Skip to main content
NCBI home page
Stem Cell Reports logoLink to Stem Cell Reports
. 2025 Sep 18;20(10):102643. doi: 10.1016/j.stemcr.2025.102643

Drivers and implications of lineage-specific cancer-related mutations in human pluripotent and adult stem cells

Jonathan Jung 1, Nissim Benvenisty 1,2,∗
PMCID: PMC12790749  PMID: 40972588

Summary

Human pluripotent stem cells (PSCs) are known to harbor mutations in tumor-associated genes, and here we aim to examine the status of adult stem cells (ASCs). We thus identify cancer-related mutations in 18% of about 600 mesenchymal stem cell samples, and in 41% of about 200 neural stem cell (NSC) samples. We show a lineage-specific profile of cancer-related genes, demonstrating that TP53 is a central mutated gene in human PSCs but not in mesenchymal or NSCs. We suggest that the lineage-specificity of tumor-associated genes correlates with their expression levels and with tumor-specific mutations in patients. We also show the consequences of mutations in oncogenes and tumor suppressor genes on the transcriptome of each specific stem cell lineage. We therefore propose a categorization of these mutated samples for further appreciation of their severity and emphasize the importance of genetic screening in pluripotent and ASC lines.

Keywords: cancer-related mutations, adult stem cells, mesenchymal stem cells, neural stem cells, embryonic stem cells, genomic integrity, point mutations, lineage specificity

Graphical abstract

graphic file with name fx1.jpg

Open in a new tab

Highlights

  • •

    Analysis for cancer mutations in ∼1,400 samples of embryonic and adult stem cells

  • •

    Cancer-related mutations are found in mesenchymal (18%) and neural (41%) stem cells

  • •

    Lineage-specific mutations correlate with expression levels and patients’ tumors

  • •

    Mutations in oncogenes and tumor suppressor genes show distinct consequences

Jung and Benvenisty identify cancer-related mutations in about 18% and 41% of mesenchymal and neural adult stem cells, respectively. They demonstrate a lineage-specific profile of cancer genes correlated with gene expression levels and tumor-specific mutations. They show the consequences of mutations in oncogenes and tumor suppressor genes, highlighting the importance of genetic screening in pluripotent and adult stem cell lines.

Introduction

Human pluripotent stem cells (PSCs) are commonly used in developmental research, disease modeling, and regenerative medicine due to their ability to differentiate into all cell types and seemingly endless proliferative capabilities (Ben-David et al., 2012; Avior et al., 2016; Kirkeby et al., 2025). Likewise, tissue-derived adult stem cells (ASCs) play a major role in biomedical research, and some of them can readily propagate in culture due to their self-renewal capacity. Specifically, mesenchymal stem cells (MSCs) isolated from a broad range of sources (bone marrow, adipose tissue, cord blood, and more) can differentiate into muscle, bone, cartilage, and adipocytes, and are used in multiple clinical trials (Kimbrel and Lanza, 2020; Rauch and Mandrup, 2021). Neural stem cells (NSCs), which are mainly isolated from fetal central nervous system and differentiate into neuronal and glial cells, are central in understanding brain development and disease, and are also used in clinical trials (De Filippis and Binda, 2012; Fernandez-Muñoz et al., 2021). The self-renewal capacity of PSCs and ASCs implies that they are constantly engaged in DNA replication, increasing their susceptibility to DNA damage and genomic defect accumulation (Branzei and Foiani, 2008). The genomic defects can range from large to small chromosomal aberrations. Indeed, large chromosomal aberrations were documented in human PSCs (Halliwell et al., 2020; Andrews et al., 2022) and adult MSCs and NSCs (Ben-David et al., 2011); however, the type of these aberrations is lineage-specific (Ben-David et al., 2011).

In addition to chromosomal aberrations, human PSCs have been shown to harbor cancer-related mutations. These were first documented in the TP53 gene in multiple lines of either human embryonic stem cells (ESCs) or induced pluripotent stem cells (iPSCs) (Merkle et al., 2017). Later, additional cancer-related genes were identified to be mutated in hPSCs by analyzing RNA sequencing (RNA-seq) (“RNA2CM” tool) (Avior et al., 2021; Lezmi and Benvenisty, 2021) or DNA-seq (genome sequencing) (Rouhani et al., 2022) data. Recently, a large-scale analysis of over 140 human PSC lines revealed cancer-related mutations in about 20% of the samples of pluripotent cells and their differentiated derivatives, and suggested the effects of such mutations on cellular differentiation (Lezmi et al., 2024). Point mutations have been studied in a few lines of MSCs (De Becker et al., 2022; Kim et al., 2017), but analysis of the extent of DNA mutations in ASCs is still lacking. Nevertheless, point mutations exist in vivo (Abyzov et al., 2017; Vijg and Dong, 2020; Fasching et al., 2021), and are often obscured in the body through mechanisms such as mosaicism and immune surveillance (Vijg and Dong, 2020; Corthay, 2014). However, the in vitro culture of stem cells alters this situation, allowing specific mutations to persist and expand. Factors such as prolonged culturing periods and medium conditions may introduce new mutagenic pressures (Kuijk et al., 2020). Here, performing a large-scale analysis, we identify cancer-related mutations in ASCs, either MSCs or NSCs.

We demonstrate lineage-specific patterning and suggest that these genes’ expression levels and involvement in human tumors may underline the lineage specificity. We also show the consequences of such mutations, activating oncogenes and selectively repressing tumor suppressor genes, on different pathways in PSCs and ASCs, in some cases, evident of potential tumorigenicity phenotypes.

