Reduced telomere length in amniocytes: an early biomarker of abnormal fetal development?

Abstract

Telomeres protect chromosome ends and control cell division and senescence. During organogenesis, telomeres need to be long enough to ensure the cell proliferation necessary at this stage of development. Previous studies have shown that telomere shortening is associated with growth retardation and congenital malformations. However, these studies were performed in newborns or postnatally, and data on telomere length (TL) during the prenatal period are still very limited. We measured TL using quantitative PCR in amniotic fluid (AF) and chorionic villi (CV) samples from 69 control fetuses with normal ultrasound (52 AF and 17 CV) and 213 fetuses (165 AF and 48 CV) with intrauterine growth retardation (IUGR) or congenital malformations diagnosed by ultrasound. The samples were collected by amniocentesis at the gestational age (GA) of 25.0 ± 5.4 weeks and by CV biopsy at 18.1 ± 6.3 weeks. In neither sample type was TL influenced by GA or fetal sex. In AF, a comparison of abnormal versus normal fetuses showed a significant telomere shortening in cases of IUGR (reduction of 34%, P < 10⁻⁶), single (29%, P < 10⁻⁶) and multiple (44%, P < 10⁻⁶) malformations. Similar TL shortening was also observed in CV from abnormal fetuses but to a lesser extent (25%, P = 0.0002; 18%, P = 0.016; 20%, P = 0.004, respectively). Telomere shortening was more pronounced in cases of multiple congenital anomalies than in fetuses with a single malformation, suggesting a correlation between TL and the severity of fetal phenotype. Thus, TL measurement in fetal samples during pregnancy could provide a novel predictive marker of pathological development.

Introduction

Telomeres are composed of tandem repeat sequences located at the ends of chromosomes and are required to maintain genomic stability and to control cell division and senescence. Owing to the ‘end-replication problem’, a portion of telomeric DNA is lost with each cell division. As soon as the length of telomeres becomes critically short, cells stop dividing and enter apoptosis (1). Thus, telomere shortening limits the proliferative capacity of most somatic cells. In postnatal life, telomere shortening is a physiological process associated with ageing. Germ and stem cells maintain their capacity to divide by regenerating telomeres via the action of an enzyme called telomerase, which is able to elongate telomeres (2). At the blastocyst stage, a telomere length (TL) reset seems to play a crucial role in establishing longer telomeres in the embryo and several studies support the role of telomere assessment as a biomarker of prognostic value in assisted reproductive therapy (reviewed in Anifandis et al. 3). In the germline, during early development and in highly proliferative organs, human telomeres are balanced between the loss of telomeric DNA with each cell division and its elongation by telomerase, but once cells are terminally differentiated or mature the equilibrium is shifted to gradual telomere shortening by repression of the telomerase enzyme.

Recent evidence suggests that during fetal life telomeres may shorten in cord blood leukocytes because of an adverse environment in utero (4). This reduction in TL could produce pathological effects during fetal life. Normal growth and organogenesis require a balance between cell proliferation and programmed cell death. Alterations in this balance could be mediated by abnormal telomere homeostasis and lead to growth retardation and congenital malformations. The maintenance of TL during early embryonal development is critically important for successful pregnancy outcomes in human and other mammalian species (5). Studies have shown the presence of shortened telomeres in cases of intrauterine growth retardation (IUGR) or congenital malformations (6–9). In these studies, TL was assessed in the placenta at birth (6–8) or from peripheral blood leukocytes in the postnatal period (9).

Data on fetal TL during pregnancy are scarce. Only four studies have reported TL measurements in prenatal samples. Mosquera (10) showed that continuous growth of human amniotic fluid (AF) cells in culture leads to telomere shortening and replicative senescence. Two studies measured TL in pathological contexts. Compared to control values, TL was reduced in cultured placental villi from pregnancies complicated by severe IUGR (8) and in cultured amniocytes from trisomy 21 conceptions (11). Of note, TL in newborns with aneuploidy is conflicting (12–15). Toutain et al. (16) observed that cell culture of placental villi produced a non-normal TL distribution and recommended estimating TL in uncultured cells. There are no reports of prenatal samples being taken from malformed fetuses. Studies of telomere biology during the prenatal period in cases of normal and abnormal fetal development could help to fill this knowledge gap.

