The influence of a high fat diet on bone and soft tissue formation in Matrix Gla Protein knockout mice

Studies suggest bone growth and development are influenced by maternal nutrition, during intrauterine and early postnatal life. This study assessed the role of MGP and a maternal high fat diet on vitamin K-dependent proteins’ gene expression and their impact on bone formation. Knockout (KO) offspring were smaller than wild type (WT) littermates, yet possessed the same volume of intrascapular brown adipose tissue. The total proportion of body fat was reduced, but only in animals on a control diet. Lung air volume was observed to be comparable in both KO and WT animals on the same diet. The degree of aortic calcification was reduced in KO animals maintained on a HF diet. KO females on the high fat diet showed reduced cortical bone volume and thickness in the femur and tibia. Gene expression levels of GGCX and VKOR were reduced in control fed KO animals suggesting a potential link between gene expression levels of MGP, GGCX, and VKOR and total volumes of bone, calcified soft tissue, and iBAT; with implications for modulation of body length and mass. Our results confirm the important role for vitamin K in bone and calcified soft tissue, but now extend this role to include iBAT.

MGP heterozygous males and females were mated. At the start of pregnancy dams were fed either a control or a high-fat diet. These diets were continued throughout pregnancy and after birth. Wild type (WT) and MGP knockout (KO) offspring were analysed at 25 days of age. The thickness of calcification of the abdominal aorta wall was determined by CT. The section of the aorta from the diaphragm to the split for the femoral arteries was analysed. Bars show calcified wall thickness for each MGP genotype and diet. Graphs show by sex, n = 4-6 per group. Graphs show mean plus 95% confidence limits.
SCIentIFIC RepORtS | (2018) 8:3635 | DOI: 10.1038/s41598-018-21650-0 volumes of iBAT, regardless of the MGP genotype. The volume of iBAT tended to be higher in the HF-fed group within the same MGP genotype, although these did not reach statistical significance. For iBAT volume as a percentage of the total fat volume, C-fed MGP KO mice of both sexes displayed a significantly higher percentage (over 2 fold) than WT animals on either diet (Fig. 2b). For iBAT volume as a percentage of the total body volume, C-fed MGP KO mice contained a significantly higher percentage (approx. 0.5 fold) than WT mice fed either diet (Fig. 2c).
When iBAT volume was normalised to body size (statistics were based on the residuals following multiple regression against mass and body length) and analysed by genotype, MGP KO animals were found to have a significantly increased iBAT volume (p = 0.036). This suggests MGP KO mice had a higher proportion of iBAT for their size and thus MGP may influence iBAT volume.
Lung structure. For each sex, lung tissue volume was significantly higher (approx. 0.6-2 fold) in WT mice maintained on a high fat diet in comparison to any of the other groups ( Fig. 3a and supplementary material). Figure 3b shows the percentage of the lung tissue which was 200 µm in width or lower upon ultrastructure examination by CT. For both male and female cohorts, the HF-fed KO animals displayed a higher proportion (approx.1.2 fold) of these smaller lung tissue structures than the other groups. These differences corresponded with a significant reduction (approx. 1.6 fold) in the percentage of lung tissue over 400 µm in size in this group, particularly in males (Fig. 3c). These differences indicated a combination of the HF diet and MGP KO genotype altered the size of the tissue composition of the lungs.
In contrast, female mice maintained on a HF diet showed a significantly higher lung air volume (approx. 1 fold) than animals fed the C diet, within the same genotype and across the different genotypes ( Fig. 3d and supplementary material). A similar trend was found in male mice, but was not statistically significant. Whilst HF-fed mice demonstrated the same lung air volume, the thickness of the airways demonstrated a different distribution. Thus, KO females displayed a lower proportion of airways less than 200 µm in thickness (Fig. 3e). Lung air volume was normalised to body size (statistics are based on the residuals following multiple regression against mass and body length). When analysed by diet, no significant differences were found (p = 0.09). However, when assessed by genotype, the KO animals were found to have a larger lung air volume than expected. Hence, MGP may influence lung air volume.
