The uORF–NMD axis and uPeptides: a new frontier in cancer therapy

1 Introduction

The development of the ribosome profiling (Ribo-seq) in the last years, a technique that provides a “picture” of actively translating ribosome positions at the transcriptomic level by high-throughput sequencing of short ribosome-protected mRNA fragments, has revealed thousands of previously unannotated initiation events in canonical initiation AUG codons, as well as in near-cognate translation start sites (AAG, ACG, AGG, AUA, AUC, AUU CUG, GUG, and UUG), located in the human genome, in regions previously thought to be non-coding (Ingolia et al., 2009; Ingolia et al., 2011; Ingolia, 2010; McGlincy and Ingolia, 2017). In fact, it has been possible to detect active protein translation outside of canonical coding sequences (CDSs), in 5′ transcript leader sequences (5′TLS) and 3′ untranslated regions (3′UTRs) of mRNAs, in intronic, intergenic, and off-frame regions, as well as in non-coding RNAs (ncRNAs), such as long ncRNAs and circular RNAs (Wu et al., 2020; Brunet et al., 2021; Zhang Q. et al., 2021). Expression of the corresponding peptides is not yet completely known and contributes to the so-called “dark” proteome (Orr et al., 2020).

If the ribosome recognises an initiation codon in a 5′TLS, it may translate an upstream open reading frame (uORF). uORFs are defined by a canonical initiation codon, or a near-cognate start site, in-frame with a translation termination codon that may be located upstream or overlapped with the main CDS (mCDS) (Barbosa et al., 2013). Bioinformatics studies have mapped genetic variants within the 5′TLSs of transcripts and identified alterations that affect the regulatory function of uORFs (Calvo et al., 2009; Ye et al., 2015; Wethmar et al., 2016). In addition, several disease-associated mutations in TLSs disrupt or create uORFs (Barbosa et al., 2013; Silva et al., 2019). Together, these data confirm the importance of uORFs in gene expression regulation.

Bioinformatics studies of Ribo-seq data have identified at least one uORF fully located within the 5′TLS, or partially overlapped with the mCDS, in approximately 13, 45, 63 and 49%–58% of yeast, mouse, zebrafish and human transcripts, respectively (Matsui et al., 2007; Calvo et al., 2009; Lawless et al., 2009; Ye et al., 2015; Johnstone et al., 2016). Many uORFs are evolutionarily conserved across species, suggesting a translational regulatory role (Calvo et al., 2009; Johnstone et al., 2016). Indeed, several reports have shown that the presence of uORFs correlates with decreased mRNA levels and/or decreased translation of the downstream CDS, resulting in a reduction of 30%–80% in mORF expression (Calvo et al., 2009; Ye et al., 2015; Johnstone et al., 2016). Recent advances in tools for RNA editing (Yan and Tang, 2024) and antisense oligonucleotide (ASO) targeting of the uORF initiation codon (Liang et al., 2017) have significantly contributed to the functional characterisation of uORFs. However, these methods are not consistently successful. An example is the targeting, with different ASOs, of a pathogenic variant that creates a uORF in the ENG 5′UTR, disrupting translation from the main ORF and contributing to hereditary haemorrhagic telangiectasia (Doisy et al., 2025). In this case, the method failed to restore the main protein (Endoglin) expression. These findings highlight the limitations of steric-blocking ASOs targeting 5′UTR variants and underscore the need for deeper mechanistic understanding of uORF-mediated translational regulation.

The presence of a uORF in a transcript may trigger it to nonsense-mediated decay (NMD) as the uORF stop codon may be recognized by the ribosome as being in a premature context (Amrani et al., 2004). Besides, translation of uORFs may produce uORF-encoded peptides (uPeptides). uPeptides have been identified as significant regulators of different pathways in normal and malignantly transformed human cells (Jürgens and Wethmar, 2022). uPeptides also contribute to the enrichment of the immunopeptidome (Jürgens and Wethmar, 2022; Nelde et al., 2022).

Here, we aim to review the importance of translation of uORFs in the regulation of gene expression and the functional roles of uPeptides. Furthermore, we wish to discuss the role of NMD in regulating uPeptide expression levels and their functions in normal and cancer cells, and how they can serve as targets to design novel cancer therapeutic approaches.

2 Mechanisms of uORF-mediated gene expression regulation

Generally, uORF-mediated translational repression can be achieved by ribosome dissociation and recycling after uORF translation (Figure 1A), by ribosome stalling during the elongation or termination processes (Figure 1B), and, in case the uORF stop codon is recognised as a premature termination codon (PTC), by inducing degradation of the mRNA through NMD (Figure 1C) (Mendell et al., 2004; Gardner, 2008; Barbosa et al., 2013). Additionally, the uORF-encoded peptide itself may act as a cis- or trans-regulatory factor that inhibits main ORF (mORF) translation (Figure 1D) (Raney et al., 2002; Pendleton et al., 2005; Lee et al., 2009; Young et al., 2015; Young et al., 2016b). In turn, uORFs may induce translation of the downstream mCDS mainly by two mechanisms: translation reinitiation at the mCDS after uORF translation, or ribosomal bypass of the uORF (Vattem and Wek, 2004; Young et al., 2015; Mohammad et al., 2017). For instance, during integrated stress response (ISR), translation of glutamyl-prolyl-tRNA synthetase (EPRS) is reduced by two inhibitory uORFs, encoded by noncanonical CUG and UUG initiation codons in the EPRS mRNA 5′-TLS (Young et al., 2015). Thus, in response to several eIF2α phosphorylation activators, ribosomes bypass these uORFs, leading to greater translation of the EPRS mCDS and helping cells adapt to different stress conditions (Young et al., 2015). On the other hand, during stress conditions that also result in the phosphorylation of eIF2α, there is a reduction in the levels of eIF2-GTP, which increases the time the scanning ribosomes need to become competent to reinitiate translation (Vattem and Wek, 2004). In the case of activating transcription factor 4 (ATF4), which has two uORFs that inhibit its mCDS expression, this delayed reinitiation allows the ribosomes to scan through the inhibitory uORF2 and reinitiate translation at the ATF4 coding region, thereby increasing ATF4 expression and, in turn, contributing to the expression of genes involved in controlling cellular stress damage (Young et al., 2015).

