RAPIDASH: Tag-free enrichment of ribosome-associated proteins reveals composition dynamics in embryonic tissue, cancer cells, and macrophages

Abstract

Ribosomes are emerging as direct regulators of gene expression, with ribosome-associated proteins (RAPs) allowing ribosomes to modulate translation. Nevertheless, a lack of technologies to enrich RAPs across sample types has prevented systematic analysis of RAP identities, dynamics, and functions. We have developed a label-free methodology called RAPIDASH to enrich ribosomes and RAPs from any sample. We applied RAPIDASH to mouse embryonic tissues and identified hundreds of potential RAPs, including Dhx30 and Llph, two forebrain RAPs important for neurodevelopment. We identified a critical role of LLPH in neural development linked to the translation of genes with long coding sequences. In addition, we showed that RAPIDASH can identify ribosome changes in cancer cells. Finally, we characterized ribosome composition remodeling during immune cell activation and observed extensive changes post-stimulation. RAPIDASH has therefore enabled the discovery of RAPs in multiple cell types, tissues, and stimuli and is adaptable to characterize ribosome remodeling in several contexts.

Keywords: cancer; embryonic development; macrophages; proteomics; ribosome; ribosome heterogeneity; ribosome-associated proteins; translational control.

Figure 1:. Characterization of RAPIDASH method in mouse embryonic stem cells (mESCs)

(A) Schematic of the RAPIDASH protocol. Cytoplasmic lysates from any sample undergo sucrose cushion ultracentrifugation and sulfhydryl-charged resin chromatography to enrich RNA-containing, high molecular weight complexes. (B) Characterization of cysteine-charged sulfolink resin. Sucrose cushion pellet samples and poly(A) RNA isolated from mESCs underwent cysteine-charged sulfolink resin chromatography. The percentage of RNA relative to input amount is plotted for the flowthrough, eluate, and resin-bound samples. N.D., not detected. Error bars are +/− standard error of the mean (SEM). (C) Characterization of binding to the cysteine-charged sulfolink resin for 80S ribosomes isolated from untreated, puromycin-treated, or RNase A-treated mESC samples. Left: the percentage of input RNA that was eluted from or bound to the resin was normalized to that for the untreated sample for three technical replicates. Error bars are +/− SEM. Right: untreated and RNase A-treated 80S samples were probed by western blotting for the presence of Pabp1, MetAP1, Rps5, and Rpl4. The horizontal black line separates sister blots loaded with the same samples. The vertical black line represents an omitted lane. (D) Boxplot of normalized log₂ fold change (FC) of RAPIDASH eluate over sucrose cushion pellet tandem mass tag (TMT) mass spectrometry ratios from three biological replicates of mESCs. RPs (red) are significantly enriched over other proteins (gray) by RAPIDASH compared to sucrose cushion ultracentrifugation alone based on Welch’s t-test. p-values: rep1 = 5.58 × 10⁻¹⁹; rep2 = 4.96 × 10⁻²²; rep3 = 3.89 × 10⁻¹⁵. (E) Western blotting of mESC sucrose cushion pellet and RAPIDASH eluate samples to assess the specificity of RAPIDASH. Approximately equal amounts of RPs (see Figure S3D) for sucrose cushion pellet and RAPIDASH eluate samples were analyzed by western blotting for Nup62, Atp5a1, Tom20. Rpl4, Rpl29, Rps5, Rps26, and Rps27 are RPs. Cytoplasmic lysate is an input control. Molecular weight markers are in kilodaltons (kDa). (F) Western blot detection of known RAPs enriched by RAPIDASH. A representative blot with 1% of the mESC cytoplasmic lysate volume and 35% of the RAPIDASH eluate volume was probed for the known RAPs Metap1, Ufl1, Upf1, Ddx1, and Nsun2^,. Molecular weight markers are in kDa. (G) Bar graph showing the translational machinery identified by Ribo-FLAG immunoprecipitation (IP) or RAPIDASH. Ribo-FLAG IP proteins were identified by FLAG IP liquid chromatography-tandem mass spectrometry (LC-MS/MS) of endogenously FLAG-tagged Rpl36 or Rps17 in mESCs. Three biological replicates of RAPIDASH were performed for each of the same cell lines and analyzed by LC-MS/MS. Proteins identified in at least three RAPIDASH samples were compared against the Ribo-FLAG IP proteins. The percentage of translational machinery components identified only in Ribo-FLAG IP (red), only in RAPIDASH (green), in both techniques (blue), or none (gray) are displayed. (H) Gene ontology (GO) term analysis of mESC proteins enriched by RAPIDASH. Proteins that were identified in RAPIDASH-enriched mESC samples were analyzed by Manteia. The ten most significant GO molecular function (GOMF) terms at level 4 or lower are shown.

