Maternal hepatocytes heterogeneously and dynamically exhibit developmental phenotypes partially via yes-associated protein 1 during pregnancy

Abstract

Pregnancy induces reprogramming of maternal physiology to support fetal development and growth. Maternal hepatocytes undergo hypertrophy and hyperplasia to drive maternal liver growth and alter their gene expression profiles simultaneously. This study aimed to further understand maternal hepatocyte adaptation to pregnancy. Timed pregnancies were generated in mice. In a nonpregnant state, most hepatocytes expressed Cd133, α-fetal protein (Afp) and epithelial cell adhesion molecule (Epcam) mRNAs, whereas overall, at the protein level, they exhibited a CD133⁻/AFP⁻ phenotype; however, pericentral hepatocytes were EpCAM⁺. As pregnancy advanced, although most maternal hepatocytes retained Cd133, Afp, and Epcam mRNA expression, they generally displayed a phenotype of CD133⁺/AFP⁺, and EpCAM protein expression was switched from pericentral to periportal maternal hepatocytes. In addition, we found that the Hippo/yes-associated protein (YAP) pathway does not respond to pregnancy. Yap1 gene deletion specifically in maternal hepatocytes did not affect maternal liver growth or metabolic zonation. However, the absence of Yap1 gene eliminated CD133 protein expression without interfering with Cd133 transcript expression in maternal livers. We demonstrated that maternal hepatocytes acquire heterogeneous and dynamic developmental phenotypes, resembling fetal hepatocytes, partially via YAP1 through a posttranscriptional mechanism. Moreover, maternal liver is a new source of AFP. In addition, maternal liver grows and maintains its metabolic zonation independent of the Hippo/YAP1 pathway. Our findings revealed a novel and gestation-dependent phenotypic plasticity in adult hepatocytes.

NEW & NOTEWORTHY We found that maternal hepatocytes exhibit developmental phenotypes in a temporal and spatial manner, similarly to fetal hepatocytes. They acquire this new property partially via yes-associated protein 1.

INTRODUCTION

As pregnancy progresses, different maternal nonreproductive organs undergo various organ-dependent adaptive changes to meet the increasing needs of the developing and growing fetus. In pregnant women, adaptive changes in the maternal brain (1), pancreas (2–5), and spleen (6) have been revealed. Certain regions of the maternal brain undergo considerable structural changes, including a decrease in gray matter volume (1). β-Cell number in the maternal pancreas increases significantly throughout gestation (3, 5). Conceivably owing to the increase in maternal blood volume (7, 8), the maternal spleen displays a robust enlargement (6). In rodents, structural and/or functional adaptations to pregnancy have been observed in the maternal brain (9), liver (10–12), pancreas (13–15), heart (16, 17), and spleen (18). This phenomenal reprogramming of maternal physiology is believed to be essential for successfully establishing and maintaining gestation. However, these phenomena remain to be comprehensively characterized.

In 1944, Kennaway et al. (19) first observed that pregnancy-induced maternal liver enlargement in mice. In the last decades, studies have shown that the maternal liver exhibits two major adjustments to pregnancy. The first is dramatic and gestation stage-dependent changes in gene expression profiles, most prominently on nutrient metabolism, organ development, and cell proliferation (10, 12). The second is maternal hepatocyte hypertrophy and hyperplasia, which occur during the second half of pregnancy (10–12, 20). Our previous study demonstrated that the maternal liver coordinates adaptations of the maternal compartment with the placental compartment to ensure the health of the offspring (21). However, many more questions remain regarding the adjustment of the maternal liver to accommodate fetal development and growth. Therefore, this study aimed to understand further the response of the maternal liver to pregnancy. We found that maternal hepatocytes exhibited phenotypes similar to fetal hepatocytes, partially regulated by yes-associated protein 1 (YAP1).

MATERIALS AND METHODS

Animal Care and Use

Protocol for the care and use of animals was prepared as per National Institutes of Health guidelines, and all animals were maintained in accordance with the regulations outlined by the Indiana University-Purdue University Indianapolis Animal Care and Use Committee. Animals had ad libitum access to food and drinking water and were fed regular chow. The temperature and relative humidity of the animal rooms were maintained at 22 ± 1°C and 40%–60%, respectively. The lighting period comprised 12 h light/dark cycles.