The inhibitory potential of a uORF can be easily understood in light of the scanning model of translation initiation. If the 43S pre-initiation complex (PIC), while scanning the 5′TLS of a transcript, recognizes the start codon of a uORF, that uORF will be translated, resulting in the consumption of a functional PIC that may not reach the downstream mORF. Yet, despite this simple model of translational repression, several transcript-specific features add multiple layers of complexity to uORF-mediated translational control, which may also result in the opposite effect, i.e., induction of mCDS translation, as is the case of uORFs containing cis-acting 5′ reinitiation-promoting elements (RPEs), which promote association of eIF3 with the uORFs present within the GCN4 5′TLS and therefore reinitiation (Mohammad et al., 2017).

These structural features include the sequence context of the uORF initiation and termination codons, the type of the initiation codon (AUG vs. near-cognate start sites), the surrounding secondary/tertiary structures of the mRNA, the length and sequence of the uORF(s), their number and position in the transcript, the distance from the uORF stop codon to the mCDS start codon (intercistronic distance), the number of uORFs, and peptide sequence (Calvo et al., 2009; Lawless et al., 2009; Ye et al., 2015; Johnstone et al., 2016; Lin et al., 2019; Jürgens and Wethmar, 2022; Zhong et al., 2024). Below, we present the principal uORF-mediated mechanisms involved in gene expression repression.

2.1 uORFs as triggers of ribosome stalling/dissociation and uORF-encoded peptides

Ribosome stalling is a mechanism intimately related to the uORF sequence. Indeed, translation of rare codons was seen to cause ribosomes to stall (Presnyak et al., 2015; Yu et al., 2015). Also, some combinations of adjacent codons (dicodons) can induce ribosome stalling and strongly affect protein synthesis, depending on their specific order (Gamble et al., 2016). By assessing the impact of thousands of variants of the coding region of yeast YAP1 uORF on its translational regulatory activity, it was found that uORF variants bearing rare codons and inhibitory codon pairs repress translation of downstream mORF, while the wild-type uORF sequence, and other variants based on common codons, could work as translational enhancers (Lin et al., 2019). Inhibition of downstream mORF translation, in these cases, likely occurs by stalling of ribosomes and consequent impairment of translation reinitiation (Young et al., 2016a; Lin et al., 2019) (Figure 1B). Interestingly, overexpression of complementary tRNAs reverses the inhibitory effects of uORF variants carrying inhibitory dicodons, suggesting that uORF codon effects are sensitive to changes in levels of charged tRNAs (Lin et al., 2019). The exact location of the rare codons within the uORFs also seems to influence their impact on translational regulation, as does the order of dicodons. Specifically, positions closer to the uORF start codon seem more sensitive to codon identity and order, and exert a greater influence on gene expression regulation. Ribosomes at such early stages in translation may be more sensitive to the availability of charged tRNAs than those that have elongated further (Lin et al., 2019). An example of a sequence-dependent regulatory uORF comes from the mRNA encoding methionine synthase, a key enzyme of the methionine synthesis pathway (Col et al., 2007). This mRNA has two inhibitory uORFs, where uORF2 is more efficiently translated. Its inhibitory function appears to be related to a clustering of rare codons for arginine and proline, which causes ribosome stalling. Mutation of the rare codons for arginine at positions +3 and +4 in uORF2 to either common codons for the same amino acid or common codons for alanine results in complete alleviation of translation inhibition. This suggests that uORF2 activity depends on the nucleotide sequence but not on the encoded peptide (Col et al., 2007). Thus, uORFs can repress translation of the downstream mCDS through codon usage. In fact, uORFs with rare codons increase mCDS repression by decreasing the ribosome scanning rate or by inducing ribosome stalling (Zhang et al., 2018; Lin et al., 2019).

To access ribosome dynamics in uORFs, in real-time with single-molecule resolution, Gaba et al. (2021) developed a new two-colour in vitro fluorescence assay, Smart-ORF, and applied it to uORF-encoded arginine attenuator peptide (AAP)-mediated translational regulation. This system allowed to find that ribosome stalling in the AAP uORF stimulates the uORF initiation efficiency and promotes clustering of slower uORF-translating ribosomes. In a different study (Bottorff et al., 2022), authors used experimental measurements on the well-studied uORF-containing 5′TLS of the human cytomegaloviral UL4 mRNA, which has 2 uORFs to test different kinetic models of uORF-mediated translational regulation. The second uORF (henceforth uORF2) in the UL4 5′TLS contains a terminal diproline motif that stalls ribosomes by disrupting peptidyl transferase center activity (Bhushan et al., 2010; Wilson et al., 2016). Using this experimental model, the authors showed that ribosomes stall in uORF2, and this phenomenon inhibits translation of the downstream mCDS due to collisions of scanning ribosomes with the stalled 80S ribosome, which causes ribosome dissociation rather than queuing. Authors also identify several human uORFs that have repressive terminal diproline motifs similar to the UL4 uORF2, which induce ribosome stalling (Bottorff et al., 2022).

