Background

Intrahepatic cholestasis of pregnancy (ICP) is a pregnancy-specific liver disease characterized by skin pruritus and abnormal liver function, such as elevated liver enzymes and increased serum TBA (≥ 10 μmol/L), that appears in the second and third trimesters of pregnancy [1]. The symptoms and biochemical abnormalities usually rapidly disappear in the early postpartum period [2]. The incidence of ICP disease ranges from 1% to 15.6% depending on geographical location [3,4,5]. The recurrence rate of ICP in the next pregnancy reaches as high as 40–60% [1]. ICP has been associated with adverse perinatal outcomes, including premature birth and intrauterine death [1, 6, 7]. An elevated level of serum TBA will increase the risk of premature delivery and stillbirth [8, 9]. Therefore, untangling the genetic basis of ICP disease is very important.

Obviously, ICP is a complex disease that depends on multiple factors, including genetic background, metabolites of progesterone, oestrogens, seasons and environmental background [4, 10, 11]. Among them, familial clustering analysis in pedigree studies indicated a genetic predisposition for ICP disease [12,13,14]. To date, several bile acid homeostasis-related genes, including NR1H4, ATP Binding Cassette Subfamily B Member 4 (ABCB4), ATP Binding Cassette Subfamily B Member 11 (ABCB11) and ATP Binding Cassette Subfamily C Member 2 (ABCC2), have been reported. Moreover, multiple previous studies have identified genetic variants of the NR1H4, ABCB4, ABCB11 and ABCC2 genes that contribute to the development of ICP [15,16,17,18,19,20]. Among them, NR1H4 plays a central role in regulating bile acid metabolism.

NR1H4 is both a key modulator of hepatocyte-protective pathways and a therapeutic target for cholestatic liver disease [21]. NR1H4 is a BA-activated transporter factor that is responsible for BA homeostasis and acts by binding to DNA response elements through the NR1H4 DNA binding domain (DBD) in the promoter of target genes (such as ABCB4, ABCB11 and ABCC2), thereby activating their transcription [22,23,24]. Moreover, the C-terminal region of NR1H4 has a highly conserved ligand binding domain (LBD), which determines the specificity of NR1H4 ligands. These ligands include farnesoid derivative, BA, unsaturated fat, hepatocyte factor-1 and steroid compound [25, 26]. NR1H4 has four different isoforms: α1, α2, α3 and α4. The first two isoforms, which are expressed in the human liver, have a different N-terminus than the other two isoforms [27, 28]. In liver tissue, when raising hepatocyte BA levels, NR1H4 regulates bile flow by directly inducing gene expression (ABCB4, ABCB11 and ABCC2) to stimulate hepatic bile export [29, 30]. Conversely, NR1H4 represses the expression of bile acid import (NTCP) [31] and key enzymes (CYP7A1 and CYP8B1) [32] in the bile acid synthesis pathway through the induction of short heterodimer partner (SHP) [31] in the liver and growth factor 19 (FGF19)/FGF15 [33] in the intestine. In addition, NR1H4−/− transgenic mice exhibited BA pool sizes [34]. Therefore, NR1H4 maintained a stable TBA level in hepatocytes by regulating TBA synthesis, transport, secretion and metabolism.

Considering that women with ICP exhibited elevated serum BAs and NR1H4 mutations resulted in altered BA levels, we hypothesized that NR1H4 mutations might also exist in ICP samples. Here, we recruited a total of 197 Han Chinese women with ICP and analysed the entire coding region of the NR1H4 gene. A total of 4 mutations, including two novel missense mutations in NR1H4, were identified in our ICP samples for the first time.

