Gene expression is the most fundamental process in creating human traits from genetic information. Transcription involves copying DNA into RNA molecules by RNA polymerases (RNAPs), and translation decodes mRNA molecules into polypeptides in ribosomes. In transcription process, RNAPs can synthesize a variety of RNA molecules in the nucleus, including mRNA for protein synthesis as well as non-coding RNA (ncRNA) that function as RNA molecules itself. Eukaryotic RNAPs undertake specialized roles in transcribing nonoverlapping gene groups. RNAP I synthesizes 5.8S, 18S, and 28S rRNAs, RNAP II synthesizes mRNAs, long noncoding RNAs (lncRNAs), and small nuclear RNAs (snRNAs), and RNAP III synthesizes 5S rRNAs and tRNAs. Given their critical roles, the defects in the biogenesis of RNAPs can disrupt the essential cellular processes, resulting in impaired growth, development, and cell death [1, 2]. They ultimately lead to severe diseases such as ribosomopathies [3], developmental disorders [4], and cancers [5].

Then, mRNAs are translated into polypeptide chains in ribosomes, a ribonucleoprotein complex consisted of ribosomal proteins (RPs) and the 5S, 5.8S, 18S, and 28S rRNAs [6]. Given the high energy consumption during the translation from nucleic acid sequences to amino acid sequences, it is subject to stringent controls. Translational control allows for more efficient adaptive responses to fluctuations in cellular environments, compared to upstream gene expression processes. For instance, the gene expression of RPs can be dynamically regulated by instantaneous translation in response to urgent cellular conditions [7]. In contrast, RP mRNAs undergo excessive production through transcriptional regulation, are stored as inactive messenger ribonucleoprotein particles, and are poised for immediate translation [8]. Dysregulation in translation can lead to abnormal proliferation, cell survival, and immune response, consequently contributing to the development of cancers.

The role of RNAPs and ribosomes in human longevity has been underscored by a Mendelian randomization study [9]. Over time, significant research efforts have been dedicated to examining the subunits of RNAPs and ribosomes. The perspective on ribosomes has evolved from being viewed as passive and indiscriminate structures to dynamic macromolecular complexes with specialized cellular functions. As the functions of ribosome subunits have been uncovered, we illuminate the intricate roles of individual genes and their expression regulations [10]. In this article, we briefly review the structure and function of constituent proteins of human RNAP I, RNAP II, RNAP III, and ribosomes as key players for gene expression processes. We explore the characterization of these proteins and their corresponding genes, focusing particularly on their expression quantitative trait loci (eQTL) identified for association with complex human traits. Fundamental terminology for this review is summarized in Table 1.

This section deals with the structure and function of RNAPs in Saccharomyces cerevisiae, better known as baker’s yeast, which is considered to be representative of eukaryotes and has contributed the most to our knowledge. A caution is, however, warranted with the names of human genes in later sections of this review because the nomenclature of many polymerase genes and proteins within and between yeast and humans is unusual and often confusing. In eukaryotes, the three RNAPs, RNAP I, RNAP II, and RNAP III, are complex enzymes composed of multiple subunits that contain core and common subunits. It is conceivable that common regulators controlling the levels of these shared subunits effectively coordinate the functional levels of the three RNAPs. In particular, the five out of the shared components, corresponding to prokaryotic core RNAP composed of catalytic ($\beta$ and $\beta$’), assembly (two $\alpha$’s), and auxiliary ($\omega$) subunits (Table 2, Ref. [11, 12, 13]), are conserved elements that have evolved from their ancestral counterparts over an extended period [14]. They additionally share four subunits of Rpb5, Rpb8, Rpb10, and Rpb12 and include one subunit with homologous N-terminal ribbon domains of Rpb9, A12, and C11 for RNAP II, I, and III, respectively [11]. Thus, regardless of the polymerase type, they work in a similar manner under common fundamental principles over the transcription process that encompasses initiation, elongation, and termination.