Human chorionic villi (CV) and AF are used in prenatal diagnosis to identify a variety of genetic and developmental disorders of the fetus. Our aim was to investigate a potential correlation between abnormal embryonic or fetal development and telomere shortening in CV and AF samples. The resulting study is the first to assess TL in malformed fetuses during the prenatal period. To avoid confounding factors, we focused on fetuses without chromosomal abnormalities and known maternal and pregnancy characteristics that lead to abnormalities of fetal development.

Results

TL in AF and CV samples

Of the 282 samples included in this study, 217 were from AF and 65 were from CV. The absolute TL in kb per diploid genome was determined for each sample using quantitative DNA-PCR (see Materials and Methods). Overall, telomeres were significantly longer in AF than in CV samples from both control (P < 10⁻⁶) and pathological groups (P < 10⁻⁶) (Fig. 1 and Table 1). We also compared the mean TL between the two groups (Fig. 1). In CV, we observed a significant reduction of TL in pathological samples (P < 10⁻³). In AF, telomere shortening was even more pronounced in fetuses with pathological development (P < 10⁻⁶). The average decrease in TL in pathological fetuses was 21% in CV samples and 33% in AF samples. This is consistent with our initial hypothesis that TL is shortened in cases of abnormal development.

Figure 1

Individual and mean TL in control and pathological CV and AF samples. Telomeres in amniocytes were longer than in CV cells. Pathological fetuses had significantly reduced TL compared to respective control cases.

Table 1

Clinical data and TL in study samples

Group	Sample type	Number of samples	Maternal age (years)	GA (weeks)	Sex M/F	TL (kb/genome)
Control	AF	52	34.5 ± 5.8	20.5 ± 5.1	26/26	1751 ± 286
	CV	17	35.3 ± 5.6	14.5 ± 1.0	7/10	1100 ± 135
Intra uterine growth retardation	AF	53	30.9 ± 6.5	27.7 ± 4.2	17/35	1157 ± 316
	CV	18	33.0 ± 7.3	25.4 ± 3.7	16/12	826 ± 222
Single congenital anomaly	AF	86	31.4 ± 5.2	25.4 ± 4.4	50/36	1244 ± 282
	CV	12	33.1 ± 6.6	17.8 ± 7.4	6/6	897 ± 271
Cardiac malformation	AF	43	31.3 ± 5.4	26.2 ± 4.1	25/18	1267 ± 274
	CV	3	36.7 ± 5.4	25.8 ± 10.2	2/1	947 ± 275
Diaphragmatic hernia	AF	13	32.0 ± 3.4	24.0 ± 3.9	9/4	1324 ± 309
	CV	4	29.8 ± 6.0	13.8 ± 0.4	2/2	1041 ± 80
Spina bifida	AF	10	31.0 ± 4.8	23.3 ± 1.5	4/6	1224 ± 213
	CV	/	/	/	/	/
Brain malformation	AF	7	29.3 ± 6.5	25.4 ± 3.6	2/5	1039 ± 244
	CV	2	35.5	18.5	0/2	816 ± 15
Limb malformation	AF	8	32.5 ± 5.9	22.1 ± 3.8	6/2	1118 ± 209
	CV	3	32.3 ± 7.7	14.8 ± 1.4	2/1	708 ± 374
Cleft lip and palate	AF	5	31.0 ± 4.2	29.6 ± 3.6	4/1	1373 ± 323
	CV	/	/	/	/	/
Multiple congenital anomalies	AF	26	32.0 ± 4.0	25.4 ± 5.1	14/12	979 ± 216
	CV	18	28.5 ± 5.0	14.5 ± 3.4	7/11	885 ± 247
Pathological (total)	AF	165	31.3 ± 5.0	26.2 ± 4.6	81/83	1174 ± 299
	CV	48	31.3 ± 6.7	19.4 ± 6.8	19/26	866 ± 246

Values of maternal age, GA and TL are mean ± standard deviation.

M, male; F, female.

Different single congenital anomalies are given in italic.

