Paternal eNOS deficiency in mice affects glucose homeostasis and liver glycogen in male offspring without inheritance of eNOS deficiency itself

Aims/hypothesis It was shown that maternal endothelial nitric oxide synthase (eNOS) deficiency causes fatty liver disease and numerically lower fasting glucose in female wild-type offspring, suggesting that parental genetic variants may influence the offspring’s phenotype via epigenetic modifications in the offspring despite the absence of a primary genetic defect. The aim of the current study was to analyse whether paternal eNOS deficiency may cause the same phenotype as seen with maternal eNOS deficiency. Methods Heterozygous (+/−) male eNOS (Nos3) knockout mice or wild-type male mice were bred with female wild-type mice. The phenotype of wild-type offspring of heterozygous male eNOS knockout mice was compared with offspring from wild-type parents. Results Global sperm DNA methylation decreased and sperm microRNA pattern altered substantially. Fasting glucose and liver glycogen storage were increased when analysing wild-type male and female offspring of +/− eNOS fathers. Wild-type male but not female offspring of +/− eNOS fathers had increased fasting insulin and increased insulin after glucose load. Analysing candidate genes for liver fat and carbohydrate metabolism revealed that the expression of genes encoding glucocorticoid receptor (Gr; also known as Nr3c1) and peroxisome proliferator-activated receptor gamma coactivator 1-alpha (Pgc1a; also known as Ppargc1a) was increased while DNA methylation of Gr exon 1A and Pgc1a promoter was decreased in the liver of male wild-type offspring of +/− eNOS fathers. The endocrine pancreas in wild-type offspring was not affected. Conclusions/interpretation Our study suggests that paternal genetic defects such as eNOS deficiency may alter the epigenome of the sperm without transmission of the paternal genetic defect itself. In later life wild-type male offspring of +/− eNOS fathers developed increased fasting insulin and increased insulin after glucose load. These effects are associated with increased Gr and Pgc1a gene expression due to altered methylation of these genes. Graphical abstract Supplementary Information The online version contains peer-reviewed but unedited supplementary material available at 10.1007/s00125-022-05700-x.

released from cauda epididymis into 5ml phosphate-buffered saline (PBS) maintained at 37℃ for 15 min incubation, then 10 ul semen was taken and placed on a MAKLER sperm counting plate and observed by an optical microscope (Olympus BX 53) at ×400 magnification to assess sperm concentration. After incubation, nylon mesh (pore size: 70 mm) was used to filter the suspension. The sperm were then treated with somatic cell lysis buffer (0.1% SDS, 0.5% Triton X in DEPC H2O) for 40 min on ice to eliminate somatic cell contamination, after which the sperm be pelleted by centrifugation at 600g for 5 minutes. after removal of suspension, the sperm pellet was resuspended and washed twice. The sperm pellet was added with TRIzol reagent, homogenated, followed by RNA extraction. Small RNA libraries were constructed according to Small RNA Sample PreKit (Illumina), the small RNA libraries were prepared followed by library quality validation for sequencing.
Deep sequencing, quality control and Small RNA-seq data analysis For each RNA library, 10 million reads (raw data) were generated by Illumina Hi-Seq. After quality control, small RNA tags were mapped to mouse genome to analyze their distribution and expression on the genome and annotated with miRNA, tRNA, rRNA and other small noncoding RNA from miRBase19, Genbank and Rfam databases using blastn. To analyze differential expression of small RNAs between L-NAME treated and normal mice sperm, miRNA reads were normalized by TPM(transcripts per million reads. Those miRNAs that had P value smaller than 0.05 and had the fold change of at least 2 were considered as significantly changed miRNAs.

testicular morphology
Small tissue samples of testicle were obtained, fixed and processed by routine histological techniques. Tissue sections of 5 μm thickness were stained with hematoxylin-eosin (H&E) and observed under a microscope (Olympus BX53).
Sections were evaluated according to the modified Johnsen scoring system as previously described [10,11]. The sloughing rate of maturing sperm cells is also calculated (at least 50 seminiferous tubules per sample were analyzed) as previously described [12].

