Interlaced roles of mitochondrial DNA in colorectal cancer: Liquid-biopsy biomarkers, nuclear mtDNA-driven genomic instability, and mito-encoded micro peptide signaling

Abstract

Colorectal cancer (CRC) is a widely occurring malignancy with significant mortality on a global scale, making up close to 10% of all diagnosed cancers in 2020. While traditional CRC diagnostics and research have focused on nuclear genomic alterations, emerging evidence has highlighted the multifaceted roles of mitochondrial DNA (mtDNA) in the pathogenesis and clinical management of CRC. In this review, we examine three interlaced aspects of mtDNA in CRC: (1) Liquid biopsy biomarkers: Cell-free mtDNA circulating in the blood serving as a minimally invasive diagnostic and monitoring tool; (2) Nuclear mtDNA (NUMT)-driven genomic instability: The somatic nuclear incorporation of mtDNA (NUMT segments, or NUMT), contributing to mutational burden and chromosomal disruption in tumours; and (3) Mitochondria-encoded micropeptide signaling - small peptides encoded by the mtDNA that modulate cellular pathways and tumour behaviour. Recent high-impact studies have demonstrated that tumour-derived mtDNA in biofluids can augment cancer detection sensitivity, although technical challenges remain due to mtDNA fragmentation and background noise. Meanwhile, genomic analyses have uncovered a significant increase in NUMT insertion events in CRC cells, linking mitochondrial genome escape to nuclear genome instability and identifying potential numtogenesis suppressor genes. A novel dimension of mito-nuclear interactions in cancer was discovered in mitochondrial microproteins, such as humanin and mitochondrial open reading frame of the 12S rRNA type-c. Humanin exhibits both tumour-promoting and cytoprotective properties under specific conditions, while mitochondrial open reading frame of the 12S rRNA type-c possesses tumour-suppressive activities under other conditions. The outcomes of clinical, mechanistic, and translational research, revealing how mtDNA-based biomarkers and involvement contribute towards early detection, prognostication, and treatment of CRC, are presented.

Key Words: Proteomic; Multi-omics; Gastrointestinal malignancy; Precision oncology; Biomarker discovery; Therapeutic resistance

Core Tip: This minireviews highlights the emerging roles of mitochondrial DNA (mtDNA) in colorectal cancer beyond its role in energy metabolism. Three interrelated domains are focused on in this paper: (1) Cell-free mtDNA as a liquid biopsy biomarker of exceptional sensitivity; (2) Nuclear mtDNA fragments as a component of genomic instability; and (3) Mitochondria-derived peptides, such as humanin and mitochondrial open reading frame of the 12S rRNA type-c, as regulators of tumour behaviour. These insights reveal novel diagnostic and therapeutic potential by targeting mitochondrial-nuclear cross-talk in colorectal cancer.

INTRODUCTION

Colorectal cancer (CRC) ranks third in how often it’s diagnosed worldwide. It is still the second leading cause of cancer death globally despite therapeutic advances[1]. In 2020 alone, around 1.9 million cases of CRC were recorded, and approximately 0.93 million died from the disease[2]. Early detection and precise molecular characterisation of CRC are critical for improving patient outcomes. However, conventional approaches often rely on invasive tissue biopsies that capture only static snapshots of the tumour. These biopsies can be limited by sampling bias and cannot easily reflect the spatial and temporal heterogeneity of tumour cell populations[3]. In recent years, liquid biopsy techniques - the analysis of tumour-derived material (DNA, cells, vesicles, etc.) circulating in body fluids - have emerged as powerful and minimally invasive complements to tissue biopsy[4,5]. Circulating tumour DNA (ctDNA) in blood plasma can reveal tumour-specific mutations and dynamic changes in real-time, correlating with disease burden and treatment response[6]. Indeed, multiple studies have shown a robust correlation between the genomic profiles of CRC tissue and matched plasma ctDNA. In some cases, plasma analysis has identified additional mutations not detected in the tissue[6]. However, detecting ctDNA, especially in early-stage patients, is technically challenging, as total cell-free DNA (cfDNA) in plasma is highly fragmented (with a modal size of 166 bp resulting from apoptotic nucleosomal fragmentation) and often diluted by an excess of wild-type DNA from normal cells[6-8]. The proportion of cfDNA originating from tumours may fall below 1% during the initial phases of disease[9]. These issues necessitate extremely sensitive methods and motivate the search for new biomarkers that can enhance the signal of tumour presence.

Mitochondrial DNA (mtDNA) has attracted considerable interest. Human cells contain hundreds to thousands of mitochondria, each with dozens of copies of a 16.6-kb circular mtDNA genome, meaning that a single cell carries vastly more copies of mtDNA than nuclear DNA[6,10,11]. mtDNA is also prone to mutations at a rate higher than that of the nuclear genome, and CRC (like many cancers) often harbours somatic mtDNA alterations[6]. These features raise the intriguing possibility that tumour-derived mtDNA may be leveraged as a cancer biomarker with heightened sensitivity. Assays directed against mtDNA would even be able to identify small tumours if dying tumour cells contributed mitochondria or mtDNA fragments into the circulation, and given the copy-number predominance of mtDNA[6,11]. Tumour cell populations become clonally marked by mtDNA mutations, such as those in respiratory-chain subunit or mitochondrial rRNA genes. Researchers have described early evidence that mutated mtDNA is easily detectable and present in the body fluids of those affected by cancer, and is 19-220-fold more abundant than mutant nuclear DNA (e.g., p53) in the same sample[12]. Read as a whole, the data indicate that mtDNA is a more discriminating liquid-biopsy analyte[12]. Despite this promise, the implementation of mtDNA-based diagnostics has been hindered by knowledge gaps and technical hurdles. Until recently, the landscape of cell-free mtDNA in cancer has remained underexplored[12].

Beyond the realm of cfDNA, mitochondria-nucleus crosstalk in cancer biology extends even further. Mitochondria are not only powerhouses of the cell but also potential instigators of genomic instability. Over evolutionary time, the nucleus has received considerable genetic input from the mitochondrial genome, leaving behind numerous nuclear-embedded mtDNA segments known as nuclear mtDNA (NUMT)[13]. While most mtDNA sequences embedded in the human genome are ancient and non-coding, recent genomic studies have revealed that mtDNA transfer into the nucleus is ongoing, even in modern humans[13]. In cancer cells, characterized by the breakdown of genomic maintenance and rampant DNA damage, the rate of new NUMT insertions appears to be significantly elevated[13]. This raises the prospect that rogue mtDNA integration events may disrupt nuclear genes or regulatory regions, contributing to tumorigenesis. The notion of numtogenesis, the generation of new NUMTs, has been proposed as a mechanism linking mitochondrial dysfunction to nuclear genome instability in cancer[14]. Unravelling this link could shed light on how mtDNA copy number and integrity influence cancer development, beyond the conventional bioenergetic and apoptotic roles of mitochondria.

