The dedicated chaperones of eL43, Puf6 and Loc1 can also bind RPL43 mRNA and regulate the production of this ribosomal protein

Abstract

The level of ribosome biogenesis is highly associated with cell growth rate. Because many ribosomal proteins have extraribosomal functions, overexpression or insufficient supply of these proteins may impair cellular growth. Therefore, the supply of ribosomal proteins is tightly controlled in response to rRNA syntheses and environmental stimuli. In our previous study, two RNA-binding proteins, Puf6 and Loc1, were identified as dedicated chaperones of the ribosomal protein eL43, with which they associate to maintain its protein level and proper loading. In this study, we demonstrate that Puf6 and Loc1 interact with RPL43 mRNA. Notably, Puf6 and Loc1 usually function as a dimeric complex to bind other mRNAs; however, in this instance, the individual proteins, but not the complex form, can bind RPL43 mRNA. Thus, Puf6 or Loc1 could bind RPL43 mRNA in loc1Δ or puf6Δ, respectively. The binding of Puf6 or Loc1 caused negative effects for eL43 production: decreased RNA stability and translation of RPL43A/B mRNA. The present results suggest that these dedicated chaperones control the protein levels of eL43 from the standpoint of stability and through regulating its production.

Graphical Abstract

Open in new tabDownload slide

Introduction

The ribosome is a large, complex biomolecule consisting of ribosomal proteins and ribosomal RNAs (rRNAs). More than 200 trans-acting (non-ribosomal) factors have been identified in the ribosome assembly process, modulating correct interaction between ribosomal proteins and rRNAs, processing and folding of rRNAs and regulating the accuracy of ribosome functional domains. The release of one trans-acting factor is usually required to load other factors, ensuring a sequential assembly process (reviewed in (1–5)).Ribosome biogenesis is necessary for cell growth and proliferation and is a major energy cost pathway in a cell; rapidly growing yeast cells generate approximately 2,000 ribosomal subunits per minute (6). To support this enormous synthesis, ~60% of total transcription is devoted to rRNA synthesis. The multiple steps of ribosome biogenesis enable precise regulation of the balance between supply and demand. Cells can regulate both RNA transcription and ribosomal protein synthesis (7). The target of rapamycin (TOR) pathway can transduce nutrient signals and growth factors into cell growth (8–10). Environmental stress (11) and internal insults such as blockage of secretory pathways (12) can quickly stop ribosome biogenesis.

Saccharomyces cerevisiae ribosome comprises 79 ribosomal proteins encoded by 138 ribosomal protein genes (RPGs) (13). Approximately 50% of RNA polymerase II (RNAP II) transcription accounts for the synthesis of ribosomal proteins and the assembly factors in the log phase (14). The supply of ribosomal proteins should meet the transcription levels of rRNAs; therefore, coordinated expressions of RPGs are required to provide similar amounts of each ribosomal protein (15). This is critical for the accuracy of ribosome synthesis and tight control of the extraribosomal functions of ribosomal proteins (16). For tight control of the synthesis of ribosomal proteins, regulation may occur at several levels. (i) Transcription. Many RPGs contain promoters with cis-elements. After Rap1 binds to the promoter, Fhl1 and Ifh1 are loaded for recruitment of RNAP II (17–19) and can also coordinate RPG transcription with rRNA synthesis (20). (ii) Splicing. Unlike higher eukaryotic cells, most genes in S. cerevisiae do not contain introns. However, 74% of RPGs contain introns, compared with 5% of non-RPGs (21). Intron splicing is blocked by different stresses (e.g. amino acid starvation and osmotic stress), decreasing the synthesis of ribosomal proteins (22). In addition, splicing could be autoregulated by its ribosomal protein and applied to regulate the isoform level (23–25). Therefore, introns provide rapid regulation of RPGs in response to environmental stimuli and protein abundance. (iii) Protein stability. Although ribosomes have very long half-lives, ribosomal proteins are very unstable, possibly because of the positive charges and long unstructured extensions of ribosomal protein structures (26). Thus, cells evolved dedicated chaperones of ribosomal proteins to maintain their stability and ensure accuracy of incorporation (26–32).