Liver morphology
Hematoxylin and Eosin Staining was done after washing the livers in PBS buffer, fixation in 4% (w/v) paraformaldehyde in PBS, embedding in paraffin and cutting into 3 μm slices using a Microm HM230 Microtom. For liver slices, the hepatic venules and their adjacent portal fields were identified by sinusoidal connection [13]. 10 lobules of every liver were thus identified using a Zeiss (Oberkochen, Germany) Axiovert 100 microscope (200x) and photographed with a Leica EC3 digital camera using LAS EZ software (Leica, Wetzlar, Germany). Linear lobular dimensions were measured from the centre of the hepatic vein to the centre of three related portal vein branches using ImageJ (version 1.410, NIH shareware). The mean radius of lobules was calculated for each animal. The extent of lobular inflammation was graded as described previously [14]: score 0, no inflammatory foci; score 1, fewer than two foci per × 200 field; score 2, two to four foci per × 200 field; and score 3, more than four foci per × 200 field.
Red Oil Staining was done as described elsewhere [15]. 30 pictures were taken per organ using an Olympus (Shinjuku, JP) BH-2 microscope (400x) and a digital camera CFW-1310C (Scion Corporation, Frederick, MD). The lipid content and the number and size of lipid droplets were quantified with the ImageJ program.
Immunostaining was followed by hematoxylin for nuclear counterstaining. The number of CD68-positive macrophages in the liver was quantified as described previously [16].

Pancreas morphology
Hematoxylin and Eosin-stained pictures of whole tissue slide and of every islet of Langerhans were taken using Zeiss Axiovert 100 microscope (25x/200x) and Leica EC3 digital camera. The islets were counted, and the islet area was measured using ImageJ software to calculate the islet density and the mean islet area per slide.
Pancreas Immunohistochemistry: Beta cell content of islets of Langerhans was measured using immunohistological staining of insulin. We used an antibody against insulin (ab7842, abcam, Cambridge, UK) and a secondary antibody (ab6907, abcam) diluted in antibody diluent (Dako, Glostrup, DK) and for visualisation the ABC staining system (sc2023, Santa Cruz Biotechnology, Santa Cruz, CA) according to the manufacturer's instructions. All islets per slide were photographed using an Olympus BH-2 microscope (200x) and CFW-1310C digital camera. Using ImageJ software, the total islet area and the beta cell area was measured (see also Figure 5). Content of βcells in islets was expressed as the percentage of positively stained area in the total islet area.

Quantitative real time PCR
We analysed a list of candidate genes involved in liver fat and carbohydrate metabolism as described recently [18]. We have choosen this list, because we wanted to investigate whether a heterozygous paternal eNOS knockout that is not transmitted to the next generation has the same effect on the offspring's phenotypea fatty liver phenotypeas we recently described in the offspring of female heterozygous eNOS mice [6]. RNA extraction from liver tissue, reverse transcription PCR and design of specific primer were done like previously described [19]with the exception that primer were obtained from Sigma-Aldrich, Eurofins (Ebersberg, GER) and Biolegio (Nijmegen, NE).
The PCR was performed on a Mx3000P thermal cycler (Stratagene, La Jolla, CA) with Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA), Sensi Mix or SensiFast low ROX kit (Bioline, London, UK) in accordance with instructions for use. All samples were analysed in triplicate. The PCR reaction efficiency has been proofed by linear regression method and the relative quantity of analysed genes was calculated with the ΔΔCt method as described elsewhere [20]. In short, the Ct values of gene of interest were normalized to the geometric mean of the reference genes HPRT, β-Actin and 18S. These values were normalized against the mean value of the reference group. Sequences of used primers are listed in ESM Table 1.

Quantification of gene specific DNA methylation
Quantification of gene specific DNA methylation was achieved with immunoprecipitation of methylated genomic DNA (MeDIP), with minor modifications as described by Weber et al. [21]. Briefly, genomic DNA was extracted from liver tissue by proteinase K treatment, RNAse digestion, phenol-chloroform extraction and precipitation with isopropanol. The DNA was sonicated to obtain random fragments between 300 and 1000 bp. 2 µg of the fragmented DNA was denatured for 10 min at 95°C and precipitated over night at 4°C with 10 µg of monoclonal antibody against 5methylcytosine (Zymo Research) in IP buffer (10 mM sodium phosphate (pH 7.0), 140 mM NaCl, 0.05% Triton X-100). 20 µl of MagnaChip protein G magnetic beads (Millipore) were added and incubated for 2 h at 4°C to capture the antibodies. Magnetic beads were washed two times with IP buffer and treated with proteinase K for 3 h at 50°C. The methylated DNA was recovered by phenol-chloroform extraction and ethanol precipitation.
For the analysis of GR exon 1A and PGC1a promoter, specific primers were created (ESM Table 1). To calculate the proportion of methylated DNA in specific target sequences the content in enriched methylated DNA and input DNA were compared and normalized against the mean value of the reference group.