Finally, the groundbreaking area of mitochondrial biology has been expanded with the discovery of mitochondria-encoded micropeptides[15]. The human mitochondrial genome, whose genetic output is assumed to comprise merely 13 proteins (components of the oxidative phosphorylation machinery), harbors several short open reading frames (ORFs) that produce bioactive peptides[15-17]. These mitochondria-derived peptides (MDPs) include humanin (HN), a 24-amino-acid peptide encoded within the 16S rRNA gene, and mitochondrial ORF of the 12S rRNA type-c (MOTS-c), a 16-amino-acid peptide encoded in the 12S rRNA gene[15,18]. MDPs can act as signaling molecules both in the intracellular and extracellular environments and have been implicated in key processes including metabolic regulation, stress response, and cell survival[15,18,19]. In the context of cancer, these peptides blur the line between mitochondria and endocrine-like signaling that modulates cell behaviour. HN, for example, is known for its cytoprotective and anti-apoptotic effects in neurons and other cell types, mediated in part by binding to a trimeric receptor complex that includes the gp130 cytokine receptor subunit[18,20]. Such pro-survival signaling could be a double-edged sword in oncology, protecting normal tissues from chemotherapy toxicity (a proposed therapeutic use of HN analogues) but also potentially protecting tumour cells from death[18]. Defining the functional role of MDPs in cancer development and refractoriness to treatment is a nascent but crucial frontier.

In this review, we present a comprehensive examination of how mtDNA contributes to CRC through three avenues: Liquid biopsy biomarkers, NUMT-driven genome instability, and MDPs signaling, drawing on recent high-impact studies. Here, we discuss the clinical relevance of cell-free mtDNA measurements in CRC and how they might enhance the current diagnostics. We then delve into genomic findings linking mtDNA integration to CRC mutation landscapes and highlighting mechanistic insights from model systems (including the identification of a NUMT suppressor gene). Finally, we explored the role of MDPs in cancer biology, illustrating how these microproteins can exert either pro-tumour or anti-tumour effects and might be harnessed for novel therapeutic strategies. By integrating clinical, mechanistic, and translational perspectives, we aimed to underscore the diverse but interconnected roles of mtDNA in colorectal carcinogenesis and to point toward future directions in this emerging field.

LITERATURE SEARCH AND SELECTION

We conducted a comprehensive literature search to gather relevant studies on mtDNA and CRC published in the last decade (2015-2025). The primary databases queried were PubMed/MEDLINE and Scopus, using combinations of keywords such as “colorectal cancer”, “mitochondrial DNA”, “mtDNA mutations”, “cell-free mtDNA”, “liquid biopsy”, “NUMT” (NUMT segment), “genomic instability”, “mitochondrial peptides”, “humanin”, and “MOTS-c”. We also manually searched the reference lists of key articles and included older seminal studies frequently cited in recent literature. Our search focused on high-impact peer-reviewed journals and recent findings to ensure an up-to-date and authoritative review. Non-peer-reviewed sources, preprints, and anecdotal reports were excluded to maintain their scientific rigour.

Inclusion criteria

We included original research articles, clinical studies, and high-quality review articles that provided data or insights on: (1) Cell-free mtDNA as a biomarker in cancer (with a focus on CRC when available, but also drawing from other cancer types for general principles); (2) The occurrence and consequences of mtDNA insertions into nuclear DNA (NUMTs) in human cancers, especially gastrointestinal malignancies; and (3) The identification and functional characterisation of mitochondria-encoded micropeptides in cancer biology. Both clinical studies (e.g., diagnostic or prognostic analyses) and mechanistic laboratory studies (e.g., cell and animal models investigating mtDNA or MDPs function) were included to cover the translational aspects.

Data extraction and synthesis

From the selected publications, we extracted key findings relevant to the review themes. For cell-free mtDNA, we noted quantitative findings (e.g., mtDNA copy number or fraction in patients vs controls), associations with disease characteristics (stage, response, survival), and technical considerations (fragment size, sequencing concordance, etc.). For NUMTs, we recorded the frequency of somatic mtDNA integrations in CRC and other cancers, any identified patterns or “hotspots” of insertion, and experimental evidence linking NUMTs to functional outcomes (such as gene disruption or correlation with clinical features). For mitochondrial peptides, we compiled evidence of their expression levels in cancers, downstream signaling pathways, and effects on tumour phenotypes (growth, apoptosis, metastasis, and therapy response). We paid special attention to studies offering clinical or translational insights, such as those measuring peptide levels in patient samples or testing interventions in preclinical models.

Quality assessment

Although formal quality scoring was not performed (given the narrative scope of this review), we prioritised larger cohort studies, high-throughput genomic analyses, and reproducible experimental findings. Contradictory results between studies were noted and discussed to provide a balanced perspective. Throughout the manuscript, we cite sources for all specific data points and direct claims, ensuring that the content is grounded in published evidence rather than conjectures.

By integrating information from these diverse sources. This article brings together contemporary insights into mtDNA’s role in CRC and identifies areas where further research is needed. The following section presents the findings in a structured manner, corresponding to the three main topics of interest: Liquid-biopsy mtDNA biomarkers, NUMT-driven instability, and mito-encoded micropeptides.

MTDNA IN LIQUID BIOPSY FOR CRC

Cell-free mtDNA: Levels and fragmentation patterns

In a healthy cell, unlike nuclear DNA, which is limited to two copies, mtDNA is found in thousands within a cell; therefore, when cells undergo apoptosis or necrosis, mitochondrial genomes are released in large numbers[10,11]. This high copy number is a double-edged sword for liquid biopsy applications. On one hand, it means that circulating cfDNA in blood often contains a substantial portion of mtDNA fragments, and it has been observed that plasma mtDNA content can be higher than that of nuclear DNA in both healthy individuals and cancer patients[6,9,21]. For example, a recent CRC study reported average plasma mtDNA levels of the order of 10³ copies per mL, significantly exceeding the approximately 10² copies/mL of nuclear cfDNA[6]. This suggests that mtDNA-based assays could potentially detect cancer signals that might be missed by nuclear DNA assays, especially if leveraged with sensitive methods[6]. However, the majority of cf-mtDNA originates from normal cell turnover, and circulating mtDNA fragments are often highly fragmented (many under 100 bp in length)[6]. Such fragmentation necessitates careful assay design [short polymerase chain reaction (PCR) amplicons or deep sequencing] to capture mtDNA signals.

Intriguingly, the foundational cf-mtDNA levels observed among patients with cancer vs healthy individuals have yielded some surprising observations. One might hypothesise that tumours, with their high cell turnover, would shed extra mtDNA into circulation. Some data support this: A 2023 pan-cancer analysis of over 650 patients (including 71 CRC cases) found that the mtDNA fraction - the proportion of cfDNA reads derived from mitochondria - increased in several cancers, notably colorectal, pancreatic, liver, prostate, and cholangiocarcinoma, compared to that in healthy controls[12]. In that study, higher plasma mtDNA fractions correlated with higher tumour-derived ctDNA fractions. Importantly, a subset of cancer patients with low detectable nuclear ctDNA still showed elevated mtDNA, suggesting that mtDNA could act as an independent diagnostic clue[12]. By integrating mtDNA metrics with nuclear DNA mutation and copy number analyses, the authors improved the accuracy of cancer detection (raising the area under the curve from 0.73 with standard markers to 0.81 with mtDNA added)[12]. This indicates that circulating mtDNA can carry tumour-specific information and boost the performance of liquid biopsy (Table 1).