Methods

Samples and features

We recruited 197 patients diagnosed with ICP disease based on clinical symptoms (skin pruritus) and laboratory investigations (fasting TBA ≥ 10 µmol/L, etc.) between 2018 and 2020. Peripheral blood samples from 197 patients with ICP disease were collected from the Department of Obstetrics, Jiangxi Provincial Maternal and Child Health Hospital in Nanchang, China. In addition, we recorded a total of twenty-nine available clinical characteristics, which included age, body mass index (BMI), gestational weeks at diagnosis, gravidity and parity; the level of ion concentration covering K, Na, Cl, Ca, Mg and P; the counts of white blood cells (WBCs), red blood cells (RBCs), platelets (PLTs), and red blood cell distribution width. SD (RDW-SD); the level of serum biochemical indices including TBA, aspartate transaminase (AST), alanine transaminase (ALT), total bilirubin (TBIL), direct bilirubin (DBIL), indirect bilirubin (IDBIL), total cholesterol (CHOL), triglyceride (TG), high-density lipoprotein (HDL), low-density lipoprotein (LDL), uric acid (UA); and the outcomes of pregnant women and newborn babies, including birth weight, bleeding count and Apgar score. The clinical features were determined as described previously [20, 35]. Briefly, the ion concentration and serum biochemical index were examined by an AU5800 automatic biochemical analyser (Beckman Coulter, Inc., USA). Routine blood tests were determined by a Sysmex-xn-2000 automatic blood cell analyser (Sysmex Corporation, Japan).

Summary statistics for all the above clinical features investigated in 197 ICP patients are shown in Table 1. Of these samples, 151 clinical data points were described in our previous study [20, 35]. In addition, 1029 samples without ICP disease were also recruited. The present study followed the tenets of the Helsinki Declaration, and the ethics approval was approved by the Institutional Review Board of Jiangxi Provincial Maternal and Child Health Hospital in China. Each participating woman gave written informed consent (Additional file 1).

Table 1 Descriptive statistics of twenty-nine clinical characteristics of 197 ICP patients
Full size table

Mutation analysis

To excavate the potential mutations of the NR1H4 gene in 197 samples with ICP disease, we designed a total of nine pairs of primers (Table 2) to sequence the entire coding regions of NR1H4 through PCR and Sanger sequencing. Briefly, 197 genomic DNA samples were isolated from peripheral blood using an Axy Prep Blood Genomic DNA Mini Prep Kit (Item No. 05119KC3, Axygen Scientific, Inc., Union City, CA, USA). A total of 25 µL PCR system, including 2 µL total 100 ng DNA, 0.5 µL of each forward and reverse primer (2 μM), 12.5 µL mixed comprising Mg2+, dNTPs and Taq polymerase (Takara Biotechnology Co., Ltd., Dalian, China), and 9.5 µL ddH2O were mixed in a reaction tube. Touch down procedures were used for PCR amplification as follows: first, DNA was initially denatured at 94 °C for 5 min, followed by 26 cycles of 94 °C for 30 s, (68–0.5) °C for 30 s, and 72 °C for 45 s, after which 19 thermal cycles for 94 °C for 30 s, 55 °C for 30 s, and 72 °C for 45 s, a final extension stage of 72 °C for 10 min, and storage at 4 °C. The obtained PCR products were then examined by 1% agarose gel electrophoresis and sequenced by an ABI 3730 Genetic Analyser (Thermo Fisher Scientific, Inc., Waltham, MA, USA). The potential mutations of the NR1H4 gene were detected by comparative analysis of 197 samples with ICP disease and 1029 controls without ICP. The mutation site was searched by bidirectional sequencing.

Table 2 Primers used for sequencing the coding regions of the NR1H4 gene
Full size table

Evolutionary conservation analysis

The evolutionary conservative analysis of p.S145Fand p.M185L were performed in 26 representative species, including Chimpanzee, Gibbon, Macaque, Olive baboon, Gelada, Marmoset, Prairie vole, Mouse, Rat, Alpine marmot, Rabbit, Domestic yak, Cow, Goat, Sheep, Sperm whale, Arabian camel, Chacoan peccary, Pig, Dog, Dingo, Cat, Leopard, Horse and Elephant, through the genomic alignments of the Ensembl Genome Browser.

Protein structural modelling

The protein template of modelling between the reference and modified (p.S145F and p.M185L) mutations of the NR1H4 gene were conducted using the SWISS-MODEL repository database (http://www.expasy.org/). Then, we compared the protein models simultaneously with the Chimera 1.14rc package.

Statistical analysis

The summary function was used to perform the descriptive statistics on the clinical data of 197 samples with ICP disease. The t.test function was conducted to analyse the potential association of 29 clinical data between ICP samples with or without NR1H4 mutations. The P values were two sided, and the results were considered significantly different at P < 0.05. The frequency significant difference for NR1H4 mutations between 197 ICP samples and databases were analysed by Fisher’s test function. All the analyses were completed with R software. Logistic regression analysis was performed to assess the clinical parameters (age, gestational age, BMI, gravidity and parity) with the mutations.

