Protocol for ribosome-protected fragment isolation from rare leukemia cell populations

Summary

Malignant cells under chemotherapy stress alter protein translation, necessitating sensitive methods to profile rare, surviving populations in vivo. We present a protocol for profiling low-input acute myeloid leukemia samples using an optimized ribosome profiling approach. We describe steps for lysing cell inputs, digesting unshielded RNA, enriching ribosome footprints, and gel-based size selection. We then detail procedures for constructing sequencing-ready libraries and using a computational pipeline to align and quantify both protected fragments and matching transcriptomes for reproducible translation analysis.

For complete details on the use and execution of this protocol, please refer to Mayerhofer et al.¹

Graphical abstract

Highlights

• •Steps for isolating ribosome footprints from as few as 10,000 rare, persistent leukemia cells
• •Instructions for sucrose cushion ultracentrifugation to purify translating ribosomes
• •Guidance on size-selecting footprints and building paired RPF/transcriptome libraries
• •Procedures for computational alignment and calculation of translation efficiency

Publisher’s note: Undertaking any experimental protocol requires adherence to local institutional guidelines for laboratory safety and ethics.

Malignant cells under chemotherapy stress alter protein translation, necessitating sensitive methods to profile rare, surviving populations in vivo. We present a protocol for profiling low-input acute myeloid leukemia samples using an optimized ribosome profiling approach. We describe steps for lysing cell inputs, digesting unshielded RNA, enriching ribosome footprints, and gel-based size selection. We then detail procedures for constructing sequencing-ready libraries and using a computational pipeline to align and quantify both protected fragments and matching transcriptomes for reproducible translation analysis.

Before you begin

Understanding the molecular mechanisms that underlie cancer cell survival is essential to uncover therapeutic vulnerabilities and improve treatment strategies. While transcriptomic profiling, including bulk and single-cell RNA sequencing, has been invaluable for mapping gene expression patterns in cancer, it does not capture the effect of translational regulation that governs protein synthesis. Translation is increasingly recognized as a central node in cancer biology, allowing cells to rapidly reprogram protein output in response to environmental stress or therapy-induced challenges. In acute myeloid leukemia (AML), a malignancy marked by high heterogeneity and poor long-term outcomes, our recent studies have shown that chemotherapy-resistant cells dynamically upregulate protein synthesis during treatment.¹ These changes occur independently of transcriptional shifts and are thus invisible to RNA-based profiling approaches.

A major obstacle to studying translation in vivo especially in rare subpopulations of cells is the need for large input material, limiting most ribosome profiling (Ribo-seq) applications to bulk samples or in vitro systems. However, cells that survive chemotherapy, and are thought to drive relapse in AML, often exist at very low frequencies and must be studied in their native microenvironment to preserve physiological relevance. Evidence suggests that these rare, drug-tolerant cells exhibit translational plasticity, selectively altering ribosome engagement with mRNAs to adapt and persist in cytotoxic conditions. These adaptive features would likely be missed without tools that can resolve the translatome with low input requirements and high sensitivity.

To address this challenge, we optimized Ribo-lite,² a low-input ribosome profiling protocol that enables translatome analysis from as few as 10,000 hematopoietic cells, without compromising data quality. This method adapts principles of Ribo-seq to suit rare populations of cells, such as residual AML blasts post-chemotherapy, and incorporates enhanced lysis, digestion, and ribosome purification steps to maximize recovery of ribosome-protected fragments (RPFs). Ribo-lite thus provides an opportunity to study translation in primary tumor models and stem cells, even when input material is limited.

In our original study, we applied Ribo-seq to investigate translational changes in chemoresistant AML cells persisting after induction therapy. We found that these cells rewire their translational program to sustain protein production and cell survival under therapeutic stress. These findings highlight the importance of studying the translatome, not just the transcriptome, in cancer progression and treatment response. While we used an MLL-AF9 driven AML model,¹ the Ribo-lite protocol described here should be compatible with any murine leukemia that includes a fluorescent marker for FACS-based isolation and can also be extended to study translation in other rare hematopoietic or malignant cell populations in vivo.

Institutional Permissions: Confirm that all procedures involving animals were approved by Institutional Animal Care and Use Committee (IACUC).

Preparation

• 1.Use RNase-free consumables and Open-Top Thickwall polycarbonate tubes for ultracentrifugation.
• 2.The Ultracentrifuge rotor can fit 10 samples, therefore preparation of maximum 10 samples per experiment is recommended.
• 3.Prepare all buffers in advance, they can be stored at 4°C for several days. On the day of the experiment, add fresh components such as cycloheximide (CHX), DTT, Triton X-100, Turbo DNase, and RNase inhibitor and keep the buffers on ice until use.
• 4.Pre-cool centrifuge rotors to 4°C. Optional: Pre-cool tip boxes by placing them in the fridge.
• 5.
  First use DNAZap, then RNaseZap to clean the workplace including pipettes and equipment such as racks.

  Note: This limits the risk of contamination by other materials.

• 6.Get new tip boxes and RNase free tubes.

Innovation

This protocol streamlines and optimizes Ribo-seq to enable robust library generation from rare leukemia populations using as few as ∼10,000 cells, addressing a limitation of conventional workflows that require substantially higher input.