Table 1 Summary of liquid-biopsy and mitochondrial pathways in colorectal cancer: From biomarkers to therapy.

Theme	Scope	Key finding	Mechanism	Clinical implication
Liquid biopsy: Cf-mtDNA for detection and monitoring	CRC cohorts (supported by pan-cancer WGS including CRC)	Adding cf-mtDNA fraction features to standard ctDNA models improves CRC detection (AUC: 0.73 to 0.81)[12]	Because cells harbor thousands of mtDNA genomes, apoptosis/necrosis releases abundant yet highly short circulating fragments that require short-amplicon or deep-sequencing assays[6,10,11]	Quantifying cf-mtDNA fractions can serve as early-warning or pharmacodynamic markers and should be integrated with nuclear features for best performance[12]
cf-mtDNA variability and pre-analytics	CRC (assay/stage-dependent)	Direction of change differs by method/cohort: Pan-cancer WGS shows elevated cf-mtDNA fraction in CRC, yet a CRC dPCR study found lower pre-treatment plasma mtDNA in patients vs controls (approximately 684 copies/mL vs approximately 1081 copies/mL)[6]	Differences likely reflect disease stage, tumor mtDNA content, patient physiology, and pre-analytical/assay factors[6,12]	Pre-analytical standardization coupled with multi-analyte panels - such as combining mtDNA abundance with nuclear mutation/protein markers or augmenting protein panels with mtDNA quantitation - can enhance early detection and follow-up surveillance[12]
NUMTs burden in CRC	CRC tissue	Tumors show approximately 4.2-fold higher mtDNA-derived nuclear reads than matched normals (proxy for somatic NUMTs); higher burden associates with worse survival[14]	New NUMTs integrate at nuclear DSBs via error-prone MMEJ, with > 90% involving non-coding mtDNA (often D-loop)[13]	Evaluation of NUMT burden as a prognostic biomarker since higher NUMT levels associate with worse survival and reflect increased genomic instability[14,26]
NUMT suppressor (YME1 L1)	CRC tissue	YME1 L1 acts as a human NUMT-suppressor; 16% of CRC tumors harbored YME1 L1 mutations and knockout increased nuclear mtDNA[14,26]	Loss of the mitochondrial protease YME1 L1 relaxes mitochondrial quality control, permitting mtDNA leakage and nuclear integration[26]	YME1 L1-deficient tumors may represent a subset with unique vulnerabilities and higher numtogenesis, suggesting potential prognostic value and therapeutic angles[14,26]
HN/GP130 chemoresistance axis	Pan-cancer/GBMs evidence	HN promotes chemoresistance and growth via GP130-ERK signaling, promotes blood-tumor barrier formation and gp130 antagonists abrogate these effects in GBMs models[18,29,37-39]	HN signals via a gp130-containing receptor complex and triggers pro-survival cascades (STAT3/MAPK/PI3K-Akt); it inhibits Bax-mediated mitochondrial apoptosis[18]; in GBMs, HN → gp130 → ERK → DDR drives chemoresistance[29]	Identify HN-high subsets and test gp130-pathway blockade to re-sensitise tumors to chemotherapy (evidence in GBMs models)[29]
MOTS-c (MDPs) tumor-suppressive signaling	Preclinical/pan-cancer	There is decline in MOTS-c levels in aggressive disease and exogenous MOTS-c suppresses growth and dissemination in vivo[43]	MOTS-c is to the nucleus under stress, activates AMPK, and inhibits mTORC1 by interfering with USP7-mediated LARS1 deubiquitination[15,30-34,43,44]	Evaluate MOTS-c in CRC as a biomarker and therapeutic candidate: In MOTS-c-low settings, consider recombinant MOTS-c/analogs[43] or targeting downstream AMPK-mTOR axis (via AMPK activation and/or direct mTORC1 inhibition)[15,30-34]

cf-mtDNA: Cell-free mitochondrial DNA; CRC: Colorectal cancer; AUC: Area under the curve; mtDNA: Mitochondrial DNA; WGS: Whole genome sequencing; dPCR: Digital polymerase chain reaction; NUMTs: Nuclear mtDNAs; DSB: Double-strand break; MMEJ: Microhomology-mediated end-joining; YME1: Yeast mitochondrial escape-1; HN: Humanin; GBMs: Glioblastoma; ERK: Extracellular signal-regulated kinase; STAT3: Signal transducer and activator of transcription 3; MAPK: Mitogen-activated protein kinase; PI3K: Phosphatidylinositol 3-kinase; Akt: Protein kinase B; DDR: DNA damage response; MDPs: Mitochondria-derived peptides; MOTS-c: Mitochondrial open reading frame of the 12S rRNA type-c; AMPK: AMP-activated protein kinase; mTORC1: Mammalian target of rapamycin complex 1; LARS1: Leucyl-tRNA synthetase 1.

However, other studies have painted a more complex picture. In a 2021 proof-of-principle CRC study, Haupts et al[6] observed that pre-treatment plasma from CRC patients had lower mtDNA copy numbers on average than plasma from age-matched healthy individuals. They quantified cf-mtDNA by digital PCR and found a mean of 684 mtDNA copies/mL in CRC patients vs 1081 mtDNA copies/mL in healthy controls - a statistically significant difference[6]. The authors described this result as unexpected and hypothesised that advanced cancer may, in some cases, depress circulating mtDNA levels because of factors such as poorer patient performance status or aggressive clearance of cfDNA[6]. Additionally, it is known that solid tumours often have fewer mtDNA copies per cell than the corresponding normal tissue (potentially as an adaptation to metabolic reprogramming); therefore, a tumour could conceivably shed less mtDNA than an equivalent mass of healthy tissue[12,22,23]. These discrepant findings between studies, with one showing higher cf-mtDNA in CRC and another showing lower cf-mtDNA, underscore that cf-mtDNA dynamics can be context-dependent. Factors such as disease stage, tumour mitochondrial content, patient physiology, and technical assay differences may influence the outcome. Larger studies are needed to clarify the scenarios in which mtDNA increases or decreases in the circulation of patients with cancer.

Mitochondrial genomic alterations as biomarkers

In addition to the total mtDNA abundance, mutations present in mtDNA could serve as highly specific cancer biomarkers. Tumours frequently accumulate mtDNA mutations, including mutations in genes that are critical for oxidative phosphorylation. In a classic study, researchers detected identical clonal mtDNA mutations in both tumours and patient bodily fluids (such as urine or plasma), and notably, mutant mtDNA was far more abundant than mutant nuclear DNA in these fluids[12]. This abundance stems from the high copy number of mtDNA; even if only a fraction of mitochondria carry the mutation (heteroplasmy), the absolute number of mutant molecules released can be high. For CRC, sequencing of tumour tissue often reveals somatic mtDNA variants, some of which may enter circulation. In principle, a panel of tumour-specific mtDNA mutations can be tracked in the plasma as a personalised marker of disease burden or minimal residual disease.

