Simultaneous measurement of nascent transcriptome and translatome using 4-thiouridine metabolic RNA labeling and translating ribosome affinity purification

Abstract Regulation of gene expression in response to various biological processes, including extracellular stimulation and environmental adaptation requires nascent RNA synthesis and translation. Analysis of the coordinated regulation of dynamic RNA synthesis and translation is required to determine functional protein production. However, reliable methods for the simultaneous measurement of nascent RNA synthesis and translation at the gene level are limited. Here, we developed a novel method for the simultaneous assessment of nascent RNA synthesis and translation by combining 4-thiouridine (4sU) metabolic RNA labeling and translating ribosome affinity purification (TRAP) using a monoclonal antibody against evolutionarily conserved ribosomal P-stalk proteins. The P-stalk-mediated TRAP (P-TRAP) technique recovered endogenous translating ribosomes, allowing easy translatome analysis of various eukaryotes. We validated this method in mammalian cells by demonstrating that acute unfolded protein response (UPR) in the endoplasmic reticulum (ER) induces dynamic reprogramming of nascent RNA synthesis and translation. Our nascent P-TRAP (nP-TRAP) method may serve as a simple and powerful tool for analyzing the coordinated regulation of transcription and translation of individual genes in various eukaryotes.


INTRODUCTION
The regulation of gene expression plays a pivotal role in di v erse biological and physiological processes such as cell dif ferentia tion, de v elopment, environmental responses, and immune responses. One determinant of gene expression is the amount of mature RN A, w hich reflects the balance between RNA synthesis and degradation (1)(2)(3)(4)(5). To understand the dynamics of RNA synthesis and degradation comprehensi v el y, metabolic RN A labeling techniques using nucleotide analogs have been de v eloped (6)(7)(8)(9)(10)(11)(12)(13)(14). For example, thiol (SH)-linked alkylation for the metabolic sequencing of RNA (SLAMseq) enables to uncover 4-thiouridine (4sU)-labeled transcripts in cDNA libraries by bioinformatic detection of specific T-to-C (T > C) conversions at the sites of 4sU incorporation ( 9 ). Combining SLAMseq with QuantSeq, a deep sequencing close to the 3 end of pol yadenylated RN As ( 15 ), allows ra pid and quantitati v e access to 4sU-labeled transcripts expression profiles ( 9 ). These techniques have addressed the dynamic aspects of gene expression regulation in various biological processes such as RNA quality control, oncogenesis, and embryogenesis (16)(17)(18).
The final output of gene expression, which indicates the expression level of proteins, is significantly influenced by RNA translation. Se v eral studies hav e reported that the primary set of RNA transcripts obtained by standard RNA-seq experiments may have a low quantitative correlation with the proteome ( 4 , 19-21 ). To overcome this issue, analyzing the translatome, which refers to mature RNAs bound to ribosomes for protein synthesis, helps in estimating individual gene expression levels (22)(23)(24). In the standard polysome profiling technique, ribosome-bound RNAs are fractionated using sucrose density gradients and analyzed using RNA-seq ( 25 , 26 ). Translating ribosome affinity purification (TRAP) is also used to analyze translatomes which r equir es the expr ession of affinity-tagged ribosomal proteins of the large (60S) ribosomal subunit by genetic modification of cells and organisms. The cell type-specific pr omoter contr ols the expression of affinity-tagged ribosomal proteins, allowing the capture of ribosome-bound RNAs from specific cells / tissues in organisms ( 22 , 27-29 ). Following the immunoprecipitation of affinity-tagged ribosomes, ribosome-bound RN A was anal yzed using RN Aseq. In a recently established ribosome profiling (Ribo-seq) technique, RNase digestion of ribosome-bound RNAs resulted in ribosome-protected RNA fragments (RPFs). The RPFs are converted to a library for deep sequencing to determine the precise position and density of ribosomes at nucleotide resolution (30)(31)(32).
Although the regulation of RNA synthesis, degradation, and translation are distinct steps, these control interplay in gene expression ( 33 , 34 ). The simultaneous measurement of these different controls takes advantage of a comprehensi v e understanding of complicated gene expression regulation. Recently, nascent Ribo-seq (nRibo-seq) was de v eloped to sim ultaneousl y measure nascent RNA synthesis and translation by combining Ribo-seq with 4sU metabolic RNA labeling ( 35 ). nRibo-seq estimates whether a short RPF is deri v ed from a 4sU-labeled nascent RNA by a binomial distribution and sim ultaneousl y measures the RN A synthesis r ate and tr anslation efficiency at the le v el of bulk or specific RNA groups. This pioneering technique re v ealed variable ribosome loading rates among the functional gene subsets. Howe v er, Ribo-seq deals with the short length of RPFs, making the reliable quantification of 4sU incorporation for individual genes difficult ( 35 ).
In this study, we de v eloped a simple method for the simultaneous measurement of nascent RNA synthesis and transla tion a t the gene le v el by combining the TRAP technique with 4sU metabolic RNA labeling. Since TRAP captures full-length RNAs bound to the ribosomes, it is compatible with QuantSeq, which analyzes the inherently uridine-rich 3 UTRs of polyadenylated RNAs, facilitating the robust quantification of T > C conversions. In contrast to the conventional TRAP technique that r equir es genetically engineered cells, we de v eloped an endogenous ribosome affinity purification method using a monoclonal antibody against an e volutionarily conserv ed ribosomal P-stalk protein. The P-stalk mediated translational ribosome affinity purification (P-TRAP) technique enables convenient access to the translatome in di v erse eukaryotes , including mammals , fish, insects and nematodes, without genetic manipulations. We applied a combination of conventional TRAP or the newly de v eloped P-TRAP with 4sU metabolic RNA labeling and defined the nascent RNA synthesis and transla tion a t the le v el of individual genes in the acute unfolded protein response (UPR) within the endoplasmic reticulum (ER). This simple and versatile translatome analysis technique would enhance our understanding of the complex regulation of gene expression in various eukaryotes.

