Optimised Ribosome Profiling Reveals New Insights Into Translational Regulation in Synchronised Chlamydomonas reinhardtii Cultures

ABSTRACT

Deep sequencing of ribosome footprints, also known as ribosome profiling (Ribo‐seq), enables the quantification of mRNA translation and a comprehensive view of the translatome landscape. Here, we report an optimised Ribo‐seq protocol and analysis pipeline for the model green alga, Chlamydomonas reinhardtiii (Chlamydomonas). Compared to the previously published data sets, the ribosome‐protected fragments generated by our protocol showed improved mapping rates to the main open reading frames, reduced bias mapping to the gene coding regions and high 3‐nt footprint periodicity. Using this optimised protocol, we employed Ribo‐seq alongside RNA‐seq to compute translation efficiency and identify genes with differential translation during the diurnal cycle. Interestingly, we found that the translation efficiency of many core cell cycle genes was significantly enhanced in cells at the early synthesis/mitosis (S/M) stage. This result suggests that translational regulation plays a role in cell cycle regulation in C. reinhardtii. Furthermore, the high periodicity of ribosome footprints allowed us to identify potential C. reinhardtii upstream open reading frames (uORFs). Further analysis revealed that some of these uORFs are differentially regulated and may play a role in diurnal regulation. In summary, we used an optimised Ribo‐seq protocol to generate a high‐quality Ribo‐seq data set that constitutes a valuable resource for Chlamydomonas genomics. The ribosome profile data is linked to the Chlamydomonas reference genome and accessible to the scientific community.

Summary statement

We report optimised ribosome profiling (Ribo‐seq) of Chlamydomonas reinhardtii. This protocol compared favourably to published data sets and facilitated studies of translational regulation, identification of uORFs and improved annotation of gene models.

1. Introduction

Ribosome profiling (also referred to as Ribo‐seq) is a high‐throughput technique that captures ribosome positions on the transcriptome and enables global translation study (Andreev et al. 2017; Brar and Weissman 2015; Ingolia et al. 2009; Wu et al. 2019). In Ribo‐seq experiments, a non‐specific ribonuclease digests the ribosome‐bound mRNA and releases ribosome‐protected mRNA fragments (RPFs, also termed ribosome footprints). RPF sequencing allows the identification and quantification of ribosome occupancy on the transcriptome. Unlike RNA‐seq, which captures random fragments of the entire mRNAs, Ribo‐seq, which does not typically cover information before the start codon and after the stop codon, enables identification of the translated open reading frames (ORFs). In addition, Ribo‐seq is a valuable technique for providing insights into the dynamic translation activities including the identification of differentially translated mRNAs (Ingolia et al. 2009; Xiao et al. 2016), upstream open reading frames (uORFs) (Hsu et al. 2016; Ingolia et al. 2011; Wu et al. 2024) and ribosome stalling sites (Duncan et al. 2018; Ingolia et al. 2009; Rubio et al. 2021).

The unicellular green alga Chlamydomonas reinhardtii (designated here as Chlamydomonas) is an important microbial system for studies of chloroplast biology, photosynthesis, cilia structure and functions, cell cycle and lipid biology (Cross and Umen 2015; Dupuis and Merchant 2023; Huang et al. 2024; Li‐Beisson et al. 2015; Marshall 2024; Salomé and Merchant 2019). The haploid genome of Chlamydomonas is especially advantageous for genetic studies, as loss‐of‐function mutations lead to observable phenotypes more readily than in diploid organisms. In addition, the fully sequenced genome (Craig et al. 2023; Merchant et al. 2007), equipped molecular toolkits (Chen et al. 2023; Emrich‐Mills et al. 2021; Kong et al. 2019; X. Li et al. 2025; Van de Vloet et al. 2025) and availability of the mutant library (X. Li et al. 2019; X. Li et al. 2016; Tulin and Cross 2014) have made Chlamydomonas an important reference for the rapidly expanding fields of algal biotechnology (Blaby‐Haas and Merchant 2019; Crozet et al. 2018).

Chlamydomonas reproduces through a multiple fission type of mitotic cell cycle, which allows temporal separation of growth and division by alternating between light and dark periods. Multiple fission has a long G1 phase during which daughter cells grow photosynthetically and can enlarge to many times the original size. Following G1, mother cells undergo n rapid rounds of alternating S phase (DNA synthesis) and M phase (mitosis) to produce 2 ⁿ daughter cells. Under a typical diurnal cycle, growth occurs during the light phase, and cell division (S/M) occurs in the dark. As a result, the cell cycle becomes naturally and highly synchronised (Fang et al. 2006).

Ribo‐seq has been used to study translational regulation in Chlamydomonas (Chung et al. 2015; Gotsmann et al. 2024). However, a thorough evaluation of data sets generated by different protocols has not been rigorously conducted. Moreover, experimental variables, such as buffer conditions and the amount of RNase I, which can be adjusted to optimise footprinting conditions (Douka et al. 2021; Hsu et al. 2016), have not been systematically tested in Chlamydomonas. Here, we compared different Ribo‐seq data sets and report an optimised protocol. Over 94% of the footprints mapped to the main open reading frames, providing high‐resolution profiles for individual transcripts and allowing for precise definitions of translated regions. These resulting data not only helped refine the annotated models in the Chlamydomonas genome but also led to the identification of many potential uORFs. In addition, we applied the optimised Ribo‐seq protocol to synchronised cell cultures collected at the G1 and S/M stages. After normalising Ribo‐seq data by the RNA‐seq reads, we identified 1688 differentially translated genes, many of which are involved in DNA replication. This result suggests that translational regulation plays a role in the cell division mechanism in Chlamydomonas.

2. Materials and Methods

2.1. Strains and Growth Conditions

C. reinhardtii strain 21gr (CC1690, MT⁺) was used in all experiments. Cells were maintained on a Tris‐acetate‐phosphate (Gorman and Levine 1965) agar plate. Cell cultures were grown in high salt medium (HSM, Sueoka 1960) under illumination at 150 μmol photons m⁻²s⁻¹ blue (450 nm) and 150 μmol photons m⁻²s⁻¹ red (630 nm) LED lights at 24°C in the presence of 0.5% (v/v) CO₂ aeration. Culture synchronisation was induced by growth in 12‐h‐light/12‐h‐dark cycles and maintained at 1–2 × 10⁶ cells per mL.

2.2. Polysome Extraction

Polysome extraction was carried out as described previously (Chung et al. 2015) with slight modifications. Briefly, ~10⁸ cells were harvested by centrifugation at 3220 × g for 5 min at room temperature (RT). Cells were flash frozen and pulverised with 700 μL of frozen polysome extraction buffer (Buffer A, B, B⁺, or C [Table 1]), supplemented with 10 U/mL RNase free DNase I (Qiagen, Cat. No. 79254), 1× protease inhibitor cocktail (Sigma‐Aldrich, P9599), 1 mM PMSF, 1 mM benzamidine and 10 μM MG132 in the presence of liquid nitrogen. The pulverised powder was allowed to thaw on ice for ~30 min, followed by centrifugation at 4200 × g for 30 min at 4°C to remove cell debris. The RNA concentration of the supernatant was determined using a Qubit RNA Broad Range assay kit (Life Technologies, Q10210). The polysome lysates were snap‐frozen and stored at −80°C.