These polymerases, however, also have distinct transcription modes evident from their different structures of other subunits. RNAP II, which targets all protein-coding genes, interacts with a broader array of regulatory factors, compared to RNAP I and RNAP III. Structurally, RNAP II has fewer permanent subunits but incorporates dissociable subunits (Rpb4 and Rpb7) and independent initiation and elongation factors (TFIIS, TFIIF, and TFIIE) as presented in Table 2. The dissociable subunits play crucial roles as a heterodimeric stalk (Rpb4/7) in elongation process. Furthermore, RNAP II has the capacity to execute a suitable termination mechanism by capturing dissociated RPB3, which plays a predominant role in regulating the 3${}^{\prime }$ end processing of RP genes [15]. In contrast, RNAP I and RNAP III have their own intrinsic components with equivalent functions. Even the core structures of RNAP I and RNAP III exhibit greater similarity, including the AC40 and AC19 heterodimer being homologous to Rpb3 and Rpb11 in RNAP II [16]. On the other hand, RNAP III-specific trimer (C82/34/31) was reported equivalent to TFIIF and TFIIE of RNAP II [17]. A RNAP III dimer, C37/53, functions equivalently to dimers (A49/34 and TFIIF$\alpha$/$\beta$) of RNAP I and RNAP II in initiation process. Interestingly, C37/53 can also work in the termination process although RNAP III has a quite unique halting mechanism compared to RNAP I and RNAP II [18].

The intrinsic subunits of RNAP I and RNAP III help conduct speedy elongation. An example is the intrinsic dimer C37/53 of RNAP III, specifically essential along with C11 for the highly efficient termination and coupled reinitiation processes in facilitating transcription of very short genes [19]. The intrinsic subunit A12 of RNAP I plays a pivotal role in RNA cleavage, facilitating proofreading, and enabling a swift resumption of elongation following a pause [16]. The intricacies involved in the resumption of elongation after pausing in RNAP II imply a necessity for specific regulatory mechanisms. This level of regulation is not essential for the comparatively simpler and faster elongation processes observed in RNAP I and RNAP III [20].

Eukaryotic ribosome, consisting of the small (40S) and large (60S) subunits, is a complex macromolecular machine that orchestrates the translation and protein synthesis as the heart of the translation machinery by collaborating with other translational apparatus molecules such as transfer RNAs (tRNAs) and translation factors. The small subunit composed of a 18S ribosomal RNA (rRNA) and 33 RPs is bound to mRNA for decoding, and the large subunit composed of a 5S rRNA, a 28S rRNA, a 5.8S rRNA, and 49 RPs is bound to aminoacyl tRNA for catalysis. The translation activity requires ribosome biogenesis through elaborate coordination of RNAP I, II, III, and over 200 ribosome assembly factors. This highly complex process inevitably requires strict regulation and smooth communication with other cellular pathways [21]. Changes in ribosome biogenesis may impact the translation process directly, influencing global gene expression during cell growth [22], differentiation [23], and disease progression (e.g., cancer metastasis potential [24]).

RPs play an inevitable role in a wide range of ribosome biogenesis, assembly, and translation as shown in Table 3 (Ref. [10, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49]). For instance, RPL33 is responsible for ribosomal-subunit joining and ribosome biogenesis. Its missense mutation (rpl33a-G76R) alters the 60S subunit, impeding ribosomal-subunit joining. This represses translation of a master transcription factor GCN4 in yeast, corresponding to ATF4 in human, and impairs the efficient processing of 35S and 27S pre-rRNAs, decreasing all four mature rRNAs responsible for the biogenesis of both ribosomal subunits [25]. RP59 is necessary for the assembly of 40S subunit, and RPL1 and RPL16 combine with 5S rRNA to produce stabilized ribonucleoprotein, which is necessary for the assembly of 60S subunit [26, 27]. RPL12 is responsible for mediating the accurate assembly of the ribosomal stalk [28]. Some RPs (RPS0, RPS14, and RPS21) participate in the cytoplasmic rRNA processing steps for the maturation of 18S rRNA [29, 30]. RPS14 is further involved in the maturation of 43S pre-ribosomes [29]. In addition, RPL25 is essential for pre-rRNA processing [31], RPL9 is crucial for the maturation of small subunit [32], and RPS15 plays a vital role in the nuclear exit of the 40S subunit precursors [50].