To generate a timed pregnancy, virgin C57BL/6J female and male mice (Jackson Laboratories, Maine, Stock No.: 000664), aged between 3 and 3.5-mo old, were housed together. The estrous cycle (22, 23) and the duration of gestation (24) of the mice are shown in Fig. A1. The following day, the female mouse was inspected for a seminal plug, which indicated successful mating. The appearance of the seminal plug was denoted as gestation day (Gd) 1, and the male was separated from the cage. The Yap1^flox/flox mouse line was purchased from Jackson Laboratories (Stock No.: 027929). In this strain, the Yap1 gene is replaced by the YAP1-flox transgene, and all mice carry this transgene under homozygous conditions. The YAP1-flox transgene is constructed by inserting two loxP sites on either side of the region containing exons-1 and 2 of the YAP1 coding sequence. Maternal hepatocyte-specific YAP1 deletion was achieved by injecting mice with the adeno-associated virus serotype 8 (AAV8) with the thyroxine-binding globulin promoter (TBG) expressing Cre [AAV8-TBG-Cre virus (Addgene, AV-8-PV1091)] via tail vein at a dose of 1 × 10¹² genomic copies/mouse. AAV8-TBG-Null virus (Addgene, AV-8-PV0148) was used as a control.

Quantitative Real-Time Polymerase Chain Reaction

Total mRNA was extracted from livers (70–100 mg/liver) using TRIzol reagent. Briefly, liver tissues were homogenized using the TRIzol reagent. cDNAs were prepared using a Verso cDNA kit (No. AB-1453B, Thermo Scientific) following the manufacturer’s instructions. Quantitative real-time polymerase chain reaction (qRT-PCR) was performed using the TaqMan gene expression assay protocol. The master mix solution was prepared using molecular biology grade water, 2× Taqman gene expression master mix (No. 4369016, Applied Biosystems), and a 20× Taqman gene expression assay probe. The probes used are listed in Table 1. qRT-PCR was performed using the ABI 7300 real-time PCR system (Applied Biosystems) and CFX connect real-time system (Bio-Rad Laboratories).

Table 1.

List of primers used for real-time quantitative polymerase chain reaction

Gene Name	Assay ID	Catalog No.
Afp	Mm00431715_m1	4331182
Cd133	Mm00477115_m1	4331182
Epcam	Mm00493214_m1	4331182
Ctgf	Mm00515790_m1	4331182
Notch2	Mm00803077_m1	4331182
Yap1	Mm00494236_m1	4351372
Albumin	Mm00802090_m1	4331182

EpCAM, epithelial cell adhesion molecule; RT-qPCR, real-time quantitative polymerase chain reaction.

In Situ Hybridization

In situ hybridization (ISH) was performed using the RNAscope HD Reagent Kit-Brown (Advanced Cell Diagnostics). The kit comprised pretreatment reagents, probe detection reagents, wash buffer, and target mRNA probes. The RNAscope HD assay was conducted according to the manufacturer’s instructions with slight modifications. Specifically, to dehydrate tissue samples, slides were passed through an ethanol gradient of 70%, 85%, 95%, and 100% ethanol. Slides were incubated in an ethanol solution for 5 min. After dehydration, the tissues were cleared by incubating the slides in xylene for 5 min. The slides were mounted in a mounting medium. The results were documented as photomicrographs obtained using a Leica microscope.

Western Blotting

Western blotting (WB) assays were performed using three different protein extracts: 1) total protein, 2) cytoplasmic protein, and 3) nuclear protein. Total protein extracts were prepared using the T-PER Tissue Protein Extraction Reagent (No. 78510, Thermo Fisher Scientific). Cytoplasmic and nuclear protein extracts were prepared using the NE-PER Nuclear and Cytoplasmic Extraction Reagent kit (No. 78833, Thermo Fisher Scientific). The protocol provided by the manufacturer was followed. Protease and phosphatase activities were inhibited by adding the Halt protease and phosphatase inhibitor cocktail (100×; No. 78442, Thermo Fisher Scientific) at a final concentration of 1× to the tissue homogenization reagent or buffer. Protein concentrations were measured using the Pierce 660 nm protein assay (No. 22660, Thermo Fisher Scientific). Protein loading samples were prepared using molecular biology grade water, 4× lithium dodecyl sulfate sample buffer (No. NP0007, Thermo Fischer Scientific), β-mercaptoethanol, and the respective sample protein extract. The protein concentration of loading samples made from total and cytoplasmic protein extracts was 0.67 µg/µL, whereas for those made from nuclear extracts was 0.16 µg/µL. SDS-PAGE was performed using NuPAGE 4%–12% Bis-Tris Protein (No. NP0336BOX, Thermo Fisher Scientific) precast gels. After electrophoresis, the separated proteins were transferred onto polyvinylidene fluoride membranes. Membranes were blocked with either 5% milk or BSA in 1× Tris-buffered saline containing 0.1% Tween 20 (TBST), depending on the primary antibody. Details of the primary antibodies used are presented in Table 2. All secondary antibodies used were conjugated with horseradish peroxidase (HRP); the details are shown in Table 3. Immune complexes were detected using the SuperSignal West Pico PLUS Chemiluminescent Substrate (No. 34577, Thermo Fisher Scientific). Signals were detected using ImageQuant LAS 4000 Mini (General Electric Life Sciences) and quantified using ImageJ software.