The core chemistry follows the original Ribo-lite protocol,² adapted and validated here for rare hematopoietic populations. The advances of Ribo-lite² over conventional methods include: (1) a simplified, small-volume lysis strategy that minimizes sample handling; (2) optimized RNase I digestion conditions scaled to low cell numbers to improve footprint yield; (3) sucrose cushion–based monosome enrichment via ultracentrifugation improving usability for limited material; and (4) a precise, gel-based RPF size-selection workflow (26–34 nt) using a parallel marker gel to enhance reproducibility and reduce cross-contamination risk during excision.

This protocol integrates paired transcriptome library preparation from one aliquot of the same lysate, enabling matched translatome/transcriptome measurements for downstream translation-efficiency analyses. Finally, we provide a fully implemented, open-source computational pipeline with separate processing scripts for RPF (including ribosomal RNA depletion) and mRNA libraries, supporting standardized trimming, alignment, deduplication/QC, and gene-count generation for reproducible end-to-end analysis.

Institutional permissions

All experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the Faculty of Arts and Sciences of Harvard University (Cambridge, MA, USA). Mice were maintained in pathogen-free conditions.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Chemicals, peptides, and recombinant proteins
BSA	Cell Signaling Technology	9998S
DNAZap	ThermoFisher Scientific	AM9890
RNaseZap	ThermoFisher Scientific	AM9780
Dynamarker for small RNA	Diagnocine	DM253S
MicroRNA Marker for small RNA	New Englad Biolabs	N2102S
TBE Buffer	ThermoFisher Scientific	15581044
TBE-Urea Gels, 15%, 10 well	ThermoFisher Scientific	EC6885BOX
GlycoBlue coprecipitant	ThermoFisher Scientific	AM9516
Gel loading dye	ThermoFisher Scientific	AM8546G
Tris-HCl Buffer	Bioworld	21420063-2
Triton X100	ThermoFisher Scientific	327371000
NaCl	Sigma-Aldrich	59222C
MgCl2	ThermoFisher Scientific	AM1028
Sodium acetate	ThermoFisher Scientific	AM9740
Sucrose	ThermoFisher Scientific	419762500
EDTA	ThermoFisher Scientific	15575020
SDS	ThermoFisher Scientific	15553027
Turbo DNAse	ThermoFisher Scientific	AM2239
RNase Inhbitor	Sigma-Aldrich	3335402001
SUPERase Inhibitor	ThermoFisher Scientific	AM2696
RNase 1	ThermoFisher Scientific	AM2295
TRIzol	ThermoFisher Scientific	15596026
Choloroform	Sigma-Aldrich	C2432
Isopropanol	ThermoFisher Scientific	149320050
Ethanol 200 proof	Decon labs	2716GEA
Dithiothreitol	ThermoFisher Scientific	15508013
RNase-free tubes	ThermoFisher Scientific	AM12450
Open-Top Thickwall Polycarbonate Tube	Beckman Coulter	343778
PCR tubes	ThermoFisher Scientific	951010022
AMPure XP	Beckman Coulter	a63881
XCell SureLock Mini-Cell and XCell II Blot Module	Invitrogen	EI0002
Cell culture dish, 15cm	Corning	CLS353025
SYBR Gold Nucleic Acid Gel Stain	ThermoFisher Scientific	S11494
Nuclease-free water	ThermoFisher Scientific	AM9932
Critical commercial assays
SMART-seq® v4 Ultra® Low Input RNA Kit	Takara	634889
D-Plex Small RNA Kit	Diagenode	C05030001
D-Plex Unique Dual Indexes for Illumina	Diagenode	C05030021
Nextera XT DNA Library Preparation Kit	Illumina	FC-131-1024
Nextera XT index set	Illumina	FC-131-2001
Qubit microRNA Assay Kits	ThermoFisher Scientific	Q32880
Qubit RNA High Sensitivity	ThermoFisher Scientific	Q32852
Qubit DNA High Sensitivity	ThermoFisher Scientific	Q32851
Experimental models: Cell lines
MLL-AF9 mouse AML	Sykes et al. and Van Gastel et al.3,4	N/A
Experimental models: Organisms/strains
C57BL/6J mice, 6–8 weeks, female or male	Jackson Laboratories	00664
Software and algorithms
Prism (version 10.0.2)	GraphPad Software	N/A
FlowJo (version 10.8.1)	Flowjo	N/A
Biorender	Biorender	N/A
Cell Ranger	10× Genomics	N/A
Rsubread	Liao et al.5	N/A
featureCounts	Liao et al.6	N/A
DESeq2	Love at al.7	N/A
clusterProfiler	Wu et al.8	N/A
RSeQC package	Wang et al.9	N/A
RiboCode	Xiao et al.10	N/A
cutadapt	Martin, M.11	N/A
bowtie2	Langmead et al.12	N/A
samtools	Danecek et al.13	N/A
fumi_tools	https://ccb-gitlab.cs.uni-saarland.de/tobias/fumi_tools	N/A
RiboWaltz	Lauria et al.14	N/A
RStudio	RStudio	N/A
Others
Refrigerated microcentrifuge 5430R with FA-45-30-11 rotor	Eppendorf	5428000010
Analytical balance	Mettler Toledo	ME54
LP Vortex Mixer	ThermoFisher	11676331
T100 Thermal Cycler	Biorad	1861096
Beckman Optima MAX-XP ultracentrifuge	Beckman Coulter	a47882
MLA-150 rotor	Beckman Coulter	357656
Magnetic rack	Diagenode	B04000001
Blue Light Transilluminator	VWR	76151-834
Qubit 4 Fluorometer	ThermoFisher Scientific	Q33238

Materials and equipment

Reagents marked in blue should be added fresh on the day of the experiment.