Yeast Rpl43 and its human homolog L37A were renamed eL43 in the new nomenclature system (33,  34). eL43 is essential and is present in archaea and eukaryotes but not in bacteria. It is a ribosomal protein in the large subunit containing 94 amino acids, located around the E-site of the 60S subunit and in close contact with uL2 (35). These ribosomal proteins are close to the peptidyl transferase centre and are required for 7S rRNA production (36). There are two paralogs of eL43 in yeast, encoded by RPL43A and RPL43B, with 90% exon DNA sequence identity and 100% amino acid sequence identity. RPL43A has a higher expression; hence, loss of this gene, but not RPL43B, results in growth defects (37). Deletion of both paralogs is lethal. Therefore, eL43 is essential in yeast (38). In a previous work, two RNA-binding proteins, Puf6 and Loc1, were required for stability and correct assembly of eL43 (37,  38). In the present study, we found that Puf6 and Loc1 also interact with RPL43 mRNA. We analysed how the binding of Puf6 and Loc1 changed the fate of RPL43 mRNA and the production of eL43 proteins.

Materials and Methods

Strains, plasmids and reagents

All S. cerevisiae strains used in this study are listed in Table I. Unless otherwise indicated, all strains were grown at 30°C in a rich medium (yeast extract peptone) or synthetic dropout medium containing 2% glucose. The plasmids used in this study are listed in Table II. Anti-Loc1, anti-Puf6, anti-eS24 and anti-eL8 antibodies were generated in this laboratory (38). Anti-green fluorescent protein (GFP) was purchased from Bioman. The gel was transferred with Trans-Blot® SD Semi-Dry Transfer Cell for 30 min at 100 V. Signals were detected using Clarity™ Western ECL substrate (BioRad) and scanned with MultiGel-21 (Top Bio, Taiwan). The signal intensity was measured by Image J for quantification.

Strains used in this study

Plasmids used in this study

Identification of the sequences of 5′ UTR and 3′ UTR

5′- and 3′ untranslated region (UTR) sequences were identified with SMARTer® RACE 5′/3′ kit (Takara). For 5′ Rapid Amplification of cDNA Ends (RACE), an RNA adaptor was ligated to the 5′ end of mRNA and cDNA was synthesized by oligo dT. For 3′ RACE, oligo dT with a known DNA sequence extended at 5′ end was used for cDNA synthesis. PCR fragment was amplified with the gene-specific primer and universal primer designed on the RNA adaptor or oligo dT in 5′ or 3′ RACE, respectively. The DNA sequence of the PCR product was analysed to determine the 5′ or 3′ sequence of the target transcript.

Yeast total RNA extraction for quantitative real-time PCR

Total RNA was extracted with the hot phenol-chloroform method. cDNA was synthesized by Reverse Transcriptase (Ambion). The levels of RNA expression were determined by quantitative real-time PCR (Applied Biosystems) with Power SYBR Green (Thermo) detection.

Polysome profile analysis

Yeast cells were grown to an OD₆₀₀ of 0.2–0.3. Cycloheximide was added to a final concentration of 50 μg/ml and incubated for 10 min. Cells were lysed by glass beads in polysome lysis buffer (20-mM Tris HCl pH 7.5, 8-mM MgCl₂, 12-mM mercaptoethanol, 100-mM KCl, 50-μg/mL cycloheximide, 1-mM PMSF, 1-mM leupeptin). 9 OD₂₆₀ units of protein extracts were loaded onto linear 7–47% sucrose gradients. After 2.5 h of centrifugation at 4°C 40,000 rpm in SW41 Ti rotor (Beckman), gradient fractions were continuously measured absorbance at 254 nm and collected (BR-188, Brandel). Sucrose fractions were extracted with Trizol® (Thermo) and chloroform for RNA extractions. The levels of RNA were quantitated by quantitative PCR (qPCR). qPCR primers used in this study are listed in Table III.

