A DNA-guided prokaryotic argonaute enables programmable RNA cleavage for sequencing and quality control of in vitro-transcribed RNAs Authors: Lizhi Liu, Eric A Hunt, Eric J Wolf, Ivan R Corrêa, Erbay Yigit, Sebastian Grünberg Published: 2026 ABSTRACT AbstractArgonaute (Ago) endonucleases are ubiquitous across all domains of life and, like Cas nucleases, can be programmed by guide oligonucleotides to target and cleave specific nucleic acid sequences. However, unlike Cas enzymes, Agos are not restricted by sequence requirements such as protospacer adjacent motifs (PAMs), enhancing their utility as molecular tools. This study demonstrates diverse applications of the DNA-guided RNA endonuclease Mucilaginibacter paludis Argonaute for in vitro mRNA processing, e.g., 3′ end polishing, and for mRNA analysis in combination with liquid chromatography-tandem mass spectrometry (LC-MS/MS). Applications of the latter include the comprehensive assessment of 5′ capping efficiency, 3′ poly(A) tail length and composition, one-pot multiplexed cleavage for RNA sequencing, modification detection, and fingerprinting.Keywords: MT: RNA/DNA Editing, argonaute, Mpa Ago, programmable RNA cleavage, LC-MS/MS, DNA-guided endoribonuclease, mRNA sequencing, RNA fingerprinting, 5′ cap analysis, poly(A) tail analysis, mRNA quality control Introduction Like CRISPR-Cas nucleases, Agos are conserved endonucleases that utilize nucleic acid guides to recognize, bind, and cleave their substrates. However, unlike Cas nucleases, Ago proteins are present across all domains of life.1,2,3 In eukaryotes, most Agos are loaded with small viral RNAs, endogenous microRNAs, or PIWI-interacting RNAs (piRNAs), functioning in RNA interference pathways as components of the RNA-induced silencing complex.4,5,6,7,8 They play crucial roles in antiviral defense, post-transcriptional regulation of gene expression, and suppression of transposon activity.9,10,11 Prokaryotic Agos (pAgos) are classified into long and short based on their domain architecture.2,3 Long pAgos consist of six domains (N-terminal, L1, PAZ, L2, MID, and PIWI) and share structural similarities with eukaryotic Agos (eAgos). However, some long pAgos are catalytically inactive due to disruptions in their catalytic tetrad, DEDX (X = D/H/N/K), which coordinates divalent metal cations, often Mg2+, that are crucial for activity.1,2,12,13,14 Short pAgos contain only the MID and PIWI domains and lack the ability to cut nucleic acids due to mutations in the catalytic tetrad.15 Short pAgos and catalytically inactive long pAgos often associate with other effector proteins that function in bacterial defense systems.16 In contrast, active long pAgos have been reported to be directly involved in cellular defense against foreign genetic elements as well as in genome replication and recombination.12,17,18,19,20,21 Active pAgos load short guides to form a nucleoprotein complex that subsequently binds and cleaves complementary target sequences between guide nucleotides 10 and 11, resulting in cleavage products with 3′-OH and 5′-phosphate ends.22 While only few eAgos have been reported to use DNA guides to cleave DNA and/or RNA targets,23,24,25 most bind RNA guides to cleave RNA targets. In contrast, most known active long pAgos utilize short DNA guides to target DNA sequences.17,18,26,27,28,29,30,31,32,33 A small subset of pAgos has been shown to use RNA guides to bind, and in some cases cleave, DNA,14,34,35,36,37 with even fewer reported to use RNA guides to target RNA.38,39 Recently, several pAgos have been described that use DNA guides to cleave either exclusively RNA40,41 or both RNA and DNA, with varying substrate preferences and cleavage efficiencies.31 The ability to direct endonuclease activity to a specific location with short nucleic acid guides makes pAgos highly versatile and valuable molecular tools with many applications in biotechnology. When compared with Cas nucleases, the lack of sequence restrictions such as protospacer adjacent motifs (PAMs) and the predominant use of short DNA guides versus RNA guides make pAgos a versatile and economical alternative, especially for in vitro applications. DNA-targeting pAgos are already used in the diagnosis of pathogens and disease-related rare mutations, DNA assembly, super-resolution microscopy, and the detection of single- and double-stranded DNA sequences.42 In contrast, applications that utilize pAgos to specifically target RNA have been limited to secondary structure probing,30 sensing of RNA modifications, and single-nucleotide mismatches.41 No applications for pAgos to access therapeutic and other functional RNAs have been reported yet. In this work, we describe RNA-specific applications of Mucilaginibacter paludis Argonaute ([Mpa Ago]; also referred to as Mbp Ago in previous reports), a DNA-guided pAgo that preferentially targets RNA substrates.40 We demonstrate that Mpa Ago can be used to efficiently cleave a target RNA at 24 sites simultaneously in a multiplex one-pot reaction, allowing for subsequent RNA sequencing using intact mass spectrometry with up to >90% sequence coverage. Custom guide mixtures generating characteristic RNA fragments enable RNA fingerprinting—a fast and unambiguous way to detect and identify RNAs by liquid chromatography. We also show that Mpa Ago can be used in 5′ cap analysis of mRNAs, allowing for the precise determination and relative quantification of the cap incorporation in a workflow that can be adapted to virtually any 5′ untranslated region (UTR) sequence and can be performed in under 1 h. Last, we demonstrate that Mpa Ago can be guided to and specifically cleave a sequence embedded in the poly(A) tail of an in vitro-transcribed (IVT) mRNA to generate mRNA populations with homogenized 3′ ends, removing non-templated additions introduced by RNA polymerases. This also allows for the accurate analysis of the poly(A) tail length and identification of unwanted potential extensions. Results Mpa Ago efficiently and specifically cleaves mRNA in vitro To assess the potential of pAgos to manipulate therapeutically relevant RNA substrates, we sought to determine the specificity and cleavage efficiency of mRNA substrates by Mpa Ago. Mpa Ago has previously been described as a programmable endoribonuclease