dsRNA Impurities: Pyroptosis, Not Ferroptosis

Research Question

When double-stranded RNA (dsRNA) accompanies an mRNA treatment, which cellular outcomes are actually supported, and what purification comparisons matter before researchers label the response as apoptosis or ferroptosis? Liu et al.’s review is useful here because it connects dsRNA impurities to innate immune sensing, translational suppression, and one defined inflammatory death pathway. It does not, however, provide direct evidence that an mRNA preparation causes apoptosis or ferroptosis. The article is therefore best read as a mechanistic background and purification-comparison review, not as experimental validation of a death-inducing reagent or delivery product. Source review.

Study Design and Methods

This is a review article rather than a new cell or animal study. The authors survey mRNA therapeutic development and synthesize published information on impurities generated during in vitro transcription, dsRNA detection, post-transcriptional purification, and strategies intended to reduce dsRNA formation during synthesis. The reviewed purification approaches include reversed-phase high-performance liquid chromatography (RP-HPLC), cellulose-based purification, RNase III treatment, oligo(dT) affinity chromatography, anion-exchange chromatography, size-exclusion chromatography, asymmetric flow field-flow fractionation, tangential-flow filtration, selective precipitation, and silica-based purification.

The review also summarizes reported cellular and in vivo observations from earlier studies, including changes in protein expression, interferon or inflammatory-factor induction, and immunogenicity. Those observations are not experiments performed by Liu et al.; the paper states that no datasets were generated or analyzed. Its pathway discussion is consequently a synthesis of prior evidence, and its purification numbers should not be treated as a single head-to-head manufacturing experiment.

Main Findings

1. The strongest death-pathway evidence is for pyroptosis

The review describes long dsRNA as a trigger of human NLRP1 inflammasome activation. In the pathway summarized by the authors, NLRP1 activation leads to caspase-1 activation; caspase-1 processes interleukin-1β, interleukin-18, and gasdermin D (GSDMD); and cleaved GSDMD forms plasma-membrane pores, producing pyroptosis. This is pathway-level evidence for an inflammatory, gasdermin-associated death program, not for apoptosis or ferroptosis.

The distinction is important for interpreting mRNA-treatment experiments. Caspase-1 activity should not automatically be treated as evidence of apoptotic signaling, because the review places it in the NLRP1 inflammasome–GSDMD pyroptosis pathway. Conversely, a loss of viability after RNA transfection would not identify pyroptosis, apoptosis, or ferroptosis without pathway evidence. The reviewed article reports no apoptosis-specific comparison, no caspase-dependent versus caspase-independent apoptosis experiment, and no rescue experiment that establishes a causal death mechanism.

2. Translation inhibition and RNA decay are not equivalent to cell death

The review separates dsRNA-associated growth inhibition from the NLRP1 death pathway. It describes protein kinase R (PKR) binding to dsRNA and phosphorylating eukaryotic initiation factor 2α, thereby suppressing cap-dependent translation. It also describes oligoadenylate synthetases activating RNase L, which degrades RNA as part of the antiviral response. These processes can reduce protein production or cell growth, but the review does not present them as sufficient evidence of a programmed cell-death mechanism.

This distinction matters particularly when the intended mRNA cargo is used to measure protein expression. A preparation containing dsRNA may appear weak because innate immune signaling suppresses translation, not because the encoded protein is intrinsically inactive. The same impurity may also alter inflammatory readouts independently of the intended mRNA activity. Thus, reduced reporter or therapeutic-protein output should be interpreted alongside dsRNA measurements and pathway controls rather than converted directly into an apoptosis or ferroptosis claim.

3. Purification methods differ in what they remove

The review’s most consequential comparison is between methods that selectively reduce dsRNA and methods that mainly remove other contaminants. Cellulose-based purification is summarized as removing approximately 90% or more of detectable dsRNA, with reported mRNA recovery of 65–90%. A wood-derived macroporous cellulose material is described as removing 98% of dsRNA within five minutes and reducing an in vivo immunogenic response in the cited work. In a separate comparison involving self-amplifying mRNA, cellulose-based purification produced higher protein expression and lower dsRNA levels than commonly used silica-based purification.

