Telomere length of both parents contribute to heritable POT1 cancer-predisposition syndrome

Abstract

Germline mutations in POT1 are linked to familial cancer predisposition, and somatic POT1 mutations occur recurrently in tumors. These mutations promote oncogenesis by enabling aberrant telomere elongation. For inherited POT1 mutations, a critical question is the extent to which elongated telomeres are transmitted to the next generation from the POT1 carrier parent and whether the inherited hyper-elongated telomeres elevate cancer risk. Using a nanopore sequencing approach that provides haplotype-specific telomere length measurements, we examined telomere inheritance in families harboring POT1 mutations. We found that individuals preferentially inherit their longest telomeres from the carrier parent, consistent with extensive telomere elongation in the carrier germline, whereas their comparatively short telomeres originate from the non-carrier parent. Analysis of carrier and non-carrier siblings showed that both sets of parental telomeres are longer in POT1 carriers, yet the shortest non-carrier-derived telomeres undergo disproportionately greater elongation than those inherited from the carrier parent. This identifies a mechanism of genetic anticipation in which the inheritance of long telomeres from one parent drives excessive extension of shorter telomeres. These findings demonstrate that telomere length inherited from both parents jointly defines the telomere-based tumor suppressor mechanism.

Summary sentence

Allele specific nanopore sequencing reveals that POT1 mutations reshape germline and somatic telomere dynamics, uncovering a novel mechanism of generational anticipation driven by preferential elongation of short inherited telomeres.

Introduction

Mutations in the gene encoding the shelterin protein POT1 occur in both sporadic and familial cancers, including chronic lymphocytic leukemia (CLL), glioma, melanoma, and other malignancies, with an overall estimated prevalence of ~3% across cancers^1-12. These mutations promote cancer by driving telomere elongation, bypassing the tumor-suppressive checkpoint normally triggered by critically short telomeres¹³. Yet, how heterozygous POT1 mutations trigger telomere elongation and give rise to a distinct cancer spectrum, including glioma, melanoma, and CLL, remains poorly understood. POT1, a core shelterin subunit, binds the telomeric single-stranded overhang to limit telomerase-mediated extension and prevent ataxia telangiectasia and Rad3-related (ATR) kinase-mediated DNA damage signaling^14-20. Inherited POT1 mutations cause autosomal dominant cancer predisposition without loss of heterozygosity^1,2,8. Engineered mutations in human pluripotent and hematopoietic stem cells show that heterozygous POT1 lead to progressive elongation in cell types with active telomerase while the chromosome end protective function of telomeres remains intact¹³.

Germline transmission of mutations that cause pathogenic telomere length changes creates a unique scenario in which not only the mutation itself is transmitted but potentially also its phenotypic consequence, namely an aberrant telomere length in the germline.

Telomere length inheritance is a well-established mechanism of genetic anticipation in telomere biology disorders that arise from short telomeres. Telomere biology disorders are a spectrum of life-threatening conditions that include bone marrow failure, liver disease, lung disease and other complications caused by mutations in telomere maintenance genes that result in short telomeres. In families with these disorders, aberrantly short telomeres inherited from a parent can lead to earlier onset and increased severity of disease symptoms in their children^21-25. Consistent with this mechanism, decreased generational telomere length following both maternal and paternal inheritance of short telomeres has been documented^24,25.

It has been recently proposed that offspring of POT1 mutation carriers who inherit the same germline variant develop cancer several decades earlier due to transmission of long telomeres⁴. However, the underlying mechanisms for this anticipation remain unclear, because telomere length measurements did not detect progressive telomere elongation across consecutive generations⁴. This raises a central question in POT1 mutation positive families: are elongated telomeres themselves transmitted to the next generation? Moreover, even if long telomeres were inherited, it remains unclear how this would drive earlier disease onset. Since a small number of critically short telomeres are sufficient to trigger replicative senescence and enforce a proliferative barrier^26-28, shorter telomeres contributed by the non-carrier parent are expected to preserve tumor suppression.

How inherited long telomeres might overcome this fundamental safeguard therefore remains unresolved, as the dynamics of inherited telomere length remain incompletely understood. Oocytes generally possess longer telomeres than sperm, yet sperm telomere length increases with age²⁹. After fertilization, zygotic telomere length more closely resembles that of sperm, supporting the model that telomeres are “reset” early in development. The mechanism underlying this resetting remains to be determined, but current evidence indicates that it is incomplete, resulting in a complex inheritance pattern with a strong paternal influence^30-33. Consequently, telomere length is heritable, and individual parental telomeres can differ by more than 6 kb at the zygotic stage³⁴.

In POT1 mutation positive families, where half of the telomeres of each generation derive from the affected parent, it remains unclear (1) whether POT1 mutations drive excessive telomere elongation in the germline, (2) whether non-carrier telomeres retain tumor suppressive function, and (3) if inheritance of long telomeres cause disease anticipation. To this end, we use Oxford Nanopore long-read sequencing to map haplotype-specific telomere inheritance in families with POT1 mutations.