Results

Cancer-related mutations in human adult stem cells

To inspect cancer-related mutations in different stem cell types, we analyzed over 1,400 RNA-seq samples, of which 606 are from human ESCs, 603 from MSCs, and 198 from NSCs (Figure 1A; Table S1). The cell identity was validated by principal-component analysis (PCA), uniform manifold approximation and projection (UMAP), and t-distributed stochastic neighbor embedding (t-SNE) presentations. Analysis of gene expression profiles with and without batch effect correction have shown defined clusters of the three cell types (Figure S1A). Analysis of cell-specific markers further confirmed cell identity, and random inclusion of male and female samples in all lineages (Figure S1B). Additional gene set enrichment analysis (GSEA) for each lineage compared to the others revealed that the top-ranking pathways (of the 100 enriched genes) were highly suggestive of the correct annotation. The top-ranking pathway in GSEA (https://www.gsea-msigdb.org/gsea/index.jsp) for ESCs was “Benporath ES 1”—related to gene sets associated with ESC identity (p value = 2.79e−28, k = 27), for NSCs was “Manno midbrain neurotypes HRGL2B”—related to neuronal progenitor (p value = 1.27e−43, k = 37), and for MSCs was “Muraro pancreas mesenchymal stroma cell”—related to mesenchymal stroma cells (p value = 1.21e−60, k = 51).

Figure 1.

Figure 1

Open in a new tab

Cancer-related mutations in embryonic and adult stem cells

(A) Fraction of the mutated samples by the number of mutations per sample.

(B) Fraction of samples categorized by source (cell line or tissue of origin).

(C) Fraction of studies with at least one mutated sample.

(D) STRING analysis of mutated genes (confidence >70%).

(A–C) Striped areas designate mutated samples (broader stripes in A indicate more than one mutation in a sample). (A–D) Graphs were included for each of the three lineages with a designated color palette in the following order: pluripotent, green; mesenchymal, red; and neural, blue (top to bottom).

We next searched for mutations in over 500 tier 1 cancer genes, where the base substitution is the same as in at least 10 patients in the Catalogue of Somatic Mutations in Cancer (COSMIC) database (https://cancer.sanger.ac.uk/cosmic) (Tate et al., 2019). We thus identified a similar percentage of samples containing at least one cancer-related mutation in MSCs and ESCs (18% and 20%, respectively, Figure 1A), and a higher mutation rate in NSC samples (41%, Figure 1A). Breakdown of the data according to sources of derivation revealed that mutations appeared in MSCs regardless of whether they originated from bone marrow, adipose tissue, cord blood, or other sources. Similarly, mutations were detected in NSCs whether derived from fetal forebrain, hindbrain, spinal cord, or other sources (Figure 1B). In addition, at least one mutated sample appears in 28% of ESC studies (38/134), 51% of MSC studies (37/73), and in 63% of NSC studies (12/19; Figure 1C). The relatively high percentage of ASC studies with mutations may be attributed to the fact that many of them include multiple cell lines from different derivations in the same study.

Lineage-specific mutation profiles and in vitro acquisition

Next, we analyzed the specific cancer-related genes emerging in each lineage (Table S2). Search Tool for the Retrieval of Interacting Genes/Proteins (STRING) analysis (Szklarczyk et al., 2023) of these genes demonstrated most of them to be interconnected, with TP53 serving as the central connector hub in ESCs (Figure 1D). Clustering mutation events by gene revealed distinct patterns across the different lineages (Figure 2A). Notably, TP53 is the predominant gene in the ESC mutation distribution (67% of mutations), consistent with previous findings (Lezmi et al., 2024). In contrast, MSCs and NSCs exhibit a wider variety of mutations, with a more evenly distributed contribution from multiple genes (Figure 2A). TP53 was only the fifth mutated gene in MSCs and NSCs, appearing in seven of the mutated samples in MSCs and in six of the mutated samples in NSCs. These patterns suggest the existence of lineage specificity in cancer-related mutations. Since many of the lines of MSCs and NSCs are uniquely analyzed in single studies, assessing which mutations are acquired in vitro and which originated from the donor tissue is difficult. There are, however, some cases where the same cell line was used in at least two studies, and a mutation appears only in one of the two. In these cases, we could suggest that the mutation was acquired in culture, as seen not only for ESCs but also for MSCs and NSCs (Figure 2B). ASCs, therefore, can acquire cancer-related mutations in vitro but may also be affected by mosaicism in the donor’s original genotype.

Figure 2.

Figure 2

Open in a new tab

Lineage-specific cancer-related mutations and their acquisition in culture

(A) Bars represent the distribution of mutated genes by lineage, counting the number of mutated samples. The dendrogram above the bars represents the similarity between the distributions.

(B) Schematic representing proof of mutation acquisition in vitro in each lineage. Cell lines are annotated on the left, mutated genes are shown in red italics, and the studies’ accession codes appear on the right.

Correlation of mutation rate with gene expression levels and in vivo tumors

We next examined the expression levels of the specific genes found to be mutated in our analysis. Interestingly, we found the top contributing genes to mutational landscape in each lineage to be highly expressed in that specific lineage when compared to the other two cell types (Figures 3A and S2A). TP53 and FGFR3 are the top contributors in ESCs and NSCs, respectively, and NOTCH2 and MET are top contributors in MSCs (Figure 3A). When compared between lineages, we observed a positive correlation between the genes’ contribution to the mutated landscape and the genes’ expression level mainly in the comparison of the MSCs to either ESCs or NSCs (Figure 3B). We speculate that this correlation suggests that highly expressed genes have also increased functional relevance. There is a notably weaker correlation between ESCs and NSCs, which may be due to their transcriptional similarity, as NSCs arise early during development. This close developmental relationship may limit transcriptional divergence and, in turn, reduce the dynamic range needed to detect a stronger correlation. An example can be observed in the Notch pathway, where NOTCH1 and NOTCH2 are homologous genes (Fan et al., 2004). In MSCs, NOTCH1 is expressed at lower levels than in ESCs and NSCs, while NOTCH2 is found to be highly expressed specifically in MSCs (Figures 3A and S2). Accordingly, NOTCH2 is the top mutated gene in MSCs, while NOTCH1 mutations appear in ESCs and NSCs but not in MSCs. Importantly, higher expression by itself did not underlie increased mutation detection (Figure S2B).

