regulated and as such we looked for features in this gene that could explain differences in translation. We did not find any significant difference in ORF size or UTR length but noted lower Kozak score and the clustering of suboptimal codons at the 5′ end of uL30B. This may explain the preferential translation of uL30A under normal condition and the resistance of uL30B to staurosporine-dependent repression of translation (Supplementary Fig. 6). We conclude that the expression of uL30A is required for ribosome biogenesis while uL30B is needed to optimize the cell tolerance to staurosporine.

The two paralogs of uL30 modulate the translation of different subsets of mRNA

By comparing the translation pattern of cells expressing two copies of uL30A or uL30B, we identified a set of paralog-dependent genes. We sequenced ribosome-associated mRNAs, in wild-type, ul30aa and ul30bb cells, both before and after staurosporine treatment, and verified the results by RT-qPCR (Fig. 3a and Supplementary Fig. 7a). We calculated the translation index as the ratio of mRNA associated with heavy (4 or more ribosomes) and light (subunits and monosome) ribosome fractions and verified the correlation of this presumed translation index or ribosome association ratio and the amount of protein produced by corresponding mRNA (Fig. 3a and Supplementary Fig. 7b). Wild-type and ul30aa cells displayed similar polyribosome association patterns with only 96 genes more- or less-associated with polyribosome in ul30aa cells, suggesting that the minor paralog is required for the translation of a small subset of genes (Fig. 3b, c). In contrast, 2094 genes were more-associated with polyribosome and 458 less-associated with polyribosome in ul30bb, than in wild-type cells (Fig. 3b, c). This paralog-dependent change in global translation did not directionally correlate with changes in RNA abundance (Supplementary Fig. 8). Therefore, the paralog-specific changes in cell growth and drug resistance are likely related to changes in translation of a defined subset of genes.

Fig. 3: Exposure to staurosporine modulates the translation of genes coding for cell periphery proteins in a paralog-dependent manner.
figure 3

a Strategy for determining translation profiles of homogenized uL30 strains. mRNA was extracted from heavy (polyribosome) and light (monosome and subunits) fractions and sequenced. Translation index was calculated as the ratio of mRNA associated with heavy to light fractions. b Translation index of mRNAs (expression > 1 TPM) in homogenized strains is compared to that of wild-type. mRNAs showing differential translation by more than log2 0.5 difference are shown in red. c Venn diagrams of number of genes with changed association to polysomes comparing homogenized strains (ul30aa and ul30bb) to WT identified in (b). d Distribution of the number of genes present in enriched component gene ontology categories (p < 0.001 with Bonferroni correction) for genes with specific changes in association to polysomes. e Map of the genetic and physical interactions of the top up-translated genes in ul30bb strain was generated with the Genemania network construction tool. Up-translated genes were grouped and colored by functional pathways, genes in black circles are not affected by ul30bb. f Translation index of mRNAs (expression > 1 TPM) in WT and homogenized strains is compared in cells grown in absence (untreated) or in presence of staurosporine (treated). mRNAs showing differential translation by more than log2 0.5 difference are shown in red. g Bar graph showing the number of genes under-translated (dark gray) or over-translated (light gray) in response to staurosporine. h Distribution of the number of genes present in enriched component gene ontology categories (p < 0.001 with Bonferroni correction) for genes with specific changes in association to polysomes after staurosporine treatment. i Map of the genetic and physical interactions of the top up-translated genes in ul30bb strain in presence of staurosporine was generated with the Genemania network construction tool. Up-translated genes associated with cell wall and membrane component category were grouped and colored in blue, genes in gray associate with other categories and genes in black circles are not affected by ul30bb. RNA sequencing was performed on n = 2 biologically independent samples.

Notably, the paralog-dependent changes in translation were not strictly linked to defects in ribosome biogenesis, or reduced ribosome numbers, since the paralog-specific effects on translation were also observed in ul30aa cells, which have similar ribosome biogenesis and ribosome number to those observed in wild-type cells (Fig. 2c and Supplementary Fig. 2f, g). Importantly we noticed that a set of 16 genes were inversely regulated (opposite effects observed in ul30aa and ul30bb cells) by uL30 paralogs, underlining the paralog-specific effect on translation (Supplementary Data 5). Together these data indicate that uL30 paralogs differentially alter translation.

