Clinical and functional characterization of telomerase variants in patients with pediatric acute myeloid leukemia/myelodysplastic syndrome

To the Editor:

Human telomeres are composed of tandem TTAGGG repeats at chromosome ends. Telomerase is composed of TERT, a specialized reverse transcriptase that is the catalytic subunit, and hTR, an integral RNA subunit that provides the template for the addition of repeats onto chromosome termini, replenishing those lost during semiconservative DNA replication. Germline mutations in telomerase and related genes are associated with very short age-adjusted telomeres and a spectrum of telomere biology disorders (TBDs) that includes dyskeratosis congenita (DC) [1]. A diagnosis of DC confers high risk for bone marrow failure (BMF), myelodysplastic syndrome (MDS), and cancer [1, 2]. Hematopoietic cells are sensitive to reduction in telomerase, so that a pathogenic variant in TERT or TERC, encoding hTR, produces progressive leukocyte telomere shortening over successive generations [3]. An unexpectedly high incidence of deleterious TERT variants is observed in adults with MDS and hematologic malignancies and lacking other DC manifestations [4].

Prolonged myelosuppression and other treatment-related toxicities are common events in pediatric acute myeloid leukemia (AML) [5], but host factor risk determinants of therapy-related toxicities are not well understood. We determined the frequency of rare or previously-undescribed TERT variants in children (age ≤ 21 years) diagnosed with de novo, non-Down syndrome-associated AML or MDS. Cases were compared with non-cancer controls and a similar racial and ethnic distribution. Rare variants were defined as those with a minor allele frequency (MAF) < 1% in the Genome Aggregation Database (gnomAD), or with MAF 1—1.5% and prior association with TBDs. Variants not present in gnomAD were labeled as ‘previously-undescribed.’ The exons and intron/exon boundaries of TERT were sequenced by the Sanger method in DNA obtained from diagnostic bone marrow and confirmed in remission marrow or nonmalignant hematopoietic tissue. Lymphocyte telomere length (LTL) was determined by flow cytometry fluorescence in situ hybridization (flow-FISH) from banked diagnostic sample. Human embryonic kidney (HEK) 293T cells were co-transfected with TERC and TERT expression plasmids containing either wild type (WT) or mutant TERT. The immunopurified telomerase enzyme complex was assessed for specific activity and processivity by the direct primer extension assay. A masked investigator reviewed medical records and tabulated presence of six pre-determined clinical criteria associated with a DC phenotype: (1) 1st degree family history of cancer, liver, or pulmonary disease, (2) persistent liver or pulmonary disease of unknown etiology and preceding stem cell transplant (SCT), (3) any-treatment delay > 60 days due to cytopenia(s), (4) prolonged cytopenia(s) (e.g. absolute neutrophil count < 500/μl and/or platelet count < 75 × 10³/μl) persisting =60 days after treatment completion, (5) second cancer, and (6) evidence for nail dysplasia, skin hyperpigmentation, or oral leukoplakia preceding SCT. See Supplementary Materials and Methods and Supplementary Figs. 1–3 for methodological details.

The study cohort included 94 cases: 84 with AML and 10 with MDS, comprising 34 European non-Hispanics (Eur, 36%), 29 Latino (Lat, 31%), 14 African -Americans (Afr, 15%), 1 East Asian (1%), 1 South Asian (1%), and 15 (16%) whose background was unknown. Locally-obtained controls included 254 adults without cancer history: 97 Eur, 97 Lat, and 60 Afr. Seven TERT variants resulting in missense changes met rare criteria and were present in seven subjects: all were heterozygous (7%: six AML, one MDS) (Table 1). The previously described variants were p.H412Y (MAF 0.32%, n = 1) and p.A1062T (MAF 1.25%, n = 4), the latter meeting inclusion criteria by prior reports of TBD association [6]: both variants are enriched in patients with TBDs [4, 6]. The remaining five variants were not cataloged in gnomAD and therefore classified as ‘previously-undescribed.’ Three of the five variants occurred in the same subject. Six out of 7 variants were confirmed germline by sequencing DNA from remission marrow (n = 5) or from non-transformed cells flow-sorted from a diagnostic sample (n = 1). We were not able to confirm germline status in two of the three cases with p.A1062T nor in the case with p.V435E due to lack of sample (Table 1). None of the controls had a previously-undescribed TERT variant, compared with three of the AML/MDS cases (p = 0.02). Five rare variants were detected in 10 out of 254 controls, 6 with the same variant, p.A1062T (2%) (Supplementary Table 1).

Table 1.