Interestingly, the first report of a uORF-mediated translational regulation in response to copper was recently shown. The authors found that, in bacteria, small molecules can interact with the nascent uORF-encoded peptide, leading to ribosomal stalling and further inhibition of downstream mCDS translation in response to intracellular copper levels (Roy et al., 2022). Further studies may reveal uORF-encoded peptides as potential interactors with small molecules, causing ribosomal stalling. This strategy may be further investigated to develop novel cancer therapies based on uORF-encoded peptides that interact with molecules to inhibit the expression of oncogenic proteins.

Opposing the above examples of ribosome stalling/dissociation, there are cases in which the uORF-encoded peptide is involved in the uORF-mediated regulatory mechanism. The ability of nascent peptides to function as translational cis-regulatory factors can be explained by their direct or indirect contact with the translational machinery, inhibiting proper ribosomal progression or termination. This seems to be the case for the S-adenosylmethionine decarboxylase (AdoMetDC) mRNA, which encodes a crucial enzyme in the polyamine biosynthesis pathway (Raney et al., 2002). The AdoMetDC transcript contains a small uORF in its 5′TLS that encodes a 6-amino acid peptide with the sequence MAGDIS. In conditions of high levels of polyamines, the 4^th and 5^th residues of the MAGDIS sequence interact with the last tRNA (tRNA^Ser). This leads to ribosome stalling, which depends on the wild-type sequence of the peptide encoded by the uORF, and suppresses translation of the mCDS (Law et al., 2001; Raney et al., 2002). In this case, the presence of polyamines is also a regulatory determinant, as these metabolites appear to stabilise the molecular interactions that stall the ribosome (Law et al., 2001; Raney et al., 2002). The involvement of the AdoMetDC uORF peptide indicates a complex method of translational control, in which the peptide influences its translation efficiency under various cellular conditions, potentially linking this regulation to cancer biology (Ruan et al., 1996).

In other cases, independent of metabolites, the direct interaction of the uORF-encoded peptide may stall or destabilise the translating ribosome, favouring its dissociation after uORF translation. Examples of this type of regulation were described for the mRNAs encoding the CCAAT-enhancer-binding protein homologous protein (CHOP) and the growth arrest and DNA damage-inducible protein 34 (GADD34). CHOP is a basic leucine zipper (bZIP) transcription factor that induces the expression of genes involved in many stress-corrective processes during stress conditions (Marciniak et al., 2004; Pakos-Zebrucka et al., 2016; Young and Wek, 2016). However, in chronic stress with continued eIF2α phosphorylation, prolonged expression of CHOP can induce apoptosis (Marciniak et al., 2004; Teske et al., 2013). A uORF in the 5′TLS of the CHOP mRNA inhibits mORF translation, resulting in its low expression during normal conditions (Jousse et al., 2001; Young et al., 2016a). The inhibitory nature of this uORF seems to be related to ribosomal elongation stall after translation of an Ile-Phe-Ile sequence (Young et al., 2016a). In the case of GADD34, its mRNA contains two uORFs, where uORF2 seems to be largely inhibitory to downstream translation (Lee et al., 2009; Young et al., 2015). It was suggested that translation of a Pro-Pro-Gly sequence juxtaposed to the stop codon of uORF2 results in an inefficient termination that increases ribosome dissociation from the mRNA (Young et al., 2015). It seems uORF2-mediated inhibition of mCDS translation is responsible for the low levels of GADD34 during basal conditions, when its activity is not required.

The uORF-encoded peptides have the additional potential of inhibiting translation in a trans-acting manner. An interesting example was described for argininosuccinate synthase (AS), an important enzyme in the recycling of citrulline to arginine, which is used in the production of the critical modulator of blood flow and blood pressure, nitric oxide, in endothelial cells (Pendleton et al., 2005). The AS gene can be transcribed from three alternative transcription initiation sites, resulting in different overlapping 5′LTSs of 92, 66, and 43 nucleotides (nts). The shortest 43-nt 5′TLS mRNA is the predominant species in endothelial cells, whereas the extended species represents less than 7% of the total AS mRNA (Pendleton et al., 2005). The extended 92- and 66-nt 5′TLSs contain an out-of-frame, overlapped uORF that inhibits expression of downstream AS mCDS (Pendleton et al., 2002). Surprisingly, this uORF found in the extended two forms, encodes a peptide that suppresses translation of the mORF at the predominant short form of endothelial AS mRNA (Pendleton et al., 2005), illustrating a different layer of translational regulation provided by uORFs.

2.2 uORFs as triggers of nonsense-mediated mRNA decay (NMD)

The mechanism of NMD is a conserved cellular surveillance pathway that targets and degrades transcripts containing premature termination codons (PTCs) (Maquat et al., 1981; Mendell et al., 2004; Silva and Romão, 2009; Brogna and Wen, 2009; Popp and Maquat, 2016; Fernandes et al., 2019; Nogueira et al., 2021). Such a mechanism prevents the production and accumulation of truncated, potentially toxic, or non-functional proteins, thus avoiding deleterious effects to the cells (Maquat et al., 1981; Mendell et al., 2004; Silva and Romão, 2009; Popp and Maquat, 2016; Fernandes et al., 2019; Nogueira et al., 2021). In general, PTCs trigger NMD if positioned more than 50–55 nts upstream of the last exon-exon junction (Nagy and Maquat, 1998; Kashima et al., 2006). This is considered the canonical rule of NMD, the 50-nts rule, and has been validated by the analyses of inherited PTCs (Rivas et al., 2015; Lindeboom et al., 2016) and those introduced by somatic mutations in human tumours (Lindeboom et al., 2016; Hu et al., 2017).