The minor paralog of uL30 induces the translation of cell periphery genes

Gene ontology analysis identified ‘cell periphery’ as the only enriched category that is inversely regulated by uL30 paralogs. The expression of the A form decreases the translation of cell periphery genes (gene ontology enrichment with p value 6.3E−4), while the B form increases the translation of genes in this category (p value 1.0E−7; Fig. 3d). Since most cell periphery genes are translated in the endoplasmic reticulum (ER) we examined the effect of uL30 paralog on the translation of ER-translated genes23. Interestingly, we also found that ER-translated genes23 are over-translated in our ul30bb strain with two copies of uL30B (Supplementary Fig. 9). This further supports the link between uL30 paralog and translation of cell periphery genes and is consistent with the proximity of uL30 and its N-terminal domain to the peptide exit channel (Supplementary Fig. 2a). Analysis of the genetic and physical interactions of the top 35 translated genes in ul30bb cells identified a tight network of proteins involved in translation, large subunit (LSU) biogenesis, ribonucleoside synthesis, cell wall and transport (Fig. 3e). The paralog-dependent changes in the translation of this network of related genes may explain the paralog-dependent tolerance to staurosporine treatment. Staurosporine affects cell wall integrity through the inhibition of protein kinase C (PKC) and as such the paralog-dependent changes in the expression of cell wall and membrane proteins may alter the drug effect on cell growth24,25. We conclude that uL30 paralogs differentially modulate the translation of cell periphery genes, and that this likely explains the paralog-dependent differences in staurosporine resistance.

The minor paralog of uL30 tempers staurosporine-dependent modification of translation

Staurosporine altered the translation of ~600 genes in wild type and this number of staurosporine-dependent genes was increased in ul30aa cells to 800 and decreased in ul30bb cells to only 150 genes (Fig. 3f, g). This indicates that the effect of staurosporine on translation is modulated in a paralog-dependent manner. This paralog-dependent modification of the translation response to staurosporine, at least in the case of ul30aa, occurs even when ribosome biogenesis and ribosome numbers are identical (Supplementary Fig. 2g). Interestingly, the only set of genes that were inhibited in wild-type and induced in ul30bb cells after staurosporine treatment were those associated with cell periphery (Fig. 3h). Genes that were most upregulated by staurosporine in the presence of the B, and not the A form were also associated with the cell wall and membrane proteins (Fig. 3i).

These data indicated that uL30B increases resistance to staurosporine by increasing the translation of cell periphery genes. We therefore hypothesized that the minor paralog of uL30 might affect the response to cell wall integrity drugs other than just staurosporine. As postulated, cells expressing the A form were sensitive, while those expressing the B version were resistant to the cell wall integrity drugs ketoconazole and caffeine (Supplementary Fig. 10). We conclude that ul30bb dependent resistance to drugs is not limited to staurosporine but extends to other drugs affecting cell wall integrity.

uL30 minor paralog selectively induces the translation of mRNAs with long open reading frames

Examining the features of the genes that are differentially translated by uL30 paralogs identified clear differences in the length of the open reading frames. Coding regions of the top 20 genes, which are more translated in the presence of uL30b, are on average >3 times longer than those that are under-translated (Fig. 4a). Comparison between the average ORF length of all the genes that are translated in a paralog-dependent manner with that of all genes in yeast indicated that in general cells expressing uL30B differentially translate long genes over the short genes (Fig. 4b). This change in translation was not linked to changes in mRNA abundance since we did not see any correlation between ORF length and RNA abundance in ul30aa and ul30bb cells (Supplementary Fig. 8c). The paralog-dependent change in translation of long ORFs is not linked to translation efficiency. Most of the long genes that are differentially translated between ul30bb and ul30aa cells have average translation index in wild-type cells (when calculated as the ratio of mRNA associated with heavy polyribosome/mRNA associated with light ribosomes (Supplementary Figs. 11a and 12). However, we found that the ribosome density decreases as the ORF length increases as previously observed (Supplementary Fig. 11b)26. Therefore, the increased translation of long ORFs in ul30bb cells is not linked to the fraction of the mRNA associated with heavy polysomes in wild type but could be linked to ribosome density. Remarkably, expression of long and short ORFs were inversely affected in ul30aa and ul30bb cells upon exposure to staurosporine. For example, while ORFs over-translated in response to staurosporine in ul30aa were relatively short, those over-translated in ul30bb were long when compared with wild-type cells (Fig. 4b, lower panel). Therefore, the observed ORF length-dependent change in translation after staurosporine is not restricted to ul30bb cells, which exhibit difference in ribosome biogenesis, but instead extends to ul30aa, which displays no major defect in ribosome production (Supplementary Fig. 2f, g). The effect of the shift in mRNA association to ribosomes on protein abundance was examined by mass spectrometry. As indicated in Fig. 4c, the protein abundance in ul30bb was reduced for genes showing decreased association with ribosome and increased for those showing increased association with ribosome. This confirms that in most cases the change in mRNA association with ribosome associates with changes in the amount of proteins produced in the cell.