Table 2.

List of primary antibodies, their dilutions, and applications

Antibody	Catalog No.	Vendor	Dilution	Application
BrdU	5292	Cell Signaling	1:100	IHC-P
GFP	ab6673	Abcam	1:200	IHC-P
AFP	LS-C123534	Lifespan Biosciences	1) 1:400	IHC-P
			2) 1:1,000	WB
CD133	PAB12663	Abnova	1) 1:3,000	IHC-P
			2) 1:1,000	WB
EPCAM	50591-R002	Sino Biological	1) 1:1,500	IHC-P
			2) 1:1,000	WB
Ki67	RM-9106-S1	Thermo Scientific	1:100	IHC-P
Beta-Catenin	610153	BD Transduction	1:100	IHC-P
MST1	3682T	Cell Signaling	1:1,000	WB
Phospho-MST1	3681S	Cell Signaling	1:1,000	WB
LATS1	3477T	Cell Signaling	1:5,000	WB
Phospho-LATS1	9157S	Cell Signaling	1:1,000	WB
MOB1	13730	Cell Signaling	1:1,000	WB
Phospho-MOB1	8699	Cell Signaling	1:1,000	WB
YAP1	14074	Cell Signaling	a) 1:200	IHC-P
			b) 1:1,000	WB
Phospho-YAP1	4911S	Cell Signaling	1:1,000	WB
TAZ	72804T	Cell Signaling	1:1,000	WB
Phospho-TAZ	59971S	Cell Signaling	1:1,000	WB
GAPDH	5174	Cell Signaling	1:1,000	WB
Arginase1	93668T	Cell Signaling	1:1,000	IHC-P
E-Cadherin	3195S	Cell Signaling	1:200	IHC-P
CYP1A2	A0062	Abclonal	1:200	IHC-P
CYP2E1	A2160	Abclonal	1:1,000	IHC-P
Glu. Synt.	Sc-376767	Abclonal	1:200	IHC-P

EpCAM, epithelial cell adhesion molecule; IHC, immunohistochemistry; WB, Western blotting.

Table 3.

List of secondary antibodies, their dilutions, and applications

Name	Catalog No.	Vendor	Dilution	Application
Bovine anti-Goat	sc-2378	Santa Cruz	1:20,000	WB
Goat anti-Rabbit	170-6515	Bio-Rad	1:20,000	WB
Horse anti-Mouse	BA-2000	Vector Laboratories	1:200	IHC-P
Donkey anti-Goat	705-065-147	Jackson ImmunoResearch	1:200	IHC-P
Goat anti-Rabbit	111-065-144	Jackson ImmunoResearch	1:200	IHC-P
Donkey anti-Mouse	715-585-151	Jackson ImmunoResearch	1:200	IHC-P
Donkey anti-Goat	705-545147	Jackson ImmunoResearch	1:200	IHC-P
Donkey anti-Rabbit	711-545-152	Jackson ImmunoResearch	1:200	IHC-P
Donkey anti-Rabbit	711-585-152	Jackson ImmunoResearch	1:200	IHC-P

IHC, immunohistochemistry; WB, Western blotting.

Immunohistochemistry

Immunohistochemistry (IHC) was performed using a standard routine staining protocol. Briefly, the tissue sections were deparaffinized and rehydrated using xylene and a series of alcohol solutions with decreasing concentration gradients (100%, 95%, 85%, and 70% alcohol). The antigenicity of the target proteins was improved or enhanced by heat-induced epitope retrieval. Tissue sections were boiled in citrate buffer (10 mM citric acid, 0.05% Tween-20, pH 6.0) for ∼20 min and allowed to cool gradually. Excess peroxidase activity was quenched using 0.3% hydrogen peroxide (H₂O₂). The quenching step was skipped for immunofluorescence staining, which used a fluorophore-tagged secondary antibody (2^oAb). After the quenching step, tissue sections were blocked with an appropriate 5% serum solution in 1X Dulbecco’s phosphate-buffered saline (1X DPBS) with calcium (Ca²⁺) and magnesium (Mg²⁺; No. 114-059-101, Quality Biological). Subsequently, the tissue sections were incubated with the appropriate primary antibody (1^oAb) overnight at 4°C. The following day, excess 1^oAb was washed with 1× phosphate-buffered saline (1× PBS), and tissues were treated with the appropriate 2^oAb. Detailed descriptions of the 1^o- and 2^o-Abs used are mentioned in Tables 2 and 3. For biotin-conjugated 2^oAbs, after incubation with 2^oAb, tissue sections were treated with avidin-horse radish peroxidase (HRP) conjugated solution and 3,3’-diaminobenzamide (DAB) substrate before being counterstained, dehydrated, and mounted. For fluorophore-conjugated 2^oAb, tissue sections were mounted directly using Prolong Gold mounting media containing DAPI counterstain (No. P36962, Invitrogen) after 2^oAb treatment.

