Gnpat does not play an essential role in systemic iron homeostasis in murine model

Abstract The GNPAT variant rs11558492 (p.D519G) was identified as a novel genetic factor that modifies the iron‐overload phenotype in homozygous carriers of the HFE p.C282Y variant. However, the reported effects of the GNPAT p.D519G variant vary among study populations. Here, we investigated the role of GNPAT in iron metabolism using Gnpat‐knockout (Gnpat−/−), Gnpat/Hfe double‐knockout (Gnpat−/−Hfe−/− or DKO) mice and hepatocyte‐specific Gnpat‐knockout mice (Gnpatfl/fl;Alb‐Cre). Our analysis revealed no significant difference between wild‐type (Gnpat+/+) and Gnpat−/− mice, between Hfe−/− and DKO mice, or between Gnpatfl/fl and Gnpatfl/fl;Alb‐Cre with respect to serum iron and tissue iron. In addition, the expression of hepcidin was not affected by deleting Gnpat expression in the presence or absence of Hfe. Feeding Gnpat−/− and DKO mice a high‐iron diet had no effect on tissue iron levels compared with wild‐type and Hfe−/− mice, respectively. Gnpat knockdown in primary hepatocytes from wild‐type or Hfe−/− mice did not alter hepcidin expression, but it repressed BMP6‐induced hepcidin expression. Taken together, these results support the hypothesis that deleting Gnpat expression has no effect on either systemic iron metabolism or the iron‐overload phenotype that develops in Hfe−/− mice, suggesting that GNPAT does not directly mediate iron homeostasis under normal or high‐iron dietary conditions.

of the HFE p.C282Y variant. 19,20 Given the low penetrance of this variant, several groups have attempted to identify additional genetic loci 21 that affect iron status both in the general population [10][11][12] and among homozygous carriers of the HFE p.C282Y variant. 13,15 Using exome sequencing, McLaren et al 22 recently found that the prevalence of the p.D519G variant in GNPAT (glyceronephosphate O-acyltransferase) is higher among males who are homozygous carriers of the HFE p.C282Y variant and present with a severe iron-overload phenotype, suggesting that GNPAT might serve as a genetic modifier of the iron-overload phenotype in these individuals. Interestingly, even among healthy individuals who do not carry the HFE p.C282Y variant, an association has been found between the GNPAT p.D519G variant and changes in several iron parameters. 23,24 Nevertheless, the precise role of the GNPAT p.D519G variant in homozygous carriers of the HFE p.C282Y variant is highly controversial. [25][26][27][28][29][30] GNPAT encodes glyceronephosphate O-acyltransferase, the first enzyme in the ether lipid biosynthesis pathway. In humans, mutations in GNPAT have been associated with rhizomelic chondrodysplasia punctata, a condition characterized by severely impaired endochondral bone formation, rhizomelic shortening of the femur and humerus, vertebral disorders, dwarfism, cataract, cutaneous lesions, facial dysmorphism and severe mental retardation with spasticity. 31 Consistent with the important role of GNPAT in development, Gnpat-deficient mice have growth deficits, male infertility, structural abnormalities in the cerebellum, and impaired myelination and Schwann cell development. 32,33 Recently, Gnpat-knockout mice were also used to demonstrate that Gnpat plays a protective role in muscle strength 33 and both hepatic steatosis and steatohepatitis. 34 Despite epidemiological evidence supporting a role of Gnpat in HFE-linked haemochromatosis, its functional role in regulating iron has not been investigated in vivo. To develop a model for functionally characterizing the role of GNPAT in iron metabolism, we generated Gnpat-knockout (Gnpat −/− ) mice and Gnpat/Hfe double-knockout mice (Gnpat −/− Hfe −/− , hereafter referred to as simply DKO mice).
using the Cre-LoxP recombination approach (see Figure 1A). In brief, two LoxP sites flanking exon 4 in the Gnpat gene were introduced into embryonic stem cells derived from C57BL/6J129S3 mice. Chimeric mice were obtained by injecting targeted embryonic stem cells into C57BL/6 blastocysts. Chimeric mice containing the Gnpat-flox allele were crossed with C57BL/6J mice to obtain Gnpatflox mice. Mice carrying the Gnpat-flox allele were then crossed with CMV-Cre transgenic mice 35 on a C57BL/6J background to generate Gnpat −/− mice. Hfe −/− mice were kindly provided by Dr Nancy C.