Results

NR1H4 mutations

We sequenced 9 exon fragments of the NR1H4 gene and detected a total of four mutations, including three missense mutations in exons 2, 3 and 4 and one synonymous mutation in exon 5 with 3 samples in 197 ICP patients.

Two out of three missense mutations were novel (novel-1, novel-2) (Fig. 1, Additional file 1, Table 3) and were identified in a 40- and 21-year-old ICP individual, respectively. Using the web-available tools SIFT and PolyPhen2, the influence of the two novel mutations on protein function was predicted to be damaging. Furthermore, these two mutations were absent from 1029 controls without the ICP, 1000G_ALL (http://www.internationalgenome.org/), and ExAC (http://exac.broadinstitute.org/) databases. There was a significant difference (P = 0.018) in the frequency for two novel mutations between 197 ICP samples and the ChinaMAP (http://www.mbiobank.com/) database.

Fig. 1
figure 1

Sequencing electropherograms of two novel mutations (novel-1 and novel-2) in the NR1H4 gene. The mutation location is marked with an arrow. For novel-1, Y represents C or T in the same sequence. For novel-2, W represents A or T in the same sequence. Novel-1 mutation from C to T and Novel-2 from A to T occurred at the 434rd and 553rd bases in the CDS region of the NR1H4 gene, respectively. The corresponding amino acids changed from serine (S) to L-phenylalanine (F) in the 145th location and methionine (M) to leucine (L) in the 185th location

Full size image
Table 3 Screening for mutations in the NR1H4 gene in 197 pregnant women with ICP disease
Full size table

The other missense mutation rs180957965 (p.Ala230Ser) was identified in a 30-year-old sample (ICP12), and the synonymous mutation rs147030757 (p.Asn275Asn) were identified in three ICP patients (ICP1, ICP69 and ICP107). These mutations were all absent in the controls and had a low frequency of databases, ranging from 0.00018 to 0.0057. There was a significant difference in the frequency of the missense mutation rs180957965 (P = 0.036) and the synonymous mutation rs147030757 (P = 1.63e−05) between 197 ICP patients and the ExAC database. In addition, rs147030757 showed a significant frequency difference between the ICP population and 1000G_ALL (P = 0.001).

Clinical features of ICP patients with NR1H4 mutations

The clinical and biochemical features of the six ICPs with 4 mutations are presented in Table 4. Serum bile acids were increased in all six patients with NR1H4 mutations. The serum TBA levels of the patients identified with novel-1 and novel-2 were 46.4 and 113.2 μmol/L, respectively (Table 4). The patient with novel-1 had one child after experiencing six previous pregnancies. The TBA level of the patient ICP12 with the missense mutation rs180957965 was 12 μmol/L, and ICP1, ICP69 and ICP107 patients with a synonymous mutation rs147030757 were had TBA levels of 18.9, 27.5 and 46.4 μmol/L, respectively. Furthermore, the concentrations of CHOL and TG for the six patients with NR1H4 mutations were higher than the reference values (CHOL: 0–5.2 mmol/L; TG: 0.34–1.69 mmol/L).

Table 4 Clinical and biochemical data in the individuals with four mutations covering six patients in the NR1H4 gene
Full size table

Evolutionary conservative analysis and protein structural modelling

Evolutionary conservation analysis showed that these two novel mutations (p.S145F and p.M185L) were highly conserved among the 26 species, ranging from human to elephant (Fig. 2).

Fig. 2
figure 2

Evolutionary conservation analysis of the NR1H4 p.S145F and p.M185L mutation among 26 vertebrates, ranging from chimpanzee to elephant. The amino acids serine (S) and methionine (M) in the red horizontal line were highly conserved

Full size image

To further investigate the possible effects of the p.S145F and p.M185L variants on protein structure, the reference and the modified protein structure of NR1H4 gene were compared using UCSF Chimera 1.14rc. These two variants were located in the DNA binding region of the NR1H4 gene (Fig. 3A). For the variant p.S145F, compared with the reference 3D model of protein structure, the mutation has a slight change in the chemical bond in the two zinc finger structures rich in Cys amino acids at positions 137, 140, 154, 173 and 192 (Fig. 3B). Similarly, for another novel missense mutation, p.M185L, there is a change in the chemical bond at positions 137, 157, 189 and 192 (Fig. 3C).