Serum Collection and Blood Chemistry Analysis

Mice were anesthetized, and blood was extracted from the posterior vena cava. Blood samples were allowed to stand and coagulate for 30 min at room temperature, after which serum was separated by centrifugation. Serum samples were sent to Eli Lilly and Company (Indianapolis, IN) for blood chemistry analysis.

Statistical Analysis

We used the “Anova” test and the “Unpaired Student’s t test” (assuming unequal variance) to calculate the P value and to test for significance. The data obtained were considered statistically significant when compared with the control group (P < 0.05). The error bars in the bar graphs represent standard errors of the mean (SEM).

RESULTS

Maternal Hepatocytes Display Heterogeneous and Dynamic Developmental Phenotypes

Liver progenitor cells or fetal hepatocytes express CD133 (25), α-fetal protein (AFP; 26–30), and epithelial cell adhesion molecule (EpCAM; 31–35) during development (36). We previously showed that maternal livers highly express Cd133 mRNA in pregnant rats (10). This finding prompted us to examine the phenotypes of maternal hepatocytes by using these three markers throughout pregnancy in mice.

Maternal livers upregulated Cd133 mRNA and protein expression as the pregnancy progressed (Fig. 1, A and B). Cd133 transcript levels increased during the first half of gestation, peaked around midgestation, then gradually decreased, and eventually returned to the prepregnancy state before parturition (Fig. 1A). Maternal liver CD133 protein was rapidly detected after copulation (gestation day 1), progressively enriched, and abundant during the second half of pregnancy (Fig. 1B). Moreover, in the nonpregnant state, Cd133 mRNA was detected in almost all hepatocytes and mainly localized in the nucleus (Fig. 1C). During pregnancy (gestation day 15), maternal hepatocytes were Cd133 transcript-positive, resembling fetal hepatocytes (Fig. 1C). This mRNA molecule was not detected in the nonpregnant hearts (negative control organ; Fig. 1C). Before pregnancy, a few hepatocytes around the portal triads expressed CD133 (Fig. 1D). In contrast, during pregnancy, almost all maternal hepatocytes were CD133-positive, similar to the fetal hepatocytes (Fig. 1D). Our data revealed that, in response to pregnancy, maternal hepatocytes increased Cd133 mRNA expression, exhibiting a CD133 protein-positive phenotype.

Figure 1.

Cd133 mRNA and protein expression in nonpregnant (NP) and pregnant mouse livers. A: qRT-PCR analysis of maternal hepatic Cd133 mRNA expression throughout pregnancy. The data indicate the mean fold change relative to NP livers ± SE (n = 5 mice per time point). *P < 0.05; ***P < 0.001; ****P < 0.0001 compared with NP livers. B: Western blotting analysis of maternal hepatic CD133, AFP, and EpCAM protein expression throughout gestation. GAPDH was the internal loading control. Each lane represents a liver sample combined from five mice with equal amounts of proteins. C: Cd133 mRNA in situ hybridization of NP livers and gestation day (Gd) 6, 10, 15, and 18 maternal livers. Fetal liver was a positive control. NP heart was a negative control. Cd133 mRNA is indicated by brown-colored spots. D: immunohistochemistry analysis of CD133 in NP and Gd 6, 10, 15, and 18 maternal livers. Fetal livers were subjected to CD133 immunofluorescence (red) with DAPI staining for the nuclei (blue). AFP, α-fetal protein; CV, central vein; EpCAM, epithelial cell adhesion molecule; PV, portal vein.

Relative to the nonpregnant state, maternal livers expressed either unchanged or reduced Afp mRNA as pregnancy advanced (Fig. 2A). Inconsistently, maternal livers expressed abundant AFP from midgestation to term (Fig. 1B). Afp mRNA was detected in most hepatocytes before and after pregnancy (Fig. 2B). As expected, the nonpregnant heart (negative control organ) was Afp mRNA-negative (Fig. 2B), and the AFP protein was undetected in nonpregnant hepatocytes (Fig. 2C). However, AFP protein expression was observed in most maternal hepatocytes after midgestation (Gd15) and was subsequently restricted to a subpopulation of maternal hepatocytes, but at an overtly higher expression level (Gd18; Fig. 2C). In addition, most fetal hepatocytes were positive for Afp mRNA (Fig. 2B) and positive for AFP protein (Fig. 2C). Together, our data demonstrate that maternal hepatocytes exhibit a phenotype of AFP protein⁺, mimicking fetal hepatocytes.

Figure 2.