NMD also regulates the levels of many physiological mRNAs, whose levels must be maintained low in the cell (Smith and Baker, 2015). NMD regulates up to 10% of the human transcriptome (Mendell et al., 2004); the depletion of UPF1, or other NMD factors, experimentally inducing their upregulation, confirms them as NMD-targets. Several features in the transcripts can trigger the NMD mechanism. Alternative splicing patterns, in which exon-junction complexes (EJCs) become located downstream of the stop codon, mark the stop codon as premature, and the corresponding mRNAs as defective, targeting them to NMD (Mendell et al., 2004; Chan et al., 2009; Yepiskoposyan et al., 2011; Tani et al., 2012; Smith and Baker, 2015; Goetz and Wilkinson, 2017). Alternative polyadenylation can also introduce splicing events downstream of the normal stop codon, with the stop codon being recognised as premature, allowing these transcripts to be degraded by the NMD pathway (Kishor et al., 2020). The presence of long 3′UTRs can also affect the stability of the translation termination complex and, therefore, may trigger NMD (Maquat, 2004; Hogg and Goff, 2010). Another feature capable of triggering NMD is the presence of uORFs in a transcript, because the uORF termination codon can be perceived by the ribosome as a PTC, if located more than 50 nts upstream of an EJC (Mendell et al., 2004; Matsui et al., 2007; Gardner, 2008; Calvo et al., 2009; Chan et al., 2009; Yepiskoposyan et al., 2011; Tani et al., 2012; Johnstone et al., 2016; Goetz and Wilkinson, 2017; Torrance and Lydall, 2018; Russell et al., 2022). Several studies reported that uORFs correlate with lower steady-state mRNA levels and shorter mRNA half-lives (Matsui et al., 2007; Calvo et al., 2009; Johnstone et al., 2016). Indeed, in a recent study where thousands of yeast uORFs were analyzed, it was reported that nearly all AUG uORFs are robust NMD inducers (May et al., 2023). In addition, it was shown that longer uORFs are associated with increased NMD susceptibility, supporting the idea that uORF length and position are key factors in NMD activation (May et al., 2023). In a different study, using a random-forest machine-learning approach, the mRNA stability of zebrafish and human cells was also analysed, and uORFs were identified as a cis-element of mRNAs that promote NMD (Musaev et al., 2024).

The ATF4 mRNA is a well-documented example of a uORF-harbouring transcript that triggers NMD. ATF4 is a transcription factor induced under stress conditions that is essential to increase the expression of genes involved in nutrient import, metabolism, and alleviation of endoplasmic reticulum and oxidative stresses (Pakos-Zebrucka et al., 2016). The 5′TLS of the ATF4 mRNA contains two uORFs: the 5′-proximal uORF1 with three codons, and the uORF2, with fifty-nine codons, which overlaps out-of-frame with ATF4 mCDS (Lu et al., 2004; Vattem and Wek, 2004). During basal conditions, after translation of uORF1, translation re-initiation at the overlapping uORF2 inhibits mORF expression, and results in low levels of ATF4 (Vattem and Wek, 2004). This translational regulatory mechanism seems to be accompanied by induction of NMD, which contributes to low levels of ATF4 mRNA (Gardner, 2008). Supporting this idea, UPF1 knockdown induced the expression and stabilization of the ATF4 mRNA, whereas UPF1 overexpression significantly decreased ATF4 steady-state levels (Gardner, 2008). In a different example, the uORF-containing human inducible nitric oxide synthase (iNOS) gene, siRNA-mediated down-regulation of UPF1, i.e., inhibition of NMD, also resulted in an enhanced expression of the endogenous cytokine iNOS in human DLD-1 cells (Gather et al., 2019).

Despite the demonstrated link between uORFs and NMD, mRNAs bearing uORFs cannot always be considered NMD-targets. For instance, the mRNA that encodes the cytokine thrombopoietin (TPO), a glycoprotein hormone that induces platelet production, contains seven uORFs that can evade NMD (Stockklausner, 2006). Accordingly, and in opposition to the previously mentioned reports, a recent genome-wide study suggested that most of the translated uORFs do not affect the mRNA level and, thus, do not trigger NMD (Aliouat et al., 2020). However, this study also demonstrated that among a list of mRNAs having a uORF, those with the initiator AUG in a strong Kozak context may be direct NMD-targets (Aliouat et al., 2020). This points out that the context of the uORF start codon can be a key determinant of its ability to induce NMD, or it just shows how NMD is a translation-dependent mechanism. In a more recent study, the uORF length was analysed as a factor that can influence mRNA stability in mammalian cells. Results suggest that the decision of whether NMD ensues after translating uORFs occurs before 40S subunits resume scanning and initiate another translation event at the downstream mCDS start codon (Russell et al., 2023). These findings stress that one should be careful when suggesting that translated uORFs commonly trigger NMD, and that each case must be addressed individually.