Femur structure. Compared to C-fed WT animals, HF-fed WT animals showed significantly different results in femoral structure for most parameters studied (Table 2), particularly in females. In contrast, HF-fed KO animals showed few significantly different results when compared to C-fed KO mice (Table 2). In addition, C-fed KO mice showed few differences compared to C-fed WT animals (particularly in males), yet most parameters when compared to HF-fed WT mice ( Table 2). The differences in the results observed may have been due to differences in animal mass between the groups (Table 1). Table 2 shows the femoral structural parameters normalised to body size (statistics are based on the residuals following multiple regression against mass and body length). HF-fed KO females showed a lower than expected cortical bone volume, cortical thickness, and midshaft wall thickness (Table 2). Tibia structure. The tibia was assessed by CT; the results are shown in Table 2. The HF-fed WT animals showed significantly different results in tibial structure for bone volume, trabecular spacing (females), cortical bone volume, and cortical wall thickness when compared to C-fed WT animals. In contrast, HF-fed MGP KO animals showed no differences when compared to C-fed KO mice (Table 2). In addition, C-fed KO mice showed significantly different results for bone volume, cortical bone volume, and cortical wall thickness when compared to C-fed and HF-fed WT animals (Table 2). However, the HF-fed KO mice did not differ to C-fed WT mice for any of the parameters assessed, except for cortical bone volume and cortical wall thickness in females (Table 2). To determine if these differences were due to differences in animal mass between the groups (Table 1), the tibial structural parameters were normalised to body size. As seen with the femur, HF-fed MGP KO females showed a lower cortical bone volume and cortical thickness than expected for their size (Table 2). Vertebra structure. Analysis of the 3 rd lumbar vertebrae showed HF-fed WT females had significant differences in length and bone volume when compared to C-fed WT animals (Table 2). HF-fed KO animals showed no differences in comparison to C-fed WT or HF-fed WT animals. C-fed KO female mice only differed in length to the other groups. When the results were normalised to body size (Table 2), C-fed KO females were found to have a higher total bone volume.
Femur qPCR. Femoral expression levels for the genes assessed did not differ between C-and HF-fed WT animals (Table 3). Periostin expression was significantly reduced (0.4-0.7 fold) in KO animals compared to diet controls in both sexes (except HF-fed females). As expected, Mgp expression was found to be 10-25 fold lower in the KO animals compared to WT (Table 3). Although levels were expected to be zero, a valid qPCR CT value of around 35 was detected in these samples (c.f. CT value of 26 for Beta-actin). Both Ggcx and Vkor displayed a lower level of expression in female KO mice; however, this was only statistically significant in the C-fed KO animals compared to HF-fed WT females (0.5 fold for Ggcx) and C-fed WT mice (0.5 fold for Vkor). Expression levels of Bmp-2 were lower (0.5 fold) in HF-fed KO males compared to C-fed KO males. Estradiol Receptor alpha (ERα) levels were reduced 0.5 fold in female C-fed KO mice compared to female C-fed WT animals. Aorta qPCR. Although found at very low levels (valid qPCR CT values of around 36 were detected, c.f. CT value of 22 for Beta-actin), Osteocalcin expression was lower in C-fed KO males compared to C-fed WT males (Table 4). Mgp expression was found to be around 20 fold lower in the KO animals compared to controls (Table 4). Although levels were expected to be zero, a valid qPCR CT value of ~32 was detected in these samples.