Afp mRNA and protein expression in nonpregnant (NP) and pregnant livers. A: qRT-PCR assay of maternal hepatic Afp mRNA expression throughout gestation. The data indicate the mean fold change relative to the NP state ± SE (n = 5 mice per time point). *P < 0.05; **P < 0.01 compared with NP livers. B: Afp mRNA in situ hybridization in NP livers and gestation day (Gd) 6, 10, 15, and 18 maternal livers. Afp mRNA is indicated by a brown color. Fetal liver was a positive control. NP heart was a negative control. C: immunohistochemistry analysis of AFP in NP livers and Gd 6, 10, 15, and 18 maternal livers. Fetal livers were subjected to AFP immunofluorescence (red) with DAPI staining for the nuclei (blue). AFP, α-fetal protein; CV, central vein; PV, portal vein.

Next, we analyzed Epcam mRNA and protein levels in the same setting. Compared with the prepregnancy state, Epcam mRNA expression was only transiently increased on gestation day 11 in maternal livers (Fig. 3A). However, relative to the nonpregnant state, EpCAM protein expression was overtly reduced after gestation day 6 but returned to equivalent levels before parturition (gestation day 18; Fig. 1B). In nonpregnant and pregnant livers, Epcam mRNA was detected in cholangiocytes, which was expected, and in most hepatocytes (Fig. 3B). Regardless of the physiological state, cholangiocytes were EpCAM protein-positive, as anticipated (Fig. 3C). In the nonpregnant state, EpCAM protein expression was predominant in pericentral hepatocytes (Fig. 3C). In striking contrast, as pregnancy advanced, EpCAM protein expression became predominant in periportal hepatocytes (Fig. 3C). It is known that, during the later liver development stage, EpCAM expression is diminished in fetal hepatocytes and restricted to fetal cholangiocytes (37). Thus, as expected, in embryonic day 18 fetal livers, Epcam mRNA and protein were only detected in cholangiocytes but not in hepatocytes (Fig. 3, B and C). Combining these data, we uncovered that pregnancy induces a switch of EpCAM protein expression from pericentral to periportal maternal hepatocytes.

Figure 3.

Epcam mRNA and protein expression in nonpregnant (NP) and pregnant livers. A: qRT-PCR analysis of maternal hepatic Epcam mRNA expression at various stages of gestation. The data indicate the mean fold change relative to the NP state ± SE (n = 5 mice per time point). *P < 0.05 compared with NP livers. B: Epcam mRNA in situ hybridization in NP livers and pregnant maternal livers. Epcam mRNA was detected as brown-colored spots. Fetal liver was a positive control. NP heart was a negative control. C: immunohistochemistry analysis of EpCAM in NP livers and Gd 6, 10, 15, and 18 maternal livers. Fetal livers were subjected to EpCAM immunofluorescence (red) with DAPI staining for the nuclei (blue). CV, central vein; EpCAM, epithelial cell adhesion molecule; PV, portal vein.

Pregnancy Does Not Affect the Activity of the Maternal Hepatic Hippo/YAP1 Pathway

The canonical Hippo pathway centrally comprises two kinases: macrophage stimulating 1/2 (MST1/2) and large tumor suppressor kinase 1/2 (LATS1/2). Phosphorylated and activated MST1/2 phosphorylates and activates LATS1/2. Activated LATS1/2 sequentially phosphorylates and inactivates the transcriptional coactivator YAP or its homolog tafazzin (TAZ), sequestering YAP/TAZ in the cytoplasm. However, dephosphorylation of MST1/2 and subsequent LATS1/2 leads to the dephosphorylation and activation of YAP/TAZ. Active YAP/TAZ is then transported to the nucleus, transactivating its target genes, including connective tissue growth factor (Ctgf) and Notch2 (38). The Hippo/YAP pathway regulates the growth of organs, including the liver (39), and modulates the phenotypic plasticity of mature hepatocytes (38). Thus, we reasoned that this pathway might control gestation-dependent maternal liver growth and maternal hepatocyte phenotypes. To test this hypothesis, we first evaluated how the components of this pathway respond to pregnancy.

Compared with the nonpregnant state, the expression of maternal hepatic cytoplasmic p-MST1, MST1, p-LAST1, LAST1, p-YAP1, YAP1, p-TAZ, and TAZ was either unchanged or reduced. In addition, the ratios of p-MAST1/total MAST1, p-LAST1/total LAST1, p-YAP1/total YAP1, and p-TAZ/total TAZ were not altered throughout gestation (Fig. 4). Moreover, although the expression of maternal hepatic nuclear YAP1 and TAZ remained the same as before pregnancy (Fig. 5A), the mRNA expression of maternal hepatic Ctgf and Notch2 was unaffected or even reduced as the pregnancy progressed (Fig. 5B). These data indicated that the activity of the Hippo/YAP1 pathway did not show a pregnancy-dependent change.

