Human RNase 4 improves mRNA sequence characterization by LC-MS/MS

Abstract

With the rapid growth of synthetic messenger RNA (mRNA)-based therapeutics and vaccines, the development of analytical tools for characterization of long, complex RNAs has become essential. Tandem liquid chromatography-mass spectrometry (LC-MS/MS) permits direct assessment of the mRNA primary sequence and modifications thereof without conversion to cDNA or amplification. It relies upon digestion of mRNA with site-specific endoribonucleases to generate pools of short oligonucleotides that are then amenable to MS-based sequence analysis. Here, we showed that the uridine-specific human endoribonuclease hRNase 4 improves mRNA sequence coverage, in comparison with the benchmark enzyme RNase T1, by producing a larger population of uniquely mappable cleavage products. We deployed hRNase 4 to characterize mRNAs fully substituted with 1-methylpseudouridine (m1Ψ) or 5-methoxyuridine (mo5U), as well as mRNAs selectively depleted of uridine-two key strategies to reduce synthetic mRNA immunogenicity. Lastly, we demonstrated that hRNase 4 enables direct assessment of the 5' cap incorporation into in vitro transcribed mRNA. Collectively, this study highlights the power of hRNase 4 to interrogate mRNA sequence, identity, and modifications by LC-MS/MS.

Figure 1.

hRNase 4 cleaves RNA primarily downstream of U and upstream of purines. (A) A schematic of the multiplexed assay to assess hRNase 4 dinucleotide cleavage specificity by LC–MS/MS. (B) The mean log₂ fold intensity change for each input oligonucleotide in a multiplexed oligonucleotide pool (Supplementary Table S1) after incubation with a dilution series of hRNase 4 relative to experiments performed in the absence of the enzyme. Results are from two independent experiments. (C) The mean total intensity of 5′ cleavage products normalized to the number of possible sites produced upon incubation of the multiplexed oligonucleotide pool with a dilution series of hRNase 4 binned by the 3′ terminal nucleotide (3′-N). Error bars represent standard deviation from two independent experiments. (D) The mean total intensity of 3′ cleavage products normalized to the number of possible sites produced upon incubation of the multiplexed oligonucleotide pool with a dilution series of hRNase 4 binned by the 5′ terminal nucleotide (5′-N). Error bars represent standard deviation from two independent experiments. (E) The distribution of sequence coverages of total and uniquely mappable cleavage products between 4 and 40 nt in length obtained from complete theoretical digestion of 1000 randomly selected human mRNAs (RefSeq) of less than 5 kB in size. The cleavage specificities of the endoribonucleases used in this study are listed in Supplementary Table S2.

Figure 2.

hRNase 4 improves mRNA sequence validation by LC–MS/MS. (A) A schematic of the digestion method combining hRNase 4 and T4 PNK to characterize mRNA by LC–MS/MS. (B) A sequence coverage map of the uniquely mappable and isomeric cleavage products greater than 4 nt from complete theoretical digestion of fLuc mRNA with hRNase 4 and RNase T1. (C) A search of the deconvoluted oligonucleotide MS¹ masses detected in hRNase 4 and RNase T1 digests of fLuc mRNA against predicted cleavage product masses from annotated human transcripts (RefSeq database) supplemented with the fLuc mRNA (see Materials and Methods). The mean identity score from three fLuc mRNA digestions is reported for each transcript. The mean signal-to-noise ratio (S/N) of the fLuc mRNA sequence score relative to all other transcripts is reported at the top of each graph. (D) The mean total number of tandem MS/MS spectra mappable to hRNase 4 or RNase T1 fLuc mRNA cleavage products as determined by NASE (5% FDR). Error bars represent standard deviation from three independent experiments. (E) The length distribution of the sequenced cleavage products detected by NASE from digestion of fLuc mRNA with either hRNase 4 or RNase T1. The total number of distinct oligonucleotides detected from three independent experiments are reported above each boxplot. (F) A sequence coverage map of positions in the fLuc mRNA detected in at least two independent digestions with either hRNase 4 or RNase T1. The percent sequence coverage is reported above each map. (G) The mean fractional coverage of fLuc mRNA sequence obtained from hRNase 4 or RNase T1 sequenced cleavage products. Error bars represent standard deviation from three independent experiments.