Table 1.

Composition of the tested polysome extraction buffers.

Buffer composition	Buffer A	Buffer B	Buffer B+	Buffer C
Tris‐HCl	20 mM, pH 7.5	20 mM, pH 7.5	20 mM, pH 7.5	100 mM, pH 7.5
KCl	140 mM	140 mM	140 mM	40 mM
MgCl2	5 mM	25 mM	25 mM	5 mM
DTT	1 mM	1 mM	1 mM	1 mM
Cycloheximide	150 μg/mL	150 μg/mL	150 μg/mL	150 μg/mL
Chloramphenicol	100 μg/mL	100 μg/mL	100 μg/mL	100 μg/mL
Sucrose	5%	5%	5%	5%
NP40	0.5%	0.5%	0.5%	0.5%
Triton X‐100	1%	1%	1%	1%
RNase Ia	1.25 μ/μg	1.25 μ/μg	3.75 μ/μg	1.25 μ/μg

^a RNase I was added only for the isolation of ribosome‐protected fragments (RPFs).

2.3. Generation of RPFs

The frozen 200 μL ( ~ 400 μg of RNA) of polysome lysate was thawed on ice and incubated with 500 units of RNase I (Thermo Fisher, Ambion, AM2294) at RT for 30 min with gentle shaking (25 rpm end‐to‐end rotation). The digestion reaction was stopped by adding 15 μL (300 units) SUPERase•In RNase inhibitor (Invitrogen, AM2694) on ice. Monosomes were isolated by applying digested lysate onto the prepared size exclusion columns (Amersham MicroSpin S‐400 HR columns, Cytiva, 27‐5140‐01) and spun at 600 × g for 2 min. The RNA fragments larger than 17 nt were further purified by RNA Clean & Concentrator‐25 kit (Zymo Research, R1017). The purified RNAs were then separated by 15% (w/v) TBE‐Urea Gels (Novex, EC6885BOX) and visualised by SYBR Gold Nucleic Acid Stain (Invitrogen, S11494). The RNAs in the range of 26 to 34 nucleotides were excised. To purify the RNA, gel slices were ruptured by centrifugation through holes at the bottom of 0.5 mL tubes nested in 1.5 mL tubes, as described previously (Uzun et al. 2021). RNAs were eluted by soaking gel pieces in the 400 μL gel elution buffer (300 mM NaOAc, pH 5.5 and 0.05% SDS) at 4°C overnight. The gel debris was removed using a Spin‐X column (Corning, 8162), and RNA was precipitated by adding 700 μL ice‐cold 100% isopropanol supplemented with 40 μg glycogen as a carrier at −20°C overnight. RNA fragments were collected by centrifugation at 16 100 × g for 30 min at 4°C, and the RNA pellet was air‐dried and dissolved in 10 μL nuclease‐free H₂O.

2.4. Ribo‐Seq Library Construction

After purification of RFPs, ~2–4 μg of purified RPFs were dephosphorylated by T4 polynucleotide kinase (T4 PNK, New England Biolabs, M0201S) at 0.5 unit/μL supplemented with the SUPERase•In RNase inhibitor (1 unit/μL, Invitrogen, AM2694) in a 20 μL reaction at 37°C for 1 h. The T4 PNK activity was inactivated by heating at 75°C for 10 min. The dephosphorylated RPFs were precipitated with 15 μg glycogen (as a carrier) by adding 0.5 volumes of 3 M NaOAc (pH 5.5) and 1.5 volumes of ice‐cold 100% isopropanol and incubating at −20°C overnight. The dephosphorylated RPFs were harvested by precipitation at 16 100 × g for 30 min at 4°C, and the RNA pellet was washed with 80% ice‐cold ethanol, centrifuged at 16 100 × g for 20 min at 4°C, air‐dried, and dissolved in 10 μL nuclease‐free H₂O. Approximately 1.5–2.5 μg of purified dephosphorylated RPFs were subjected to Ribosomal RNA (rRNA) depletion using RiboMinus Plant kit (Invitrogen, A1083808, Ribo‐Minus) or Ribo‐off rRNA Depletion Kit (Plant) (Vazyme N409, Ribo‐Off) following the user manuals. The RNA quantification and quality were analysed by Agilent 2100 Bioanalyzer using the Agilent Small RNA Kit (Agilent, Part Number: 5067–1548). Approximately 1−2 ng of rRNA‐depleted RPFs were used to make cDNA libraries using the SMARTer smRNA‐Seq kit for Illumina (Takara, 635030) as described in the user's manual. The cDNAs were amplified by 17 cycles of PCR, and the quality of cDNA libraries was validated by Agilent TapeStation using the High Sensitivity D1000 ScreenTape System (Agilent, Part Number: 5067–5584). Equal molarity of the barcoded libraries was pooled for single‐end 75‐nt sequencing in a MiniSeq or NextSeq. 550 platform.

2.5. Polysome Profiling

The sucrose gradients were prepared using a Gradient Master 108 (BioComp Instruments) with a polypropylene centrifuge tube (Beckman, No. 331374). Approximately 400 μg of RNA of the polysome lysate was loaded on a pre‐cooled continuous 10%–50% (w/v) sucrose (HPLC‐grade) gradient containing 20 mM Tris‐HCl (pH 7.5), 140 mM KCl, 25 mM MgCl₂, 1 mM DTT, 100 μg/mL cycloheximide, 100 μg/mL chloramphenicol and 20 units/mL SUPERase•In RNase inhibitor (Invitrogen, AM2694). The samples were spun in a SW 40 Ti swinging‐bucket rotor (Beckman Coulter) at 260 343 × g for 2 h at 4°C. Ribosomes were fractioned using a density gradient fractionation system (Brandel). The noise setting and sensitivity were set to 0.5. Polysome profiles were monitored by optical density measurement (A254) using a UA‐6 Detector (Teledyne ISCO) and collected by FoxyR1 fraction collector (Teledyne ISCO).

2.6. Immunoblotting

Fifteen microliters of the ribosome fraction was resolved in a 12% Bis‐Tris Plus Bolt (Invitrogen) and transferred to Immnobilon‐P PVDF membrane (Merck Millipore). Blots were blocked in 1× TBST (20 mM Tris base, 150 mM NaCl, 0.1% [v/v] Tween 20) containing 5% (w/v) non‐fat milk for 1 h at RT. For the RPL30 protein, blots were incubated with anti‐RPL30 antiserum (Lin et al. 2020) diluted 1:3000 in 1× TBST with 5% (w/v) non‐fat milk at 35°C for 1 h. For the histone 3 (H3) and β subunit of ATP synthase (AtpB) proteins, blots were incubated with anti‐H3 antiserum (Agrisera AS10710) or anti‐AtpB antiserum (Agrisera AS05085) diluted 1:5000 at 4°C overnight. Blots were then washed three times for 15 min each and then incubated with horseradish peroxidase–conjugated goat‐anti‐rabbit‐IgG (1:20 000, PerkinElmer NEF812) diluted in 1× TBST with 5% (w/v) non‐fat milk for 1 h at RT. After washing in 1× TBST for 15 min three times, the blots were processed for chemiluminescence detection using Advansta WesternBright ECL (for H3 and AtpB proteins) or Amersham ECL Select (for RPL30 protein) Western blot analysis detection reagent. The image was acquired by a ChemiDoc XRS+ imager (Bio‐Rad).