Ggcx levels were reduced (0.5 fold) in C-fed KO animals compared to C-fed WT mice. Bmp-4 levels were lower (0.5 fold) in KO animals compared to controls, but this was only statistically significant in males (Table 4). Leptin levels were higher (approx. 2 fold) in KO animals compared to controls, but this was only statistically significant in males (Table 4). Sclerostin levels were reduced 0.6 fold in KO animals, particularly in males. ERα levels were reduced 0.5 fold in C-fed KO animals, particularly in females (Table 4). Angiotensinogen levels were raised 2.5 fold in C-fed KO animals. Vegf 120 levels were increased 2 fold in female C-fed KO females compared to C-fed WT  (Table 4). Glyceraldehyde-3-Phosphate Dehydrogenase (Gapdh) levels were reduced 3-10 fold in KO animals compared to WT mice ( Table 4).
All the genes examined, with the exception of femoral Gas6, showed a significant negative correlation to percentage of iBAT; the highest r 2 values were observed with Ggcx (femoral and aortic), Mgp (femoral and aortic), femoral Periostin, and aortic Sclerostin and Bmp-1 (Table 5).
Aortic expression of Sclersotin, Ggcx, Vkor, and Mgp together with femoral Periostin expression showed highly significant positive correlations with body length and mass, whereas aortic Angiotensinogen expression showed significant negative correlations (Table 5).
There was negligible correlation observed between any of the genes analysed and tissue volume; either total body tissue or lung tissue ( Table 5).
The relevance of the femoral expression of the 5 genes and bone volume was confirmed by multivariant analysis (Table 6). In addition, the relevance of aortic expression of the 7 genes in body length and mass are shown in Table 6. Femoral and aortic expression of the selected genes was relevant in the volume of calcified soft tissue, total fat volume, and iBAT. There was no evidence of a significant correlation between femoral and aortic gene expression with body soft tissue volume or lung tissue volume (Table 6).

Discussion
We have previously shown that MGP expression levels together with a maternal and offspring HF diet altered bone development. This study assessed the effect of MGP knockout offspring, with a maternal HF diet, on bone phenotype and body composition of the offspring, due to the close inter-relationship between bone and fat, and the regulation of vascular calcification by MGP. Interestingly, over-expression of MGP is lethal soon after birth 11 due to a lack of branching of the lung vasculature.
We assessed the gene expression, in bone and aorta, of a number of proteins dependent on vitamin K, bone development, and vascular function. These were the vitamin K-dependent proteins (Osteocalcin, Gas6, Periostin, and Mgp), enzymes involved in the vitamin K cycle (Ggcx and Vkor), proteins involved in bone growth and metabolism (Bmp-1, Bmp-2, Bmp-4, Leptin, Leptin Receptor, Sclerostin, Estradiol, Estradiol Receptor α, and Estradiol Receptor β), proteins involved in the renin-angiotensin system (Angiotensin Converting Enzyme and Angiotensinogen), and proteins involved in angiogenesis (Vegf120, Vegf164, and Vegf188).  It is interesting that one of the commonly used housekeeping genes for qPCR analysis (Gapdh) was reduced in expression in MGP KO animals fed either diet. One possible explanation is reduced GAPDH production from the increased requirements of NADPH 12 due to the increased oxidative stress of the HF diet 13 , and the process of calcification inducing stress 14 and apoptosis in the aorta of the MGP KO animals. Gene expression levels of clotting factors were not assessed as hepatic vitamin K levels were not expected to reduce during the 3 week study period, as both diets contained the same amount of vitamin K.