TL according to maternal age, GA and fetal sex

Fetal TL was not influenced by maternal age. Linear regression analysis showed that TL was not correlated with maternal age in control AF (r = 0.21, P = 0.13 NS) and CV (r = −0.08, P = 0.60 NS) samples. Similarly, TL was not associated with maternal age in pathological AF (r = 0.09, P = 0.21 NS) and CV (r = −0.23, P = 0.38 NS) samples.

We analyzed whether TL could be influenced by variations in GA and by fetal sex. Prenatal samples were performed at GA of 18.1 ± 6.3 weeks for CV and 25.0 ± 5.4 weeks for AF. CV sampling can be performed after 11 weeks of gestation, but AF sampling is only possible from 15 weeks of gestation. We divided GA into several stages to cover the period of CV and AF sampling according to the term of pregnancy (Fig. 2). There was no significant variation in TL according to GA in CV and AF samples in either the control group (P = 0.32 and P = 0.26, respectively) or the pathological group (P = 0.32 and P = 0.85, respectively).

Figure 2

TL in CV (A) and AF (B) samples according to GA. The mean TL was not related to GA in control and pathological samples.

The sex ratio M/F was well balanced in control (26/26) and pathological (81/83) AF samples. In CV samples, there was a slight excess of female fetuses in both the control (7/10) and pathological (19/29) groups. No significant differences in TL were observed between male and female fetuses in control CV and AF samples (P = 0.82 and P = 0.45, respectively) or in pathological samples (P = 0.65 and P = 0.40, respectively) (Fig. 3).

Figure 3

TL in CV (A) and AF (B) samples according to sex. The mean TL was not different in male (M) and female (F) control and pathological samples.

TL in fetuses with growth delay and congenital anomalies

Intrauterine growth retardation

TL in cases of IUGR has been previously studied in chorionic and placental villi, but AF cells have not yet been analyzed. We found that TL was significantly shorter in IUGR than in CV samples from control cases (P < 0.001, Fig. 4A). Of note, TL in AF cells from IUGR cases was also significantly shorter than in control fetuses (P < 10⁻⁷, Fig. 4B).

Figure 4

TL in control fetuses and in fetuses with growth retardation and different congenital anomalies in CV (A) and AF (B) samples. In CV, TL was significantly shorter in fetuses with IUGR, isolated BM and LM, and MCAs than in controls. No telomere shortening was observed in fetuses with CM and DH. In AF samples, telomeres were significantly shorter in cases of IUGR, single malformations including SB and cleft lip and palate (CLP) and in multiple malformations than in controls.

Single congenital anomaly

In CV samples, fetuses with single congenital anomalies had significantly shorter telomeres than controls (Table 1, P = 0.015). Considering the type of single anomaly, no significant difference in TL was found between fetuses with isolated cardiac malformation (CM) or diaphragmatic hernia (DH) and controls (Fig. 4A). In fetuses with isolated brain or limb malformation (LM), TL was significantly reduced compared to that of controls (P < 0.05 and P < 0.01, respectively).

In AF samples, the mean TL of fetuses with a single congenital anomaly was shorter than the mean TL of control samples (Table 1, P < 10⁻⁶). A significant reduction in TL was observed for all anomaly types (Fig. 4B). The shortest telomeres were observed in amniocytes from fetuses with brain and LM (Table 1).

Multiple congenital anomalies

Fetuses with multiple congenital anomalies (MCAs) had significantly shorter telomeres in CV (P < 0.01) and AF (P < 10⁻⁷) samples than in the respective control groups. Interestingly, TL in CV samples did not differ between MCA and single malformation cases (885 ± 247 vs 897 ± 271, P = 0.91, NS). In AF samples, fetuses with MCAs had significantly shorter telomeres than those with isolated malformations (979 ± 216 vs 1244 ± 282, P < 10⁻⁴).