Figure 3.

Figure 3

Open in a new tab

Correlation of cancer-related mutation rates with gene expression levels and in vivo tumors

(A) Comparison of expression by log2(CPM [count per million]) of top mutational landscape contributor genes in the three lineages (top inset), and their contribution to the mutational landscape of each lineage (bottom inset). All gene expression differences per lineage are significant by the Kruskal-Wallis U-test and Post-hoc Dunn’s test.

(B) Correlation between expression and mutational contribution. Results are shown as a comparison between ESCs and MSCs (left inset), NSCs and MSCs (middle inset), and ESCs and NSCs (right inset). Green area, pluripotent tendency; red area, mesenchymal tendency; blue area, neural tendency. FC, fold change.

In (A and B), the TP53 gene is excluded from all plots except the TP53 plots themselves.

(C) Bars show the genes ranked by proportion of tumor samples harboring a mutation in that gene, in teratomas and teratocarcinomas (left inset), unclassified sarcoma (center inset), and neuroblastomas (right inset). The data in this figure are based on the COSMIC database. The ranking of TP53 is marked in dark red.

Though this “expression-mutation tendency” correlation exists, it is not very strong and is clearly not the only contributor to the frequency of mutation events, suggesting that additional driving forces should be considered. Following this notion, we revisited the COSMIC database (Tate et al., 2019) to find if the TP53 gene had any lineage-specific correlation with in vivo tumors. Interestingly, when assessing tumors derived from the various lineages, we saw that in teratomas and teratocarcinomas, which tend to be derived from pluripotent cells, the most abundant mutation is in the TP53 gene (Figure 3C). In contrast, in tumors related to the other lineages, TP53 ranks much lower (Figure 3C). Notably, the disproportionate contribution of TP53 to the mutational landscape of ESCs favors analysis of the source of lineage specificity, while MSCs and NSCs do not exhibit such a dominant pattern, and therefore, a similar correlation analysis is less relevant. Following this observation, we suggest that different cell types harbor distinct cancer-related mutations, and that factors beyond the culturing period contribute to the predominance of TP53 mutations in ESCs. These factors could affect other genes as well.

Consequences of mutations in oncogenes and tumor suppressor genes

The mutated genes we found in our analysis are classified by the COSMIC database (Tate et al., 2019) as either oncogenes, tumor suppressor genes, or both (Figure 4A top). When focusing on each lineage separately, we still see that all classifications contribute to each lineage (Figure 4A middle). In oncogenes, a single aberrant allele can underlie potential transformation, shown by an allelic frequency of 0.5, presenting a homogeneous population of heterozygous mutations (Figure 4A bottom). However, in tumor suppressor genes, there may be a need to induce mutations in both alleles, unless a dominant negative feature is involved (Willis et al., 2004). As such, tumor suppressor genes (and those classified as both tumor suppressor genes and oncogenes) can also be found in higher allelic frequencies, up to 1.0, suggesting a loss-of-heterozygosity (LOH) event in some samples (Figure 4A bottom). By applying e-SNP-karyotyping (Weissbein et al., 2016), we confirmed monoallelic expression of chromosome 17p in ESCs, where TP53 is located, further supporting our assumption of LOH (Figure S3). Importantly, monoallelic expression can occur from other reasons, such as RNA stability or epigenetic silencing of the normal allele; however, monoallelic expression of a cluster of genes in the same chromosomal region, that is not parentally imprinted, most likely results from LOH.

Figure 4.

Figure 4

Open in a new tab

Involvement of oncogenes and tumor suppressor genes in different stem cell lineages, and their transcriptomic consequences

(A) Top: a Venn diagram of all cancer-related genes found in this study, classified by their role in cancer. Middle: bars represent the distribution of all mutated genes by lineage and their role in cancer. Bottom: distribution of all allelic frequencies of each mutation based on their role in cancer.

(B) Lollipops represent top downregulated gene-sets by normalized enrichment score, −log10(p value) as the lollipop size. Color representation: purple, all mutated samples; blue, samples mutated in oncogenes; red and yellow, samples mutated in tumor suppressor genes or in genes classified as both tumor suppressor genes and oncogenes (classification according to inset A), all compared to WT samples.

(C) Two-dimensional density plot presenting a PCA based on the expression of genes contributing to the top two downregulated pathways in each lineage (pathways from B). The largest areas for each group represent 40% of the density for each category.

Analyses for (B and C) were performed for each lineage separately: top, ESCs; middle, MSCs; and bottom, NSCs.