Results

Complex cancer spectrum in families with POT1 mutations

To assess the impact of cancer-associated POT1 mutations on telomere length inheritance, we analyzed two unrelated families with germline POT1 mutations (see Supplementary Table 1 for all samples analyzed in this study).

Proband 1 harbored the previously reported heterozygous POT1 c.1851_1852del; p.Asp617GlufsTer9^10,35,36 mutation, which was paternally inherited and is located in the C-terminus of POT1, resulting in truncation of the C-terminal TPP1 binding domain³⁷. The proband presented with a posterior fossa ependymoma at age 3 years. Family members available for study included both parents, healthy at the time of the sample collection, with the father being the POT1 carrier.

Proband 2 carried a maternally inherited heterozygous POT1 c.265_273delinsAATCTT; p.Tyr89_Lys91delinsAsnLeu mutation, which is located in the first Oligonucleotide/Oligosaccharide-Binding (OB) fold of POT1. The proband was diagnosed with Hodgkin lymphoma at age 12 years and was managed with chemotherapy, radiation therapy, and autologous bone marrow transplant. Unfortunately, he experienced disease relapse at 13 and 17 years of age, followed by osteosarcoma at 22 years of age, for which he was treated with chemotherapy and radiation. At 27 years of age, he experienced osteosarcoma relapse and expired at age 29 years. Available family members included his mother, sister, and five nephews (his sister’s children), with mutation status confirmed by sequencing. The mother, sister and two of five nephews were confirmed to harbor the same POT1 variant as the proband. The proband’s sister was diagnosed with melanoma and CLL at the age of 32 years. Her melanoma was treated with wide excision and her CLL was managed with watchful waiting for 8 years before starting therapy on a clinical trial. The proband’s mother was diagnosed with thyroid cancer at 64 years. The nephews, ranging from 9 to 16 years of age, have all been healthy. All POT1 carriers are followed annually in a cancer predisposition clinic.

In addition, we studied two other unrelated probands with germline POT1 mutations for whom family members were not available. Specifically, Proband 3 harbored POT1 mutation c.1087C>T; p.Arg363Ter and was diagnosed at 2 years of age with high-risk neuroblastoma to which he succumbed to at age 5 years. Proband 4 was found to harbor POT1 mutation c.1071dup; p.Gln358SerfsTer13 after being diagnosed with embryonal rhabdomyosarcoma of prostate at 13 years of age. He experienced recurrence at 14 years before expiring at 16 years. Three unrelated control samples from healthy adults (age 56, 48, and 38) were also evaluated (Figure 1A, Supplementary Table 1).

Figure 1: Nanopore long read sequencing of POT1 cancer families.

(A) Pedigrees for families with inherited caPOT1 mutations and associated cancer spectrums. Abbreviations as follows: P=Proband, M=Mother, F=Father, S=Sister, N=Nephew. Parenthetical numbers indicate age at time of the study (see also Supplementary Table 1).

(B) Bulk telomere length measurements by nanopore sequencing of the indicated individuals. Color groups denote independent families (or unrelated control samples) whereas a hashed fill identifies individuals with no POT1 mutation and solid fill identifies carriers of a caPOT1 mutation. The total fraction of long telomeres (>10kb) for each sample was compared to Control Sample A using a Fisher’s exact test with FDR correction for multiple testing. Only significant comparisons are shown (**** = p-value<0.0001)

Telomere length profiling of POT1 families by Nanopore sequencing

We applied Nanopore-based telomere-targeted sequencing to evaluate telomere length in this cohort (Supplementary Figure 1)^38-42. Bulk telomere length measurements closely matched results obtained by telomere restriction fragment (TRF) analysis (Supplementary Figure 2). In total, we sequenced peripheral blood samples from 18 individuals: 9 carrying POT1 mutations (6 individuals with cancer, 3 unaffected carriers), 4 non-carrier family members, and 3 unrelated controls (Figure 1B, Supplementary Table 1). Across samples, 1.3k–7k reads containing telomere ends were detected per individual, averaging 0.44% of total reads, representing a 44-fold enrichment compared to whole-genome sequencing. Proband samples from Families 1 and 2 were sequenced twice from independent clinical samples collected 5 years apart, showing strong concordance between replicates, and individual sample measurements were consistent across sequencing rounds (Supplementary Figure 3). On average, probands displayed median telomere lengths ~1.9 kb longer than our three controls (age 56, 48, and 38), and 7 of 11 POT1 carriers had a significantly enriched fraction of long telomeres (>10 kb) compared to controls (Figure 1B).

