Combined Effect of Telomere Length and Mitochondrial DNA Copy Number As a Potential Biomarker Indicating PE Risk: a Case-Control Study in a Chinese Population

Purpose To explore changes of Telomere length (TL) and mitochondrial copy number (mtDNA-CN) in preeclampsia (PE) and to evaluatethe combined effect of maternal TL and mtDNA-CN on PE risk. Methods A case-control study of 471 subjects (130 PE cases and 341 age frequency matched controls) was conducted in Relative telomere length (RTL) and mtDNA-CN were measured using quantitative polymerase chain reaction (qPCR) and PE risk was calculated between groups by logistic regression analyses. Results PE patients displayed longer RTL (0.48 versus 0.30) and higher mtDNA-CN (3.02 versus 2.00) in maternal bloodas well as longer cord blood RTL(0.61 versus 0.35) but lower mtDNA-CN (1.69 versus 5.49) in cord blood (all p<0.001). Exercise during pregnancy exerted an obvious effect of prolonging maternal telomere length. Multiparous, women with folic acid intake during early pregnancy and those delivered vaginally showed longer telomere length while those factors imposed no or opposite effect on RTL in PE cases. Furthermore, RTL and mtDNA-CN were positively correlated in controls (in maternal blood r=0.18, p<0.01; in cord blood r=0.19, p<0.001), but this correlation was disrupted in PE cases, no matter in maternal blood or in cord blood. Longer maternal RTL and higher mtDNA-CN were associated with higher risk of PE, and the ROC curve of RTL and mtDNA-CN in predicting PE risk presented an AUC of 0.755(95%CI: 0.698-0.812). Conclusions Interaction of TL and mtDNA-CN may play an important role in pathogenesis of PE and it could be a potential biomarker indicating PE risk. blood and cord blood in women with PE, and further to explore their combined effect of on PE risk. The results of present study may aid improvement in the current understanding of PE, through identifying the joint involvement of TL and mtDNA-CN in PE pathogenesis. based on last date Both activity and into four groups by levels of (qPCR) FlexReal-Time PCR The ratio of telomere repeat copy (T) single copy 36B4 number (S) was computed to reect relative telomere length (RTL). Primers sequences for telomere PCR were TEL1,5′-GGTTTTTGAGGGTGAGGGTGAGGGTGAGGGTGAGGGT-3′ and TEL2,5′-TCCCGACTATCCCTATCCCTATCCCTATCCCTATCCCTA-3′ and the single-copy gene(36B4) primers sequences were 36B4u, 5′-CAGCAAGTGGGAAGGTGTAATCC-3′, and 36B4d, 5′-CCCATTCTATCATCAACGGGTACAA-3′. We also determined mtDNA-CN as the ratio of mitochondrial encoded NADH dehydrogenase-1(ND-1) to nuclear gene (hemoglobin subunitβ, HGB) by simultaneous amplications of ND1 and HGB genes. secondhand T-test of normal distributed and nonparametric continuous variables in cases and controls. Wilcoxon signed rank test mtDNA-CN matched Kruskal Wallis by levels categorized into two their The odds calculated by logistic regression analyses. Then a backward stepwise logistic regression carried explore the independent factors on PE risk. Receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC) were calculated estimate the feasible use of maternal RTL and mtDNA-CN as possible markers in determining PE risk. A two-sided P< 0.05 was considered statistically signicant, and p values with signicance are marked in bold in tables. In conclusion, loss of positive correlation between RTL and mtDNA-CN in our study may induce the initiation or progression of PE pathogenesis; additionally, maternal RTL and mtDNA-CN before delivery were positively associated with the risk of PE, suggesting that increased levels of maternal RTL and mtDNA-CN were risk factors of PE and their combined effect had a predictive ecacy for PE risk. This study demonstrates the contribution of interplay between RTL and mtDNA-CN to pathogenesis of PE and opens a new perspective for PE prediction.


Introduction
Preeclampsia (PE), a serious multisystem disorder, is de ned as a new onset of hypertension with either proteinuria or end-organ dysfunction after 20 weeks of gestation [1]. With an incidence of 2-8% worldwide, PE is one of the leading causes of maternal and perinatal morbidity and mortality [1]. Although the pathophysiology of PE is not completely understood, early poor perfusion or placental ischemia-reperfusion injury leading to increased oxidative stress has been widely accepted as one of the main pathological processes responsible for the development of PE [2][3]. There are some markers of oxidative stress raised in PE circulation [4], but it is not well established about the speci c changes of oxidative stress biomarkers in the context of PE and insights into the relationship between oxidative stress markers and PE will help to illustrate the mechanisms of oxidative stress in the context of PE and to discuss perspectives on disease prediction.
Telomere length (TL) and mtDNA copy number (mtDNA-CN) are emerging biomarkers of oxidative stress and have been related to various age-related diseases. They are highly susceptible to oxidative stress [5] and in ammation so that have been suggested as sensitive indexes of cellular oxidative stress, mitochondrial dysfunction, aging process, and age-related diseases [6].Recent research has indicated that telomeres and mitochondrial are functionally linked and telomere dysfunction could induce p53 represses PGCs and result in metabolic and mitochondrial compromise, suggesting disorder of telomeremitochondria axis may be an important and early event in biological aging diseases [7][8]. Hyunsu et al found that loss of the association between telomere length and mitochondrial DNA copy number may induce the initiation of Colorectal Carcinogenesis and co-regulation of telomere and mitochondrial may play an important role in the genesis and development of Oxidative stress-related diseases [9]. However, evidence referring to TL and mtDNA-CN in preeclamsia was scarce and the results were mixed [10][11].
Therefore, we hypothesized that alterations in leukocyte TL and mtDNA-CN may re ect the cumulative exposure to oxidative stress and underlie the pathogenesis of PE. However, no previous study has been conducted on the association between TL and mtDNA-CN in PE cases. Here, we designed a case-control study with 130 PE cases and 341 controls in Chinese population to investigate TL and mtDNA-CN alterations in maternal peripheral blood and cord blood in women with PE, and further to explore their combined effect of on PE risk. The results of present study may aid improvement in the current understanding of PE, through identifying the joint involvement of TL and mtDNA-CN in PE pathogenesis. Study recruitment PE cases and controls were recruited from Nanjing Drum Tower Hospital in Nanjing, Jiangsu province of eastern China between January 2019 and June 2020. According to the report of the American College of Obstetricians and Gynecologists' Task Force on Hypertension in Pregnancy, PE is diagnosed by specialist doctors at admission [12]. During the same period, women who were healthy without any complications and age frequency matched were recruited as controls. Inclusion criteria of the study were single viable pregnancy, any maternal age, maternal BMI, and parity status. Exclusion criteria were multiple pregnancies, pregnancy with fetal anomalies, pregnancy with preexisting chronic disease such as chronic hypertension, diabetes mellitus et al and other pregnancy complications such as gestational diabetes, prelabor rupture of membranes, chorioamnionitis, placenta abruptions and so on. Finally, 130 cases and 341 controls were included. All subjects were informed about the study and after signing the informed consent; they were interviewed bywell-trained interviewers with structured questionnaires. Data available in the questionnaire include demographic data and lifestyle related factors during pregnancy such as education, folic acid intake, occurrence of threatened abortion, physical activity, secondhand smoking, BMI at labor admission, medical and reproductive characteristics. Delivery related information such as gestational age, delivery mode, birth weight, and occurrence of postpartum hemorrhage (PPH) were obtained from obstetric records. Gestational age was calculated based on last menstrual period or ultrasound-based estimated date of conception. Both physical activity and secondhand smoke were divided into four groups by levels of activities.
Maternal peripheral venous blood samples were collected from 471 participants right before or during delivery. Paired umbilical cord blood samples were collected immediately after birth from the cord vein of newborns. Blood samples were shipped by cold chain equipment to the laboratory and stored until analysis. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This study was approved by the ethics committees of Nanjing Drum Tower Hospital (No.2018-017).
Measurement of relative telomere length and mitochondrial DNA copy number Genomic DNA was extracted from leukocytes of maternal peripheral blood and cord blood. Telomere length and mtDNA-CN were analyzed on modi ed quantitative polymerase chain reaction (qPCR) by QuantStudio 7TM FlexReal-Time PCR System (Applied Biosystems). The ratio of telomere repeat copy number (T) to single copy gene 36B4 number (S) was computed to re ect the relative telomere length (RTL). Primers sequences for telomere PCR were TEL1,5′-GGTTTTTGAGGGTGAGGGTGAGGGTGAGGGTGAGGGT-3′ and TEL2,5′-TCCCGACTATCCCTATCCCTATCCCTATCCCTATCCCTA-3′ and the single-copy gene(36B4) primers sequences were 36B4u, 5′-CAGCAAGTGGGAAGGTGTAATCC-3′, and 36B4d, 5′-CCCATTCTATCATCAACGGGTACAA-3′. We also determined mtDNA-CN as the ratio of mitochondrial encoded NADH dehydrogenase-1(ND-1) to nuclear gene (hemoglobin subunitβ, HGB) by simultaneous ampli cations of ND1 and HGB genes. Primers sequences were as follows: ND1 forward 5'-CCCTAAAACCCGCCACATCT-3'; ND1 reverse 5'-GAGCGATGGTGAGAGCTAAGGT-3'; HGB forward 5'-GAAGAGCCAAGGACAGGTAC-3', and HGB reverse 5'-CAACTTCATCCACGTTCACC-3'. Reference DNA (pooled from 5 healthy controls) was used to generate a standard curve for quanti cation. After exclusions of outliers, average cycle threshold (Ct) values of the remain samples were calculated. Each reaction system contains 10µl SYBR ® Green PCR Master Mix (Applied Biosystems) with a nal DNA concentration of 5 ng/µl. All samples were assayed in duplicate and three quality controls were employed in each plate to analyze variability. qPCRs were executed by investigators blinded to clinical data and disease status. RTL and mtDNA-CN were calculated based on Cawthon's formula [13][14]:

Statistical analysis
All statistical analyses were performed with Stata version 15.1 (Stata Corp, College Station, TX). Sample characteristics were described as means (SD), median (IQR) or percentages. Pearson χ2 test was used to test differences between cases and controls for categorical variables (maternal age group, education, folic acid intake, occurrence of threatened abortion, physical activity, secondhand smoke exposure status, BMI at delivery and mode of delivery). T-test and Wilcoxon rank test were employed to compare the differences of normal distributed and nonparametric continuous variables in cases and controls. Wilcoxon signed rank test was utilized for the comparations of RTL and mtDNA-CN between matched maternal blood and cord blood. Kruskal Wallis rank test was used to compare nonparametric variables among groups. Correlations of RTL and mtDNA-CN in PE cases and controls were analyzed by Spearman's rank correlation. To estimate the relative association between preeclampsia and levels of maternal RTL and mtDNA-CN,we categorized maternal RTL and mtDNA-CN into two groups according to their median distribution. The odds ratios (ORs) and 95% con dence intervals (95% CIs) of RTL and mtDNA-CN associated with PE risk were calculated by logistic regression analyses. Then a backward stepwise logistic regression was carried out to explore the independent factors on PE risk. Receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC) were calculated to estimate the feasible use of maternal RTL and mtDNA-CN as possible markers in determining PE risk. A two-sided P< 0.05 was considered statistically signi cant, and p values with signi cance are marked in bold in tables.

Baseline characteristics of study participants in PE cases and controls
Women with PE tended to be less educated, exerted with more gravidities, lower folic acid intake rate, higher threatened abortion rate in early pregnancy, more secondhand smoke exposure, smaller gestational age, higher BMI at delivery, higher cesarean section rate, lower birth weight, higher PPH incidence yet with an increase in physical activity compared to the controls(p<0.05).
No signi cant differences were found for parity between cases and controls. Maternal RTL was signi cantly longer in PE patients as compared with controls (median:0.48 versus 0.30; p<0.001), and this trend was also observed from cord blood (median: 0.61 versus 0.35; P<0.001). Maternal mtDNA-CN of PE cases was signi cantly higher than that of controls (median: 3.02 versus 2.00; p<0.001); however, in cord blood, lower mtDNA-CN was found in PE cases (median: 1.69 versus 5.49; p<0.001) ( Table 1. Fig. 1).  Association between RTL and mtDNA-CN in maternal blood and cord blood As shown in Fig. 2, a positive correlation of maternal RTL and mtDNA-CN was determined in normal pregnancy(r=0.18, p<0.01, Fig. 2A), but this correlation was disappeared in PE patients (p=0.52, Fig. 2B). We also found positive correlation between RTL and mtDNA-CN in cord blood within controls(r=0.19, p<0.001, Fig. 2C) while a negative correlation was observed in PE cases(r=-0.23, p<0.01, Fig. 2D

Effects of maternal RTL and mtDNA-CN for PE prediction
We further applied logistic regression models with backward stepwise procedures to investigate the independent factors associated with PE risk (Fig. 3). Model 1 incorporated all the maternal characteristics associated with PE from Table 1

Discussion
Given the dearth of ndings on the role of RTL and mtDNA-CN in the pathophysiology of PE, this study aimed to compare RTL and mtDNA-CN in PE patients and normal pregnant controls, in the meanwhile, to evaluate predictive value of maternal RTL and mtDNA-CN on PE risk. Our study demonstrated that PE patients displayed longer RTL and higher mtDNA-CN in maternal blood as well as longer cord blood RTL, but lower cord blood mtDNA-CN compared to normal pregnant women. Multiparous, exercise during pregnancy and rst trimester folic acid intake exerted lengthening effect on maternal telomere length. Furthermore, RTL and mtDNA-CN were positively correlated in healthy pregnant women and newborns while this correlation was disrupted in PE cases. We also found longer maternal RTL and higher mtDNA-CN were associated with higher risk of PE, followed by presenting the evidence that combination of maternal RTL and mtDNA-CN was effective in prediction of PE risk. Up to our knowledge, this was the rst study to examine the combined prediction effect of maternal RTL and mtDNA-CN on PE risk.
Previous studies reported shorter telomeres in placental samples from pregnancies complicated with PE and suggested abnormal telomere homeostasis was closely related to the pathogenesis of PE and raised TL could be a potential biomarker of PE [15].Evidence has supported consistency of telomere length from different tissue, thus easily accessible leukocyte was widely employed as a substitute of tissue in telomere length measurement [16][17][18][19]. However, research on peripheral leukocyte TL and PE was scarce. One previous study was conducted with 50 cases of PE and 50 controls in Washington State. Harville et al analyzed telomere length in peripheral blood and found women in the highest tertile showed a trend of higher PE risk (OR 1.08) compared to those in the lowest tertile when adjusted with age although the difference was not statistically signi cant [10], which may be explained by the limited sample size. As TL shortening has been associated with various aging related diseases, it is somewhat surprising that we found longer leukocyte TL was associated with PE in the current study. More physical activity among our PE patients may partially explain this inconsistency; most of our PE patients were more likely to follow doctors' advice during pregnancy and kept good lifestyles although they had been in initial abnormal state. Evidence supported that telomere length of leukocyte cells were positively associated with healthy living and physical activity may confer protection against telomere length shortening [20][21][22]. Thus, the modi ed life style may lengthen TL enough to compensate TL attrition when most PE cases in our study were mild and well controlled with imperceptible shortening. Furthermore, although in cells of most human tissue telomeres shorten throughout human life, it is heavily confounded by, among other factors, the variable levels of telomerase activity-and hence variable capacities for telomere length replenishment-in stem cells. These can constantly renew somatic tissue cells. Telomerase is enriched in germline lineage cells and embryonic stem cells [23][24][25][26]; previous data have reported that telomerase reverse transcriptase (TERT) and telomerase activity were signi cantly higher in preeclamptic placenta when compared with control group according to different gestational age [27][28], which may cause a lengthening effect of TL in PE cases as well. Additionally, despite consistency of telomere length among tissue, leukocyte TL may not completely re ect TL changes in placenta in PE cases. More studies are warranted to elucidate the correlations between leukocyte TL and placenta TL throughout the progression of PE.
Mitochondrial dysfunction contributing to the pathogenesis of preeclampsia has been proposed by Torbergsen et al [29] and Widschwendter et al [30], the latter indicating that defects in trophoblastic mitochondria may be the initial step in the pathophysiological process of PE. In this study, we found higher mtDNA-CN were associated with higher PE risk, which corroborates previous ndings [31]. Previous data also reported that the odds of preeclampsia were positively associated with maternal blood mtDNA-CN [32]. These results further support the idea that elevated peripheral blood mtDNA-CN may serve as a risk marker for PE. It has been demonstrated that mtDNA is highly susceptible to oxidative stress and mitochondrial damage, as re ected by changes in mtDNA-CN may alter mitochondrial gene expression and cause oxidative phosphorylation de ciency and surge of ATP by glycolysis [33]. In addition, oxidative stress, an important pathogenesis pathway involved in preeclampsia [34], may alter mitochondrial function and increase mtDNA-CN through several mechanisms, it is possible that elevated systematic reactive oxygen species (ROS) may impair or disrupt cellular components such as mitochondrial lipid membranes [35]. ROS may also in uence mitochondrial function by impairing DNA and damaging election chain transport; and a compensatory response to this cellular stress may lead to an increase in mtDNA-CN [33,36]. This result is highlighted by experimental animal studies demonstrating increased mitochondrial damage and mtDNA-CN with increasing exposure to pro-oxidants [35]. These data altogether suggest that that association of increased mtDNA-CN with preeclampsia is biologically plausible. However, several other studies reported inconsistent ndings. The heterogeneity in results between studies may be related to population diversity, lifestyle modi cations, exposure levels, time windows and ability of mitochondrial DNA compensation [37][38].
The initial or newborn setting of TL represents a critically important characteristic of an individual's telomere biology system [39]. In our study, we found that RTL of cord blood were longer than that of maternal blood both in controls and PE cases. But we did not observe a reverse association between age and RTL in pregnant women. Our narrow age range may have accounted for this lack of association; the rate of changes in telomere length varies with age and is greatest in childhood and elderly while relatively stable in adults [40][41]. In addition, only around 10 percent of women were at or older than 35 and less than 10 percent were younger than 25 in our study, so the small sample in those subgroups may yield some deviation. The above reasons may dilute the adverse effects of aging on telomere length in the current study. Recent studies emphasized telomere length had association with diet and lifestyle determinants [42][43][44]. In our study, we also found both folic acid intake and regular exercise exerted prolonging effect on maternal TL in normal pregnancy, which further support the idea that TL could be adjusted by modifying lifestyles. This result underlined that TL was sensitive enough to re ect alterations of oxidative stress in pregnant women and could be a reliable biomarker of oxidative stress during pregnancy.
Although mtDNA-CN has been directly related to obesity in child population [45], we did not observe this effect in our study.
Higher mtDNA-CN was shown associated with less exercise and more secondhand smoke in normal pregnancy in this study. The excessive oxidative stress generated by less exercise and more secondhand smoke may lead to increased mtDNA-CN synthesis as a compensatory mechanism to ensure cell survival. In addition to population diversity, variations in levels and duration of exposure to oxidative stress [37,46]  . This duality in response to mild vs. excessive oxidative stress might also explain previous mixed ndings on mtDNA content. It is worthy to note that umbilical cord blood mtDNA-CN in PE cases was much lower than that of controls, which was in contrast to our result of maternal blood; in all probability pregnancy is a period when newborns are extremely susceptible to oxidative stress as this time mitosis is highly active, with a result that mitochondria in cord blood of PE patients exceeded the extent of compensation and showed a decrease in copy number. This nding is in line with a recent study which observed air pollution during pregnancy was associated with decreased mtDNA content in cord blood and suggested heightened sensitivity to oxidative stress during the speci c prenatal window of life stage [48].
In our control subjects, there were positive correlations between mtDNA-CN and telomere length in maternal blood and cord blood suggesting co-regulation of telomeres and mitochondrial function in mothers and newborns in normalpregnancy.
Telomere length and mtDNA-CN have largely been examined as independent contributors to oxidative stress related diseases, yet there is growing evidence that these two markers are functionally linked or at least combining them together may better predict disease development. Several recent studies concerning school age children and healthy adults as well as elderly women reported a positive association between mtDNA-CN and telomere length [49][50][51][52]. Studies involving animal model and cell culture experiments havealsoshown that telomere dysfunction is associated with abnormal mitochondrial biogenesis and function [53][54]. However, just limited evidence with small sample size [55] has documented the dependence between these two biomarkers in pregnant women. Therefore, we found out strong correlations between RTL and mtDNA-CN when they have been measured both in maternal and cord blood during normal pregnancy. Mechanism of this association remains to be determined; mitochondria effects of p53 activation from telomere dysfunction [7] and TERT effects on mtDNA repair may be involved [56]; Telomere shortening can reciprocally lead to cellular mitochondrial endangerment and diminished mitochondrial biogenesis via diminution of PGC-1α, the master regulator of mitochondrial biogenesis [7]. In addition, TERT, a catalytic subunit of telomerase with canonical role of telomere maintenance, contains both nuclear localization signal and mitochondrial targeting sequence, and might be transported from nuclei to mitochondria under increased oxidative stress conditions to regulate mitochondrial function and protect mtDNA from oxidative damage [57]. However, this association was not observed in PE cases implying that the pathways shared by regulating telomere length and mitochondrial biogenesis might be disturbed by the pathophysiology of PE. The present ndings provide evidence that telomeres and mitochondria are co-regulated in normal pregnant women and their newborns, and this co-regulation was interrupted when PE occurs, with increased RTL and mtDNA-CN been associated with increased PE risk. In this sense, the pathophysiology of PE may play a role in the mechanisms regulating the association between telomere length and mtDNA-CN and the complex interplay between TL and mtDNA-CN could be a potential effective predicting factor of PE risk. Considering this, we rstly investigated the cumulative effect of TL and mtDNA-CN on PE risk. As a result, we found that combination of maternal leukocyte mtDNA-CN and RTL can effectively predict the risk of PE, contributing to recent investigations concerning improvement in PE prediction models. Further studies are still needed to verify our results and more precise knowledge of the regulatory pathways governing interaction of RTL and mtDNA-CN to PE process is also necessary to delineate both in its onset and pathogenesis.
Our study has some strength. We are the rst to investigate the combined effect of maternal RTL and mtDNA-CN on PE prediction and rstly revealed the disruption of positive dependence between these two biomarkers participate in the process of PE even without knowing the causative or just a compliance. Second, many lifestyle factors such as smoking status, folic acid intake, exercise during pregnancy as well as BMI before delivery, which may in uence RTL and mtDNA-CN, were applied and adjusted for in the present study to examine the independent effect of RTL and mtDNA-CN on PE risk. Moreover, unlike other studies regarding BMI and pregnancy, we chose to use BMI before delivery rather than pre-pregnancy because this variable would have more of impact on RTL and mtDNA-CN measured in our study as it takes into account weight gain during pregnancy.
Several limitations also need to be acknowledged. First, we only measured RTL and mtDNA-CN before delivery and the alterations of RTL and mtDNA-CN in leukocytes during different stages of pregnancy and disease progression remain unclear.
Second, our study is restricted to Han Chinese; the generalizability to other ethnic cohort needs further evaluation. Third, due to the sample size, our study treated all PE cases as a whole; types of PE need to be investigated separately as early onset preeclampsia processes different pathogenesis as that of late onset. Fourth, although we have incorporated a series of lifestyle related factors in uencing RTL and mtDNA-CN into our study, many other l factors such as stress, environmental pollutants and detailed diet habits were not involved as they are di cult to measure objectively. Finally, we only carried out the association analyses among RTL, mtDNA-CN, and risk of PE. The underlying mechanisms that account for pathways of leukocyte mtDNA content and RTL effect PE pathophysiology need further investigation.

Conclusions
In conclusion, loss of positive correlation between RTL and mtDNA-CN in our study may induce the initiation or progression of PE pathogenesis; additionally, maternal RTL and mtDNA-CN before delivery were positively associated with the risk of PE, suggesting that increased levels of maternal RTL and mtDNA-CN were risk factors of PE and their combined effect had a predictive e cacy for PE risk. This study demonstrates the contribution of interplay between RTL and mtDNA-CN to pathogenesis of PE and opens a new perspective for PE prediction.