Plasmid and antibodies
A plasmid for the stab le e xpression of enhanced green fluorescent protein (EGFP) fused to the N-terminus of the ribosomal protein L10a (EGFP-L10a) in mammalian cells was constructed using standard procedures. Briefly, a synthetic DNA fragment of EGFP-L10a was cloned into an expression vector pcDNA5 / FRT / TO (V652020, Thermo Fisher Scientific). Although the human ribosomal protein L10a is r eferr ed to as uL1 in standard nomenclature ( 36 ), we used L10a here following the de v elopment of the TRAP method.

Cell lines and animals
Flp-In T-REx 293 cells (R78007, Thermo Fisher Scientific; mentioned as HEK293 in the text and figures) wer e cultur ed at 37 • C in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum, 100 U / ml penicillin, and 100 g / ml streptomycin. pcDNA5 / FRT / TO EGFP-L10a was co-transfected with pOG44 (V600520, Thermo Fisher Scientific) into Flp-In T-REx 293 cells to generate a stable cell line inducibly expressing EGFP-L10a and selected in media supplemented with 50 g / ml hygromycin according to the manufacturer's instructions. EGFP-L10a expression was induced for 24 h by the addition of doxy cy cline at 3 g / ml. Zebrafish were grown at 28 • C in E3 embryo medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl 2 and 0.33 mM MgSO 4 ) using standard methods ( 38 ). Drosophila S2 cells were maintained in Schneider's Drosophila medium (21720024, Thermo Fisher Scientific), supplemented with 10% fetal bovine serum and 1 × antibiotic-antimycotic (15240096, Thermo Fisher Scientific) a t 27 • C . Prepara tion of synchronized KH1668 smg-

PAGE 3 OF 13
Nucleic Acids Research, 2023, Vol. 51, No. 14 e76 2 ( yb979 ) worms were prepared as previously described ( 39 ). Briefly, synchronized gravid worms were cultivated in Scomplete medium supplemented with E. coli strain OP50 and bleached with a standard bleach solution ( 40 ). The embryos were harvested and washed three times with M9 buffer. The embryos were then incubated in M9 buffer for 18 h at 20 • C with gentle agitation for hatching, and L1 larvae were harvested and washed three times with M9 buffer.
For immunoprecipitation of ribosomes from Drosophila S2 cells, 1.5 × 10 6 cells were harvested by centrifugation at 600 × g for 5 min. The cell pellets were lysed in 400 l lysis buffer (50 mM HEPES-NaOH pH 7.5, 10 mM MgCl 2 , 150 mM NaCl, 1% (v / v) NP-40, 0.5 mM DTT, protease inhibitor cocktail, phosphatase inhibitor cocktail, 100 g / ml cy clohe ximide). The lysates were clarified by centrifugation at 20,000 × g for 10 min at 4 • C. The supernatant was divided into two fresh microcentrifuge tubes and incubated with 1 g of 9D5 antibody or isotype control IgG at 4 • C for 60 min. The subsequent procedures were performed as described above.
For immunoprecipitation of the ribosome from zebrafish, 30 whole zebrafish embryos (4 days after fertilization) were resuspended in 1 ml lysis buffer (50 mM HEPES-NaOH pH 7.5, 100 mM KCl, 12 mM MgCl 2 , 1% (v / v) NP-40, 0.5 mM DTT, protease inhibitor cocktail, phosphatase inhibitor cocktail, 100 g / ml cy clohe ximide, 1 mg / ml heparin, 0.2 unit / l RNasin) and homogenized by douncing on ice. The lysates were then clarified by centrifugation at 10,000 × g at 4 • C for 10 min. The supernatant was divided into two fresh microcentrifuge tubes and incubated with 1 g of 9D5 antibody or isotype control IgG at 4 • C for 60 min. Subsequent procedures were performed as described.
For immunoprecipitation of the ribosomes from the worms, synchronized L1 larvae were harvested by centrifu-ga tion a t 600 x g for 1 min. The worm pellets were washed twice with 5 ml of M9 buffer and resuspended in 1 ml worm lysis buffer (50 mM HEPES-NaOH pH 7.5, 10 mM MgCl 2 , 150 mM NaCl, 1% (v / v) NP-40, 0.5 mM DTT, protease inhibitor cocktail, phosphatase inhibitor cocktail, 100 g / ml cy clohe ximide). After homogenization by sonication, the lysates were clarified by centrifugation at 20,000 × g at 4 • C for 10 min. The supernatant was divided into two fresh microcentrifuge tubes and incubated with 1 g of 9D5 antibody or isotype control IgG at 4 • C for 60 min. The subsequent procedures were performed as described above.

