Telomeres and replicative cellular aging of the human placenta and chorioamniotic membranes

Recent hypotheses propose that the human placenta and chorioamniotic membranes (CAMs) experience telomere length (TL)-mediated senescence. These hypotheses are based on mean TL (mTL) measurements, but replicative senescence is triggered by short and dysfunctional telomeres, not mTL. We measured short telomeres by a vanguard method, the Telomere shortest length assay, and telomere-dysfunction-induced DNA damage foci (TIF) in placentas and CAMs between 18-week gestation and at full-term. Both the placenta and CAMs showed a buildup of short telomeres and TIFs, but not shortening of mTL from 18-weeks to full-term. In the placenta, TIFs correlated with short telomeres but not mTL. CAMs of preterm birth pregnancies with intra-amniotic infection showed shorter mTL and increased proportions of short telomeres. We conclude that the placenta and probably the CAMs undergo TL-mediated replicative aging. Further research is warranted whether TL-mediated replicative aging plays a role in all preterm births.


Results
Characteristics of TL parameters measured by SB and TeSLA, and telomere dysfunction measured by TIF. The principle measurements employed in this work (SB, TeSLA and TIF) provide different but vital information about the potential role of TL-mediated replicative aging in the biology of the human placenta and CAMs. Although the SB method is the 'gold standard' of TL measurements against which the validity of all other techniques have been judged 11 , it was originally designed to measure mTL, i.e., SBmTL, which reflects the mean length of the 92 telomeres of the q and p arms of the 23 human chromosome pairs. When used in population-based telomere research and in this study (Fig. S1a), SB typically generates a weak TRF signal below 3 kb; the analysis is applied, therefore, to a range of 3-20 kb 11 . TeSLA, in contrast, generates a signal of TRFs for single telomeres that extends below 3 kb (Fig. S1b) 12 . Although the two methodologies of TL measurements inter-relate, they do not produce identical information, because the TeSmTL includes telomeres shorter than 3 kb (Fig. S2).
In the final analysis, the buildup of short telomeres is expressed in telomere dysfunction, which is captured by TIF [7][8][9][10] . We found it essential, therefore, to examine the relation in the placenta and CAMs between TL parameters and TIF. The key for this analysis is the simultaneous signals of DNA damage response (DDR), as detected using antibodies against p53 binding protein 1 (53BP1), and telomeres labeled with a Cy3-conjugated peptide nucleic acid (PNA) complementary to telomeric repeats (illustrated in Fig. S3).
SBmTL differed between CAMs, being longer in AM than the CM ( Fig. 1; t = 4.29, P = 0.0009). It was shorter in the CM than in the placenta (t = 5.25, P < 0.0001), and did not differ from SBmTL in UCB (t = 1.97, P = 0.30). In contrast, SBmTL in the AM did not differ from SBmTL in placental samples (t = 0.77, P = 0.94), or in UCB (t = 2.52, P = 0.11). An overview of all pair-wise comparisons is provided in Table S1A.
TeSmTL at full term also varied significantly between tissues (mixed model, including subject as random effect: tissue: F 4,48.0 = 17.07, P < 0.0001; Fig. 1). This was due to TeSmTL being longer in UCB compared to all other tissues (all t > 4.7, P < 0.003), while pair-wise comparisons revealed no significant differences in TeSmTL between the other tissues (all t > 2.47, P > 0.11). A comparison between tissues of the proportion of telomeres shorter than 3.0 kb yielded the same result, with significant variation between tissues (F 4,48 = 19.98, P < 0.0001) which was due to fewer short telomeres in UCB compared with all other tissues (all t > 5.1, P < 0.0002), and no significant differences between the other tissues (all t < 2.77). An overview of all pair-wise comparisons is  www.nature.com/scientificreports/ provided in Table S1B. There were modest inter-individual differences in TeSmTL across tissues (ICC individual identity: 0.21, SE = 0.13). Placental findings above are for samples obtained from the central placenta. Previous work found no SBmTL difference between the central and peripheral placenta 3 , but we found a trend for SBmTL to be longer in central placenta samples compared to peripheral samples ( Fig. S4; paired-t-test; difference ± SE: 0.47 ± 0.22 kb, t 12 = 2.09, P = 0.059). However, there was no such pattern for the TeSmTL (paired-t-test; difference ± SE: 0.08 ± 0.093 kb, t 12 = − 0.86, P = 0.40), consistent with previous single TL analysis of the p arms of the X and Y chromosomes 13 .
At full term, the proportion of cells with TIF, measured in the placenta and the CAMs, differed significantly between tissues ( Fig. 2; F 2,38 = 7.48, P < 0.002). This was due to the proportion of cells with TIF being higher in placenta compared to both the CM and AM (t > 3.2, P < 0.007), while there was no difference in the proportion of cells with TIF between the CM and AM (t = 0.21, P = 0.98). TL parameters and TIF at 18-week gestation. We examined the CAMs as a single tissue at 18-weeks of gestation (gestational age 18.2 ± 2.57; mean ± SD; n = 10) because of technical difficulties of separating the CM from the AM at early stages of pregnancy. There were substantial inter-individual differences in SBmTL across tissues (ICC individual identity ± SE: 0.45 ± 0.23). There was a trend for SBmTL , to be longer in placenta compared to the CAMs within subjects (paired-t-test, difference ± SE: 0.75 ± 0.34 kb, t 9 = 2.91, P = 0.056), but there were no differences in TeSmTL across tissues (ICC = 0.0 ± 0.19). Consistent with the absence of a difference in TeSmTL, there was also no difference in the proportion of telomeres shorter than 3.0 kb or in TIF between membranes and placenta at 18 weeks of gestation (proportion < 3.0 kb: t 9 = 0.89, P = 0.40; TIF: t 9 = 0.32, P = 0.76).
Changes in TL parameters and TIF during gestation. We compared SBmTL, TeSmTL and TIF in a set of placentas and CAMs between 18-weeks of gestation (n = 10) and full term. SBmTL of placenta samples did not differ significantly between gestational stages ( Fig. 3a; difference ± SE: + 0.33 ± 0.51 kb at term; t 21 = 0.64, P = 0.53). However, TeSmTL was significantly shorter at full term ( Fig. 3b; − 0.66 ± 0.16 kb; t 21 = 4.09, P = 0.0005), which was at least partially due to a higher proportion of telomeres shorter than 3.0 kb at full term ( Fig. 3c; t = 3.83, P < 0.001).
SBmTL in the AM was significantly longer at full term than in the CAMs at 18 weeks ( Fig. 4a; + 0.97 ± 0.35 kb; t 18 = 2.78, P = 0.012), while in the CM at full term, SBmTL did not differ significantly from that in the CAMs at 18 weeks (− 0.26 ± 0.33 kb; t 18 = 0.79, P = 0.44). In contrast, TeSmTLs in both the AM and the CM were shorter at full term compared to the CAMs at gestation age 18 weeks ( Fig. 4b; AM: − 0.71 ± 0.26 kb, t 21 = 2.75, P = 0.012; CM: − 1.05 ± 0.24 kb, t 21 = 4.31, P = 0.0003). Consistent with this result, the proportion of telomeres shorter than 3.0 kb in CM was higher at full term than in the CAMs at gestation age 18 weeks ( Fig. 4c; t 21 = 3.09, P = 0.006), but this comparison was not significant for the AM (t 21 = 1.21, P = 0.24).
The proportion of cells in placenta showing TIF was higher at full term than at 18 weeks ( Fig. 2; t 28 = 6.92, P < 0.001). The proportion of cells in the CAMs showing TIF was also higher at full term than at 18 weeks (AM www.nature.com/scientificreports/ at term versus CAMs at 18 weeks: t 28 = 4.73, P < 0.0001; the same comparison for CM: t 28 = 4.11, P < 0.0001). The effect of pregnancy stage on TIF was stronger in placenta compared to the CAMs (interaction tissue * pregnancy stage: F 1,74 = 5.87, P = 0.023). TeSmTL was shorter in placenta samples with a higher proportion of cells with TIF ( Fig. 5a; linear regression t 18 = 4.40, P = 0.0003) and this association was independent of pregnancy stage (P = 0.8 when added to a regression model containing TIF only). The same result emerges when TeSmTL is replaced with the proportion of telomeres shorter than 3.0 kb (linear regression t 18 = 3.36, P = 0.003). In CAMs, there was no significant association of TeSmTL with the proportion of cells with a TIF when results of the AM and the CM were pooled ( Fig. 5b; t 28 = 0.84, P = 0.41), and this did not change selecting either AM or CM at full term (P > 0.33), or when replacing TeSmTL with the proportion of telomeres shorter than 3.0 kb (P > 0.58).

Discussion
Our work shows that at the end of normal pregnancy, the human placenta displays evidence of TL-mediated replicative senescence. This conclusion is based on TeSLA measurements, coupled with TIF analyses. It is supported by the following set of observations: (a) diminished TeSmTL and increased proportion of telomeres www.nature.com/scientificreports/ shorter than 3 kb in the placenta between 18 weeks of gestation and full term; (b) a buildup in TIF between the two gestational stages; and (c) correlation between TeSmTL and TIF across the two gestational stages (18 weeks and full term). We could not reach this conclusion based on TL measurements by SB, as SBmTL did not show shortening between 18 weeks of gestation and full term. Moreover, at full term, SBmTL was longer in the placenta than in cord blood, confirming previous findings, which originally led to a conclusion that the placenta cannot be an aging organ 3,4 . However, SBmTL results do not include the critical population of telomeres shorter than 3 kb, which apparently signals TL-mediated replicative senescence 5,6 . The association of TIF with TeSmTL, but not with SBmTL, in the placenta at full term (and at 18 weeks of gestation) further supports this conclusion.
A recent study showed a buildup during gestation of short telomeres in the placenta and CAMs of mice 14 . Most mice have exceedingly long telomeres for their approximately 2-year lifespan; unless genetically engineered, mice are thus a sub-optimal model of TL-mediated replicative aging 15 . That said, the finding in mice support our observations in humans. We note that in and of itself, a buildup of short telomeres is insufficient to prove TL-mediated senescence. In our study, the CAMs displayed not only shortening of TeSmTL but also buildup of TIF between 18 weeks of gestation and full-term. However, we found no significant correlation between TeSmTL and TIF in the CAMs either at 18 weeks of gestation or at full term. We therefore cannot be certain, at present, that the human CAMs experience TL-mediated replicative aging during gestation.
TIF may result from processes other than TL shortening with repeated cell divisions 16 . As double stranded DNA breaks that occur in telomeres resist normal DNA repair activities, any genotoxic stress has the potential to activate a persistent DDR in telomeric sequences 7,17 . TL independent TIF formation can therefore occur due to a number of stresses, including those that that cause stalling of telomeric DNA replication forks. As repetitive and G-rich telomeric sequences are prone not only to developing a greater abundance of DNA lesions such as oxidative damage, but also form secondary structures called G4-DNA, telomeres are regions on chromosomes that are susceptible to replication fork stalling 18 . Failure to restart DNA replication at stalled forks often leads to formation of double stranded DNA breaks, which persist in telomeres and cause cellular senescence. In addition, increased levels or reactive oxygen species (ROS) 19 or certain pro-inflammatory cytokines that raise intracellular ROS levels 20,21 can also lead to TL independent TIF formation, likely due to mechanisms involving telomeric DNA replication stress. As a result, even a relatively long telomere can be damaged, resulting in TIF and senescence.
The mechanistic pathways that trigger term and preterm parturition and engender normal/abnormal intrauterine growth are still poorly understood. Based on mTL data 4,22-24 , it has been proposed that TL-mediated replicative aging of the placenta and CAMs might be a common pathway that contributes to the timing of term/ preterm parturition and play a role in intra-uterine growth restriction 1-3 . Our pilot study examined, therefore, associations of TL parameters, measured by SB and TeSLA, with PTB, PTBI, and SGA pregnancies. We found that TL was short in the placenta (based on SBmTL) and CAMs (based on SBmTL and TeSmTL) only in PTBI. We cannot conclude, however, based on these findings that TL-mediated accelerated aging of the placenta, and CAMs in particular, contribute to PTBI. That is because the CAMs and placenta of PTBI display massive infiltration by maternal inflammatory cells 25 with telomeres that are likely shorter than those in fetal structures. Therefore, in women with intra-amniotic infection, the inflammatory processes (e.g. inflammasome activation) induced by invading microbes may be the main cause of preterm labor and birth 26,27 .
Finally, we note the following study limitations: First, the sample size is small and for obvious reasons findings are based on cross-sectional data for evaluation of TL and TIF changes between 18 weeks of gestation and full term. Still, using TeSLA and TIF analysis, we were able to show that the placenta displays a buildup of short telomeres and telomere dysfunction at term. We were also able to show the same findings in CAMs but because of lack of correlations between TeSmTL and TIF in the CAMs, we are uncertain of the biological meaning of these findings in these tissues. Second, PTB and SGA are highly heterogeneous in nature and replicative aging might play a role in a subset of these obstetrical syndromes, which we might have missed because of the small sample size. Third, our findings are based on samples donated by subjects of African American and Hispanic ancestries. A recent study reported that placental mTL (measured by quantitative real-time polymerase chain reaction) was shorter in African Americans compared to whites of European ancestry 28 , while the consensus is that leukocyte mTL is longer in African Americans 29,30 . It is essential, therefore, to examine the profiles of SBmTL and TeSmTL in the placenta and CAMs across individuals from different ancestries.
In conclusion, TeSLA and TIF findings are consistent with the thesis that the human placenta experiences TL-mediated replicative aging during the course of pregnancy.

Methods
Subject characteristics. Samples were collected at the University Hospital (UH) of the New Jersey Medical School, Rutgers University, Newark New Jersey. At the UH, we collected samples from the placenta, CAMs, UCB, and maternal blood from 13 full-term normal newborns and their African American (n = 6) and Hispanic (n = 7) mothers [aged 27.62 ± 5.61 years (mean ± SD)]. We also obtained placenta and CAM samples from elective abortions donated by African American mothers (n = 10). In this set, the CM and the AM were collected as one tissue sample. The Rutgers University Institutional Review Board (IRB) approved research on the progression of pregnancy from 18 weeks of gestation to full term, which was performed at the New Jersey Medical School. All mothers for this component of this work signed a written informed consent.
Human placental and CAM samples for the pilot study of pregnancy outcomes were obtained at the Hutzel Women's Hospital, Detroit Medical Center, Wayne State University (Step 3). The following groups of women (all African Americans, 9 women per group) were included: normal term; PTB; and PTBI based on high levels of IL-6 (> 2.6 ng/ml) 31 and detectable microorganism in amniotic fluid [32][33][34] . SGA was defined as a term neonate born at 37-42 weeks of gestation with birthweight below the 10th centile for gestational age and with a cerebral placental ratio below the 5th centile for gestational age 35  Telomere shortest length assay. TeSLA measurements were performed as previously described 12  Immunostaining and telomere-immunoFISH. All procedures were performed essentially as described 8 . In brief, tissue samples were cryopreserved, cut, fixed in 4% paraformaldehyde/PBS, permeabilized in PBS/Tween (0.1%), and blocked overnight in 4% BSA/PBS. Samples were subsequently rinsed, dehydrated and nuclear DNA denatured in hybridization buffer containing Cy3 conjugated telomere specific peptide nucleic acid. Samples were then incubated with primary antibodies and subsequently with secondary antibodies as indicated. Slides were washed and mounted using DAPI containing mounting medium (Vector Laboratories, Burilngame, CA, USA). Immunostaining used primary polyclonal anti-53BP1 antibodies (Novus, Centennial, CO, USA) and secondary AlexaFluor 488-conjugated goat anti rabbit antibodies (Invitrogen, Carlsbad, CA, USA).
Microscopy. Images were acquired using a Zeiss AxioObserver Z1, an AxioCam MRm camera (Zeiss, Pleasanton, CA, USA), and Zen Blue 2.3 software (Zeiss, Pleasanton, CA, USA). To analyze and quantitate colocalization between telomere signals and 53BP1 foci, images from 5 fields were acquired as z-series through the thickness of telomere signals (4-6 images, 0.5 μm optical slices) using a motorized stage and a 63X/1.4 oil immersion lens. To quantitate 53BP1 foci, images from 12 fields were acquired through the thickness of 53BP1 foci as z-stacks using a 40×/1.4 oil immersion lens. Stacks were merged into a single image using the Zen software for easier counting.
Statistics. Data were analyzed using (mixed) general linear models using R Core Team 37 and Tukey posthoc tests. Proportions were arcsine square root transformed prior to analysis to ensure homoscedasticity.
Ethics declarations. All subjects gave their informed consent for inclusion before they participated in the study.

Approval for human experiments. This study received IRB approvals from Rutgers, New Jersey Medical
School, The State University of New Jersey (Protocol No. 2018001630) and Wayne State University (Protocol No. 1011008995 and 0510003068). All participating women provided written informed consent prior to sample collection. All methods involving human participants were carried out in accordance with relevant guidelines and regulations.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.