Sorting buffer (0.5% BSA in PBS)

For 500 mL buffer, dissolve 2.5 g BSA in 500 mL PBS, mix until fully dissolved.

Lysis Buffer

Reagent	Final concentration	Amount
1 M Tris-HCl pH 7.4	20 mM	200 μL
5 M NaCl	150 mM	300 μL
1 M MgCl2	5 mM	50 μL
H2O	N/A	9.450 mL
Total	N/A	10 mL

Store the basic lysis buffer at 4°C for several days until use.

On the day of the experiment, use 3410.75 μL of the basic buffer to prepare 3.5 mL Ribo-lite lysis buffer (sufficient for 10 samples) by adding the following components.

Reagent	Final concentration	Amount
Basic lysis buffer	N/A	3410.75 μL
add fresh: 1 M DTT	1 mM	3.5 μL
add fresh: 50 mg/mL CHX	100 μg/mL	7 μL
add fresh: Triton X-100	1%	35 μL
add fresh: 2 U/μL Turbo DNase	25 U/mL	43.75 μL
Total	N/A	3.5 mL

Sucrose cushion

To prepare the sucrose cushion, first add approximately half of the required volume of nuclease-free water to the tube, then add the sucrose and allow it to dissolve for 30 min at 18-22°C before adding the remaining water and other components.

Reagent	Final concentration	Amount
Sucrose	1M, 34,23% w/v	6.846 g
H2O	N/A	First, add 10 mL of H2O to the sucrose and allow it to dissolve completely, then adjust the final volume to 18.9 mL
1 M Tris-HCl pH 7.4	20 mM	400 μL
5 M NaCl	150 mM	600 μL
1 M MgCl2	5 mM	100 μL
Total	N/A	20 mL

Store the basic sucrose cushion at 4°C for several days until use.

On the day of the experiment, use 8,964 μL of the basic sucrose cushion to prepare 9 mL sucrose cushion (sufficient for 10 samples) by adding the following components.

Reagent	Final concentration	Amount
add fresh: 1 M DTT	1 mM	9 μL
add fresh: 50 mg/mL CHX	100 μg/mL	18 μL
add fresh: 20 U/μL SUPERase Inhibitor	20 U/mL	9 μL
Total	N/A	9 mL

RNA extraction buffer

Remove the RNA extraction buffer from the fridge 3–4 h before use to ensure complete dissolution of SDS.

Note: If time is limited, the solution can be placed in a 37°C water bath for a few minutes until the solution clears. The solution should be allowed to cool back to 18°C–22°C before use.

Reagent	Final concentration	Amount
3 M Sodium acetate pH 5.5	300 mM	1000 μL
0.5 M EDTA	1 mM	20 μL
10% SDS	0.25%	250 μL
H2O	N/A	8730 μL
Total	N/A	10 mL

Store the RNA extraction buffer at 4°C for several days until use.

Pellet buffer

Remove the pellet buffer from the fridge 3–4 h before use to ensure complete dissolution of SDS.

Reagent	Final concentration	Amount
1 M Tris-HCl pH 7.4	10 mM	50 μL
10% SDS	1%	500 μL
H2O	N/A	4450 μL
Total	N/A	5 mL

Store the pellet buffer at 4°C for several days until use.

Step-by-step method details

Day 1

Sample cell lysis

Timing:25–30min

Here, we describe steps for preparation of the cells and sample lysis.

• 1.
  Preparation of the cells:

  • a.
    Place cells in flow-buffer on ice.
  • b.
    Sort cells using a 70 μm nozzle, with gating based on a fluorescent marker (GFP) and purity mode;
  • c.
    If sorted cells are used, lyse them directly after sorting.

Note: Cell number is determined from the sorter's event count. Sorting success can be additionally verified by visual inspection of pellet size following centrifugation. Independent cell counting (e.g., cell counter) can optionally be performed for further verification prior to digestion. Accurate knowledge of cell number is required to determine the appropriate RNaseI volume (in step 10).

• 2.Centrifuge sorted (or counted cells) in 0.5% BSA buffer at 500 × g, 7 min, 4°C.

Note: Ensure correct cell counts to avoid over-digestion.

• 3.
  Place cells on ice.

  • a.
    Carefully remove the supernatant.
  • b.
    Add 20 μL ice-cold lysis buffer.
  • c.
    Flick the tube to resuspend the pellet and mix.
• 4.Incubate on ice for 10 min.

Note: The split-lysis step follows the co-profiling approach of Zou et al.¹⁵

• 5.For transcriptome, withdraw 2 μL of lysate (10% of the sample) and process using the SMART-seq® v4 Ultra® Low Input RNA Kit protocol.

Note: We recommend placing the lysate in 1 mL Trizol, freezing at −80°C, and proceeding with the protocol on Day 6. We do not recommend measuring the RNA concentration at this stage. Instead, the RNA input should be measured at step 60.

• 6.Add 300 μL additional lysis buffer to the remaining lysate (90% of the sample).
• 7.Incubate on ice for 10 min.