2.7. Ribo‐Seq Data Preprocessing and Quality Analysis

The raw reads were processed by Cutadapt v.1.14 (Martin 2011) with parameters (–m 18, –u 3, –M 40 and –a AAAAAAAAAA) to remove the adapter sequences and to select sequences ranging from 18 to 40 nt. The low‐quality reads were further removed using the FASTX‐toolkit v.0.0.13 (Pearson et al. 1997) with parameters [parameter: –q 20 –p 85]. The rRNA, tRNA, snRNA and snoRNA non‐coding RNA sequences were removed from the remaining reads using Bowtie 2 v.2.4.1 (Langmead and Salzberg, 2012). The sequence IDs of rRNA, tRNA, snRNA and snoRNA are described in the Supporting Information: Methods. The rest of the reads were then aligned to the Chlamydomonas reference genome v.6.1 (Craig et al. 2023) using the split‐aware aligner STAR v.2.7.9a (Dobin and Gingeras 2016), allowing three mismatches. The options used for aligning were: ‐‐outFilterMismatchNmax 3, ‐‐outFilterMultimapNmax 20, ‐‐outSAMtype BAM SortedByCoordinate, ‐‐outSAMmultNmax 1, ‐‐outMultimapperOrder Random, ‐‐alignEndsType EndToEnd, ‐‐outSAMstrandField intronMotif, ‐‐outSAMattributes All, ‐‐quantMode TranscriptomeSAM GeneCounts and ‐‐outReadsUnmapped Fastx. It is important to use the ‘‐‐outSAMattributes All’ option to include SAM attributes (such as alignment scores, read group tags and sequence tags) during alignment, ensuring compatibility with the subsequent Ribo‐TISH analysis. After alignment, the BAM files were sorted and indexed by Samtools v.1.9. Only uniquely mapped reads were used for subsequent analysis.

Raw sequencing data from Chung et al. (accession: ERX558438) were downloaded from the NCBI Sequence Read Archive database and processed using the procedures described above. Processed reads from Gotsmann et al. (Chlamy_v.6‐1_translatome. bam) were downloaded directly from the Phytozome Data Portal. TPM values were calculated by RSEM v.1.3.3 (B. Li and Dewey 2011) with parameter settings: ‘rsem‐calculate‐expression’, ‘‐‐fragment‐length‐mean’ and ‘‐‐fragment‐length‐sd’. The correlation of samples and replicates was analysed and plotted by R v.4.3.3 packages: Hmisc, reshape2, ggplot2, extrafont, smplot2 and psych. The ‘ribotish quality’ of Ribo‐TISH v.0.1.10 (Zhang et al. 2017) was used to analyse length distribution, meta‐gene analysis, P‐site offsets and triplet periodicity of the RPFs. The genomic features (5’ UTR, coding sequence [CDS], introns, 3’ UTR and intergenic regions) were assigned to the reads using RSeQC v.4.0.0 (Wang et al. 2012) with the ‘read_distribution. py’ setting.

To improve the conceptualisation of the represented ORFs, P‐site tracts (corresponding to Frame 1) for the individual transcripts were displayed in Figures 5, 6, 7 and Figure S5.

Figure 5.

PHEROPHORIN‐CHLAMYDOMONAS HOMOLOG2 (PHC2) and PHC35 genes showed differential expression between ZT8 and ZT12. (a) RNA‐seq and P‐sites in ribosome footprints are shown for PHC2 (Cre14.g620600). (b) RNA‐seq and P‐sites in ribosome footprints are shown for PHC35 (Cre05.g238687). Annotated gene model and chromosome coordinates are indicated under each Ribo‐seq profile. Within the gene model: grey box, 5′‐UTR; black box, CDS; grey triangle, 3′‐UTR. Ribo‐seq reads are shown by plotting their first nucleotide of the P‐site. Three reading frames are shown in red (the corrected frame according to the predicted start codon), green and blue. Note that the highly repetitive sequences (Supporting Information: Methods) in part of the exon 3 of PHC2 and part of exon 5 of PHC35 result in very low or a lack of read mapping. (c) Quantification of ribosome‐protected PHC2 and PHC35 mRNAs by RT‐qPCR, with SE shown by error bars. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 6.

uORF predicted by ribosome profiles. (a) RNA‐seq and P‐sites in ribosome footprints are shown for the predicted uORFs of DCA1 (Cre03.g205900). The yellow, pink and orange boxes represent different uORFs. The position of the predicted start codon of uORF is indicated by a black, pink, or orange dashed line, respectively. (b) RNA‐seq and P‐sites in ribosome footprints are shown for the predicted uORF of Suppressor of fil1 (SFI1, Cre04.g215800). The position of the predicted start codon of uORF is indicated by a black dashed line on the Ribo‐seq profile panel. Notice the footprints of both genes display clear 3‐nt periodicity. Annotated gene model and chromosome coordinates are indicated under each Ribo‐seq profile. Within the gene model: grey box, 5′‐UTR; black box, CDS; grey triangle, 3′‐UTR. Ribo‐seq reads are shown by plotting their first nucleotide of the P‐site. Three reading frames are shown in red (the corrected frame according to the predicted start codon), green and blue. Insets show rescaled graphs to visualise footprints of predicted uORFs. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 7.

Coding sequence annotation by ribosome profiles. (a) RNA‐seq and P‐sites in ribosome footprints are shown for RPD3/HDA1 type histone deacetylase (HDA4, Cre03.g162050). (b) RNA‐seq and P‐sites in ribosome footprints are shown for DIACYLGLYCEROL PYROPHOSPHATE PHOSPHATASE 1‐RELATED PROTEIN 1 (DPP1, Cre05.g230900). Notice the footprints of both genes display clear 3‐nt periodicity. Annotated gene model and chromosome coordinates are indicated under each Ribo‐seq profile. Within the gene model: grey box, 5′‐UTR; black box, CDS; grey triangle, 3′‐UTR. Ribo‐seq reads are shown by plotting their first nucleotide of the P‐site. Three reading frames are shown in red (the corrected frame according to the predicted start codon), green and blue. The positions of the annotated and predicted start codons are indicated by grey and black dashed lines, respectively, on each Ribo‐seq profile panel. [Color figure can be viewed at wileyonlinelibrary.com]

2.8. RNA‐Seq Library Construction, Sequencing and Data Analysis

In addition to collecting cells for Ribo‐seq, ~10⁷ cells were harvested (centrifugation at 3220 ×g for 5 min at RT) for RNA‐seq analysis. Total RNA was isolated as previously described (Fang et al. 2006). The quality and quantity of RNA were determined by Agilent 2100 Bioanalyzer using the Agilent RNA 6000 Nano kit (Agilent, Part Number: 5067–1511). Approximately 2 μg of purified RNA was used to construct a cDNA library by the Agilent SureSelect XT HS2 mRNA Library Preparation Kit with Index Primer Pairs 193–288, 96 Reactions (Agilent, Part Number: G9997C) as described in the user manual. The cDNAs were amplified by 10 cycles of PCR, and the quality of cDNA libraries was validated by Agilent TapeStation using the D1000 ScreenTape System (Agilent, Part Number: 5067‐5582). Equal molarity of the barcoded libraries was pooled for paired‐end 150‐nt sequencing in a NovaSeq X plus (10B) platform. Raw reads were trimmed using Trimmomatic v.0.38 (Bolger et al. 2014) to remove adapter and low‐quality reads. Removal of the non‐coding RNA, mapping to the reference genome, calculation of TPM values and genomic feature assignment of the RNA‐seq data were carried out as described for Ribo‐seq analysis.