qPCR primers used in this study

In vitro RNA binding assay and RNA stability assays

PUF6, LOC1 and RPL43 were cloned into pGEX-4T3 or pET28 vectors and overexpressed in Escherichia coli BL21 (DE3). Protein expressions were induced with 0.5-mM IPTG. Cell extracts were prepared by sonication (Chrom Tech, Taipei, Taiwan) in lysis buffer (25-mM Na-phosphate, 150-mM NaCl, 3-mM KCl, 1-mM MgCl₂, 0.1% Tween 20, pH 7.3). Whole-cell extracts containing GST-tagged or His₆-tagged proteins were incubated with glutathione beads or NTA beads for 1 h at 4°C and washed with chilled buffer three times. To purify the Puf6/Loc1 complex, GST-Puf6/Loc1-His₆ or GST-Loc1/Puf6-His₆ were co-purified with glutathione beads.

T7 promoter sequence (GCATGCTAATACGACTCACTATAGGG) was added at the 5′ end of the primer. The desired DNA fragment was amplified by PCR. 1-μg DNA, NTP buffer mix and T7 polymerase (NEB) were incubated at 37°C for 2 h to transcribe target RNA in vitro. RNA was added and incubated with protein-coated glutathione beads in lysis buffer with RNase inhibitors (Invitrogen) for 1 h at 4°C. After washing three times, the bound RNAs were extracted with Trizol® and quantitated using RT-qPCR.

For in vitro RNA stability assay, the RNA containing P_43B-I-GFP-3′UTR or P_43B-I-GFP was transcribed by T7 polymerase (NEB) in vitro. 4-pmol RNA was mixed with 12-pmol purified GST, GST-Puf6 or GST-Loc1 protein on ice. Then, 0.7-μg RNase A (Thermo; DNase and proteinase-free-grade) was added in each reaction. After a 5-min incubation at 37°C, 500-μL Trizol® was added to stop the reaction. The remaining RNA was quantitated by qPCR. To detect RNA stability in vivo, normalized protein extracts were incubated at 37°C for 10 and 20 min. RNA were extracted with Trizol® and quantitated by qPCR. The RNA levels at T₀ were set as 1 and the relative amounts were calculated.

Results

Individual Puf6 or Loc1 proteins but not the dimeric complex can bind RPL43 mRNA

In our previous study, Puf6 and Loc1 are shown as the dedicated chaperones of eL43 (37,  38). Puf6 and Loc1 are RNA-binding proteins (39,  40). Proteins containing the PUF domain have been shown to recognize the 5′ or 3′ UTR and control stability, translation and localization of mRNAs (reviewed in (41)). Loc1 has been shown to work with Puf6 on ASH1 mRNA (39). Thus, we examined if they could bind and regulate RPL43 mRNA.

The transcript sequences of these two paralogs were first identified. Their 5′- and 3′ UTR sequences were identified with 5′ or 3′ RACE. Their 5′ UTRs were similar in size, 31 and 27 nucleotides in RPL43A and RPL43B, respectively. RPL43B has a longer 3′ UTR, whereas RPL43A has a longer intron. The first exon of both RPL43A and RPL43B only contains the first two nucleotides of the start codon (Fig. 1A). The amino acid sequences of Rpl43A and Rpl43B are identical, and their exon DNA sequences show 89.6% identity (Supplementary Fig. S1).