The role of MGP in lung development has been studied by the Boström 11,15-17 and Rannels 18,19 groups. Their data showed; (i) MGP is expressed in the lung near the epithelial tips of distal buds 19 , (ii) over-expression of Mgp inhibits vascular growth in the lung by limiting BMP-4 activity and Vegf expression 11,15 , (iii) MGP KO mice exhibit excessive vascular branching 15 and finally, (iv) MGP acts as an inhibitory morphogen, acting antagonistically to BMP-4 in an activator/inhibitor reaction-diffusion dynamic system in lung vascular development 11 . While Boström's group studied lung from 4 week old KO animals, they did not discuss airway volume or lung tissue volumes. Interestingly, our diets appeared to alter lung air volume, given KO animals fed a HF diet has significantly higher air volume than WT animals fed the C diet, despite both groups of animals having the same mass, length and body composition. Interestingly lung tissue volume was the same in all groups except the WT HF group which had a larger volume. Hence, lung tissue volume may be proportional to size, or the abnormal lung vasculature reported by Boström 15 in KO animals may prevent further tissue growth expected with the HF diet. An abnormal vasculature and accompanying reduced nutrient supply would explain the increased proportion of small volume tissue and reduced proportion of large volume tissue we found in the KO animals fed a high fat in the current study. However, the HF-fed KO animals displayed a lung air volume the same as HF-fed WT animals with a lung tissue volume half that of the WT animals. This was as a consequence of the KO animals having a higher proportion of wider airway than WT animals. Hence, our studies demonstrated diet and genotype alter lung phenotype.
While the total body fat volume appeared to alter with size across all groups, the volume of iBAT was the same in both WT and KO animals on the same diet, despite these animals being different in size, indicating MGP protein levels are potentially linked to iBAT volume, but not white adipose tissue (WAT) volume. Bone morphogenetic proteins (BMPs) are known to influence adipose tissue; BMP-4 promote the differentiation of WAT to beige adipose 20 , whereas BMP-7 directs iBAT adipogenesis 21 . In addition, MGP is known to bind BMP-2 22 and    24 . If MGP also bound BMP-7, then KO mice would be expected to produce more iBAT for their size. This occurs in C-fed KO animals, but not in HF-fed KO animals. Therefore, the HF-diet itself may be reducing BMP-4 and/or BMP-7 protein levels. However, the HF-diet did not alter gene expression levels of BMP-4. It is possible the increased proportion of iBAT seen in the C-fed KO mice is simply due to their smaller size and therefore an increased need for heat for thermoregulation (from iBAT), which reduces their overall fat volume (energy source). The additional mass produced by the HF-fed KO animals made them indistinguishable from C-fed WT mice in body composition.
The total volume of calcified tissue was comparable in control and high fat fed KO animals. High fat fed KO animals were significantly larger than control fed KO mice. Hence, the HF-fed group showed an 11% reduction in calcified wall thickness compared to C-fed KO animals. Again, this demonstrates how diet and MGP genotype can altered the phenotype of the animals. It is not clear what mechanisms would reduce the level of calcified tissue on a HF diet, especially as this diet can be used to induce diabetes and subsequent vascular calcification. It is known that vascular calcification, in the absence of MGP, is due to uninhibited BMP protein levels 25 . Therefore, to reduce calcification, the diet must reduce BMP protein levels, or suppress other osteogenic pathways implicated in this calcification process 26,27 . It is possible the HF diet is inhibiting osteoblast activity 28 , thereby reducing the level of vascular calcification. This would also account for the reduced body proportion of bone seen in the HF-fed KO group compared to the C-fed KO animals.
Our data suggested KO animals fed a C diet had a higher proportion of bone for body size (particularly in the females), compared to the other groups. There was no increase in bone volume in the HF-fed KO females. The "lack of increase" expected due to the MGP KO genotype appears to be due to a reduction in cortical bone volume in HF-fed KO females, seen when normalised to body size. These results demonstrate the HF diet had altered the effect the MGP KO genotype had on growth and bone structure compared to the C diet. With altered cortical thickness, observed in our study, sclerostin levels from osteocytes might also be expected to be altered.  Expression levels for the Vitamin K-dependent protein (VKDP) genes assessed did not differ between C-and HF-fed WT animals. This is in keeping with our previous data, where we showed no differences at 6 weeks of age, between C-fed WT offspring and WT offspring fed a HF diet for gestation and the first 3 weeks postnatally, before being switched to the C diet 10 . This latter group is more representative of the HF feeding schedule used here.