ER stress and 4sU metabolic RNA labeling
The HEK293 cell line used for the inducib le e xpression of EGFP-L10a was seeded in a 10 cm dish at 5.0 × 10 5 cells per dish. The expression of EGFP-L10a was induced by 3 g / ml of doxy cy cline for 24 h. The ne xt day, the medium was quickly replaced with a fr eshly pr epar ed medium containing 200 M 4sU with either 0.02% (v / v) DMSO or 1 M thapsigargin. The cells were incuba ted a t 37 • C for 3 h and lysed in 1 ml lysis buffer (20 mM HEPES-NaOH pH 7.5, 2.5 mM MgCl 2 , 150 mM NaCl, 1% (v / v) Triton X-100, 0.5 mM DTT, protease inhibitor cocktail, phosphatase inhibitor cocktail, 100 g / ml cy clohe ximide). Whole cell lysates were clarified by centrifugation at 10,000 x g at 4 • C for 10 min, and the supernatant supplemented with 1 unit / l RNasin Plus was used for further procedure. For cytosolic RNA preparation, 200 l of the supernatant was mixed with 500 l of ISOGEN II (311-07361, NIPPON GENE) supplemented with 1 mM DTT and the pr epar ed sample solution was stored at -20 • C. For P-TRAP, 400 l of the supernatant were incubated with 4 g of 9D5 antibody at 4 • C for 60 min. Afterward, 28 l of Dynabeads Protein G was added to the mixture and incuba ted a t 4 • C for 40 min. After the beads were washed three times with wash buffer (20 mM HEPES-NaOH pH 7.5, 2.5 mM MgCl 2 , 150 mM NaCl, 0.05% (v / v) Tween 20), the beads were directly mixed with 500 l of ISOGEN II supplemented with 1 mM DTT and the sample solution was stored at -20 • C. For EGFP-L10a TRAP, 400 l of the supernatant were incubated with 25 l of GFP-Trap ® Magnetic Particles M-270 (gtd-100, proteintech) at 4 • C for 60 min. After the beads were washed three times with wash buffer, the beads wer e dir ectly mix ed with 500 l of ISOGEN II supplemented with 1 mM DTT, and the sample solution was stored at -20 • C. The manufactur er's instructions wer e f ollowed f or RNA extraction using ISOGEN II, and the RNA samples were assessed for quality and quantity on MultiN A ca pillary electrophoresis instrument (Shimadzu).

SLAMseq and QuantSeq
Cytosolic and immunoprecipitated RNAs were processed according to the standard SLAMseq protocol described previously ( 9 ). In brief, 5 g of cytosolic RNA or immunopr ecipitated RNA wer e incuba ted in 50 l of alkyla tion buffer (50 mM NaPO 4 buffer pH 8.0, 50% (v / v) DMSO, 10 mM iodoacetamide) at 50 • C for 15 min. The reaction was quenched by adding 1 l of 1 M DTT, followed by ethanol precipitation. For each sample, 500 ng of cytosolic RNA or immunoprecipitated RNA was used as an input  for QuantSeq 3 mRNA-Seq Library Prep Kit FWD for Illumina (Lexogen). The cDNA library was pr epar ed according to the manufacturer's instructions. Libraries were assessed for quality using a MultiN A ca pillary electrophoresis instrument (Shimadzu), multiplexed to equimolar concentrations, and sequenced using the HiSeq 2500 system (Illumina) in PE-150 mode with a yield of 20-30 million reads per sample.

Bioinformatic analysis
Total and nascent RNAs were quantified based on sequence data using SLAM-DUNK v 0.4.2, a pipeline for SLAMseq data analysis, with default parameters ( 9 , 41 ). Briefly, 12 bases from the 5 end were trimmed as adaptorclipped reads (-trim-5p 12), and then fiv e or more subsequent adenines from the 3 end were regarded as the remaining poly(A) tail and removed (-max-polya 4). Up to 100 regions with multiple mapped reads were allowed (-topn 100). As QuantSeq mainly targets the 3 end of individual RNAs in a poly(A) tail-dependent manner, the sequence data were aligned on genome-wide 3 UTR sequences generated based on the human genome sequence (GRCh38.p13) and annota tion da ta (gencode.v41.annota tion). Dif ferential gene expression analysis was performed using DE-Seq2 v 1.3.8 ( 42 ). Differentially transcribed genes and dif ferentially transla ted genes were identified using the deltaTE method ( 43 ). For DESeq2 and deltaTE analyses, we used the ReadCount (for total RNA) and TcRead-Count (for nascent RNA) columns of the tcount file (output from SLAM-DUNK) as input. To analyze the TcRead-Count column with the DESeq2 and deltaTE method, we calcula ted global normaliza tion factors using the Read-Count column. The proportion of nascent transcripts to the total transcripts was estimated by normalizing the number of T > C reads to the total reads. Gene ontolo gy anal ysis was performed by over-r epr esentation analysis using a biological process database (WebGestalt 2019) ( 44 ).

Combination of TRAP with 4sU metabolic RNA labeling for simultaneous measurement of nascent transcripts and their translation
SLAMseq distinguishes 4sU-labeled nascent transcripts from pre-existing RNA prior to labeling by bioinformatics detection of specific T > C conversions at the sites of 4sU incorpora tion. The combina tion of SLAMseq with QuantSeq allows the robust quantification of T > C conversions ( 9 ). As QuantSeq r equir es r e v erse transcription from poly(A), it is compatible with polysome profiling and TRAP techniques for recovering pol yadenylated RN As bound to ribosomes. Polysome profiling allows for the fractionation of monosomes and polysomes using a sucrose density gradient to monitor acti v e translation. Howe v er, it r equir es specialized equipment (e.g. ultracentrifuges and gradient fractionation systems), is labor-intensi v e, and does not allow the handling of many samples in parallel ( 24 ). To ensure experimental throughput, we designed a combination of TRAP and SLAMseq / QuantSeq (Figure 1 ).