Nanopore sequencing robustly resolves telomere length inheritance patterns in POT1 families

To trace the inheritance of telomeres across generations, we used a recently published method, Telogator2, to cluster telomere reads based on the Telomere Variant Repeat (TVR) regions located between the telomere and sub-telomere⁴³. Extensive polymorphism between TVR regions (Figure 2A) provide haplotype-specific markers for chromosome ends and are subject to Mendelian inheritance. Therefore, while TVRs complicate alignments to the human reference sequence (Supplementary Figure 4), leveraging TVRs to cluster nanopore reads from the same chromosome end in a reference-free manner, as implemented in Telogator2, eliminates the need for a family-specific reference genomes to trace telomere inheritance. By clustering individual reads based on their shared TVR region, we can confidently identify telomeres from single chromosome ends, enabling us to trace their inheritance via comparison to related samples (Figure 2B, C).

Figure 2: TVR Clustering enables allele correlation between family members.

(A) Schematic of the Telomere Variant Repeat (TVR) region situated between canonical TTAGGG repeats of the telomere and the subtelomere comprised of unique chromosomal DNA

(B) Example of individual telomere clusters that are shared between Proband 1 and their mother (top) and father (bottom). Individual points show a single measurement of telomere length for the indicated sample, with the grey bar denoting the 75th percentile telomere length for that allele. The TVR region is identified with a dashed line and colored bars within the TVR denote unique telomere variant sequences. For a full table of variant sequences identified and colors used, please refer to the Materials and Methods.

(C) As in panel A for an allele identified within patient Family 2 which is shared across six unique DNA samples from patient Family 2.

In total, after filtering out telomere-associated nanopore reads with no detected TVR and potential interstitial alleles, we identified an average of 87 unique telomeric clusters per sample representing 95% of expected 92 chromosome ends. We then compared clusters from related individuals by aligning TVR consensus sequences using a Levenshtein ratio cutoff of 0.85 (Supplementary Table 2, see Materials and Methods). With this threshold, we found that an average of 97% of TVR haplotypes could be identified between independent samples from each of Proband 1 and Proband 2. In contrast, comparisons between unrelated individuals show only limited haplotype sharing (2%, or 17 of a total of 838 unique haplotypes from 18 samples) due to the high variability of TVRs across humans. Further, for both Probands 1 and 2, the telomere length of individual haplotype clusters showed strong correlation between independent samples of the respective Proband (Supplementary Figure 5). Consistent with original Telogator2 observations⁴³, the coefficient of determination (R²) for the 75th percentile of telomere length was stronger than either mean or median, likely because it is less affected by outliers e.g. the presence of short telomere fragments generated during sample preparation. Therefore, the 75th percentile telomere length was used for subsequent cluster comparisons⁴³.

The longest telomeres of children with POT1 mutations are preferentially inherited from their carrier parent

Robust assignment of telomere inheritance allowed us to analyze the Family 1 trio with telomere length data from both parents and the proband. Absolute telomere length comparisons are confounded by donor age, cancer presentation, treatment, and sample composition, particularly across generations due to the significant age differences (see Supplementary Table 1). To overcome these limitations, we rank ordered telomeres by relative telomere length within each sample, evaluated whether each telomere was inherited from the carrier or non-carrier parent, and compared rank order across family members. This rank-based approach is internally controlled and independent of absolute telomere length, allowing us to isolate the contribution of inherited telomeres and obtain a direct readout of mutation-driven inheritance patterns.

Rank-ordering revealed a significant enrichment of telomeres inherited from the POT1 mutation carrier father among the longest telomeres (Figure 3A and Supplementary Figure 6), with a corresponding depletion in the shortest, consistent with germline elongation in the carrier parent. This pattern was not unique to Family 1; across all families analyzed, carrier-derived haplotypes were consistently overrepresented among the longest third of telomeres and underrepresented among the shortest third, highlighting germline elongation as a reproducible and mutation-specific effect (Figure 3B). The shortest telomeres, in contrast, were predominantly inherited from the non-carrier parent, suggesting that these telomeres may still impose functional limits on proliferative capacity even when long telomeres are transmitted from the carrier parent.

Figure 3: Longest telomeres are preferentially inherited from the caPOT1 mutation carrier parent.

(A) All telomere allele clusters identified for Proband 1 rank ordered by 75th percentile telomere length. Points show a single telomere length measurement, whereas bar colors denote whether that allele cluster corresponds to one from the carrier parent (red), the non-carrier parent (blue), or whether the origin could not be identified with our clustering parameters (grey).

(B) Percentage of alleles determined to arise from a carrier parent in an indicated fraction normalized to the total percentage of detected carrier alleles: A=All alleles (normalization group), R=random sampling of one third of all telomere alleles for each sample (datapoints represent three independent random sampling events for each sample), 1=shortest third for telomere length for each sample, 2=middle third, and 3=longest third. Each group was compared to the random sampling control using an unpaired t-test with Welch's correction. Only significant comparisons are shown (*** = p-value<0.001,**** = p-value<0.0001)

(C) Shared alleles inherited from the non-carrier parent in Family 1 independently rank-ordered by 75th percentile telomere length for Proband 1 (x-axis) and the non-carrier parent (y-axis). Independent samples for Proband 1 are plotted separately and denoted a and b. Ordinary least squares linear regression was performed using the statsmodels.api OLS module.