However, in practice, detecting tumour-specific mtDNA mutations in the plasma has proven challenging. The aforementioned 2021 CRC study attempted deep sequencing encompassing the mitochondrial genetic material in its entirety in plasma cfDNA and matched tumour tissue[6]. They found that although each tumour had several mtDNA mutations, only 20% of the patients had any of the mutations detectable in plasma cfDNA[6]. Even in patients with metastatic disease, many plasma mtDNA mutations did not correspond to those in the tumour, and the plasma harboured additional mtDNA variants not found in the tumour tissue[6]. This discordance suggests that many mtDNA mutations found in the plasma might originate from sources other than the primary tumour, such as clonal hematopoietic cells (ageing-related mtDNA mutations in blood cells) or other benign processes. Technical sensitivity is another issue; due to the lower absolute quantity of mtDNA in plasma compared to tissue (often over 30-fold lower per unit DNA input), sequencing depth must be very high to catch low-frequency mutant fragments[6]. Additionally, plasma mtDNA is highly fragmented. One study showed that targeting shorter mtDNA amplicons (about 60 bp) doubled the measured mtDNA content compared to targeting longer regions (about 180 bp), indicating that many mtDNA fragments are truncated[6]. Therefore, comprehensive mtDNA mutation profiling from plasma may require specialised error-corrected ultra-deep sequencing or mutation enrichment methods. Digital PCR assays can be used to track known hotspot mutations or deletion junctions with high sensitivity; however, they are limited to mutations known a priori[6,24]. Tumour-agnostic approaches, such as untargeted sequencing for mtDNA mutations, face the hurdle of distinguishing true tumour mutations from technical artefacts and innocuous variants.

However, promising advances have been made in this regard. van der Pol et al[12] employed whole-genome sequencing of cfDNA to capture the total cf-mtDNA fraction as a collective signal, bypassing the need to call specific mutations. As noted, this approach showed that certain cancers, including CRC, have a higher mtDNA fraction than controls, and integrated models using this feature improved the detection rate[12]. An alternative strategy is to examine mtDNA fragmentation patterns or topology in plasma. The mtDNA is released not only as linear fragments, but also as intact or circular molecules and might be packaged within extracellular vesicles[12]. Differences in fragment size distribution or methylation status of mtDNA in cancer vs normal plasma could potentially serve as biomarkers (an area of ongoing research)[12]. Additionally, there is an interest in examining mtDNA in other biofluids for CRC, such as stool samples. While current multitarget stool DNA tests for CRC (e.g., Cologuard) focus on nuclear DNA mutations and methylation, mtDNA mutations or deletions in shed colonocytes or microbiome interactions should be explored in future studies.

Variation of NUMTs and cf-mtDNA in CRC progression and detection

In tumour tissue, the level of mtDNA typically diminishes as the disease progresses. To give an example, van Osch et al[11] established that tissues of colorectal carcinoma contained significantly lower quantities of mtDNA when compared to matched adenomas from the corresponding patients. Meanwhile, numtogenesis represents the process when mtDNA becomes incorporated into the nuclear genome. Srinivasainagendra et al[14] indicated that the genomes of colorectal adenocarcinoma on average contain 4.2 times more reads of mtDNA when compared to normal tissue, indicating numerous somatic insertions of mtDNA (NUMTs), which are not present in normal cells. Notably, an increase in the number of NUMT constructions is connected with reduced survival rates[14]. This may indicate that reduced function among the mitochondria reduces the level of oxidative stress, thereby preserving the tumor cells from reactive oxygen species-mediated damage and stimulating tumor progression[11].

A 2023 study with over 650 patients found that the amount of mtDNA (the part of DNA from mitochondria) in the blood was higher in several cancers, including CRC, than in healthy people[12]. In that study, adding mtDNA measurements helped find cancer earlier (the area-under-curve went up from 0.73 to 0.81 when mtDNA features were included)[12]. Haupts et al[6] found that CRC patients had lower levels of plasma mtDNA (about 684 copies/mL) compared to healthy people of the same age (about 1081 copies/mL). This suggests that biological and physical processes that affect how cfDNA is released and removed can vary among different groups[6]. Reliability of cf-mtDNA quantitation depends greatly on conditions before analysis; the way the received samples will be processed or analyzed may change cfDNA composition and content, hence would change results acquired[12]. Such results suggest that both cf-mtDNA and NUMTs significantly change with the progression of CRC.

NUMT-DRIVEN GENOMIC INSTABILITY IN CRC

Somatic mtDNA integration into the nucleus

The human nuclear genome carries numerous fragments of mtDNA acquired over evolutionary time, termed NUMTs. Historically, these NUMTs have been regarded as molecular fossils, most of which are ancient insertions that occurred in primates or earlier evolution[13]. However, recent high-resolution genome studies have upended this view by demonstrating that in human, relocation of mtDNA fragments into nuclear DNA continues as an active process[13,25]. In cancer, where DNA double-strand breaks and repair mishaps are frequent, the rate of new NUMT formation appears markedly elevated. Wei et al[13] analysed whole-genome sequences from over 12500 cancer cases and found de novo NUMT insertions in approximately 2.3% of tumours[25,26]. This corresponded to an estimated incidence of 3.6 × 10^-2 new NUMTs on a per-genome basis in cancer, which was an order of magnitude higher than the rate in germline cells (approximately 10^-3 per genome per generation)[13]. In other words, while a new NUMT might occur in only one in 10000 births, it occurs in approximately one in 1000 cancers[13]. This striking finding indicates that numtogenesis accelerates in the permissive environment of the cancer genome.

In this regard, not all cancers are equal. The frequency of tumour-specific NUMTs varied by cancer type: Renal cell carcinomas and CRCs showed four-fold fewer new NUMTs than breast cancers, and approximately 7.5-fold fewer than bladder cancers, which had the highest NUMT burden[13]. The reasons for these differences are not fully clear but could be related to the tissue-specific expression of factors that facilitate or inhibit mtDNA integration or differences across the scope of nuclear DNA damage and restoration processes active in these cancers[27,28]. Interestingly, across all cancers, > 90% of new NUMTs are involved in the insertion of non-coding mtDNA segments (such as pieces of the D-loop control region)[13]. This bias suggests that mtDNA most often integrates during attempts to repair double-strand break in the nucleus, a process that might capture any available DNA as a filler. The majority of NUMTs arise by microhomology-mediated end-joining[26]. Support for this mechanism comes from short mtDNA - chromosome microhomology tracts at the breakpoints[13]. In agreement with PRDM9-driven recombination, NUMT insertion is focused close to its binding sites, connecting the phenomenon to meiotic recombination machinery, and in repetitive genomic regions such as satellite DNA[13]. With these patterns in sight, association of NUMT formation with complex genomic rearrangements is frequently seen. Mito-chromothripsis in some tumours involves chromosomal fragmentation and reassembly with insertion of multiple mtDNA fragments[13,28].