Affinity purification of endogenous translating ribosome from human cultured cells using an antibody against ribosomal Pstalk
TRAP r equir es genetically engineered cells expressing affinity-tagged ribosomal proteins of the large ribosomal subunit to purify translating ribosome-bound RNAs. The expression of the affinity-tagged ribosomal proteins with specific promoters enables the measurement of cell-, tissueand de v elopmental stage-specific translatome ( 22 , 27-29 ), which is a major advantage of TRAP. Howe v er, the genetic manipulation of endogenous ribosomal proteins reduces experimental throughput and is unsuitable for cells that cannot be genetically engineer ed. Ther efor e, we aimed to de v elop a high-throughput system that allows affinity purification of endogenous non-tagged ribosomes from various cells and organisms. To this end, we focused on the ribosomal P-stalk which is a multimeric ribosomal protein complex in the eukaryotic 60S ribosomal subunit. The Pstalk consists of one copy of ribosomal protein P0 and two heterodimers of ribosomal proteins P1 and P2, forming a pentametric P0-(P1-P2) 2 complex ( 45 ). P0, P1 and P2 share homologous C-terminal intrinsically disordered r egions (IDRs); ther efor e, the P-stalk contains fiv e copies of C-terminal IDRs ( 46 ). As the homologous C-terminal IDRs of the P-stalk are exposed outside the ribosome, the antibody that recognizes the C-terminal IDRs of the P-stalk is expected to be suitable for the immunoprecipitation of endogenous ribosomes (Figure 2 A) ( 47 ). Metz et al. recentl y a pplied an endo genous ribosome imm unoprecipitation approach using 5.8S rRNA as an epitope as well ( 48 ). Besides, our method uses multiple epitopes in the ribosome instead of a single epitope for efficient immunoprecipitation.
We tested whether translating ribosomes could be immunoprecipitated from cultured human cell extracts using a monoclonal antibod y (9D5) tha t recognizes the Cterminal IDRs of P0, P1 and P2 ( 49 ). The isotype control of 9D5 was used as a negati v e control for immunoprecipitation. As expected, the 9D5 antibody immunoprecipitated with ribosomal proteins uL3, uS2 and uS6, and mRNA-binding proteins CBP80, eIF4A3 and PABP4 (Figure 2 B). RNase A treatment markedly decreased the intensity of CBP80, eIF4A3 and PABP4 but not that of ribosomal proteins, re v ealing that these mRNA-binding proteins wer e co-immunopr ecipitated via ribosome-bound mRNA. In addition, the RNA electrophoresis profiles confirmed that r epr esentati v e ribosomal RNA signals (28S and 18S) and smear patterns of ribosome-bound RNAs were obtained by co-immunoprecipitation (Figure 2 C). These results indicate that the 9D5 antibody against the ribosomal P-stalk enables the purification of endogenous non-tagged ribosomes and ribosome-bound RNAs.

Affinity purification of endogenous ribosome via ribosomal P-stalk from diverse eukaryotes
A C-terminal 22 amino acids sequence in the IDRs of the P-stalk recognized by the 9D5 antibody is well-conserved in di v erse eukaryotes from yeast to humans (Figure 3 A). Ther efor e, the immunopr ecipitation of endogenous ribosomes using the 9D5 antibody may be applied to di v erse eukaryotes. We next tested whether the immunoprecipitation of endogenous ribosomes using the 9D5 antibody could be performed in zebrafish, fruit flies, and worms (Figures 3 B-D). In these experiments, the 9D5 antibody immunoprecipitated with the ribosomal proteins uL3, uS3, and uS15, indica ting tha t the 9D5 antibod y can be used to purify endoge-  nous 80S (translational) ribosomes from di v erse eukaryotes without genetic manipulation. We termed the anti-P0 antibod y-media ted purifica tion of endogenous ribosomes as P-stalk-mediated translational ribosome affinity purification (P-TRAP).