2.9. Culture Synchronisation and Cell‐Cycle Analysis

Chlamydomonas cells were cultured in HSM liquid medium in a 12 h:12 h light:dark cycle with equal fluences of blue (450 nm) and red (630 nm) light (150 μmol photons m⁻²s^{− 1} each). Mitotic index was determined as described previously (Fang et al. 2006; Huang et al. 2024; Umen and Goodenough 2001). Cells were fixed as described previously to determine the DNA content (Fang et al. 2006). Briefly, ~2 × 10⁶ cells were pelleted in the presence of 0.005% Tween 20 and fixed by resuspending in 10 mL of ethanol/acetic acid (3:1) at RT for 1 h. Fixed cells were then washed once with 10 mL of FACS buffer (0.2 M Tris [pH 7.5], 20 mM EDTA), resuspended in 1 mL of FACS buffer, and stored at 4°C. Before image cytometry, cells were incubated in FACS buffer at 37°C with 100 μg/mL RNase A for 3 h, washed once with 1 mL of PBS, and resuspended in 100 μL of PBS. DNA was stained with 1 μM Sytox Green (Invitrogen) and imaged by the NucleoCounter NC‐3000 system (Chemometec). The DNA content data was analysed using Modtif LT, and the plot was generated by FlowJo (BD Biosciences).

2.10. RNA Isolation and Quantitative RT‐PCR

Total RNA was isolated as previously described (Fang et al. 2014). Approximately 1–5 μg of purified RNA was used for cDNA synthesis. cDNAs were reverse‐transcribed in the presence of a mixture of oligo dT and random primers (9:1) at 55°C for 10 min using SuperScript IV First‐Strand Synthesis System (Invitrogen, Cat.18091050) following the manufacturer's instructions. Each 10 μL of qPCR reaction contained 2.5 μL of 1/20 diluted cDNA, 0.2 μM of primers and 5 μL of 2× KAPA SYBR FAST master mix (KAPA Biosystems). The PCR programme was performed on a Bio‐Rad CFX96 (Bio‐Rad) as follows: 95°C for 2 min, 40 cycles of 95°C for 5 s and 63°C for 25 s. PCR was performed in triplicate. G PROTEIN BETA SUBUNIT‐LIKE PROTEIN (GBLP, Cre06.g278222) was used as an internal control. For quantitative RT‐PCR of polysome fractions, RIBOSOMAL PROTEIN L30 (RPL30, Cre10.g420750) was used as an internal control. The primers used for qPCR are listed in Table S5.

2.11. Identification of Differentially Translated Genes

Read counts per gene were computed by mapping the Ribo‐seq and RNA‐seq reads to the CDS of annotated protein‐coding genes using STAR with the option ‘‐‐quantMode TranscriptomeSAM GeneCounts’. Translation efficiency is defined as the ratio of normalised ribosome footprint reads to RNA‐seq reads for individual transcripts. These read counts were used to quantify translation efficiency using the ‘TE. py’ script of RiboDiff (Flanagan et al. 2022; Zhong et al. 2017). RiboDiff loads the input files and normalises the raw count of each replicate sample according to its sequencing library size. To minimise false‐positive gene identification, transcripts with TPM values below 1 in all samples were removed from further analysis. Multiple test correction was conducted using the Benjamini‐Hochberg method (Benjamini and Hochberg 1995) to control the false discovery rate and an adjusted p value threshold of < 0.05 was applied.

2.12. Identification of Potential UORFs and Differentially Translated UORFs

After adaptor and quality trimming, the pooled Ribo‐seq reads from two buffer B⁺ samples and six synchronised samples were mapped to the Chlamydomonas genome (v.6.1) by STAR aligner to generate a BAM file. This BAM file was used as input for ‘ribotish predict’ analysis (with the ‘‐‐longest’ option, Ribo‐TISH v0.1.10), and p < 0.05 cutoff was applied. The reads with lengths of 26, 27 and 28 nucleotides, and respective P‐site offsets 11, 11 and 12 nucleotides (determined by ‘ribotish quality’), were used for prediction. The selected read lengths (26, 27 and 28 nucleotides) were chosen because they have relatively high triplet periodicity (Table S4). The output file contains the unannotated ORFs and annotated main open reading frames (mORFs).

The unannotated ORFs were extracted for uORF prediction. To identify uORFs, additional filters including peptide size (larger than or equal to 10 amino acids, and less than or equal to 100 amino acids), genome position (within the annotated 5′ UTR) based on the suggested criteria (Egorov and Atkinson 2023; Guo et al. 2023) were applied. We excluded the ORFs that are part of long transcripts but were identified as uORFs of the shorter alternatively spliced transcripts of the same gene. To improve identification confidence, pooled Ribo‐seq reads (from two buffer B⁺ samples and six synchronised samples) were mapped to the custom‐made uORF structural annotation gtf file by STAR aligner. To increase identification stringency, only the uORFs with over 10 read counts were selected as putative uORFs.

Three criteria were considered when selecting near‐cognate codons for uORF identification. First, the evolutionarily conserved uORF that regulates S‐adenosylmethionine decarboxylase (DCA1) (Hanfrey et al. 2005) had to be identified as it was the only documented Chlamydomonas uORF before our analysis. Second, since AUG is the dominant translation initiator in animals and plants (ranging from 41% in humans to 80% in tomatoes (Y. R. Li and Liu 2020), the cutoff for AUG percentage was set to equal or more than 50%. Third, the most prevalent translation initiation codons—AUG, CUG and ACG—were selected for our analysis based on their enrichment identified through translation initiation sequencing (TI‐seq) (Y. R. Li and Liu 2020).

To identify the differentially translated uORFs, the translational efficiency (TE) of the uORFs was calculated as described in the ‘Identification of differentially translated genes’ section.

