Abstract

Introduction

   Emerging data suggests liver disease may be initiated during
   development when there is high genome plasticity and the molecular
   pathways supporting liver function are being developed.

Methods

   Here, we leveraged our Collaborative Cross mouse model of developmental
   vitamin D deficiency (DVD) to investigate the role of DVD in
   dysregulating the molecular mechanisms underlying liver disease. We
   defined the effects on the adult liver transcriptome and metabolome and
   examined the role of epigenetic dysregulation. Given that the parental
   origin of the genome (POG) influences response to DVD, we used our
   established POG model [POG1-(CC011xCC001)F1 and POG2-(CC001xCC011)F1]
   to identify interindividual differences.

Results

   We found that DVD altered the adult liver transcriptome, primarily
   downregulating genes controlling liver development, response to
   injury/infection (detoxification & inflammation), cholesterol
   biosynthesis, and energy production. In concordance with these
   transcriptional changes, we found that DVD decreased liver cell
   membrane-associated lipids (including cholesterol) and pentose
   phosphate pathway metabolites. Each POG also exhibited distinct
   responses. POG1 exhibited almost 2X more differentially expressed genes
   (DEGs) with effects indicative of increased energy utilization. This
   included upregulation of lipid and amino acid metabolism genes and
   increased intermediate lipid and amino acid metabolites, increased
   energy cofactors, and decreased energy substrates. POG2 exhibited
   broader downregulation of cholesterol biosynthesis genes with a
   metabolomics profile indicative of decreased energy utilization.
   Although DVD primarily caused loss of liver DNA methylation for both
   POGs, only one epimutation was shared, and POG2 had 6.5X more
   differentially methylated genes. Differential methylation was detected
   at DEGs regulating developmental processes such as amino acid transport
   (POG1) and cell growth & differentiation (e.g., Wnt & cadherin
   signaling, POG2).

Conclusions

   These findings implicate a novel role for maternal vitamin D in
   programming essential offspring liver functions that are dysregulated
   in liver disease. Importantly, impairment of these processes was not
   rescued by vitamin D treatment at weaning, suggesting these effects
   require preventative measures. Substantial differences in POG response
   to DVD demonstrate that the parental genomic context of exposure
   determines offspring susceptibility.

   Keywords: vitamin D, DOHaD, liver disease, parental origin,
   susceptibility, omics

1. Introduction

   Nonalcoholic fatty liver disease (NAFLD) is a metabolic disease
   characterized by excessive liver fat accumulation (fat content exceeds
   5% of liver weight) and inflammation ([47]1–[48]3). NAFLD, associated
   with underlying metabolic syndrome, including type 2 diabetes, high
   blood pressure, elevated triglycerides and cholesterol, was recently
   renamed “Metabolic dysfunction-Associated Fatty Liver Disease” (MAFLD)
   ([49]4). The worldwide prevalence of NAFLD is increasing annually, with
   recent (2020) rates of 25% among individuals aged 2-70+ ([50]5–[51]8).
   NAFLD alone can have severe health consequences, and when left
   untreated, NAFLD can progress into more severe end-stage liver disease
   such as hepatocellular carcinoma (HCC). HCC is one of the leading
   causes of cancer-related death worldwide ([52]6, [53]7). Therefore, it
   is critical to elucidate the early causes and mechanisms underlying
   NAFLD.

   The most marked characteristic of NAFLD is liver fat accumulation
   caused by metabolic dyshomeostasis ([54]8), whereby energy production
   is disrupted in favor of energy storage. While metabolic dyshomeostasis
   can occur at any point in time, in utero development has long been
   recognized as a sensitized window of exposure ([55]9) when offspring
   are particularly vulnerable to nutrient fluctuations ([56]10, [57]11)
   that can alter this balance. For example, depletion of vitamin D levels
   during in utero development in Wistar rat models caused severe liver
   fat accumulation in adulthood ([58]12). This effect was attributed to
   underlying mitochondrial dysfunction and abnormal liver lipid
   metabolism ([59]12). Sprague-Dawley rat models also showed that
   maternal vitamin D deficiency during pregnancy is associated with
   offspring insulin resistance, a key regulator of energy storage in the
   liver ([60]13). In this model, the insulin resistance was attributed to
   deficiency-induced upregulation of inflammatory cytokines in the liver
   and serum ([61]13).

   Chronic inflammation is another key characteristic of NAFLD ([62]3),
   which when untreated left can progress to fibrosis and HCC ([63]14,
   [64]15). In Sprague-Dawley rat models of developmental vitamin D
   deficiency (DVD), DVD induced elevated liver and pancreatic oxidative
   stress levels in adult offspring despite vitamin D repletion ([65]16).
   Following a similar exposure model, Masako et al. found that DVD in
   C57BL/6J mouse models caused permanent changes in the proportions of
   inflammatory cells in the adult liver and expression of genes
   regulating lipid metabolism ([66]17). One hypothesis is that vitamin D
   deficiency during in utero development creates an inflammatory state in
   the fetus. Supporting this hypothesis, vitamin D has been shown to
   regulate placental inflammation during human pregnancy ([67]18) and in
   C57BL/6J mice, developmental vitamin D deficiency promotes infiltration
   and activation of liver macrophages ([68]17). However, the role of DVD
   in NAFLD-related liver inflammation has not been directly investigated.

   Severe maternal vitamin D deficiency in humans during pregnancy impairs
   fetal growth and development, causing low birth weight and small for
   gestational age (SGA) ([69]19, [70]20). Both low birth weight and SGA
   are recognized risk factors for NAFLD, especially under conditions of
   rapid postnatal weight gain ([71]14, [72]15). This is in part due to
   underlying disruptions in metabolic-endocrine maintenance ([73]21,
   [74]22), making the offspring more susceptible to liver disease
   outcomes.