For C-fed KO animals, Periostin, Vkor, and ER alpha (in females) gene expression levels were reduced in the femur. In contrast, for HF fed KO animals, only Periostin gene expression was reduced in the femur. Thus, the reduction in gene expression of Periostin was independent of diet, whereas the other gene expression levels were diet specific. Similarly for the aorta, C-fed KO animals showed increased Angiotensinogen and Vegf120 gene expression, and reduced gene expression of Ggcx, Bmp-4, and Gapdh. HF-fed KO animals showed reduced gene expression of Bmp-4 and Gapdh. Therefore the reduction in Bmp-4 and Gapdh gene expression was independent of diet, whereas the other gene expression levels were diet specific. Therefore, specific VKDP gene expression was altered at specific sites by both MGP levels and the diet. In a model system of type 1 diabetes in rats, Doyon et al. 33 studied the aorta and found increased calcification corresponded with reduced levels of carboxylated MGP (cMGP) and carboxylase (produced by GGCX), but no alteration in Vkor expression. Kaesler et al. 34 used an adenine model of chronic kidney disease with aortic calcification in rats, and reported reduced GGCX activity and increased uncarboxylated MGP in the aorta. Using a similar model, McCabe et al. 35 found reduced Vkor and Ggcx, but increased Mgp expression in the aorta. These studies are in agreement with our data and would indicate that in the calcifying aorta there are low levels of cMGP and reduced activity of GGCX. Carboxylation of MGP is required for MGP binding activity to BMPs. Clearly, a reduction in cMGP levels would increase aortic calcification. As stated above, the high fat diet may be altering the calcification process and thereby reducing the effects of the low cMGP levels.
Multivariant analysis of the expression levels of 3 vitamin K-relevant genes from the aorta (Ggcx, Vkor, and Mgp) showed significant correlations to body length and mass. Similar analysis for 5 genes from the femur (Ggcx, Vkor, Mgp, Gas6, and Periostin) showed a significant correlation to bone volume as a percentage of total body volume, as well as lung airway volume. Furthermore, multivariant analysis of these 8 genes combined showed significant correlations to the volume of calcified soft tissue, and iBAT.
We believe that the high fat diet caused sequestration of vitamin K within fat depots. The lack of availability of vitamin K to extra-hepatic tissues causes a reduction in the function of MGP, thereby affecting vascular calcification by not inhibiting BMPs, which in turn may result in altered iBAT. In addition, these low levels of vitamin K may alter GGCX and VKOR levels, resulting in altered vitamin K recycling.
Through gene expression analysis and Micro-CT analysis, we have confirmed, and extended, the ways in which MGP can alter lung structure. In addition, we suggest MGP expression and the HF diet alters bone formation in a sex specific manner. We also suggest a potential link between gene expression levels of MGP, GGCX, and VKOR and total volumes of bone, calcified soft tissue, and iBAT; with implications for modulation of body length and mass. Overall, our results confirm the important role for vitamin K in bone and calcified soft tissue, but now extend this role to include iBAT.

Materials and Methods
Experimental Design and Animal Care. All  High Fat Diet. At 13-18 weeks of age, female MGP+/− C57BL/6 strain mice were mated with aged-matched MGP+/− C57BL/6 males. Females were mated to separate males, and after confirmation of mating (presence of a vaginal plug) were individually housed and allocated either a control (C, n = 11 dams) diet of standard chow RM-1 (7.5% kcal fat, 17.5% kcal protein, 75% kcal carbohydrate; Special Diet Services, Witham, Essex, UK) or a high-fat diet (HF, n = 14 dams; 45% kcal fat, 20% kcal protein, 35% kcal carbohydrate; Special Diet Services diet 824053). These C or HF diets were continued throughout pregnancy and lactation until the offspring were harvested at 25 days of age. Animals were killed by cervical dislocation. Where possible for each dam, one wild type and one knockout (for both male and female) were included in the analysis. All dams produced litters containing knockout offspring. MGP Genotyping. Wild type MGP gene was detected using primers MGP-5 + : GCCACAATTTCTGC ATCCT-GC and OMGP-1: CGGGAAAGATGAGGAAGAAGGG, producing a 450 bp PCR product. The MGP null gene mutation was detected using primers MGP-E4: TGCCTGAAGTAGCGGTTGTA and NEO-1: TGAATGAACTGCAGGACGAGG, producing a 1 kb PCR product. PCR conditions were 94 °C for 2 minutes followed by 94 °C for 45 s, 57 °C for 40 s, and 72 °C for 60 s for 35 cycles.