with strong guide-mediated substrate specificity that can cleave RNA substrates of up to 352 nucleotides (nt) in length.40 We tested site-specific RNA cleavage by Mpa Ago on two different IVT mRNAs: pSG90 IVT RNA, which represents a shortened version of the replication-competent Sindbis virus genome43 (1,678 nt), and Fluc (Firefly luciferase) IVT RNA (1,766 nt). We designed 16-nt long 5′-phosphorylated guides targeting unique sequences in the pSG90 and Fluc IVT RNAs, spaced 100 or 72 nt apart throughout the complete mRNA sequences. Cleavage of pSG90 RNA by Mpa Ago using individual guides in separate reactions showed overall robust RNA cleavage, with guide-dependent variation in efficiency relative to the no-guide control (Figure S1A). In contrast, Mpa Ago cleaved Fluc RNA more efficiently, with most guides producing near-complete substrate cleavage (Figure S1B). Notably, the trends in cleavage efficiency across guide-defined sites were reproducible, indicating that the observed trends are robust and not due to experimental variability. To challenge Mpa Ago further, we tested cleavage of a uridine-depleted version of the human erythropoietin (Epo) IVT RNA with a GC content (excluding the poly(A) tail) of 59% (pSG95; Figure S2). The ability to cleave GC-rich substrates is a desirable property, as uridine depletion and/or modification has been shown to reduce immunogenicity and is commonly applied to RNA therapeutics design.44 All but one guide enabled cleavage with efficiencies close to those observed in the other two tested mRNAs; the outlier, guide 95.2, reproducibly only cleaved ∼50% of the substrate, possibly due to local secondary structures that were not fully resolved under the conditions tested. To expand on the initial characterization of Mpa Ago by Li et al.,40 we tested the efficiency of Mpa Ago to cleave long RNAs (>1,000 nt) at various temperatures and incubation times (Figure S3A). We observed very little cleavage of the 1,678-nt pSG90 IVT RNA at 4°C even after a 16-h incubation (Figure S3B). Increasing the temperature to 24°C and 37°C significantly increased the cleavage efficiency, with much of the substrate being cut after 30-min incubation at 37°C. Cleavage by Mpa Ago was highly effective at 50°C, with most of the substrate completely cleaved after 10 min. Importantly, both Mpa Ago and a guide were required for specific cleavage as neither Mpa Ago nor the guide alone induced cleavage of the substrate RNA when incubated for up to 16 h (Figure S3B). We consistently observed a background RNA smear surrounding the uncleaved RNA substrates in all reactions containing long IVT RNA. This background signal appeared largely independent of incubation temperature, reaction time, or the presence or absence of Mpa Ago or guide DNAs. We speculate that it arises from incompletely denatured IVT RNA species. Additional Mpa Ago- and guide-independent degradation of the RNA substrate was observed at 37°C or 50°C when incubation times exceeded 4 h (Figure S3), likely resulting from Mg2+-catalyzed phosphodiester bond hydrolysis.45 Incubation of Mpa Ago with Fluc-targeting guides and pSG90 RNA did not yield detectable cleavage in a PAGE assay (Figure S4), further supporting the specificity of Mpa Ago-mediated cleavage. Although the resolution of the PAGE assay and detection limits of the SYBR Gold staining cannot fully exclude low-abundance off-target cleavage, the data suggest that any such events occur only at low levels. Previous work reported that guide length may affect cleavage efficiency on short RNA substrates.40 We repeated cleavage of Fluc IVT RNA with Mpa Ago using 18-nt DNA guides and found similar cleavage efficiencies compared to that using 16-nt guides (Figure S5). We next tested if the DNA guides could be efficiently removed using E. coli exonuclease I (Exo I), which would enable effective removal of residual DNA prior to downstream processes. While the theoretical carryover of short 16-nt guides presents a very low immunological risk, as such short single-stranded DNA species are unlikely to activate known cellular DNA sensors, their removal may be necessary to meet purity thresholds for therapeutic RNAs. Exo I is a 3′-5′ exonuclease with strong preference for single-stranded DNA and efficiently degraded the majority of the guides from the pAgo reaction (Figure S6). We hypothesized that any residual DNA guides were bound to Mpa Ago and thereby protected from exonucleolytic degradation. Consistent with this hypothesis, removal of Mpa Ago by proteinase K treatment prior to Exo I digestion resulted in complete removal of the DNA guides (Figure S6, lane 4). Multiplexed cleavage of mRNA After confirming that pAgos specifically and efficiently cleave RNAs longer than 1,000 nt, we next investigated whether Mpa Ago could generate the corresponding cleavage fragments in a single reaction when supplied with a pool of guides targeting multiple positions along an RNA, analogous to yeast Ago (Figure 1A).25 Although multiplexed guide use in pAgo-based workflows has been described previously,46,47,48 these studies focused on DNA cleavage, and direct multiplex targeting of RNA by pAgos has not been reported. Consistent with the single-cleavage reactions shown in Figure S1A, incubating Mpa Ago with two guides (90.1 and 90.2) resulted in specific cleavage of the full-length pSG90 RNA at two individual sites (Figure 1B). The length of the resulting ∼100-nt cleavage fragment (Figure 1B, lane 4) is consistent with the distance between the target sequences of the two guides. To completely cleave the full-length Fluc RNA into unique 72-nt fragments (plus a 29-nt 5′ fragment and an 81-nt 3′ fragment), we combined twenty-four 16-nt DNA guides in a single reaction. Using a ratio of substrate to enzyme to total guide pool of 0.1:1:1, complete cleavage of Fluc IVT RNA was achieved (Figure 1C, experiment 2). However, several fragments of sizes >72 nt were still visible, suggesting that not all guides enabled complete cleavage of their respective target sequences. The addition of 0.5% bovine serum albumin (BSA) to the reaction enhanced cleavage efficiency, collapsing the banding pattern to only three bands besides the unique 5′ and 3′ end fragments: the desired 72-nt fragments, one prominent