Figure 4.

Expression of cytoplasmic components of the Hippo/YAP1 pathway in maternal livers. Timed pregnancies were generated in 3–3.5-mo-old virgin female C57BL/6J mice. Nonpregnant (NP) and pregnant maternal livers (gestation days 8, 15, and 18) were collected, and cytoplasmic fractions were prepared. A: Western blotting analysis of the cytoplasmic components of the Hippo/YAP pathway. GAPDH was the internal loading control. B: quantification of relative protein expression of cytoplasmic components of the Hippo/YAP1 pathway. Relative fold changes were calculated after normalizing to GAPDH. Data are indicated as means ± SE (n = 3 mice per time point). *P < 0.05; **P < 0.01 compared with NP livers. YAP1, yes-associated protein 1.

Figure 5.

Expression of nuclear YAP1 and TAZ in maternal livers. Timed pregnancies were generated in virgin female C57BL/6J mice. Nonpregnant (NP) and pregnant maternal livers (gestation days 8, 15, and 18) were collected, and nuclear fractions were prepared. A: Western blotting analysis of the nuclear YAP1 and TAZ in maternal livers. The total nuclear protein was stained using ponceau S stain and was the internal loading control. B: quantification of relative protein expression of nuclear YAP1 and TAZ. Relative fold change was calculated after normalizing to total nuclear protein. Data are shown as means ± SE (n = 3 mice per time point). C: mRNA expression of maternal hepatic Ctgf and Notch2. Total mRNA was extracted from NP and pregnant maternal livers (gestation days 1, 4, 6, 8, 11, 13, 15, and 18) of C57BL/6J mice. qRT-qPCR was performed to quantify the mRNA levels of Ctgf and Notch2. Data represent the mean fold changes relative to NP ± SE (n = 5 mice per time point). *P < 0.05; **P < 0.01; ****P < 0.0001 compared with NP livers. TAZ, tafazzin; YAP1, yes-associated protein 1.

Presence of YAP1 is Essential for CD133 Protein Expression in Maternal Livers

Although pregnancy does not activate maternal hepatic YAP1, we evaluated whether pregnancy-dependent events in the maternal liver require the presence of YAP1. Timed pregnancies were generated in Yap1^f/f mice by mating them with wild-type male mice. We injected the AAV8-TBG-Cre virus into gestation day 6 Yap1^f/f mice and deleted the Yap1 gene specifically in maternal hepatocytes during the second half of pregnancy. The AAV8-TBG-null virus was used as a control. We collected samples on gestation day 18 for various endpoint analyses.

As anticipated, maternal hepatic Yap1 mRNA and protein expression were largely diminished in maternal hepatocyte-specific Yap1 knockout (hYap1^−/−) pregnant mice compared with Yap1^f/f pregnant ice. The data indicated the high efficiency of the virus-mediated loss of function of Yap1 (Fig. 6, A–D). In the maternal livers without Yap1, Cd133 mRNA levels were unaffected, whereas CD133 protein expression almost vanished (Fig. 6, A–D). The absence of Yap1 led to increased Afp transcript levels (Fig. 6B) but unchanged expression and distribution of AFP protein. The presence or absence of Yap1 did not affect the expression of Epcam mRNA and protein or the distribution of EpCAM protein (Fig. 6, A–D). Collectively, these data demonstrate that the lack of Yap1 posttranscriptionally prevents CD133 protein expression in maternal hepatocytes; therefore, Yap1 contributes to the regulation of pregnancy-dependent phenotypes of these cells.

Figure 6.

Effects of maternal hepatocyte-specific Yap1 gene deletion on maternal hepatocyte phenotypes. Timed pregnancies were generated in virgin female Yap1^f/f mice by mating with wild-type males. On gestation day (Gd) 7, mice were administered with the AAV8-TBG-Cre virus to delete the Yap1 gene specifically in maternal hepatocytes (Yap1^−/− mice) or administered the AAV8-TBG-Null virus to serve as the control mice (Yap1^f/f mice). Gd18 maternal livers were collected from both genotype groups of mice. A: Western blotting analyses of YAP1, CD133, AFP, and EpCAM expression in maternal livers. GAPDH was an internal loading control. B: qRT-PCR analyses of Yap1, Cd133, Afp, and Epcam mRNA expression in maternal livers. Data represent mean fold change relative to Yap1^f/f controls ± SE (n = 3 mice per genotype). *P < 0.05; **P < 0.01. C: quantification of YAP1, CD133, AFP, and EpCAM protein expression in maternal livers. Data were normalized with GAPDH. Data represent mean fold change relative to Yap1^f/f controls ± SE (n = 3 mice per genotype). **P < 0.01; ****P <0.0001. D: immunohistochemistry analyses of YAP1, CD133, AFP, and EpCAM protein distribution in maternal livers. AFP, α-fetal protein; CV, central vein; EpCAM, epithelial cell adhesion molecule; PV, portal vein; YAP1, yes-associated protein 1.