(D) As in panel C for alleles inherited from the carrier parent.

Direct comparison of Proband 1 rank order with each parent revealed two findings: telomere rank order correlated strongly with the non-carrier mother (Figure 3C) but not with the carrier father (Figure 3D), suggesting that germline elongation “scrambled” the original telomere length order in the carrier. This differential pattern was also observed in absolute haplotype-specific telomere length measurements (Supplementary Figure 7). Together, these results indicate that the excessively long telomeres caused by POT1 mutations are not fully reset during early embryogenesis. This is consistent with prior studies showing that parental telomere length can be inherited and remain detectable in offspring, reflecting incomplete telomere reprogramming in the zygote^30-33. Conversely, the persistence of short telomeres on non-carrier haplotypes suggests that telomere-based tumor-suppressive mechanisms may remain at least partially intact, as critically short telomeres can still enforce replicative barriers and limit unchecked proliferation.

A novel mechanism of genetic anticipation through biased telomere elongation

We next analyzed Family 2, comparing telomeres from the two nephews who harbor the maternally inherited POT1 variant to their three non-carrier siblings. Rank-order comparison revealed that telomere length in the unaffected nephews correlated strongly with that of their mother (Figure 4A), whereas this correlation was lost in the affected nephews (Figure 4B). Similar to Family 1, this indicates that telomere length patterns from the affected parent are significantly altered upon transmission to the next generation. However, the absence of paternal samples in Family 2 makes it difficult to attribute this effect exclusively to germline elongation in the mother versus post-fertilization elongation affecting both maternal and paternal haplotypes.

Figure 4: Comparison between carrier and non-carrier children identifies sporadic telomere elongation in POT1 mutation carriers.

(A) Shared alleles between Sister 2 and the caPOT1-negative Nephews independently rank-ordered by 75th percentile telomere length for the Nephews (x-axis) and the Sister (y-axis). Ordinary least squares linear regression was performed using the statsmodels.api OLS module. Adjusted R² = 0.366, slope = 0.6091, Intercept = 7.8188, and p-value = 7.75e-14.

(B) As in panel a for the caPOT1-positive Nephews. Adjusted R² = −0.009, slope = 0.0645, Intercept = 18.2415, and p-value = 0.570.

(C) 75th percentile telomere length for individual telomere allele clusters identified in common between at least two nephews, split into caPOT1-negative (left) and caPOT1-positive (right) nephews and by inheritance from either the carrier parent (red-Sister1) or likely the non-carrier parent (grey). Group comparisons were performed using Welch’s t-test with FDR-correction for multiple testing. Only significant differences are shown (** = p-value<0.01,**** = p-value<0.0001)

(D) For each allele that was shared between a caPOT1 positive and caPOT1 negative nephew, 75th percentile telomere length for the cluster was normalized to the caPOT1-negative nephew and split by allele inheritance (maternal/carrier or likely non-carrier). Sample comparison performed using Welch’s t-test, *=p-value < 0.05.

(E) Standard deviation in telomere length within an individual cluster by binned 75th percentile telomere length for that cluster. All alleles derived from carrier nephews were compared to non-carrier nephews using Welch’s t-test. Only significant p-values are shown (** = p-value<0.01, *** = p-value<0.001,**** = p-value<0.0001)

To explore post-fertilization dynamics, we directly compared absolute telomere lengths between carrier and non-carrier nephews, who are similar in age, share the same germline origin, and lack cancer or hematopoietic malignancies. In non-carrier nephews, haplotypes from the carrier parent were significantly longer than those from the non-carrier parent (median of haplotype-specific telomere lengths 7.5kb vs 6.6kb), indicating that long telomeres inherited from carriers persist through development without being fully reset (Figure 4C). By contrast, in POT1-positive nephews, there was no significant telomere length difference between parental carrier versus non-carrier haplotypes (Figure 4C). This suggests that POT1 mutations strongly exacerbate telomerase-mediated elongation of the shortest telomeres²⁶, inherited from the non-carrier in the next generation.

To directly test this hypothesis, we compared telomere length differences between POT1-positive nephews and their non-carrier siblings. For each telomere haplotype shared between siblings, 75th percentile telomere length in the POT1-positive nephew was normalized to the length of the corresponding telomere cluster in the non-carrier nephews. This enabled quantification of relative post-fertilization elongation at a per-allele level. Grouping by parental origin revealed a modest but significant increase in elongation of non-carrier derived telomeres (Figure 4D; Welch’s t test, p < 0.05), indicating that POT1 mutations promote preferential extension of telomeres inherited from the non-carrier parent. These results suggest a potential mechanism for genetic anticipation, in which the presence of long telomeres inherited from the carrier parent shifts under conditions where telomerase is limiting the process of elongation towards the shorter, non-carrier derived telomeres in the next generation. Importantly, elongation was not entirely confined to short ends; telomeres across all length classes exhibited sporadic extension, revealing a stochastic component of elongation in POT1 carriers that contributes to increased telomere length heterogeneity in POT1 mutation carriers (Figure 4E).