band running at 144 nt, and one faint fragment with a size of ∼220 nt (Figure 1C, experiment 4). Guide multiplexing with Mpa Ago produces RNA fragments suitable for LC-MS/MS analysis (A) Schematic showing guide multiplexing with Mpa Ago. Guides targeting different regions of an RNA are pooled and used in a single reaction to cleave an RNA into smaller fragments compatible with LC-MS/MS analysis. (B) pSG90 IVT RNA (∼1.7 kb) was subject to Mpa Ago cleavage in the absence of a guide (lane 1), with single 5′ phosphorylated guide (lanes 2 and 3), or with two guides in a single reaction (lane 4), at 50°C for 30 min. The substrate:Ago molar ratio was 1:5, and the Ago:guide molar ratio was 1:5. Cleavage products were analyzed on a denaturing 6% TBE-urea gel stained with SYBR Gold. The resulting short cleavage fragments are indicated by asterisks. The target sites for the two guides were 100 nt apart, producing a new 100-nt fragment (black asterisk) when used together. This fragment migrated faster than the 100-nt marker likely due to different buffer compositions or unresolved secondary structure elements. (C) Fluc IVT RNA (∼1.8 kb) was subjected to Mpa Ago cleavage with a pool of 24 5′ phosphorylated guides, with or without 0.5% BSA or enhancer mix at 50°C for 30 min as indicated above the gel. The substrate:Ago molar ratio was 1:10, and the Ago:guide molar ratio was 1:1. Cleavage products were analyzed on a denaturing 6% TBE-urea gel stained with SYBR Gold. The distance between target sites of adjacent guides was 72 nt, such that complete cleavage would produce fragments of uniform 72-nt length in addition to unique 5′ and 3′ fragments of 29 nt and 81 nt, respectively. Bands ≥72 nt represent incomplete cleavage fragments. (D) Cleavage products from two biological replicates of the reactions shown in (C) were analyzed by LC-MS/MS, and the average sequence coverage of Fluc RNA (%) is shown in a bar graph. Experimental conditions: 1, no guides; 2, a pool of 24 guides, no additives; 3, 24 guides together with 0.5% BSA and the enhancer mix; 4, 24 guides with 0.5% BSA; and 5, 24 guides with the enhancer mix. Consistent with the single-guide experiments, cleavage efficiencies in the multiplexed reactions varied across target sites. RNA fragments migrating at approximately 144 and 220 nt were detected, likely corresponding to fragments that include sites at which cleavage occurred less frequently or did not occur under the conditions tested. We then tested additional supplements to further increase cleavage efficiency. In the absence of BSA, the addition of an 1x enhancer mix containing 0.5 M betaine, 20 mM potassium glutamate, 20% glycerol, and 5% PEG 3350 to the reactions resulted in a similar fragment distribution as did the addition of 0.5% BSA (Figure 1C, experiment 5). Combining the enhancer mix with 0.5% BSA resulted in the complete cleavage of Fluc mRNA into the programmed RNA fragment sizes (Figure 1C, experiment 3). Cleaved fragments from the reactions containing the guide pool (experiments 2–5) were submitted to liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis, indicating average sequence coverages between 79% and 88% (Figures 1D and S7). Interestingly, even the reaction without additives, in which cleavage at certain target sites was reduced, resulted in ∼79% average Fluc mRNA sequence coverage (Figures 1D and S7, experiment 2), indicating that high mapping coverage can be achieved even when cleavage efficiency varies across sites. We next tested 18-nt guides to determine if longer guides could further improve Mpa Ago-mediated cleavage efficiency. While the 16-nt guides were selected to generate uniform 72-nt RNA fragments, the 18-nt guides were designed to generate laddering RNA fragments ranging from 69 to 81 nt and thus simplify oligonucleotide analysis by LC-MS/MS. Under conditions similar to those used with the 16-nt guides, Fluc and pSG90 IVT RNAs were incubated with Mpa Ago and their respective 18-nt guide pools (Figure S8A). Consistently, the addition of the enhancer mix significantly increased RNA site-specific cleavage, collapsing most of the RNA fragment sizes to around 60–80 nt (Figure S8A, compare experiments 2 and 3 and experiments 5 and 6). The average sequence coverage for Fluc and pSG90 IVT RNAs ranged from 74% to 95%, based on LC-MS/MS analysis (Figures S8B–S8D). To assess whether an RNA modification commonly used in therapeutic RNAs affects Mpa Ago cleavage efficiency, we incubated it with a Fluc IVT RNA in which all uridines were replaced with N1-methylpseudouridine (m1Ψ) and the same 18-nt guide pool as above. Gel electrophoresis revealed partially cleaved higher-molecular-weight fragments even in the presence of the enhancer mix (Figure 2A), suggesting that m1Ψ may affect the cleavage efficiency. Even though cleavage of the fully m1Ψ-modified Fluc was not complete, the ensemble of identified cleavage products was sufficient to generate RNA mapping coverages of around 75% (without enhancer mix) and 85% (with enhancer mix), as determined by LC-MS/MS analysis (Figures 2B and S9), which were only slightly lower than those of unmodified Fluc (∼78% and 95%, respectively). While this does not provide direct evidence that m1Ψ is cleaved efficiently by Mpa Ago, it suggests that Mpa Ago can be utilized to analyze m1Ψ-modified RNA species. MpaAgo cleaves fully m1Ψ-modified Fluc IVT RNA (A) Fully m1Ψ-modified Fluc IVT RNA was subject to cleavage by Mpa Ago with or without a pool of 22 5′ phosphorylated 18-nt guides, and with or without enhancer mix as indicated above the lanes. The substrate:Ago molar ratio was 1:10, and the Ago:guide molar ratio was 1:1. Cleavage products were analyzed on a denaturing 6% TBE-urea gel stained with SYBR Gold. Experiments 4 and 5 are replicates using RNA substrates synthesized in independent IVT reactions. (B) Sequence coverage in percent derived from the Fluc cleavage products from three biological replicas of the reaction described in (A) as determined by LC-MS/MS. Overall, these data suggest that Mpa Ago can be combined with multiple guides in a single reaction to cleave long RNA substrates into fragments of defined sizes, effectively reducing the complexity of the fragment pool (compared with non-specific RNA chemical cleavage or commonly used RNases with short recognition motifs) and enabling high sequence mapping coverage using oligonucleotide LC-MS/MS analysis. We next investigated whether pooled guide DNAs could be employed to simultaneously target multiple IVT RNA species in a one-pot reaction. The resulting cleavage products then serve as species-specific molecular fingerprints, which can be detected by LC-MS/MS to include or exclude RNA species. In reactions containing Fluc or pSG90 IVT RNA—individually or in combination—we introduced Mpa Ago along with a guide set designed to produce three distinct 72-nt fragments from Fluc and a single 100-nt fragment from pSG90 (Figures 3A and S10). Following cleavage, the reactions were analyzed by LC-MS/MS. In reactions containing only one RNA species, the corresponding target fragments from either Fluc (Figure 3B) or pSG90 (Figure 3C) were successfully enriched and identified over unassigned Ago-independent background fragments. Notably, in the reaction containing both RNA species, specific fragments derived from both Fluc and pSG90 were identified (Figure 3D), indicating that Mpa Ago in combination with multiplexed guide sets enables precise RNA fingerprinting and thus identification of specific species in mixed IVT RNA populations. Mpa Ago enables RNA fingerprinting (A) Schematic representation of an RNA fingerprinting experiment, where an equimolar mix of Fluc and pSG90 IVT RNA substrates was incubated with Mpa Ago and guides that were designed to generate three specific 72-nt cleavage fragments of Fluc (fr. 1–3) and one 100-nt cleavage fragment of pSG90 (fr. 4) RNA. LC-MS/MS analysis of (B) the specific Fluc or (C) pSG90 fragments post-cleavage using the guide pool. A no-enzyme control is shown in the top panel. The mass in kDa of the fragments is depicted on the x axis versus the relative intensity on the y axis. The name of the identified fragment (salmon) is indicated above each bar. Unidentified background fragments are shown in cyan. (D) LC-MS/MS analysis of cleavage fragments from a reaction containing both Fluc and pSG90 IVT RNA substrates. Generation of custom RNA ladders using pAgos The ability of pAgos to cleave a long RNA substrate into fragments of predetermined lengths defined by sequence-specific guides allows for the design and generation of custom RNA ladders containing fragments of desired lengths. To this end, we synthesized a 1,576-nt IVT RNA featuring six unique sequences of lengths ranging from 56 to 514 nt that were separated by a specific 20-nt sequence (5′- … CCUACCGUUAAUGACAGGUC … -3′, Figure S11A). A single guide targeting this 20-nt region enabled simultaneous Mpa Ago-mediated cleavage at sites between the adjacent sequences, releasing the six sequences from the parental RNA, thus generating a set of fragments with pre-defined lengths. Like in the reactions with multiplexed guides, the addition of the enhancer mix increased the cleavage efficiency and resulted in near-complete cleavage of the substrate into a clean and precise RNA ladder, as shown on a denaturing 6% Tris-borate-EDTA (TBE)-urea gel (Figure S11B). RNA 5′ cap analysis using Mpa Ago The N7-methyl guanosine cap structure at the 5′ end of eukaryotic mRNAs is important for nuclear export, stability, and translation of mRNA in situ. Determining the efficiency of mRNA 5′ capping is an essential quality control step in modern mRNA vaccine/therapeutics manufacturing.49,50,51 We sought to investigate whether Mpa Ago could be deployed for mRNA 5′ cap analysis. Capillary electrophoresis (CE) was used to assess cleavage efficiency and determine if Mpa Ago could be programmed to cleave near the 5′ end of an RNA. A 5′-FAM-labeled synthetic 33-nt oligoribonucleotide with a sequence representing the 5′ UTR sequence of Fluc mRNA was incubated with Mpa Ago, and with one of four guides designed to induce cleavage 20 nt downstream of the 5′ end with either a complementary (g1A) or a non-complementary (g1C, g1G, g1T) 5′-most nucleotide (Figure S12A). We found that Mpa Ago efficiently and precisely cleaved at the intended site, regardless of the identity of the 5′ nucleotide, resulting in the expected 20-nt cleavage fragment (Figure S12B). Next, we extended this approach to full-length mRNAs. To facilitate capping efficiency testing in a straightforward and cost-effective gel-based assay, we designed guides that specifically targeted a site either 29 nt (Fluc mRNA) or 30 nt (pSG90 mRNA) downstream of their 5′ end (Figure 4A). In both cases, the capped and uncapped cleavage products showed sufficient difference in mobility on a 15% denaturing polyacrylamide gel to observe successful capping (Figures 4B and 4C). mRNA 5′ cap analysis (A) Schematic of Mpa Ago-facilitated 5′ cap analysis, in which a 5′ phosphorylated guide targets Mpa Ago cleavage to a site close to the 5′ end of a mRNA. Upon substrate release, the resulting fragments are analyzed via LC-MS/MS. (B) 5′ cleavage fragments (29 nt) of reactions containing Fluc mRNA that was either capped in vitro (+) or uncapped (−) were resolved on a 15% TBE-urea gel. Capped fragments migrate slower in the gel compared with uncapped fragments. The 16-nt guide DNA is indicated. (C) 5′ cleavage fragments (30 nt) of reactions containing pSG90 mRNA that was either capped in vitro (+) or uncapped (−) were resolved on a 15% TBE-urea gel. (D and E) Heatmaps showing the relative intensity of detected 5′ fragments starting with a triphosphate (3p), diphosphate (2p), guanosine triphosphate (Gppp), cap 0 (1Me7Gppp), and N7,2′-O-dimethylguanosine triphosphate cap (2Me7Gppp) for Fluc mRNA (D) and pSG90 mRNA (E). To gain more detailed insight into the nature of the cap structure and estimate the capping efficiency, we turned to LC-MS/MS for the analysis of Mpa Ago-mediated cleavage products. We targeted the 5′ Fluc-UTR using three distinct guides designed to generate 29-nt (guide Fluc_A1), 26-nt (guide Fluc_A1.2), and 32-nt (guide D891) 5′ fragments, respectively. We observed precise Fluc 5′ cleavage, producing fragments of the expected lengths, 29-mer (90%), 26-mer (87%), and 32-mer (73%). Some low levels of 5- and 12-nt fragments were also detected, accounting to 6%–20% of the fragments detected (Figure S13A). However, these short fragments were observed regardless of the presence of Mpa Ago or guides and likely represent abortive transcripts carried over from the IVT reaction (Figure S13B). The cleavage of the pSG90 5′ UTR, too, was highly precise, predominantly yielding the intended 30-nt (D831 guide, >92%) or 21-nt (D892 guide, >94%) 5′ fragments (Figure S13C). Similarly, Mpa Ago- and guide-independent 5- and 12-nt fragments were the only other notable fragments detected. As expected, uncapped Fluc IVT RNA predominantly exhibited 5′ triphosphate termini (94%) and fewer diphosphate (5%) ends. In contrast, capped Fluc mRNA was composed exclusively of the Cap-0 (Me7 Gppp) structure (Figure 4D). Similarly, uncapped pSG90 IVT RNA consisted of 94% triphosphate and 4% diphosphate 5′ ends, whereas capped pSG90 mRNA comprised a mixture of 85% Cap-0, 7% diphosphate, and 3% triphosphate species (Figure 4E). These data show that Mpa Ago can be used to successfully and quickly access the capping status and efficiency—either in a simple, cost-effective gel-based assay or by precise LC-MS/MS analysis. Poly(A) tail analysis and homogenization of RNA 3′ ends Most current high-resolution methods to access poly(A) tail length and composition include the release of the poly(A) tail by RNase T1, which cleaves 3′ of guanosines.52 However, when we attempted to use RNase T1 to measure poly(A) tail length of an IVT RNA, we found that RNase T1 also cleaved non-guanosines in a concentration-dependent matter (Figures 5A, S14A, and S14B), necessitating optimization of RNase T1 concentration and digestion conditions for reliable poly(A) tail measurements. Here, we show that for a more detailed analysis of poly(A) extensions and/or measurement of the poly(A) tail length using LC-MS/MS, Mpa Ago can either be programmed to cleave directly upstream of the poly(A) tail (Figure 5B) or a target sequence embedded in the tail as described above (Figure 5C). We utilized a slightly modified version of the poly(A) tail sequence utilized in the BNT162b2 vaccine,53 consisting of a stretch of 22 adenosines, followed by a non-poly(A) sequence (GCATATGACTAAAAAACAT), 36 terminal adenosines, and a 3′-terminal non-poly(A) sequence (GCTATCGGATC) in a 698-nt IVT RNA sequence (pSG120). The length of the poly(A) tail in these experiments was shortened to accommodate for the resolution of the LC-MS/MS and to assure full capture of non-templated additions by T7 RNA polymerase without exceeding the range of the LC-MS/MS. Mpa Ago was complemented with guides targeting a sequence upstream of the poly(A) tail (D834) or a sequence in the poly(A) tail (D758), resulting in 91- or 56-nt 3′ cleavage fragments, respectively (Figure 5D). Cleavage reactions were performed as described above and the cleavage products analyzed by intact LC-MS/MS. Reactions with either guide allowed for the sequence determination of the 3′ cleavage fragments. As expected, in addition to the encoded poly(A) sequence, we observed multiple non-templated addition events by T7 RNA polymerase. In fact, we found that the majority of fragments had one or more additions to the encoded sequence (Figures 5E and 5F). A significant fraction of fragments from both cleavage reactions was found to contain a single additional cytosine, an additional cytosine with one to three adenines, or one to three adenines without cytosine (Tables S3 and S4). These findings agree with what has been reported for non-templated T7 RNA polymerase additions.54 mRNA 3′ end processing by Mpa Ago (A) Bar graph of the summed relative intensity of full-length RNase T1 digestion products (Full-Length), shortened poly(A) products (Cleaved), and unidentified masses from oligo-DT isolations of pSG120 IVT RNA following digestion with varying concentrations of RNase T1. (B) Schematic representation of a reaction where Mpa Ago is used to cleave and thus remove the complete poly(A) tail from an IVT RNA (pSG120) for analysis of the full poly(A) tail by LC-MS/MS. (C) Diagram of mRNA 3′ end trimming. A non-poly(A) sequence island in the poly(A) tail of an mRNA is targeted and cleaved by Mpa Ago, resulting in an mRNA population with homogenized 3′ ends, while the 3′ fragments can be further analyzed for their composition. (D) Sequences of the 3′ poly(A) tail of the pSG120 IVT RNA and two guides (D758 and D834). The respective cleavage sites and the lengths of the resulting 3′ fragments are indicated. (E and F) LC-MS/MS analysis of the 3′ fragments post-cleavage using the D834 and D758 guides, respectively. The mass in kDa of the fragments is depicted on the x axis versus the relative intensity on the y axis. The lengths in nt are indicated above each bar. (G) A 723-nt IVT RNA (pSG95) containing a non-poly(A) island was cleaved at four adjacent sites, using four guides (D758–D761 as indicated above the gel, Figure S9B), and the cleavage products were resolved on a denaturing 6% TBE-urea gel. These undesired 3′ extensions by RNA polymerases are a common by-product of in vitro mRNA synthesis using T7 and other RNA polymerases and were first reported in the 1980s.55,56 Importantly, these 3′ extensions have been reported to negatively impact the efficacy of therapeutic mRNAs.54,57,58 We showed that Mpa Ago could be employed to selectively target and cut a specific sequence incorporated into the poly(A) tail of an mRNA for poly(A) analysis. Thus, we envisioned that Mpa Ago could be used to process IVT RNA (e.g., RNA-therapeutics) to remove unwanted extensions. Such 3′ end trimming would generate an mRNA population with homogenous 3′ ends as shown in Figure 5C. We designed a slightly longer poly(A) tail sequence consisting of a stretch of 30 adenosines, followed by a non-poly(A) sequence (GCATATGACTAAAAAACAT), another stretch of 61 adenosines, and a terminal GCT in a 723-nt IVT RNA sequence (pSG95), to more closely resemble the poly(A) tail length found in the BNT162b2 vaccine. Guides D758–D761 were designed to induce Mpa Ago cleavage in the non-poly(A) sequence (Figure S14C). After cleavage, the reaction products were resolved on a denaturing polyacrylamide gel. We found that guides D758, D760, and D761 enabled the complete cleavage of the full-length substrate at the respective target site, while a small fraction of the substrate remained uncleaved in the reaction with guide D759 (Figure 5G). These data in combination with the LC-MS/MS data above show that Mpa Ago successfully homogenized the 3′ ends of IVT products, efficiently cleaving a target embedded in the poly(A) tail. These data show that Mpa Ago in combination with a guide targeting the poly(A) tail can be used to precisely cleave the 3′ end of mRNAs, thus creating mRNA molecules with homogenous ends. The 3′ cleavage fragments can be further analyzed by LC-MS/MS for poly(A) tail length measurement and assessment of non-templated additions by RNA polymerases. Discussion Although the properties of DNA-targeting pAgos have long been acknowledged and utilized in biotechnological applications, RNA-targeting pAgos have largely been overlooked, likely due to their relative scarcity and limited characterization. In this study, we demonstrate that Mpa Ago can serve as a powerful molecular tool, significantly enhancing or simplifying existing analytical workflows, such as RNA sequencing, fingerprinting, and 5′ cap and poly(A) tail analysis. Mpa Ago also uniquely enables the generation of mRNA populations with homogeneous mRNA 3′ ends. We show that Mpa Ago can be programmed with multiplexed guides to cleave multiple sites within a single target IVT RNA, enabling controlled fragmentation of long transcripts. This capability enables Mpa Ago to address current challenges in analyzing large RNA molecules, including self-amplifying RNAs (saRNAs), where conventional RNase-based digestion generates short (10- to 20-nt) fragments59,60,61,62 that produce complex LC-MS/MS datasets that can pose significant challenges for sequence reconstruction and/or modification detection. Mpa Ago enables cleavage of long RNAs into fragments that are short enough for robust LC-MS/MS detection yet long enough to effectively reduce the complexity of the resulting oligoribonucleotide pool. In initial experiments using 16-nt guides designed to produce uniform 72-nt fragments from Fluc IVT RNA, addition of an enhancer mix substantially increased cleavage efficiency and mapping depth, likely due to molecular crowding that promotes Ago/guide-target interactions. Combining BSA with the enhancer mix slightly reduced coverage, possibly because nonspecific BSA-RNA interactions interfered with target accessibility, suggesting that BSA is unnecessary under crowding conditions. We speculated that uniform fragment lengths may contribute to incomplete coverage, as fragments of similar mass can be difficult to resolve and be assigned by LC-MS/MS. To further reduce complexity and improve sequence identifiability, we designed 18-nt guides that generated fragments binned across a range of lengths (69–81 nt). Although cleavage appeared slightly less efficient by gel analysis, the average sequence coverages were comparable to those obtained with 16-nt guides and improved further in reactions containing the enhancer mix (∼95%), indicating that increased guide specificity and variable fragment lengths can facilitate data deconvolution. However, we observed that a defined region of the Fluc transcript (positions 1037–1144) remained undetectable with both guide sets across multiple reaction conditions. While RNA structure prediction did not indicate any obvious structural or template sequence irregularities in this region, guides from both the 16-nt and the 18-nt guide sets targeting this region (including guide Fl.7 [16 nt; targeting position 1037] and 18-nt guides F2.2 [1062] and G2.2 [1144]) consistently displayed lower cleavage efficiency across experiments (Figures S1B and S5). The underlying cause of the reduced cleavage efficiency at these sites is unclear, but this effect may contribute to a lower-than-average abundance of RNA fragments of this region in multiplexed cleavage experiments. Importantly, the presence of RNA fragments flanking this region indicates successful Mpa Ago binding and cleavage at these sites. One possible explanation for the absence of the respective fragments in the LC-MS/MS analysis is that cleavage products originating from this region co-elute with other RNA cleavage fragments and are, therefore, not resolved in the chromatogram. Alternatively, these specific fragments may exhibit reduced recovery during RNA cleanup due to unknown sequence or structural features. Although the RNA substrates examined in this study were shorter than typical saRNAs, and 24 multiplexed guides represents the maximum number of combined guides we tested, our data suggest that multiplexed Mpa Ago cleavage could be applied to clinically relevant long RNAs. Notably, full substitution of uridine with m1Ψ, a common feature in therapeutic mRNAs, only had modest effects on average sequence coverage. Mpa Ago used with a limited set of multiplexed guides that generate characteristic substrate cleavage patterns enables RNA fingerprinting. This approach could help identify various RNA species within mixed samples (e.g., to determine the composition of mRNA pools in lipid nanoparticle formulations) or allow detection of target RNAs in diagnostic contexts. pAgo-driven multiplexed detection of multiple DNA species in parallel has been demonstrated previously.46,63,64 However, only a single method has shown direct detection of RNA within mixed RNA pools using the HD3 deoxyribozyme.65 Due to its single-turnover character, this method requires multiple rounds of cleavage and re-annealing, resulting in digestion times of up to 24 h for complex substrates. In contrast, Mpa Ago achieves efficient fragmentation in as short as 15 min and is not restricted to the limited trinucleotide motifs required by HD3, offering greater speed and flexibility in target-site selection. In eukaryotic mRNA, the N7-methyl guanosine cap structure is a 5′ end modification that is important for RNA stability, nuclear export, and translation efficiency, making effective cap incorporation a critical quality attribute of therapeutic mRNAs. Current cap-analysis workflows typically rely on LC-MS/MS characterization of a defined 5′ fragment generated by selective cleavage, using either trans-acting deoxyribozymes or ribozymes,66,67,68,69 DNA-directed approaches employing RNase H,70 or sequence-specific RNases (e.g., RNase 4, MC1, and RNase T1)71,72 to release the cap-containing fragment from the 5′ UTR. Our data show that guided Mpa Ago efficiently releases the cap-containing 5′ termini of mRNAs, enabling straightforward and precise assessment of capping efficiency and cap structure. Although pAgos have relatively low turnover26,33 and require higher enzyme concentrations than non-guided endoribonucleases, they offer several advantages for 5′-end analysis. Unlike RNase H or sequence-specific RNases, Mpa Ago does not require blocking oligos or the introduction of specific recognition motifs, and it generates only two defined fragments rather than fragmenting the entire downstream RNA. This simplifies workflows and avoids the need to enrich 5′ fragments before analysis. Mpa Ago-driven cap analysis is also compatible with transcripts that contain heterogeneous or poorly defined endogenous 5′ ends,73 where blocking-oligo protection may be insufficient. Additionally, the transient Ago/guide-RNA binding at 50°C may help to resolve inhibitory RNA structures in the 5′ UTR and thus reduce bias and improve cleavage consistency. Mpa Ago leaves single-stranded fragments with a 5′-phosphate group on the downstream and a chemically competent 3′-hydroxyl group at the terminus of the upstream fragment, allowing for downstream enzymatic modifications, such as fill-in labeling or ligation reactions, without complication from double-stranded fragments with 3′ phosphate or cyclic-phosphate overhangs generated by many RNases. Mpa Ago further significantly reduces workflow time and produces cleavage products that can be analyzed directly by LC-MS/MS or, due to the generation of only two fragments, by PAGE for rapid and inexpensive capping-efficiency assessment. In a recent study, a type III toxin endoribonuclease was used for cap analysis by similarly generating two major fragments.74 However, this required incorporation of the nuclease-specific target sequence and exhibited relative rather than absolute specificity, making off-target cleavage and sequence-design constraints significant drawbacks for mRNA therapeutics. Finally, unlike RNase H-based methods, which typically yield heterogeneous cleavage near the desired cut site and thus require optimization of the blocking oligo sequence,70 Mpa Ago displayed minimal off-target activity across multiple guides targeting short or full-length capped RNA substrates as assessed by CE and LC-MS/MS. Together, these features position Mpa Ago as a versatile and robust tool for high-resolution analysis of mRNA 5′ ends. We further demonstrate that Mpa Ago-mediated site-directed cleavage overcomes current limitations in characterizing RNA 3′ ends, which are critical determinants of mRNA stability and translation. While high-resolution methods like RNase T1 digestion are widely used,75,76 they are incompatible with modern therapeutic mRNAs, such as BNT162b2, which incorporate G-containing non-A sequences in the poly(A) tail to enhance polymerase accuracy. RNase T1 targets these guanines, generating complex fragment pools. Furthermore, we show that RNase T1 exhibits concentration-dependent non-specific degradation of poly(A) regions, further complicating tail assessment. In contrast, Mpa Ago allows for precise cleavage to capture defined poly(A) segments without the need for purification or enrichment and thus facilitates noise-free, single-nucleotide resolution analysis via LC-MS/MS. In addition, Mpa Ago uniquely enables 3′ end polishing of IVT RNA, producing transcripts with uniform termini. Non-templated addition of nucleotides by phage RNA polymerases in vitro is a source of 3′ end heterogeneity, which can facilitate formation of double-stranded RNA. Our data suggest that targeted 3′ trimming by Mpa Ago can homogenize 3′ RNA ends and may significantly reduce or prevent the formation of immunogenic dsRNA in RNA manufacturing, addressing a main concern in RNA therapeutics development. Because Mpa Ago, its DNA guide, and the cleaved 3′ RNA extensions can easily be removed by proteinase K/DNase treatment, and/or sizing/silica-based adsorption columns, this strategy could be integrated into scalable RNA manufacturing workflows. Taken together, we show that programmed cleavage of mRNA by Mpa Ago enables faster, more precise, and straightforward workflows for multiple important RNA analysis workflows, like 5′ capping analysis, RNA sequence/modification mapping, and poly(A) tail analysis—all essential for therapeutic RNA quality assessment. In addition, we show that Mpa Ago can be used to generate RNA populations with homogeneous 3′ ends, which may help to minimize the presence of immunogenic dsRNA and thus address one of the most critical hurdles in therapeutic RNA production. Materials and methods In vitro RNA cleavage assay 200 ng of Fluc RNA (∼1.8 kb, ∼0.37 pmol), a shortened version of the pTsin saRNA (pSG90; ∼1.7 kb, ∼0.37 pmol), or Epo RNA with a modified sequence (pSG95; ∼0.7 kb, ∼0.86 pmol), all synthesized using in vitro transcription (New England Biolabs, #E2040) according to the manufacturer’s instructions, were cleaved with Mpa Ago in a 10-μL reaction at 50°C for either 15 min (single and double cleavages, 1:5 substrate:Ago molar ratio) or 30 min (multiplexing, 1:10 substrate:Ago ratio) in 1x ThermoPol reaction buffer (New England Biolabs, #B9004S). The sequences of all IVT RNAs used in this study are shown in Table S1, and the sequences of the guides used are shown in Table S2. The Ago:guide molar ratios were 1:5 for single and double cleavages, 1:1 for multiplexing experiments, and 1:10 for cleaving polycistronic RNA. DNA guides were pooled using the Echo Acoustic Liquid handler (Beckman Coulter) and phosphorylated using T4 polynucleotide kinase (New England Biolabs, #M0201S) according to the manufacturer’s protocols. For multiplexing experiments and when indicated, a 1x enhancer mix containing 0.5 M betaine, 20 mM potassium glutamate, 20% glycerol, and 5% PEG3350 was added to the reactions. BSA was added to 0.5% (v/v) when indicated. Removal of DNA guides was performed by adding 20 U of exonuclease I (New England Biolabs, #M0293), followed by incubation at 37°C for 15 min. Cleavage experiments were stopped by the addition of 0.08 U of proteinase K (New England Biolabs, #P8107S) at room temperature for 5–10 min. After