Figure 2:. Characterization of Dhx30 as a bona-fide mRNA-independent RAP

(A) Analysis of GO terms in forebrain MS data. RAPIDASH was performed on E12.5 mouse forebrain samples. These samples were analyzed by LC-MS/MS. The top 5 GOMF terms at level 4 or lower as analyzed by Manteia are shown. (B-C) Sucrose gradient and western blotting analysis of Dhx30. Sucrose gradient fractionation was performed on mESC cytoplasmic lysate without (B) or with (C) EDTA treatment. The proteins from each fraction were precipitated and analyzed by western blotting; lanes were loaded with equal volumes of precipitated protein sample. RPs are markers for the 40S and 60S ribosomal subunits. Molecular weight markers are in kDa. (D) Sucrose gradient fractionation and western blotting analysis of Dhx30 after RNase A treatment. The mESC cytoplasmic lysate was treated with (blue) or without (black) RNase A, fractionated, and probed by western blotting as in Figure 2B. Rpl4 is a marker for large ribosomal subunit, and Pabp1 is a marker for mRNA-dependent ribosome association. Molecular weight markers are in kDa. (E) Effect of Dhx30 knockdown on global protein synthesis. Protein synthesis in mESCs was measured by O-propargyl-puromycin (OP-Puro) incorporation. Top: OP-Puro median fluorescence intensity (MFI) between siFluc (a control siRNA) and siDhx30 (siRNA against Dhx30) across four biological replicates. Error bars are +/− SEM. Bottom: western blotting of Dhx30 in mESCs treated with siFluc or siDhx30. Actin is a loading control. Molecular weight markers are in kDa. (F) Illustration of Dhx30 constructs. Oligonucleotide/oligosaccharide binding (OB) fold domain; and double-stranded RNA-binding domains (dsRBDs). (G) Ribosome association of Dhx30 mutants. mESCs were transfected with V5-tagged Dhx30 mutants, lysed, and subjected to sucrose cushion ultracentrifugation. The lysate, pellet, and supernatant (sup.) were probed by western blotting for the presence of the V5, Rps6, and Gapdh. Top: V5-Dhx30, ∆OB-fold, and R805/8A constructs. Bottom: V5-Dhx30, ∆dsRBD-1, and ∆dsRBD-1/2 constructs. Molecular weight markers are in kDa. (H) Quantification of the proportion of Dhx30 mutants in the sucrose cushion pellet or supernatant. Three biological replicates of mESCs were transfected with V5-Dhx30, ∆OB-fold, and R805/8A constructs as in Figure 2G and analyzed by western blotting for the presence of Dhx30. Data are plotted as a percentage of the total amount of Dhx30 found in the combined pellet and supernatant fraction. Error bars are +/− standard deviation.