   Previously, we found that DVD exposure in Collaborative Cross (CC) mice
   caused increased adult body weight and fat mass despite full recovery
   of vitamin D sufficiency in adulthood ([75]23). Furthermore, a
   comparison of offspring from reciprocal crosses between strains
   CC011/Unc (CC011) and CC001/Unc (CC001) showed that this effect
   differed based on the parental origin of the genome (POG) ([76]23).
   F[1]males from CC001-dams x CC011-sires were susceptible to DVD-induced
   increased adult adiposity, while F[1] males from CC011-dams x
   CC001-sires were resistant to DVD-induced increased adult adiposity
   ([77]23).

   Here, we leveraged this unique POG model in an ancillary study to
   investigate a novel role for vitamin D in the developmental origins of
   liver metabolic dysfunction. Using liver samples collected from our
   previously published DVD model ([78]23), we defined DVD-induced changes
   in liver transcriptional pathways that regulate energy metabolism,
   cholesterol biosynthesis, inflammation, growth & development, and liver
   detoxification; examined evidence for a role in perturbing liver
   metabolic processes; and investigated the role of epigenetic mechanisms
   in the persistence of these effects. In addition, the use of our
   Collaborative Cross POG model allowed us to define interindividual
   differences in susceptibility to DVD-induced adult liver disease.

2. Materials and methods

2.1. Animal husbandry, dietary treatment, and breeding

   All animals were handled in accordance with the Guide for the Care and
   Use of Laboratory Animals under an approved animal use protocol at the
   University of North Carolina (UNC) at Chapel Hill. As previously
   published ([79]23), Collaborative Cross inbred mouse strains, CC001/Unc
   (CC001) and CC011/Unc (CC011), were purchased from the UNC Systems
   Genetics Core Facility (Chapel Hill, NC) ([80]24). Inbred CC001 and
   CC011 virgin dams, aged 4-6 weeks, were placed on one of two isocaloric
   purified diets for 5 weeks prior to the start of breeding, which is the
   timeline shown to induce deficiency in mice (25): VDS (vitamin D
   sufficient diet, 1000 IU/kg vitamin D3, AIN-93G, 110700, Dyets Inc.,
   PA) or VDD (vitamin D depleted diet, 0 IU/kg vitamin D3, AIN-93G,
   #119266, Dyets Inc., PA). Sires stayed on the VDS diet except when
   mating with VDD dams (14 days). Dams remained on diet during and after
   reciprocal mating to CC001 and CC011 sires to generate male F1
   offspring that were primarily genetically identical other than
   different parental origin of the genome (POG) and different
   mitochondrial and Y chromosomes. (CC001-dam xCC011-sire)F[1] and
   (CC011-dam x CC001-sire)F[1] males were maternally exposed to either
   VDS or VDD diets from conception to weaning. At PND21, mice were weaned
   onto standard rodent chow diet (2400 IU/kg vitamin D3; Teklad diet
   #8604, Harlan Laboratories, Germany) and remained on this vitamin D
   sufficient diet for 5-6 weeks, which is a timeline shown to be
   sufficient for vitamin D repletion in mice ([81]25). At 8-9 weeks of
   age, all F1 adult male offspring were euthanized by CO[2], and livers
   were collected and flash-frozen in liquid nitrogen and stored at −80°C.
   Throughout the study, all mice were housed and maintained at a vivarium
   temperature of 21-23°C with a 12-h light cycle and ad libitum access to
   sterilized water and rodent chow ([82]23).

2.2. Sample selection (total RNA-sequencing, metabolomics,
bisulfite-sequencing)

   For RNA-seq, six adult male F[1] offspring liver samples were selected
   from at least three different litters for each diet and POG group
   (n=6/diet/POG, 24 samples total). A subset of these samples
   (n=3/diet/POG from 3 litters, 12 samples total) were used for the
   metabolomics and bisulfite-seq experiments. Whole liver samples were
   pulverized and mixed while frozen and then aliquoted while frozen for
   each experiment.

2.3. Total RNA-Seq

   Total RNA was isolated from adult male F1 offspring livers using Trizol
   reagent following the manufacturer’s protocol (#155960926, Life
   Technologies, NY). All remaining sample prep and sequencing were
   performed by the UNC High-Throughput Sequencing Facility (HTSF) at UNC
   Chapel Hill. RNA was prepped for RNA-Seq using the RNA Clean &
   concentratorTM-5 kit according to the manufacturer’s protocol (#R1013,
   Zymo Research, CA). All RNA analytes were assayed for RNA integrity,
   concentration, and fragment size. Samples for total RNA-Seq were
   quantified using RNA Qubit (#[83]Q32855, Invitrogen, MA). All samples
   had a RIN>7.0 (average RIN = 8.9). 500 ng of input RNA was used for
   library preparation. Fragmentation was performed for 6 minutes at 85°C
   following the KAPA Stranded RNA-Seq Kit with RiboErase protocol
   (#8098131702, Illumina Inc., CA). Indexed libraries were prepared and
   run on NovaSeq XP (#20043131, Illumina Inc., CA) to generate an average
   of 221 million paired-end reads (100 bp) per sample library. The read
   length was 2 x 100bp.

   Raw sequencing reads were filtered for a quality score > 20 in at least
   90% of bases using fastq_quality_filter (version 0.0.14) ([84]26).
   Sequence adapters were then trimmed using Cutadapt ([85]27) (version
   1.12). We then adopted a 2-pass alignment approach utilizing CC001 and
   CC011 pseudo-references (retrieved from