Short telomeres remain functionally relevant in a cancer-context-dependent manner

Next, we compared telomeres of the proband and his sister in Family 2, who both carry the POT1 mutation and share ~50% of their genetic information but presented with different hematopoietic cancers (HL vs. CLL) at 12 and 32 years of age, respectively. Bulk measurements revealed striking differences: the proband had very long telomeres in bone marrow samples that were free of cancer cells at the time of collection (median 8.5 kb), whereas his sister had significantly shorter telomeres (median 4.3 kb) in a peripheral blood sample obtained after her CLL diagnosis (Figure 1B). These differences underscore the challenges of interpreting absolute telomere length across individuals with distinct disease courses, age, treatments, and cellular compositions.

Despite these dramatic differences in absolute telomere length, rank-order analysis revealed a clear hereditary signature, with haplotypes from the POT1 carrier mother enriched in the longest third of telomeres and depleted in the shortest third in both siblings (Figures 5A and B, Supplementary Figure 6). Rank-order comparison between the siblings’ shared haplotypes also showed significant correlation (Figure 5C). Notably, the sister’s shortest chromosome-arm telomeres displayed constrained heterogeneity, consistent with these telomeres being critically short and being preferentially elongated by telomerase (Figure 5D).

Figure 5: Cancer presentation significantly alters telomere length profile in carrier siblings.

(A) All telomere allele clusters identified for Proband 2 rank ordered by 75th percentile telomere length. Points show a single telomere length measurement, whereas bar colors denote whether that allele cluster corresponds to one from the carrier parent (red) or whether the origin could not be confidently identified because there is no sample from the non-carrier parent (grey).

(B) As in panel A for Sister 2.

(C) Shared alleles between Proband 2 and Sister 2 independently rank-ordered by 75th percentile telomere length for the Proband (x-axis) and the Sister (y-axis). Ordinary least squares linear regression was performed using the statsmodels.api OLS module. Adjusted R² = 0.352, slope = 0.6051, Intercept = 9.0824, and p-value = 6.62e-06.

(D) Standard deviation in length for individual telomere clusters plotted against 75th percentile telomere length for both Proband 2 samples and Sister 2. Quadratic regression was performed using the scipy.optimize curve_fit module, resulting in the following equation: $1.88\mathrm{e}\text{-}5\ast \mathrm{x}2\text{-}0.068\ast \mathrm{x}+1552.55$. The grey region indicates the confidence interval defined by the square root of the covariance matrix.

These results reiterate that excessively long telomeres from POT1 mutations are not fully reset in embryogenesis, and, depending on the cancer context, short telomeres inherited from non-carriers persist and remain functionally relevant under the high replicative stress of CLL. Thus, telomere length from both parents contributes to heritable POT1 cancer-predisposition syndromes.

Discussion

POT1 mutations promote cancer by elongating telomeres in a telomerase-dependent manner^12,13,15,44. Telomerase expression is tightly regulated during development and across cell types, creating contexts in which POT1 mutations can facilitate tumorigenesis. Early telomerase expression ensures that telomeres are sufficient for proliferation; later restricted activity maintains replicative potential in adult stem cells and aberrant reactivation in cancers allows bypass of senescence. In germ cells, telomerase preserves telomere length across generations. Understanding how abnormal telomere changes are inherited has been difficult due to limited samples and lack of haplotype-specific resolution. Nanopore-based chromosome arm-specific measurements now make this possible, and we applied this approach to measure telomere length in families with cancer-associated POT1 mutations.

Our data show that telomeres inherited from POT1 carrier parents are consistently enriched among the longest third of haplotype-grouped telomeres, while short telomeres are preferentially inherited from the non-carrier parent. This pattern was preserved across both families, indicating that germline elongation in carriers is counteracted by persistence of short non-carrier haplotypes. These findings argue against a conventional model of genetic anticipation, as the rank order of short telomeres is preserved and in some cases, such as the sister of proband 2 with CLL, telomeres eroded to the point of constrained heterogeneity characteristic of CLL^8,45,46.

However, inheritance of long telomeres from the carrier parent suggests a distinct mechanism of anticipation. Because hyper-elongated telomeres are less efficiently extended by telomerase, under conditions where telomeres is limiting a greater proportion of telomerase activity may act on the shorter telomeres inherited from the non-carrier parent than would occur if all zygotic telomeres started at comparable lengths. Analysis of affected and unaffected nephews from family 2 supports this model, providing evidence that preferential elongation of the short telomeres could represent a mechanism for genetic anticipation in POT1 cancer predisposition syndromes.