Figure 3:. LLPH is a RAP with a role in neurodevelopment

(A) LLPH binding location on the ribosome. Previously published cryo-EM data show LLPH binds near the sarcin-ricin loop (SRL) of the human pre-60S particle (Protein Data Bank ID: 6LSS). Helices 91, 95 (SRL), and 97, as well as RPL9, RPL3, and LLPH are colored. (B) Sucrose gradient and western blotting analysis of LLPH co-fractionation with ribosomes. P493-6 cell lysate treated with (red) or without (black) EDTA was fractionated, and the proteins from each fraction were precipitated and analyzed by western blotting as in Figure 2B. RPL8 is a marker for large ribosomal subunit-containing fractions. Molecular weight markers are in kDa. (C) Measurement of traced primary neurite lengths of individual LLPH^+/+ and LLPH^Nterm/Nterm human Ngn2-induced neurons (hiNs) at days in vitro (DIV) 30. The median neurite length is displayed. The p-value was calculated using the Wilcoxon signed-rank test. (D) Representative fluorescence images of fixed DIV 30 wild-type and LLPH^Nterm/Nterm hiNs. Wild-type and LLPH^Nterm/Nterm hiNs were fixed and stained with a primary antibody against MAP2 and 4',6-diamidino-2-phenylindole) (DAPI). (E) Comparison of Ribo-seq and RNA-seq data for DIV 14 LLPH^+/+ and LLPH^Nterm/Nterm hiNs (n = 3 biological replicates). Blue genes significantly change (Benjamini-Hochberg procedure false discovery rate (FDR) < 0.1 and absolute fold change (FC) ≥ 2) in mRNA abundance only; red genes change in ribosome occupancy only; purple genes change in mRNA abundance and ribosome occupancy. (F) Representative genes with lower mean translation efficiency differences in LLPH^Nterm/Nterm vs. LLPH^+/+h hiNs. Top: representative genes with downregulated translation efficiency in LLPH^Nterm/Nterm compared to LLPH^+/+ that are involved in building the extracellular matrix. Bottom: representative genes known to be linked to neurodevelopmental defects. (G) Boxplots of coding sequence (CDS) length for genes that are translationally upregulated, downregulated, or unchanged when comparing LLPH^Nterm/Nterm and LLPH^+/+h hiNs. Significance was calculated using the Mann-Whitney U test. The median CDS lengths are displayed. Only genes with CDS lengths shorter than 6,000 nucleotides are displayed; the full plot is shown in Figure S4F.

Figure 4:. Characterization of tissue-specific RAPs in E12.5 mouse embryos

(A) Schematic of the strategy to quantify tissue-specific RAPs by TMT mass spectrometry. Four biological replicates of forebrain, limbs, and liver tissues from mouse embryos underwent RAPIDASH. Peptides derived from the enriched proteins were labeled with TMT reagents and analyzed by LC-MS/MS. (B) Volcano plots comparing RAPs identified in different tissues. Putative RAPs identified in: liver to limbs (left), forebrain to limbs (center), forebrain to liver (right) are shown using volcano plots graphing −log₁₀(p-value) against log₂(FC). Proteins present in at least three out of four biological replicates with absolute log₂FC ≥ 1 and FDR < 0.10 are defined as significantly differentially enriched (red). Significance was calculated using a paired Welch’s t-test. (C) Characterization of Elavl2 as a forebrain-enriched RAP. E12.5 mouse embryonic forebrain, limb, and liver tissues were lysed, treated with (red line) or without (black lines) EDTA, and separated by sucrose gradient fractionation. The proteins from each fraction were precipitated and analyzed by western blotting for the presence of Elavl2 or Rps5, a marker for the ribosome. Lanes were normalized by equal volume of samples. Molecular weight markers are in kDa.

Figure 5.. RAPIDASH identifies RAPs in macrophages following TLR stimulation

(A-B) Volcano plots of murine bone marrow derived macrophages (BMDMs) stimulated with lipopolysaccharide (LPS) (A) or polyinosinic-polycytidylic acid (poly(I:C)) (B) for 6, 12, or 24 hours prior to isolation of ribosome complexes for TMT-MS analysis. X-axes show the log₂FC of ribosome complex composition in activated versus unstimulated macrophages at 6 hours (left panels), 12 hours (middle panels), and 24 hours (right panels). Significance was calculated using Student’s t-test. (C-F) Characterization of candidate RAPs in unstimulated or stimulated BMDMs. Lysates from unstimulated or LPS-stimulated BMDMs (C-E) or macrophages differentiated from HoxB8-immortalized progenitor cells (F) were subjected to sucrose gradient. Total protein was extracted from individual fractions and subjected to western blot analysis for the indicated proteins. Lanes were loaded with equal volumes of precipitated protein. Puromycin and RNase treatments were performed as described in the Methods section. Molecular weight markers are in kDa.