RNA modifications: importance in immune cell biology and related diseases

Abstract

RNA modifications have become hot topics recently. By influencing RNA processes, including generation, transportation, function, and metabolization, they act as critical regulators of cell biology. The immune cell abnormality in human diseases is also a research focus and progressing rapidly these years. Studies have demonstrated that RNA modifications participate in the multiple biological processes of immune cells, including development, differentiation, activation, migration, and polarization, thereby modulating the immune responses and are involved in some immune related diseases. In this review, we present existing knowledge of the biological functions and underlying mechanisms of RNA modifications, including N⁶-methyladenosine (m⁶A), 5-methylcytosine (m⁵C), N¹-methyladenosine (m¹A), N⁷-methylguanosine (m⁷G), N⁴-acetylcytosine (ac⁴C), pseudouridine (Ψ), uridylation, and adenosine-to-inosine (A-to-I) RNA editing, and summarize their critical roles in immune cell biology. Via regulating the biological processes of immune cells, RNA modifications can participate in the pathogenesis of immune related diseases, such as cancers, infection, inflammatory and autoimmune diseases. We further highlight the challenges and future directions based on the existing knowledge. All in all, this review will provide helpful knowledge as well as novel ideas for the researchers in this area.

Introduction

Chemical modification occurs on many types of biological macromolecules, such as nucleic acids, sugars, lipids, and proteins, and is specific and efficient for regulating their functions.^1,2,3,4 RNA modifications, such as N⁶-methyladenosine (m⁶A), 5-methylcytosine (m⁵C), N¹-methyladenosine (m¹A), N⁷-methylguanosine (m⁷G), N⁴-acetylcytosine (ac⁴C), pseudouridine (Ψ), uridylation, and adenosine-to-inosine (A-to-I) RNA editing, are RNA features that alter the canonical AUGC bases and function as emerging and critical post-transcriptional regulators.^1,2,3,5 The RNA modifications catalyzed by “writer” enzymes can be removed by “eraser” enzymes.^6,7 However, some modifications are further modified by enzymes, which we term for the first time as “modifiers”. The modifications are identified by RNA-binding proteins (RBP) known as “readers” to participate in various physiological and pathological processes.^1,2,3,6,7

The immune cell abnormality in human diseases is also a research focus and progressing rapidly these years. In 2005, a study demonstrated that RNA modifications, such as m⁶A, m⁵C, m⁵U, s²U, or Ψ, may influence the activations of dendritic cells (DCs) and toll-like receptor (TLR)-expressing cells.⁸ Although this research was a preliminary exploration, it gave us knowledge that RNA modification could affect the biology of immune cells. In recent years, advances in new technologies and ideas have led to an increasing number of researchers focusing on the influence of RNA modifications on immune cell biology, and their roles in immune related diseases. m⁶A is the modification most frequently studied and there already have some reviews summarized its important roles in immune processes.^9,10 Although not thoroughly and comprehensively, other RNA modifications have also been confirmed to participate in the immune cell biology and immune related diseases. However, due to the complexity of RNA modification and the diversity of immune cells, the interaction network between RNA modification and immune cells remains largely unclear, which needs to be further consummated. Hence, based on the research status, we write this review, which we hope to be helpful for researchers and promote progress in this area.

In this review, we clarify the current understanding of eight RNA modifications and focus on their critical roles in regulating immune cell biology and immune related diseases. We also highlight questions that remain to be addressed in this area and provide perspectives for further studies.

RNA modifications

N ⁶-methyladenosine

The methylation of adenosine at position N6, m⁶A modification has emerged as the most prevalent and abundant mRNA modification in eukaryotes (m⁶A/A = 0.1–0.6%).^11,12 It appears in the full-length sequence but is enriched in the vicinity of the stop codon and the 3ʹ untranslated region (3ʹUTR) of mRNAs, within the consensus motif RRACH (R = G or A; H = A, C, or U).^13,14 It also occurs in most non-coding RNAs, including ribosomal RNAs (rRNAs), small nuclear RNAs (snRNAs), small nucleolar RNAs (snoRNAs), microRNAs (miRNAs), long non-coding RNAs (lncRNAs), and circular RNAs (circRNAs) (Fig. 1).^15,16,17

Fig. 1

RNA modifications and their distributions on different RNA subtypes. a Chemical structures of eight RNA modifications. b Distribution of RNA modifications on different RNA subtypes. Indicated modifications are labeled at the corresponding modification sites. m⁶A N⁶-methyladenosine, m⁵C 5-methylcytosine, m¹A N¹-methyladenosine, m⁷G 7-methylguanosine, ac⁴C N⁴-acetylcytidine, ψ pseudouridine, A-to-I editing adenosine-to-inosine RNA editing, CDS coding sequence, UTR untranslated regions, pri-miRNA primary microRNA, pre-miRNA precursor microRNA

m⁶A deposition in mRNA is mostly mediated by the m⁶A methyltransferase complex (MTC) (Fig. 2 and Table 1).^18,19,20 The key MTC components are methyltransferase-like 3 (METTL3), METTL14, Wilms’ tumor 1-associating protein (WTAP), vir-like m⁶A methyltransferase-associated (VIRMA, also known as KIAA1429), Cbl proto-oncogene-like 1 (HAKAI), zinc finger CCCH-type containing 13 (ZC3H13), and RNA-binding motif protein 15/15B (RBM15/15B).^18,21,22 Among them, METTL3 is considered the only putative S-adenosylmethionine (SAM)-dependent methyltransferase with its own catalytic ability and can form a tight heterodimer with METTL14 to perform catalysis.^23,24,25 The other aforementioned writers act as regulatory factors.^15,20 Zinc finger CCCH-type containing 4 (ZCCHC4) and METTL5 mediate m⁶A formation on 28S and 18S rRNA, respectively, to accelerate the global translation rate (Fig. 3 and Table 1).^{26,27,28,29,30} In U6 snRNA, m⁶A is executed by METTL16 to participate in RNA splicing regulation (Fig. 4 and Table 1).^31,32,33

Fig. 2

The machinery of RNA modifications and their molecular functions in mRNA. All RNA modifications included in this review can be installed on mRNA through their writers, and m⁶A as well as m¹A modifications can be removed by indicated erasers, therefore making these RNA modifications dynamically reversible processes. Some of the RNA modifications can be recognized by their respective reader proteins, which changes the fates of target RNAs via altering generation, transportation, function and metabolization. m⁶A N⁶-methyladenosine, m⁵C 5-methylcytosine, m¹A N¹-methyladenosine, m⁷G 7-methylguanosine, ac⁴C N⁴-acetylcytidine, ψ pseudouridine, U Uridine, A-to-I editing adenosine-to-inosine RNA editing, dsRNA double-stranded RNA

Table 1 Characteristics of reviewed RNA modifications

Fig. 3

The machinery of RNA modifications and their molecular functions in rRNA. The indicated RNA modifications are installed on rRNA via their writers. These modifications occurred on rRNA alter the RNA structure, thereby regulating the function of ribosomes, which in turn affects the translation rate. The same modification can be installed by different writers in different parts of the cell. Besides, m⁶A modifications on different subunits of the ribosome can be catalyzed by different writers. Some writers also need to form a heterodimeric complex with methyltransferase activators to gain metabolic stability in cells, such as METTL5-TRMT112. m⁶A, N⁶-methyladenosine; m⁵C, 5-methylcytosine; m¹A, N¹-methyladenosine; m⁷G, 7-methylguanosine; ac⁴C, N⁴-acetylcytidine; ψ, pseudouridine

Fig. 4

The machinery of RNA modifications and their molecular functions in snRNA, snoRNA and miRNA. The indicated RNA modifications are installed on snRNA, snoRNA and miRNA through respective writers. m⁶A modifications on snRNA and miRNA can be removed by FTO, while m⁷G modifications on snRNA and snoRNA can be removed by H29K, making RNA modifications on snRNAs, snoRNA or miRNAs dynamically reversible process. Besides, m⁷G installed snRNA and snoRNA can be further modified as m^2,2,7G by modifier TGS1. RNA modifications affect the function of these non-coding RNAs via altering their structures, facilitating fine-tuning in various physiological processes. m⁶A N⁶-methyladenosine; m⁷G 7-methylguanosine, ψ pseudouridine, U Uridine, A-to-I adenosine-to-inosine, m₃^2,2,7G 2,2,7-trimethyl guanosine, pri-miRNA primary microRNA, pre-miRNA precursor microRNA

To date, two m⁶A erasers have been identified, both belonging to the AlkB family of the Fe (II)/α-ketoglutarate-dependent dioxygenase superfamily.^34,35,36,37 The first eraser is fat mass and obesity-associated protein (FTO) that mediates mRNA and snRNA demethylation in m⁶A and m⁶A_m residues, and tRNA in m¹A residue (Figs. 2, 4, 5 and Table 1).^35,36,38 The other m⁶A eraser, AlkB homolog 5 (ALKBH5), only oxidatively reverses m⁶A in mRNA (Fig. 2 and Table 1).^39,40