| Serum iron, tissue non-heme iron and haematology parameters
Tissue non-heme iron was measured as previously described. 37 To measure serum iron and transferrin saturation, whole blood was collected by heart puncture and allowed to coagulate at room temperature for 2 hours; serum was then obtained by centrifugation at 2400 g. Serum iron (SI) concentration and unsaturated iron-binding capacity (UIBC) were measured using a colorimetry-based assay

Immunohistochemical detection of intestinal ferroportin and Perls'
Prussian blue iron staining were performed as previously described. 38 The protein expression levels of ferroportin are quantified using ImageJ software.

| Plasmids and cell culture
The wild-type GNPAT cDNA (Ensembl transcript ID: ENST00000366647.8) was cloned from the HepG2 cell line. The GNPAT cDNA encoding the p.D519G mutation was generated by site-directed mutagenesis. Both the wild-type and p.D519G GNPAT cDNAs were cloned into the pm-Cherry vector. The sequences of both constructs were confirmed by sequencing. For transfection experiments, HepG2 and Huh-7 cells were cultured in 6-well plates at 37°C in humidified air containing 5% CO 2 .
The culture medium (Dulbecco's modified Eagle's medium containing 4.5 g/L glucose) was supplemented with 10% fetal calf serum and 1% penicillin-streptomycin. The cells were cultured for 24 hours, mock-transfected or transfected with 10 ng/mL BMP4 (as a positive control), wildtype GNPAT, or p.D519G GNPAT using Lipofectamine 3000 (Thermo Fisher Scientific), then cultured for an additional 36 hours.

| siRNA transfection
HepG2 cells were plated in 12-well plates and cultured at 37°C in 5% CO 2 with Dulbecco's modified Eagle's medium (Gibco) containing 10% heat-inactivated fetal bovine serum (Gibco) (v/v). The cells were then transfected with siRNA targeting human GNPAT/mouse Gnpat gene or control non-specific siRNA (30 pmol/well; TransSheep; siRNA sequences were listed in Table S2) at 70% confluence with Lipofectamine 3000 (Invitrogen). Twenty-four hours after transfection, human recombinant BMP6 (R&D systems) was added to a final concentration of 20 ng/mL and cells were incubated for additional 12 hours. The cells were then collected for RNA extraction and quantitative real-time PCR detection. Isolation of mouse primary hepatocytes was performed as previously described. 39 The isolated primary hepatocytes from either wild-type or Hfe −/− mice were cultured in Dulbecco's modified Eagle's F I G U R E 2 Loss of Gnpat expression does not affect iron metabolism under high dietary iron condition. A-F, At 6 weeks of age, the indicated groups of mice were fed a high-iron diet for 10 d, and hepatic non-heme iron (A), splenic non-heme iron (B), serum iron concentration (C), transferrin saturation (D), hepatic Hamp1 mRNA (E) and hepatic Bmp6 mRNA (F) were measured (n = 5-7 female mice per group) medium containing 10% fetal bovine serum. The siRNA transfection and BMP6 treatment steps are comparable with HepG2 cells.

| RNA extraction and quantitative real-time PCR
RNA was extracted from cells and liver tissues and analysed using quantitative real-time PCR analysis as previously described. 40

| Western blot analysis
Hepatic tissue was lysed using RIPA lysis buffer, and total protein (40 µg/sample) was loaded on a 10% sodium dodecyl sulphate polyacrylamide gel and separated by electrophoresis. After transferring the proteins to a membrane, the following primary antibodies were used for Western blot analysis: rabbit anti-phospho-Smad1/5/8 (Cell Signaling Technology), rabbit anti-Smad1 (Cell Signaling Technology) and mouse anti-β-actin (Sigma-Aldrich).

| Statistical analysis
All summary data are presented as the mean ± SD. Where indicated, groups were compared using the analysis of variance (ANOVA) and Tukey's test for multiple comparisons. All statistical analyses were performed using GraphPad Prism 8, and differences with a Pvalue < .05 were considered significant.