2.3 uORFs as determinants of the downstream mCDS translational efficiency

Mammalian cells are exposed to various stress conditions, including hypoxia, oxidative stress, unfolded protein-induced endoplasmic reticulum (ER) stress, and metabolic stress, to which they need to respond. In fact, failing to respond to these stress conditions promptly leads to cell death, tissue dysregulation, and tumorigenesis (Barbosa et al., 2013; Silva et al., 2019). It is now known that uORFs are essential for the cell to respond to stress conditions. Indeed, uORFs are essential for regulating mammalian gene expression in response to these types of cellular stress. As previously discussed, if the presence of uORFs in a transcript does not trigger NMD, they can function as a determinant of translation efficiency, as they often act as roadblocks that inhibit translation of the downstream mCDS, allowing some proteins to be maintained at low levels under normal physiological conditions (Barbosa et al., 2013; Silva et al., 2019). However, under stress conditions, including disease states like cancer, while global canonical translation is impaired to avoid unnecessary waste, the uORFs are usually leaked by the ribosome, allowing the efficient translation of the downstream mORF and its upregulation (Barbosa et al., 2013; Silva et al., 2019). This mechanism occurs in a set of transcripts, for example, those that are part of the ISR, which must be simultaneously upregulated to respond to external cell stress. Paradigmatic examples of such transcripts are ATF4, CHOP and GADD34. These three transcripts encode proteins that are translated in conditions of eIF2α phosphorylation due to cellular stress, as part of the ISR stress-protective pathway (Pakos-Zebrucka et al., 2016; Young and Wek, 2016). Of note, this mechanism is conserved across species as an essential cellular strategy to cope with stress (Xiao et al., 2022; Dasgupta and Prensner, 2024). In a more recent study, where the global view of mRNA translation was analysed by ribosome profiling in human HEK293T cells following X-ray exposure, data showed that a significant subset of cellular mRNAs is subjected to translational upregulation by ribosomal bypassing of their uORFs (Hou et al., 2025). Together, these results illustrate well the role of uORFs in translational control in response to stress.

Interestingly, while specific uORF-mediated mechanisms allow the expression of proteins that are necessary for the cell to cope with stress, other mechanisms also involving uORFs inhibit translation not only in homeostatic conditions, but also in senescent cells (Vattem and Wek, 2004; Lee et al., 2009; Palam et al., 2011). In fact, in senescent cells, there are high levels of phosphorylated eIF2α, but maladaptive ISR is a hallmark of senescence that desensitizes senescent cells to stress; as a consequence, in these conditions, there is a translational deregulation of ATF4, which has been shown to occur due to the ineffective bypass of the ATF4 uORFs (Payea et al., 2024). In a different study, where the mechanistic responses of acute or chronic stress on the heart were analysed, a uORF localised in the 5′TLS of the tumour suppressor folliculin (FLCN) was identified as responsible for the downregulation of FLCN observed in cardiac hypertrophic growth (Doroudgar et al., 2019). This highlights the uORFs’ contribution to achieving the adequate translational control mechanism that appropriately regulates gene expression according to the cellular context (Young and Wek, 2016).

Other data have shown that uORFs, rather than controlling translation efficiency in response to external stimuli, can also regulate translation in response to internal pressures, such as cell differentiation, microtubule organisation or circadian rhythms (Figure 2). This is exemplified by the overlapping uORFs present in the 5′TLS of the human tumour necrosis factor alpha-induced protein 2 (TNFAIP2) mRNA, which were shown to be major regulators of TNFAIP2 levels during myeloid differentiation (Scholz et al., 2020). Based on a ribosome profiling dataset from human MCF7 breast cancer cells treated with interleukin 1β (IL1β), these authors identified TNFAIP2 as a novel target of uORF-mediated translational regulation, suggesting that a major uORF-dependent translational switch occurs during macrophage differentiation (Scholz et al., 2020). Thus, although there is a massive induction of TNFAIP2 mRNA expression in response to IL1β, TNFAIP2 protein levels remain low in this cell line. Since TNFAIP2 mRNA expression did not change between monocytes and macrophages, translational regulation of TNFAIP2 seems to be a major determinant of TNFAIP2 expression during differentiation. This suggests that during monocyte-to-macrophage differentiation, while the uORFs efficiently block TNFAIP2 protein synthesis in monocytes, they are inactivated in mature macrophages, thereby allowing a massive increase in TNFAIP2 protein expression (Scholz et al., 2020).

uORFs also have important roles in other cellular programmes, such as microtubule organisation (Phan et al., 2022) and circadian rhythm (Sun et al., 2025). For example, it has been shown that Drosophila circadian rhythm genes are enriched in uORFs (Sun et al., 2025). Regarding the function of these uORFs in the regulation of circadian rhythm and sleep, it was shown that the uORFs of the core clock gene Clock (Clk) rhythmically and substantially attenuate CLK protein translation in Drosophila, with pronounced translation suppression during daylight hours (Sun et al., 2025). This study by Sun and collaborators identified functional uORFs in Clk that repress CLK protein translation in both circadian and circannual modes, promoting morning wakefulness in adaptation to seasonal photoperiod changes and modulating global gene expression (Sun et al., 2025). Knocking out Clk uORFs increased CLK protein levels during the day and shortened the circadian cycle, along with a broad shift in clock gene expression rhythms (Sun et al., 2025). Also, uORF abrogation reduces dopaminergic activity, which augments morning sleep. In addition to this, mutants also showed decreased fecundity and lower resistance to starvation, showing that the physiological consequences of Clk ablation are far beyond the regulation of circadian rhythms and sleep (Sun et al., 2025).

Regarding the function of uORFs in microtubule organisation, the dosage-sensitive Polo-like kinase 4 (PLK4) abundance, the enzyme that limits centriole duplication to once per cell cycle, is translationally controlled through conserved uORFs within its 5′ TLS to prevent excess protein synthesis (Phan et al., 2022). Using genetically engineered mice with homozygous Plk4 uORF knockouts, these authors demonstrated that fertility in these mice was dramatically reduced due to a significant depletion of primordial germ cells (PGCs) (Phan et al., 2022). The persistent PGCs in knockout mice contained extra centrioles, which are associated with increased cell division errors (Phan et al., 2022). Such PGCs undergo hypertranscription and hence have more Plk4 mRNA than somatic cells, but reducing Plk4 mRNA levels in knockout mice restored PGC numbers and completely rescued fertility (Phan et al., 2022). These data introduce a new pathway for regulating PLK4 expression levels and reveal a crucial role for uORFs in regulating translational control during gametogenesis (Phan et al., 2022).