Fig. 4: Paralogs of uL30 modulate cell response to drug by altering the translation of long genes.
figure 4

a Top 20 genes over- or under-translated in ul30bb in absence (top panel) or presence (lower panel) of staurosporine are plotted relative to ORF length in nucleotides. Genes are ordered on the X-axis according to the magnitude of change in translation. b Box plots comparing the ORFs length of all genes and genes that are over-(OT) and under-translated (UT) in ul30aa and ul30bb in absence (top panel) or presence (lower panel) of staurosporine. Box limits represent the first and third quartiles, the middle line is the median and whiskers extend to minima and maxima. ORFs numbers in each category is stated at bottom. The results of two-tailed Mann–Withney test indicated on top. c The violin plots show the relative change in the protein abundance of the top 50 OT and UT genes in ul30bb. The data obtained from n = 4 (WT) and n = 2 (ul30bb) biologically independent samples. The result of two-tailed Mann–Withney test indicated on top. d Translation index of long (Glt1, 6438 nt) and short (Egd1, 474 nt) ORFs determined by RT-qPCR and normalized to middle size ORF (Osh6, 1344 nt). The black circles indicate data from n = 2 biologically independent samples. e Relative enrichment of Glt1 and Egd1 was monitored across sucrose gradients fractions as described in (d). P1 are di-trisomes, P2 are quadra-penta-hexasomes and P3 are heptasomes to decasomes fractions. f GLT1 ORF was substituted by EGD1 ORF (top panel) in strains containing or lacking major paralog of uL30. Graph shows the relative change in translation for the endogenous EGD1, GLT1, and the chimeric gene (glt1Δ::EGD1) upon deletion of uL30A in absence (−) or presence (+) of staurosporine. The number of independent biological replicates is indicated at bottom. g ORFs length of cell periphery proteins (CP) genes and genes that are under- (UT) or over-translated (OT) in ul30aa and ul30bb strains are represented as in (b). The RNA sequencing data in (a, b, g) are generated using n = 2 biologically independent samples.

The ORF length-dependent translation was next validated by RT-qPCR using the long GLT1 (6438 nt) and short EGD1 (474 nt) model ORFs (Fig. 4d). Once again, the translation of long and short ORFs were inversely regulated in ul30aa and ul30bb cells. The differences in the translation of mRNAs of different lengths was not due to differences in the early initiation phases, as the amount of free mRNA did not vary in cells expressing different paralogs (Fig. 4e). Instead, we observed increased association of the long ORFs with the heavy polyribosome in cells expressing uL30B. Curiously, this minor isoform also decreased the ratio of long/short ORF co-sedimentation with the 40S and 60S fractions, mostly due to an increased association of the short ORF mRNA in these fractions (Fig. 4e). It is not clear if this unusual co-sedimentation pattern with the 60S fraction represents an abnormal association with the subunit or a coincidental co-sedimentation of an independent mRNA protein complex. In all cases, it is clear that uL30B favors the optimal translation of long ORFs, through increased association with ribosomes.

To further investigate the importance of ORF length to uL30B-dependent translation, we replaced GLT1’s long ORF with the short EGD1 ORF and monitored translation before and after staurosporine treatment. Strikingly, changing the ORF length abolished both paralog- and staurosporine-dependent translation, confirming the importance of ORF length to uL30B effect on translation (Fig.4f and Supplementary Fig. 13). Interestingly, ORFs coding for cell periphery proteins, as a group, are longer than the average ORF length, explaining their increased expression in the presence of uL30B (Fig. 4g). We conclude that the uL30 paralog-dependent response to staurosporine is mediated by differential translation of genes with long ORFs.

uL30 minor paralog-dependent drug resistance is not due to differences in ribosome biogenesis