Furthermore, the loss of function of maternal hepatic Yap1 did not change maternal liver weight, maternal liver-to-total body weight ratio, or maternal hepatocyte density (Fig. 7A). These results implied that Yap1 might not mediate pregnancy-induced maternal liver growth. Blood chemistry analyses revealed that hYap1^−/− pregnant mice displayed increased serum alkaline phosphatase (ALP), decreased serum cholesterol, and unchanged other parameters compared with Yap1^f/f pregnant mice (Fig. 7B). These data suggested that maternal hepatic Yap1 is involved in cholesterol metabolism during pregnancy.

Figure 7.

Effect of maternal hepatocyte-specific Yap1 gene deletion on maternal liver growth and blood chemistry. Timed pregnancies were generated in virgin female Yap1^f/f mice by mating with wild-type males. On gestation day (Gd) 7, one group of mice was administered the AAV8-TBG-Cre virus to delete the Yap1 gene specifically in maternal hepatocytes (Yap1^−/− mice), whereas the other group was administered the AAV8-TBG-Null virus to act as the control group (Yap1^f/f mice). Gd18 maternal livers and blood were collected. A: maternal liver weights, maternal liver-to-total body weight ratios, and maternal hepatocyte densities were shown. B: maternal serum concentrations of a subset of blood chemistry parameters. Data are shown as means ± SE (n = 3 mice per genotype). *P < 0.05; **P < 0.01.

In addition, YAP1 has been shown to regulate liver metabolic zonation (40, 41). Deleting YAP1 from hepatocytes increases zone three hepatocytes expressing glutamine synthetase (GS; 40). Therefore, we first evaluated whether pregnancy affects the metabolic phenotypes of maternal hepatocytes by examining the distribution of pericentral hepatocyte markers CYP1A2, CYP2E1, and GS and the periportal hepatocyte markers arginase 1 and E-cadherin. The distribution of these marker proteins did not show overt gestation-dependent changes compared with those in the nonpregnant state (Fig. 8A). We then compared the distribution of these proteins between gestation day 18 Yap1^f/f and hYap1^−/− maternal livers. We also did not observe overt Yap1-dependent alterations (Fig. 8B). Hence, these data suggest that neither pregnancy nor YAP1 modulates maternal liver metabolic zonation or hepatocyte metabolic phenotypes.

Figure 8.

Effect of pregnancy and maternal hepatocyte-specific Yap1 gene deletion on maternal liver metabolic zonation. Livers of nonpregnant (NP) C57BL/6J mice and maternal livers of gestation day (Gd) 15 and 18 mice of the same strain were collected. A: distribution of pericentral hepatocyte marker CYP1A2, CYP2E1, and glutamine synthetase (GS) and periportal hepatocyte marker E-cadherin and arginase 1 was analyzed with immunohistochemistry. Timed pregnancies were generated in virgin female Yap1^f/f mice by mating with wild-type males. On gestation day (Gd) 7, mice were injected with the AAV8-TBG-Cre virus to delete the Yap1 gene exclusively in maternal hepatocytes (Yap1^−/− mice) or administered the AAV8-TBG-Null virus to act as the controls (Yap1^f/f mice). Gd18 maternal livers were harvested from both genotype groups of mice. B: distribution of CYP1A2, CYP2E1, GS, E-cadherin, and arginase 1 in maternal livers was visualized via immunohistochemistry. CV, central vein; PT, portal triad.

DISCUSSION

Most interestingly, we discovered for the first time that maternal hepatocytes acquire heterogeneous and dynamic developmental phenotypes during pregnancy. During development, fetal hepatocytes are initially Cd133⁺, Afp⁺, and Epcam⁺ and subsequently Cd133⁺, Afp⁺, and Epcam⁻ at the mRNA level. Correspondingly, fetal hepatocytes display an early-stage CD133⁺/AFP⁺/EpCAM⁺ phenotype and a later-stage CD133⁺/AFP⁺/EpCAM⁻ phenotype at the protein level (28–30, 32, 37). Surprisingly, in the nonpregnant state, most adult female hepatocytes retain the mRNA expression of Cd133, Afp, and Epcam. However, in this state, all hepatocytes are AFP⁻, the vast majority of which are CD133⁻, and pericentral hepatocytes are EpCAM⁺. Remarkably, pregnancy induces most maternal hepatocytes to express CD133 and AFP proteins, thus allowing them to exhibit a developmental phenotype of CD133⁺/AFP⁺, simultaneously switching EpCAM protein expression from pericentral to periportal hepatocytes. Obviously, these observations imply a pregnancy-dependent mechanism associated with posttranscriptional regulation of the expression of these genes. This process is dynamic. On gestation day 15, most maternal hepatocytes are CD133⁺, whereas, on gestation day 18, all maternal hepatocytes are CD133⁺. In contrast, on gestation day 15, most hepatocytes are AFP⁺, whereas, on gestation day 18, AFP expression is restricted to a randomly distributed subpopulation of maternal hepatocytes. It is well known that adult hepatocytes exhibit heterogeneous metabolic phenotypes. In this study, we revealed heterogeneous developmental phenotypes in this particular physiological state (pregnancy). The exhibition of behaviors by maternal hepatocytes that partially resemble fetal hepatocytes is intriguing. The biological significance of this new feature of adult hepatocytes is unclear and requires further investigation.