NUMTs as a source of genomic instability

Additional intranuclear DNA is best viewed as a determinant with functional consequences and not an idiosyncrasy. Disruption may follow if a NUMT integrates into genic or regulatory DNA sequence. The article in Nature (2022) reported probable inactivation of a tumour suppressor gene by NUMT insertion in a myxoid liposarcoma sample, which was deemed a driver mutation central to the genesis of that tumour[13]. The genome-wide cancer analysis in the article identified NUMTs co-occurring with highly characterised cancer genes - ranging from fragile regions (e.g., FHIT) to repair machinery (e.g., FANCI) - in subsets of the cohort)[13]. In macro-genomic alteration, mitochondrial insertions influence the structure in which somatic variants are formed in cancer. A greater NUMT number may reflect a record of more pervasive genomic turbulence. A viability screen intrinsic to the circumstance removes overly large and deleterious NUMTs - especially essential-locus hits - before tumour outgrowth, which explains their low frequency in catalogues[13]. After filtering by biology, the size spectra and genomic location of remaining inserts are determined by both random events and filtering by evolution[13]. Further low-prevalence NUMTs occur in cancers, though they tend to be in noncoding regions and lack driver marks. Sequenced tumour genomes rarely show the most disruptive events, which are presumed to be cell-lethal[13]. The high diversity in NUMTs seen in human tumours (14.2% of individuals in that study had ultra-rare NUMTs unique to < 0.1% of the population) suggests a heightened state of genetic alteration burden in cancer, coupled with the omission of long-term purifying alternative that would otherwise remove such insertions over generations[13].

MDPs and the colorectal tumor-immune ecosystem

Cytoprotection across multiple tissues is attributed to the mitochondria-encoded micropeptide HN, with regulation of the mitochondrial apoptotic pathway achieved through interaction with Bcl-2 family proteins[18]. Through intracellular binding to Bax, tBid, and BimEL, HN antagonizes pro-apoptotic Bcl-2 signaling and blocks mitochondrial cytochrome-c release[18]. Neuronal protection is inferred from HN engagement with receptor complexes comprising ciliary neurotrophic factor receptor/WSX-1/gp130[20]. HN elicits protumorigenic effects through GP130-dependent activation of extracellular signal-regulated kinase (ERK)[29]. Cheng et al[29] mentioned specific mediation of tumor-supportive effects are done by extracellular HN.

Segregation of glioblastoma (GBMs) into high- vs low-sensitivity groups toward HN has been documented[29]. Bidirectional signaling between Glioma-associated microglia or macrophages and GBMs induces HN, conferring increased temozolomide (TMZ) tolerance in the HN-responsive subset[29]. Suppressed apoptotic signaling, reduced chemotherapeutic antineoplastic and anti-metastatic impact, and potentiated oncogenic drive characterized peripheral HN exposure in triple-negative breast cancer (TNBC) models together accelerating primary tumor expansion and spontaneous pulmonary dissemination[18,30]. Maintenance of metabolic homeostasis by MOTS-c (MDPs) involves AMP-activated protein kinase (AMPK)-dependent signaling and the direct regulation of adaptive nuclear gene expression subsequent to nuclear translocation[31]. Here, concordant lowering of MOTS-c in serum and intratumoral compartments characterizes ovarian cancer and portends an unfavorable prognosis[32,33]. Administration of MOTS-c from exogenous sources limits osteoclast cell propagation and dissemination capacity and drives checkpoint arrest and apoptosis[32,33]. In vivo MOTS-c exposure rebalances mouse serum cytokines toward an anti-inflammatory profile with mechanistic attribution to AMPK activation and inhibition of the mitogen-activated protein kinase/c-Fos cascade mediating analgesia and anti-inflammation[15].

Gene Ontology biological process analysis demonstrates enrichment across metabolism, oxidative-stress regulation, immune activity, and nuclear trafficking meeting the criterion of false discovery rate < 15%[31]. Levels of HN, HUS1, and GP130 stratify TMZ responsiveness[29]. GP130-ERK1/2 activation by nanomolar HN generates an ataxia telangiectasia and Rad3-Related protein-dependent DNA damage response profile that provides resistance to TMZ resistance[29]. In support of TMZ resistance reversibility came from the inhibitor: Inhibition of MEK/ERK restored sensitivity, but GP130 inhibition resulted in larger re-sensitization[29]. With favorable tolerability demonstrated, MOTS-c caused notable in vivo growth suppression of ovarian cancers[33]. MOTS-c deficiency in CRC renders invading T cells ill-prepared, while restoration of pathway provokes a proportionate anti-tumor response.

CRC-specific findings

Focusing on CRC, earlier studies provided the first clues that numtogenesis is active in this disease. Srinivasainagendra et al[14] analysed whole-genome sequences from 57 colorectal tumours and their matched normal blood DNA. They found that CRC genomes contained, on average, a 4.2-fold higher proportion of mtDNA reads than normal genomes, indicating a significant excess of somatic mtDNA integrations in tumours[14]. In other words, mtDNA jumped into the nucleus during tumour development, leaving multiple new NUMTs per tumour that were absent from the patient’s normal DNA. In this study, the authors introduced the term “numtogenesis” to denote the process of mtDNA migration to the nucleus in somatic cells[14,26]. Strikingly, there was a gender difference: Women’s CRC tumours had higher NUMT content than men’s tumours, although the reasons for this remain speculative (potentially related to hormonal influences on the mitochondria or differences in oxidative damage)[14]. They also observed that tumours with higher NUMT levels were associated with worse survival, although the sample size was limited to draw firm conclusions[26]. This raises the provocative idea that rampant numtogenesis might be a marker of aggressive disease, perhaps because it signifies a breakdown in genomic integrity.

A particularly important discovery from the 2017 CRC study was the identification of a NUMT suppressor gene. The authors noted similarities between the numtogenesis phenomenon in CRC and a classic yeast mutant called yeast mitochondrial escape-1 (yme1), in which the loss of the YME1 gene causes mtDNA to leak into the nucleus[14,26]. Humans have a homolog of yeast YME1, known as YME1 L1, which encodes a mitochondrial inner-membrane protease involved in protein quality control and mitochondrial maintenance[26]. Upon examining sequencing data, 16% of the CRC tumours had mutations in YME1 L1, a significant enrichment[26]. Moreover, mining the The Cancer Genome Atlas Pan-Cancer data revealed that YME1 L1 is mutated at a notable frequency in several other tumour types as well[26]. This suggests that the loss of YME1 L1 may permit mtDNA escape. To directly test this hypothesis, researchers knocked out YME1 L1 in human cells in vitro. The result showed a dramatic increase in mtDNA presence in the nuclear fraction of cells[26]. In essence, eliminating YME1 L1 function led to a surge of numtogenesis, confirming that YME1 L1 normally acts to suppress the movement of mtDNA to the nucleus[14]. Conversely, in a clever cross-species experiment, introducing human YME1 L1 into yeast cells lacking YME1 partially rescued the hyper-numtogenesis phenotype, reinforcing the conserved role of this gene across species[26]. YME1 L1 thus emerged as the first bona fide NUMT suppressor identified in humans[14]. Mechanistically, YME1 L1 can mediate the breakdown or sequestration of leaked mtDNA fragments into the cytosol or guarantee mitochondrial integrity to prevent leakage of mtDNA when there is mitochondrial stress.