Fig. 5

The machinery of RNA modifications and their molecular functions in tRNA. The indicated RNA modifications are installed on tRNA through indicated writers, and m¹A modifications can be removed by ALKBH3 and FTO, while m⁵C modification on pre-tRNA can be converted into hm⁵C or f⁵C by TET2. These modifications on tRNA can alter the structure of tRNA, thereby regulating its functions to affect the translation efficiency. The same modification can be installed by different writers on tRNAs in different parts of the cell. A-to-I editing on different tRNA positions can be added by different writers. ac⁴C writer NAT10 modifies tRNAs assisted by the adaptor Tan1/THUMPD1. m⁵C 5-methylcytosine, m¹A N¹-methyladenosine, m⁷G 7-methylguanosine, ac⁴C N⁴-acetylcytidine, ψ pseudouridine, A-to-I adenosine-to-inosine, hm⁵C 5-hydroxymethylcytidine, f⁵C 5-formylcytidine

Many reader proteins influence the fate of m⁶A RNAs in various ways, which is largely determined by their subcellular localization. The most studied readers are the YT521-B homology (YTH) domain family members, which share the m⁶A-recognizing YTH domain but exert different effects on RNA fate.^41,42 The YTH domain family includes YTHDF1–3 and YTHDC1–2.^41,42 YTHDF1 and YTHDF3 actively promote protein synthesis by interacting with translation machinery, whereas YTHDF2 recruits RNA-degrading enzymes or adaptor proteins to trigger the rapid degradation of its target mRNA.^{43,44,45,46,47} YTHDC1 not only facilitates the decay of m⁶A-modified chromosome-associated regulatory RNAs to influence the open chromatin state and downstream transcription but also mediates mRNA splicing by recruiting and regulating pre-mRNA splicing factors.^48,49,50 YTHDC2 may participate in mRNA stability and translation in an m⁶A-dependent or -independent manner.^51,52,53 Insulin-like growth factor 2 mRNA-binding proteins (IGF2BPs, which include IGF2BP1–3), identify m⁶A through K homology domains to enhance mRNA stability and translation.^54,55 The heterogeneous nuclear ribonucleoprotein (HNRNP) family members, which include HNRNPC, HNRNPG, and HNRNPA2B1, can identify m⁶A on precursor (pre)-mRNAs and/or primary (pri)-miRNAs to mediate splicing and/or nucleocytoplasmic trafficking.^{56,57,58,59,60} Eukaryotic initiation factor 3 (eIF3) promotes cap-independent translation upon the induction of cellular stress by recruiting the 43S complex to initiate translation.^61,62 Both proline rich coiled-coil 2A (PRRC2A) Staphylococcal nuclease and tudor domain-containing 1 (SND1) function as m⁶A readers, facilitating the stabilization of modified RNA.^63,64 In particular, m⁶A participates in transcription termination by promoting co-transcriptional R-loop formation to prevent the readthrough activity of Pol II, while it is unclear whether other readers are involved in this process.⁶⁵ The specific roles of m⁶A in various RNA molecules are presented in Figs. 2 to 4 and Table 1.

Collectively, m⁶A modification plays a regulatory role in various cellular processes by affecting transcription, maturation, localization, function, and metabolism in different RNA classes.^14,22,66,67 (Table 1) In mRNAs, m⁶A can affect transcription, maturation, localization, translation and degradation, eventually influencing the proteins encoded (Fig. 2).^7,68 In rRNAs, the m⁶A1832 modification in 18s rRNA as well as the m⁶A4220 modification in 28S rRNA are required for global translation (Fig. 3).^26,27 In snRNAs and snoRNAs, m⁶A modification may regulate snRNA pre-mRNA or pre-lncRNA splicing processes (Fig. 4).^31,69 In miRNAs, m⁶A facilitates pri-miRNA processing by recruiting the miRNA microprocessor complex protein DGCR8 depending on HNRNPA2B1 (Fig. 4).^56,70,71 In lncRNAs and circRNAs, m⁶A has been verified to modulate generation, structure, distribution, function and metabolism (Fig. 1), and m⁶A is also a critical translation initiator in circRNAs with coding potential.^70,72

Although m⁶A has been widely investigated, there are still some important questions remain to be solved. For example, m⁶A is a widespread modification that can affect a variety kind of RNA, existing studies mainly focus on the effect of m⁶A on mRNA, and its effect on non-coding RNA may be of good research interest. The current study shows that the effect of m⁶A on RNA stability is bidirectional, i.e., increasing stability or promoting degradation. For this issue, it is necessary to fully consider the modification sites of m⁶A in RNAs as well as the imbalance of different readers, such as YTHDF2 and IGF2BPs. m⁶A may affect circRNA generation and circRNA-mRNA imbalance by mediating pre-mRNA splicing, which is also a potential study focus due to the recent research upsurge on circRNAs. Although existing inhibitors can interfere m⁶A levels by inhibiting writers,^25,73,74 they are often nonspecific and affects overall m⁶A levels. It is more significant to explore gene specific m⁶A interference.

5-methylcytosine

m⁵C methylation occurs at position 5 of the cytidine residues of both DNA and RNA. Identified in 1958, m⁵C is described as a widespread mark in the epitranscriptome on tRNA, rRNA, mRNA, enhancer RNA (eRNA), and miRNA and is most abundant in eukaryotic tRNAs and rRNAs (Fig. 1).^75,76,77,78

In eukaryotes, m⁵C methylation is introduced by the NOL1/NOP2/SUN domain (NSUN) family members, NSUN1–7, and DNA methyltransferase-like 2 (DNMT2) as presented in Table 1.^76,78,79 Specific m⁵C writers catalyze different RNA subsets. According to current knowledge, cytoplasmic tRNAs are methylated by NSUN2, NSUN6, and DNMT2, while mitochondrial tRNAs are catalyzed by NSUN2 and NSUN3 (Fig. 5 and Table 1).^{80,81,82,83,84} rRNAs are methylated by NSUN1 and NSUN5 in the nucleus and by NSUN4 in the mitochondria (Fig. 3 and Table 1).^85,86,87 mRNAs are methylated by NSUN2 and NSUN6, whereas ncRNA and eRNAs can be modified by NSUN2 and NSUN7, respectively (Fig. 2 and Table 1).^{76,78,79,84,88}

Recently, some m⁵C erasers or modifiers have been concerned in RNA molecules. The known erasers/modifiers include ten-eleven translocation (TET) proteins (TET1–3) and α-ketoglutarate-dependent dioxygenase ABH1 (ALKBH1), which possess the activity of oxidizing m⁵C to 5-hydroxymethylcytidine (hm⁵C) (Fig. 5 and Table 1).^89,90,91 In DNA, TETs successively convert m⁵C to hm⁵C, 5-formylcytosine (f⁵C), and 5-carboxylcytosine, the latter two of which are identified and removed by thymine DNA glycosylase;^92,93,94,95 whereas in RNA, TETs has been only reported to convert m⁵C to hm⁵C (Fig. 5).^96,97,98 ALKBH1 can successively convert m⁵C into hm⁵C and f⁵C at position 34 of cytoplasmic and mitochondrial tRNA, and the process in mitochondria is required for mitochondrial functions (Fig. 5).^91,99 Under all circumstances, hm⁵C formation from m⁵C reduces the m⁵C modification, which is why previous studies considered TETs and ALKBH1 erasers.

Like m⁶A modification, m⁵C also involves binding proteins to change the fate of the modified RNA. The first identified mRNA m⁵C reader was RNA and export factor-binding protein 2 (ALYREF), a well-known protein complex that facilitates the nuclear export of mRNAs (Fig. 1 and Table 1).^100,101 Y-box-binding protein 1 (YBX1) is a reader located in the cytoplasm that enhances the stability of m⁵C-modified mRNA by recruiting ELAV like RNA binding protein 1 (ELAVL1), an mRNA stability maintainer (Fig. 2 and Table 1).^102,103 Besides, YTHDF2, also an m⁶A reader protein, could directly bind to m⁵C in RNA to modulate the distribution of m⁵C in both coding and noncoding RNA and influence rRNA maturation by regulating m⁵C levels (Table 1).¹⁰⁴ Recently, Lan and colleagues presented a novel m⁵C reader, fragile X messenger ribonucleoprotein 1 (FMRP), which could be recruited to DNA damage sites by DNMT2 and promote TET1-mediated RNA m⁵C demethylation in DNA:RNA hybrids (Table 1).⁹⁰

Generally, m⁵C plays a critical role in stabilizing both non-coding and coding RNAs. In tRNAs, m⁵C regulates RNA structure and stability and is required for translation accuracy (Fig. 5 and Table 1).^78,80,105 m⁵C methylation at C2278 within a conserved region of 25S rRNA stabilizes the structural conformation of the ribosome, ensures translation fidelity, and recruits oxidative stress-responsive mRNA subsets to polysomes (Fig. 3 and Table 1).⁸⁶ m⁵C methylation on vault RNAs affects their processing into derived small RNAs, while m⁵C in eRNAs protects them from degradation (Table 1).¹⁰⁶ In mRNAs, m⁵C is vital for modulating stability, nuclear export, and translation (Table 1).^{101,102,107,108,109} For example, a subset of mRNAs with hypermethylated m⁵C sites was stabilized in an NSUN2- or YBX1-dependent manner, which influenced bladder carcinogenesis or embryonic development in zebrafish.¹⁰⁷ NSUN2 enhanced the recognition of cyclin-dependent kinase inhibitor 1A (CDKN1A) mRNA by ALYREF, which functionally promoted the nuclear export capacity and translation of CDKN1A mRNA in 3T3-L1 preadipocytes.¹⁰¹

As we described above, m⁵C is abundant and required for maintaining RNA structure and stability in eukaryotic tRNAs and rRNAs, which are vital molecules in maintaining the normal physiology of almost all types of eukaryotic cells. Thus, targeting m⁵C as a therapeutic approach may have a long way to go. Fortunately, different RNAs possess different writers, and targeting specific writers can affect the function of specific RNAs. For instance, a recent study revealed that targeting NSUN3 to regulate site-specific mitochondrial RNA m⁵C modification shows therapeutic effects in combating cancer metastasis.¹⁰⁵

N ¹-methyladenosine

Identified in the 1960s, m¹A was reported as the methylation of adenosine at position N1 and has been observed in tRNAs, rRNAs, mRNAs, and lncRNAs.^110,111 m¹A is inextricably linked with m⁶A modification, as not only does m¹A rearrange to m⁶A under alkaline conditions (Dimroth rearrangement), they also share some regulators (Fig. 1).^36,112

The current reported human m¹A writers include nucleomethylin (NML, also known as RRP8) (for rRNA), the tRNA methyltransferase 6 non-catalytic subunit (TRMT6)–RNA methyltransferase 61A (TRMT61A) complex (for mRNA and mitochondrial tRNA), TRMT61B (for mitochondrial tRNA and rRNA), TRMT10B (for tRNA), and TRMT10C (for mitochondrial tRNA and mRNA).^{113,114,115,116,117} m¹A erasers, including FTO (for tRNA) and the ALKBH family members ALKBH1 (for mitochondrial tRNA), ALKBH3 (for tRNA and mRNA), and ALKBH7 (for mitochondrial tRNA), overlap or are closely related to some m⁶A erasers. Accordingly, it has been verified that some m⁶A readers, i.e., YTH domain family proteins including YTHDF1–3 and YTHDC1, identify m¹A modifications (Figs. 2, 3, 5 and Table 1).^{36,99,118,119,120,121,122,123}

Generally, m¹A affects RNA base pairing and subsequently influences the target RNA molecule structure and function.^115,119,124 Human rRNAs and tRNAs contain many different m¹A modification sites. For example, m¹A at position 1322 of 28S rRNA promotes 60S ribosomal subunit formation and m¹A at position 947 is essential for mitoribosomal structure and function.^113,116 m¹A at position 58 of tRNA is essential for tRNA structure, stability, and translational initiation; in this position, absent m¹A may promote the generation of tRNA-derived small RNAs (tDRs), enhancing ribosome assembly and causing malignant phenotypes.^118,121 In mRNA, m¹A is distributed in every mRNA segment, which includes the coding sequence (CDS), 5′UTR, and 3′UTR, and its roles appear region- or subcellular location-dependent.^119,125 Near the start codon, m¹A might regulate translation initiation by altering the secondary/tertiary structure or reader recognition of translation initiation sites (TISs), thereby promoting translation.¹²⁵ In mitochondria, m¹A in the 5′UTR or CDS repressed translation, probably by affecting ribosome scanning or translation (Figs. 2, 3, 5 and Table 1).^115,126

Because m¹A shares some regulators, such as YTHDF1–3 and YTHDC1, with m⁶A modification, the research ideas of m⁶A can provide reference for m¹A. Since m¹A modification can affect RNA base pairing, we expect that it might affect the binding of miRNAs with other RNA structures, such as mRNA 3ʹUTR, lncRNA and circRNA. Competing endogenous RNAs (ceRNA) regulatory network is attracting much attention these years, and m¹A modification may add novel conceptions to this theory.

N ⁷-methylguanosine

m⁷G refers to the RNA methylation of guanine at position N7 and is present in approximately 0.4% of all guanosine, a level similar to that of m¹A modification.^127,128,129 m⁷G is well known for the formation of the 5ʹ cap (m⁷GPPPN) structure of mature mRNA, snRNA, and snoRNA; moreover, it is enriched in all three transcript segments of mRNA 5ʹUTR, CDS, and 3ʹUTR and in pre-mRNAs.^{127,130,131,132} m⁷G is also present in noncoding RNAs, such as position 46 of tRNA, G1575/G1639 of 18S rRNA, and even mature and pre-miRNAs (Fig. 1).^133,134,135

RNA guanine-7 methyltransferase (RNMT), METTL1–WD repeat domain 4 (WDR4) complex, and Williams–Beuren syndrome chromosomal region 22 protein (WBSCR22, also known as BUD23) are considered m⁷G writers. Activated by RNMT-activating mini-protein (RAM), RNMT is required for efficient cap methylation.^136,137,138 By forming a complex with WDR4 or other partners, METTL1 has m⁷G methyltransferase activity for tRNA, internal mRNA, and pri-miRNA/miRNA.^{127,135,139,140} Requiring the methyltransferase adapter protein TRM112, WBSCR22 specifically methylates m⁷G in 18S rRNA.¹⁴¹ In most non-coding RNAs, the m⁷G cap can be lost during maturation by cleavage or further modification to m^2,2,7G trimethylguanosine.¹⁴² For example, trimethylguanosine synthase 1 (TGS1) might function as a modifier, which hypermethylated the m⁷G caps of snRNAs and snoRNAs to a m^2,2,7G cap structure, leading to their concentration in nuclear foci.¹⁴³ The m⁷G cap can be recognized by eIF4E and the cap-binding complex (CBC) composed of CBP80 and CBP20, thereby affecting RNA maturation, nuclear export, and translation (Figs. 2 to 5 and Table 1).^144,145,146

On mRNA, the m⁷G cap regulates multiple stages of mRNA processes, including pre-mRNA slicing, nuclear export, transcription elongation, translation, and degradation and indirectly augments ribosome synthesis and translation rates.^127,142,147 On internal mRNA, m⁷G methylation might influence mRNA translation.¹²⁷ m⁷G in tRNAs remodels the mRNA translatome by maintaining tRNA structural integrity to promote its stability, translation ability, and reduce ribosome pausing.^{133,140,148,149} However, the effects of m⁷G on rRNA have not been studied in-depth. In miRNA, m⁷G promoted miRNA processing by antagonizing G-quadruplex structures in pri-miRNA (Figs. 2 to 5 and Table 1).^135,150,151

As we know, m⁷G is widely present in mRNAs and is a critical regulator in the translation process, therefore, it may not be a good therapic target in human diseases. The roles of m⁷G regulators may vary in different RNAs and diseases. For example, m⁷G modification on tRNA promoted the progression of lung cancer,¹³⁹ while m⁷G modification on let-7 miRNA showed the opposite effect.¹³⁵ m⁷G modification promoted the progression of hepatocellular carcinoma and bladder cancer,^152,153 while it exerted an opposite opposite effect in teratoma.¹⁵⁴

N ⁴-acetylcytosine

Aside from m⁵C and hm⁵C, ac⁴C (acetylation of the N4 position of cytosine) is another conserved modification in cytidine and is currently the only acetylation event described in eukaryotic RNA.^155,156,157 As with many RNA modifications, ac⁴C was detected initially in tRNA and rRNA, followed by mRNA.^158,159 In rRNA, ac⁴C is distributed in helix 34 and helix 45 near the decoding site of mammalian 18S rRNA; in tRNA, it is detected at the D-stem of tRNA^Ser/Leu in eukaryotes.^{160,161,162,163} In mRNA, the deposition of ac⁴C sites is detected mainly in the CDS region, and also in the 5ʹUTR (Fig. 1).¹⁵⁸

N-acetyltransferase 10 (NAT10), an essential ATP-dependent RNA acetyltransferase, is currently considered the only writer of ac⁴C.¹⁶⁴ It catalyzes ac⁴C modification in 18S rRNA, tRNA, and a broad range of mRNA.^{158,160,164,165} Two additional proteins are required in ac⁴C formation in human rRNA or tRNA, respectively. The first is the box C/D snoRNA U13, which is essential and specific for 18S rRNA acetylation by timely pre-rRNA folding.¹⁶⁰ The other is THUMP domain-containing 1 (THUMPD1), a specific RNA adaptor protein harboring an RNA-binding motif that can interact with NAT10 to cooperate in tRNA acetylation (Figs. 2, 3, 5 and Table 1).^160,162

In 18S rRNAs, ac⁴C is critical for pre-rRNA processing and ribosome synthesis and influences translation ability possibly by turning the 18S rRNA 3′ end into an environment rich in base modifications to interact with mRNA or tRNA.¹⁶⁰ The function of ac⁴C formation in tRNA is not fully understood, but ac⁴C of tRNA can promote its stability and is considered a monitoring indicator of eukaryotic tRNA maturation due to the rapid tRNA degradation pathway.^166,167 Furthermore, ac⁴C can influence mRNA translation. The presence of ac⁴C on mRNA CDS region robustly boosts mRNA stability and promotes protein translation, probably by affecting its interaction with cognate tRNAs during translation.^158,168,169 However, ac⁴C modification on 5ʹUTR mainly affects translation initiation by directly and indirectly mediating exquisite locational specificity: ac⁴C modification immediately adjacent to a strong AUG start codon can repress translation, while ac⁴C modification downstream of a weak translation initiation site can facilitate translation (Figs. 2, 3, 5 and Table 1).¹⁷⁰

As a newly identified RNA modification, ac⁴C remains largely unknown, particularly its regulators and molecular functions. Only one writer and no erasers or readers have been identified. The functions of ac⁴C in rRNA, tRNA, and the CDS as well as UTR regions of mRNAs have been reported, however, relevant studies are rare. More investigations are required.

