Variability in newborn telomere length is explained by inheritance and intrauterine environment

Telomeres are nucleoprotein structures formed of non-coding tandem repeats of hexamer TTAGGG at the end of the chromosomes. Functionally, telomeres play an important role in maintaining genomic stability and protect chromosomes from end-to-end fusion and degradation [1]. Telomeres shorten with each cell division, reflecting the age of a cell and the time until senescence [1]. Telomere (TL) is a biomarker of biological aging and has been associated with physiological and environmental stress, age-related diseases, and early mortality [2, 3]. Sex differences have also been reported in multiple studies with females having longer TL than males [4]. Shortened leukocyte TL has been associated with smoking [5], type 2 diabetes [6], higher 2-h post-load glucose concentration [7], lower insulin-like growth factor 1 [8], and higher leptin [9] and homocysteine levels [10]. Multiple studies reported telomere shortening to be associated with hypertension [11], cardiovascular disease [12], and mental disorders [13]. Some studies reported diet [14], parity [15, 16], and educational attainment [17] to explain inter-individual variation in TL.

In addition to diet, lifestyle, and other environmental exposures, heritable factors such as genetic variants have also been recognized to play a significant role in determining an individual’s TL. Genome-wide association studies (GWAS) have been performed on TL in many adult studies, which have identified TL-linked genetic variants at multiple loci in the human genome: TERC (3q26.2) [18‐22], TERT (5q15.33) [20, 21, 23], RTEL1 (20q13.33) [21, 24], OBFC1 (10q24.3) [20, 21], NAF1 (4q32.2) [21], ZNF208 (19p12) [21], and ACYP2 (2p16.2) [21].

Newborn TL is the initial setting of TL and highly variable among individuals [25]. It has important implications on telomere dynamics and molecular longevity over the lifespan [26, 27]. Current advances in newborn TL research indicate plasticity in the programming of telomere biology and the initial setting of TL [28]. The in utero environment has been suggested to be a significant contributor to this effect. Cord blood TL association studies with specific in utero exposures such as maternal nutrient status (e.g., folate [29] and vitamin D [30]), smoking status [31], educational attainment [32], or metabolic [33‐35] and mental health [36] have provided valuable insights. Offspring TL has also been reported to associate with parental age and TL [37‐41]. However, as the existing studies have primarily focused on individual exposures, a comprehensive understanding of the magnitude of their independent effects on offspring TL is lacking in the field. Also, though leukocyte TL in newborn is known to differ by sex, the sexual dimorphism in the factors explaining newborn TL variation is underexplored.

To the best of our knowledge, this is the first study to investigate newborn TL variation in umbilical cord tissue and report a comprehensive analysis of the effects of heritable factors, socioeconomic status, antenatal maternal health, and nutrition on this variation. It is known that TL is maximal at birth and decreases progressively with advancing age, and thus is considered a marker of biological aging [26]. In our study, we found relative average TL of newborns to be longer than the length of TL observed in their mothers (average maternal age, 31 years old). Maternal TL negatively correlated with maternal age ranging from 18 to 46 years. Newborn TL was not associated with gestational age but showed positive association with parental age. Comparison of the paternal and maternal age effects in the same regression model showed paternal age to have a dominant effect. This finding suggests that offsprings of older mothers could have longer TL in the analysis simply because the offsprings’ fathers were also older. Supporting this hypothesis, we indeed found a strong correlation between the parents’ age in our cohort. Previous studies [37‐39] have reported increase in sperm TL with age as a potential reason for offsprings of older fathers to inherit longer telomeres. As oocytes are produced prenatally, while sperm are continually produced throughout life, it is believed that there is greater potential for TL plasticity with age in sperms than in oocytes. Effect of paternal age on newborn TL is intriguing as it potentiates a scenario of intergenerational genetic plasticity in which the DNA passed on to the offspring is systematically changed based upon the reproductive age of one’s father.