Altogether, these three examples illustrate well the contribution of uORF-mediated translational regulation pathways to maintain cell homeostasis.

3 The function of uORFs and uPeptides in tumorigenesis

As referred above, uORFs regulate gene expression, but their translation can also originate uPeptides. uPeptides can be released to the cytosol where they enrich the cellular peptidome and function in regulatory pathways individually or in protein complexes (Silva et al., 2019; Zhong et al., 2024). However, uPeptides can also be processed by the endoplasmic reticulum resident aminopeptidases for presentation on MHC class I molecules to function as neoantigens with a great relevance in cancer immunotherapy (Nogueira et al., 2021; Jürgens and Wethmar, 2022).

3.1 uPeptides as neoantigens

Genome-wide cancer mutation analyses have revealed an extensive landscape of functional mutations, not only within the coding, but also in the noncoding genome. In effect, during tumorigenesis, mutations in genomic DNA and other alterations can originate aberrant transcripts, including abnormal lncRNAs, microRNA-target mRNAs, and NMD-target mRNAs (for example, splicing-, nonsense- or frameshift-mutated mRNAs), which may translate into aberrant peptides/proteins, i.e., neoantigens. If recognised by the immune system, such neoantigens may offer a promising avenue for immunotherapy by targeting these unique cancer-associated peptides. Accordingly, it has been shown that NMD inhibition, by using siRNAs targeting NMD factors in cultured human cancer cells, resulted in the upregulation of proteins encoded by splicing-, nonsense- and frameshift-mutated mRNAs, which may potentially lead to the expression of tumour-specific neoantigens (Nogueira et al., 2021; Xie et al., 2023).

Recent advances in the discovery of immunogenic neoantigens have, in fact, been instrumental for the development and approval of tumour immunotherapies, including cancer vaccines, adoptive T-cell therapy and antibody-based therapies, especially for solid tumours. Neoantigens are recognised as non-self and trigger an immune response that is not subject to central and peripheral tolerance. Compared to tumour-associated antigens, the highly immunogenic and tumour-specific neoantigens provide emerging targets for personalised cancer immunotherapies, and their presence serves as a prospective predictor of tumour survival and immune checkpoint blockade responses (Rech et al., 2018; Duperret et al., 2019; Duperret et al., 2023). Indeed, neoantigens can be presented by tumour cells’ major histocompatibility (MHC) class I and II molecules. When presented on the cell surface, the neoantigens can serve as the target of CD4⁺ helper T lymphocytes and CD8⁺ cytotoxic T lymphocytes for tumour infiltration, thus enhancing the immunity to cancer and paving the way for immunotherapy targeting neoantigens, which promote tumour eradication in patients (Minati et al., 2020; Jou et al., 2021; Nogueira et al., 2021; Zhang Z. et al., 2021; Zhu et al., 2021; Zhu et al., 2023; Liu et al., 2022; Xie et al., 2023; Du et al., 2024). Such neoantigens can offer a promising avenue for immunotherapy.

Neoantigens can also be generated by tumour cells as a result of various tumor-specific alterations, including dysregulated NMD, or “simple” mutations in genomic DNA, and other alterations that originate aberrant RNAs with different and/or altered uORFs, and/or different susceptibility to NMD, which, if not degraded, may be translated into tumour-specific aberrant uPeptides (Figure 3). These cancer cell-specific uPeptides also have the potential to be processed by the proteasome and aminopeptidases and presented to MHC molecules to function as neoantigens (Minati et al., 2020; Jou et al., 2021; Nogueira et al., 2021; Zhang Z. et al., 2021; Meraviglia-Crivelli et al., 2022; Xie et al., 2023; Du et al., 2024; Rosenberg-Mogilevsky et al., 2025). Indeed, uPeptides seem to represent a previously underestimated source of tumour antigens, as evidenced by studies that highlighted their utility in immunotherapy by leveraging the immune system’s capacity to recognise these unique antigens (Nelde et al., 2022; Lozano-Rabella et al., 2023; Kowar et al., 2025). For instance, Nelde et al. provided evidence that specific uORF-derived peptides can be processed into tumour-associated HLA I-presented peptides detectable in primary human samples, emphasising their role as immunogenic targets in cancer (Nelde et al., 2022) and, therefore, a promising avenue for developing targeted cancer therapies. Still in the field of immunotherapy, peptides arising from PD-L1 uORFs may play crucial roles in the development of immune evasion strategies in tumours, as they can act as self-antigens, which are crucial for tailoring cancer immunotherapy (Barragan-Iglesias et al., 2021). Also, the HMGA1-derived peptide enhances the presentation of immunogenic peptides in cancer cells, particularly those that are mitotically arrested (Kowar et al., 2025). The scope of uORF research continues to expand, with an increasing acknowledgement of their role in the tumour immunopeptidome (Zhu et al., 2021; Jürgens and Wethmar, 2022). This positions uORF-derived peptides as promising candidates for future cancer therapies and biomarker development.

3.2 uPeptides as functionally active units

Studies have shown that uORFs can also generate bioactive peptides that participate in regulatory networks influencing major oncogenic pathways. For example, uPeptides can affect cancer cell behaviour and the tumour microenvironment (Sendoel et al., 2017; Jürgens and Wethmar, 2022; Nelde et al., 2022). The significance of uORFs and their derived uPeptides in cancer development is an expanding research area with key implications for understanding cancer biology and developing treatments.