Pseudouridine

Identified nearly 70 years ago, Ψ is the C5-glycoside isomer of uridine, of which the C5 atom (instead of N1) of the heterocyclic ring is bonded to the C1′ atom of the pentose.^171,172,173 Ψ is present in almost all kinds of RNAs, including coding and non-coding RNAs, and is highly conserved among species (Fig. 1).^79,171,174

Thirteen writers for Ψ have been identified in humans, one of which is Dyskerin pseudouridine synthase 1 (DKC1), a catalytic subunit of the H/ACA snoRNP complex that catalyzes rRNA pseudouridylation, which requires an RNA guide for its catalytic activity.^175,176,177 The remaining 12 writers are RNA-independent single pseudouridine synthases (PUSs): PUS1, PUSL1, PUS3, TRUB1, TRUB2, PUS7, PUS7L, RPUSD1–4, and PUS10; these enzymes have specific cellular localizations and RNA targets.^{178,179,180,181} To date, there are no known Ψ erasers and readers. The absence of erasers may be due to the relatively inert C–C bond formed by the ribose sugar and base, leading to the pseudouridylation process being irreversible (Figs. 2 to 5 and Table 1).¹⁷⁵

Previous studies have shown that Ψ plays functional roles in RNA biogenesis, structure, stability, and function to participate in regulating gene expression.⁷⁹ tRNAs contain many pseudouridylation sites, which are critical for maintaining stable tRNA structure and mediating tRNA codon–anticodon base pairing and are thereby involved in translation processes.^{175,176,182,183,184} Ψ also represses aberrant protein synthesis by altering the properties of tRNA-derived fragments.¹⁸⁵ Similar to that in tRNA, Ψ is also abundant and present in various rRNA regions, aiding the formation of stable structures.^186,187,188 Moreover, Ψ contributes to ribosome processing and function to ensure translational fidelity in protein synthesis.^178,189 In snRNAs, Ψ was predicted to influence structure and RNA–RNA or RNA–RBP interactions to function in pre-mRNA splicing.^{172,190,191,192} Ψ is also involved in regulating pre-mRNA processing, mRNA structure, stability, translational fidelity, and termination, which is another mechanism of translation control apart from tRNA and rRNA modification (Figs. 2–5 and Table 1).^{176,178,193,194,195,196}

Despite being identified several decades ago, the contributions of Ψ to multiple cellular processes are just starting to be revealed. Similar to ac⁴C, new things always take time to be understood. The elucidation of erasers and readers of Ψ will be one of the key directions in the future. Specially, Ψ have already been applied to generating highly effective COVID-19 mRNA vaccines,¹⁹⁷ which is the clinical application of this modification and has potential value for further research.

Uridylation

In addition to the most widespread homomeric poly (A) tails, uridylation, which consists of the untemplated addition, appears to be the second most prevalent modification at the 3ʹ RNA termini.^{198,199,200,201} Virtually, uridylation can occur on all classes of eukaryotic RNAs including mRNAs, and noncoding RNAs including U6 spliceosomal RNA, guide RNA (gRNA), small interfering RNA (siRNA), miRNA, Piwi-interacting RNA (piRNA), rRNA, and tRNA. Uridylation also targets viral RNA tagging (Fig. 1).^198,199,202

In different substrates, uridylation is catalyzed by different terminal uridylyltransferases (TUTases), which belong to the noncanonical terminal nucleotidyltransferases (TENTs).^203,204 In nuclear U6 snRNA, the U6 TUTase (TUT1) specifically added or restored at least four uridines at the 3′ end.^205,206 TUT4 and/or TUT7 belonging to the TENT3 subfamily are the predominant writers of other cellular uridylation.^{205,207,208,209,210} Uridylation erasers or modifiers have not been reported and the uridylation readers include the LSM1-7 complex (for oligouridylation), DIS3L2 (for oligouridylation and polyuridylation), and La protein (Fig. 2 and Table 1).^{211,212,213,214,215,216}

Uridylation alters RNA fate from diverse aspects, including RNA maturation, function, stability, and decay. Uridylation is essential for U6 snRNA maturation and 3′ stabilization to perform splicing function and initiating gRNA maturation.^{206,217,218,219} The functions of uridylation in miRNAs are diverse. For example, TUTs-mediated pre-miRNA uridylation is a critical step in miRNA biogenesis, which involves repairing or removing defective pre-miRNAs, arm switching, and Dicer processing.^220,221,222 Uridylation on the miRNA 3′ end can recognize noncanonical target sites; on the other hand, it may abrogate target gene repression by directly affecting miRNA 3ʹUTR interactions.^223,224 Moreover, the 3ʹ addition of uridine promotes miRNA degradation, which also applies to other small RNAs, such as siRNAs and piRNAs.^225,226 Many studies have demonstrated that uridylation facilitates 5ʹ-to-3ʹ or 3ʹ-to-5ʹ mRNA decay, which is mediated by the recruitment of deadenylases, decapping enzymes, and exonucleases.^202,227 Uridylation also regulates translation efficiency via various mechanisms, for example, mRNA destabilization, and rRNA and tRNA turnover.^{202,209,228,229} Moreover, viral RNA uridylation is involved in antiviral defense.^230,231,232 Uridylated ncRNAs appear overrepresented in exosomes, indicating that uridylation directs RNA sorting into exosomes (Figs. 2, 4 and Table 1).²³³

Uridylation can act on almost all classes of RNAs in eukaryotic cells, further identification of writers and their auxiliary factors in recognizing specific RNA substrates, as well as of erasers and readers that regulate the deuridylation and decide the fate of uridylated transcripts, will no doubt be key to further understanding the regulatory network. The cell-type and disease-specific patterns of uridylation are also crucial in unraveling the roles of uridylation in cellular biology. Viral RNAs are also targets of uridylation, and the contributions of uridylation in fighting viruses and controlling transposons might be interesting topics of future research considering the current epidemic situation of COVID-19.

Adenosine-to-inosine editing

A-to-I editing, which converts adenosines to inosines by deamination in RNA molecules, is a widespread co-transcriptional and post-transcriptional modification in mammals.^234,235 A-to-I editing occurs widely in pre-mRNAs, mRNAs, noncoding RNAs such as miRNAs, lncRNAs as well as tRNAs, and even in virus RNAs (Fig. 1).^{236,237,238,239,240} The most common targets of A-to-I RNA editing are dsRNA hairpin structures forming from inverted Alu repetitive elements, which are located mainly within introns and untranslated regions and fewer in coding exons.^241,242,243

A-to-I editing is the direct conversion of adenosine residues to inosine residues, which is not a conventional “writing” process, so the writers of A-to-I editing are also editors. Adenosine Deaminase TRNA Specific 1 (ADAT1) is responsible for the deamination of adenosine 37 to inosine in eukaryotic tRNA,^239,243,244 while A-to-I conversion at position 34 of certain tRNAs is catalyzed by ADAT2 and ADAT3 (Fig. 5 and Table 1).^245,246,247 Other A-to-I editing events are catalyzed by adenosine deaminases acting on RNAs (ADAR) family members, which are conserved in mammals.^{248,249,250,251,252} ADARs share similar functional domain structures of dsRNA-binding domains (dsRBDs) and a larger catalytic adenosine deaminase domain.^253,254 There are three ADAR members: ADAR1 and ADAR2 deaminate double-stranded (ds)RNA, whereas ADAR3 binds to dsRNA as well as single-stranded (ss)RNA .^236,237,255 ADAR3 lacks editing activity, and it may competitively bind to dsRNAs with other ADARs to decrease the efficiency of these enzymes.^256,257,258 Different from other additive chemical modifications, A-to-I editing may not be further regulated by erasers/modifiers and readers.

Generally, this specific adenosine editing can cause transcriptomic diversity and influence the functions of the target RNAs.^{259,260,261,262} Although the probability of A-to-I editing occurrence in the coding regions is relatively low,^241,263 studies have revealed its role in impacting the protein translation and function by altering the protein codon.^{264,265,266,267} For example, in colorectal cancer, the A-to-I editing of ras homolog family member Q (RHOQ) transcripts results in the substitution of asparagine with serine at residue 136 of RHOQ protein, leading to increased RHOQ activity and cancer invasion potential.²⁶⁴ In UTRs, A-to-I RNA editing can regulate RNA processes including transport, translation, and degradation.^{259,268,269,270,271} For example, ADAR1 directly edits 3ʹ UTR of XIAP and MDM2 mRNAs to promote nuclear retention of these mRNAs.²⁷² A-to-I editing facilitates the recruitment of the stabilizing RNA-binding protein human antigen R (HuR) to the 3ʹ UTR of the CTSS mRNA, thereby enhancing the stability and translation CTSS mRNA.²⁷³ In addition, the A-to-I RNA editing in 3ʹUTR has a potential to block the interaction between miRNAs and target genes to hinder the post-transcriptional repression activity.²⁷⁴ The A-to-I RNA editing in introns usually regulates alternative splicing processes.^268,275 For example, ADAR1 deficiency may cause alternative splicing in intron 27 of the ABCB1 gene to produce transcripts with retained intron, resulting in nonsense-mediated mRNA decay and decreased ABCB1 mRNA stability.²⁷⁶ The A-to-I RNA editing in miRNAs may influence the biogenesis and function of miRNAs.^277,278 A-to-I RNA editing also represses Alu elements in introns to form dsRNA structures, leading to altered linear mRNA and circRNAs generation (Figs. 2, 4 and Table 1).^279,280,281

Some reports indicate that A-to-I RNA editing in pri- or pre-miRNA may induce local structural conformation changes, leading to cleavage suppression and decreased mature miRNA biogenesis.^{282,283,284,285} Conversely, some A-to-I RNA editing may not interfere or promote miRNA biogenesis.^282,286 Specially, ADARs may also bind directly to miRNA precursors to promote miRNA processing by acting as an RNA-binding protein, independent of adjacent A-to-I editing events.^287,288,289 The A-to-I editing in mature miRNAs or siRNAs may impact their target mRNA selection and silencing efficiency.^{290,291,292,293,294,295,296,297} In lncRNAs, A-to-I editing can affect their secondary structures, stability and interactions with other molecules.^238,298,299 For example, the A-to-I RNA editing in lncRNAs may impact lncRNA-miRNA interactions and, consequently, change their miRNA sponge function.³⁰⁰ In tRNAs, A-to-I editing is closely associated with their decoding capacity (Fig. 5 and Table 1).²⁴⁵ In virus RNAs, A-to-I RNA editing can directly target the genome or transcriptome of RNA viruses to regulate viral pathogenicity as well as host innate immune response, which we will discuss in detail below.^301,302,303

Nonetheless, there are many questions that remain to be answered in this field. How the target sites of A-to-I editing are chosen by editors? Although previous studies have identified lots of A-to-I editing sites in human RNA molecules, the significance of such editing for the vast majority of RNA sites remains unclear. Due to A-to-I editing can regulate gene expression through multiple mechanisms, it may be a potential approach to assist or replace RNA interference. Except for influencing miRNA-3ʹUTR and miRNA-lncRNA interaction, A-to-I editing may also affect miRNA-circRNA interaction, which has not been validated yet. Further investigations of RNA editing may provide lessons for precise gene editing.

Roles of RNA modification in immune cell biology

RNA modifications and T lymphocytes

T lymphocytes originate from bone marrow progenitors, mature in the thymus, and are transported to the periphery to fulfill immune functions after activation, proliferation, and differentiation.^304,305 The m⁵C methyltransferase NSUN2 mediates hyperhomocysteinemia-induced interleukin-17A (IL-17A) upregulation by methylating IL17A mRNA and enhancing its translation in T lymphocytes.³⁰⁶ A recent study discovered that the m⁷G cap methyltransferase RNMT plays critical roles in T cell activation by specifically regulating ribosome synthesis.¹⁴² Enzymes modulating miRNA uridylation and uridylated miRNAs are regulated during T cell activation; TUT4 is critical for maintaining miRNA uridylation in the steady state of T lymphocytes and is downregulated during T cell activation, leading to the degradation of uridylated miRNAs.²²⁶ A-to-I RNA editing induced by ADAT1 prevents the sensing of endogenous dsRNAs by MDA5 to participate in thymic T cell maturation, which includes negative selection.^307,308 Specifically, m⁵C and Ψ mRNA modification may be promising in the systemic delivery of nanoparticle formulations for regulating T cell immunity and inflammation.³⁰⁹ Many studies have uncovered the key functions of RNA modifications in the biology of multiple T lymphocyte subsets, which are presented below.

CD4⁺ T cells

Naive CD4⁺ T cells exit the thymus as Th0 cells and differentiate into various cell subsets following different activation signals.^310,311 The best understood effector cell subsets include T helper (Th) cells (Th1, Th2, Th9, Th17, Th22, et al.), T follicular helper (Tfh) cells, and T regulatory (T_reg) cells.^312,313,314 Up to now, there have been some studies revealed that m⁶A participates in the biology of CD4⁺ T cells, as well as several subsets.

First, m⁶A can affect the functions of CD4⁺ T cells. For example, ALKBH5 decreases m⁶A levels in CXCL2 and IFNG mRNA to enhance mRNA stability and translation, thereby promoting CD4⁺ T cell responses.³¹⁵ m⁶A can also influence CD4⁺ T cell differentiation and subset functions, which are discussed in detail below. In particular, as CD4⁺ T cells are the target cells of HIV infection, HIV infection leads to an extensive increase in m⁶A levels in both host and viral mRNAs, thereby influencing HIV replication and viral RNA nuclear export.³¹⁶ During the latent phase of HIV-1 infection, NSUN1 binds with HIV-1 TAR RNA at the 5′ long terminal repeat and generates its m⁵C methylation, and NSUN1 binding with TAR competes with Tat–TAR interaction, leading to hampered HIV-1 transcriptional elongation and viral latency in CD4⁺ T cells.³¹⁷

m⁵C levels and NSUN2 expression are decreased in the CD4⁺ T cells of systemic lupus erythematosus (SLE) patients, and hypermethylated m⁵C in SLE is closely associated with the immune- and inflammation-related pathways, including the immune system, cytokine signaling, and interferon (IFN) signaling.³¹⁸ In the CD4⁺ T cells of SLE patients, ac⁴C modification in mRNAs is highly conserved and enriched in mRNA CDS regions and participates in critical immune and inflammatory signaling in SLE pathogenesis.³¹⁹

Th1/Th2 cells Th1 cells are characterized by the expression of the transcription factor T-bet and IFN-γ secretion, and participate in immune responses against intracellular pathogens.^320,321,322 Th2 cells are characterized by the expression of the transcription factor GAΤA3 and IL-4/5/13 and participate in immune responses against larger extracellular pathogens.^320,321,322 A preliminary study using the clustering method demonstrated that m⁶A may be involved in the Th1/Th2 imbalance and the occurrence of allergic asthma.³²³

Th17 cells Defined by expression of the master transcription factor RORγt and the production of the lineage cytokines IL-17/IL-22, Th17 cells participate in the elimination of bacteria and fungi and in the pathogenesis of autoimmune diseases.^324,325,326 In enterotoxigenic Bacteroides fragilis-induced intestinal inflammation and tumorigenesis, METTL14-dependent m⁶A modification promoted the splicing and generation of miR-149-3p to regulate Th17 differentiation.³²⁷

Tfh cells With Bcl6 as the lineage-defining transcription factor, Tfh cells are a specialized CD4⁺ T cell subset essential for germinal centers and B cell responses.^328,329,330 METTL3/METTL14-catalyzed m⁶A modification of ICOS mRNA suppressed ICOS expression, resulting in impaired Tfh cell differentiation.³³¹ METTL3-catalyzed m⁶A modification on the Tcf7 mRNA 3ʹUTR enhanced the stability of Tcf7 mRNA, ensuring TCF-1 expression in maintaining Tfh differentiation.³³²

T_reg cells Specifically expressing FoxP3 in the nucleus, T_reg cells play immune regulatory roles in maintaining immune cell homeostasis and preventing immunopathology.^333,334,335 Mettl14 deficiency led to the inability to maintain the differentiation of naïve T cells into induced T_reg cells and the Mettl14-deficient T_reg cells exhibited impaired function in suppressing naïve T cell–induced inflammation.³³⁶ Mettl3/m⁶A deficiency in T_reg cells increased Socs mRNA levels, leading to deactivation of the IL-2–STAT5 signaling which is integral in maintaining the functions and stability of T_reg cells.³³⁷

CD8⁺ T cells

Naive CD8⁺ T cells proliferate and differentiate into various effector and memory cell types following different activation signals. CD8⁺ T cells can persist for years and are involved in protective immunity against intracellular pathogens and tumors.^{338,339,340,341} Many studies have demonstrated that m⁶A methylation regulators are closely associated with CD8⁺ T cell infiltration in various cancers.^{342,343,344,345} Furthermore, m⁶A methylation regulators are involved in regulating CD8⁺ T cell functions. For example, Ythdf1-deficient mice exhibited an elevated antigen-specific CD8⁺ T cell antitumor response.³⁴⁶ In tumor-associated macrophages, METTL14 deficiency led to anomalous CD8⁺ T cell differentiation, driving CD8⁺ T cell dysfunction and repressing CD8⁺ T effector cell activation.³⁴⁷ Tumor-intrinsic FTO restricted the activation and effector states of CD8⁺ T cells; knockdown of FTO impaired tumor cell glycolytic activity, which restored CD8⁺ T cell function.³⁴⁷ m¹A levels were negatively related to CD8⁺ T effector cell proliferation in colon cancer.³⁴⁸

RNA modifications and B lymphocytes

Generally, B lymphocytes are well known for their function of producing antibodies in the adaptive immune response; they are also key modulators of the innate immune response.^349,350,351 Under antigen stimulation, mature B cells are activated and differentiate into memory B cells or plasma cells, which secrete antibodies.^352,353,354 Some studies have reported that m⁶A modification also participates in B cell biology. For example, METTL14 deficiency inhibits mRNA m⁶A methylation in developing B cells and blocks IL-7-induced pro-B cell proliferation and the large-pre-B to small-pre-B transition, resulting in severe B cell development stagnation in mice.³⁵⁵ The RNA exosome cofactor MPP6, m⁶A modification, and m⁶A readers play vital roles in modulating lncRNA processing, DNA recombination, and development in B cells.³⁵⁶ m⁶A methylation was significantly decreased in the plasma cells of patients with multiple myeloma, which was due to the upregulation of FTO; FTO facilitated multiple myeloma cell proliferation, migration, and invasion by targeting HSF1–HSPs in a YTHDF2-dependent manner.³⁵⁷ In addition, ADAR1 is essential for normal B lymphopoiesis in the bone marrow and peripheral maintenance.³⁵⁸

RNA modifications and DCs

DCs are key regulators of the innate and adaptive immune responses. They integrate signals from pathogens or other damage and present processed antigens to naïve T cells to control T cell differentiation.^{359,360,361,362} Similar to other immune cell types, the expression of the m⁶A methylation regulators in diseases is also associated with DC infiltration or depletion.^363,364 Specially, chemokine receptor 7 (CCR7) increases lnc-Dpf3 expression by reducing m⁶A modification to prevent its degradation, and lnc-Dpf3 functions in the feedback control of DC migration and inflammatory responses by coupling the epigenetic and metabolic pathways.³⁶⁵ YTHDF1 identifies m⁶A-modified mRNAs encoding lysosomal proteases and promotes the translation of these transcripts in DCs, thereby suppressing the cross-presentation of wild-type DCs.³⁴⁶ DCs exposed to m⁵C-, m⁶A-, m⁵U-, s²U-, or Ψ-modified RNAs express decreased cytokines and activation markers, suggesting that nucleoside modifications repress the latent capacity of RNAs to activate DCs.⁸ Mettl3-mediated m⁶A modification maintained DC maturation and activation by promoting the translation of key factors, including CD40, CD80, and the TLR signaling adaptor TIRAP.³⁶⁵ Recognition of mRNA m⁶A methylation by YTHDF1 promoted the translation of lysosomal proteases in DCs and suppressed cross-priming of CD8⁺ T cells, resulting in defective immune recognition and tumor immune evasion.³⁴⁶ ADAR1 is required for the differentiation, functionality, and survival of DCs and alveolar macrophages, which involves the A-to-I editing of several coding genes and lncRNAs.³⁶⁶

RNA modifications and natural killer cells

Natural killer (NK) cells are cytotoxic lymphocytes of the innate immune system characterized by target cell killing and cytokine production, functioning in controlling viral and intracellular bacterial infections and tumors, as well as regulating other immune cells.^{367,368,369,370} m⁶A also influences the functions of NK cells. METTL3-mediated m⁶A methylation guaranteed the sufficient response of AKT and MAPK signaling to IL-15 by raising SHP-2 expression, thus exerting critical roles in maintaining NK cell homeostasis and anti-tumor immunity.³⁷¹ YTHDF2 is increased in NK cells activated by cytokines, tumors, and cytomegalovirus infection, and is essential for maintaining NK cell homeostasis and maturation; YTHDF2 is also required for IL-15–mediated NK cell survival, proliferation, and effector functions by forming a STAT5–YTHDF2 positive feedback loop.³⁷² In addition, YTHDF2 modulates NK cell proliferation and division partially via reducing Tardbp mRNA stability.³⁷²

RNA modifications and monocytes or macrophages

Monocytes and macrophages play an essential role in the innate immune system and present phagocytic activity to exhibit antimicrobial, homeostatic, and immunoregulatory functions.^{373,374,375,376} Due to the wide application of monocyte/macrophage cell lines such as THP-1 and RAW264.7,^377,378 there have been many studies investigating the roles of RNA modifications in monocytes or macrophages.

Some recent studies demonstrate that m⁶A modification plays critical roles in the antiviral immunity of monocytes and macrophages. For example, m⁶A-modified HIV-1 RNA escaped RIG-I-mediated RNA sensing and IFN-I-mediated innate antiviral immune responses in differentiated human monocytic cells and primary monocyte-derived macrophages.³⁷⁴ After vesicular stomatitis virus infection, METTL3 in monocytes/macrophages translocated to the cytoplasm to promote m⁶A modification of viral RNAs. Then, the m⁶A-modified viral RNAs were reshaped with decreased double-stranded RNA loads to restrain innate sensing efficacy by MDA5 or RIG-I, resulting in inactivation of the global innate immune signaling pathways.³⁷⁹ In response to viral infection, macrophages impaired ALKBH5 enzymatic activity and induced m⁶A modification-mediated inactivation of the OGDH–itaconate pathway to inhibit viral replication.³⁸⁰ In response to DNA viruses, HNRNPA2B1 promoted m⁶A modification and nucleocytoplasmic trafficking of CGAS, IFI16, and STING mRNAs, thereby triggering the downstream cytoplasmic TBK1–IRF3 signaling in macrophages.³⁸¹ DDX46 recruited ALKBH5 via its DEAD helicase domain to demethylate m⁶A-modified antiviral transcripts, impeding their nuclear exportation and translation and resulting in impaired IFN production and antiviral innate responses.³⁸² Under homeostatic conditions, YTHDF3 cooperated with PABP1 and eIF4G2 to enhance FOXO3 translation by binding to the translation initiation region of FOXO3 mRNA and functioning as a negative regulator of antiviral immunity.³⁸¹ BCG vaccine exposure can cause increased ADAR1 expression and subsequent enhanced A-to-I editing events in human macrophages to participate in trained immunity.³⁸³

m⁶A modification also regulates monocyte inflammation and immune activity. METTL3-mediated m⁶A modification and YTHDF2-mediated recognition promoted PGC1A mRNA degradation, leading to insufficient ATP production and excessive reactive oxygen species accumulation in monocyte inflammation.³⁸⁴ In the peripheral blood immune cells from patients with colorectal cancer, m⁶A modification was the most abundant in monocytes, and the m⁶A levels in the monocytes were negatively related to the monocyte immune response.³⁸⁵

m⁶A modification is involved in various aspects of macrophage biology, including polarization, differentiation, activation, inflammation, and pyroptosis.^{386,387,388,389} For example, IGF2BP2 reads the m⁶A modification on TSC1 and PPARG mRNA to regulate TSC1 and PPAR-γ expression, thereby skewing M1 macrophages to M2 activation through the TSC1–mTORC1 pathway and PPAR-γ-mediated fatty acid uptake.³⁸⁶ METTL3-mediated m⁶A modification of Irakm mRNA accelerated its degradation, resulting in TLR signaling-mediated macrophage activation.³⁸⁶ METTL3 increased MALAT1 levels through m⁶A methylation to downregulate USP8; the reduced USP8 decreased TAK1 ubiquitination and degradation, which promoted macrophage pyroptosis and inflammation.³⁸⁹ TUT7 functioned as a regulator in TLR4-mediated inflammation in macrophages by uridylating and thereby destabilizing the mRNAs of inflammatory mediators, including Zc3h12a.²²⁷ Through targeting the miR-21 precursor, ADAR1 reduces the generation of mature miR-21, then facilitating the polarization of macrophages toward the M2 phenotype via regulating the Foxo1-IL-10 axis.³⁹⁰

RNA modifications and granulocytes

It is well known that granulocytes are divided into three types—neutrophils, eosinophils and basophils.^391,392 There are relatively few and superficial studies on m⁶A regulation of granulocytes. m⁶A modification on c-Rel and Rela mRNA inactivated the NF-κB pathway to suppress IL-8 secretion, thereby inhibiting neutrophil infiltration in papillary thyroid cancer progression in a METTL3- and YTHDF2-dependent manner.³⁹³ Other studies only found that the expression of m⁶A methylation regulators in tumors was associated with the infiltration of granulocytes, especially neutrophils.^393,394,395

To summarize, for functions, RNA modifications regulate various biological processes of immune cells, including development, differentiation, activation, migration and polarization, thus modulating the immune responses. For molecular mechanism, RNA modifications target immune cell RNAs that are responsible for those biological processes and influence RNA processes including generation, transportation, function and metabolization, leading to alterant immune cell biology. However, there are many kinds of RNA modifications and their functions are complex; immune cells are also diverse, and each cell type has its own unique cellular processes. Therefore, although the mechanism presented above is a common one explicating the interaction between RNA modifications and immune cells, the roles of a certain RNA modification in a specific immune cell need to be concretely investigated. Since the relevant research is still in its infancy, more work is needed to further improve the interaction network between RNA modification and immune cells.

Roles of RNA modifications in immune related diseases

The immune system, consists of innate and adaptive immune, functions in the host defense against harmful antigens and immune homeostasis.^396,397,398 Immune cells are important constituents of the immune system, the dysregulation of which can result in immune related diseases, such as cancers, infection, inflammatory disorders, and autoimmune diseases.^{399,400,401,402} Therefore, via regulating the biological processes of immune cells, RNA modifications can participate in the pathogenesis of immune related diseases.

Cancers

Although previous research on cancers mainly focused on the malignant phenotypes of the cancer cell itself, in recent years, there are more and more studies on anti-tumor immunity, such as immune checkpoint, immune cell infiltration, cancer immune escape, and cancer immunotherapy.^{403,404,405,406,407} RNA modifications have been widely investigated in cancers, and they play vital roles in various cellular biology aspects of cancer cells, such as proliferation, metastasis, metabolism, apoptosis, and treatment resistance.^{1,15,18,79,175} Relatively, the roles of RNA modifications mediating immune cell biology in tumor immunization are not extensively and profoundly considered.

The most reported RNA modification mediating immune cells in cancers is their influence on immune cell infiltration of tumors.^408,409,410 m⁶A modification and multiple m⁶A regulators have been verified to be closely associated with the infiltration of various immune cells in plenty of human cancers.^408,409,411 Relatively less, m⁵C, m¹A, m⁷G, ac⁴C, and Ψ are also found to be related to immune cell infiltration in cancers (Fig. 6).^{412,413,414,415,416,417} However, most of these studies investigated the RNA modifications and regulators in cancer tissues and cells, but not in immune cells, and they did not elaborately explain how RNA modification disorders affect immune cell infiltration. Here, we propose an idea that chemokines secreted by tumors may be an intermediate medium regulated by RNA modification in this process.

Fig. 6

RNA modifications and immune cells in diverse cancers. RNA modifications, especially m⁶A modification, mainly play a positive role in regulating immune function in various cancers as illustrated in the figure. In melanoma and lung cancer, FTO-mediated m⁶A demethylation in tumor cells elevates the transcription factors c-Jun, JunB, and C/EBPβ, allowing the rewiring of glycolytic metabolism, thereby restricting the function of CD8⁺ T cells and inhibiting tumor growth. Besides, ablation of Mettl3 in myeloid cells promotes tumor growth and metastasis via impairing the YTHDF1-mediated translation of SPRED2, which enhances the activation of NF-kB and STAT3 through the ERK pathway, thereby increasing M1/M2-like tumor-associated macrophage and regulatory T cell infiltration into tumors. In melanoma, upregulation of YTHDF2 in NK cells promotes NK cell effector function and is required for IL-15–mediated NK cell survival and proliferation by targeting Tardbp. In melanoma and colorectal carcinoma, loss of YTHDF1 in classical DCs enhanced the cross-presentation of tumor antigens and the cross-priming of CD8⁺ T cells via increasing the m⁶A sites on transcripts encoding lysosomal proteases recognized by YTHDF1, which could be written by METTL14. In colorectal carcinoma, Mettl3- or Mettl14-deficient macrophages showed faster tumor growth via slowing down the degradation of Irakm, encoding a negative regulator of TLR4 signaling, or driving CD8⁺ T cells to dysfunctional ones by directly targeting Ebi3, respectively. Moreover, elevation of METTL14 in colorectal cancer cells promotes the differentiation of CD4⁺ T cells into Th17 cells via exosomes included miR-149-3p. In Thyroid carcinoma, METTL3 reduction in PTC cells recruits tumor-associated neutrophils into tumor tissue through IL-8, thereby further promoting tumor development, while in ovarian carcinoma, depletion of METTL3 in NK cells inhibits cell infiltration ability and function, leading to accelerated tumor development via reducing SHP-2 expression as well as the activation of AKT and MAPK signaling pathway. In multiple myeloma, upregulation of the demethylase FTO in plasma cells plays a tumor-promoting and pro-metastatic role in MM by targeting HSF1 which could be recognized by YTHDF2. In pancreatic ductal adenocarcinoma, LncRNA-PACERR increased the number of M2-polarized cells and facilized cell proliferation, invasion and migration via binding to IGF2BP2 to enhance the stability of KLF12 and c-myc, thereby activating KLF12/p-AKT/c-myc pathway through binding to miR-671-3p. Extensive bioinformatics analysis revealed the potential key roles of RNA modifications other than m⁶A modification in immune cell infiltration in diverse types of tumors. DC dendric cell, NK T cell natural killer T cell, PTC papillary thyroid carcinoma, HCC hepatocellular carcinoma, ccRCC clear cell renal cell carcinoma, TAM tumor-associated macrophage, HSF1 heat shock factor 1

There have been some studies exploring the RNA modification dysregulations in immune cells in the tumor immune microenvironment (TIME) and their roles in cancer progression.^{13,418,419,420} As expected, m⁶A is the most extensive and in-depth modification inquired. METTL3 is downregulated in tumor-infiltrating NK cells, which affects Ptpn11 m⁶A modification and downstream IL-15-induced signaling, leading to homeostasis disruption, impaired infiltration, and function of NK cells in TIME, resulting in cancer development.³⁷¹ Through targeting and inhibiting the stability of Tardbp mRNA, YTHDF2 is involved in maintaining NK cell homeostasis, maturation, IL-15–mediated survival, and antitumor activity.³⁷² m⁶A modification can also influence macrophage reprogramming by mediating SPRED2 translation, and, thereby regulating the activation of NF-κB and STAT3 signaling; METTL3 deficiency impairs the YTHDF1-mediated translation of SPRED2, orchestrating growth, metastasis, and anti-PD-1 therapeutic efficacy of cancer.⁴²¹ Another study confirmed this biological effect of m⁶A in macrophages from another aspect that loss of Mettl3 impairs the TLR4 signaling in macrophage activation by reducing Irakm mRNA degradation.³⁸⁶ Additionally, lncRNA-PACERR induces the polarization of pro-tumor macrophages in an IGF2BP2 and m⁶A-dependent manner.⁴²² YTHDF1 negatively regulates the anti-tumor immune responses of DCs by promoting the translation of m⁶A-modified mRNAs encoding lysosomal proteases to impair immune recognition, leading to tumor immune evasion.³⁴⁶ In T cells, m⁶A modification targets specific genes to control T cell differentiation and maintains the suppressive effects of Tregs, functioning as a negative regulator in the anti-tumor immune responses.³³⁷ CD8⁺ T cells are direct effector cells of anti-tumor immunity, but there are few studies revealing the m⁶A disorders in this cell type in cancers. Nevertheless, m⁶A can influence the anti-tumor response of CD8⁺ T cells via controlling the biological processes of other related cells in the TIME, such as tumor cells, macrophages, and DCs.^346,347,423 Especially, the progression of multiple myeloma, a B-cell lymphoma, is mediated by m⁶A in an FTO and YTHDF2 dependent manner.³⁵⁷ Relatively, there are no comprehensive studies on the biological functions and molecular mechanisms of other modifications regulating immune cells in cancers till now. (Fig. 6 and Table 2).

Table 2 Main functions of RNA modifications on various immune cells

Despite all this, the roles of RNA modifications mediating immune cell biology in cancer immunization remains largely unclear. There seem to be some paradoxes as well as enlightenments. For instance, as we discussed above, m⁶A deficiency will lead to the disability of some anti-tumor immune cells, whereas YTHDF1 deficiency enhances anti-tumor immune responses. In this regard, we think researchers should comprehensively consider the other roles of writers, erasers, and readers, not just focusing on their regulation of RNA modifications. Moreover, as we know, RNA modifications are vital modulators of normal cell biology, and it is easy to understand that their delicacy may cause the disability of immune cells; but we don’t know what will happen if these modifications are excessive, and whether there is a balance. Besides, the level and functions of some modifications, such as m⁶A, are diverse between cancer cells and infiltrated immune cells.^1,15 This indicates researchers to separate cancer cells and infiltrated immune cells when analyzing human or animal tumor samples.

Infectious diseases

Similar to cancers, the pathogenesis and development of infectious diseases are closely related to the immune defense deficiency that involves the deficiency of the immune system itself and immune escape from pathogens.^{424,425,426,427} According to existing literature, RNA modifications are critical participators in the progression of infectious diseases by affecting the biology of immune cells.

In recent years, many studies have focused attention on the roles of RNA modifications in viral infection.^{53,302,428,429,430} On one aspect, RNA modifications, such as m⁶A, m⁵C, ac⁴C, Ψ, and RNA editing, directly act on viral RNAs, thus influencing RNA structure, RNA nuclear export, translation, stability, and replication.^{53,430,431,432,433} On the other aspect, RNA modifications can regulate host responses to viral infection by mediating viral RNA sensing and signaling, cytokine responses, as well as immune cell functions, which are emphasis discussion of this text. The roles of RNA modifications in regulating immune cell functions in antiviral infection can also be explicated from two perspectives. The first one is that RNA modifications on viral RNAs repress innate immune signaling pathways. For example, m⁶A-modified HIV-1 and vesicular stomatitis virus RNAs restrain the innate sensing efficacy of MDA5 or RIG-I and thereby impaired IFN-I-mediated innate antiviral immune responses in monocytes and macrophages.^374,379 The other one is that RNA modifications affect the key factors of antiviral immunity in immune cells, especially in innate immune cells. For example, in monocytes and macrophages, m⁶A modification affects antiviral transcripts including CGAS, IFI16, STING, Mavs, Traf3, Traf6, and FOXO3, as well as signaling pathways including OGDH–itaconate, TBK1–IRF3, and IFN signaling to function in inhibiting viral replication and antiviral innate immunity.^{380,381,382,434} In NK cells, except for the antitumor activity, YTHDF2 is also essential for the antiviral activity of NK cells by targeting Tardbp.³⁷² Especially, as CD4⁺ T cells are the target of HIV infection, RNA modification participates in the viral processes including replication, nuclear export, transcriptional elongation, and viral latency in CD4⁺ T cells via modulating biological processes inside CD4⁺ T cells as described above. After COVID-19 infection, A-to-I editing of endogenous Alu RNAs is decreased in normal human lung cells and in lung biopsies, which may represent the responses of the hosts.⁴³⁵ ADAR1 mediated RNA editing on extensive duplex RNA structures can lead to repressed innate immune responses and is profitable for viral replication, which indicates that A-to-I editing prevents autoimmunity while also favoring pathogens (Table 3).⁴³⁶

Table 3 RNA modifications and immune cells in infectious, inflammatory and autoimmune diseases

Unfortunately, there is no report on the roles of other RNA modifications other than m⁶A or A-to-I editing in regulating immune cells in infectious diseases. Also, there is few study on the roles of RNA modifications regulating immune cells in infectious diseases induced by other pathogens such as bacteria and fungi. In addition, as we reviewed above, the roles of RNA modifications in antiviral processes are reported only in innate immunity, while its functions in adaptive immunity are ignored and may be of good research interest. Due to the pathogenesis similarity in immune defense deficiency, the research ideas on cancer may provide lessons for infectious diseases in this field.

Inflammatory and autoimmune diseases

Inflammation and immune responses are critical in opposing harmful stimuli and injury, while their overreaction or being out of control will lead to inflammatory and autoimmune diseases, causing tissue damage and organ dysfunction.^437,438,439 Generally speaking, inflammatory diseases and autoimmune diseases are different, but they share some similar pathogenesis,^440,441,442 so we discuss them together here. To date, there have some studies verified that RNA modifications may exert biological functions during inflammation and autoimmunity by regulating immune cells.

In hyperhomocysteinemia, NSUN2 upregulates IL-17A expression by inducing IL-17A mRNA m⁵C modification in T lymphocytes to mediate chronic inflammation.³⁰⁶ In allergic asthma, m⁶A may participate in the Th1/Th2 imbalance.³²³ In inflammatory bowel disease (IBD), m⁶A modification may affect immune infiltration and therapeutic response,¹⁵³ and Mettl14 deficiency can cause impaired induction of naïve T cells into iTreg cells by decreasing RORγt expression, thereby leading to spontaneous colitis.³³⁶ Deficiency of Adar1 mediated impaired A-to-I editing in CD4⁺ T cell will lead to abnormal thymic T cell maturation and impaired negative selection, thereby resulting in autoimmunity disease such as spontaneous colitis.^307,308 By regulating various aspects of macrophage biology, including polarization, differentiation, activation, inflammation, and pyroptosis, m⁶A has been found to serve as a mediator in inflammatory diseases including ulcerative colitis, cytokine storm after bacterial infection and liver fibrosis (Table 3).^387,389,443

In SLE, m⁵C level and NSUN2 expression are decreased in CD4⁺ T cells, and hypermethylated m⁵C is related to the immune- and inflammation-related pathways.³¹⁸ ac⁴C modification in mRNAs of SLE CD4⁺ T cells is highly conserved and enriched in CDS regions and involved in the immune and inflammatory signaling of SLE pathogenesis.³¹⁹ ADAR1 mRNA was significantly up-regulated in SLE T cells, which may be a potential mechanism accounting for the mutations in the RI alpha subunit of type 1 protein kinase A.⁴⁴⁴ In SLE and some other autoimmune diseases, increased A-to-I RNA editing is involved in generating or elevating the autoantigen load to facilitate autoimmunity progression.⁴⁴⁵ In autoimmune encephalomyelitis, ablation of ALKBH5 resulted in increased m⁶A modification on IFNG and CXCL2 mRNA, as well as impaired responses of CD4⁺ T cells to repress autoimmunity.³¹⁵ In systemic autoimmunity characterized by chronic or acute type I IFN pathway such as systemic sclerosis and other disease contexts, ADAR1p150 isoform mediated A-to-I editing in PBMCs are closely related to type I IFN responses.²⁷⁰ ADAR1 can edit putatively immunogenic dsRNA substrates to evade MDA5-mediated dsRNA sensing to suppress inflammatory and autoimmune diseases.⁴⁴⁶ In the allogeneic transplant model, ADAR1 mediated RNA editing can suppress the host antigraft response and promote graft survival through the ADAR1-miR-21-Foxo1-IL-10 axis (Table 3).³⁹⁰

Although there are few studies on RNA modification regulating immune cells in inflammatory and autoimmune diseases, as we described above, many studies have confirmed the modulation of RNA modifications on the biology of immune cell closely related to these diseases, which build potential connections between cell biology and human diseases. For example, RNA modifications can influence the biology of DCs, monocytes and macrophages, and mediate inflammatory cytokine secretion,^{8,142,365,384,447} they also probably function in many inflammatory diseases. RNA modifications can affect the activation of T cells and differentiation of Th17 cells, Tfh cells and Treg cells,^{142,327,332,337} which play critical roles in the immune dysregulation of some autoimmune diseases such as SLE and multiple sclerosis,^448,449 they are likely to participate in the pathogenesis of these diseases. In addition, the research on cancers and infectious diseases listed above may also provide feasible ideas. Due to the contrary roles of immune cell activation and infiltration in cancers and infectious diseases with inflammatory and autoimmune diseases, it is possible that the roles of RNA modifications regulating immune cells are opposite in these diseases.

Conclusions and perspectives

In this review, we introduced eight RNA modifications including m⁶A, m⁵C, m¹A, m⁷G, ac⁴C, Ψ, uridylation, and A-to-I editing, and summarized their influence on the biology of immune cells, as well as their roles in immune related diseases by regulating immune cells. The modifications involve various kinds of RNAs, such as mRNAs, noncoding RNAs, tRNAs, rRNAs, or even exogenous RNAs (eg., viral RNAs and synthetic RNAs). These modifications are executed by RNA-modifying enzymes such as writers, erasers and readers, and influence RNA processes including generation, transportation, function, and metabolization. Based on these molecular functions, RNA modifications participate in various biological processes of immune cells, including development, differentiation, activation, migration, and polarization, thus modulating the immune response and participating in the pathogenesis of immune related diseases (Fig. 6 and Table 3).

Currently and a long time in the future, m⁶A will continue to be a research hotspot in this field, especially in the field of anti-tumor immunity, with relatively adequate background knowledge and mature research technology available. In addition, RNA modifications are mostly investigated in T lymphocytes and monocytes/macrophages, which are relatively easy to obtain, and in T cells or monocytes/macrophages related immune processes and diseases. Moreover, due to the global epidemic of COVID-19, RNA modifications that are involved in antiviral immunity may also be of promising research interest.

There remain many questions that should be addressed to fully understand the impact of RNA modifications on immune cell biology. Nevertheless, there is continuous progress in this field and there are presently many meaningful points to study. First, to what extent do RNA modifications regulate immune cell biological processes? Are they the main mediators or auxiliary participants? In other words, how helpful would it to rectify RNA modification abnormalities in the treatment of immune related diseases? This is important for designing therapeutic targets and a critical issue to be solved. Second, some RNA modifications have been proven feasible in the systemic delivery of nanoparticle formulations for regulating both immune cell immunity and inflammation in the laboratory.^309,450,451 Clinical practice is still the grandest challenge, which will be a valuable research direction. Third, essentially, RNA modifications and their functions are regulated by writers, erasers and readers. However, many studies have failed to elucidate the mechanisms causing the anomalies of these regulators. As RNA modifications can target many downstream RNAs and there are no effective interfering drugs, research on the upstream factors is particularly important. Fourth, further exploration of the role of RNA modifications in various immune cells needs to be done, since the current research is mainly focused on a few modifications (e.g., m⁶A) and immune cells (e.g., T lymphocytes) while overlooking other modifications and immune cells. Immune cells are interconnected and RNA modifications regulate various immune and inflammatory-related factors and signaling pathways. This is a complicated regulatory network that requires us to consummate. Most important, targeting RNA modifications as a treatment in immune related diseases remains in the theoretical stage, and there are no clinical application examples at present. RNA modifications may influence almost all types of RNA, and interfering RNA modifications may cause a wide range of effects. Therefore, gene-specific RNA modification interference is a vital research bottleneck in this area.