The mechanism described above also offers a potential explanation for a long-standing unresolved question: why individuals with elongated telomeres, even in the absence of POT1 mutations, show increased cancer risk^47-50. This has been difficult to reconcile with the fact that tumor suppression is enforced by only a small subset of the shortest telomeres^26-28. We propose that long telomeres are not directly oncogenic. Instead, their presence, even without POT1 mutations, may merely exacerbate preferential elongation of the shortest telomeres, which in turn weakens the telomere-based tumor suppressor checkpoint.

This study is constrained by the limited number of families and incomplete banking across family members, reflecting the relative rarity of inherited POT1 mutations. Larger cohorts will be needed to define more precisely how inheritance of telomere patterns occur. For example, we cannot yet determine whether the germline "scrambling" of the relative telomere rank order observed in family 1 are specific to POT1 mutations or reflect general features of male germ cells. Evaluating the degree to which telomere rank order is correlated across families with different POT1 mutations will be challenging, as this correlation likely depends on the penetrance with which individual familial POT1 mutations cause telomere elongation in somatic tissues post fertilization. More penetrant mutations will induce more dramatic telomere extension post fertilization and thereby accelerate erosion of the inherited telomere length correlation with the parents. Notably, it also remains unclear why the rank order of haplotype-specific telomere length was more preserved between the proband 2 and his sister than between carrier nephews, which may reflect differences in paternal germline length or differences in sample composition.

Nonetheless, our findings establish that long telomeres elongated in the germline of POT1 carriers are transmitted to the next generation, while short telomeres from the non-carrier parent persist and remain functionally relevant. Thus, telomere length from both parents jointly defines inheritance patterns in POT1 families and contributes to the cancer predisposition syndrome.

Material and Methods

Patient sample collection and DNA isolation

The study was conducted in accordance with the US Common Rule and was deemed exempt by the Institutional Review Board at St. Jude Children’s Research Hospital (SJCRH; IRB Number: 24-1642, Study Number: POT1). Patients with POT1 tumor predisposition syndrome who underwent genetic counseling and germline genetic testing in the genetic predisposition clinic at SJCRH were included in this study. Demographic, clinical, family history, and germline genetic test result data were collected through review of electronic medical records of patients and families affected by POT1 Tumor Predisposition. Due to the retrospective nature, it was determined that written informed consent was not required. Note that for Proband 1, two peripheral blood samples were collected at clinic visits separated by four months. For Proband 2, bone marrow samples were collected and banked at age 12, shortly after HL diagnosis and prior to initiating radiation and chemotherapy, and again at age 17 at the time of HL recurrence. Both bone marrow samples were negative for malignant cells. The peripheral blood sample from proband 2’s sister was collected and banked at age 33 following her diagnosis with CLL.

Germline samples were collected and banked in the SJCRH biorepository under the SJFAMILY (NCT03050268) and/or TBANK (NCT01354002) studies. Patient family members were enrolled in SJFAMILY, a study that aims to learn more about the genetic causes of cancer. Patients were enrolled in SJFAMILY and/or TBANK, a study that aims to provide a biorepository of tumor and germline samples for research purposes. Through SJFAMILY and/or TBANK, germline samples were collected and banked for study participants. Samples consisted of blood or healthy bone marrow, aliquoted as cell suspensions or extracted DNA. Primary investigators on both protocols provided permission to utilize the samples in this study. An application was submitted and approved by the SJCRH Tissue Resource Distribution Committee. A Tissue Use Agreement was provided and a Materials Transfer Agreement was executed so the samples could be shared with the Hockemeyer Lab. All samples were de-identified.

Telomere length nanopore sequencing

Nanopore libraries were prepared following modified library preparation and enrichment protocols previously described by others^38-41,51. Specifically, 15-20 ug of purified gDNA were ligated to 2uM of Telobait (Supplementary Table 3) at 35C overnight with 4000 U of T4 DNA ligase in 200-300 uL reaction volume. Ligation reactions were then heat inactivated at 65C for 10 min followed by digestion with 5uL EcoRI-HF . Subsequently, digested DNA samples and ligation ligation reaction were cleaned up with 0.5X ratio of DNA size selection beads (AMPure XP alternative) from the UC Berkeley DNA Sequencing Facility (https://ucberkeleydnasequencing.com/dna-storeroom). Telobait-ligated DNA was resuspended in 100uL Milli-Q water (MQW) and enriched with 60uL of M280 Streptavidin Dynabeads incubation at RT overnight. Next day, beads were washed twice with 1X Binding and Wash (B&W) buffer (prepared from 2X: 10 mM Tris-HCl pH 7.5, 1 mM EDTA, 2 M NaCl as recommended by manufacturer, Thermo Scientific), equilibrated with 1X Cutsmart buffer for 5 min (without disturbing the beads) then subject to on-bead incubation with 5U T4 DNA polymerase (NEB) fill-in for 30 min at 37C. Beads were then washed again twice with 1X B&W buffer, equilibrated with Cutsmart and resuspended in 100uL of Cutsmart+EcoRV-HF (5uL) to elute the ligated Telobait for at least 6 h at 37C. Eluted Telobait fragments were purified with 1.1X Pronex beads and allowed to dry for at least 10 minutes up to 30 minutes prior to elution ensuring no residual ethanol carry-over. Eluted DNA was quantitated using the Qubit-HS dsDNA kit. Telobait fragments were end-prepped for Nanopore sequencing according to the Native Barcoding Kit 24 V14 (SQK-NBD114-24) protocol up to 14 sample barcodes with the following key modifications. For DNA-cleanup steps, 1.1X relative sample volume of Pronex beads were used instead of AMPure XP beads except for the final cleanup post-NA sequencing adapter ligation whereby 0.5X volume of AMPureXP beads were used instead. Bead drying steps were also extended to 10-15 minutes for cleaning up EcoRV eluates and after end-prep steps. Notably, drying was extended to 30-60 minutes when cleaning-up pooled barcoded Telobait fragments, in order to ensure no residual ethanol carryover prior to NA sequencing adapter ligation. Telobait fragments were eluted in EB buffer according to manufacturer’s protocol and the entire library loaded on the Promethion (R10.4.1) flow cell. Sequencing runs were performed up to 72 h with SUP (superaccuracy) basecalling with a FASTQ quality score cutoff > 8, min read length filter of 500 bp, barcode demultiplexing, with trimming corresponding to the Native Barcoding Kit 24 V14 (SQK-NBD114-24).