Simultaneous measurement of nascent transcriptome and translatome in the UPR by combining SLAMseq / QuantSeq with P-TRAP or conventional L10a-TRAP
We combined SLAMseq / QuantSeq with TRAP to analyze nascent RNAs and their translation at the le v el of individual genes in the unfolded protein response (UPR). In the UPR, transcription and translation are dynamically reprogrammed to reduce unfolded and misfolded proteins in the ER and r estor e protein homeostasis ( 50 , 51 ). To compare P-TRAP with conventional TRAP, we generated a HEK293 cell line for the stab le e xpression of doxy cy clineinducible enhanced green fluorescent protein (EGFP) fused with the N-terminus of ribosomal protein L10a (EGFP-L10a), an initially reported TRAP technique in mammals (Supplementary Figure S1A) (22,(27)(28)(29). The expression of EGFP-L10a enabled the affinity purification of translating ribosomes containing EGFP-L10a using GFP nanobodyconjugated magnetic beads (Supplementary Figures S1B,  C). We subjected the EGFP-L10a-expressing cells with 4sU for meta bolic RNA la beling and tha psigargin, w hich inhibits the sar coplasmic / endoplasmic r eticulum calcium ATPase (SERCA) and leads to the accumulation of unfolded proteins in the ER for 3 h ( 52 ). The cell lysate was divided into three factions: cytosolic RNA (fraction I), ribosome-bound RNAs obtained by P-TRAP (fraction II), and conventional EGFP-L10a-TRAP (fraction III), followed by the QuantSeq (here we termed cytosolic RNAseq, P-TRAP-seq and L10a-TRAP-seq). Principal component analysis (PCA) re v ealed that cytosolic RNA-seq, P-TRAP-seq, and L10a-TRAP-seq were distinguishable in the DMSO-and thapsigargin-treated groups (Supplementary Figure S2). In particular, the cytosolic RNA group was separated from the other two groups in PC1 (46.5% and 49.1% for total and T > C r eads, r especti v ely) and PC2 (15.8% and 8.3% for total and T > C r eads, r especti v ely), re-flecting a low correlation between the transcriptome characterized by standard RNA-seq and the translatome ( 4 , 19-21 ) both in the presence or absence of thapsigargin. The similar trend was also confirmed by calculating correlation coefficients based on counts per million (CPM) for each method (Supplementary Figure S3).

T r anslatome analysis in the UPR by P-TRAP-seq and L10a-TRAP-seq
To validate whether translatome analysis with P-TRAPseq and L10a-TRAP-seq captured gene expression reprogramming in the UPR, we performed differential gene expression analysis using total reads (Figure 4 A, Supplementary Table S1) to identify differ entially expr essed genes (DEGs). In comparison to cytosolic RNA-seq, more DEGs were detected in P-TRAP-seq and L10a-TRAP-seq (656 genes in cytosolic RNA-seq, 788 genes in P-TRAP-seq, and 781 genes in L10a-TRAP-seq). The expression of typical genes involved in the UPR ( HSPA5 , DDIT3 , PDIA4 , DNAJC2 and HERPUD1 ) was significantly increased in P-TRAP-seq analysis similar to cytosolic RNA-seq and conventional L10a-TRAP-seq (Figure 4 B). Notably, specific genes pr efer entially translated in the UPR as determined by polysome profiling ( 53 ) were up-regulated in the P-TRAPseq and L10a-TRAP-seq but not in the cytosolic RNA-seq (Figure 4 C). These results indicate that P-TRAP-seq detects changes in translation without changing the RNA abundance and is well suited for translatome analysis like the conventional L10a-TRAP method and polysome profiling.
Although not as significant as the difference between tr anscriptomes and tr anslatomes, P-TRAP-seq and L10a-TRAP-seq were separated in the PCA analysis (Supplementary Figure S2). This suggested that P-TRAP-seq and L10a-TRAP-seq may have ca ptured slightl y different RNA subsets. Indeed, the DEGs identified by P-TRAP-seq and L10a-TRAP-seq showed incomplete overlap , with n umerous genes uniquely detected by one of the two protocols (Figure 4 D). This discrepancy likely reflects ribosomal heterogeneity and consequent differences in the RNA subsets captured by each technique (see Discussion).