Statistics
A formal power calculation to plan group size was not possible due to two reasons first the litter size of wildtype offspring cannot be exactly planned a priori. Second, we did not know the resulting phenotype of wildtype offspring from heterozygous eNOS knockout fathers. Thus, there was no clear assumptions about the effect size and variation that we might expect. Given that, we decided that we needed to have a least 10 animals per group and planed our experiments based on this assumption. This approach to plan the study was approved by the responsible animal welfare committee.
Statistical analysis was done using GraphPad Prism 6 software (GraphPad, La Jolla, CA). All values are presented as mean ± SEM. For the statistical analysis of IPGTT glucose and insulin, two-way analysis of variance (2-way ANOVA) test followed by Bonferroni post-hoc test was conducted. The unpaired Student's t-test and Pearson correlation analysis were applied for normally distributed data, while the Mann-Whitney U test and Spearman correlation analysis were used for non-normally distributed data. The data were non-normally distributed; thus the Mann-Whitney-U test and Spearman's rank correlation were applied. To correct for multiple testing in the gene expression analysis, a false discovery rate (FDR) cut off was set at 0.05 (observation of no more than 5% false positives) [22,23]. Statistically significant differences were considered as p ≤ 0.05. microRNA mmumiR-615-3p is reduced under conditions of endoplasmic reticulum (ER) stress, wherein it regulates the expression of C/EBP homologous protein (CHOP) and determines cellular sensitivity to cell death. [24] CircZNF609 is involved in the pathogenesis of focal segmental glomerulosclerosis by sponging miR-615-5p [25] mmu-miR-193a-5p miR-193a alleviates diabetic neuropathic pain in a mouse model through the inhibition of HMGB1 expression [26] miR-193a/b-3p overexpression attenuates liver fibrosis through suppressing the proliferation and activation of HSCs. [27] mmu-miR-193 influenced embryo implantation by regulating growth factor receptor-bound protein 7 expression. [28] mmu-miR-199b-5p miR-199b-5p is an important regulator in medullary TEC proliferation through targeting Fzd6 to activate Wnt signaling and cell cycle signaling. [29] miR-199b as a regulator of the phenotypic switch during vascular cell differentiation derived from iPS cells by regulating critical signaling angiogenic responses [30] miR-199b is a direct calcineurin/NFAT target gene that increases in expression in mouse and human heart failure [31] mmu-miR-144-3p miR-144 is involved in extracellular matrix remodeling post MI and its loss leads to increased myocardial fibrosis and impaired functional recovery.

ESM
[32] Downregulation of microRNA-144 inhibits proliferation and promotes the apoptosis of myelodysplastic syndrome cells through the activation of the AKAP12-dependent ERK1/2 signaling pathway [33] miR-144 maybe a potential regulator of the development of atherosclerosis via changes in vimentin signaling. [34] Circulating exosomal miR-144-3p inhibits the mobilization of endothelial progenitor cells post myocardial infarction via regulating the MMP9 pathway. [35] mmu-miR-132-3p miR-132-3p priming enhances the effects of mesenchymal stromal cell-derived exosomes on ameliorating brain ischemic injury [36] Targeted silencing of miRNA-132-3p expression rescues disuse osteopenia by promoting mesenchymal stem cell osteogenic differentiation and osteogenesis in mice [37] Brown adipocyte-derived exosomal miR-132-3p suppress hepatic Srebf1 expression and thereby attenuate expression of lipogenic genes [38] MiR-132 controls pancreatic beta cell proliferation and survival through Pten/Akt/Foxo3 signaling [39] mmu-miR-8114 No functional related literature found ESM  eNOS knockout mice of the C57BL/6J strain and their wild-type (wt) littermate were used. Female wt mice were cross-bred with homozygous male eNOS knockout mice. The resulting male heterozygous eNOS knockout (eNOS +/-) mice were then again crossed with female wt mice. Only wt offspring of this breeding procedure (F2 generation) entered the study. These mice were compared to wt mice resulting from crossing male wt and female wt mice. Heterozygous animals used for breeding of the F2 generation were all derived from different dams i.e siblings were not used.