Skip to main content

National Institutes of Health

Eunice Kennedy Shriver National Institute of Child Health and Human Development

2021 Annual Report of the Division of Intramural Research

RNA Metabolism in Cell Biology, Growth, and Development

Richard Maraia
  • Richard J. Maraia, MD, Head, Section on Molecular and Cellular Biology
  • Sergei Gaidamakov, PhD, Biologist
  • Sandy Mattijssen, PhD, Staff Scientist
  • Alex Vassilev, PhD, Staff Scientist
  • Alan Kessler, PhD, Postdoctoral Fellow
  • Abdul Khalique, PhD, Visiting Fellow
  • Amitabh Ranjan, PhD, Visiting Fellow
  • Leah Pitman, BS, Postbaccalaureate Fellow
  • Jack Prochnau, BS, Postbaccalaureate Fellow

We are interested in tRNAs and mRNAs as well as some of their interacting proteins and how the pathways involved in their biogenesis, maturation, and metabolism intersect with processes critical to cell proliferation, growth, and development during health and disease. tRNAs are produced at over 10-fold higher molar levels than are ribosomes during cellular proliferation. After synthesis by RNA polymerase III (RNAP III, also known as Pol III), post-transcriptional tRNA processing steps and modifications occur prior to nuclear export and additional modifications. Failure to support programmed mRNA translation is evident by developmental and other diseases attributable to defective tRNA biogenesis.

RNAP III synthesizes high levels of tRNA by a conserved process of transcription termination–associated reinitiation, relevant to development and cancer. Termination occurs within a short tract of T residues in the non-template (NT) DNA strand at the ends of RNAP III–transcribed genes. The nascent transcripts bear a copy of this terminator, (U)UU-3′OH, a recognition motif for the nuclear La protein, which binds in a sequence- and length-dependent manner. Notably, 3′U(n) length-dependence of La binding was found to be one residue shorter than the minimal T-length required for efficient RNAP III termination, the latter of which is 6, 5, and 4 Ts for S. cerevisiae, S. pombe, and human, respectively. This suggests a La link with tRNA expression, a link that fits with human (h)RNAP III that evolved a minimal 4T termination mechanism, and data suggest that this may direct some post-transcriptional events. Studies on development, structure, and its gene variants indicate that hRNAP III evolved beyond housekeeping via tRNA expression, to more intricate control of gene-regulatory programs.

The human La protein is a target of auto-antibodies in patients with chronic inflammatory disease such as systemic lupus erythematosus, Sjögren’s syndrome (SS) (La is also known as SS antigen-B, SSB), and neonatal lupus. Although La is active in other immune pathways, evidence of its primary involvement in autoimmunity is lacking. As noted above, La binds to RNAP III transcripts in all cells and serves as a chaperone during their nuclear maturation. La not only protects pre-tRNAs from 3′ exonucleases, it directs and temporally orders the first step in the 5′ processing by RNase P pathway. We propose a role for La in a novel pathway, different from tRNA maturation. Our data suggest that La is a determinant for a subset of pre-tRNAs (genes) to enter an alternative pathway to specific activation of the type-I interferon (IFN) response. Accordingly, some dual-activity tRNA genes may serve as endogenous immune adjuvants or may contribute to autoimmunity. We are currently exploring these possibilities.

We also work on La-related proteins (LARP) 1 and 4, which directly bind to 3′ poly(A) and poly(A)–binding protein (PABP) to modulate mRNA stability. LARP4 mRNA itself bears a coding determinant that can sense tRNA levels, with potential for translation homeostasis. A larger theme is that translational fidelity via tRNAs, to meet mRNA demands, is a key element of health and disease. Our goal is to develop knowledge and potential for translational research.

A view of tRNA synthesis, processing, modification, and mRNA activity

A major part of our work centers on processes involved in tRNA biogenesis and metabolism and on their function in translation. The rate of RNA production by RNAP III was shown to affect the efficiency by which pre-tRNAs acquire the common m2,2G26 modification, which increases tRNA activity. Disrupted tRNA biogenesis, attributable to gene variants, leads to developmental and other diseases, and to neurodegeneration and intellectual impairment. About 120 modifications occur on tRNAs, of which about 40 have been documented for human cytoplasmic (cy-) tRNAs and several others for human mitochondrial (mt-) tRNAs. Many tRNA–modification enzymes (TME) are multisubunit, and some composite modifications require several gene products. Thus, mutations in numerous TMEs cause disease.

Critical modifications to tRNA anticodon loops fine-tune base pairing for optimal decoding and wobble decoding of synonymous codons, which is an important component of a tunable system of high-fidelity cy-translation. It is important to note that the 61 sense codons for 20 amino acids (aa) are decoded by cy-tRNAs, which collectively carry only about 45 anticodons, a feature of tRNA genes, named anticodon-sparing, that is widely conserved, although the number and identity of the 'missing' anticodons differ among kingdoms, anti-correlated with wobble-base modifications. In humans, about 15 standard codons must be wobble-decoded by cy-tRNAs whose activities are controlled by anticodon modifications. Thus, RNAP III transcription of tRNA genes is only one level on which the translation of a range of cy-mRNAs with biased codon content can be regulated. While this constitutes a nuclear-cytoplasmic translation system with much potential for intricate regulation, the complexities of overlapping/redundant decoding activities also create potential for translation infidelity in cases of faulty modification or tRNA pool imbalance.

All TMEs are nuclear-encoded, and a modest subset modify cy- and mt-tRNAs, with potential to synchronize translation in both compartments. Distinct decoding rules apply in mitochondria, in part because each mt-DNA encodes only one tRNA for each of 18 amino acids. Synonymous codons occupy mt-mRNAs but are not distinguished by different tRNAs, as is the case for cy-mRNAs. For example, while four codons each for Ala, Gly, Pro, Thr, and Val are decoded by three cy-tRNAs, a single mt-tRNA for each must wobble-decode its four cognate codons. Furthermore, these mt-tRNAs use unmodified U34 as their wobble base, whereas cy-tRNA wobble bases are part of an elaborate modification system, especially for U34. Each mt-tRNA is critical; point mutations impair mt-translation with unique association with oxidative phosphorylation diseases, whereas most nuclear tRNA genes are buffered by multiple copies; a known exception is a unique, single-copy, brain-specific tRNA. Yet, some diseases caused by a mutation in a mt-tRNA exhibit a phenotype similar to that resulting from a nuclear gene mutation to a TME whose substrate is the mt-tRNA.

Mechanisms linking tRNA homeostasis and neurologic disease may vary. One model is that nerve tissue is highly susceptible to low-fidelity translation, leading to protein aggregation and altered proteostasis, as observed with TME deficiencies (also in yeast). We showed that La deletion from mouse brain soon after birth perturbed tRNA processing and initiated inflammation and neurodegeneration. La is a general pre-tRNA factor that acts redundantly with TMEs, and thus tRNA homeostasis is likely to be disrupted by its deficiency or dysfunction.

Numerous pathogenic alleles that encode defective subunits of RNAP III cause a type of leukodystrophy known as hypomyelinating leukodystrophy (HLD), whose pathobiology is consistent with poor formation of axonal myelin sheaths rather than demyelination. While RNAP III produces several ncRNAs (non-coding RNAs) other than tRNAs, HLDs and disorders of RNAP III–associated factors are considered tRNA-opathies that disrupt translation. We proposed models in which tRNA imbalances may alter cell type–specific mRNA translation profiles. Even subtle tRNA imbalances applied across mRNA populations may challenge proper translation in the developing CNS. Changes in the rate of RNAP III output can not only alter relative tRNA levels but also their modification status in a way that can differentially impact their decoding activities. Curiously, codon use by brain-specific genes is more conserved than by other tissue-specific genes, and function-related genes are intolerant of variation in synonymous codons. Other models suggest that deficiency in brain-specific RNAP III transcripts as well as tRNAs contribute to hypomyelinating leukodystrophy and other CNS–specific phenotypes.

We suggest that another disruption mechanism to a tRNA biogenesis–CNS network is also possible. As TME defects often manifest as neurological, reflecting tissue reliance on mitochondria, and because cy- and mt-translation are synchronized including by tRNA modifications and other factors, disruption of general tRNA homeostasis may impair mt-function with pathophysiologic contribution to disease, including in developing oligodendrocytes.

mRNA levels are determined in significant part by levels of cognate tRNAs, from yeast to human. In addition, certain mRNAs are particularly sensitive to tRNA levels. The LARP4 mRNA has a tract of about 70 codons with a very poor match to cellular tRNA levels. Thus, unprogrammed changes in tRNA output (e.g., genetic mutation in RNAP III) could shift mRNA stabilities and translation efficiencies in unpredictable ways with uncertain outcomes.

Translation of LARP4 mRNA produces LARP4, which binds to poly(A) and also to poly(A)–binding protein (PABP/PABPC1). LARP4 stabilizes mRNAs by opposing deadenylation of their poly(A) tails, substrates of Ccr4-Not deadenylase (a multiprotein complex that functions in gene expression in the nucleus, where it regulates transcription, and in the cytoplasm, where it associates with translating ribosomes and RNA processing bodies). Potential for regulation is that Ccr4-Not monitors mRNA–ribosomes for codon–tRNA match. For ribosomal protein–encoding mRNAs stabilized by LARP4, this supports a working model in which LARP4 mRNA senses tRNA levels and relays this by producing LARP4 to regulate ribosome biogenesis, perhaps with LARP1.

Human RNAP III and associated factors have been known to be dysregulated in cancer. More recently, functionally related mRNAs favoring proliferative or differentiation states were shown to be biased in synonymous codons and in the cognate tRNAs differentially expressed in those cells. Strikingly, the first hRNAP III structures provide insight into activities specific to the higher eukaryotic 17–subunit complex. First, the cancer-associated RPC7a subunit paralog (encoded by POLR3G) appeared to interfere with binding of the RNAP III–negative regulator and tumor suppressor MAF1, whereas the RPC7b (POLR3GL) paralog subunit is enriched in cells programmed for differentiation (limited proliferation). Yeast RNAP III has only one homolog. The second example involves the most striking feature of higher eukaryote–specific RNAP III, the multi-domain expansion of RPC5 and hypothesized associated higher eukaryote promoter–type specificity, and its link to termination-reinitiation recycling.

Mechanistic control of ncRNA genes that appear to have evolved activities that extend hRNAP III beyond the housekeeping activities of its yeast counterpart

This fits a view of hRNAP III in self vs. non-self surveillance functions. Notable are RNAP III Vault (Vt) ncRNAs with involvement in two activity types, in innate immune surveillance, and differentiation vs. maintenance of undifferentiated states. Both RNAP III Vt and snaR (small nuclear factor 90–associated RNA) ncRNAs are processed to miRNAs that exert downstream effects on mRNA profiles and/or differentiation/cancer. Our proposal includes the study of a dual-activity tRNA gene(s) for which La is a key determinant of whether the nascent 4T–terminated transcripts are directed to tRNA maturation or to an alternate pathway of innate immune activation.

Activities of RNA polymerase III (RNAP III) and associated factors

The RNAP III multisubunit enzyme complex consists of 17 subunits, several with homology to subunits of RNAPs I and II. The transcription factor TFIIIC, composed of six subunits, binds to A- and B-box promoters (promoter elements of tRNA genes) and recruits TFIIIB to direct RNAP III to the correct start site. TFIIIB–RNAP III complexes appear highly stable and demonstrate great productivity in supporting the many cycles of initiation, termination, and re-initiation necessary to produce the more than tenfold molar excess of tRNAs relative to ribosomes that is required to drive translation during growth and development. In contrast to all other multisubunit RNA polymerases, termination and re-initiation by RNAP III are functionally, if not physically, linked. Our laboratory developed methods for in vivo and biochemical studies to examine the unique mechanisms used by RNAP III. Hereditary mutations in RNAP III cause hypomyelinating leukodysplasia (HLD), as well as defects in innate immunity. In addition to its being essential for cell proliferation, RNAP III is also linked to aging.

Transcription termination delineates 3′ ends of gene transcripts, prevents otherwise runaway RNAP from intruding into downstream genes and regulatory elements, and enables release of the RNAP for recycling. While other RNAPs require complex cis signals and/or accessory factors to accomplish these activities, eukaryotic RNAP III does so autonomously with high efficiency and precision at a simple oligo(dT) stretch of 5–6 bp. A basis for this high-density cis information is that both the template and non-template strands of the RNAP III terminator carry distinct signals for different stages of termination. High-density cis information is a feature of the RNAP III system that is also reflected in the dual functionalities of the tRNA promoters as both DNA and RNA elements. Furthermore, the TFIIF–like RNAP III subunit C37 is required for this function of the non-template strand signal. Our results reveal the RNAP III terminator to be an information-rich control element. While the template strand promotes destabilization via a weak oligo(rU:dA) hybrid, the non-template strand provides distinct sequence-specific destabilizing information through interactions with the C37 subunit.

Control of the differential abundance or activity of tRNAs can be an important determinant of gene regulation. RNAP III synthesizes all tRNAs in eukaryotes, and its derepression is associated with cancer. MAF1 is a conserved general repressor of RNAP III under the control of TOR (target of rapamycin), which acts to integrate transcriptional output and protein-synthetic demand toward metabolic economy. We used tRNA–Hydro-Seq (hydrolysis-based tRNA sequencing) to document that little change occurred in the relative levels of different tRNAs in maf1 mutated cells. By contrast, the efficiency of N2,N2-dimethyl G26 (m2,2G26) modification on certain tRNAs was reduced in response to maf1 deletion and associated with anti-suppression, which we validated by other methods. Overexpression of Trm1 (tRNA dimethyl transferase), which produces m2,2G26, reversed maf1 anti-suppression. The model that emerges is that competition by elevated tRNA levels in maf1-delta cells leads to m2,2G26 hypo-modification resulting from limiting Trm1, thus reducing the activity of suppressor tRNASerUCA (UCA is the anticodon for serine) and accounting for anti-suppression. Consistent with this, RNAP III mutations associated with HLD reduce tRNA transcription, increase m2,2G26 efficiency, and reverse anti-suppression. Extending this more broadly, a reduction in tRNA synthesis by treatment with rapamycin leads to increased m2,2G26 modification, a response that is conserved among highly divergent yeasts and human cells [Arimbasseri AG et al., PLoS Genetics 2015;11:e1005671].

The ability of RNAP III to efficiently recycle from termination to re-initiation is critical for abundant tRNA production during cellular proliferation, development, and cancer. We used two tRNA–mediated suppression systems to screen for Rpc1 (subunit of RNAP III) mutants with gain- and loss-of-termination phenotypes in Schizosaccharomyces pombe. We mapped 122 point mutation mutants to a recently solved 3.9 Å structure of the yeast RNAP III elongation complex (EC); they cluster in the active-center bridge helix and trigger loop, as well as in the pore and funnel formation, the latter indicating involvement in termination of the RNA–cleavage domain of the C11 subunit of RNAP III. Biochemical, kinetic, and genetic data indicate that mutants with the RT (readthrough) phenotype synthesize more RNA than wild-type cells and, surprisingly, more than can be accounted for by the mutants' increased elongation rate. Importantly, similar mutations in spontaneous cancer suggest this as an unforeseen mechanism of RNAP III activation in disease.

Role of La-related protein-4 (LARP4) in poly(A)–mediated mRNA stabilization

Ubiquitous in eukaryotes, La proteins are involved in two broad functions: first, metabolism of a wide variety of precursor tRNAs and other small nuclear RNAs by association with these RNAs’ common UUU-3′OH–transcription termination elements; and second, translation of specific subsets of mRNAs, such as those containing 5′ IRES (internal ribosome entry site) motifs. LARP4 emerged later in evolution, and we found it to be an mRNA–associated cytoplasmic factor associated with poly(A)–binding protein C1 (PABPC1, PABP). LARP4 uses two regions to bind to PABPC1. We showed that the N-terminal domain (NTD, amino acids 1–286) of LARP4, consisting of an N-terminal region (NTR, amino acids 1–111) followed by two tandem RNA–binding motifs known as an ‘La module’ (111–285), exhibits preferential binding to poly(A). The NTR contains a unique PAM2w motif that binds to the MLLE (a peptide-binding domain) of PABP. The group of our collaborator Maria Conte showed that the N-terminal region (NTR) itself is responsible for most of the poly(A) binding and that, moreover, this involves conserved residues unique to the PAM2w of LARP4. The La module is flanked by a different motif on each side, each independently interacting with PABP. LARP4 is controlled at the level of mRNA stability: one level of control is by an A+U-rich element (ARE) in its 3′ UTR via interactions with the protein tristetraproline (TTP), the latter of which is regulated in mammals by tumor necrosis factor alpha (TNFα); a second level of control was found for the LARP4 mRNA–coding sequence in an unusual group of synonymous codons with poor match to cellular tRNA levels [Reference 1]. The LARP4 protein controls the metabolism/homeostasis and translation of heterologous mRNAs by affecting their poly(A) tail length. Working with researchers in the NICHD Molecular Genomics Core facility, we developed a single-molecule, high-throughput nucleotide-resolution poly(A)–tail sequencing method referred to as SM-PAT-Seq, which yielded insights into LARP4 function and mechanism. LARP4 is a global factor involved in mRNA poly(A) length homeostasis and appears to effect mRNA stabilization by opposing the action of deadenylases when poly(A) tails are short.

Fission yeast as a model system for the study of tRNA metabolism and function in translation

More than 20 years ago, we began developing, refining, and advancing a tRNA–mediated suppression (TMS) system in the fission yeast Schizosaccharomyces pombe, which provides a ‘red-white’ phenotypic real-time assay that can be used to investigate various aspects of tRNA biogenesis, maturation, and metabolism of tRNAs in vivo. In S. pombe, the human La protein can replace the tRNA–processing/maturation function of Sla1p, the S. pombe equivalent of the La protein. Moreover, in S. pombe, human La is faithfully phosphorylated on Ser-366 by protein kinase CKII, the same enzyme that phosphorylates Ser-366 in human cells, and this phosphorylation event promotes pre-tRNA processing. We use the system to study transcription by RNAP III, post-transcriptional processing, and tRNA modifications by conserved enzymes that produce tRNA isopentenyl-adenosine-37 and dimethyl-guanosine-26 (N2,N2-dimethyl G26, m2,2G26).

The fission yeast Schizosaccharomyces pombe as a model organism

Figure 1

Click image to view.

Red-white colony differentiation by tRNA–mediated suppression

tRNAs, codon use, and mRNA metabolism in growth and development

A major interest of ours is to decipher what we refer to as ‘secondary information’ in the genetic code, information that is derived from mRNAs' biased use of synonymous codons. This can produce a layer of information beyond the amino acid sequence of a protein; i.e., in addition to providing the template for the sequence of a protein, the use of certain synonymous codons can also produce additional biochemical effects, which we refer to as ‘secondary information.’ The effects can be related to ribosome pausing, which can affect protein folding, or to alterations in the stability of the mRNA. Other types of secondary information can also be encoded in synonymous codons; for example, sets of mRNAs that share similar patterns of synonymous codon bias are similarly sensitive to tRNAs with the same anticodon modification and exhibit similar patterns of efficiency of translation elongation. The components of the secondary information system are the tRNA pool, the tRNA–modification enzymes, and the codon-bias distribution among the mRNAs. We recently found that synonymous codon use by the human LARP4 mRNA is a key determinant in the control of the expression levels of its mRNA and protein, and that increases in otherwise limiting tRNAs that are cognate to these codons increase LARP4 production, which in turn activates LARP4 to promote a net increase in the poly(A) tail length of heterologous mRNAs, including those that encode ribosomal protein subunits [Reference 1], which may be important because ribosome production is regulated during growth and development, and because the potential circuit involving LARP4 control by tRNA could be an important point of control.


  1. Mishra S, Hasan SH, Sakhawala RM, Chaudhry S, Maraia RJ. Mechanism of RNA polymerase III termination-associated reinitiation-recycling conferred by the essential function of the N terminal-and-linker domain of the C11 subunit. Nat Commun 2021;12:5900.
  2. Mattijssen S, Kozlov G, Gaidamakov S, Ranjan A, Fonseca BD, Gehring K, Maraia RJ. The isolated La-module of LARP1 mediates 3' poly(A) protection and mRNA stabilization, dependent on its intrinsic PAM2 binding to PABPC1. RNA Biology 2021;18:275–289.
  3. Mattijssen S, Iben JR, Li T, Coon SL, Maraia RJ. Single molecule poly(A) tail-seq shows LARP4 opposes deadenylation throughout mRNA lifespan with most impact on short tails. eLife 2020;9:e59186.
  4. Khalique A, Mattijssen S, Haddad AF, Chaudhry S, Maraia RJ. Targeting mitochondrial and cytosolic substrates of TRIT1 isopentenyltransferase: specificity determinants and tRNA-i6A37 profiles. PLoS Genet 2020;16:e1008330.
  5. Coon SL, Li T, Iben JR, Mattijssen S, Maraia RJ. Single-molecule polyadenylated tail sequencing (SM-PAT-Seq) to measure polyA tail lengths transcriptome-wide. Methods Enzymol 2021;655:119–137.


  • Leonid V. Chernomordik, PhD, Section on Membrane Biology, NICHD, Bethesda, MD
  • Maria R. Conte, PhD, King's College, University of London, London, United Kingdom
  • Steven Coon, PhD, Molecular Genomics Core, NICHD, Bethesda, MD
  • Kalle Gehring, PhD, McGill University, Montreal, Canada
  • James R. Iben, PhD, Molecular Genomics Core, NICHD, Bethesda, MD


For more information, email or visit

Top of Page