Note: If multiple samples require extended sorting times, lysates can be kept on ice during this period. If same-day processing is not feasible, lysates may be frozen at −80°C up to one week, the longest duration tested and no deterioration in sample quality was observed within this window. As is generally recommended for RNA-based protocols, prompt processing i.e., completing Day 1 steps as soon as possible after sorting/lysis, is nonetheless strongly recommended as a precaution to preserve RNA integrity.

Clarification and RNase I digestion

Timing: ∼1 h

Here, we describe steps for RNase digestion.

• 8.Clarify the lysate by centrifugation at 20,000 × g, 10 min, 4°C.

Note: There will be no cell debris pellet visible.

• 9.Transfer supernatant to a clean nuclease-free 1.5 mL tube.
• 10.
  Add RNase I:

  • a.
    1 μL for 10,000–50,000 cells.
  • b.
    2 μL for 50,001–200,000 cells.

Note: The 1 μl dose is taken unchanged from the parent Ribo-lite protocol.² For inputs above 50,000 cells we increased the dose to 2 μl empirically. The adequacy of both doses was confirmed from the sequencing data, which show the expected 28–30 nt footprint length distribution and 3-nucleotide periodicity (Figure 1). Incubate at 18°C–22°C with gentle mixing for 45 min.

• 11.Stop reaction with 10 μL SUPERase Inhibitor.
• 12.Place samples on ice and proceed to next step.

Figure 1.

Quality metrics confirming recovery of ribosome-protected fragments at 10,000 and 100,000 cell input

(A) Metagene analysis of 28–30 nt reads around annotated start codons at 10,000 cell input. Reads accumulate at the initiation site and show pronounced 3-nucleotide periodicity with a dominant frame (red), consistent with elongating ribosomes.

(B) Metagene around stop codons (10,000 cells), showing in-frame periodicity across the coding sequence and a sharp drop downstream of the stop codon.

(C) Length distribution of aligned reads (10,000 cells), peaking at 28–30 nt as expected for ribosome footprints.

(D) Read density across a metagene transcript (10,000 cells), showing enrichment over the coding sequence (CDS) relative to the 5′ and 3′ UTRs.

(E–H) The corresponding analyses at 100,000-cell input: start-codon metagene (E), stop-codon metagene (F), read-length distribution (G), and transcript read density (H). Across both inputs, the correct footprint length, 3-nucleotide periodicity, start/stop-codon architecture, and CDS enrichment confirm that the recovered reads represent bona fide ribosome-protected fragments, with comparable quality at 10,000 cells relative to a ten-fold higher input.

Sucrose cushion ultracentrifugation

Timing: ∼5 h

Here, we describe steps for sucrose cushion ultracentrifugation.

Before you start, precool the ultracentrifuge. Use Open-Top Thickwall polycarbonate tubes for ultracentrifugation; unlike thinwall tubes, these can be run partially filled.

• 13.
  Layer lysate on top of 700 μL sucrose cushion.

  • a.
    Place a reference mark on the side of the ultracentrifuge tube facing the outside of the rotor to identify the predicted pellet position.

Note: This step is crucial because the pellet will not be visible.

• 14.Ultracentrifuge at 260,000 × g for 4 h at 4°C using an MLA-150 rotor.
• 15.Carefully remove the supernatant on the opposite side of the mark to not disturb the pellet.

Note: If residual volume remains, perform a second aspiration to ensure the complete removal of the sucrose cushion.

• 16.Resuspend ribosome pellet in 50 μL pellet buffer.

Note: Be careful not to touch the presumed position of the pellet.

• 17.
  Proceed to a certified chemical fume hood for the steps involving Trizol:

  • a.
    Add 1 mL Trizol to the sample (to the pellet buffer).
  • b.
    Mix by pipetting carefully and transfer to a new nuclease-free 1.5 mL tube.
  • c.
    Place the samples directly at −80°C or proceed with RNA extraction.

Note: Samples stored in Trizol are stable for less than 2 weeks (minimizing storage time is recommended).

Day 2

RNA extraction for RPF sample

Timing: ∼1 h

Here, we describe steps for RNA extraction of the RPF sample.

• 18.
  Proceed to a certified chemical fume hood for the steps involving Trizol and chloroform.

  Note: Wear appropriate personal protective equipment (nitrile gloves, lab coat, and eye protection/goggles). Work strictly inside the fume hood, as both Trizol (containing phenol) and chloroform are highly toxic, corrosive, and volatile. Avoid inhalation of vapors, contact with skin, and wear splash-proof protection. Collect all liquid and solid waste in designated hazardous phenol/chloroform waste containers.

  • a.
    Let the Trizol sample thaw and add 200 μL chloroform.
  • b.
    Securely cap the tube, then thoroughly mix by shaking.

    Note: This shaking step is relevant and should already start the separation. Significant sample losses can occur if not thoroughly mixed.

  • c.
    Incubate for 2–3 min.
• 19.Centrifuge the sample at 15000 × g for 15 min at 4°C.
• 20.
  Extract aqueous phase.

  Note: The mixture separates into a lower red phenol-chloroform phase (or organic phase), an interphase, and a colorless upper aqueous phase.

  • a.
    Transfer the aqueous phase containing the RNA to a new nuclease-free tube by angling the tube at 45° and pipetting the solution out.
• 21.
  Add 750 μL isopropanol and 20 μg GlycoBlue.