Besides point mutations and chromosomal aneuploidy, intercompartmental DNA transfer is important in mutation burden and evidence proves that it produces a new cancer instability. NUMT formation is the incorporation of mitochondrial remnants into nuclear chromosomes which is most common as single insertions and reiterated or structurally variant forms and enables one to promote the concept of the genomic chaos of cancer[13]. Second, lots of benefit found in numerous tumours as there’s association identified between YME1 L1 mutation and increased numtogenesis due to their ability to capitalize on losing this mitochondrial quality-control gene. NUMTs provide selective benefit of the cell through the disruption of tumour-suppressive loci or the encoding of proliferative variants of crucial genes. When it comes to driving or not driving the disease, the NUMTs record a genome-scale disaster that the tumor successfully survived through - tolerance that a normal cell does not have.

Translational perspectives

The potential for translating NUMTs into the clinic in CRC is only now becoming evident. One intriguing idea is their use as tumour markers. Each new NUMT insertion represents a unique sequence junction (mitochondrial-to-nuclear DNA breakpoint) that is not present in normal cells. In theory, highly sensitive PCR across that junction could detect a single cancer cell in the background of normal cells. The University of Alabama at Birmingham group that discovered numtogenesis in CRC developed a single-molecule fibre-FISH technique to rapidly map NUMTs[26]. They suggested that this could be used to distinguish and monitor cancer stages and progression[26]. For example, a blood test might capture rare cells or DNA fragments harbouring a tumour-specific NUMT junction, signaling the presence of CRC. However, practical implementation would require knowing the sequence of the NUMT in each patient’s tumour. This implies having access to tumour tissue and broad sequencing, limiting its workability as an initial screening tool. Another therapeutic angle is that if loss of YME1 L1 contributes to cancer, restoring its function or compensating for its absence could slow tumour growth. YME1 L1, a mitochondrial protease of the AAA+ family, is not a traditional drug target, but its downstream effects - such as the accumulation of certain proteins or mtDNA stress signals - could be targeted[14,26]. More broadly, the concept that mitochondria-to-nucleus DNA transfer drives genomic instability has positioned the mitochondrial genome as an unexpected player in cancer genomics. This emphasises that preserving mitochondrial integrity might be crucial for genomic stability. It serves as a reminder that treatments which exacerbate mitochondrial damage (such as certain oxidative drugs) could inadvertently foster numtogenesis and new mutations[14,26]. Conversely, agents that stabilise the mitochondria or promote mitophagy (clearance of defective mitochondria) may reduce the chance of mtDNA escape.

MITOCHONDRIA-ENCODED MICROPEPTIDES AND CANCER SIGNALLING

Discovery of MDPs

The contribution of the mitochondrion to cellular physiology extends beyond bioenergetics and metabolism; it also actively communicates with the rest of the cell and even distal tissues through signaling molecules. In the past two decades, a series of small peptides encoded by the mitochondrial genome have been identified, collectively known as MDPs[17]. These include HN, a 24-amino acid peptide encoded by the MT-RNR2 gene (16S rRNA), MOTS-c, a 16-amino acid peptide encoded by an ORF in the MT-RNR1 gene (12S rRNA), and several smaller HN-like peptides encoded in the 16S rRNA region[15,18,27]. MDPs were initially discovered in contexts unrelated to cancer. For example, in 2001, HN was initially reported as a peptide that protected neurons from cytotoxicity associated with Alzheimer’s disease[28]. This was the first hint that the tiny mitochondrial genome harboured hidden genes with significant effects. These peptides can be synthesised within the mitochondria or in the cytosol, with some having dual targeting signals. Many are secreted to produce autocrine or endocrine effects[18,29]. Functionally, MDPs express cytoprotective properties and act as metabolic regulators. They have also been shown to modulate apoptosis, oxidative stress responses, insulin sensitivity, and inflammatory pathways[15,30,31]. For instance, under metabolic stress, nuclear translocation of MOTS-c occurs and upregulates tension response genes, improving glucose metabolism and insulin action[15,31,32]. These adaptive effects make MDPs attractive age-related disease therapeutic targets and analogues such as HN(G) (a potent variant of HN) have been investigated in models of neurodegeneration and metabolic syndrome[33-35].

HN: A double-edged sword

From the perspective of cancer, the actions of MDPs can be a double-edged sword. HN, in particular, is emerging as a factor that tumours may co-opt for survival advantages. The normal role of HN is anti-apoptotic, binding to receptors that trigger pro-survival signaling cascades (such as signal transducer and activator of transcription 3, mitogen-activated protein kinase, and phosphatidylinositol 3-kinase/protein kinase B pathways), and inhibiting Bax-mediated mitochondrial apoptosis[18,36]. HN can signal through a cell-surface receptor complex that involves the gp130 subunit (encoded by IL6ST), which is shared with interleukin-6 (IL-6) family cytokines and the ciliary neurotrophic factor receptor α chain, among possibly other components[18]. In essence, HN can mimic a growth factor or cytokine-like signal. While this is beneficial for protecting normal cells from stress, it can unfortunately also protect cancer cells from therapeutic stress.

Recent studies have demonstrated the pro-tumour effects of HN in certain cancers. In TNBC, the upregulation or administration of HN has been shown to promote tumour progression. Moreno Ayala et al[18] demonstrated that treating TNBC cells with exogenous HN made them more resistant to the chemotherapeutic drug (doxorubicin) treatment, as HN exposure partially rescued cell proliferation and reduced apoptosis in the presence of chemotherapy. Conversely, the knockdown of HN in these cells sensitises them to the drug. In mice bearing TNBC tumours, HN injections accelerated tumour growth and abrogated the antitumour efficacy of doxorubicin[18]. Notably, HN-treated tumours had fewer apoptotic cells and exhibited more lung metastases compared to control[18]. These findings suggest that HN provides a protective niche for cancer cells, allowing them to evade chemotherapy-induced cell death and even enhance their metastatic potential, presumably by inhibiting anoikis or other cell death pathways that normally curtail metastasis. The TNBC study highlighted that a cytoprotective peptide in normal tissue becomes a chemoresistance factor in tumours. Similarly, GBMs, an aggressive brain cancer, appears to harness humans in its microenvironment. Cheng et al[29] found that HN is abundantly expressed in GBMs tumours, particularly by tumour-associated myeloid cells (microglia or macrophages) in the tumour milieu. Glioma-associated microglia or macrophages in GBMs showed high levels of the 16S rRNA transcript and HN peptide. Notably, HN was more abundant in GBMs tissues than in normal brain[29]. In an organotypic culture model of brain slice repopulated with human microglia, co-culture of GBM cells and microglia strengthen human expression, indicating the tumour cells trigger microglia to synthesise HN[29].