3D Computed Tomography. Whole animals were scanned using a Skyscan 1176 in vivo micro-CT scanner (Bruker microCT, Kontich, Belgium). All scans were taken at 50 kV, 500 µA with 0.5 mm aluminium filter, and 0.5 ° rotation step. Individual 2D cross-sectional images were reconstructed using Bruker NRecon software version 1.6.10.2. Voxel resolution was 18 µm. All reconstructed bones were set to the same orientation with the transverse plane perpendicular to the long axis of the bone using Bruker Dataviewer version 1.5.1.2. Reconstructed images were analysed using Bruker CTAn software version 1.16.4.1. Whole body scans from the C3 vertebra to the second caudal vertebra were used and reconstructed with a dynamic range of 0-0.08. Within CTAn, thresholds to determine SCIentIFIC RepORtS | (2018) 8:3635 | DOI:10.1038/s41598-018-21650-0 body composition of fat, soft tissue, and bone were set at 17-24 (BAT 17-20; only measured within the interscapular region), 25-50, and 51-255 respectively. For the lungs, the chest cavity was reconstructed from a region within the trachea to just below the lower extremity of the lungs with a dynamic range of 0-0.025. Within CTAn, a volume of interest was produced by manually selecting around the area of the lung tissue. Within this new volume, the airways were selected using a threshold of 0-19, and lung tissue selected using a threshold of 20-60. The uppermost limit of the analysed volume was set as the dividing point of the bronchi. For the vertebra, the vertebral body of L3 was analysed. Height and width measurements were taken at the rearward end of the vertebra. For the femur, a section 1.76 mm in height and 0.7 mm behind the growth plate at the distal end was analysed for trabecular and cortical parameters, and the midshaft assessed for cortical parameters. For the tibia, a section 1.76 mm in height and 0.53 mm behind the growth plate at the proximal end was analysed for trabecular and cortical parameters.
The trabecular parameters measured were BvTv (volume of bone within a measured total volume), BsBv (surface to volume ratio of trabecular bone), trabecular thickness, number of trabeculae per mm, trabecular spacing (distance between trabeculae), structural model index (SMI, measure of surface convexity where a lower SMI is indicative of a more connected, plate-like trabeculae), trabecular pattern factor (indicator of connectivity of trabeculae where values closer to zero, both positive or negative, represent a more connected structure), and connectivity density.
Sample Collection. Mice. Following the CT scan (detailed above), the thoracic aorta was removed, washed and flushed of blood in phosphate buffered saline (PBS) then immediately placed in Trizol Reagent (Invitrogen). The left femur was removed, cleaned of soft tissues, and cut in half. The marrow was flushed out of each half using PBS through a fine gauge needle (25 G). The halves were then cut into smaller pieces, approximately 2 mm in length, then immediately placed in Trizol Reagent.

Statistics.
All data for all diet groups was observed to be normally distributed using the Shapiro-Wilks test.
The effect of the gene knockout and sex was determined by two-way ANOVA followed by Bonferroni post hoc test using PASW version 21 (SPSS UK, Woking, Surrey, United Kingdom). Data are presented as mean ± Standard Deviation unless otherwise stated; significance was determined with a P value of 0.05 or lower. For variables correlated with size, we adjusted for differences using multiple regression for the variable of interest versus body mass and length and comparing residuals 37 . Relationships between variables were tested using linear regression analysis (univariant and multivariant) where a P value < 0.05 was considered statistically significant. Data Availability. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.