3. Results and Discussion

3.1. Optimisation of rRNA Depletion, Polysome Extraction and Isolation of Ribosome‐Protected Footprints

rRNA removal is critical for Ribo‐sequencing analysis because rRNA contamination reduces the yield of informative sequence data (Brar and Weissman 2015; Gotsmann et al. 2024; Ingolia et al. 2012; Zinshteyn et al. 2020). Unfortunately, the Ribo‐Zero Plant Leaf rRNA removal Kit (Illumina; MRZPL1224) previously used to remove Chlamydomonas rRNA (Cavaiuolo et al. 2017) was discontinued. Therefore, we tested two other commercially available rRNA removal kits: RiboMinus Plant kit (Ribo‐Minus) and Ribo‐off rRNA Depletion kit (Plant) (Ribo‐Off). The polysome extraction buffer previously used for Chlamydomonas (Chung et al. 2015) with slight modification (Buffer A, Table 1) was used to prepare polysome lysates. After preparing the polysome lysates containing the immobilised ribosomes, polysomes were subjected to RNase I digestion to obtain monosomes (see Section 2 for details). RNAs ranging from 26 to 34 nucleotides were purified before rRNA depletion by Ribo‐Minus or Ribo‐Off. Depleting rRNA after RNA size selection has been shown to improve read coverage (Wu et al. 2024). After rRNA depletion, the RPFs were purified before library construction. Small‐scale Ribo‐seq was used to assess the rRNA mapping rate. The efficiency of rRNA removal by the Ribo‐Minus and Ribo‐Off was evaluated using the rRNA mapping rate, analysed with Bowtie 2 (Langmead and Salzberg 2012). Our data showed that approximately 39.4% of rRNA remained in the RNA reads after Ribo‐Off treatment, while about 93.5% remained after Ribo‐Minus treatment (Figure 1a and Table S1). Compared to the methods used by Chung et al. (2015), with 65.3% rRNA remaining, and Gotsmann et al. (2024), with ~64% rRNA remaining, the Ribo‐off kit was more efficient at rRNA removal (Table S1). Moreover, after manually removing non‐coding RNAs, including tRNA, snoRNA, snRNA and residual rRNA sequences, ~60% of RNA reads remained for follow‐up analysis (see Section 2 for details). In contrast, only about 6% of RNA reads remained in the Ribo‐Minus‐treated sample, and approximately 34% remained in the DSN‐treated sample (Figure 1a and Table S1). Hence, the Ribo‐off rRNA Depletion kit was chosen for the rest of the experiments.

Figure 1.

Comparison of different rRNA removal methods and polysome extraction buffers. (a) Pie chart showing the distribution of RNA reads in the indicated categories after rRNA removal by the indicated method. Note that the data from the duplex‐specific nuclease experiment were obtained from Chung's study. (b) Polysome profiles of polysome extracts prepared with the three extraction buffers before RNase digestion. Notice that buffer B had an increased amount of heavy polysomes. The profiles highlighted by the red rectangle are magnified and shown in the lower panel. (c) Immunoblotting (IB) of ribosome fraction number as indicated in (b) by RPL30 antibodies.

Buffer conditions such as ionic strength, buffering capacity and magnesium concentration have been shown to affect polysome quality (Davies and Abe 1995), RNase digestion efficiency and 3‐nt periodicity (Douka et al. 2021; Hsu et al. 2016; Ingolia et al. 2012). To optimise polysome isolation, three polysome extraction buffers with varying magnesium concentration, ionic strength and buffering capacity were tested (Table 1). We found that Buffer B improved the yield of heavy polysomes (Figure 1b), suggesting that the higher MgCl₂ concentration stabilised ribosome‐RNA interaction during the purification process. In contrast, the heavy polysome yield isolated by Buffer A and Buffer C was reduced, while light polysome levels increased, presumably due to unstable ribosome‐RNA interaction during purification. This experiment was repeated multiple times with similar results. The increased amount of heavy polysomes under the Buffer B conditions was confirmed by the accumulation of the RIBOSOMAL PROTEIN L30 (RPL30) protein (Figure 1c and Figure S1a,b). Therefore, Buffer B was more effective than the other buffers in stabilising heavy polysomes during purification. To evaluate the quality of RPFs extracted with different buffers, Ribo‐seq was performed on samples from two independent unsynchronised cell cultures. Each sample was prepared using one of the three different buffers. Our data showed a fairly good correlation (Pearson correlation $\ge$ 94%) of RPF reads in the sample prepared with different buffers (Figure S2a). Interestingly, the correlation between biological replicates decreased, with Pearson correlation coefficients ranging from 0.87 to 0.92. This is likely due to the heterogeneous diurnal and/or cell‐cycle states in the two different cultures. The RPF reads were mapped to the Chlamydomonas genome v.6.1 (Craig et al. 2023) by STAR aligner (Dobin et al. 2013). The mapping statistics are summarised in Table S2. RPF meta‐gene analysis showed that a 27‐nt footprint size was dominant across all buffer conditions (Figure S2b). This finding is consistent with previous reports (Chung et al. 2017; Chung et al. 2015) but differs from another study, which reported a dominant 30‐nt footprint size (Gotsmann et al. 2024). The discrepancy in footprint size may be due to differences in buffer composition. Ribo‐TISH (Zhang et al. 2017) was then applied to assess the 3‐nt periodicity in the P‐site signals on the annotated nuclear protein‐coding genes. Ribo‐TISH calculated the distribution of RPF counts in the dominant frame (f_d) of the annotated protein‐coding genes; high 3‐nt periodicity is expected to have a higher f_d in the actively translated reading frame than the other two frames. Our analysis showed that most of the 27‐nt footprints mapped to the second nucleotide position of codons (Frame 2) and had f_d greater than 0.6, regardless of the buffer used (Figure 2a). This suggests that the RPFs generated by our protocol exhibited relatively good 3‐nt periodicity (Zhang et al. 2017). For 27‐nt footprints, the ribosome protects 11 nt 5′ of the P‐site codon (with a P‐site offset of 11 bp) regardless of the buffer used. This indicates that the codon translated at the P‐site is located between the 12th and 14th nucleotide within a 27‐nt footprint (Figure 2b), which is consistent with a previous report (Chung et al. 2015). When translation reaches termination and the A‐site encounters a stop codon, the last in‐frame footprints cover the 14th nucleotide upstream of the stop codon, indicating the A‐site is located between the 15th and 17th nucleotide within a 27‐nt footprint (Figure 2b).

Figure 2.

Meta‐gene analysis of Ribo‐seq reads. (a) Count distribution of the 5′ end of 27 nt RPFs uniquely mapped near the annotated translation start or stop site. The data corresponding to the first, second and third reading frames are coloured in orange, light blue and megenta, respectively. fd, the fraction of the RPF counts in the dominant frame. The P‐site offset is 11 nts in all data sets. (b) The inferred 27‐nt ribosome footprint positions related to the ribosomes near the start and stop codon are shown. Three reading frames are shown in orange (the main open reading frame, Frame 1), light blue (Frame 2) and megenta (Frame 3). Most footprints are mapped within the CDS and show enrichment for the Frame 2. Footprints at the translation initiation and termination revealed that the ribosomal P‐site is located between the 12th and 14th nucleotides, whereas the ribosomal A‐site is located between the 15th and 17th nucleotides. A‐site, the entry point for the aminoacyl‐tRNA. P‐site, the position where peptide bond forms. E‐site, the exit site of the uncharged tRNA. [Color figure can be viewed at wileyonlinelibrary.com]

Although RPFs isolated with buffer B had a lower f _d value compared to those isolated with buffer A, buffer B was preferred for its ability to maintain the integrity of heavy polysomes (Figure 1b). Since increasing RNase I concentration has been shown to improve 3‐nt periodicity (Douka et al. 2021), the RNase I concentration in Buffer B was adjusted to 3.75 u/μg RNA (Buffer B⁺, Table 1). Encouragingly, increasing the RNase I concentration in Buffer B (Buffer B⁺) resulted in a higher f _d value of the 27‐nt footprints in both biological replicates (0.87 and 0.82, Figure 2a). Therefore, buffer B⁺ was used for the remainder of the experiments.