  • a.
    Pipette mix and precipitate 12–18 h at −20°C.

Day 3

RPF size selection

Timing: ∼5–6 h/1 day

Here, we describe steps for RPF size selection via gel purification.

Preparation: Clean all equipment (including gel running chambers) with water and then RNaseZap.

• 22.Pellet RPFs by centrifugation at 12,000 × g for 40 min, 4°C.
• 23.Remove the supernatant immediately (the pellet may be instable).

Note: Do not touch the RNA pellet indicated by GlycoBlue.

• 24.Carefully remove any remaining isopropanol by brief centrifugation, gentle pipetting, and air drying.

Note: Drying times may vary, and this step may take some time. Avoid over-drying the pellet.

• 25.
  Resuspend RPF RNA in 6 μL nuclease-free water.

  • a.
    Add 6 μL Gel Loading Buffer II.
  • b.
    Keep the sample on ice while preparing the gels to allow the RNA pellet to fully dissolve.
• 26.
  Load one sample per gel without a marker to avoid cross-contamination.

  • a.
    Run a separate parallel gel with RNA markers for size reference.
  • b.
    Place the 15% (w/v) polyacrylamide TBE-urea gel into the electrophoresis chamber.
  • c.
    Fill with 1× TBE running buffer.

Note: We prepare TBE using double-distilled water (ddH₂O); for extra caution use nuclease-free water.

• 27.
  Pre-run the 15% TBE-urea gel at 200 V for 30 min (without the sample).

  • a.
    Wash out the wells by pipetting buffer in the wells to remove bubbles or remaining gel pieces.
• 28.Begin by loading the marker gel for size reference, including both a prestained RNA marker (e.g., Dynamarker) and a microRNA marker.
• 29.
  Individually denature each sample at 80°C for 90 seconds, then immediately load one sample per gel.

  • a.
    Run the gels together at 200 V for 65 min.

Note: This is critical because the marker gel needs to run in parallel. This avoids marker-to-sample cross-contamination; a potential risk of differing migration remains, so the light and dark blue dye fronts within the sample lane serve as internal landmarks.

• 30.
  Carefully remove the gels and place them in a 15 cm dish containing running buffer.

  • a.
    Stain the marker gel with SYBR Gold (1:10,000 dilution) for 5 min.
  • b.
    Then use a blue transilluminator (utilize the orange protection shield!) to estimate the appropriate size range for excising the RNA fragment region from the sample gels.

Note: 10,000 cells is the functional lower limit for hematopoietic cells, regardless of the visual appearance of the gel, as the bands at lower inputs may primarily consist of fragmented rRNA rather than viable RPFs.

• 31.Excise the 26–34 nucleotide region from the sample gel based on the position of the blue loading dye and the RNA oligo markers run on the parallel marker gel.
• 32.Stain the gels as well with SYBR gold and confirm correct excision of samples (Figure 2; Figures S1 and S2).

Note: In doubt, we recommend to perform the cut more inclusively given that the size adjustment can be optimized bioinformatically.

• 33.Extract the RNA by placing the gel slices in 400 μL RNA extraction buffer in a fresh nuclease-free 1.5 mL tube.
• 34.Snap-freeze the tubes on dry ice for 20 min, thaw at 37°C for 2 min (or until fully thawed).

Note: Freeze-thaw cycles have been added to improve RNA extraction from the gel due to initial failure with this step. If a pause in the protocol is necessary, samples can be stored at −80°C for up to 48 h before resuming.

• 35.Repeat this step.
• 36.Incubate the samples 12–18 h at 18°C–22°C with gentle shaking.

Figure 2.

Excised sample gel (left, 75.000 and 10.000 cells input) and marker gel (right) after SYBR gold staining

The excision window (in red) was defined using a visible light blue marker at the upper boundary and a lower boundary placed approximately halfway between the light blue and dark blue markers. Pre-stained RNA and micro RNA markers were included for orientation. Shorter RNA fragments can be excluded during bioinformatic processing.

Day 4

Isopropanol precipitation

Timing: <1 h

Here, we describe steps for isopropanol precipitation.

• 37.
  Collect the eluate to a clean nuclease-free 1.5 mL tube.

  • a.
    Add 500 μL isopropanol and 20 μg GlycoBlue (1.3 μL using 15 mg/mL concentration).
• 38.Precipitate 12–18 h at −20°C.

Day 5

RPF library preparation and sequencing

Timing: 1 day

Here, we describe steps for library preparation of the RPF sample.

• 39.Pellet RPFs by centrifugation at 12,000 × g for 40 min, 4°C.
• 40.Remove the supernatant immediately (the pellet may be instable).

Note: Do not touch the RNA pellet indicated by glycogen blue.

• 41.Carefully remove any remaining isopropanol by brief centrifugation, gentle pipetting, and air drying.

Note: Drying times may vary, and this step may take some time. Avoid over-drying the pellet.

• 42.
  Resuspend RPF RNA in 6 μL nuclease-free water.

  • a.
    Keep the sample on ice while preparing the library kit to allow the RNA pellet to fully dissolve.

Note: Critical step, resuspending small samples may take some time to fully dissolve.

• 43.Measure small RNA concentration by Qubit.
• 44.Prepare libraries using the D-Plex Small RNA-seq Kit and purify the libraries using AMPure XP beads (1.5× beads).