For translation function, RPS3 conducts a vital regulation at the preliminary translational initiation: the conserved residues R116/117 within RPS3 stabilize interactions between the ribosome and mRNA at the mRNA entry pore, R146 and K148 contribute to the accuracy of start codon selection; K62 functions ribosome-based mRNA quality control, and residues at 60 to 63 ensure the proper structure of the 48S preinitiation complex [33]. RPS26 promotes a selective translation of specific mRNAs under specific cellular environment by recognizing Kozak sequence and interacting with initiation factors eIF3a and eIF3d [10, 34, 35]. In particular, RPS26 is a detachable component for fine regulation of mRNAs largely with Kozak sequence according to specific situation [51]. RPS20 is responsible for the mRNA binding and subunit docking, and its deletion reduced mRNA binding and decreased 70S complexes, leading to initiation defects to the small subunit [36]. RPS15 located in the decoding center plays a pivotal role in accommodating aminoacyl-tRNA at the A site. A high-resolution cryoelectron microscopy captured the C-terminal tail of RPS15, interacting with the tRNAs located in A- and P-sites in the decoding center [37]. RPL36a plays a critical role in the elongation of peptide chains by interacting with the 3${}^{\prime }$-end of deacylated tRNA at the P site for peptide bond formation [38]. The hydroxylation of RPL27a was stressed to maintain stability of E site that binds to free tRNA to exit. Mutated RPL27a (His39Ala) at the hydroxylation site led to specific changes to the repertoire in mRNA translation [39].

RPL23 plays a regulatory role in protein biosynthesis and chaperone-assisted protein folding as a chaperone docking site [40], and RPL23a along with RPL35 also showed an important role during signal peptide recognition and insertion of the peptide into the translocation channel by repositioning a signal recognition particle (SRP54) [41]. In addition, RPS4, RPS13, and RPS28 contribute to translational accuracy [42], and RPL3, RPL5, and RPL41 contribute to peptidyltransferase activity [43, 44, 45]. In yeast, RPL5 further plays an important role in anchoring peptidyl-tRNA to the P-site [45]. RPS4 and RPS5 are essential for preserving the accuracy of protein translation [52]. In contrast, RPS12 enhances the translation rate at the expense of a higher error rate in protein synthesis [53].

Nucleotide sequence variants that regulate expression of the genes responsible for RNAPs and RPs can have profound implications in conducting gene expression. Insufficient or altered components can delay expression, cause misinterpretations during expression, or, in some cases, a complete halt in the process, resulting in deficient or undesirable proteins. The nucleotide variants are, therefore, likely to be associated with a variety of complex traits and diseases, highlighting the pivotal role of a properly functioning RNAPs/ribosomes in maintaining cellular health.

With the genes responsible for RNAPs/ribosomes, we uncovered 12 EFO parent terms of phenotypes associated with their regulatory variants. They differed in proportion from those with all the available genes. Among them, 8 EFO parent terms for RNAPs/ribosomes showed a larger portion than those for nominal genes. Notably, terms such as ‘hematological measurement’, ‘cancer’, and ‘immune system disorder’ were critically associated with protein synthesis that is closely coupled to RNAPs and ribosomes. This result concurred with previous studies in which global protein synthesis plays a role in both quiescence and differentiation of hematopoietic stem cell [58], colorectal cancer [59], hepatocellular carcinoma [60], systemic lupus erythematosus [61], and ankylosing spondylitis [62].

The differences in EFO proportions between RNAPs/ribosomes and nominal genes were considerably attributed to ribosomes. GWAS–eQTL pairs for ribosomes accounted for 83.4% of the total pairs for RNAPs/ribosomes, highlighting the significant cellular effort devoted to ribosome production, which encompasses more than half of all transcription and translation processes [63]. We hypothesize that numerous eQTL might yield ribosomes of insufficient quality and/or quantity, resulting in unfavorable translation, which could impact disease susceptibility. Heterogeneity in ribosomes results in distinct interactions with specific mRNAs, determining translation priorities in particular cell types or in response to specific environmental cues [64], exemplified by the ribosome-mediated response with RPS26 under stress [10]. The accumulation of ribosomes with altered protein stoichiometry [65] or mutations in maturing ribosomes with quality control bypass [66] may predispose human cells to cancer. In addition, a limited number of ribosomes can induce ribosome competition among cellular mRNAs, thereby altering the translation efficiency of subsets of mRNAs [67]. The undesirable translation by ribosome variability could largely influence the three parent terms ‘hematological measurement’, ‘cancer’, and ‘immune system disorder’. Hematological measurements, including corpuscular hemoglobin, hemoglobin level, and red cell distribution width, have demonstrated that highest rates of protein synthesis occur in erythroid lineage commitment [59]. Reduced ribosome levels in hematopoietic cells can affect the translation of a specific subset of mRNAs, particularly GATA1, a master regulator of hematopoiesis, impairing erythroid lineage commitment [68]. Conversely, increased RPs may regulate the p53 pathway via Mdm2 and Mdm4, inducing apoptosis and suppressing cell proliferation [69]. Enrichment analysis revealed ribosomal protein-synthetic pathways associated with GWAS signals for autoimmune diseases such as multiple sclerosis, rheumatoid arthritis, and systemic lupus erythematosus [70].