Genetic variants are additional factors known to have heritable effects on TL. The genome-wide association analysis in this study identified a LD block within the 3q26.2 region to be significantly associated with both maternal TL and newborn TL. LRRC34-rs10936600 was the top variant in this LD block. By using LDproxy and CHB population data [https://​analysistools.​nci.​nih.​gov/​LDlink], we were able to extract the full list of genetic variants in a strong LD (R² > 0.85) with rs10936600 (Additional file 1: Table S9). This analysis identified TERC variant rs2293607 (R² = 0.98) to also be a member of this LD block. TERC encodes a long non-coding RNA found in eukaryotes that is a component of telomerase, the enzyme used to extend telomeres. The T allele of this variant (correlated to A allele in rs10936600) has been reported to be associated with an increase in TERC expression and TL [56]. Our finding is consistent with this reported trend as the dosage of A allele in rs10936600 is positively associated with TL (Fig. S4A). Further, the association results of previously reported genetic variants (TERT, RTEL1, OBFC1, NAF1, ZNF208, and ACYP2) are shown in Additional file 1: Table S10. They showed a weak association with TL in our study. Among them, TERT-rs2853677 (telomerase reverse transcriptase) showed the most significant association with maternal TL (β = 0.18, P = 1.85E−4).

For the previously reported factors associated with cord blood TL [25‐36], we similarly found significant associations for sex, parental age, maternal TL, antenatal stress, maternal educational attainment, maternal smoking during pregnancy, and genomic region at 3q26.2, but the associations with maternal pre-pregnancy BMI, GDM status, hypertension, plasma folate, and vitamin D concentrations were not found in our study. As our pilot study (the “Methods” section) showed cord blood and cord tissue TLs were highly correlated (R² = 0.64, Fig. S10), it is not surprising that we find similar associations with many factors. However, the discrepancy between our findings with the previous reports could be due to the sample size, population effects, and tissue-specific differences. In our study, we identified five new significant factors influencing newborn TL: ethnicity (Asian population), plasma fasting glucose concentration, plasma IGFBP3 level, plasma DGLA%, and vitamin B12 level (Table 2).

Sex-specific analysis identified distinct male vs female effects of factors influencing newborn TL. Interestingly, heritable factors such as maternal TL and newborn LRRC34-rs10936600 showed significant effects on both male and female newborn TLs with different effect sizes. LRRC34-rs10936600 showed a stronger effect on male newborns TL, while maternal TL had more pronounced effects on the female newborn TL. For other influencing factors, sex-specific effects were also observed. Female newborn TLs were more susceptible to the variation in maternal mental health (depression/anxiety) and vitamin B12 levels, while male newborn TLs were strongly affected by the variation of paternal age, maternal educational attainment, plasma fasting glucose concentration, plasma DGLA%, and IGFBP3 level. Although the negative impact of antenatal maternal distress on newborn TL has been widely reported, this study is the first to note that anxiety scores (STAI state and trait) have much stronger effects than depression scores (EPDS) on TL, and all scores exhibit female-specific effects. Low vitamin B12 level was linked to negative impacts on cognitive, motor, and growth outcomes for fetal development and was related to depression in mothers [57]. Interestingly, vitamin B12 level showed similar female-specific effects as did antenatal depression/anxiety. The effect of paternal age on offspring’s TL was linked to telomere elongation in sperm from older men as reported previously [39]. It is interesting to observe the male-specific effect of paternal age on newborn TL. For maternal educational attainment, the male-specific effect is consistent with a recent report that highlighted the male-specific effect of parental SES on newborn TL [58]. Impaired glucose metabolism and the related hormonal imbalance may generate physiological stress for the growing fetus [59] and may lead to TL attrition, especially for the male offspring. IGFBP3 is the main insulin-like growth factor transport protein in the bloodstream and plays an important role in senescence as an aging marker [60]. We observed a negative effect on male newborn TL. For all the TL-associated factors, we identified eight independent predictors of newborn TL (Table 3). These include sex, ethnicity, newborn LRRC34-rs10936600, maternal TL, paternal age, antenatal anxiety, plasma fasting glucose, and IGFBP3. The contributions of these factors influencing the variation in newborn TL were compared and ranked in three best multivariate models using all, female-only, and male-only subjects.