adding RNA loading dye (New England Biolabs, #B0363), cleaved RNAs were resolved by gel electrophoresis on 6% TBE-urea gels (Thermo Fisher Scientific, #EC6265), stained by 0.5x SYBR Gold (Thermo Fisher Scientific, #S11494), and analyzed using the Typhoon biomolecule imager (Cytiva) or Odyssey M imager (LICORBio). For mass spectrometry analysis, reactions were scaled up 10-fold in ThermoPol reaction buffer (i.e., 2 μg of RNA) at 50°C for 30 min. Cleavage products were then column purified (New England Biolabs, #T2030S) and eluted in 30 μL of nuclease-free water. Purified RNA was subsequently filtered with Ultrafree-MC Centrifugal Filter (Millipore Sigma, # UFC30GV0S) for 5 min at 10,000 × g prior to LC-MS/MS analysis. For RNA fingerprinting, cleavage assays were performed as described above, but with a limited set of guides targeting Fluc IVT RNA and pSG90 IVT RNA, generating three 72-nt or one 100-nt RNA species-specific cleavage fragments, respectively. 3.5 pmol of Fluc IVT RNA, 3.5 pmol of pSG90 IVT RNA, or a mix of 1.75 pmol Fluc IVT RNA and 1.75 pmol pSG90 IVT RNA were incubated with 17.5 pmol of the guide pool in the presence or absence of 17.5 pmol Mpa Ago in 1x ThermoPol reaction buffer and the enhancer mix for 30 min at 50°C. Reactions were stopped by the addition of 0.8 units proteinase K and incubation at 24°C for 10 min. Following a column cleanup, samples were either run on a denaturing 6% TBE-urea gel or analyzed by LC-MS/MS. Capping analysis For 5′ mRNA cap analysis experiments, 50 μg (∼83 pmol) or 25 μg (∼44 pmol) of Fluc or pSG90 IVT mRNA, respectively, were initially denatured for 5 min at 65°C and immediately put on ice. Both mRNAs were capped using FCE (New England Biolabs #M2081S) according to the manufacturer’s manual. The capping reactions were subsequently column purified (New England Biolabs, #T2040S). 200 μL reactions containing 3.5 pmol of capped and uncapped Fluc or pSG90 mRNA were then incubated for 15 min at 50°C in 1x ThermoPol reaction buffer, 30% glycerol, 160 units of Murine RNase Inhibitor (New England Biolabs, #M0314S), 67.4 pmol Mpa Ago, and 66.4 pmol guide (A1:/5Phos/ATCTCCTTCTTAAAGT, A1.2:/5Phos/TCCTTCTTAAAGTTAAAC, D891:/5Phos/TATATCTCCTTCTTAA targeting Fluc mRNA, or D831:/5Phos/GTCGGCTGTTTGATTC, D892:/5Phos/TTGATTCAATAGTGTG targeting pSG90 mRNA). Reactions were incubated at 50°C. After 15 min, the reactions were put on ice, 120 units of exonuclease I (New England Biolabs, #M0293S) was added, and the reactions incubated at 37°C for 15 min to remove unbound DNA guides. The reactions were stopped by the addition of EDTA to 20 mM and incubation at room temperature for 5 min. Finally, reactions were column purified and submitted for LC-MS/MS analysis. 3′ end homogenization and poly(A) tail analysis For the analysis of the poly(A) tail composition and length and to homogenize the 3′ ends of an mRNA population, Mpa Ago reactions were scaled up to a 400 μL reaction volume. 67 pmol of IVT pSG95 (723 nt) or pSG120 (698 nt) RNAs were incubated with 162 pmol Mpa Ago, 800 pmol D758 (/5Phos/TATGTTTTTTAGTCATA), or D834 (/5Phos/TTTTTTTATTAGGAGCAG) guides in 1x ThermoPol reaction buffer with 640 units of Murine RNase Inhibitor for 30 min at 50°C. Reactions were stopped by the addition of 1.6 units proteinase K and 10-min incubation at room temperature. Reactions were column purified, followed by enrichment of the poly(A)-containing cleavage fragments. The volume of the purified cleavage fragments was adjusted to 50 μL and then 1 volume (50 μL) of high-salt buffer (20 mM Tris/Cl pH 7.5, 0.5 M LiCl2, 10 mM EDTA) was added. 250 μL of oligo d(T)25 magnetic beads (New England Biolabs, #S1419S) were washed once with 100 μL low-salt buffer (20 mM Tris/Cl pH 7.5, 50 mM LiCl2, 10 mM EDTA) and twice with high-salt buffer. The RNA was added to the beads and incubated at room temperature on a rotator for 20 min. Bead-bound RNA was washed once with low-salt buffer, twice with high-salt buffer, and one final time with low-salt buffer. To elute the enriched RNA, the beads were incubated with 30 μL of H2O for 2 min at 50°C in a heat shaker (400 rpm). The eluted RNA was recovered, incubated for 5 min at 65°C, and spun through Ultrafree-MC Centrifugal Filter (Millipore Sigma, #UFC30GV0S) for 5 min at 10,000 × g. The enriched and filtered samples were then subjected to LC-MS/MS analysis. Data and code availability Raw data, including unedited gel images underlying the figures, and supplemental figures, are available in the Harvard Dataverse at https://doi.org/10.7910/DVN/JSUJRW. Acknowledgments The authors want to thank Katherine Marks for purifying Mpa Ago. We also want to thank Thomas C. Evans Jr. and Eric Cantor for their advice and strong support of this research. Author contributions L.L., E.A.H., and S.G conceived the study, designed the experiments, performed the biochemical experiments, and analyzed the data. E.J.W. assisted with LC-MS/MS analysis and data interpretation. I.R.C. and E.Y. supervised the study and provided conceptual guidance. S.G. and L.L. wrote the manuscript with input from all authors. All authors reviewed, edited, and approved the final manuscript. Declaration of interests At the time of this work, the authors were employed by New England Biolabs, Inc., which funded this work and is the manufacturer of reagents, some of which were used in this manuscript. Even though the authors were employed and funded by New England Biolabs, Inc., this did not detract from the objectivity of data generation or its interpretation. The authors are inventors on a patent application related to the methods described in this manuscript (application no. NEB-483-US). Footnotes Supplemental information can be found online at https://doi.org/10.1016/j.omtn.2026.102983. Supplemental information Document S1. Figures S1–S14, Tables S1–S4, and supplemental methods Document S2. Article plus supplemental information References Associated Data Supplementary Materials Document S1. Figures S1–S14, Tables S1–S4, and supplemental methods Document S2. Article plus supplemental information Data Availability Statement Raw data, including unedited gel images underlying the figures, and supplemental figures, are available in the Harvard Dataverse at https://doi.org/10.7910/DVN/JSUJRW.