Unlike ribosomes, we found that EFO proportion for each of RNAP I, II, and III differed from those for RNAPs/ribosomes, showing the largest difference in the portion of ‘cancer’ for RNAP III. We hypothesize that cancer susceptibility could be driven by RNAPs/ribosomes, especially RNAP III, which directly plays a pivotal role in translation from nucleic acid sequences to amino acid sequences. Dysregulation of RNAP III has been consistently observed in various cancers, including prostate cancer, lung cancer, hepatocellular carcinoma and breast cancer [71, 72, 73]. RNAP III can alter the abundance and availability of specific tRNAs in cancer cells [74], enhancing translation efficiency of EXOSC2 and GRIPAP1 genes with codons corresponding to tRNAs [75].

Individual genes with many GWAS–eQTL pairs possess some characteristics in structure and/or function of RNAPs/ribosomes. RPS26, with the largest number of pairs among RP eGenes, is located in the mRNA exit channel in ribosomes and contributes to mRNA-specific translation by recognizing the Kozak sequence. Ribosome lacking RPS26 could decrease the translation of essential mRNA with the Kozak sequence while increasing the translation of mRNA with a long 5${}^{\prime }$ untranslated region or a weak Kozak sequence [10]. Such a control was observed in wild type yeast, where RPS26 was released and reincorporated from ribosomes under stress conditions [35]. This study suggests that RPS26 plays a crucial role in controlling translation for cell survival across all tissues, emphasizing its importance as a key player in translating mRNA with the Kozak sequence. Dysregulation of RPS26 might extensively influence susceptibility to many human diseases.

For RNAP III, we found the largest GWAS–eQTL pairs in the brain among tissues, whereas the other subtypes of RNAPs/ribosomes showed a high correlation (r${}^{\mathrm{2}}$ = 0.72~0.96) between the number of eQTLs retrieved from GTEx and the number of GWAS–eQTL pairs identified in this study, showing the largest numbers in muscle. This remarkable number (110 out of 294 in total) of brain-specific pairs was largely attributed to those regulating POLR3H. Their GWAS phenotypes were largely assigned to 21 ‘neurological disorders’ and 34 ‘other measurements’ terms, which might be influenced by dysregulating POLR3H. The gene POLR3H encodes RPC8 that can act together with RPC9 (encoded by CRCP) in the differential transcription initiation of pre-tRNA${}^{\text{Tyr}}$ transcripts, as shown in yeast [76, 77]. Interestingly, POLR3H and CRCP are all eGenes identified in the brain. Thus, a decreased level of RNAP III with RPC8 might cause a deviation from a normal tRNA composition, subsequently increasing a susceptibility to brain diseases. Brain tissue is particularly susceptible to imbalances in tRNA composition [78, 79] because of the remarkably conserved codon usage of brain-specific genes [80]. Furthermore, this vulnerability may be exacerbated by the polarized morphology of neurons, which have distinct functional compartments such as dendrites and axons. This study also suggests that RNAP III in the brain might be a factor for susceptibility to ‘neurological disorders’ and ‘other measurements’; ‘neurological disorders’ includes Alzheimer’s disease, insomnia, major depressive disorder, neuroticism, schizophrenia, and depression, and ‘other measurements’ includes anxiety, feeling nervous/tense/worry, irritable mood, neurociticism, and tea/caffeine/cannabis consumption.

This study showed the most abundant pairs (310 pairs) associated with eGenes belonging to RNAP I, and this abundance was largely attributed to the eGene (POLR1H) encoding PRA12 (246 pairs). PRA12 contributes to RNAP I passage through nucleosomes, conveyed with RNA cleavage, enzyme backtracking and proofreading, and transcription termination [81, 82, 83]. Depletion of its proofreading function for rRNA transcripts may result in a reduction in the overall fidelity of transcription. A recent study has confirmed that deletion of PRA12 reduces transcription fidelity, accompanied by many G $\to$A transitions in 18S rRNA, affecting the secondary structure and thereby interfering with interactions with other rRNAs [84]. Additionally, mutations in the 28S rRNA, which functions as the peptidyl transferase center, severely affect peptidyl transferase activity [85] or cause read-through errors [86]. Furthermore, transcription fidelity could be decreased by the differential expression of RPB9 in RNAP II, the functional analog of PRA12 in RNAP I [87]. Consequently, PRA12 is an essential built-in transcription factor for the high transcriptional efficiency of RNAP I, which exclusively synthesize rRNA occupying ~90% of total RNA by mass, consuming high-energy expense in cellular homeostasis [88].