Telomere length restriction fragment analysis

Genomic DNA was prepared as described previously. Briefly, genomic DNA was digested with MboI, AluI, and RNase A overnight at 37°C. The resulting DNA was normalized, and 2 μg of digested DNA was resolved via pulse-field gel electrophoresis and subject to in-gel hybridization with radiolabelled telomeric probe as described previously²⁰. Gels were washed three times for 10 min in 0.2× SSC at room temperature then exposed on a phosphorimager screen. Exposed TRF images were analyzed for telomere length as described previously⁵². Analogous to binned lane intensities, phosphorimager gels were analyzed with Fiji/ImageJ2⁵³ using the Plot Profile tool for each lane starting from the well to the end of the gel to generate pixel-distance lane intensity profiles. The non-linear regression function of MW to pixel distance was fitted to the below equation model: $\mathrm{a}+\mathrm{b}∕(1+{(\mathrm{x}∕\mathrm{c})}^{\wedge }\mathrm{d})$ where $\mathrm{x}$ is the distance from the well and $\mathrm{y}$ is the DNA MW in (kb) and was performed using MyCurveFit.com (MyAssays Ltd.). This equation was then used to interpolate MWi at a given intensity. Subsequently 20 row averages were taken for MWi and Inti prior to calculating: $\sum (\text{Inti})∕\sum (\text{Inti}∕\text{MWi})$ for each lane was performed as described to determine the average telomere length.

Chromosome-end-specific alignments

We used the recently published Telometer analysis pipeline for chromosome-end-specific alignments³⁸. Initially fastq files were aligned and indexed with minimap2 against the combined T2T-CHM3, appended subtelomere reference sequence (available also on the Github) as recommended. Telomere reads were then identified with a minimum read length of 1000 bp and telomere repeat gap tolerance of 20 bp (“-m 1000, -g 20”) using V1.0 of the published pipeline available at https://github.com/santiago-es/Telometer. Telomere mapping was evaluated as the fraction of telomere reads with MAPQ filter > 30 out of total reads without mapping.

TVR Cluster analysis and Nanopore telomere length measurement

TVR clustering and telomere length measurements were performed on a per-sample basis with Telogator2 (https://github.com/zstephens/telogator2) using default settings for oxford nanopore sequencing except for the for the following adjusted parameters to account for the digestion step in our telomere enrichment protocol: minimum read length (-l) was lowered to 1000bp, minimum length of detected subtelomeric sequence (--filt-sub) was lowered to 200bp, and the minimum number of reads for defining clusters (-n) was set per-sample by identifying the number of reads passing the previous telomeric thresholds and assuming a 5-fold difference in read depth for identified clusters (# telomeric reads/92 chromosome ends/5) or a minimum of 5 reads, whichever was higher.

Following Telogator2 clustering, inter-sample cluster correlation was calculated by pairwise alignment of TVR consensus sequences across all samples sequenced. As reads without TVR sequences could not be compared in this manner, they were excluded from the analysis. Haplotype clusters were considered a “match” if the TVR sequence similarity exceeded a Levenshtein ratio of 0.85, given the inherent sequencing error of nanopore sequencing and the repetitive nature of telomere sequences. This threshold was experimentally determined to maximize the number of aligned haplotypes between family members while minimizing cross-family TVR matching. Because TVR sequences often contain large stretches of canonical telomeric repeats, the total telomere length for individual reads was calculated by the length of the TVR + the length of the terminal canonical telomeric repeats. All results of this clustering and individual read telomere length calculation are included in Supplementary Table 2.

Results were plotted using matplotlib and the seaborn library using custom python scripts. TVR sequences were encoded as reported for Telogator2⁴³ and plotted using the following color schema:

Category	Sequence	Symbol	Plot Color (Python Colorname)
Canonical	TTAGGG	C	lightgrey
C-type	TCAGGG	D	gold
G-type	TGAGGG	E	seagreen
J-type	TTGGGG	F	indianred
	CGAGGG	G	navy
	CTAGGG	H	yellowgreen
Common Variations	CTGGGG	I	royalblue
	TAAGGG	K	darkseagreen
	TCCGGG	L	cornflowerblue
	TTCGGG	M	deepskyblue
	CTAGG	N	goldenrod
Rare Variations	TCGGG	P	darkorange
	TGGGGG	Q	steelblue
	TTAAGGG	R	rebeccapurple
	TTAGAGGG	S	mediumvioletred
	TAGG (also includes TAGGG and TAGGGG)	T	darkmagenta
Errors	TTGG (also includes TGGG and TTGGG)	V	teal
	GGGGGG	W	firebrick

Supplementary Material

figure-S1

Supplementary Figure 1: Telomere library preparation for nanopore sequencing

Schematic of telomere enrichment and nanopore library preparation protocol

figure-S2

Supplementary Figure 2: (A) Telomere restriction fragment (TRF) length analysis of healthy unrelated controls compared against corresponding POT1 patient DNA samples used in this study from Family (1), (2), (3) as shown in Figure 1A. Mean telomere length determined by densitometry analysis of TRF are indicated at the bottom of each lane alongside ethidium bromide (EtBr) staining. (B) Comparison of telomere length measurements by TRF and nanopore sequencing. TRF mean telomere lengths from (A) versus 75th percentile bulk nanopore sequencing telomere lengths. For B, calculated values of 75pct telomere length (ATL = adjusted telomere length. ATL = TL_len+tvr_len) from Nanopore sequencing shown in Figure 1A in Supplementary Data 2 and mean telomere length of TRF calculations are shown in Supplementary Data 3.

figure-S3

Supplementary Figure 3: Comparison of sample consistency between sequencing rounds

Bulk telomere length measurements for all samples split by sequencing round. Each sequencing round refers to independent library preparation and subsequent sequencing from the same starting DNA sample. Median and interquartile range are indicated.

figure-S4

Supplementary Figure 4: Chromosome-end specific telomere alignment to standard reference genome using Telometer³⁸

Chromosome-end specific telomere length measurements for Proband 2 samples a (top panel) and b (bottom panel) with p-arm alignment indicated in red and q-arm alignment shown in blue. Sample median displayed for each aligned sample. Due to the high heterogeneity of subtelomeric regions and loss of distal unique sequences during enrichment, on average, fewer than 61% of telomeric reads mapped confidently to unique chromosome ends with a MAPQ > 30 filter. Moreover, the distribution of aligned reads was skewed, with up to 430-fold more reads mapping to the most represented telomere compared to the least. These findings are consistent with results from alternative pipelines, underscoring the inherent challenges of chromosome-end assignment without a sample-specific telomere-to-telomere reference genome^38-42. Moreover, chromosome-end specificity could in principle identify individual telomere pairs, but it does not resolve maternal vs. paternal inheritance or the distinct dynamics of allelic telomeres.

figure-S5

Supplementary Figure 5: Correlation of cluster-specific telomere length

75th percentile telomere length for each TVR cluster identified in both samples for Proband 1 (A) and Proband 2 (B). Ordinary least squares linear regression was performed using the statsmodels.api OLS module.

figure-S6

Supplementary Figure 6: Significant enrichment of carrier-inherited alleles among the longest telomeres

Intra-sample telomere rank order (A) and 75th percentile telomere length (B) plotted for each telomere allele, split by inheritance from the carrier (red) or non-carrier (blue, grey if non-carrier presumed but no non-carrier parental sample available) for the following samples: Proband 1 (left), Proband 2 (middle), and Sister 2 (right). Sample comparison performed using Welch’s t-test (** = p-value<0.01, *** = p-value<0.001,**** = p-value<0.0001)

figure-S7

Supplementary Figure 7: Correlation of Proband 1 telomere length with parental telomere length Proband (x-axis) and parental (y-axis) 75th percentile telomere length for shared alleles inherited from the carrier (red) and non-carrier (blue) parents. Ordinary least squares linear regression was performed independently for carrier and non-carrier alleles using the statsmodels.api OLS module.

Data availability

Nanopore sequencing fastq reads generated in this study are available at NCBI’s Sequence Read Archive (SRA) and is available through the accession number <redacted>. Remaining source data is provided with this paper.

Code availability

The Telogator2 package used for TVR and telomere read analysis of Nanopore reads is available at Github (https://github.com/zstephens/telogator2). Jupyter notebook and python code used for downstream analysis is available on Github at <redacted>. The accompanying PatientKey Excel file used for this analysis is in Supplementary Data 2.