Functionally, secreted HN from myeloid cells was highly active against GBM cells by making the tumour cells chemotherapy-resistant and assisted in strengthening the blood-tumour barrier which is a drug delivery prevention[29,37-39]. Overall effect was bidirectional protection of the tumour by chemotherapy[29,37]. Mechanistically, the research on GBM revealed gp130 as the mediator of the effect of HN. HN-sensitive GBM cells showed strong gp130 (IL6ST) expression, and pharmacological gp130 inhibition completely abolished HN-induced growth stimulation and therapy resistance[29]. Two different gp130 antagonists - SC144 and bazedoxifene suppressed the pro-tumour action of HN in vitro[29]. HN signaling in GBM cells was shown to activate the ERK pathway downstream of gp130 by triggering a DNA damage response program to increase the viability of cells following chemotherapy[29,40-42]. Furthermore, HN itself feedback that gp130 activity increased intratumoural expression of HN in GBMs, a positive feedback loop whereby HN-activated microglia stimulate further human expression by gp130 signaling[29,41,42]. These findings position the HN-gp130 pathway as a viable therapeutic target and inhibition of this pathway can withhold tumours of a crucial survival signal. It is a striking example of how a product of a mitochondrial gene ends up hijacking an established cancer-linked cytokine pathway (gp130/IL-6) for tumour progression.

Although specific studies in CRC are sparse, it is plausible that similar phenomena occur in the CRC microenvironment. CRC tumours often have abundant infiltrating macrophages and other stromal cells, which could be sources of MDPs. If HN is expressed in CRC tissue, it may contribute to chemotherapy resistance (for instance, to 5-fluorouracil- or oxaliplatin-based regimens) and may also protect cancer stem cells. However, this issue remains unresolved.

MOTS-c as a tumour suppressor peptide

The flip side of the coin is MOTS-c, which appears to have more tumoursuppressive properties than HN’s tumour-promoting properties. MOTS-c is primarily known for its metabolic effects, which can translocate to the nucleus during stress and help regulate genes related to metabolism and oxidative stress, often activating AMPK and antagonising the mammalian target of rapamycin (mTOR) pathway[15,30-34]. Cancer cells frequently rely on mTOR signaling for growth and survival, and the downregulation of mTOR can inhibit tumour proliferation. New findings highlights the role of MOTS-c that it perform exactly in certain cancers. A study by Yin et al[43] demonstrated that amount of MOTS-c shows lower in the serum and tumour tissues of patients with ovarian cancer than in healthy controls, and low MOTS-c is associated with poorer prognosis. This implies that ovarian tumours might somehow downregulate this peptide to escape its growth inhibitory effect. When researchers treated ovarian cancer cell lines and tumour-bearing mice with MOTS-c, they discovered robust anti-cancer activity: MOTS-c suppressed cancer cell growth, and in mouse xenografts, reduced tumour volumes and weights than the control group[43]. These findings show that MOTS-c can effectively control tumour growth, lost or blunted in virulent disease.

Some studies linked the activity of MOTS-c to an effect on leucyl-tRNA synthetase 1 (LARS1) protein homeostasis[43,44]. LARS1 is a sensor of intracellular leucine levels and a positive regulator of mTOR complex 1 (mTORC1) under nutrient-rich conditions, LARS1 helps activate mTORC1 and promotes cell growth. The study found that MOTS-c interfered with the deubiquitination of LARS1 by the enzyme USP7, leading to LARS1’s degradation[43,44]. As LARS1 levels decrease, mTORC1 activity is attenuated, thereby inhibiting the growth and proliferation of cancer cells. MOTS-c pushes the cells into a more metabolic stress-like state, which is unfavourable for rapid tumour expansion. This novel insight connects a mitochondrial peptide to the central mTOR pathway, a key target in many cancers, and opens the door to the potential use of MOTSc or its analogues as therapeutic agents. Notably, these findings align with the role of MOTS-c in metabolic homeostasis; specifically, by dampening anabolic processes via mTOR inhibition, MOTS-c produces an anti-proliferative outcome in tumour cells.

While direct studies of MOTS-c in CRC have not yet been published, to our knowledge, parallels in metabolic reprogramming between ovarian cancer and CRC suggest that it could have similar effects. CRC cells often activate mTOR signaling (for example, through phosphatidylinositol 3-kinase/protein kinase B pathway mutations or loss of phosphatase and tensin homolog) and thrive on anabolic metabolism. Introducing MOTS-c might counteract this by reinforcing mTOR-driven processes. There is also evidence that exercise, which elevates MOTS-c in circulation, is associated with better CRC outcomes, although it involves many factors[15,32,35]. It is plausible that part of the benefit of a healthy lifestyle or exercise on CRC risk could be mediated by MDPs like MOTS-c, which improve the metabolic resilience of cells and potentially reduce cells tipping into malignancy. Although speculative, this should be investigated in the future.

CLINICAL AND TRANSLATIONAL IMPLICATIONS

Converging evidence on the diverse roles of mtDNA in CRC has several important implications for clinical practice and future research.

Enhanced early detection and monitoring

Liquid biopsy approaches for CRC could be bolstered by incorporating mtDNA-based biomarkers with existing ctDNA assays. For example, a blood test that measures the mtDNA fraction or absolute mtDNA copy number (possibly via digital PCR for speed and cost-effectiveness) could flag individuals with an elevated cancer risk. In population studies, adding mtDNA quantification to panels of protein markers or ctDNA mutations might improve the sensitivity for detecting early-stage CRC[12]. Likewise, for patients with known CRC, tracking mtDNA levels over time could help monitor treatment responses or detect minimal residual disease. If a patient’s plasma mtDNA level drops during therapy and then starts rising, it might indicate tumour recurrence earlier than conventional imaging. However, standardisation will be key: The field will need to determine threshold values or patterns that constitute a positive mtDNA signal, accounting for age and other factors that influence baseline cf-mtDNA (e.g., inflammatory conditions can also release mtDNA). Large prospective trials are warranted to evaluate mtDNA as an adjunct in CRC screening (especially in synergy with stool DNA tests or circulating tumour cells) and as a surveillance tool post-surgery.

New insight into tumour biology and potential targets

The discovery of numtogenesis in CRC has revealed a previously unappreciated source of genetic instability. This suggests that therapies aimed at protecting mitochondrial integrity may indirectly stabilise the nuclear genome. For instance, some researchers have proposed the use of mitochondria-targeted antioxidants or mitophagy inducers in cancer treatment to reduce oxidative stress and remove dysfunctional mitochondria. One could hypothesise that such interventions might also reduce the incidence of mtDNA fragments escaping into the nucleus, thereby lowering the accumulation of deleterious NUMTs. Although targeting the process of numtogenesis itself is not straightforward (preventing a cell from incorporating DNA during double-strand repair is essentially difficult), understanding it helps to complete the picture of how CRC genomes evolve. The identification of YME1 L1 as a suppressor of mtDNA migration is especially intriguing; tumours with YME1 L1 mutations might represent a subset with unique vulnerabilities[14,26]. YME1 L1 has other cellular roles (e.g., degrading pro-apoptotic factors in mitochondria); therefore, YME1 L1-mutant tumours might be more dependent on alternate pathways for mitochondrial protein quality control, potentially revealing a synthetic lethal interaction. In addition, the presence of tumour-specific NUMT junctions, as previously mentioned, could serve as an ultra-specific biomarker for tumour DNA. PCR assays for a patient’s unique NUMT can be designed if the whole-genome sequencing of the tumour is available. This might detect a single tumour cell among billions of normal cells, offering a personalised monitoring strategy.