Furthermore, uPeptides may exert translation-independent biological functions in the cell that may not even be related to the functions of the protein encoded by the mCDS of the corresponding mRNA. For instance, the McKusick-Kaufman syndrome (MKKS) transcript contains two uORFs that are translated into highly conserved peptides that are translocated to the mitochondrial membrane, while the MKKS protein produced by the mORF localises at the cytoplasm (Akimoto et al., 2013). This shows that the uORF-encoded peptides may be functional in vivo in a distinct way from the main protein. More recently, it has been shown that a uORF located in the mRNA of the asparagine synthetase domain-containing 1 (ASNSD1) encodes a protein that is part of the chaperone complex called PAQosome, which is responsible for the biogenesis of a vast number of protein complexes in humans, while ASNSD1 is involved in the unrelated process of asparagine biosynthesis (Cloutier et al., 2020).

Additionally, uPeptides, including those from oncogenes such as MYC, have been implicated in malignancies due to their growth-promoting capacities. The peptide derived from MYC, MPEP, has shown a proclivity to enhance glioblastoma growth by interacting with tropomyosin receptor kinase B and downstream AKT-mTOR signalling pathways, highlighting its oncogenic potential in aggressive brain tumours (Li et al., 2024; Zhong et al., 2024). On the other hand, the peptide MP31, derived from PTEN, has been shown to inhibit tumorigenesis while sensitising glioblastoma cells to chemotherapy, suggesting a pivotal role in tumour suppression and underscoring the dual roles of uORF-derived peptides in cancer (Li et al., 2024; Zhong et al., 2024).

In an additional example, authors showed that the transcript encoding the human transcription factor AP-4 (TFAP4) contains a 66-codon uORF that functions as a tumour suppressor gene in gliomas (Wang et al., 2020). In fact, the corresponding peptide is downregulated in glioma cells and tissues, with this condition correlated with high levels of TFAP4 expression, and induction of cell proliferation, migration, and invasion. The same group also described the role of the zinc ribbon domain containing 1 antisense RNA (ZNRD1-AS1) 144 amino acid-uORF-encoded peptide in the vasculogenic mimicry formation of gliomas (Wang et al., 2020). Here, authors showed that upregulation of the 144-amino acids peptide inhibits proliferation, migration, invasion and vasculogenic mimicry formation within glioma cells, and induces degradation of the ZNRD1-AS1 through the NMD pathway (Wang et al., 2020). A similar example had previously been described for the transcription factor nuclear receptor subfamily 2 group C member 2 (NR2C2) uORF (Fan et al., 2018). Furthermore, the findings regarding GATA4 suggest a significant potential for uPeptides to act as essential components that could promote or worsen tumour development in specific contexts, such as cardiovascular tumours (Hedaya et al., 2025).

uPeptides can also form complexes with their associated main proteins, acting through both cis- and trans-regulatory mechanisms. This flexibility allows them to respond to cellular signals, potentially influencing translation in response to microenvironmental changes, particularly relevant in cancer contexts (Jürgens et al., 2021). The ability of these peptides to dynamically regulate gene expression could play a significant role in tumour progression. The uPEP2 peptide, a kinase inhibitory peptide, which inhibits PKCη activity, is encoded by a uORF upstream of the PKCη mCDS (Belsamma et al., 2023). This peptide directly inhibits PKCη activity, a crucial regulator of cancer cell survival and migration, particularly in triple-negative breast cancer, exemplifying how specific uORF-derived peptides can serve as cancer agents, potentially contributing to reducing cancer cell survival (Jayaram et al., 2021; Belsamma et al., 2023). As for renal cancer, the VHLα-CUG alternative translation regulates hnRNPA2B1 via NMD, with complex interactions within these cancer pathways (Liu et al., 2020). Table 1 summarises a selection of uPeptides reported in the literature as implicated in tumorigenesis.

TABLE 1. Example of upstream open reading frame (uORF)-encoded peptides (uPeptides) relevant to tumorigenesis.

Regulation of translation by uORFs and uORF-encoded peptides open new avenues for cancer therapy. Modulating uORF activity could control oncogenes or tumour suppressor genes, affecting tumour growth and suggesting novel treatments. Studying uPeptides offers key insights into their role in cancer, such as boosting immune recognition and altering cancer cell signalling. These findings highlight the importance of further uORF research in cancer biology, potentially leading to new therapies.

3.3 The interplay between NMD, uORFs and uPeptides

Because of its key role in transcriptome modulation, NMD has been implicated in several biological processes and in the pathophysiology of numerous human genetic diseases, including cancer. The involvement of NMD in cancer development has been extensively studied during the past few years. It was shown that different cancer contexts can exploit both NMD quality control and regulatory functions to potentiate disease development (Popp and Maquat, 2018). For example, tumour-suppressor genes were shown to exhibit NMD-inducing features, providing an opportunity for NMD to regulate their expression and potentiate cells’ sensitivity to cancer development. Indeed, cancer cells may utilise NMD to suppress tumour-suppressor genes or escape immune detection mechanisms (Reddy et al., 1995; Perrin-Vidoz, 2002; Ware et al., 2006; Pinyol et al., 2007; Anczuków et al., 2008; Mort et al., 2008; Lindeboom et al., 2016; Pawlicka et al., 2020). On the other hand, in some tumours, NMD factors harbour disabling mutations that impair NMD-mediated degradation and its protective role against tumorigenesis. In fact, abnormal regulation of NMD components can cause the accumulation of mRNAs that support oncogenic behaviours, a phenomenon observed across various cancer types (Liu et al., 2014; Lu et al., 2016). In lung cancer, Baudu et al. demonstrated that low levels of NMD factors during epithelial-mesenchymal transition (EMT) are inversely related to GABARAPL1 expression. This suggests that GABARAPL1 may be regulated post-transcriptionally by NMD, implying a role in cancer cell adaptability during EMT (Baudu et al., 2021). In line with this, NMD inhibition, observed under stress conditions, such as those involved in carcinogenesis, was shown to upregulate the expression of many important factors that can aid tumorigenesis, including proteins involved in cell growth, cell cycle, apoptosis and cell migration (Wang et al., 2011). Moreover, some data have shown that inhibiting NMD leads to increased mRNA levels of genes often linked to cancerous traits, suggesting NMD might serve as an essential tumour suppressor pathway (Ottens et al., 2017).