That comparison does not establish apoptosis or ferroptosis. It shows that impurity burden can influence translation and immune activation. The review also emphasizes that cellulose is not a complete purification solution: template DNA, RNA polymerase, unincorporated nucleotides, truncated or uncapped RNA, and other contaminants may remain in the flow-through with the desired mRNA. Therefore, an apparent benefit after cellulose treatment cannot automatically be attributed only to dsRNA removal unless the other impurities are characterized.

RP-HPLC is presented as a higher-resolution option that can separate single-stranded RNA from dsRNA and is compatible with modified RNA constructs. Its tradeoffs include operating cost, solvent handling, and scale-up complexity. The review summarizes earlier work in which RP-HPLC-purified mRNA showed markedly greater protein expression and lower interferon or inflammatory-factor induction in cell assays, but these are results from cited studies rather than new experiments in this review.

Other methods have different blind spots. Oligo(dT) affinity chromatography captures poly(A)-containing mRNA but can leave dsRNA because it selects by poly(A) hybridization rather than dsRNA structure. Anion-exchange chromatography can provide scalable mRNA purification but may not remove dsRNA completely and can require desalting. Size-exclusion chromatography separates short transcripts and nucleotides from larger mRNA, yet dsRNA of similar size can overlap with the product. RNase III directly targets dsRNA, but incomplete digestion may leave immunostimulatory fragments, structured regions of the intended RNA could theoretically be affected, and the enzyme and cleavage products require removal.

4. Tangential-flow filtration can leave the key confounder behind

Tangential-flow filtration (TFF) is described as useful for concentration and buffer exchange. The review summarizes examples in which membrane filtration removed small molecules such as nucleotides and salts, reduced the protein-to-RNA ratio, and preserved mRNA integrity. A vibration-assisted single-pass format is also reported to concentrate mRNA and reduce nucleoside triphosphates and protein content.

However, the review explicitly states that TFF is ineffective for dsRNA removal because dsRNA is often similar in size to the intended mRNA. This creates a critical negative comparison: a workflow can improve chemical cleanliness or reduce protein contamination while retaining an impurity capable of activating PKR, RIG-I-like sensors, MDA5, TLR3, or NLRP1. TFF should therefore not be treated as evidence that a preparation is free of the dsRNA-related confounder relevant to inflammatory signaling and cell-death assays.

5. What is missing before apoptosis or ferroptosis can be assigned?

The review notes that J2- or K2-antibody assays detect dsRNA duplexes of approximately 40 base pairs or longer, but it also states that universally accepted dsRNA thresholds for immune activation, translation suppression, or regulatory acceptance have not been established. Effects depend on dsRNA length and heterogeneity, sequence, dose, formulation, administration route, and the targeted cell or tissue.

For apoptosis research, the paper supplies no direct evidence for either caspase-dependent or caspase-independent apoptosis. For ferroptosis research, it supplies no evidence of iron dependence, lipid-peroxidation dependence, or ferroptosis-specific rescue. A viability decrease, translational arrest, inflammatory cytokine response, or generic caspase signal would therefore be insufficient on its own. Mechanistic attribution should require pathway evidence and, where feasible, a rescue experiment that distinguishes the proposed mechanism from dsRNA-driven pyroptosis or nonlethal translational stress.

Limitations

This review does not perform a standardized head-to-head comparison across identical mRNA sequences, doses, cell types, time points, formulations, or analytical thresholds. Reported recovery and impurity-removal values come from different cited studies and should not be ranked as universal performance specifications. The reviewed material also does not establish how purification effects vary across primary cells, immune cells, or tumor models, and it does not provide durability data for a specific therapeutic preparation. The LNP discussion is background on delivery components and is not an LNP formulation comparison. Most importantly for apoptosis and ferroptosis researchers, the paper contains no direct apoptosis or ferroptosis experiments and no rescue data for either mechanism.

Source

Liu J-H, Zang J, Xu J-R, et al. The current landscape of mRNA therapy and the strategies for mRNA purification and dsRNA removal. Journal of Biomedical Science. 2026;33:79. https://doi.org/10.1186/s12929-026-01277-4