To analyze the consequences of cancer-related mutations in human ESCs, MSCs, and NSCs, we performed a differential expression analysis to unveil possible aberrant phenotypes common to all mutated samples. By initial PCA visualization, we did not identify clear expression profiles of the mutated samples compared to all other samples (even without batch effect correction). This is likely due to the lineage-specific expression patterns, shadowing the more subtle molecular effects promoted by each mutated gene (Figure S4A). Hence, the samples were first segregated by lineage to nullify the clear distinct expression profiles between them. We then isolated the differentially expressed genes in each group by comparing the mutated to unmutated samples, followed by a GSEA (Figure 4B). The analysis identified pathways that were significantly depleted in the mutated samples. Each lineage shows distinct pathways that characterize the mutated vs. unmutated samples. For example, in ESC samples’ analysis, the top depleted pathways are “response to transforming growth factor β (TGF-β) signaling” and “signal transduction by p53.” The latter can be explained by the abundance of the TP53 mutated samples, conferring a downregulation of p53 signaling. Response to TGF-β may have appeared due to its tumor suppressive nature (Massagué, 2008). In MSCs, the most depleted process is aerobic respiration, which may be due to the important role of mitochondria in these cells (Ren et al., 2020; Zhang et al., 2022). The second depleted pathway was “response to TGF-β signaling.” However, in MSCs, TGF-β may potentially play a bigger role in differentiation mechanisms, as this annotation appears mostly in the MSC samples with mutated oncogenes, which impede developmental and differentiation processes, expanded further in the following text. NSCs’ depleted pathways consist of “cell adhesion” and “neuropeptide signaling.” Both may suggest a strayed phenotype from the typical characteristics of the cells (Morante-Redolat and Porlan, 2019; Zaben and Gray, 2013).

Subdivision of genes based on their role in cancer was also performed, separating oncogenes and tumor suppressor genes. Interestingly, the samples mutated in oncogenes showed general differentiation and developmental capacity depletion (Figure 4B). The pathways depleted were tube and epithelium development in ESCs, differentiation through TGF-β and circulatory system development in MSCs, and neural precursor proliferation and olfactory lobe development in NSCs. The tumor suppressor gene mutated samples presented depletion of general cellular processes (Figure 4B). Notably, these processes varied between the lineages as well. By identifying the genes contributing to the top depleted pathways in each lineage, we refined our data for the ability to identify intricate nuances between the different samples. By density visualization of samples of PCA of the genes contributing to the lineage-depleted pathways, we managed to observe clear segregation of unmutated, oncogene, tumor suppressor gene, and TP53 mutated samples (Figure 4C). Once more, distinct cancer-related categories appear differently in the various stem cell types, providing lineage-specific patterns. We can see that in the neural lineage, both oncogene and tumor suppressor gene mutated samples diverge from the unmutated samples to a similar extent (Figure 4C, bottom). In the pluripotent state, we observe divergence of all mutated samples, although in different trends (Figure 4C, top). However, in the mesenchymal lineage, only the samples mutated in oncogenes deviate from the unmutated cluster, suggesting that they have a stronger aberrant phenotype than the samples mutated in tumor suppressor genes (Figure 4C, middle). This observation does not include samples with TP53 mutations, which show an even greater aberrant phenotype in the mesenchymal lineage. To further explore the functional consequences of TP53 mutations, we performed GSEA comparing TP53-mutated and wild-type (WT) samples from the same lineage. Proceeding with the analysis for ESCs (as they had the most TP53-mutated samples) showed significant depletion of pathways related to “positive regulation of programmed cell death” and “apoptotic process” (Figure S4B). These pathways were also significantly depleted in samples with TP53 mutations in MSCs and NSCs, suggesting a global functional effect of TP53 mutations in human stem cells (Figure S4B).

Discussion

Here, we performed extensive analyses and identified cancer-related mutations in PSCs, MSCs, and NSCs. Similar to ESCs, in MSCs, about 20% of the samples harbor a cancer-related mutation. This is a relatively high percentage of mutations for an ASC lineage. Moreover, in NSCs, a staggering 40% of the samples had at least one cancer-related mutation (Figure 1A). This high rate may result from the origin of these samples, which was predominantly fetal tissue from abortions. The isolation of NSCs in this developmental period of organogenesis, specifically fetal brain development, is known for high rates of mutagenesis (Bae et al., 2017; Derks and Van Boxtel, 2023). This, coupled with difficulties in isolation and maintenance of NSCs in culture, may have led to the increased selective advantage of cells with cancer-related mutations. Indeed, it is challenging to conclude whether the mutations documented in ASCs come from the donor tissue or are acquired in culture. However, due to selected cases where the same cell line has shown normal and mutated genotypes, we conclude that at least some of the mutations identified in ASCs were acquired during culturing (Figure 2B). The allelic frequencies of the mutations observed fit the mutated gene’s role in cancer, whether oncogenes or tumor suppressor genes, with rare documentation of LOH in tumor suppressor genes (Figure S3).

Notably, we have discovered that the genes involved in the cancer-related mutations are lineage-specific. This is most evident with TP53, which is prevalent in PSCs, but not in ASCs (Figure 2A). TP53 is indeed the most mutated gene in human tumors (Levine, 2020), but is not the most mutated in every type of tumor (Figure 3C). Another intriguing feature in our analysis is the positive correlation between lineage-specific expression of the cancer-related genes and their rate of mutation per the respective lineage (Figure 3B). The differential expression analysis, followed by GSEA, offered a broader view of the pathways that appear depleted in mutated samples within each lineage, with additional patterns emerging when considering their roles in cancer (Figure 4). Both of these observations derived from the samples’ expression may suggest that some genes and pathways are more prominent in specific stem cells as compared to others; thus, mutations in them might be associated with lineage-specific selective pressures. Even depleted general cellular processes were still lineage-specific, possibly reflecting the differing hierarchy of these general pathways in tumorigenic transformation per lineage. Nonetheless, TP53-mutated samples across all lineages showed consistent depletion of pathways related to programmed cell death and apoptosis (Figure S4B), suggesting a conserved functional impact of TP53 mutations across different stem cells.