Discussion

Maintenance and lengthening of TL during early embryo development are very important for successful pregnancy outcomes in human and other mammalian species (5,17–19). Schaetzlein and Rudolph (20) showed that early mammalian embryos have a telomerase-dependent genetic program that elongates telomeres to a defined length, possibly required to ensure sufficient telomere reserves for viability of fetuses and responses to DNA damage during the developmental process. This telomere resetting takes place at the morula–blastocyst transition (21,22). Blastocysts with critically short telomeres could not implant (23). TL in CV samples from spontaneously eliminated product of conception at 5–12 weeks of GA is characterized by short telomeres in comparison to induced abortion samples, which also suggested a strong correlation between TL and the viability of embryos at this stage of development (24). These findings show that TL plays an important role in the development of the embryo and that an alteration of this elongation program could be the cause of congenital malformations.

Fetal development can be affected by genetic or environmental factors or both, but in most cases the etiology and pathogenesis of organ malformations and growth retardation are unknown (25). Congenital malformations have been associated with shorter telomeres in postnatal samples (9). Newborns and fetuses with IUGR also have shorter telomeres (6–8). However, fetal telomere homeostasis in humans is still largely understudied. Our objective was to evaluate TL in AF and CV samples from fetuses with normal and abnormal development.

We found that telomeres were significantly longer in AF than in CV cells in both control and pathological samples. To date, there are no documented reports comparing TL between CV and AF cells. CV and AF have different embryological origins and different rates of proliferation according to the stage of development. In CV that contain trophoblastic and mesenchymal core cells, proliferation and telomerase activity are high at the beginning of the first trimester of pregnancy and decrease as the pregnancy progresses (26–29). Of note, we studied CV samples obtained after 11 weeks of gestation. AF samples contain essential cells from the developing fetus, including stem cells with long telomeres and high telomerase activity (30,31), which could explain why TL was greater in AF samples.

We also investigated the potential impact of GA on TL and found no influence of GA in CV and AF samples after 11 weeks of gestation. The GA at which placental biopsy is performed does not appear to have an effect on TL (8). TL in the skin and umbilical arteries of newborns was not shorter than in the corresponding tissues of fetuses of 15–19 weeks gestation, suggesting that TL is maintained from 20 weeks of gestation to birth, which is consistent with our results (32).

We found no difference in TL between male and female fetuses. These results are in line with previous observations showing the absence of sex impact on TL at birth in the placenta, white blood cells, umbilical artery and foreskin (33–35). In contrast, a study of umbilical cord blood showed that telomeres were slightly but significantly longer in girls than in boys (36). In another study, male sex was associated with shorter placental telomeres (37). Thus, results are still contradictory and the potential effect of sex on TL in newborns warrants further research.

In our study, telomere shortening was observed in CV (25% shorter) and AF (34% shorter) samples from fetuses with IUGR. Previous studies showed that TL was significantly reduced in placental villi collected in the prenatal period or at delivery from pregnancies complicated by IUGR (6–8). Conversely, Wilson et al. (37) observed no significant difference in TL between IUGR and control placentas. These conflicting results can be partly explained by the use of different methods for sample processing (with or without enzymatic treatment, cultured or uncultured cells) and varying techniques of TL assessment. The TL of cord blood cells was previously found to be identical in IUGR and control pregnancies, whereas placental TL was reduced in fetuses with IUGR (6). Toutain et al. (8) hypothesized that TL would not be reduced in amniocytes during pregnancies complicated by an IUGR because it is a placental pathology. Our results run counter to this hypothesis. Telomere shortening could also be the cause of impaired fetal growth caused by insufficient cell proliferation. Further studies are needed to evaluate the usefulness of TL markers in prenatal samples from fetuses with IUGR.

We observed significant telomere shortening in CV samples from fetuses with brain and LMs and MCAs. Telomere shortening in AF samples was even more marked and was observed for all types of malformations included in our study. AF mostly contains epithelial cells, which originates from the epiblast prior to or at the time of primitive streak formation, and is more representative of fetal tissues than chorionic cells (31,38), suggesting that AF is a more suitable sample than CV for studies of the role of TL in fetal development.