Fig. 3
figure 3

The genetic features of NR1H4. A The distribution of the NR1H4 variants. NR1H4 is a 486-amino acid protein containing two DBD regions and one LBD region. Schematic representation of NR1H4 NM_001206993.1 cDNA and protein showing the locations of two novel possible pathogenic variants p.S145F and p.M185L detected in two out of 197 patients with ICP disease. Effects of NR1H4 p.S145F. B and p.M185L variants. C on the protein structure. The three-dimensional models of reference and modified (p.S145F and p.M185L) NR1H4 showed gold and blue rounded structures, respectively. The enlarged portion showed that the two DBD regions have small changes in the chemical bond lengths. DBD: DNA-binding domain; LBD: ligand-binding domain. D Comparison of the expression level of the NR1H4 gene between two healthy pregnant women and 4 patients with ICP. The expression level of NR1H4 was higher in the ICP group than in the healthy group. The difference did not reach the significance level (P = 0.22)

Full size image

To further explore the genetic basis of NR1H4, we analysed the mRNA expression level of the NR1H4 gene in placental tissue between two healthy pregnant women and four patients with ICP using NCBI GEO databases (GEO accession: GSE46157) from the Du Q et al. report [36]. The results showed that the expression of NR1H4 was upregulated in the ICP group (Fig. 3D), even though the difference was not significant (P = 0.22).

Correlation analysis

The potential correlation of NR1H4 four mutations and 29 available clinical and laboratory data are presented in Table 5. The results showed that the mutation group had higher TBA levels, TBIL levels, and bleeding amounts and a lower Apgar score. In addition, it was found that only the level of Na ions was significantly (P = 0.014) higher in the mutation group (139.50 mmol/L) than in the wild-type group (137.27 mmol/L). The associations between the clinical parameters (age: odds ratio (OR) = 0.965; 95% confidence intervals (CI): 0.823–1.132; gestational age (OR = 1.001; 95% CI: 0.982–1.019); BMI (OR = 0.806, 95% CI: 0.605–1.074); gravidity (OR = 1.143, 95% CI: 0.720–1.814); parity (OR = 1.398, 95% CI: 0.572–3,416) and the mutations were shown by logistic regression analysis.

Table 5 The potential correlation of NR1H4 mutations with clinical and laboratory data in 197 ICP patients
Full size table

Discussion

NR1H4 is required for the basal maintenance of enterohepatic circulation and is responsible for bile acid homeostasis. Milona et al. reported that increased hepatic bile acid concentrations during pregnancy in mice are associated with reduced NR1H4 function [24], which is consistent with the results of Castano et al. [37]. Castano et al. also demonstrated that impaired NR1H4 function during pregnancy may be associated with elevated levels of serum bile acids [37]. Our results also found that the expression level of NR1H4 was higher in the ICP group than in the normal group using GEO data. Furthermore, previous studies have demonstrated that functional variants in NR1H4 are associated with ICP disease/progressive familial intrahepatic cholestasis [16, 21]. In this study, we also detected four mutations, including three missense mutations, S145F, M185L, and rs180957965, and one synonymous mutation, rs147030757. Saskia et al. identified the missense variant M173T in NR1H4 and conducted cell function analysis [16]. They found that the M173T variant located in the DBD region caused lower transcription levels of bile acid transport-related genes, including ABCB11 and IBABP. In the present study, the two novel mutations S145F and M185L were also located in the first and second zinc finger of the DBD of NR1H4. The mutant has a slight change in the chemical bond of the structure for the NR1H4 gene compared to the wild-type (Fig. 3B, C). Therefore, we speculated that NR1H4 mutations result in changes in NR1H4 function (Fig. 3D) and the expression level of its target genes, thus increasing the level of bile acids in vivo. The exact mechanism of action remains elusive and requires further experimental study.