3.2. Optimised Ribosome Profiling Improved Data Quality Compared to Published Data Sets

To directly assess the quality of different Ribo‐seq data sets, we used Ribo‐TISH to compare our data to the previously published Ribo‐seq data (Chung et al. 2015; Gotsmann et al. 2024). Regardless of the extraction buffers used, the coding sequence (CDS) mapping rate (between 93.7% and 96.9%) of our data was higher than those (92.9% and 88.4%, respectively) of previous data sets (Table S3 and Figure 3a). Interestingly, we noticed a biased mapping to 5′ or 3′ UTR using the previously published data sets (Figure 3a,b). In contrast, our data produced RPFs that were distributed across the CDS (Figure 3b) and exhibited a pattern similar to high‐quality RPFs, with sharp initiation at the start codon and sharp termination at the stop codon (Brar and Weissman, 2015). The 3‐nt periodicity of the published data sets was assessed by Ribo‐TISH. A fairly high periodicity (f _d = 0.95) was found in the Ribo‐seq data generated by Chung et al. 2015, while a moderate periodicity (f _d = 0.76) was observed in the data of Gotsmann et al. 2024 (Figure 3c). A direct comparison of RPF sizes, f _d values and 5′ P‐site offsets from different Ribo‐seq data sets is listed in Table S4.

Figure 3.

Comparison of different Ribo‐seq data sets. (a) The distribution of Ribo‐seq reads across different genome features annotated in Chlamydomonas genome. For RPFs prepared with B⁺ buffer, two biological replicates were pooled together. For synchronised cultures, ZT8 (Zeitgeber time = 8 h after the onset of illumination) and ZT12 Ribo‐seq data of three replicates were pooled together. (b) The dominant‐sized RPF count profile throughout the protein‐coding regions across three reading frames. The annotated translation start or stop site was labelled. (c) Distribution of RPF 5′ end counts near the annotated translation start or stop site. The data corresponding to the first, second and third reading frames are coloured organe, light blue and megenta, respectively. f _d, the fraction of the RPF counts in the dominant frame. The P‐site offset is 11 nts from RPF data of the current study and Chung et al. (2015). The P‐site offset is 12 nts from the RPF data of Gotsmann et al. (2024). (d) Number of protein‐coding genes identified by Ribo‐TISH with different sequencing depths. [Color figure can be viewed at wileyonlinelibrary.com]

Because sequencing quality and depths influence the number of the identified ORFs in Arabidopsis (Hsu et al. 2016), we used random sample reads to evaluate the ability of different Ribo‐seq data sets to identify ORFs. A fixed number of RPF reads were randomly selected using the Seqtk v.1.3 package. The number of RPF reads was plotted against the number of identified ORFs. As expected, the number of identified ORFs was positively correlated with sequencing depths across different data sets (Figure 3d). The ability to predict ORFs was similar across the different Ribo‐seq data sets. Notably, the number of identified ORFs appeared to plateau at approximately 10 million mapped reads and increased only slightly at 15 million mapped reads. This suggests that 10–15 million reads is a good starting point for designing Chlamydomonas Ribo‐seq experiments.

3.3. Chlamydomonas Cell Cycle May be Regulated at the Translational Level

We were interested in applying this optimised Ribo‐seq protocol to study translational regulation between G1 (ZT8, Zeitgeber time = 8 h after the onset of illumination) and S/M (ZT12) stage of the synchronised cell cultures. Chlamydomonas cell culture was synchronised by alternating light and dark cycles (see Section 2). To prevent the effect of light on overall translation rate (Gotsmann et al. 2024), ZT12 samples were collected under illumination conditions. Synchrony of wt culture was monitored by mitotic index analysis, DNA content and expression of cell cycle marker genes, CYCLIN‐DEPENDENT KINASE B1 (CDKB1), PROLIFERATING CELL NUCLEAR ANTIGEN (PCNA) and CYCLIN A1 (CYCA1) (Figure 4a). The synchrony of two more independent cultures was validated (Figure S3a) and included in the statistical analysis of this translational study. Pearson correlation analysis showed that replicates of Ribo‐seq data sets were highly correlated (r = 0.89–0.96, Figure S3b), indicating our protocol was robust in generating reproducible data. A strong positive correlation (Pearson correlation, r = 0.97–1, Figure S3c) was also found across the replicates of RNA‐seq data. As expected, the CDS mapping rate of Ribo‐seq reads (~95%) was higher than that of RNA‐seq reads (~80%, Figure 4b), confirming the ability of Ribo‐seq to decode the translated mRNAs.

Figure 4.

Quality evaluation of Ribo‐seq and RNA‐seq data from synchronised wild‐type Chlamydomonas cultures. (a) Synchronisation was confirmed by analyses of mitotic index, DNA content and the expression of the S/M phase markers CDKB1, PCNA and CYCA1. The cell cycle phases are indicated by the bars above the graph. The synchronised culture was maintained in 12‐h‐light/12‐h‐dark cycles. Cells entered the S/M phase at approximately ZT12 (Zeitgeber time = 12 h after the onset of illumination). Cultures were sampled at ZT8 and ZT12. For RT‐qPCR analysis, the standard error is represented by error bars. (b) The distribution of the Ribo‐seq and RNA‐seq reads across different genomic features annotated in Chlamydomonas genome v.6.1. (c) Volcano plot showing differentially translated genes. The red dashed line represents a p value of 0.05. Only differentially translated genes with p < 0.05 are displayed. (d) Immunoblot detection of Histone 3 (H3) proteins in the total protein lysate of the ZT8 and ZT12 samples with anti‐H3 antibody. Protein loading for each lane was normalised either by total protein amount (10 μg) or by equal cell number (5 × 10⁴ cells). ATP synthase subunit beta (AtpB) protein was used as a loading control. Ponceau S staining showing equal loading of total protein amount. Results from three independent experiments (#1, #2 and #3) are shown. [Color figure can be viewed at wileyonlinelibrary.com]

We used RiboDiff (Zhong et al. 2017) to compute the TE of the expressed genes. For further analysis, low‐abundance genes (TPM < 1 across both Ribo‐seq and RNA‐seq data sets) were removed to reduce the possibility of identifying false‐positive candidates. We were interested in transcripts whose translation was differentially regulated between ZT8 (G1) and ZT12 (S/M) and identified 1688 differentially translated genes using a cutoff of p < 0.05. Gene Ontology analysis revealed that biological processes related to transport and localisation, as well as molecular functions such as structural molecule activity, structural constituent of the cell wall and protein binding are associated with cells collected at ZT8 (Figure S4). In contrast, metabolic processes and molecular functions related to heterocyclic and organic acid compound binding are associated with cells collected at ZT12 (Figure S4).