An association between TL and an isolated congenital malformation was first reported by Mazumdar (9), who observed the presence of telomere shortening in peripheral blood mononuclear cells of patients with congenital limb anomaly. We found that the TL of fetuses with LM was reduced by 36% in both CV and AF samples. In AF samples, we also observed a TL reduction of 30% in fetuses with spina bifida (SB) and of 41% in fetuses with brain malformations (BMs). Interestingly, structural brain anomalies like cerebellar hypoplasia have been found in 57% of patients with telomere biology disorders (39). Furthermore, TL was reduced by 28% in AF samples from fetuses with isolated heart defects. Telomerase activity and telomeres are of great importance in many periods of heart ontogenesis (40). Telomere dysfunction was identified as a critical physiological signal for cardiomyocyte cell-cycle arrest and ample telomere reserves were shown to be essential for cardiomyocyte proliferation and cardiac regeneration in newborn mice (41).

MCAs are defined as the occurrence of two or more structural malformations in different organ systems not related to a known chromosomal or genetic syndrome. The prevalence of cases with MCAs was estimated to be 15.8 per 10 000 births in 19 population-based European registries [EUROCAT (42)]. Our results suggest that telomeres are shorter in fetuses with MCAs than in those with a single malformation. Thus, the degree of TL seems to be related to the severity of fetal phenotype and may therefore have a prognostic value. We hypothesize that the decrease in fetal TL below a certain threshold alters the replicative capacity and cell proliferation/death balance, thereby leading to defects in organs that require more cell divisions for their formation. The presence of critically short telomeres in several developing organs could result in multiple malformations, but the causality remains to be investigated in further experimental and mechanistic studies.

Genetic factors, maternal lifestyle and health status during pregnancy can accelerate telomere shortening in early life. In newborns, reduced TL has been associated with maternal smoking, stress, nutritional and sleeping disorders and diseases such as hypertension and diabetes (43–46). The mechanisms underlying telomere shortening during embryonic and fetal development are largely unknown, but oxidative stress could play a major role in the process. Oxidative stress induces single- and double-stranded breaks in telomeric DNA in vitro and in vivo and a consequent loss of telomeric sequences (47,48).

Pregnancy enhances oxidative stress because the higher metabolic demand of the growing fetal tissues contributes to the increased production of reactive oxygen species (49). Exposure to environmental and lifestyle factors can also contribute to oxidative stress. Interestingly, preconception TL shortening in women, associated with chronic exposure to oxidative stressors such as poor nutrition and lifestyle, appears to lead to an increased risk of SB in the offspring (50). Shorter TL was associated with the presence of type 2 diabetes, which can be partially attributable to high oxidative stress (51). During embryogenesis, a positive correlation was observed between hyperglycemia and the risk for congenital malformations in infants of diabetic mothers. Hyperglycemia increases the production of reactive oxygen species, which in turn generates oxidative stress conditions in the embryo (52,53). Although we excluded from our study fetuses of mothers with gestational diabetes, preeclampsia, heavy smoking and infection, milder but significant oxidative stress mediated by various endogenous and exogenous factors might have occurred in certain pathological cases in our series and led to telomere shortening and abnormal development. Figure 5 summarizes the potential impact of oxidative stress on TL shortening during pregnancy and the consequences of short telomeres for embryonic and fetal development.

Figure 5

Schematic presentation of the proposed mechanism for the role of telomere shortening in abnormal development.

Conclusion

Our results show that TL is significantly shortened in AF samples in case of abnormal development, irrespective of GA and sex, and that it appears to be correlated with the severity of the fetal phenotype and thus could be an early predictive marker of abnormal development. Our hypothesis is that an embryo with short telomeres cannot ensure the cell divisions necessary for the development of certain organs or can be more sensitive to environmental, metabolic or genetic factors, which results in the occurrence of malformations. Identifying the factors that contribute to telomere shortening during the embryonic period and the mechanistic pathways linking telomere shortening to abnormal development warrants further investigation. Uncovering these relationships is important because of their potential utility in early identification of risk groups among diverse fetal phenotypes.

Materials and Methods

Ethics statement

This retrospective monocentric study was performed on samples remaining after routine diagnostic tests. Written informed consent was obtained from patients to use the residual samples for research purposes. All samples and data were anonymized. The study was approved by the local Institutional Review Board (IRB00008526, CPP SUD-EST VI Clermont-Ferrand, 2021/CE17).