Purified recombinant GST and GST-tagged Puf6 and Loc1 proteins were immobilized on glutathione beads and applied in RNA-binding assays (Fig. 1B–E). To exclude the efficiency of different primers in qPCR analysis, exon 2 was replaced with GFP, 5′ UTR, intron and 3′ UTR from RPL43A or RPL43B were constructed as indicated on the figure and used as templates for in vitro transcription (Fig. 1B). The same amount of RNAs were added and incubated with protein-coated beads. The levels of RNAs bound on the GST–Puf6 or GST–Loc1 were quantitated with qPCR using primers recognized GFP and were calculated against the background, the bound RNAs on GST.

Puf6 could bind both RPL43A and RPL43B mRNAs. The presence of an intron enhanced Puf6 binding on RPL43A mRNA but did not change the binding with RPL43B mRNA. By contrast, Loc1 slightly bound intron-containing RPL43A mRNA. Removal of the intron enhanced Loc1 binding on the mRNAs of both paralogs (Fig. 1B). To further dissect the potential binding region, the 5′ UTR, intron and 3′ UTR RNAs from RPL43A or RPL43B were transcribed separately in vitro and applied in binding assays (Fig. 1C and D). Compared with the 5′ UTR- and intron-containing sequences, Puf6 (Fig. 1C) and Loc1 (Fig. 1D) had a higher binding ability with the 3′ UTR from both paralogs.

Because Puf6 and Loc1 have been shown to work together on ASH1 mRNA (42) or eL43 protein (37,  38), we examined if the Puf6–Loc1 complex showed stronger binding with RPL43B mRNA. Notably, the complex showed reduced binding with this mRNA (Fig. 1E). To further check if the protein interaction competed with mRNA binding, the Loc1–RPL43 mRNA complex was first constituted. A different amount of Puf6 was added, and the remaining RPL43 mRNA on the Loc1 was analysed. Consistently, increasing the addition of Puf6 decreased the Loc1-bound RPL43 mRNA (Fig. 1F). This result suggests that the dimeric Puf6 and Loc1 complex does not bind preferentially to RPL43 mRNA; instead, each protein may only bind RPL43 mRNA when present alone (i.e. in puf6Δ or loc1Δ strains) or when one dissociates from the other.

Loss of Puf6 or Loc1 disturbs the transcription of reporters containing 5′ UTR, intron or 3′ UTR of RPL43B

To further dissect how Puf6 and Loc1 changed the fate of RPL43 mRNA, reporters with different parts of the RNA (i.e. 5′ UTR, intron and 3′ UTR) were constructed to drive the synthesis of GFP (Fig 2A and Supplementary Fig. S2A).

When GFP was driven by the RPL43A promoter (Supplementary Fig. S2A), the presence of the RPL43A intron or 3′ UTR did not change either the protein level (Supplementary Fig. S2B) or mRNA level of GFP (Supplementary Fig. S2C). We compared the expression levels of the reporters in puf6∆ and loc1∆ and found the GFP levels remained unchanged in either mutant and were similar to those of wild type (Supplementary Fig. S2B).

However, when GFP was driven by the RPL43B promoter (Fig. 2A), the presence of the corresponding 3′ UTR decreased the GFP level (Fig. 2B, lanes 2 versus 4). To see if this 3′ UTR sequence also decreased the expression of other promoters, we put it in conjunction with the promoter of another ribosome biogenesis factor, SSF1, and found it also reduced the expression of reporters by ~80% (Fig. 2B, lanes 5 versus 6).

We subsequently added the RPL43B intron back to the system for further analysis. The presence of the RPL43B intron alone did not change the GFP level when driven by its promoter (Fig. 2B, lanes 3 versus 4). By contrast, the intron could relieve the repression effects of the 3′ UTR (Fig. 2B, lanes 1 versus 2). We further quantitated the GFP mRNA levels by qPCR and obtained similar results (Fig. 2C). Therefore, the 3′ UTR can negatively regulate the level of RPL43B mRNA, and the presence of its intron can neutralize this effect.