Note: A successful preparation for cell inputs ranging from 10,000 to 100,000 leukemia cells should yield more than 2 ng of small RNA, sufficient for amplification using 13 PCR cycles with unique dual index library preparation.

• 45.Pool barcoded libraries equimolar for sequencing (see Figure 3 for representative Tapestation results).

Note: Libraries should be normalized to 2–4 nM per library (following standard Illumina recommendations) prior to pooling. Consult with your sequencing core facility for optimal results, as certain sequencing platforms may not perform well with small RNA insert libraries that are shorter than standard genomic fragments and are inherently low-diversity due to residual rRNA content (Figure 4). Co-pooling RPF libraries with higher-diversity libraries (e.g., standard RNA-seq) may help mitigate low-diversity sequencing issues. A higher PhiX spike-in may also be beneficial, although this has not been tested by us.

• 46.Sequence samples according to the recommendation of the sequencing core facility or use an Illumina sequencing platform (e.g., NovaSeq 6000) with paired-end 2 × 100 sequencing aiming for 30–40 million reads/sample.

Note: Only Read 1 (representing the sense strand) is used for analysis, as RPFs are typically short (26–34 nt) and fully captured within Read 1 in paired-end sequencing.

Figure 3.

Tapestation traces of RPF libraries

Representative Tapestation electropherograms of a small RNA-seq library prepared from 10,000 and 100,000 leukemia cells, following library preparation with the D-Plex Small RNA-seq Kit. Expected is a single dominant library peak (average size 208–209 bp).

Figure 4.

Read distribution and alignment statistics of Ribo-lite libraries

(A) Relative distribution of sequencing reads across murine AML samples. Primary alignments (yellow) represent usable ribosome-protected fragments (RPFs). The predominant fraction of ribosomal RNA (rRNA, blue) is a characteristic result of the low-input Ribo-lite protocol, which omits rRNA depletion to maximize recovery from rare cell populations.

(B) Absolute read counts per sample, illustrating the total sequencing depth and the respective proportions of trimmed (purple), unmapped or multimapped (green), and aligned reads.

Day 6

RNA extraction for transcriptome sample

Timing: 1 day

Here, we describe steps for RNA extraction of the transcriptome sample.

The transcriptome fraction is 2 μl of the initial 20 μl lysate (10% of the initial sample), taken at step 5 before the lysis volume is increased.

• 47.
  Proceed to a certified chemical fume hood for the steps involving Trizol.

  • a.
    Let the Trizol sample thaw and add 200 μL chloroform.
  • b.
    Securely cap the tube.
  • c.
    Thoroughly mix by shaking.
  • d.
    Incubate for 2–3 min.
• 48.Centrifuge the sample at 12000 × g for 15 min at 4°C.
• 49.
  Extract aqueous phase.

  • a.
    The mixture separates into a lower red phenol-chloroform, an interphase, and a colorless upper aqueous phase.
  • b.
    Transfer the aqueous phase containing the RNA to a new nuclease-free tube by angling the tube at 45° and pipetting the solution out.
• 50.
  Add 500 μL isopropanol and 20 μg glycogen blue.

  • a.
    Pipette mix and precipitate for a minimum of 10 min on ice.
• 51.Centrifuge the sample at 12000 × g for 10 min at 4°C.
• 52.Remove the supernatant immediately (the pellet may be instable).

Note: Do not touch the RNA pellet indicated by glycogen blue.

• 53.
  Resuspend the pellet in 75% ethanol.

  • a.
    Mix by flicking the tube.
• 54.Centrifuge the sample at 7500 × g for 5 min at 4°C.
• 55.Remove the supernatant immediately (the pellet may be instable).

Note: Do not touch the RNA pellet indicated by glycogen blue.

• 56.Carefully remove any remaining ethanol by brief centrifugation, gentle pipetting, and air drying for 5 min.

Note: Avoid over-drying the pellet.

• 57.
  Resuspend the RNA in 6 μL nuclease-free water.

  • a.
    Keep the sample on ice while preparing the library kit to allow the RNA pellet to fully dissolve.
• 58.Measure RNA concentration by Qubit (alternatively by Tapestation).

Note: This yielded 0.7–2.1 ng/μL at RINe 7.0–9.8, corresponding to approximately 4–12 ng total RNA per sample (from 10,000 to 100,000 cells). Normalize to 3 ng input and amplify with 9 PCR cycles; for other input amounts, follow the cycle number recommended by the kit manual.

• 59.Proceed with SMART-Seq v4 Ultra Low Input RNA Kit as recommended by the manufacturer.

Note: Do not pool the samples!

Day 7

cDNA library preparation

Timing: <1 day

Here, we describe steps for cDNA preparation of the transcriptome sample.

• 60.Measure cDNA concentrations from step 59 by Qubit (alternatively by Tapestation).
• 61.
  Proceed with the samples for cDNA library preparation using the Nextera XT kit.

  • a.
    purify the libraries using AMPure XP beads.

Note: We recommend 2.3× bead ratio.

• 62.Measure cDNA concentrations by Qubit (alternatively by Tapestation).
• 63.Pool the libraries for sequencing according to the sequencing core facility (or pool with the RPF samples) and perform paired-end 2 × 100 sequencing aiming for 20–30 million reads/sample.