The gravest safety concern in stem cell therapy is the development of tumors. Indeed, transplantation of cells derived from autologous PSCs into a diabetic patient has already resulted in a malignant immature teratoma (Han et al., 2022). Induction of tumors has also been documented after transplantation of allogeneic NSCs into the central nervous system, which is an immune-privileged site (Amariglio et al., 2009; Berkowitz et al., 2016; Baloh et al., 2022). In one case, a child with ataxia telangiectasia transplanted with NSCs was diagnosed with multifocal donor-derived brain tumors, four years after transplantation (Amariglio et al., 2009). In another case, a 66-year-old patient, after a stroke, was transplanted with neural fetal stem cells and developed glioproliferative lesions in the spinal cord (Berkowitz et al., 2016). In a third case, transplantation of neural progenitor cells secreting glial cell line-derived neurotrophic factor in amyotrophic lateral sclerosis (ALS) patients resulted in the discovery of benign neuromas in the injection sites at postmortem (Baloh et al., 2022). Notably, all reported cases of tumor formation following ASC transplantation are of neural lineage. This may be partially explained by differences in the immune environments into which these cells are introduced. NSCs are usually transplanted into the brain, an immune-privileged organ, while MSCs are typically delivered into more immunocompetent tissues, where allogenic transplanted tumorigenic cells may be more effectively cleared. Taken together, our data suggest that cancer-related mutations are prevalent in both PSCs and ASCs and should be inspected before their use in cell therapy, as their impact may vary depending on lineage and transplantation environment.

Furthermore, we propose a heuristic framework to divide the cancer-related mutations identified in stem cells into different categories based on the expected severity of the type of mutation (Table 1). This classification is intended as a guiding model, based on our interpretation of the literature, primarily from Online Mendelian Inheritance in Man (OMIM) (https://omim.org). Starting from the lowest severity—heterozygous mutations in tumor suppressor genes with no evidence of potential tumorigenicity. No known disease phenotype exists for the individuals carrying a germ-line heterozygote mutation; for example, individuals with a heterozygote mutation in the ERCC5 gene do not have increased susceptibility to develop tumors. In these cases, which harbor 4.6% of the samples, we are not sure they harbor a higher risk for tumorigenicity. The next level of severity is of heterozygous mutations in tumor suppressor genes, where individuals carrying a heterozygote mutation have a higher risk of tumors. An example is the APC gene, where a single mutated allele increases susceptibility to adenomatous polyposis and colorectal cancer. About 54.1% of the samples have such a mutation. In this category, dominant negative mutations are a special class with a higher risk. The following level of severity is of oncogenes, where, by definition, a single mutation is needed to turn the proto-oncogene into a cancerous oncogene. An example is HRAS, where point mutations can result in constitutive pathway activation. About 25.7% of the samples had a mutation in an oncogene. We suggest that even a higher risk exists in cells where biallelic hits occurred in a tumor suppressor gene, such as LOH. An example found in this study is of LOH in TP53. About 7.9% of the samples had such an event. The highest severity involves samples found with mutations in multiple cancer-related genes. For example, a sample of mutated TP53 and NF1 may reflect cooperative effects or advanced selection during culture. About 8.6% of the samples had mutations in multiple genes. While we cannot assign definitive tumorigenic risk levels to these categories without functional validation, this classification provides a useful framework for interpreting mutation severity in stem cells. Our data suggest that cancer-related mutations are prevalent in PSCs and ASCs, and more information should be collected to characterize them and their consequences. Furthermore, it highlights the need to genetically inspect these stem cells before their use in cell therapy, especially in cases where the cells propagated in culture for a long period.

Table 1.

Categorization of mutated samples

Category of mutated sample Example Number of samples Percentage (%)
A - 2/3 mutated cancer-related genes TP53 + NF1 26 8.6
B - Tumor-suppressor genes + LOH TP53 24 7.9
C - Oncogenes HRAS 78 25.7
D - Tumor-suppressor genes (tumor-prone carriers) APC 164 54.1
E - Tumor-suppressor genes (normal carriers) ERCC5 14 4.6
Open in a new tab

Ranking of mutated samples by presumed severity of identified mutations. In category B, loss-of-heterozygosity (LOH) is perceived as mutations with allelic frequencies of more than 0.8. In category D, genes are known to promote tumor formation in heterozygous individuals, and in category E, genes have no increased tumorigenicity in heterozygous individuals.

Limitations of the study

This study demonstrates cancer-related mutations in MSCs and NSCs, with lineage specificity when compared with ESCs. However, it has several limitations that should be considered. First, the stringent filtration criteria and coverage thresholds for mutation identification, although yielding high-confidence results for known cancer-related mutations, may lead to overlooked additional mutations. Second, the cell passaging number, which may correlate with the abundance of mutations, was unavailable for many samples, preventing us from fully assessing the effects of passaging on our findings. Moreover, although we could detect mutations that were acquired in vitro across all three stem cell lineages, our analysis was limited by the available information on each cell line upon its generation, and during its propagation in culture. This limitation is particularly relevant for MSCs and NSCs, where this lack of information prevents definitive conclusions about whether the mutations reflect pre-existing somatic mosaicism or appeared upon culture adaptation. Furthermore, while we have identified many recurrently mutated genes, functional analysis was somewhat limited by sample size, and the functionality of a single gene was only possible for TP53.