We further compared the expression levels of the reporters in puf6∆ and loc1∆. Notably, the ability of the intron to neutralize the negative effect of the 3′ UTR was much weaker in puf6∆ (Fig. 2D, lanes 1 versus 2 and 2E) and slightly decreased in loc1Δ (Fig. 2F and G). Thus, in the absence of Puf6, the interconnection between intron and 3′ UTR was disturbed.

Puf6 and Loc1 recognize the unique secondary structure instead of the sequence of RPL43B 3′ UTR

To further examine how Puf6 and Loc1 affected RPL43B mRNA, the interaction between Puf6 and RPL43B mRNA was selectively disrupted. Sequence searching revealed a potential Puf6 binding site containing two TTGT tetranucleotides (40) within the RPL43B 3′ UTR, but not in the RPL43A 3′ UTR (Fig. 3A, red site). To test if this sequence altered Puf6 binding, it was deleted in a P_L43B–I–GFP–3′ UTR reporter (PKL680) to generate a P_L43B–I–GFP–(3′ UTRΔPuf6) reporter (PKL874). The GFP expression level of PKL874 was examined and found to be much lower than WT (PKL680), similar to an intron-less reporter (PKL752; Fig. 3B). To exclude the possibility that the shorter 3′ UTR may have different repressing strength, another 16-nt sequence just downstream of the stop codon (Fig. 3A, blue site) was also deleted. In contrast to the previous result, this constructs (Fig. 3B, PKL875) maintained an expression level similar to WT (Fig. 3B, PKL680).

To further verify whether Puf6 indeed recognized this binding site sequence, the two TTGT tetranucleotides were mutated to AACA, which was demonstrated to abolish Puf6 binding on ASH1 mRNA (43). However, no changes in GFP protein levels were detected (Fig. 3C, PKL924). We examined binding between Puf6 and Loc1 with an RNA fragment containing a mutation of the Puf6 binding site (Fig. 3D). Notably, Puf6 and Loc1 showed increased binding with 3′ UTR(AACA) and even higher binding with 3′ UTR(ΔPUF6) mutated RNA. These data suggest that the decreased expression of reporter PKL874 was not from loss but from gaining of Puf6.

Because the dimeric Puf6/Loc1 did not bind RPL43 mRNA (Fig. 1E), the reduced expression of the reporter in puf6Δ or loc1Δ might result from an adverse effect from the binding Loc1 or Puf6, respectively. A mutation of the 3′ UTR, which showed stronger binding with Puf6 or Loc1, also reduced expression of the GFP reporter (Fig. 3D). We elevated each single protein by overexpressing either Puf6 or Loc1 in cells to verify this conjecture. Both manipulations reduced the expression levels of GFP reporters in the cells (Fig. 3E), consistent with the previous results. Thus, the binding of Puf6 or Loc1 negatively affected RPL43 mRNA.

Unlike other PUF family proteins, the PUF domain of Puf6 exhibited a unique L-like shape. It was proposed to recognize target RNA in a structure-dependent (but not sequence-dependent) manner (44). From RNA secondary structure prediction (MXfold2 Server) (45), 3′ UTR has four stem-loops. The 16-nt deletion site (Fig. 3F, labelled by the blue line) is at Stem 1, and the potential Puf6 binding site (Fig. 3F, labelled with red dots) is at Stem 3. Point mutations of the Puf6 binding site (Fig. 3F, AACA: labelled by green dots) changed the structure and position of Stem 3. Truncation of the potential Puf6 site abolished the extended structure of Stem 3. Thus, Stem 3 is probably the primary recognition site by Puf6 and Loc1.

Binding of Puf6 or Loc1 destabilizes RPL43B mRNA

As Puf6 and Loc1 can bind RPL43 mRNA, we further dissected their potential influences. Previous results (Fig. 2D–G) indicated that the respective introns decreased the ability to alleviate repression by the RPL43B 3′ UTR in puf6Δ and loc1Δ. We first investigated whether Puf6 and Loc1 changed the splicing efficiency. RNA was extracted from WT, puf6Δ and loc1Δ, and the levels of both mature and immature mRNAs of RPL43A and RPL43B were quantitated by qPCR using the primers indicated in Fig. 4A. However, no increases of nascent mRNAs were observed (Fig. 4A).