The phenotypic changes of maternal hepatocytes occur in the second half of pregnancy. During this period, maternal hepatocytes are exposed to an environment enriched in hormones associated with pregnancy, such as estrogen, progesterone, and placental lactogens. In the nonpregnant state, during the estrous cycle, blood levels of estrogen, progesterone, and gonadotrophins fluctuate in large magnitudes. In our current studies, we did not examine the stages of the estrous cycle in the virgin control mice. To the best of our knowledge, whether these hormones regulate the expression of CD133, AFP, and EpCAM in the liver or other tissues remain elusive.

We also revealed for the first time that maternal hepatocytes produce AFP. AFP is a fetal plasma glycoprotein primarily synthesized by developing fetal hepatocytes and, to a lesser extent, by yolk sac cells. Its presence in the maternal serum is due to fetal and maternal blood exchange during gestation. Clinically, it is used to assess the health of the developing fetus and pregnancy. Abnormally high levels of AFP in maternal serum are considered a potential indicator of various fetal defects or diseases (42). Our findings revealed the maternal liver as an additional source of AFP, implying that abnormally high AFP levels could also be a consequence of maternal liver pathology. Therefore, it is critical to verify this finding in humans. However, obtaining human maternal liver samples is challenging because biopsying this organ is not a routine clinical practice.

The Hippa/YAP pathway modulates liver growth, liver metabolic zonation, and hepatocyte phenotype in adults (43–47). However, our studies indicated that the maternal liver does not rely on this pathway to grow and maintain its metabolic landscape. We arrived at this conclusion based on two pieces of evidence. First, the pathway remains silent in response to pregnancy. Second, deletion of the Yap1 gene, specifically in maternal hepatocytes, does not affect maternal liver growth or metabolic zonation. However, genetic manipulation of the Yap1 gene substantially diminished CD133 protein expression without affecting Cd133 mRNA expression in maternal hepatocytes. Therefore, we can conclude that the presence of YAP1 is indispensable for posttranscriptional expression of CD133 protein and that pregnancy modulates maternal hepatocyte phenotypes partially via YAP1 at the posttranscriptional level. This finding enabled us to gain some mechanistic insight into the regulation of pregnancy-dependent phenotypic plasticity in maternal hepatocytes. In addition, YAP1 has been reported to regulate lipid metabolism, and activation of YAP1 in hepatocytes causes fatty liver (48–51). In this study, we observed that the absence of YAP1 in maternal hepatocytes decreased cholesterol concentrations in the maternal circulation, consistent with these studies. Our findings warrant further studies on how YAP1 is required to maintain pregnant-dependent maternal lipid homeostasis.

In summary, we have demonstrated a novel aspect of pregnancy physiology. As pregnancy progresses, maternal hepatocytes heterogeneously and dynamically adopt developmental phenotypes that mimic fetal hepatocytes, partially via YAP1. In addition, the growth of the maternal liver and the maintenance of its metabolic zonation are independent of the Hippo/YAP1 pathway. These findings enabled us to understand further how the maternal liver adapts to pregnancy.

GRANTS

This research was supported by the National Institutes of Health Grant 1R01DK117076 (to G. Dai).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

S.M.N. and G.D. conceived and designed research; S.M.N., J.L., J.A.Y., V.G., and H.J. performed experiments; S.M.N., J.L., J.A.Y., V.G., and H.J. analyzed data; S.M.N., J.L., and G.D. interpreted results of experiments; S.M.N. prepared figures; S.M.N. and G.D. drafted manuscript; S.M.N. and G.D. edited and revised manuscript; G.D. approved final version of manuscript.

APPENDIX

Figure A1 shows the estrous cycle (22, 23) and the duration of gestation (24) of the mice.

Figure A1.

The lengths of the estrous cycle and pregnancy of the C57/BL6 mice. The estrous cycle in the mice have four stages, sequentially from the proestrus stage to the estrous, metestrus, and diestrus stage. The estrous cycle repeats every 4–5 days with variabilities in duration in each stage. The gestation length is 20 days.