Sample collection

The AF and CV samples were obtained by obstetricians using their standard clinical procedures in the context of prenatal chromosomal diagnosis. The samples were taken between 11.9 and 37.9 weeks of gestation depending on the indication of the sampling, the feasibility of the procedure and the type of the sampling (amniocentesis can only be done from 16 weeks of pregnancy and choriocentesis from 11 weeks).

AF and VC samples were obtained from 69 control fetuses (52 AF and 17 CV) and 213 pathological fetuses (165 AF and 48 CV) diagnosed by abnormal antenatal ultrasound. The samples were collected at the time of fetal ultrasound diagnosis. Among 282 pregnancies, 279 came from natural conceptions and only 3 were obtained after in vitro fertilization. All fetuses had normal karyotype and array CGH (Agilent 180K, threshold 400 kb). Fetal female samples contaminated with maternal blood or decidua were excluded by short tandem repeats analysis. Control fetal samples were obtained from women who had normal ultrasound and undergone prenatal diagnosis for advanced maternal age, risk > 1/250 of having a child with trisomy 21 (after maternal serum screening ± nuchal translucency measurement) or with a family history of chromosome abnormalities.

Of the 213 fetuses with abnormal development, 98 had a single congenital malformation (single organ defect) and 44 had MCAs, defined as two or more anomalies involving distinct organs or systems. The remaining 71 fetuses had severe IUGR with fetal biometrics less than the third percentile and no cytomegalovirus infection, congenital malformation, preeclampsia or heavy maternal smoking. Characteristics of the samples are given in Table 1.

Exclusion criteria were maternal disease (diabetes, high blood pressure, infection), multifetal pregnancy and fetal chromosome abnormality. Samples from pregnancies with mechanical factors such as amniotic bands, anhydramnios or severe oligohydramnios were also excluded.

The samples were collected, prepared and frozen with the same procedures over a short period of time. DNA extractions and quantitative PCR (qPCR) were also performed with identical methods for all samples.

Measurement of TL using qPCR

DNA was extracted directly from AF and CV after enzymatic digestion. The aim of enzymatic treatment by trypsin and collagenase was to remove the trophoblast and enhance the proportion of mesenchymal core cells. DNA isolation was performed using a Maxwell® 16 Instrument according to the manufacturer’s instructions (Promega, Charbonnières-les-Bains, France). DNA purity and concentration were assessed with a NanoDrop 1000 spectrophotometer (ThermoScientific, Wilmington, DE, USA).

TL was measured by qPCR in a LightCycler 480 System (Roche Diagnostics, Meylan, France) using SYBR Green I technology (SYBR Green Kit, Roche Diagnostics), as described elsewhere (54,55). Briefly, two PCRs were performed in triplicate to measure the template amounts of telomere repeats and a reference single-copy gene (glyceraldehyde-3-phosphate dehydrogenase, GAPDH). Twenty nanograms of DNA were used in each reaction. Oligomer standards were introduced in each PCR run to determine the total TL in kb and the diploid genome copy number for each sample. Serial dilutions of the telomeric standard containing synthetic (TTAGGG)₁₇ DNA in the range from 10⁷ to 10³ kb were used to generate a standard curve to measure the total content of telomeric sequences per sample in kb. The number of diploid genome copies per sample was measured using a standard curve obtained with serial dilutions of the synthetic GAPDH gene standard (97 pb) containing from 10⁶ to 10² copies. The absolute TL in kb per diploid genome was then calculated for each sample by dividing the telomere kb value by the number of diploid genome copies.

Statistical analysis

We assessed the normality of the TL distribution using the Kolmogorov–Smirnov test. The datasets followed a normal distribution. Continuous variables were presented as mean ± standard deviation. The comparisons of the means were performed using the t-test. A two-sided difference in TL was expected to be one-third of the standard deviation with α set at 5% (false positive error) and power at 95% (false negative error β = 0.05). P < 0.05 was considered to be statistically significant. In all significant comparisons, the difference in TL was greater than one-third of the standard deviation (ranged from 0.44 to 2.9) and the number of cases was sufficient to reach the power of 99.9%.

Acknowledgements

We are grateful to the patients for their consent and to Farida Godeau and Delphine Voisin for technical support.

Conflict of Interest statement. The authors declare no conflict of interest.