Among the differentially translated genes, 752 transcripts (725 nucleus‐encoded transcripts, 20 chloroplast‐encode transcripts and 7 mitochondria‐encoded transcripts) had increased TE at ZT8 (hereafter referred as the ZT8 TE transcripts) (Figure 4c and Data S1). We also identified 936 nuclear‐encoded transcripts whose TE was increased at ZT12 (Data S1). Among them, 20 ZT8‐enriched chloroplast‐encoded transcripts correspond to actively translated genes important for photosynthesis (Gotsmann et al. 2024). These include photosystem I subunits (psaA and psaC), photosystem II subunits (psbD, psbE, psbF, psbJ, psbK, psbL, psbH), ATP synthase subunits (atpA, atpE, atpF, atpH), cytochrome b6/f subunits (petB and petD), chloroplastic ribosomal protein subunits (rpl23, rpl36, rplS12) and the chlorophyll synthesis gene chlN. Hence, our result suggests translational control of these photosynthetic genes is likely diurnally regulated. Interestingly, translation of mitochondrial NADH: Ubiquinone oxidoreductases (Complex I) nad1, nad2, nad4, nad5 and nad6, Complex III component cytochrome b (cob) and complex IV component cytochrome c oxidase (cox1) were also increased at ZT8, suggesting translational control of the mitochondrial function may be important during the light and/or G1 stage. Because mitochondrial Complex I, III and IV are involved in pumping protons from matrix to the intermembrane space to drive ATP production through ATP synthase (Massoz et al. 2017), cells at ZT8 may have higher ATP demand than cells at ZT12. It is also possible that ZT8‐associated mitochondrial complex I‐dependent NAD⁺ homoeostasis, which is crucial for nucleotide metabolism and DNA replication (Munk et al. 2023), needs to be activated before entering the S/M phase.

Because cells at ZT12 were at the early S/M phase when transcripts of many core cell cycle genes, including CDKB1, PCNA and CYCA1 (Figure 4a) were upregulated (Bisova et al. 2005; Zones et al. 2015), we were interested in determining whether any of the core cell cycle genes were regulated at the translation level. After surveying our data (Data S1), we observed translation efficiency of 21 of the core cell cycle genes was enhanced significantly at the ZT12 (Table 2). They consist of six MINICHROMOSOME MAINTENANCE PROTEINS encoded by MCM2, MCM3, MCM4, MCM5, MCM6 and MCM7; four DNA polymerases (POLA1, POLA2, POLD1 and POLE1); two anaphase promoting complex subunits (APC1 and APC2); DNA replication factor C complex subunit 1 (RFC1); DNA replication initiation factor 1 (CTD1); germinal‐centre associated nuclear protein (FEE1); ribonucleoside‐diphosphate reductase R1 (RIR1); cohesin complex subunit 3 (SCC3), structural maintenance of chromosomes protein 5B (SMC5B); replication factor A1 (RFA1); retinoblastoma protein (MAT3); and cyclin‐dependent kinase G1 (CDKG1). Increased translation of MCM genes during G1/S stage has been reported in mammalian cells (Zeng et al. 2023), indicating that translation regulation of the cell cycle genes may be evolutionarily conserved. Interestingly, despite the relatively abundant mRNA at early S/M (Zones et al. 2015), the translation efficiency of the G1/S regulator CYCLIN D1 (CYCD1) was slightly but significantly decreased at ZT12. This suggests a complex regulation for CYCD1. Together, our data suggest that the Chlamydomonas core cell cycle genes were not only regulated at the transcriptional level but also potentially through translational mechanisms.

Table 2.

The core cell cycle genes that are differentially translated in ZT8 (G1) and ZT12 (S/M). Histone annotation is as described previously (Rommelfanger et al. 2021).

Gene ID	Annotation	Categorised cell cycle genes (Zones et al. 2015)	log2FC (ZT12/ZT8)
Cell cycle genes
Cre07.g338000	MCM2	MCM	1.2
Cre06.g295700	MCM3	MCM	1.4
Cre07.g316850	MCM4	MCM	1.3
Cre01.g023150	MCM5	MCM	0.6
Cre03.g178650	MCM6	MCM	0.8
Cre10.g455850	MCM7	MCM	1.2
Cre04.g214350	POLA1	DNA polymerase	1.1
Cre01.g017450	POLA2	DNA polymerase	0.8
Cre01.g015250	POLD1	DNA polymerase	0.7
Cre03.g179961	POLE1	DNA polymerase	1.4
Cre12.g521200	RFC1	Replication factor C	1.6
Cre03.g163300	CTD1	Other replication genes	0.9
Cre02.g143600	FEE1	Other replication genes	0.6
Cre12.g492950	RIR1	Other replication genes	0.8
Cre17.g736400	SCC3	Cohesin	0.6
Cre10.g440200	SMC5B	SMC5/6	1.5
Cre13.g579100	APC1	APC/C	0.9
Cre10.g460532	APC2	APC/C	0.6
Cre17.g718850	RFA1	Replication factor A	0.6
Cre06.g255450	MAT3	RB pathway	0.7
Cre06.g271100	CDKG1	RB pathway	0.7
Cre11.g467772	CYCD1	d‐type cyclin	‐2.2
Histone genes
Cre01.g062172	HBV1 (H2B variant)		2.6
Cre06.g264750	H2A (HTA14)		3.8
Cre06.g268350	H3 (HTR9)		4.1
Cre06.g268400	H4 (HFO9)		3.6
Cre06.g271250	H3 (HTR8)		6.6
Cre06.g275750	H3 (HTR12)		4.6
Cre06.g276650	H4		6.5
Cre13.g590800	H2A (HTA37)		0.8

Among the differentially translated nuclear genes, many histone genes (Rommelfanger et al. 2021) were found to be preferentially translated at ZT12 (Table 2). To validate the Ribo‐seq data, protein lysates of ZT8 and ZT12 cells were analysed using two immunoblots: one loaded with equal amounts of protein per lane, and the other with lysates normalised by cell number. Immunoblotting confirmed increased levels of H3 proteins in ZT12 samples compared to ZT8 samples, regardless of the protein normalisation method used (Figure 4d).

Similar to the animal histones (Hentschel and Birnstiel 1981; Marzluff 1992), the 3′ untranslated regions of Chlamydomonas histone genes contain no poly A signal, but a stem‐loop sequence (Fabry et al. 1995) that is essential for histone translation (Allard et al. 2005; Gallie 1996; Gorgoni et al. 2005; Sànchez and Marzluff 2002). Because histones are essential to stabilise the newly replicated DNA at S phase (Armstrong and Spencer 2021; Chari et al. 2019; Duronio and Marzluff 2017; Soto et al. 2004), it is likely that Chlamydomonas histones were preferentially translated at early S/M phase.

Consistent with our observation, translational control has been reported to play a crucial role in regulating the cell cycle. For example, translation of the density‐regulated re‐initiation and release factor has been shown to promote mitotic protein translation and cell division in animal cells (Clemm von Hohenberg et al. 2022). Active translation activity in mitosis is supported by the enhanced translation of terminal oligopyrimidine tract‐containing transcripts, which include many ribosomal proteins and translational factors (Park et al. 2016). Moreover, the cytoplasmic polyadenylation element binding protein‐dependent phase‐specific regulation of poly(A) tail length modulates translation to facilitate cell‐cycle progression (Novoa et al. 2010). Although how translation regulation is coupled to the cell cycle remains unclear, studies on translational regulation are crucial for comprehensively understanding cell‐cycle control mechanisms.