Because multiple PUF proteins were shown to regulate the stability of ribosome biogenesis transcripts in yeast (46), we further tested if the binding of Puf6 or Loc1 changed the stability of RPL43B mRNA (Fig. 4B). The P_L43B–I–GFP–3′ UTR RNA fragment was transcribed by T7 RNA polymerase in vitro and incubated with purified GST–Puf6 and GST–Loc1 proteins. RNase A was subsequently added and incubated for 5 min. The levels of remaining RNA were detected by qPCR. Compared with the control, the addition of Puf6 or Loc1 decreased the RNA. To further check the effect was due to the specific binding of Puf6 or Loc1 to 3′ UTR, 3′ UTR was removed from the mRNA for parallel comparisons. Although Puf6 addition decreased RNA abundance compared to the control, the level was 4-fold higher than 3′ UTR containing RNA. Thus, the decrease might be from the impurities in the purified proteins. In contrast, Loc1 addition showed comparable RNA levels as the control (Fig. 4B).

To further check RPL43 mRNA stability in vivo, we tracked RNA stability in WT, puf6∆ and puf6∆loc1∆ (Fig. 4C). The decay rate of ACT1 mRNA was similar in the three strains. While RPL43 mRNA was decreased slightly faster in puf6∆, it was stabilized in puf6∆loc1∆ (Fig. 4C). The data above suggest that binding of Puf6 or Loc1 enhance the decay rate of RPL43 mRNA.

Binding of Puf6 or Loc1 decreases the translation efficiency of RPL43 mRNA

Since Puf6 and Loc1 could bind RPL43 mRNA (Fig. 1), they may also regulate the translation of RPL43 mRNA. The translation situations of RPL43 mRNAs were examined in WT, puf6∆, loc1∆ and puf6∆loc1∆ across sucrose gradients. Cell extracts were fractionated into free proteins, the 40S, 60S, 80S and polysome (Fig. 5A), and each fraction was collected for RNA extraction. cDNAs were prepared by reverse transcriptase from mRNAs and quantified by qPCR. The distribution of mRNAs in the gradient correlates with translation efficiency. Non-translated mRNAs are at free and non-polysome fractions. Translated mRNAs are further separated as heavy and light polysome peaks, standing for actively and less actively, respectively (Fig 5B).

Although the major translation pools of actin (ACT1) mRNA were at heavy polysome fractions, the major distribution of RPL43A and RPL43B mRNAs was at light polysome peaks in WT (Fig. 5C). These might be because of the relative sizes of the respective mRNAs: the exon sizes of ACT1 and RPL43 are 1,128 and 276 bp, respectively. The distribution of ACT1 was not significantly altered in WT, puf6∆, loc1∆ or puf6∆loc1∆ (Fig. 5C, top panel). For RPL43A/B mRNA, ~60–70% (sum of light and heavy polysome fractions) was at translation status in WT (Fig. 5C, middle and bottom panels). The translated fraction of RPL43A/B mRNA decreased in puf6Δ and even more in loc1Δ but partially restored in puf6∆loc1∆ (Fig. 5C). These data suggest that Puf6 or Loc1 binding decreased the translation efficiency of RPL43 mRNAs.

Loc1 was shown only distributed in the nucleus, but Puf6 could be exported and repressed the translation (47). To check if Puf6 tethered on the RPL43 mRNA in loc1∆, the cell extracts of WT and loc1Δ with or without Puf6-myc were fractioned through sucrose gradients (Fig. 5D). The free RNAs, the 40S, 60S, 80S and polysome peaks were collected separately and applied for immunoprecipitation. The RNA levels bound by Puf6 were quantitated by qPCR (Fig. 5E). When compared Puf6-myc and no-tag control in WT, no increase in binding levels were observed. Thus, Puf6-myc did not bind RPL43 mRNA in WT. Compared with the binding levels between WT and loc1Δ, Puf6 bound more RPL43 mRNA at the 40S and 60S fractions. These data imply that Puf6 bound on RPL43 mRNA in loc1∆ and represses its translation at initiation status.