Data analysis

We developed a pipeline to analyze the data and made it available under https://github.com/emayerhofer/riboseq_leukemia. This repository contains two scripts: process_rpf.sh, for extracting gene counts of the RPF fraction, and process_transcriptome.sh, for extracting gene counts of the transcriptome. The two processes are different due to underlying preparation methods.

The pipeline relies on a Linux-based high-performance cluster with installed command-line tools specified in environment.yml, at least 32 GB of memory and 4 CPU cores, and at least 35GB of free disk space for the genome index creation and approximately 20 GB of temporary storage per sample (for intermediate aligned files, trimmed FASTQs, and output matrices). The estimated runtime is 90 min per sample (depending on the number of cores and memory available, but the first time will take substantially longer (+60 min) for genome index creation.

mRNA-seq analysis

Transcriptome RNASeq data is processed according to the recommendation of the manufacturer (Smart-Seq v4 Ultra Low Input RNA Kit, Takara Bio USA) and the original Ribo-lite publication.² Reads are trimmed with cutadapt¹¹ and then aligned using STAR.¹⁶ Read distribution is calculated using read_distribution.py from RSeQC package,⁹ and metaplots are generated with the python package RiboCode.¹⁰ Finally, featureCounts⁶ is used to detect exons, and reads are summarized per gene.

Ribo-seq/RPF analysis

Ribo-seq data is processed according to the recommendation of the manufacturer (Diagenode user guide C05030001) and the original Ribo-lite publication.² The UMI is copied from the read to the read identifier with fumi_tools (https://ccb-gitlab.cs.uni-saarland.de/tobias/fumi_tools) to allow subsequent deduplication. Trimming is performed with cutadapt¹¹ and reads are aligned to mouse using bowtie2.¹² Reads that can’t be aligned to rRNA are from the RPF fraction and are thus aligned to the mouse genome reference genome using STAR.¹⁶ Deduplication of reads is performed (to remove PCR duplicates) using fumi_tools (https://ccb-gitlab.cs.uni-saarland.de/tobias/fumi_tools). Read distribution is then calculated using read_distribution.py from RSeQC⁹ package, and metaplots are generated by RiboCode.¹⁰ Finally, featureCounts⁶ is used to detect and reads are summarized per gene.

Expected outcomes

Ribosome profiling is a powerful technique that captures mRNAs actively engaged in translation by sequencing RPFs. Classical Ribo-seq, however, requires large input material (>10⁷ cells), making it unsuitable for rare cell populations. The Ribo-lite protocol adapts Ribo-seq for low-input settings,² enabling high-quality translatome data from as few as 10⁴ hematopoietic cells (three orders of magnitude less than conventional protocols). This is achieved by optimizing RNAse digestion conditions, omitting rRNA depletion, and using a single-tube ultrasensitive small RNA library preparation strategy.² These adjustments minimize sample loss and contamination. Due to the omission of rRNA depletion, a step specifically bypassed to maximize recovery from rare cell populations, the majority of sequencing reads (average 65%) map to ribosomal RNA. Primary alignments, representing the usable RPF reads, account for approximately 4% to 10% of total reads (Figure 4). While this fraction is lower than in conventional high-input protocols, the Ribo-lite method consistently yields sufficient coverage to identify actively translated mRNAs (of over 9500 protein coding genes, Figures S3) and maintains high data fidelity. To quantify complexity, we calculated the unique read fraction (N deduplicated/N mapped) for RPF footprints mapped to CDS regions. Across all sample inputs including the 10,000-cell input we observed a high library complexity ratio of ∼0.83 (83% unique reads), indicating low PCR duplication rates (∼17%) (Figure S3).

Importantly, Ribo-lite preserves the hallmarks of conventional Ribo-seq, including three-nucleotide periodicity and frame enrichment as characteristic indicators of ribosomal engagement. The periodicity arises because ribosomes decode mRNAs in codon-by-codon steps during elongation, resulting in a precise 3-nt signature in sequencing reads. Codon-level resolution is achieved by isolating 26–34 nucleotide fragments that predominantly map to coding sequences (Figure 5).

Figure 5.

Read length distribution and 3-nt periodicity of ribosome footprints

Expected length of translatome reads (A) and analysis of periodic reads (B).

Smaller reads including fragments 18–25 nt have also been shown to exhibit periodicity and may result from the library preparation method in this protocol.^1,2

The method generates two libraries from each sample the RPF library (representing the translatome) and the mRNA library (representing the transcriptome). This paired dataset allows the calculation of translation efficiency (TE) for each gene, defined as the ratio of Ribo-seq signal to RNA-seq signal (Figure 6). TE provides a direct measure of how effectively a transcript is being translated, adding a layer of functional interpretation to transcriptomic data.

Figure 6.

Ribo-seq with paired generation of transcriptome and translatome for calculation of translation efficiency

RNA-seq measures total mRNA abundance, which lacks the three-nucleotide (3 nt) periodicity characteristic of active translation, Ribo-seq captures ribosome-protected fragments (RPFs); their strong 3 nt periodicity marks coding regions actively engaged by ribosomes.

While transcription is relatively permissive and low-cost, translation is the most energy-consuming cellular process.¹⁷ As such, translatome profiling offers a more accurate snapshot of the cell’s immediate functional state and resource allocation. It reveals which transcripts are prioritized for critical protein synthesis in settings such as stress, metabolic restriction, or cell fate decisions. This is particularly relevant in leukemia, where protein synthesis is tightly linked to chemoresistance and outgrowth of persister cells.