Figure 3.

hRNase 4 discriminates between mRNAs modified with m¹Ψ and mo⁵U. (A) The mean log2 fold intensity change of oligonucleotides in a multiplexed pool comprising uridine modifications (Supplementary Table S3) after incubation with either hRNase 4 or MC1 relative to the corresponding conditions in the absence of an endoribonuclease. Results are from two independent experiments. (B) (Left) A search of the deconvoluted oligonucleotide MS¹ masses detected in hRNase 4 or RNase T1 digests of U-, m¹Ψ- or mo⁵U-modified EPO mRNA against predicted cleavage product masses from annotated human transcripts (RefSeq database) supplemented with the EPO mRNA (see Methods). U-, m¹Ψ- or mo⁵U-modified RNA sequences were utilized for each corresponding search. The mean identity score was calculated from two independent experiments as described in Figure 2C. (Right) The mean signal-to-noise ratio (S/N) of the score of each U-, m¹Ψ- or mo⁵U-modified EPO mRNA sequence relative to all other transcripts. (C) The mean total number of sequenced cleavage products in hRNase 4 or RNase T1 digests of U-, m¹Ψ- or mo⁵U-modified EPO mRNA (5% FDR). Cleavage products are grouped by true (green) and false (red) assignments to a modified EPO mRNA. Error bars represent standard deviation from two independent experiments. (D) A sequence coverage map of positions in the U-, m¹Ψ- or mo⁵U-modified EPO mRNA detected in two independent digestions with either hRNase 4 or RNase T1. The percent sequence coverage is reported above each map. (E) The mean fractional coverage of the U-, m¹Ψ- or mo⁵U-modified EPO mRNA sequences obtained from hRNase 4 or RNase T1 sequenced cleavage products. Error bars represent standard deviations from two independent experiments.

Figure 4.

hRNase 4 discriminates between uridine-depleted mRNAs. (A) (Top) A schematic of the three uridine-depleted cLuc mRNAs used in this study. (Bottom) A table of the percent identity between the three uridine-depleted cLuc mRNAs (Clustal Omega). (B) A search of the deconvoluted oligonucleotide MS¹ masses detected in hRNase 4 digests of the three uridine-depleted cLuc mRNAs against predicted cleavage product masses from annotated human transcripts (RefSeq database) supplemented with each uridine-depleted cLuc mRNA. The mean identity score was calculated from two independent experiments as described in Figure 2C. The mean signal-to-noise ratio (S/N) of the score of each uridine-depleted cLuc mRNA sequence relative to all other transcripts is reported at the top of each graph. (C) The mean total number of sequenced cleavage products in hRNase 4 digests of three uridine-depleted cLuc mRNAs (5% FDR). Cleavage products are grouped by true (green) and (red) false positive assignments to uridine-depleted cLuc mRNAs. Error bars represent standard deviation from two independent experiments. (D) The distribution of search engine (NASE) scores of true (green) and false (red) assigned cleavage products to each uridine-depleted cLuc mRNA detected in two independent digestions with hRNase 4. (E) A sequence coverage map of positions in each uridine-depleted cLuc mRNA from two independent digestions with hRNase 4. The percent sequence coverage of true (green) and false (red) assigned cleavage products are reported above each map.

Figure 5.

Characterization of a 4187 nt mRNA encoding the BNT162b2 sequence with hRNase 4. (A) (Top) A schematic the BNT162b2 mRNA sequence unmodified or modified with m¹Ψ. (Bottom) A search of the deconvoluted oligonucleotide MS¹ masses detected in hRNase 4 or RNase T1 digests of U- or m¹Ψ-modified BNT162b2 mRNA against predicted cleavage product masses from annotated human transcripts (RefSeq database) supplemented with the BNT162b2 mRNA sequence. The mean identity score was calculated from two independent experiments as described in Figure 2C. The mean signal-to-noise ratio (S/N) of the BNT162b2 mRNA sequence score relative to all other transcripts is reported at the top of each graph. (B) A sequence coverage map of positions in the U- or m¹Ψ-modified BNT162b2 mRNA sequence detected by NASE in two independent digestions with hRNase 4 or RNase T1. The percent sequence coverage is reported above each map. (C) The mean fractional coverage of the U- or m¹Ψ-modified BNT162b2 mRNA sequence obtained from hRNase 4 or RNase T1 sequenced cleavage products. Error bars represent standard deviations from two independent experiments.

Figure 6.

hRNase 4 enables characterization of an m⁷GpppAm capped mRNA. (A) A schematic of the capped and poly-adenylated EPO mRNA. (B) A sequence coverage map of positions in either uncapped or capped EPO mRNA detected in at least two independent digestions with hRNase 4. The percentage sequence coverage is reported above each map. (C) A comparison of hRNase 4 cleavage products identified by oligonucleotide MS¹ mass analysis of uncapped and capped EPO mRNA. hRNase 4 cleavage products identified in two independent experiments are reported. (D) The mean intensity of 5′ terminal hRNase 4 cleavage products with a maximum of three missed cleavages from the 5′ end and variable addition of possible terminal cap structures (see Materials and Methods for details). Error bars represent standard deviation from two independent experiments.