Discussion

In a previous work, we demonstrated that Puf6 and Loc1 form a heterotrimer with eL43 for protection of stability and accuracy of accommodation (37,  38). The concave surface of the PUF domain is a large RNA-binding interface, which binds the recognition sequence of mRNAs. PUF proteins can either repress or activate mRNA expression and target mRNAs to specific subcellular localizations for the spatial expression of proteins (41,  48). Puf6 and Loc1 have been shown to form a heterodimer and regulate the cellular distribution of ASH1 mRNA (42,  43,  49). Protein levels of eL43 are reduced in puf6Δ or loc1Δ from destabilization (37,  38). In the present study, we found that Puf6 and Loc1 also regulate eL43 production.

Puf6 or Loc1 alone but not the dimeric complex can bind RPL43 mRNA

In contrast to the ASH1 mRNA model, either Puf6 or Loc1 alone could interact with RPL43 mRNA, but the dimer exhibited substantially decreased binding. One possible reason is that the Puf6–Loc1 dimer binds preferentially to the nascent eL43 protein. However, each monomer may only bind the mRNA in the absence of the other one to attenuate the expression level of eL43. Since RNA-binding proteins play crucial roles in the fates of mRNAs, regulating the activity of these proteins is critical. RNA-binding proteins may be regulated in higher eukaryotic cells through transcription, splicing, posttranslational modifications and miRNA (50). In S. cerevisiae, there are no introns in PUF6 and LOC1 and no miRNAs. Hence, the regulation would most likely be through transcription and posttranslational modifications. The phosphorylation sites of Puf6 and Loc1 were identified in several phosphoproteome assays. The potential phosphorylation sites on Puf6 are Ser34, Ser35 and Ser39, in response to TORC1 activation (51,  52) and Cdk1 kinase (53). The only phosphorylation site of Loc1 is at Ser24, in response to Cdk1 kinase (53). Therefore, the association of Puf6 or Loc1 with mRNA might be controlled in response to cell status to regulate the levels of eL43. Notably, we found that the protein level of Puf6 but not Loc1 decreased rapidly in response to nitrogen starvation (Supplementary Fig. S3A). Loc1 but not Puf6 was detectable at stationary-phase cells (Supplementary Fig. S3B t = 0). Their protein levels increased at a nutrition-rich medium (Supplementary Fig. S3B  t = 60, 120). Thus, the protein levels of Puf6 and Loc1 may change in response to environmental conditions and regulate the expression of Rpl43.

Binding of Puf6 or Loc1 to RPL43 mRNA negatively regulates the production of eL43

To study how its intron and 3′ UTR potentially regulate the RPL43 mRNA, the intron or 3′ UTR sequence was included or excluded in the reporters driven by the gene’s promoter. Notably, the expression levels of reporters were not changed by either the RPL43A intron or 3′ UTR. However, the 3′ UTR of RPL43B alone decreased the reporter expression considerably; the presence of the corresponding intron could restore the expression level.

Our results indicate that in puf6Δ, the intron lost the ability to eliminate suppression by the RPL43B 3′ UTR. This suggests that (1) the presence of Puf6 is required for restoration and (2) Loc1 binding on RPL43B mRNA alters the intron effect. We believe the second possibility is more likely. Because the Puf6–Loc1complex exhibits decreased binding with RNA, Loc1 may only stably associate with RPL43 mRNA in puf6Δ.