A small number of RNAP II-related pairs might have been caused by a small number of eGenes (Supplementary Table 1). This is largely because of its unique structural organization with minimal subunits to efficiently transcribe a wide range of genes with distinct regulatory requirements. Unlike other RNAPs, RNAP II achieves flexibility through a combination of fewer permanent subunits and the presence of dissociable subunits (RPB4/7). Additionally, it utilizes independent initiation (TFIIF and TFIIE) and elongation (TFIIS) factors, whereas other RNAPs contain built-in equivalent factors. Notably, the subunits of RNAP II are most likely observed in other polymerases. For instance, POLR2L, the most frequent eGene, encodes the RPABC5 protein, an essential subunit of all RNAPs for assembly and coordination [89]. The second most eGene was POLR2D, encoding RPB4, which constitutes RPB4/7, integral throughout the gene expression process. This forms a stalk module of RNAP II, playing a crucial role in transcription initiation and elongation, co-transcriptional mRNA splicing, and 3${}^{\prime }$ end processing [90]. The RPB4/7 may conduct co-transcriptional mRNA imprinting as the dissociable heterodimer of RNAP II, affecting mRNA export, decay, and translation [91, 92, 93]. Subsequently, this contributes to bidirectional controls in mRNA transcription and decay.

Highlighted above are efforts to understand the functions and characteristics of individual subunits within RNAPs and ribosomes, particularly in relation to complex traits. Alongside addressing priority subunits, there is a growing interest in subunit-specific functions. For example, Rpb3 can directly bind to tissue-specific transcription factors, such as myogenin in skeletal muscles [94] and ATF4 in fibroblasts [95]. Notably, the N-terminus of Rpb3 exhibits selective inhibition in proliferating hepatocellular carcinoma cells that overexpress Rpb3 [96]. Additionally, Rpb9 has been recognized as a driver gene within a regulatory network associated with atherosclerosis, specifically targeting arterial wall tissue, as revealed by GWAS [97]. Understanding the individual genes encoding subunits of RNAPs and ribosomes, along with their specific functions, would greatly aid in comprehending the etiology, pathogenesis, and treatment of diseases.

This review suggests that differences in gene expression related to alleles of genes encoding RNAPs and ribosomes could potentially influence susceptibility to complex human diseases. However, it emphasizes caution regarding the underlying mechanisms of disease pathogenesis. While some data are sourced from GWAS involving patients, an equal proportion comes from analyzing gene expression patterns in healthy individuals with genetic variants. Experimental investigations are imperative to validate and elucidate the role of these genes and their variants in disease susceptibility, particularly in the context of patients or relevant animal models.

This study provides insightful characteristics of the regulatory factors for the RNAP I, II, III, and ribosomes that can influence human complex traits. These regulatory factors were predominantly attributed to RP, suggesting critical ribosomal perturbations in the traits. In particular, RPS26 was the notable eGene with the largest and most widespread impact across all tissues. Among RNAP III genes, POLR3H, with the largest regulatory factors, was highlighted to have brain-dominant regulatory factors associated with neurological disorders. Many regulatory factors implicate RNAP I and RNAP II genes with remarkable functions. While POLR1H encodes a built-in transcription factor, POLR2D encodes dissociable stalk module.

These results suggest the critical impact of genes and proteins responsible for RNAPs and RPs on complex human traits. The emphasis is on RPS26, POLR1H, POLR2D, and POLR3H, highlighting their significance. This information contributes to a better understanding of RNAPs/ribosomes and their regulatory factors, serving as potential prognostic and therapeutic targets in precision medicine.

Conceptualization, CL; formal analysis, JR; writing—original draft preparation, JR; writing—review and editing, CL; visualization, JR and CL; supervision, CL. Both authors have read and agreed to the published version of the manuscript. Both authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work.

Not applicable.

The authors would like to thank two anonymous reviewers for their insightful comments on the first version of the manuscript.

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2021R1A6A1A10044154).

The authors declare no conflict of interest. Given his role as Editorial Board member, Chaeyoung Lee had no involvement in the peer-review of this article and has no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to Nobuo Shimamoto and Yudong Cai.

Supplementary material associated with this article can be found, in the online version, at https://doi.org/10.31083/j.fbl2905185.