All in all, modulation of NMD in cancer development is not straightforward, as it can play both pro- and antitumorigenic roles depending on the context. Cells can use NMD to inhibit gene expression of several tumour suppressors, like BRCA1, p53, and Wilm’s tumour 1, that acquire mutations during tumour progression, and that would otherwise convert into dominant-negative oncoproteins if NMD would not eliminate the corresponding transcripts [reviewed in Rehwinkel et al. (2006)]. On the other hand, some tumours inhibit NMD to stabilise the expression levels of transcripts encoding proteins that allow them to adapt to the tumour microenvironment. Among such mRNAs are those encoding components of the ISR, which are needed to restore cellular balance upon stress, as previously referred (Popp and Maquat, 2018).

Research has shown that the NMD machinery frequently recognises uORF-containing mRNAs. Russell et al. demonstrated that translation termination at uORFs could be detected by the NMD machinery in a manner similar to that of real PTCs, leading to degradation of the corresponding mRNAs (Russell et al., 2022). This suggests that NMD modulates the stability/levels of uORF-containing transcripts, with significant implications for cell growth and differentiation (Nyikó et al., 2009), features often disrupted in cancer. However, some studies indicate that not all uORFs trigger NMD in all cases. Under specific physiological conditions, certain uORFs may help regulate translation without causing mRNA decay, thereby protecting oncogenic transcripts (Wang et al., 2020). This highlights a complex regulatory role where uORF translation supports normal cellular activities, but misregulation could promote tumour development. In addition, the interplay between uORFs and NMD is a vital regulatory mechanism with essential implications for tumour development, as dysregulation of either mechanism can promote tumorigenesis, as previously referred.

Additionally, malignant transformation occurs concomitantly with transcriptomic reprogramming (Huang et al., 2025; Wang et al., 2025), driven by distinct mechanisms, including NMD inhibition (Nogueira et al., 2021; Patro et al., 2024). Since NMD can modulate the stability of mRNAs with uORFs, its inhibition can increase the expression of diverse uPeptides in diverse cancer types (Lindeboom et al., 2019; Pawlicka et al., 2020; Becker et al., 2021; Meraviglia-Crivelli et al., 2022). Some of these uPeptides may be highly immunogenic, and thus, able to trigger a significant antitumor immune response. For example, it has been shown that NMD affects MHC-I expression, thereby influencing how immune cells recognise tumours, highlighting the broader implications of NMD in immuno-oncology (Becker et al., 2021; Nasif et al., 2023; Yang et al., 2025). The presence of a sequence within the 5′TLS capable of encoding a uPeptide can lead to ribosome stalling, which induces NMD and therefore tumour cells become resistant to immunotherapy, promoting tumour growth and development (Figure 3, left-hand panel). Also, in mouse models of metastasis, inhibition of NMD has decreased tumour growth and increased T-cell infiltration (Pastor et al., 2010; Litchfield et al., 2020). Thus, inhibition of the NMD pathway may have therapeutic benefits for different cancers. In this case, the expressed uPeptides function as neoantigens, contributing to the immunopeptidome and being presented to MHC-I by T cells, enabling tumour cells to respond to immunotherapy (Figure 3, right-hand panel).

As discussed, the link between uORFs and NMD is essential for controlling the expression of uPeptides, affecting cellular responses that may contribute to tumour development and treatment. Table 2 summarises cancers in which there is a demonstrated interplay between NMD and uORFs, revealing another layer of complexity in gene expression regulation in malignancy. Further investigation into this relationship is likely to uncover additional complexities relevant to cancer biology and treatment.

TABLE 2. Example of transcripts (dys)regulated by the interaction between nonsense-mediated decay (NMD) and upstream open reading frames (uORFs), along with the observed effects in various cancer types.

4 Conclusion

The work reviewed here shows the importance of uORF-mediated translational regulation pathways, NMD, and uORF-encoded uPeptides, for the maintenance of cellular homeostasis. In addition, we discuss how perturbations in cellular homeostasis may contribute to the redefinition of a novel cellular transcriptome, in which novel uORFs are recognised by the ribosome and impact gene expression, promoting malignant transformation. Taking into consideration the relevance of uORF-mediated translation regulation pathways in tumorigenesis and since it has been shown that, during uORF translation, small molecules can interact with nascent peptides and cause stalling of the ribosome to inhibit downstream translation, we can hypothesise that the use of uPeptide-interacting molecules may open up a new therapeutic strategy for cancer. In addition, it might be advantageous to screen for and identify specific small molecules that may induce ribosomal stalling and NMD induction of uORF-containing transcripts that encode oncogenes, in order to inhibit their expression, as a potential cancer therapy approach.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Tatsuaki Kurosaki, University at Buffalo, United States

Reviewed by: Mateusz Wawro, Jagiellonian University, Poland

Lara Jürgens, University Hospital Münster, Germany