Our sequence search revealed one potential Puf6 binding site (TTGTNNTTGT) in the RPL43B 3′ UTR but not found within the RPL43A 3′ UTR. Although the reporter with point mutations at this site maintained normal expression levels, the reporter with truncation of this sequence exhibited decreased expression levels as in puf6Δ. In contrast to our expectation, Puf6 and Loc1 demonstrated increased interaction with 3′ UTR(AACA) and even stronger interaction with 3′ UTR(ΔPuf6) sequence in vitro. From the previous study, Puf6 and its human homolog Puf-A exhibited distinctive L-shapes and did not demonstrate sequence specificity towards mRNA or rRNA (44). From the RNA structure prediction, the mutations changed the structures and positions of stem 3 but not to other stems. Taken together, these results suggest that Puf6 and Loc1 may recognize the unique RNA secondary structure of the 3′ UTR.

Our findings also indicated that Puf6 or Loc1 binding did not change the splicing but decreased mRNA stability. We found Loc1 showed higher binding with RPL43 RNA with no intron. Thus, the presence of intron may change the secondary structure of 3’UTR and alter the protein binding. Since the binding of Puf6 or Loc1 enhanced RNA susceptibility to RNase, one possibility is that protein binding at 3′ UTR changed the mRNA folding, leading to rapid degradation of this molecule. In WT, Puf6 and Loc1 formed a dimeric complex and did not interact with RPL43 mRNA. Thus, the negative regulation only happens in puf6∆, loc1∆ or physiological conditions that cause dissociation between two proteins and destabilization of one of the two proteins.

In addition, Puf6 and Loc1 binding decreased the translation efficiency of RPL43 mRNA. Although the translation of ACT1 was not altered among different stains, RPL43A/B mRNAs displayed a 20–30% reduction in puf6Δ and a 30–50% reduction in loc1Δ. The reduction of RPL43B was more significant than that of RPL43A. To demonstrate the Puf6 role in translation repression, the translation of RPL43A/B was tracked in puf6Δloc1Δ. Indeed, the deletion of Puf6 restored the translation efficiency. Thus, the translation repression in loc1Δ was partially from Puf6 binding. While the translation levels were similar in puf6Δ and puf6Δloc1Δ, it implies that Loc1 binding in puf6Δ did not decrease translation.

How Puf6 binding in loc1Δ represses the translation of RPL43 mRNA remains unknown. In ASH1 mRNA, the translational repression of Puf6 requires an RNA-dependent interaction with the general translation initiation factor Fun12 (eIF5B) (49), which interferes with the conversion of 48S to 80S during initiation. The binding of Puf6 on 5′ UTR of STE12 mRNA represses its translation via regulation of RNA helicase Dhh1 (54). We speculate that Puf6 may repress the protein synthesis of eL43 through a similar mechanism. In puf6Δ, the translation of RPL43 mRNA slightly decreased. Loc1 has been demonstrated as a nuclear protein (39), and it was not found at 80S fractions in puf6Δ. Thus, the potential mechanism still requires to be elucidated.

We previously showed that both Puf6 and Loc1 were required for the correct assembly and stability of eL43 (37,  38). In this study, we found loss of either one decreased the production of eL43 post-transcriptionally. Thus, this is also an additional mechanism for the dedicated chaperone to regulate its associated ribosomal protein level.

Authors’ contributions

KYL conceived and coordinated the study and wrote the paper. LYY, YTT and CYC designed, performed and analysed the experiments. All authors reviewed the results and approved the final version of the manuscript.

Supplementary Data

Supplementary Data are available at JB Online.

Acknowledgments

We thank the staff of the Taiwan Yeast Bioresource Center at the First Core Labs, National Taiwan University College of Medicine, for bioresources sharing.

Funding

This work is financially supported by the Ministry of Science and Technology of Taiwan (MOST 106-2313-B-002-031-MY3 and MOST 109-2313-B-002-023-MY3) and the National Taiwan University Career Development Project (108-110 L7862).

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.

References

Author notes