TERT variants present in an AML/MDS cohort (n = 94) and associated telomerase activity, processivity, constitutional lymphocyte telomere length, and number/type of features of dyskeratosis congenita.

Case ID	Dx	Race/ethnicity	TERT missense changea	Included in gnomADb	dbSNP ID	Confirmed germline	Reported activity/processivity	Telomerase activity	Telomerase processivity	LTL at diagnosis	LTL interpretationc	DC features
1333	AML	Eur	P308Q	N	None	Y	N/A	Similar to WT	Similar to WT	Sample not available	N/A	None: early death from progressive disease
			V713E	N	None	Y	N/A	0.05 ±0.01	Not quantified
			F928L	N	None	Y	N/A	0.49 ± 0.06	0.57 ± 0.04
260	MDS	U	V435E	N	rsl561213530	Not possible (MDS)	N/A	Similar to WT	Similar to WT	Sample not available	N/A	Data not available
1413	AML	Lat	A1062T	Y	rs35719940	Y	Normal to mildly reduced (12, 36, 37)	0.90 ± 0.08	Similar to WT	5.3 kb	Very low	None
242	AML	Lat	A1062T	Y	rs35719940	Sample not available				Sample not available	N/A	Two: pulm. disease, skin/nail changes
453	AML	U	A1062T	Y	rs35719940	Sample not available				Sample not available	N/A	Data not available
652	AML	Eur	H412Y	Y	rs34094720	Y	Normal to reduced (35–37)	Similar to WT	Similar to WT	2.8 kb	Very low	Two: tx delays, prolonged cytopenias
1488	AML	Lat	P404R	N	rs1060503008	Y	N/A	Similar to WT	Similar to WT	9.4 kb	Normal	Two: pulm. disease, prolonged cytopenias

AML acute myeloid leukemia, MDS myelodysplastic syndrome, DC dyskeratosis congenita, LTL lymphocyte telomere length, WT wild type, Eur European, U Unknown, Lat Latino.

^aBased on TERT isoform NP_937983.2.

^bgnomAD searched on October 8, 2019.

^cNormal: 10th–90th percentile for age; very low: <1 st percentile for age.

Telomerase in vitro specific activity (ability of the telomerase enzyme to extend the DNA primer) and processivity (ability of telomerase to add multiple telomere repeats after a single binding event) was assayed from the immunoprecipitated telomerase complex after transfection with WT or mutant constructs. The deleterious missense control was TERT p.K570N, documented to have markedly reduced telomerase activity and processivity [7]. Mutant proteins that contained the previously-undescribed variants p.P308Q, p.P404R, and p.V435E, and the rare variant p.H412Y had specific activity and processivity that was similar to WT (Fig. 1a, b, e). Two previously-undescribed variants identified in the same patient were each individually associated with <50% telomerase activity: p.V713E and p.F928L (Fig. 1a, b). The p.A1062T variant found in both cases and controls conferred slightly (but not significantly) reduced activity (p = 0.894), consistent with most previous studies [6, 8, 9].

Fig. 1. Telomerase activity and processivity determined from the direct primer extension assay for all TERT mutants, compared with WT and K570N as controls.

a Direct primer extension by WT and mutant telomerase of 20 nM of an 18-nt substrate primer [(TTAGGG)3], with a 12-mer loading control (LC). Each band represents the addition of a hexamer repeat to the 18-nt DNA substrate, with the number of nucleotides added shown on the left. The WT lane on the left shows normal activity and processivity. b Comparative telomerase activity between WT and mutant telomerase. Error bars represent the standard error around the mean (n = 3–4). WT, wild type. ****p < 0.0001. c K_M calculated from the direct assay (V713E and K570N, n = 3, see Supplementary Fig. 3) or by Telospot (all others, n = 3, see Supplementary Fig. 2 and Supplementary Methods). The intensity of extension products was summed and normalized against the intensity of the ³²P-labeled 12-nt LC to give relative activities, then expressed as a percentage of the reaction with maximal activity and plotted against DNA (substrate) concentration. Curves were fitted to the equation y = [V_max(S)/[K_M + (S)]. ****p < 0.0001. d Quantitation of telomerase processivity (see Supplementary Methods). The natural log of 1 minus the ‘fraction left behind’ plotted against repeat number and fitted to a linear regression equation, with the slope inversely proportional to processivity. e Summary plot showing final processivity values calculated from three separate experiments for each mutant, ****p < 0.0001.