Unlike proteomics, which reflects steady-state protein levels and often requires >10⁶ cells, the translatome captures dynamic changes in gene regulation on shorter timescales and with significantly lower input. This makes Ribo-lite ideally suited for interrogating rare cell types such as stem cells and specialized populations.

The resulting data can be analyzed using established computational pipelines, including Principal Component Analysis (PCA), Differential gene expression (transcriptome and translatome), Translation efficiency analysis, Pathway and gene set enrichment analyses (e.g., GO, KEGG, Reactome) and Codon occupancy. Overall, this protocol enables robust and quantitative ribosome profiling from low-input material and opens a new window into translational regulation in rare and clinically relevant AML cell populations.

Limitations

We validated the protocol down to 10,000 cells (Figure 1), which we therefore recommend as the lower input limit; inputs below this threshold (we tested 1,000 cell input) yielded reduced quality-control metrics including lower fraction of CDS-mapping reads. Lysis efficiency cannot be verified at these inputs. Input RNA quality critically influences footprint integrity. Finally, all data presented here were generated from freshly isolated cells. We have not tested cryopreserved material; because freeze–thaw cycles could in principle affect ribosome integrity, we recommend that performance be validated before applying the protocol to cryopreserved samples. This workflow offers no quality control checkpoint before sequencing. All data presented here were generated in murine AML cells; validation in primary human AML is not shown. Users applying the protocol to human material should note that the analysis pipeline aligns against mouse rRNA and genome references, which must be substituted, and that cell size and lysis efficiency may differ. We recommend confirming 3-nucleotide periodicity and footprint length distribution as a quality check.

Troubleshooting

Problem 1

Smearing or missing bands on gel (related to step 31). Possible cause: incomplete sample denaturation, gel not prerun.

Potential solution

Ensure sample is denatured at 80°C for 90 s; pre-run gel for 30 min in 1×TBE.

Problem 2

Poor library quality (related to step 44). Possible cause: RNA degradation; reagent degradation; freeze-thaw effects.

Potential solution

Use fresh reagents; avoid multiple freeze-thaw cycles; validate RNA integrity before library preparation.

Problem 3

Low RPF yield (related to steps 4–11). Possible cause: Incomplete cell lysis or RNase I digestion; degraded RNA.

Potential solution

Lysis efficiency cannot be verified; ensure the described conditions are met (correct buffer volumes, freshly added components, full incubation times) and extend the lysis incubation if incomplete lysis is suspected; optimize RNase I concentration and digestion time; increase input if possible; Assess input RNA integrity at the start of the workflow by TapeStation (RNA ScreenTape) to identify degraded starting material.

Problem 4

Poor RPF periodicity (related to steps 4–11 and step 31). Possible cause: Impure RNA or poor excision of RPF band.

Potential solution

Check input quality of RNA (e.g., using Tapestation). Optimize RNase I concentration and digestion time; increase input if possible. Ensure precise excision of the 26–34 nt region using properly aligned marker gel.

Problem 5

Low complexity or PCR duplicates (related to step 44). Possible cause: Overamplification or too low input.

Potential solution

Reduce PCR cycles if yield allows. The recommended PCR cycles in the library preparation protocol (SMART-seq v4 Ultra Low Input RNA protocol for RNA fraction or D-Plex Small RNA protocol for RPF fraction) are a good starting point.

Pro-tip: For an additional layer of security, users may opt for low-binding tips.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, David T. Scadden (dscadden@mgh.harvard.edu).

Technical contact

Technical questions on executing this protocol should be directed to and will be answered by the technical contact, Christina Mayerhofer (christina.mayerhofer@uniklinik-freiburg.de).

Materials availability

This study did not generate new reagents.

Data and code availability

This study did not generate new data. The code is available under https://github.com/emayerhofer/riboseq_leukemia (https://doi.org/10.5281/zenodo.22083026).

Acknowledgments

C.M. was supported by a Walter-Benjamin fellowship from the German Research Foundation (Deutsche Forschungsgemeinschaft [DFG], GZ: MA 9452/1-1 and MA 9452/2-1) and the DKTK Junior Clinical Cooperation Unit (ID: FR07-SOB-TRANSMETA-MAY) funded by the German Cancer Consortium. T.K. was supported by the Swedish Research Council’s International Postdoc grant. D.T.S. was supported by the Gerald and Darlene Jordan Professorship, the Edward P. Evans Foundation, and by NCI P50CA206963.

Author contributions

C.M. designed and conducted experiments and drafted the manuscript. E.M. performed bioinformatic analysis and wrote the code for the pipeline. D.L. and T.K. conducted the experiments. T.F. and D.T.S. supervised the work and proofread the manuscript.

Declaration of interests

D.T.S. is a founder, director, and stockholder of Lightning Biotherapeutics. He is a director of Agios Pharmaceuticals, Editas Medicine, and Carisma Therapeutics, and a founder and stockholder of Fate Therapeutics and Stratus Therapeutics. E.M. is an employee of Regeneron Pharmaceuticals. The current address for T.K. is Tessera Therapeutics, Cambridge, MA. D.T.S. has received consulting fees from VcanBio.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the authors used ChatGPT and Claude in order to refine the language and improve overall clarity. After using this tool/service, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication.