**4. Autophagy in Placentation**

The MAP1LC3 protein families: MAP1LC3A, MAP1LC3B, and MAP1LC3C, are expressed in both the labyrinth zone and decidua basalis in mouse models. Expression of MAP1LC3A and MAP1LC3B were higher in the decidua basalis than in the labyrinth layer [22]. Autophagy activation was observed in human EVTs, which invade the maternal decidua basalis at the implantation site, at week 7 of gestation [23]. Autophagy plays an important role in trophoblast functions, including invasion and vascular remodeling in EVTs, for normal placental development [23]. This was confirmed using a mouse model, in which the Atg7 gene, essential for autophagy, was deleted in trophoblast cells, but not fetuses, by a lentiviral vector [24]. The Atg7 knockout placentas were smaller than the wild, suggesting autophagy deficiency mediates poor placentation, a feature of preeclamptic placentas (Figure 2) [24]. The Atg7 knockout placentas were characterized by shallow trophoblast invasion and failure of vascular remodeling. This functional impairment was confirmed by autophagy-deficient human EVT cell lines, which are constructed by stably transfecting Atg4BC74A, an inactive mutant of Atg4B that inhibits autophagic degradation and lipidation of MAP1LC3B paralogs in hypoxia [23,25]. Physiological hypoxia during early pregnancy, with approximately 2% oxygen tension, induces autophagy in primary trophoblasts [23,26]. Although hypoxia inducible factor1 α (HIF1 α) is required for EVT invasion regardless of oxygen tension; failure of EVT invasion was provoked by hyper-expression of HIF1 α by cobalt chloride, and excessive autophagy activation by glucose oxidase in HTR8/SVneo cells, an EVT cell line [27–29]. Thus, physiological hypoxia regulates autophagy by adjusting trophoblasts to cope with harsh conditions during early placentation.

**Figure 2.** Placental autophagy inhibition inducing gestational hypertension and poor placentation. (Left figure) Trophoblast-invasion and vascular remodeling are fundamental for normal placentation (the black arrows indicate the place of invasion and vascular remodeling). Autophagy deficiency impairs the functions of trophoblasts in the trophoblast-specific Atg7 knockout mouse model, resulting in poor placentation (the red "T" bars indicate the inhibition). PlGF mRNA levels, but not sFlt1 mRNA levels, are decreased in the knockout placentas (the red arrow indicates the decrease, and the black arrow indicates the stable). (Right figure) Also, the dams bearing the knockout placentas showed hypertension, but not proteinuria (the red arrow indicates the induction of hypertension by the placenta).

Trophoblastic stem cells differentiate to syncytiotrophoblasts as well as EVTs. Autophagy regulates the differentiation of invasive trophoblasts via reduction of galectin-4, which is required for normal placental development, as seen in a rat model [30,31]. Autophagy activation is expected during syncytialisation of BeWo cells, a choriocarcinoma cell line [32,33]. During this process, p53 negatively regulates autophagy activation based on high levels of p53 in the nuclei of cytotrophoblasts, but not in syncytiotrophoblasts [33]. However, as these experiments used choriocarcinoma cell lines, this experiment should be replicated using the primary human trophoblast differentiation model [34].

HIF1α is a key factor for EVT invasion. HIF1α expression, induced by hypoxia, was not affected by autophagy suppression in trophoblast cells [23]. Interestingly, CMA partially controls HIF1α expression in lysosomes [35]. Hypoxia stabilizes HIF1α by blocking proteasome-mediated degradation, but HIF1α is degraded via CMA in response to nutrient deprivation, but not hypoxia in the liver of rats [35]. CMA may be important for EVT invasion via modulating HIF1α expression levels, because the placenta, especially in intervillous space, develops under in conditions of hypoxia and low glucose during the first trimester [36,37].

### **5. Autophagy in Pregnancy-Related Complications**

#### *5.1. Preeclampsia or Fetal Growth Restriction (FGR)*

It has been reported that the expression of BECN1, involved in autophagosome formation in mammalian placentas, is higher in the presence of FGR without preeclampsia, but not when preeclampsia is present [38,39]. However, BECN1 increase has been reported recently in preeclamptic placentas compared to those in age-matched controls [40]. A substrate of autophagy, p62, is highly expressed in EVT cells in human placental bed biopsies obtained from preeclampsia, suggesting that autophagy inhibition is present in EVTs of preeclamptic placentas. Sera from preeclamptic patients induce hypertension and proteinuria in pregnan<sup>t</sup> interleukin 10 (IL-10) knockout mice, suggesting that factors in blood, including soluble endoglin (sENG) and soluble fms-like tyrosine kinase (sFlt1), induce preeclampsia-like features in mice [41]. The sera from normotensive women, but not from women with preeclampsia, induced autophagy in peripheral blood mononuclear cells [42]. In the sera of preeclamptic women, sENG, which blocks transforming growth factor-β1 (TGF-β1) signals, suppressed invasion and vascular remodeling via autophagy inhibition in EVT cell lines. This effect was reversed by administration of TGF-β [23]. Pregnant women with donor oocytes would be at a greater risk of preeclampsia and gestational hypertension than pregnan<sup>t</sup> women with their own oocytes [43–45]. Accumulation of p62, an indicator of autophagy inhibition, in EVTs was significantly higher in women with donor oocytes, suggesting autophagy inhibition correlates with preeclampsia [46]. Conversely, some reports sugges<sup>t</sup> activation of autophagy in preeclamptic placentas. An electron microscopic study showed autophagic vacuoles in both syncytial layers and endothelium in preeclamptic placentas [40]. An increase in MAP1LC3-II and decrease in p62 were reported in the placentas of women with hypertensive disorder, compared to those in normotensive pregnancies, which indicates autophagy activation [47]. Ceramide overload-induced autophagy impaired placental function in preeclampsia in cooperated with oxidative stress-reduced hydrolase activity [48]. Autophagy is clearly involved in the pathophysiology of preeclampsia, but the effect on preeclamptic placentas remains unclear. As mentioned earlier, it is still impossible to accurately estimate autophagy flux in fixed tissues because autophagy is a dynamic mechanism to maintain homeostasis in cells. A placental autophagy-deficient model is required to solve this problem. Dams bearing Atg7-knockout placentas, which were smaller than wild dams, showed hypertension without proteinuria, suggesting that autophagy deficiency in placentas, but not in maternal bodies, induced gestational hypertension [24]. Autophagy deficient placentas, in which mRNA levels of placental growth factor (PlGF), but not sFlt1, decreased, appear to affect maternal circulation, but not endothelial dysfunction [24]. Atg9a mediates autophagosome formation and is ubiquitous in multiple human organs. Atg9b, a homolog of Atg9a, is found only in the placenta and pituitary gland [49]. The role of autophagy under preeclamptic dams was reported using Atg9a knockout mice mated with heterozygous p57Kip2 mice, which develop hypertension and proteinuria in dams [50]. The incidence of fetal death increased in pups with hetero- or homozygous deletion of Atg9a compared to that in the wild type [51]. In addition, the body weights in Atg9a homozygous knockout pups were significantly lower than those in Atg9a heterozygous knockout or wild type pups. Taken together, autophagy protects placental and fetal growth from stress under preeclampsia.

Autophagic vacuoles are more likely to be present in the syncytiotrophoblast layer of human FGR placentas, which indicates autophagy activation [52,53]. Higher expression of BCLN1 in FGR placentas might support this notion [38]. On the other hand, a recent paper reported that the birth weight of fetuses delivered from dams with labyrinth layer-specific Atg7-deleted placentas were significantly lower than the birth weight of dams with normal placentas, indicating that inhibition of autophagy was also related to FGR [54]. In addition, Hirota et al. reported that an autophagy inducer, rapamycin, which is used for preventing preterm birth, did not affect the body weight of pups [55]. There is still some controversy for autophagy status in placentas with FGR between human and mouse.

Protein aggregation caused by autophagy suppression has been reported in several neurodegenerative diseases, including Alzheimer's disease, Parkinson's disease, and Huntington's disease [56,57]. Recently, protein aggregation has been reported in preeclampsia [58]. Transthyretin, a transporter of thyroxine and retinol, and amyloid precursor proteins; which are proteins that accumulate in neurodegenerative diseases are also seen in preeclamptic placentas [59,60]. Furthermore, aggregated amyloid proteins were detected in higher levels in the urine of women with preeclampsia than in healthy pregnan<sup>t</sup> women [59,61]. Thus, autophagy would prevent protein aggregation in trophoblasts. Aggregated proteins might disturb placental development through induction of apoptosis, and cellular senescence. Cellular senescence is known to be triggered by aging or autophagy suppression, in trophoblasts, and results in telomere shortening or dysfunction. This process is seen in early onset preeclampsia and FGR and is related to placental aging that accompanies the pro-inflammatory phenotype [62]. Senescent cells also alter their microenvironment by the secretion of proinflammatory cytokines, chemokines, growth factors, and proteases, collectively known as the senescence-associated secretory phenotype (SASP) [63]. Three pathways have been proposed for cellular senescence with

DNA damage: the p16INK4a pathway, the p53 pathway, and the autophagy-mediated GATA Binding Protein 4 (GATA4) pathway. Increased expression of p53, p21, and p16INK4a proteins has been reported in preeclampsia [64,65]. GATA Binding Protein 4, which is essential for embryonic development, is selectively degraded by p62 [66,67]. Therefore, GATA4 stabilization mediated by autophagy inhibition may contribute to cellular senescence with inflammation in preeclampsia.

#### *5.2. Gestational Diabetes Mellitus (GDM) and Obesity*

Gestational diabetes mellitus (GDM) is a type of diabetes that develops during pregnancy and affects 3–30% of pregnan<sup>t</sup> women [68–70]. Gestational diabetes mellitus increases the risk of fetal morbidity and mortality, as well as incidence of preeclampsia in mothers [69]. The role of autophagy in GDM remains controversial. Ji et al. reported that autophagy activation, manifested by increases in MAP1LC3-II and Atg5, and a decrease in p62, was observed in GDM placentas [71]. In addition, high glucose increased autophagy in HTR8/SVneo cells. Although the opposite result has also been reported, which included a decrease in BCLN1, and increases in MAP1LC3-II and p62 [72]. Placentas from obese women with GDM showed downregulation of protein kinase AMP-activated catalytic subunit alpha 2 (PRKAA2, also known as AMPK) and upregulation of mTOR which caused an increase in ribosomal protein S6 kinase B1 (RPS6KB1), suggesting autophagy inhibition in GDM placentas [73]. Muralimanoharan et al. constructed a labyrinth layer-specific Atg7 knockout mouse model on the basis of findings that autophagic activity decreased in the placentas of obese women [54]. Interestingly, weight gain in the offspring of animals with these knockout placentas was significantly greater than that in the wild type counterparts and was accompanied by hyperglycemia. This was thought to be due to greater sensitivity to a high-fat diet. Placental autophagy deficiency in this context supports the developmental origins of health and disease (DOHaD) hypothesis correlating poor fetal nutrition in utero with chronic diseases in adulthood such as obesity and certain cancers [74].