To determine the mechanism by which p.V713E and p.F928L result in reduced enzyme activity, we measured the Michaelis constant (K_M) for all mutant proteins, as a measure of DNA affinity (Fig. 1c). The K_M for proteins containing p.V713E and the p.K570N deleterious control were significantly greater than WT (~8-fold and ~60-fold increase, respectively), providing evidence for decreased DNA binding affinity. Given that the DNA concentration used in initial activity assays would be sub-saturating, these assays were repeated at 10-fold higher DNA concentration. The activity of p.K570N was partially rescued, whereas that of p.V713E was not rescued (Supplementary Fig. 4), implying defective DNA binding as the primary mechanism underlying reduced activity for the p.K570N mutant but suggesting other contributing factors to p.V713E reduced activity. The p.F928L mutant protein was less processive than WT (Fig. 1d, e), and both p.V713E and p.K570N proteins had substantially reduced processivity that was not quantifiable due to their exceptionally low activity. Processivity for the remaining mutants was similar to WT (Fig. 1d, e).

Telomerase activity and processivity that are similar to WT do not exclude the possibility of telomere maintenance defects resulting from impaired binding of telomerase-associated proteins or hTR. Therefore, for variants with normal to borderline-low in vitro telomerase activity and processivity (i.e. four variants occurring in six patients), we determined germline LTL in nonmalignant cells sorted from the AML diagnostic sample. Two cases had very low LTL that was < 1st percentile for age (one each with p.H412Y and p.A1062T) (Supplementary Fig. 5). The case with p.P404R had normal LTL (between the 10th and 90th percentile for age). The other two p.A1062T cases and p.V435E case had no viable cells available for testing (Table 1).

Out of 84 pediatric AML cases, 71 survived beyond the first week after diagnosis and had clinical information available for review. Of the five cases with a rare or previously-undescribed variant, two had no DC-like features and three had ≥2 DC-like features that were evident pre-SCT (Table 1). Although 10 cases without a TERT variant had a single DC-like feature, cases with a rare or previously-undescribed TERT variant were more likely to have ≥2 DC-like features than cases without a variant (3/5 vs. 4/66, p = 0.006). Presence of a variant did not impact 5-year EFS or OS (EFS p = 0.87, OS p = 0.98), nor did presence of ≥2 features of DC (EFS p = 0.84, OS p = 0.79).

In this pediatric AML/MDS cohort, 7% of cases had a rare or previously-undescribed TERT variant, consistent with prior reports of a 5–10% frequency of rare TERT variants in adults with hematologic malignancies but differing from a prior report in a similarly-sized pediatric AML cohort [10]. Reasons for this discrepancy are unclear, but likely represent expected variation of a relatively rare event. Previously-undescribed TERT variants were more common in cases than controls (p = 0.02). Cases with a rare or previously-undescribed variant were more likely to have two or more clinical features of DC than cases without a variant (p = 0.006). Although significantly reduced telomerase activity and processivity was noted in only two variants tested, four out of seven cases with a variant had age-adjusted LTL < 1st percentile and/or ≥2 DC features (Table 1), suggesting that defects in telomere maintenance may underlie a proportion of AML/MDS diagnoses in children. The presence of rare missense TERT variants in control populations emphasizes the importance of applying functional testing to demonstrate further evidence of a deleterious effect [11], an approach critical to characterizing rare variants found in a disease context.

Both TERT p.V713E and p.F928L lie within the highly-conserved TERT reverse transcriptase domain, a region critical to telomerase function [12], so that reduced telomerase catalytic activity associated with these variants is not surprising. Neither TERT p.A1062T nor p.H412Y are noted as pathogenic in ClinVar and had essentially normal telomerase activity, but both were associated with very low age-adjusted LTL, suggestive of a concurrent mutation in an untested gene or a defect that does not rely on telomerase catalytic function, such as those that affect telomerase recruitment to telomeres [13]. For example, the very low LTL observed in the case with p.H412Y, which lies in the TERT RNA binding domain, may be explained by reduced binding of hTR to TERT and would not be detected by our assays.

Given that individuals with DC are at a 200-fold increased risk for AML and a ~2000-fold increased risk for MDS [2], a high index of suspicion should be maintained for this rare disorder, recognizing that clinical features of DC may be subtle and that TBDs may manifest as an isolated phenotype (e.g., AML) [1]. Recognition of TBDs prior to AML treatment initiation and particularly SCT is critical due to the associated risk for therapy-related toxicities and recommendation for reduced SCT conditioning in such cases [14, 15]. Children diagnosed with AML/MDS should be evaluated for features of DC, and those who experience excessive toxicities during treatment should be screened for TBDs by assessing LTL and/or the presence of pathogenic mutations in telomerase and related genes.