In the association study between maternal TL and antenatal maternal health, our results were similar to the findings from previous studies on TL associations with age [1], diabetes [6], 2-h post-load glucose concentration [7], educational attainment [17], and smoking status during pregnancy [5]. Although n-3 PUFA supplementation has been reported to affect TL [14], our study for the first time showed that higher plasma n-3 PUFA% and lower n-6:n-3 PUFA ratio are associated with longer maternal TLs. In addition, previous reports showed telomere shortening was associated with cardiovascular disease [12]. Elevated plasminogen activator inhibitor-1 (PAI-1) is a risk factor for thrombosis and atherosclerosis and associated with major adverse cardiovascular events (MACE) [61]. Notably, our study showed that shorter maternal TL was associated with higher plasma PAI-1 level. Finally, plasma IGF2 levels were positively associated with maternal TL. Higher circulating levels of IGF1, a related protein of IGF2, are known to be associated with longer TL in healthy subjects [62]. IGFs are known to play an essential role in the pathogenesis of several age-related diseases, including dementia, cardiovascular, and metabolic diseases; hence, their levels in circulation play a significant role in predicting healthspan and biological aging. The association of maternal TL with plasma PAI-1 and IGF2 levels should be interpreted with caution as their associations lost significance in the sensitivity analysis.

While comparing the factors influencing the newborn and maternal TLs, we found some consistent trends. For example, higher maternal educational attainment showed a positive association, and smoking during pregnancy showed a negative association with TL in both the mother and the offspring. The mediation analysis showed these two factors affected newborn TL through the mediating effect of maternal TL (Fig. S11). We also observed both mother and newborn TL to be associated with antenatal glycemia. As fasting and 2-h post-load glucose concentrations have different underlying etiologies and pathophysiologies, different effects on newborn/maternal TL are expected. It was noteworthy that maternal GDM or higher 2-h post-load glucose concentration was significantly associated with reduced TL only in mothers, while fasting glucose concentration had an adverse impact only on offsprings. Impaired glucose tolerance has been previously shown to associate with impaired telomerase activity [63] which may explain the shortening of TL in mothers. Although both impaired glucose tolerance and impaired fasting glucose can be harmful to the growing fetus [59], our findings showed that the latter had a more detrimental effect on fetal TL. Finally, we found that the association between antenatal depression/anxiety and TL was observed in the offspring but not in the mothers. This may be attributed to a high vulnerability of the fetus to stress during intrauterine development.

Our study has some limitations. First, the GUSTO cohort had a disproportionate number of participants within the Chinese, Malay, and Indian groups. Therefore, analyses of ethnic-specific effects of TL associated factors were limited by sample size. Second, due to the absence of paternal TL data in the cohort, we were unable to compare it with the effects of maternal TL on newborn TL. Finally, pre-pregnancy maternal weight, smoking status and alcohol consumption before/during pregnancy, and SES (maternal education and household income) were drawn from self-reported questionnaires. Bias may exist in the self-reported variables. Pre-pregnancy weight was assessed to be reliable as it highly correlated with the maternal weight at the first-trimester visit. In order to avoid underreported smoking, maternal smoking status during pregnancy was determined by the combination results of self-reported questionnaires and plasma cotinine level. As the DNA from maternal blood was extracted using two different methods (manual and automated), this technical variation could have impacted the TL measures. Hence, we adjusted the analysis models for this technical variation.

Telomeres serve as the biological timekeepers of cellular health, and hence, their attrition relative to chronological age is an indication of advanced biological aging, existence of a stressed environment, and potential risk to disease. Our study found antenatal maternal health to be a crucial determinant of TL programming in utero, which could potentially impact subsequent offspring health outcomes over the life span, including aging and longevity. Hence, improving antenatal health of mothers by targeting modifiable factors can help prevent in utero telomere attrition and enhance cellular longevity. As evident from the findings of this study, smoking during pregnancy, maternal nutrient insufficiency (i.e., vitamin B12), and cardiometabolic and mental health adversities are suboptimal conditions for fetal programming of telomere biology. In addition, our findings in sex-specific associations suggest different focus of antenatal care may be required for mothers carrying babies of different sex.

We found evidence that the genetic variants at 3q26.2, paternal age, maternal TL, and antenatal maternal health have a significant impact on newborn TL. The comparative analysis of these factors identified the differences in the magnitude of their effects. Sex stratification analysis provided new insights into the factors explaining the male vs female TL variation. We also found that mother’s TL was significantly associated with her own metabolic health and nutrient status, which may have transgenerational effects on offspring TL. Our findings provide a comprehensive understanding of the heritable and environmental factors and their relative contributions to the initial setting of TL and programing of longevity in early life.