Categorical analysis of nascent transcriptome and translatome in the UPR
In SLAMseq, 4sU incorporated into nascent RNAs is alkylated during library preparation, and T > C conversion occurs via re v erse transcription. Consistent with a previous report ( 9 ), we observed a median rate of 2.3-2.7% for T > C conversion and < 0.2% for any other conversion in total RNA in the presence of 200 M 4sU ( Supplementary Figure S4). We extracted sequence reads with T > C conversion (T > C reads) using the SLAM-DUNK pipeline and obtained the nascent transcriptome (cytosolic RNA-seq) and nascent translatome (P-TRAP-seq and L10a-TRAPseq) during the UPR. Differential gene expression analysis of nascent RNAs re v ealed 443, 663 and 551 DEGs in the cytosolic RNA-seq, P-TRAP-seq and L10a-TRAP-seq, respecti v ely (Figure 5 A, Supplementary Table S1). Similar to the total reads analysis, the expression of typical genes involved in the UPR increased in the nascent transcriptome and translatome (Figure 5 B). Up-regulated DEGs detected by RNA-seq, P-TRAP-seq, and L10a-TRAP-seq in the nascent RN A anal yses were less than those in the total RN A anal ysis. On the other hand, down-regulated DEGs were detected to be in less quantity by RNA-seq as compared to detection by P-TRAP-seq and L10a-TRAPseq (Figure 5 C). In the total reads analysis (Figure 4 D), the DEGs identified by P-TRAP-seq and L10a-TRAP-seq showed incomplete overlap, with numerous genes uniquely detected by one of the two methods (Figure 5 D).
The different ratios of up-and down-regulated DEGs in the total reads analysis to the T > C reads analysis in the transcriptome and translatome suggest that the regulation of nascent RNAs and pre-existing RNAs is different between the transcriptome and translatome (Figure 5 C). To investigate the regulation of nascent RNAs and preexisting RNAs in the transcriptome and translatome, we anticipa ted tha t it would be ef fecti v e to subtract T > C reads from total reads to obtain non-T > C reads which may reflect the abundance of pre-existing RN A. Importantl y, the non-T > C r eads ar e expected to include some reads that are deri v ed from nascent RNAs but do not contain T > C conversion. If the proportion of non-T > C reads derived from these nascent RNAs is large, then the estimation of the proportion of pre-existing RNAs based on the subtracted read counts will be inaccurate. To evaluate the sensitivity of the detecting of T > C reads under the present experimental conditions, we focused on the three genes ( CHAC1 , DDIT3 and HERPUD1 ) that showed the highest ratio of increased RNA le v els after the addition of 4sU and thapsigargin in the cytosolic RNA-seq. In these genes, most reads increased after the addition of 4sU and thapsigargin could be considered to be deri v ed from nascent RNA. By calculating the proportion of reads detected as T > C reads out of all these reads, we estimated the sensitivity of detecting T > C reads under the present experimental conditions and determined that a pproximatel y 82-86% of the reads deri v ed from nascent RNAs were detected as T > C reads (Supplementary Figures S5A and B).
In the acute phase of the UPR, the ER stress sensor PERK phosphorylates translation initiation factor 2 ␣ (eIF2 ␣) at serine 51, leading to global r epr ession of translation initiation, and the whole protein synthesis rate is transiently reduced to 15-20% ( 54 , 55 ). The phosphorylation of eIF2 ␣ also promotes the pr efer ential translation of mR-NAs with upstream open reading frames (uORFs), such as  ( HSPA5 , DDIT3 , PDIA4 , DNAJC3 and HERPUD1 ) and a control gene ( GAPDH ) in the cytosolic RNA-seq (RNA), P-TRAP-seq (P0), and L10a-TRAP-seq (L10a) in response to DMSO-(black) or TPG-(red) treatments. The means and standard deviations of three replicates are shown. ( C ) Venn diagr ams over lapping DEGs detected by the cytosolic RNA-seq, P-TRAP-seq and L10a-TRAP-seq between the total reads analysis (grey) and the T > C r eads analysis (r ed). ( D ) Venn diagr ams over lapping DEGs detected by the cytosolic RNA-seq (red), P-TRAP-seq (blue), and L10a-TRAP-seq (yellow) in the T > C reads analysis. transcription factors ATF4 and CHOP. In the non-T > C reads anal ysis, the RN As enriched in the pol ysome fraction upon eIF2 ␣ phosphorylation were significantly upregulated in both P-TRAP-seq and L10a-TRAP-seq, but not that in cytosolic RNA-seq (Figure 6 A). In contrast, the elevated levels of these RNAs wer e r elati v ely mild in the T > C reads analysis suggesting that the phosphorylated eIF2 ␣-dependent pr efer ential translation primarily occurs to pre-existing RNAs in the UPR.
We next analyzed gene expression reprogramming for the essential branches of the UPR, the ATF4 and CHOP pathway (ATF4 / CHOP target) and the ATF6 and XBP1 pathway (ATF6 / XBP1 target). Consistent with previous reports ( 56 , 57 ), the RNA le v els of A TF4 / CHOP and A TF6 / XBP1 targets were significantly elevated in the T > C reads analysis upon thapsigargin treatment, but not in the non-T > C reads analysis (Figure 6 B). These results are consistent with the enhanced transcriptional activity of ATF4 / CHOP and ATF6 / XBP1 during the UPR, and support that the non-T > C analysis could separate pre-existing RNAs from nascent RNAs in this experimental condition. The abundance of ATF6 / XBP1 targets was also significantly elevated in nascent translatomes obtained by P-TRAP-seq and L10a-TRAP-seq, indica ting tha t the nascent RNAs transcribed by ATF4 / CHOP and ATF6 / XBP1 were translated during the UPR (Figure 6 B).
In the acute phase of the UPR, the total amount of proteins synthesized on the ER membrane is reduced to decrease protein loading to the ER ( 50 , 51 ). The expression of genes encoding membrane and secretory proteins was significantly downregulated in both the T > C reads analysis and the non-T > C reads analysis in the cytosolic RNA-seq data in response to ER stress (Figure 6 C). Notably, the reduction in RNA le v els localized to the ER membrane was more prominent in the non-T > C reads analysis than in the T > C reads analysis. Similar patterns were observed in the P-TRAP-seq and L10a-TRAP-seq data. These results indica te tha t the UPR likel y destabilizes pre-existing RN As localized to the ER membrane, known as the output of regulated IRE1-dependent decay (RIDD) ( 58 , 59 ).

Analysis of nascent transcriptome and translatome in the UPR for individual genes
One of the advantages of combining TRAP-seq and SLAMseq / QuantSeq is the robust estimation of translation le v els of nascent RNAs for individual genes as compared to nRibo-seq. In order to further investigate the translational states in the UPR, we performed deltaTE analysis ( 43 ) for total or nascent RNAs from cytosolic RNA-seq and P-TRAP-seq data (Figure 7 A, Supplementary Table  S2). The deltaTE algorithm classified genes into four categories: (i) genes regulated at the RNA le v el (forwar ded), (ii) genes regulated at the translation le v el (e xclusi v e), (iii) genes regulated by the synergistic effect of RNA and translation le v els (intensified), and (iv) genes regulated by the levels of RNA and translation against each other (buffered). Consequentl y, deltaTE anal ysis re v ealed that the acti v ely transcribed RNAs during UPR (categorized as forwarded)  were loaded to the ribosomes indica ting tha t these acti v ely transcribed RNAs relati v ely escaped from the global repression of translation initiation by eIF2 ␣ phosphorylation (Figure 7 B, expressed genes). These acti v ely translated RNAs included genes involved in the UPR ( HSPA5 , HER-PUD1 , and PDIA4 ), and enrichment analysis showed significant enrichment of genes responsible for ER stress tolerance (Figure 7 B). Notabl y, deltaTE anal ysis re v ealed that a certain subset of nascent RNAs still undergo translational r epr ession (Figur e 7 A). These r esults ar e consistent with previous observa tions tha t the UPR leads to massi v e translational r epr ession associa ted with eIF2 ␣ phosphoryla tion, followed by the synthesis of proteins involved in the proper folding or degradation of unfolded and misfolded proteins in the ER ( 60 , 61 ). Previous study showed that there is a lag in the accumulation of nascent RNAs between the transcriptome and the transla tome a t stead y-sta te condition ( 35 ). Ther efor e, changes in the nascent translatome in the UPR may be influenced by how ra pidl y nascent RN As are loaded onto ribosomes. We estimated the proportion of nascent RNAs in the transcriptome and translatome by normalizing the T > C reads to total reads and investigated whether the UPR influences ribosomal loading of nascent RNAs. At the le v el of bulk RN As, onl y slight differ ences wer e observed in steadystate condition ( Figure 7C). Howe v er, a significant decrease in the proportion of nascent RNAs was observed in the tr anslatome, in contr ast to the tr anscriptome, indicating a reduced ribosomal loading of nascent RNAs in the UPR (Figure 7 C). Notably, the proportion of nascent RNAs of the up-regulated genes during UPR was not significantly decreased in the translatome (Figure 7 D, up-regulated genes). This indicated that ribosomes are ra pidl y loaded on these genes after transcription. On the other hand, the proportion of nascent RNAs of down-regulated genes was significantly reduced in the transla tome, suggesting tha t riboso-mal loading of these genes is decreased in the UPR (Figure 7 D, down-regulated genes). These data suggest the existence of uncharacterized mechanisms, to escape or to be translationally r epr essed in certain gene subsets.
ORF length has been discussed as one of the parameters that determine the ribosomal loading rate ( 35 , 62 ). We investigated whether ORF length might influence the ribosomal loading of nascent RNAs. Although, no significant differ ences wer e detected in stead y-sta te condition, thapsigargin tr eatment decr eased the proportion of nascent RNAs in the translatome in an ORF length-dependent manner (Figure 7 E). These results indicate that ORF length is one of the parameters that influences ribosomal loading of nascent RNAs in the UPR. Further analysis is r equir ed to understand the complexity of gene expr ession r eprogramming in the UPR.