Methods

RNA-seq data collection and processing

Bulk RNA-seq datasets of embryonic, mesenchymal, and NSCs were collected from the NCBI Sequence Read Archive (https://www.ncbi.nlm.nih.gov/sra). Mutation identification was performed using established pipelines and tools (RNA2CM [Lezmi and Benvenisty, 2021], Trimmomatic [Bolger et al., 2014], and STAR [Dobin et al., 2013]). The detailed pipeline and thresholds are provided in the supplemental information.

Analysis of gene expression

The gene expression analysis was performed using R packages: EdgeR (Robinson et al., 2010), limma (Ritchie et al., 2015), fgsea (http://bioconductor.org/packages/fgsea), and dimensionality reduction was applied via three separate methods (PCA [Venables and Ripley, 2002], UMAP [McInnes et al., 2018], and t-SNE [Van Der Maaten and Hinton, 2008]). Loss-of-heterozygosity was examined using an adapted e-SNP-Karyotyping approach (Weissbein et al., 2016). Detailed protocols and statistical methods are provided in the supplemental information.

Datasets used

All RNA-seq data are available in NCBI’s Sequence Read Archive, referenced through accession codes provided in Table S1.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Nissim Benvenisty (nissimb@mail.huji.ac.il).

Materials availability

No new materials or reagents were generated in the study.

Data and code availability

Sequencing data used in this project are all from previous studies and are publicly available. Accession numbers are listed in Table S1. All original code has been deposited at GitHub and is publicly available at https://github.com/JungJonathan/ASC2CM’ as of the date of publication.

Acknowledgments

We thank all members of The Azrieli Center for Stem Cells and Genetic Research for critical reading of the manuscript. The research was partially supported by the Azrieli Foundation, the Rosetrees Trust, the Israel Science Foundation (no. 2054/22), the ISF-Israel Precision Medicine Partnership Program (no. 3605/21) and the HEAL project, funded by the European Union, EU Horizon (no. 101056712). Views and opinions expressed are, however, those of the authors only and do not necessarily reflect those of the European Union or EU Horizon; neither the European Union nor EU Horizon can be held responsible for them. N.B. is the Herbert Cohn Chair in Cancer Research.

Author contributions

J.J. and N.B. designed the research. J.J. gathered, verified, and analyzed the data. J.J. and N.B. interpreted the results and prepared the manuscript. N.B. supervised the study and secured funding.

Declaration of interests

The authors declare no competing interests.

Published: September 18, 2025

Footnotes

Supplemental information can be found online at https://doi.org/10.1016/j.stemcr.2025.102643.

Supplemental information

Document S1. Figures S1–S4
mmc1.pdf (4.6MB, pdf)
Table S1. Stem cell samples with their identified mutations in cancer-related genes
mmc2.csv (72.2KB, csv)
Table S2. Type of cancer-related mutations in all aberrant stem cell samples
mmc3.csv (33.7KB, csv)
Document S2. Article plus supplemental information
mmc4.pdf (14.9MB, pdf)