| Association between GNPAT p.D519G and iron-overload phenotype in homozygous carriers of the HFE p.C282Y variant in human populations
We summarized published population studies on GNPAT p.D519G variant in Table S1. Following the first report of the increased prevalence of the GNPAT p.D519G variant among homozygous carriers of
We then investigated the effects of deleting Gnpat in Hfeknockout (Hfe −/− ) mice, which develop a phenotype similar to HFE-linked haemochromatosis in human. We therefore generated Gnpat/Hfe double-knockout mice (referred to hereafter as DKO mice) by crossing our Gnpat −/− mice with Hfe −/− mice. To our surprise, we found no significant difference between Gnpat −/− mice and Gnpat +/+ mice or between Hfe −/− mice and DKO mice with respect to hepatic non-heme iron ( Figure 1E), splenic non-heme iron ( Figure 1F), serum iron concentration ( Figure 1G) or transferrin saturation ( Figure 1H).
In addition, we found no significant difference in hepatic Hamp1 mRNA ( Figure 1I), hepatic Bmp6 mRNA ( Figure 1J), the Bmp6/liver non-heme iron ratio ( Figure 1K) or hepatic phospho-Smad1/5/8 levels ( Figure 1L and M). Finally, deleting Gnpat had no effect on iron distribution in the liver ( Figure 1N) or ferroportin protein in the small intestine ( Figure 1O, the ferroportin expression was quantified in Figure S1). Taken together, these results indicate that deleting Gnpat does not significantly affect iron metabolism, nor does it affect the iron-overload phenotype in Hfe −/− mice.

| Iron metabolism in both Gnpat −/− and DKO mice is not affected by high dietary iron
Next, to explore whether Gnpat plays a role in regulating either the absorption of dietary iron or the deposition of iron in the liver, we fed 6-week-old Gnpat +/+ , Gnpat −/− , Hfe −/− and DKO mice an iron-rich diet for 10 days. Similar to our results obtained with mice that were fed a standard diet, we found no significant difference between

| Knockdown of human GNPAT in HepG2 cells and mouse Gnpat in primary hepatocytes
We used small interfering RNA (siRNA) targeting human GNPAT in HepG2 cells ( Figure 4A in the presence of 20 ng/mL BMP6 in wild-type or Hfe −/− primary hepatocytes ( Figure 4D).

| Expressing the GNPAT p.D519G variant does not affect hepcidin expression in human cells expressing the HFE p.C282Y variant
Lastly, we examined whether the GNPAT p.D519G variant affects hepcidin expression in two human hepatic cell lines. Both HepG2 cells ( Figure 4E) and Huh-7 cells ( Figure 4F) were either mock-transfected or transfected with wild-type human GNPAT or the GNPAT p.D519G variant, and hepcidin mRNA was measured. Consistent with our in vivo results, we found that expressing either wild-type GNPAT or the p.D519G variant had no effect on hepcidin expression; as a positive control, overexpressing BMP4 significantly increased hepcidin mRNA levels in both cell lines.  22 The authors also reported that the frequency of the GNPAT p.D519G allele among male patients with a severe iron phenotype (38.6%) was significantly higher than among Americans of European descent (20.6%); in contrast, none of the males who presented with a mild iron phenotype were carriers of the GNPAT p.D519G allele. 22 In subsequent studies, this apparent increased prevalence of the GNPAT p.D519G variant was also confirmed in two additional populations. 25,41 Moreover, the GNPAT p.D519G variant was reported to be associated with increased iron parameters both in homozygous carriers of the HFE p.C282Y variant 41 and in healthy individuals. 23,24 Interestingly, however, several other groups found no increase in F I G U R E 4 GNPAT knockdown in HepG2 cells and primary hepatocytes of wild-type or Hfe −/− mice. HepG2 cells were transfected with siRNA targeting human GNPAT gene (siGNPAT) or control non-specific siRNA (NC). GNPAT (A) and hepcidin (B) expressions were detected under basal condition or in the presence of 20 ng/mL BMP6. Primary hepatocytes from wild-type or Hfe −/− mice were transfected with siRNA targeting mouse Gnpat gene (siGnpat) or control non-specific siRNA (NC). Gnpat (C) and hepcidin (D) expressions were detected under basal condition or in the presence of BMP6. HepG2 (E) and Huh-7 (F) cells were transfected with the indicated constructs, and hepcidin mRNA was measured 36 h after transfection. Groups labelled without a common letter were significantly different (P < .05; analysis of variance) the prevalence of the GNPAT p.D519G variant in large cohorts of homozygous carriers of the HFE p.C282Y variant, 26,[28][29][30] and no significant association was found between the GNPAT p.D519G variant and serum ferritin levels. [25][26][27][28][29][30] McLaren et al also found that knocking down GNPAT expression in HepG2 cells significantly reduced hepcidin expression but had no effect on phosphorylated Smad1/5/8 level, leading to the hypothesis that GNPAT may regulate iron metabolism via hepcidin independent of the Smad1/5/8 pathway. 22 In contrast, we found that global deleting Gnpat or hepatocyte-specific deleting

| D ISCUSS I ON
Gnpat had no significant effect on hepatic hepcidin ( Figure 1I and

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.