To evaluate the contribution of their chromosomal loci and regulatory sequences to the paralogs’ functional specificity, we compared cells expressing the cDNA of either paralog from identical plasmids and promoters (Fig. 5a, b). The cDNAs were transcribed from the promoter of the 40S subunit protein eS28A/RPS28A as the sole source of uL30 protein in the cell. Unlike the native copy, the plasmids borne copies of uL30A or uL30B equally increased the abundance of total uL30 mRNA produced in the cell (Fig. 5c and Supplementary Fig. 14a). As would be expected, the increased mRNA abundance did not increase the amount of uL30 proteins, as free RPs are rapidly degraded (Supplementary Fig. 14b)27. Importantly, plasmids did not alter the amount of other RPs or rRNA suggesting that the ribosome numbers in these cells remain similar (Supplementary Fig. 14c–e). However, expression of the ORFs from the plasmid eliminated differences in the paralogs’ Kozak scores and thus reduced the preferential association of the A mRNA with heavy polyribosomes (Supplementary Fig. 6). The lower Kozak scores and reduced association with heavy polyribosome of uL30B that favors the translation of uL30A are altered when it is expressed from plasmid. This indicates that the native and plasmid-borne copies of the uL30 are not translated in the same way. Together these results indicate that cells expressing uL30 copies from plasmids generate equal amount of protein forms and ribosomes but differ from wild-type cells in paralog mRNA abundance and translation pattern.

Fig. 5: uL30B promotes translation of long mRNAs and staurosporine resistance independent of the noncoding regulatory sequences.
figure 5

a Endogenous wild-type genes are shown on top and those of plasmids expressing paralogs cDNA from RPS28A promoter and ADH1 terminator sequences are below. Blue, orange and white boxes indicate exons of uL30A, uL30B and introns. Lines show 5′ and 3′ UTRs. P, E1, E2, E3 and snR indicate position of promoter, exon 1–3, and snoRNA. b Strategy for creating haploid yeast strains expressing a single copy of uL30 genes from plasmids. c uL30 mRNA expression from plasmid was quantified by RT-qPCR relative to endogenous mRNA. d Growth of cells harboring R28P-30A or R28P-30B plasmids compared to that of WT in absence (left panel) or presence (right panel) of staurosporine. The results of two-tailed paired t test indicated at bottom. e Polysome profiles of cells expressing uL30 from plasmid. Curves are representative examples from n = 3 biologically independent samples. f The ratio of 25S to 18S rRNA was determined by capillary electrophoresis. g Translation index of mRNAs (expression > 1 TPM) detected in R28P-30A or R28P-30B was compared to that of wild-type. Differentially translated mRNAs (difference > log2 0.5) are shown in red. h Venn diagrams of the number of genes showing differential polyribosome association in (g). i Number of differentially translated genes in each enriched component gene ontology category (p < 0.001 with Bonferroni correction). j Box plots representing the length of differentially translated ORFs. Box limits represent the first and third quartiles, the middle line is the median and whiskers extend to minima and maxima, the number of ORFs in each category is stated under the boxes, the results of two-tailed Mann–Withney test are stated above. k Bar graph showing the relative translation index for B/A paralogs according to mRNA sizes. RNA sequencing for (gk) was performed on n = 2 biologically independent samples. The number of biologically independent samples shown in (c, d, f) is stated above the bars representing the means.

Comparison between cells expressing different versions of uL30 genes from plasmids indicated, that unlike the chromosomal copies, the plasmid-borne versions equally supported growth and subunits production under normal conditions (Fig. 5d–f). This might be due to the difference in expression and translation of the plasmid-borne copies. However, despite the lack of differences in ribosome biogenesis, cells expressing the B form from the plasmid were more resistant to staurosporine than those expressing the A form (Fig. 5d). The B form-specific increase in translation of long ORFs and cell periphery genes was also maintained in cells expressing the paralogs from plasmids (Fig. 5g–i). The differential effects of the uL30 paralog on the translation of long ORFs was also maintained when they are expressed from plasmid and was confirmed using RT-qPCR (Fig. 5j, k and Supplementary Figs. 14f, 15). This indicates that uL30 paralog may differentially alter the translation pattern independent of any regulatory or intervening sequence (e.g., introns or snoRNA) or significant variations in ribosome production or number (e.g., the 60S biogenesis defects). We conclude that the uL30 minor paralog-specific translation and drug resistance phenotypes do not depend on defects in the synthesis of the 60S subunit.

The different functions of the uL30 paralogs are driven by differentially acetylated N-terminal domain sequences

Most of the differences between the uL30 copies are found in the first 42 amino acids of the N-terminal domain (Fig. 6a and Supplementary Fig. 6b). Indeed, the N-terminus includes 4 out of the 5 non-identical amino acid that distinguish between the uL30 versions and features paralog-specific acetylation. In total extract, the N-terminal serine residue of the B form is 100% acetylated, while the alanine residue at the corresponding position of the A version is mostly unmodified (Fig. 6a, b). The acetylation level of the A, and not the B, form depends on the levels and ratios of the two paralogs. In strains containing plasmids encoding the uL30A paralog under the control of a heterologous promoter the percent of acetylated A form was 75%, compared to 10% in wild type. In contrast, expression from a plasmid had no effect on the mostly acetylated B form (Supplementary Fig. 16c). This lack of difference in B form acetylation may explain the reduced difference between the A and B phenotypes when the paralog is expressed from plasmids.