References

  1. Abyzov A., Tomasini L., Zhou B., Vasmatzis N., Coppola G., Amenduni M., Pattni R., Wilson M., Gerstein M., Weissman S., et al. One thousand somatic SNVs per skin fibroblast cell set baseline of mosaic mutational load with patterns that suggest proliferative origin. Genome Res. 2017;27:512–523. doi: 10.1101/gr.215517.116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Amariglio N., Hirshberg A., Scheithauer B.W., Cohen Y., Loewenthal R., Trakhtenbrot L., Paz N., Koren-Michowitz M., Waldman D., Leider-Trejo L., et al. Donor-derived brain tumor following neural stem cell transplantation in an ataxia telangiectasia patient. PLoS Med. 2009;6:e1000029. doi: 10.1371/journal.pmed.1000029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Andrews P.W., Barbaric I., Benvenisty N., Draper J.S., Ludwig T., Merkle F.T., Sato Y., Spits C., Stacey G.N., Wang H., Pera M.F. The consequences of recurrent genetic and epigenetic variants in human pluripotent stem cells. Cell Stem Cell. 2022;29:1624–1636. doi: 10.1016/j.stem.2022.11.006. [DOI] [PubMed] [Google Scholar]
  4. Avior Y., Lezmi E., Eggan K., Benvenisty N. Cancer-related mutations identified in primed human pluripotent stem cells. Cell Stem Cell. 2021;28:10–11. doi: 10.1016/j.stem.2020.11.013. [DOI] [PubMed] [Google Scholar]
  5. Avior Y., Sagi I., Benvenisty N. Pluripotent stem cells in disease modelling and drug discovery. Nat. Rev. Mol. Cell Biol. 2016;17:170–182. doi: 10.1038/nrm.2015.27. [DOI] [PubMed] [Google Scholar]
  6. Bae T., Tomasini L., Mariani J., Zhou B., Roychowdhury T., Franjic D., Pletikos M., Pattni R., Chen B.J., Venturini E., et al. Different mutational rates and mechanisms in human cells at pregastrulation and neurogenesis. Science. 2018;359:550–555. doi: 10.1126/science.aan8690. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Baloh R.H., Johnson J.P., Avalos P., Allred P., Svendsen S., Gowing G., Roxas K., Wu A., Donahue B., Osborne S., et al. Transplantation of human neural progenitor cells secreting GDNF into the spinal cord of patients with ALS: a phase 1/2a trial. Nat. Med. 2022;28:1813–1822. doi: 10.1038/s41591-022-01956-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Ben-David U., Kopper O., Benvenisty N. Expanding the boundaries of embryonic stem cells. Cell Stem Cell. 2012;10:666–677. doi: 10.1016/j.stem.2012.05.003. [DOI] [PubMed] [Google Scholar]
  9. Ben-David U., Mayshar Y., Benvenisty N. Large-scale analysis reveals acquisition of lineage-specific chromosomal aberrations in human adult stem cells. Cell Stem Cell. 2011;9:97–102. doi: 10.1016/j.stem.2011.06.013. [DOI] [PubMed] [Google Scholar]
  10. Berkowitz A.L., Miller M.B., Mir S.A., Cagney D., Chavakula V., Guleria I., Aizer A., Ligon K.L., Chi J.H. Glioproliferative lesion of the spinal cord as a complication of “stem-cell tourism”. N. Engl. J. Med. 2016;375:196–198. doi: 10.1056/NEJMc1600188. [DOI] [PubMed] [Google Scholar]
  11. Bolger A.M., Lohse M., Usadel B. Trimmomatic: A flexible trimmer for Illumina Sequence Data. Bioinformatics. 2014;30:2114–2120. doi: 10.1093/bioinformatics/btu170. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Branzei D., Foiani M. Regulation of DNA repair throughout the cell cycle. Nat. Rev. Mol. Cell Biol. 2008;9:297–308. doi: 10.1038/nrm2351. [DOI] [PubMed] [Google Scholar]
  13. Corthay A. Does the immune system naturally protect against cancer? Front. Immunol. 2014;5:197. doi: 10.3389/fimmu.2014.00197. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. De Becker A., Heestermans R., De Brouwer W., Bockstaele K., Maes K., Van Riet I. Genetic profiling of human bone marrow mesenchymal stromal cells after in vitro expansion in clinical grade human platelet lysate. Front. Bioeng. Biotechnol. 2022;10:1008271. doi: 10.3389/fbioe.2022.1008271. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. De Filippis L., Binda E. Concise review: self-renewal in the central nervous system: neural stem cells from embryo to adult. Stem Cells Transl. Med. 2012;1:298–308. doi: 10.5966/sctm.2011-0045. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Derks L.L.M., Van Boxtel R. Stem cell mutations, associated cancer risk, and consequences for regenerative medicine. Cell Stem Cell. 2023;30:1421–1433. doi: 10.1016/j.stem.2023.09.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Dobin A., Davis C.A., Schlesinger F., Drenkow J., Zaleski C., Jha S., Batut P., Chaisson M., Gingeras T.R. STAR: ultrafast universal RNA-seq aligner. Bioinformatics. 2013;29:15–21. doi: 10.1093/bioinformatics/bts635. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Fan X., Mikolaenko I., Elhassan I., Ni X., Wang Y., Ball D., Brat D.J., Perry A., Eberhart C.G. NOTCH1 and NOTCH2 have opposite effects on embryonal brain tumor growth. Cancer Res. 2004;64:7787–7793. doi: 10.1158/0008-5472.CAN-04-1446. [DOI] [PubMed] [Google Scholar]
  19. Fasching L., Jang Y., Tomasi S., Schreiner J., Tomasini L., Brady M.V., Bae T., Sarangi V., Vasmatzis N., Wang Y., et al. Early developmental asymmetries in cell lineage trees in living individuals. Science. 2021;371:1245–1248. doi: 10.1126/science.abe0981. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Fernandez-Muñoz B., Garcia-Delgado A.B., Arribas-Arribas B., Sanchez-Pernaute R. Human neural stem cells for cell-based medicinal products. Cells. 2021;10:2377. doi: 10.3390/cells10092377. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Halliwell J., Barbaric I., Andrews P.W. Acquired genetic changes in human pluripotent stem cells: origins and consequences. Nat. Rev. Mol. Cell Biol. 2020;21:715–728. doi: 10.1038/s41580-020-00292-z. [DOI] [PubMed] [Google Scholar]
  22. Han L., He H., Yang Y., Meng Q., Ye F., Chen G., Zhang J. Distinctive clinical and pathologic features of immature teratomas arising from induced pluripotent stem cell-derived beta cell injection in a diabetes patient. Stem Cells Dev. 2022;31:97–101. doi: 10.1089/scd.2021.0255. [DOI] [PubMed] [Google Scholar]
  23. Kim M., Rhee J.K., Choi H., Kwon A., Kim J., Lee G.D., Jekarl D.W., Lee S., Kim Y., Kim T.M. Passage-dependent accumulation of somatic mutations in mesenchymal stromal cells during in vitro culture revealed by whole genome sequencing. Sci. Rep. 2017;7 doi: 10.1038/s41598-017-15155-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Kimbrel E.A., Lanza R. Next-generation stem cells - ushering in a new era of cell-based therapies. Nat. Rev. Drug Discov. 2020;19:463–479. doi: 10.1038/s41573-020-0064-x. [DOI] [PubMed] [Google Scholar]