Therapeutic modulation of mitochondrial peptide pathways

Perhaps, the most immediate actionable area is the modulation of MDPs and their signaling pathways. The HN-gp130 axis is one example of drugs that inhibit gp130 (some of which are in development for inflammatory diseases), which can be repurposed to help overcome chemoresistance in CRC or other cancers that show high HN activity[29]. It is important to identify CRC patient subsets with elevated HN levels in tumours or circulation, potentially those with chemotherapy-refractory disease, who might benefit from such an approach. Conversely, augmenting MOTS-c levels could be beneficial. There is tantalising evidence that exercise interventions increase MOTS-c levels and are correlated with reduced cancer recurrence in survivors, although direct causation has not been proven[45]. Nonetheless, one could envision a therapeutic MOTSc analogue (modified for a longer half-life) being tested as an adjuvant therapy to slow tumour growth or as a metabolic therapy to counter cachexia and metabolic derangements in late-stage patients.

Comprehensive biomarker panels

Ultimately, the integrative approach is likely to be the most powerful. For instance, a mitochondrial signature in CRC could include mtDNA copy number in the tumour (low mtDNA copy in tumour tissue has been linked to certain oncogenic mutations), mtDNA mutation burden (some studies correlate high mtDNA mutation load with worse outcomes), expression of mitochondrial biogenesis regulators (such as peroxisome proliferator-activated receptor gamma coactivator 1 alpha), and the levels of key MDPs such as HN/MOTS-c[22,23,46]. Together with NUMT occurrence, these factors could form a composite score reflecting mitochondrial involvement in a patient’s cancer. Some of these components have prognostic associations individually. For instance, a higher mtDNA mutation burden might reflect defects in mtDNA repair or higher oxidative stress in the tumour, possibly correlating with aggressive behaviour. If validated, clinicians could use such mitochondrial biomarkers to stratify patients for certain treatments. For example, perhaps tumours with depressed mtDNA content respond differently to hypoxia-targeting drugs or to metabolic inhibitors.

Mitochondrial genomics in precision oncology

The era of precision oncology has largely focused on nuclear gene mutations (KRAS, BRAF, mismatch repair genes in CRC, and others). However, alterations in mtDNA may also lead to therapy. An interesting idea is theranostics based on mtDNA mutations, which has been explored using mitochondrial-targeted drugs or even mitochondria-directed gene therapy that would specifically affect cells with certain mtDNA mutations that can cause metabolic vulnerabilities. While not in routine use, there are cases where mtDNA mutations (such as in the MT-ND4 gene) make cancer cells dependent on certain substrates, hinting at niche therapeutic windows[12]. Drugs that further disrupt mitochondrial activity may prove especially potent in cancers whose mtDNA deletions have already crippled oxidative phosphorylation and the additional mitochondrial insults can convert that pre-existing weakness into a lethal strike.

Orthogonal mitochondrial signals for CRC stratification and detection

During ATP generation, anabolic flow, and intracellular signaling converge, stress-sensitive mitochondria reorganise cellular structure towards environmental demands[46]. Across cohorts, a consistent association emerges where mitochondrial genome copy number correlates with canonical subtypes such as lower in BRAF-mutant tumors (P = 0.027, 0.006) and in microsatellite instability (MSI) cases (P = 0.033, < 0.001), but KRAS mutation elevation (P = 0.004)[11], implying mitochondrial reprogramming that is dependent on subtype context. Up to 4.2-fold NUMT elevation over matched normals in colorectal adenocarcinoma suggests that risk can be subclassified within current subtypes (e.g., MSI-high) based on evidence[14]. In liquid biopsy, adding mDNA information to a copy-number increases sensitivity when mito-signals are combined with standard cDNA 24. KRAS/BRAF readouts by lifting the area under the curve from 0.73 to 0.81 for detection and monitoring[12]. Together, these mitochondrial readouts complement MSI/KRAS/BRAF by capturing orthogonal metabolic/genome-stability states and enabling within-stratum risk refinement and better cDNA-based detection.

Potential side effects and considerations

If the target is mitochondrial pathways, it should be done so carefully because normal cells also depend on mitochondria. For example, systemic gp130 inhibition might dampen HN’s protective effects on normal tissues and cause adverse events. Given that gp130 is also used by IL-6, IL-11, etc., the immune and hematopoietic side effects should be monitored. Peptide therapies such as MOTS-c might have off-target effects on metabolism that require monitoring (e.g., hypoglycaemia if it greatly increases insulin sensitivity). However, the fact that these peptides are natural and usually well-tolerated in contexts such as exercise or ageing is encouraging.

Research directions

This review also highlights key research directions. In CRC specifically: More studies are needed to map out how often numtogenesis occurs in different stages, whether it is an early event during the adenoma-carcinoma transition or a later event in carcinoma progression. Do metastases have more NUMTs than primaries? The role of the immune system in clearing or responding to cell-free mtDNA can act as a danger signal via toll-like receptor 9 or cGAS-STING pathways. Does the chronic release of mtDNA from tumours trigger inflammation that promotes cancer (such as colitis-associated cancer scenarios), or does it help the immune system recognise the tumour? One could measure cGAS-STING activation in mtDNA-rich tumours. Interplay between mitochondrial genetics and the microbiome: The gut microbiome in CRC is a large area, so interestingly, bacterial DNA can also integrate into the human genome (albeit rarely). Could there be an interaction between mitochondrial dysfunction and microbial dysbiosis that promotes CRC? In addition, can lifestyle interventions (diet and exercise) that impact mitochondria be formally tested as adjuvant therapies for CRC? This answer could directly benefit patients in terms of recommendations beyond surgery or chemotherapy.

CONCLUSION

Overall, mitochondria-centric research illuminates previously dark corners of CRC biology. This encourages a more holistic view of tumour cells as systems where nuclear and mitochondrial genomes co-evolve during cancer development. As this field progresses, information obtained from the CRC will translate to many other cancers, given the fundamental nature of mitochondrial function in cell biology. The hope is that unravelling these mitochondrial mysteries will fuel the development of innovative interventions, such as combining mtDNA biomarker-based early detection with metabolic therapy to combat cancer through a multi-angle attack. Due to continuous initiatives to mitigate the global burden of CRC, mtDNA has offered new hope for earlier diagnosis and more effective treatment of this formidable disease.