Fig. 6: Post-translation modification alters function of uL30 paralogs.
figure 6

a Amino acid sequences of uL30 paralogs were aligned using Clustal/W2. Different amino acids are indicated in red, amino acids encoded by suboptimal codon are highlighted in cyan and position of N-terminal acetylation is underlined. b N-terminal acetylation was measured by mass spectrometry in wild-type, ul30aa and ul30bb cells. The number of biologically independent samples is indicated at bottom. c N-terminal acetylation level of uL30 proteins detected in light (60S) and heavy (heptamers and heavier) polyribosome fractions in WT cells using mass spectrometry. The number of biologically independent samples is indicated at bottom. The ratio of non-acetylated uL30A over acetylated uL30B in each fraction is reported. d N-terminal acetylation was measured by mass spectrometry in wild-type and uL30B-Ant mutant. The number of biologically independent samples is indicated at bottom. e Relative growth rates of WT and uL30B-Ant mutant after exposure to staurosporine. The results of the two-tailed t test assuming unequal variance are indicated on top. The number of biologically independent samples is indicated at bottom. f Translation index of 5 long mRNAs (3945-6438 nt) that were found down regulated in ul30aa by RNA sequencing was analyzed by RT-qPCR in WT and uL30B-Ant. The result of the paired two-tailed t test of the change in the translation of the indicated genes is shown on the right. g 18S and 25S rRNA was quantified by capillary electrophoresis and reported relative to WT. The number of biologically independent samples is indicated at bottom. h Polysome profiles of WT and uL30B-Ant. Curves are representative examples of biologically independent samples n = 2 WT and n = 4 uL30B-Ant.

The acetylated uL30 proteins were more abundant in heavy polyribosomes, and exposure to staurosporine increased the amount of ribosome incorporating the 100% acetylated B form (Figs. 6c, 2g and Supplementary Fig. 16). Notably, the small proportion of uL30A proteins that were acetylated were also preferably found in heavy polyribosome (Supplementary Fig. 16a). This indicates that ribosomes with acetylated uL30 are more likely to be in heavy polyribosomes, which normally are associated to longer ORFs. We conclude that the minor paralog of uL30 is differentially acetylated and incorporated into actively translating ribosome.

The differential acetylation of uL30 paralogs and their resulting effects on translation and staurosporine resistance depend on the sequence of the N-terminal domain. As indicated in Fig. 6d, swapping the first 42 amino acids of the chromosomal copy of the B form with the A version abolished the hyper-acetylation of the B form. As expected, this swap (uL30B-Ant), which only modifies the amino acid sequence of the uL30B N-terminus, without affecting the overall chromosomal structure of either A or B genes or ribosome numbers, also increased cell sensitivity to staurosporine and reduced the translation of long ORFs (Fig. 6e–h).

Next, we swapped the N-terminal domains of uL30A and uL30B in the context of the same plasmid and promoter sequences (Supplementary Fig. 17a). Swapping the N-terminus on a plasmid thus produced the chimeric protein as the sole source of uL30. This clearly also swapped the effect of the paralogs on growth, staurosporine resistance and translation of long ORFs, with little effect on ribosome numbers, biogenesis, or overall polyribosome profile (Supplementary Fig. 17). The lack of connection between the paralog-specific translation and tolerance to staurosporine is also clear form the general comparison of the rRNA and growth rate of the different strains used in this study (Supplementary Figs. 18 and 19). All strains except those showing uL30A dependent defect in ribosome biogenesis (ul30a∆ and ul30bb) displayed growth rates and rRNA amount similar to that detected in wild-type cells. Exposure to staurosporine also had no effect on ribosome number in any of the strains tested compared to wild type (Supplementary Fig. 19a, b). Interestingly, we found that exposure to staurosporine induces the expression of uL30B when uL30A is absent (ul30a∆ and ul30bb) and represses the expression of uL30A when uL30B is absent (ul30b∆) (Supplementary Fig. 19c). This is consistent with the presumed cells’ preference for expressing uL30B when exposed to staurosporine. We conclude that the N-terminal domain of the uL30 paralogs differentially affects cells tolerance to staurosporine.

Leave a Comment