  25. Kirkeby A., Main H., Carpenter M. Pluripotent stem-cell-derived therapies in clinical trial: A 2025 update. Cell Stem Cell. 2025;32:10–37. doi: 10.1016/j.stem.2024.12.005. [DOI] [PubMed] [Google Scholar]
  26. Kuijk E., Jager M., van der Roest B., Locati M.D., Van Hoeck A., Korzelius J., Janssen R., Besselink N., Boymans S., van Boxtel R., Cuppen E. The mutational impact of culturing human pluripotent and adult stem cells. Nat. Commun. 2020;11:2493. doi: 10.1038/s41467-020-16323-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Levine A.J. p53: 800 million years of evolution and 40 years of discovery. Nat. Rev. Cancer. 2020;20:471–480. doi: 10.1038/s41568-020-0262-1. [DOI] [PubMed] [Google Scholar]
  28. Lezmi E., Benvenisty N. Identification of cancer-related mutations in human pluripotent stem cells using RNA-seq analysis. Nat. Protoc. 2021;16:4522–4537. doi: 10.1038/s41596-021-00591-5. [DOI] [PubMed] [Google Scholar]
  29. Lezmi E., Jung J., Benvenisty N. High prevalence of acquired cancer-related mutations in 146 human pluripotent stem cell lines and their differentiated derivatives. Nat. Biotechnol. 2024;42:1667–1671. doi: 10.1038/s41587-023-02090-2. [DOI] [PubMed] [Google Scholar]
  30. Massagué J. TGFβ in cancer. Cell. 2008;134:215–230. doi: 10.1016/j.cell.2008.07.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. McInnes L., Healy J., Saul N., Großberger L. UMAP: uniform manifold approximation and projection. J. Open Source Softw. 2018;3:861. doi: 10.21105/joss.00861. [DOI] [Google Scholar]
  32. Merkle F.T., Ghosh S., Kamitaki N., Mitchell J., Avior Y., Mello C., Kashin S., Mekhoubad S., Ilic D., Charlton M., et al. Human pluripotent stem cells recurrently acquire and expand dominant negative p53 mutations. Nature. 2017;545:229–233. doi: 10.1038/nature22312. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Morante-Redolat J.M., Porlan E. Neural stem cell regulation by adhesion molecules within the subependymal niche. Front. Cell Dev. Biol. 2019;7:102. doi: 10.3389/fcell.2019.00102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Rauch A., Mandrup S. Transcriptional networks controlling stromal cell differentiation. Nat. Rev. Mol. Cell Biol. 2021;22:465–482. doi: 10.1038/s41580-021-00357-7. [DOI] [PubMed] [Google Scholar]
  35. Ren L., Chen X., Chen X., Li J., Cheng B., Xia J. Mitochondrial dynamics: fission and fusion in fate determination of mesenchymal stem cells. Front. Cell Dev. Biol. 2020;8 doi: 10.3389/fcell.2020.580070. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Ritchie M.E., Phipson B., Wu D., Hu Y., Law C.W., Shi W., Smyth G.K. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucl. Acids Res. 2015;43:e47. doi: 10.1093/nar/gkv007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Robinson M.D., McCarthy D.J., Smyth G.K. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26:139–140. doi: 10.1093/bioinformatics/btp616. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Rouhani F.J., Zou X., Danecek P., Badja C., Amarante T.D., Koh G., Wu Q., Memari Y., Durbin R., Martincorena I., et al. Substantial somatic genomic variation and selection for BCOR mutations in human induced pluripotent stem cells. Nat. Genet. 2022;54:1406–1416. doi: 10.1038/s41588-022-01147-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Szklarczyk D., Kirsch R., Koutrouli M., Nastou K., Mehryary F., Hachilif R., Gable A.L., Fang T., Doncheva N.T., Pyysalo S., et al. STRING database in 2023: protein-protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucl. Acids Res. 2023;51:638–646. doi: 10.1093/nar/gkac1000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Tate J.G., Bamford S., Jubb H.C., Sondka Z., Beare D.M., Bindal N., Boutselakis H., Cole C.G., Creatore C., Dawson E., et al. COSMIC: the catalogue of somatic mutations in cancer. Nucl. Acids Res. 2019;47:941–947. doi: 10.1093/nar/gky1015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Van Der Maaten L., Hinton G. Visualizing high-dimensional data using t-SNE. J. Mach. Learn. Res. 2008;9:2579–2605. http://jmlr.org/papers/v9/vandermaaten08a.html [Google Scholar]
  42. Venables W.N., Ripley B.D. Springer; 2002. Modern Applied Statistics with S. [DOI] [Google Scholar]
  43. Vijg J., Dong X. Pathogenic mechanisms of somatic mutation and genome mosaicism in aging. Cell. 2020;182:12–23. doi: 10.1016/j.cell.2020.06.024. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Weissbein U., Schachter M., Egli D., Benvenisty N. Analysis of chromosomal aberrations and recombination by allelic bias in RNA-Seq. Nat. Commun. 2016;7 doi: 10.1038/ncomms12144. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Willis A., Jung E.J., Wakefield T., Chen X. Mutant p53 exerts a dominant negative effect by preventing wild-type p53 from binding to the promoter of its target genes. Oncogene. 2004;23:2330–2338. doi: 10.1038/sj.onc.1207396. [DOI] [PubMed] [Google Scholar]
  46. Zaben M.J., Gray W.P. Neuropeptides and hippocampal neurogenesis. Neuropeptides. 2013;47:431–438. doi: 10.1016/j.npep.2013.10.002. [DOI] [PubMed] [Google Scholar]
  47. Zhang L., Liu Q., Hu H., Zhao L., Zhu K. Progress in mesenchymal stem cell mitochondria transfer for the repair of tissue injury and treatment of disease. Biomed. Pharmacother. 2022;153 doi: 10.1016/j.biopha.2022.113482. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Document S1. Figures S1–S4
mmc1.pdf (4.6MB, pdf)
Table S1. Stem cell samples with their identified mutations in cancer-related genes
mmc2.csv (72.2KB, csv)
Table S2. Type of cancer-related mutations in all aberrant stem cell samples
mmc3.csv (33.7KB, csv)
Document S2. Article plus supplemental information
mmc4.pdf (14.9MB, pdf)

Data Availability Statement

Sequencing data used in this project are all from previous studies and are publicly available. Accession numbers are listed in Table S1. All original code has been deposited at GitHub and is publicly available at https://github.com/JungJonathan/ASC2CM’ as of the date of publication.

Articles from Stem Cell Reports are provided here courtesy of Elsevier

RESOURCES