To date, an increasing number of researchers have found rare (MAF < 0.01) and low-frequency (0.05 ≤ MAF ≤ 0.01) variants associated with human pregnancy diseases, such as spontaneous preterm birth, cardiomyopathy and preeclampsia, by whole-exome sequencing [38, 39]. Consistent with this, in our study, frequency analysis of all four mutations in NR1H4 in 197 ICP samples, 1029 controls and 3 website databases covering much larger cohorts suggests that these variants are rare. The allele frequencies of the three missense mutations (MAF = 0.0025) and one synonymous mutation (MAF = 0.007) were lower in this study. According to previous studies, low-frequency and rare variants with large effect sizes contribute to complex traits and diseases [40,41,42]. Therefore, we hypothesized that the allele frequency and the size effect of mutations have a larger effect on TBA levels. In this study, combining the prediction results with the website available tools SIFT and PolyPhen2 and protein structural modelling, we suspected that the novel mutations contributed more to the development of ICP than the other two. Therefore, it is also likely reasonable that there is no significant difference in TBA levels between wild-type and NR1H4 mutations even though the mutation group tended to be associated with higher TBA levels when considering the allele frequency and size effect. Except for the ICP caused by the NR1H4 mutations, we speculated that other gene mutations (such as ANO8, ATP-binding cassette transporter family, bile acid receptors) [20, 35, 43], epigenetic regulators (microRNAs, DNA methylation and histone modification) [44,45,46], oestrogen and progesterone sulfate metabolites [10, 47], hypoxia [48] and the immune system [49], among other factors [50], may be responsible for the remaining ICP patients in this study.

Considering that BAs are toxic to the body, the excessive increase in BA levels has been depicted in different pathological contents. Moreover, several previous studies demonstrated that BAs have the ability to promote lipid absorption and biliary cholesterol secretion [16, 51, 52], indicating that BAs are associated with abnormalities in lipids. Saskia et al. reported that six out of 11 pregnant women with ICP having NR1H4 variants had symptomatic gallstones [16], and the remaining five did not have gallstone symptoms but had a family history of gallstones. The formation of gallstones is likely determined by the relative concentrations of TBA, CHOL and phospholipids in bile. In the present study, according to the clinical characteristics of 6 ICP cases with NR1H4 mutations, we found that the TBA levels, CHOL levels and TG levels were higher than the reference values. Therefore, we speculated that these ICP cases with NR1H4 variants have a high risk for gallstones. Bergheim et al. demonstrated that the possible mechanism of gallstones is the decrease in the expression of the NR1H4 gene [53]. Furthermore, Moschetta et al. prevented cholesterol gallstone disease by NR1H4 agonists in a mouse model, indicating that NR1H4 could be associated with cholesterol [54]. In addition, NR1H4 dysfunctions may occur during the progression associated with inflammatory bowel disease, colorectal cancer in the gut [55, 56], fibrosis and hepatocellular carcinoma in the liver [57, 58]. These results suggest that the variants affecting the structure and functions of NR1H4 lead to gut-liver axis diseases, and in the future, NR1H4 will be proposed as an emerging therapeutic target for both cholestatic and multiple metabolic diseases.

Our present study had several advantages. First, to our knowledge, only a few pathogenic mutations of the NH1R4 gene, such as M173T, R176* and Tyr139_Asn140insLys, have been identified thus far [16, 21]. Our findings broaden our understanding of the mechanism of NR1H4’s action on ICP disease. Second, NR1H4 mutations have been detected in ICP families [16, 21]. To date, no studies have uncovered genetic mutations in NR1H4 genes of hepatic disease among pregnant patients from a relatively large nationally representative sample (n = 197) in China and 1029 local healthy pregnant women. Third, the 29 clinical data of 197 ICP patients are relatively complete, which provides data supporting correlation analysis between mutations and clinical data. However, even though our results provided possible pathogenic variants, the causality between the two potential interesting candidate loci and ICP disease needs to be verified by validation functional experiments.

Conclusions

In summary, we reported two potential damaging mutations (p.S145F and p.M185L) in the NR1H4 gene in two out of 197 Chinese patients with ICP for the first time. Our findings provide new insights into the genetic architecture of ICP disease and suggest potential candidate variant targets for ICP clinical treatment.