DISCUSSION
In this study, we established a simple protocol for the simultaneous measurement of nascent RNAs and their transla tion a t the gene le v el. Recently, Schott et al. de v eloped nRibo-seq for the simultaneous measurement of nascent RNA synthesis and transla tion ef ficiency a t the le v el of bulk RNAs or specific gene categories by combining Ribo-seq with 4sU metabolic RNA labeling ( 35 ). Instead of Riboseq, we combined TRAP-seq with 4sU metabolic RNA labeling. Because TRAP-seq is compatible with QuantSeq, enabling the accurate quantification of T > C conversions from a small number of sequence reads, it is suitable for the simultaneous measurement of nascent RNA synthesis and translation efficiency at the le v els of indi vidual genes. In addition, TRAP-seq does not r equir e specialized equipment such as ultracentrifugation and gradient fractionation systems, is not labor-intensi v e, and enab les the recovery of ribosome-bound RNAs from a small number of cells. each total or nascent RNAs in cytosolic RNA-seq and P-TRAP-seq da ta a t the le v els of the individual genes analyzed by the deltaTE method (FDR < 0.05 and log 2 fold change > |0|). Translationally forwarded genes (cyan), intensified genes (b lue), e xclusi v e genes (red) and buffered genes (purple) are highlighted. Right: Regulation modes categorized by deltaTE analysis ( 43 ). ( B ) The enrichment ratios in over-r epr esentation analysis (ORA) of up-regulated translationally forwarded genes in deltaTE analysis (T > C reads) performed by WebGestalt 2019 analysis ( 44 ). Resulting biological processes and FDR values are sho wn. ( C ) Bo x plots of the proportion of nascent RNAs at the le v el of bulk RNAs in the cytosolic RNA-seq (RNA) and P-TRAP-seq (P0) in response to DMSO-(black) or TPG-(red) treatment. n.s., not significant, * P < 0.05, **** P < 0.0001 (Mann-Whitney U test). (D, E) Box plots of the proportion of nascent RNAs in the groups of RNAs according to expression sta tes (up-regula ted genes and down-regula ted genes) ca tegorized by deltaTE analysis ( D ) or groups of RNAs according to ORF length ( E ). n.s., not significant, * P < 0.05, *** P < 0.001, **** P < 0.0001 (Mann-Whitney U test).
These advantages of TRAP-seq make it compatible with automated libr ary prepar ation systems for high-throughput sequencing procedures, such as single-cell RNA sequencing. Furthermore, in contrast to conventional TRAP-seq, which r equir es genetic manipula tion for the af finity purification of ribosomes, the newly developed P-TRAP-seq method can be applied to immunoprecipitate endogenous ribosomes without genetic manipulation. Ther efor e, measuring nascent RNAs and their translation using P-TRAPseq with 4sU metabolic RNA labeling, nascent-P-TRAPseq (nP-TRAP-seq), is a simple and robust method for investigating the crosstalk between transcriptional and transla tional regula tion. Furthermore, non-T > C reads may be analyzed as pre-existing RNAs under experimental conditions that ensure high detection sensitivity for T > C reads.
As shown in the present study, we observed pr efer ential tr anslational reprogr amming of both nascent and preexisting RNAs in response to the UPR. Although individual translation efficiency can be monitored using massspectrometry-based techniques such as PUNCH-P ( 63 ) and BONCAT ( 64 ), our sequencing-based method is simple, low-cost, and provides highly accurate measurements.
It should be noted that the translatome measured using TRAP-seq only reflects ribosome-bound RNA le v els. In contrast to Ribo-seq, TRAP-seq cannot determine the position of individual ribosomes in RNAs and has less resolution for complex translational regulation (e.g. translation efficiency for multiple open reading frames in a single transcript). TRAP-seq relies on a simple immunoprecipitation technique to extract ribosomes, including monosomes and polysomes; ther efor e, it cannot selecti v ely fractionate acti v ely translating polysomes, similar to the polysome profiling technique ( 65 ). Since the polysome profiling also yields full-length ribosome-bound RNA and is compatible with QuantSeq, a combination of polysome profiling and 4sU meta bolic RNA la beling is pr eferr ed w hen anal yzing more ef ficiently transla ted RNAs (i.e. RNAs bound to multiple ribosomes).
Ther e ar e two major ad vantages to the imm unoprecipitation of endogenous ribosomes with C-terminal IDRs of the ribosomal P-stalk. The first is the high specificity and efficiency of immunoprecipitation. The C-terminal IDRs of the P-stalk were present in fiv e copies on the 60S large ribosomal subunit and exposed on the solvent side (Figure 2 A). Although these IDRs bind to multiple translational GT-Pase factors (trGTPases), such as eEF1A and eEF2 ( 66 ), the 9D5 antibody binds pr efer entiall y and stabl y to the P-stalk rather than to trGTPases ( 49 , 67 ). In addition, because Pstalk incorporation into the pre-60S ribosomal subunit occurs in the late cytoplasmic stage of ribosome assembly, P-TRAP onl y ca ptur es matur e translating ribosome particles ( 68 , 69 ). Ther efor e, antibodies r ecognizing the C-terminal IDRs of the P-stalk ensure stable, specific, and robust immunoprecipitation of endogenous translating ribosomes.
Secondly, the C-terminal IDRs of the P-stalk are conserved across most, if not all, eukaryotes ( 47 ). Indeed, antibodies that recognize the highly conserved C-terminal IDR allow the purification of endogenous ribosomes from cultured human cells , zebrafish, flies , and nematodes without any genetic manipulations ( Figure 3 ). P-TRAP can shorten the experimental period compared with conventional TRAP, which r equir es genetic manipulation. Furthermor e, immunopr ecipitation of endogenous ribosomes in P-TRAP allows translatome analysis from non-model organisms without the need for genetic manipulation techniques or from human specimens in clinical r esear ch. Mor eover, the immunoprecipitation of ribosomes using P-stalks can be applied to other experimental systems that r equir e ribosome collection, such as Ribo-seq. Highly efficient ribosome collection without ultracentrifugation is expected to significantly reduce the time and effort required for experiments and simplify protocols. These advantages of P-TRAP will acceler ate tr anslatome r esear ch and contribute to our understanding of gene expr ession r egulation in various biological fields such as UPR ( 50 , 51 ), nonsense-mediated mRN A decay ( 70 ), imm une r esponses ( 71 ), and drug r esponses ( 44 ).
Both P-TRAP-seq and L10a-TRAP-seq captured the changes of the translatome in the UPR, but they identified se v eral different DEGs (Figure 4 D and 5 C). Recent proteomic analyses have shown diversity in the composition of ribosomal proteins, r eferr ed to as ribosomal heterogeneity, resulting in the preferential translation of specific RNA. For example, L10a is present at substoichiometric le v els in mouse embryonic stem cells (mESCs), and ribosomes containing L10a have an altered affinity for some transcripts ( 72 ). Performing TRAP-seq using substoichiometric ribosomal proteins as bait for immunoprecipitation may cause translatome bias. The P-stalk is present in polysomes with 1:1 stoichiometry in mESCs ( 72 ), indica ting tha t the translatome obtained with P-TRAP-seq would have less trans-lation bias due to ribosome heterogeneity, making the Pstalk an ideal bait for the TRAP-seq technique. Howe v er, in se v eral e xperimental designs, P-TRAP has a disadvantage compared to the conventional TRAP technique because of the purification of ribosomes without genetic manipulation. Conventional TRAP captures gene expression from particular cells / tissues of genetically modified organisms expressing affinity-tagged ribosomal proteins using cell / tissue-specific promoters, such as the Gal4-UAS system in the fly or the Cre-lox system in mice ( 27 , 28 , 73 ). In contrast to conventional TRAP, P-TRAP cannot fractionate ribosomes from specific cells or tissues. Ther efor e, it is essential to use conventional TRAP and / or P-TRAP according to the experimental design.

DA T A A V AILABILITY
The deep sequencing data reported in this study have been deposited in the DDBJ (DRA015587). All codes r equir ed to perform the computational analysis are available on Zenodo ( https://zenodo.org/record/7943421 ).

SUPPLEMENT ARY DA T A
Supplementary Data are available at NAR Online.