In addition to cell cycle genes, we observed that translation of many PHEROPHORIN‐CHLAMYDOMONAS HOMOLOG (PHC) genes was downregulated at ZT12 (Data S1). PHC genes encode hydroxyproline‐rich rod‐like domain glycoproteins that are abundant in the extracellular compartment (Hallmann 2006). The decrease in ZT12‐associated translation of PHC genes including PHC1, PHC2, PHC35 and PHC38 (Figure 5a,b and Figure S5a,b) suggests that cell‐wall composition may be dynamically regulated during the diurnal cycle. This result was validated by RT‐qPCR analysis of the ribosome‐protected PHC2 and PHC35 mRNAs (Figure 5c). Note that the highly repetitive sequences (see Supporting Information: Methods) in part of exon 3 of PHC2 and part of exon 5 of PHC35 result in very low or a lack of read mapping in both our data set (Figure 5a,b) and Chlamydomonas genome v.6.1 data set (Figure S6).

3.4. Ribosome Profiling Enables Identification of Potential uORFs and Study of Differentially Regulated uORFs

Even though bioinformatic analysis suggested that uORFs are prevalent in the Chlamydomonas genome (Cross 2016), currently, no experimental evidence supports this supposition. Because most of the P‐sites are mapped to the main reading frame of CDS in translated ORFs, the 3‐nt periodicity has been used to identify the uORFs (Bazzini et al. 2014; Hsu et al. 2016; Ji et al. 2015; Liu et al. 2023; Wu et al. 2024; Wu et al. 2019). We took advantage of the high‐quality Ribo‐seq data sets and used Ribo‐TISH (Zhang et al. 2017) to identify the potential Chlamydomonas uORFs (Data S2). The most prevalent translation initiators, including canonical (AUG) and non‐canonical (CUG and ACG) codons of plants and humans (Y. R. Li and Liu 2020) were used to predict uORFs. Peptide sizes ranging from 10 to 100 amino acids were applied. In total, we identified 926 translated uORFs (Data S2). Among them, the evolutionarily conserved uORF that regulates the key polyamine biosynthetic enzyme S‐adenosylmethionine decarboxylase (DCA1) (Hanfrey et al. 2005) was identified (Figure 6a). An additional uORF that starts with the CUG codon was identified upstream of DCA1. This finding suggests that our analysis pipeline is effective for identifying Chlamydomonas uORFs.

Approximately 30%–70% of plant genes contain potential AUG‐initiated uORFs (Wu et al. 2024; Zhang et al. 2021). Only ~5% (900/16 801) of Chlamydomonas genes were found to have one uORF in this study, suggesting the number of the identified Chlamydomonas uORFs may be underestimated. Some translated uORFs may have gone unidentified due to their low abundance, short length, or overlap with mORFs (Wang et al. 2024) or other uORFs. For example, the tiny translated uORF (nine amino acids) of DCA1 (Hanfrey et al. 2005) could have been identified if the length criteria were relaxed (Figure 6a). Also, other non‐AUG initiation sites may be used by Chlamydomonas that are not included in our studies. Intriguingly, many predicted uORFs were associated with genes encoding proteins presumably associated with primary cilium homoeostasis and basal body proteome (Data S2). A similar observation has been reported in a murine model (Hoang et al. 2019). This suggests that uORF may be a regulatory mechanism to control the translation of flagella/cilium proteins. In addition, AUG‐initiated uORFs (59 out of 458) appeared enriched in genes encoding protein kinases and phosphatases. This observation is consistent with the report in Arabidopsis (Kim et al. 2007).

To identify biologically relevant uORFs, we focused on those with expression differentially regulated between ZT8 and ZT12, using a cutoff of adjusted p < 0.05. Ninety‐two differentially regulated uORFs were identified: 50 with AUG, 24 with CUG and 18 with ACG as translation initiators (Data S3). Many of them are associated with genes that are not functionally annotated. Intriguingly, some are associated with genes encoding proteins in flagella proteome and centrosome functions. They are FLAGELLUM ASSOCIATED PROTEINS 131 (FAP131), FAP192, FAP246 (Pazour et al. 2005), CiliaCut protein SSA16 (Merchant et al. 2007) and Sfi1 centrin‐interacting protein (Keller et al. 2005) (Figure 6b) that is important for spindle pole body duplication in budding yeast (Kilmartin 2003) and essential for centriolar architecture and ciliogenesis in humans (Laporte et al. 2022). Chlamydomonas flagella are cell organelles with microtubule‐based structures arranged in a nine‐fold symmetry, resembling the structure of animal cilia (Dutcher 2020; Dutcher and O'Toole 2016). Flagella must resorb before cell division, to free their basal bodies for spindle organisation during mitosis and cytokinesis (Cross and Umen 2015; Parker et al. 2010). The link between cell cycle‐dependent uORF translation and flagella/basal body‐associated genes suggests uORFs may play a role in the assembly and disassembly of flagella during the cell cycle. Whether these potential uORFs regulate functions of flagella and central bodies remains to be determined.

3.5. Ribosome Profiling Data Can be Used to Improve Annotated Transcripts

Because the 3‐nt periodicity is useful in fine‐tuning annotated ORFs (Bazzini et al. 2014; Calviello et al. 2016; Hsu et al. 2016; Wu et al. 2019), we used the ‘ribotish predict’ output from Ribo‐TISH analysis to evaluate the annotated v.6.1 Chlamydomonas mORFs. While substantial improvements have been made in the start codon assignments of Chlamydomonas v.6.1 annotations (Craig et al. 2023; Gotsmann et al. 2024), our manual inspection of the mORF revealed genes that use AUG codons upstream of the annotated start sites. For example, the transcript encoding the RPD3/HDA1 type histone deacetylase (HDA4, Cre03.g162050) initiates translation at AUG upstream of the annotated start codon (Figure 7a). Another example is DIACYLGLYCEROL PYROPHOSPHATE PHOSPHATASE 1‐RELATED PROTEIN 1 (DPP1, Cre05.g230900), which may also start translation upstream of the annotated AUG (Figure 7b). Whether this long in‐frame DPP1 is an alternative isoform warrants further study. These examples illustrate how ribosome profiling can assist in improving genome annotation

In summary, we established an optimised Ribo‐seq protocol and demonstrated its ability to generate high‐quality RPF reads. We also present a detailed Ribo‐TISH analysis pipeline and identify genes with differential translation between synchronised cells collected at the G1 and S/M stages. In addition, the high 3‐nt periodicity of RPFs enabled the prediction of uORFs, some of which may be diurnally regulated. Overall, this ribosome‐profiling data provides an invaluable resource for uncovering novel ORFs, such uORFs, and may be a valuable tool to improve Chlamydomonas genome annotation.

Supporting information

20250501 Data S1.

20250501 Data S2.

20250501 Data S3.s

20250501 Supplemental Tables and methods.

20250526 Sup Figures.

Data Availability Statement

All gene IDs are based on the Chlamydomonas genome v.6.1 annotation (https://phytozome-next.jgi.doe.gov/). Ribo‐seq raw sequenced reads are available in the National Center for Biotechnology Information (NCBI) Sequence Read Archive repository: SAMN43934484, SAMN43934485, SAMN43934486, SAMN43934487, SAMN43934488, SAMN43934489, SAMN43934490, SAMN43934491, SAMN43934492, SAMN43934493, SAMN43934494, SAMN43934495, SAMN43934496, SAMN43934497. RNA‐seq raw sequenced reads are available in the NCBI Sequence Read Archive repository: SAMN43934498, SAMN43934499, SAMN43934500, SAMN43934501, SAMN43934502, SAMN43934503.