Effects of different training modalities on phosphate homeostasis and local vitamin D metabolism in rat bone

Introduction

Mechanical forces associated with physical exercise cause deformation of bone tissue. These deformations activate an acute response in osteocytes, in particular production of cytokines and signalling molecules such as nitric oxide (NO), bone morphogenetic proteins (BMPs), Wnts, and prostaglandin E2 (PGE2) (Klein-Nulend, Bacabac & Bakker, 2012). These factors affect osteoblast and osteoclast recruitment, differentiation and activation locally at the sites that are subjected to high strains. Since whole bone response to mechanical loading depends on the strain rate and magnitude (Rubin & Lanyon, 1985; Turner, Owan & Takano, 1995), peak power training with high peak loads with a high frequency is generally considered to result in a higher bone response than endurance training. Moreover, it was reported that a combination of training types even better maintained bone mineral density in postmenopausal women (Martyn-St James & Carroll, 2009).

Recently, it was proposed that mechanical loading affects gene expression and local protein activity of fibroblast growth factor 23 (FGF23) (Sapir-Koren & Livshits, 2014). FGF23 mRNA expression in mouse bone has been shown to increase after 6 days of endurance training (Gardinier et al., 2016). Moreover, it was suggested FGF23 might be a ‘molecular mediator of the whole-body effects of exercise originating from bone’ (Qi, Liu & Lu, 2016). Indeed, serum FGF23 increased after endurance exercise in rat and human (Li et al., 2016; Lombardi et al., 2014), indicating a potential systemic effect of local mechano-response in bone; FGF23 is produced exclusively in bone but acts on the kidneys, where it decreases the reabsorption and increases excretion of phosphate and also suppress 1α-hydroxylase (CYP27b1), reducing its ability to activate vitamin D and subsequently impairing calcium absorption  (Bonewald & Wacker, 2013). To our knowledge, the direct effect of different training types on FGF23 expression within bone has not been reported yet.

A known stimulator of FGF23 production is systemic 1,25-dihydroxyvitamin D3 (1,25(OH)2D3) (Saji et al., 2010) or in vitro locally produced 1,25(OH)2D3 (Tang et al., 2010). Inhibitors of FGF23 production are mineralization-regulating proteins such as phosphate-regulating gene with homologies to endopeptidases on the X chromosome (PHEX) and dental matrix protein 1 (DMP1) (Martin et al., 2011), possibly through alterations in bone mineralization. Mechanical loading could either affect FGF23 gene expression directly, or indirectly by alterations in the expression of FGF23-regulating proteins known to be stimulated by mechanical loading, such as DMP1 (Gluhak-Heinrich et al., 2003; Yang et al., 2005) and matrix extracellular phosphoglycoprotein (MEPE) (Reijnders et al., 2013) as well as presumably locally converted 1,25(OH)2D3.

The active metabolite 1,25(OH)2D3 is hydroxylated from 25-hydroxyvitamin D3 (25(OH)D3) by 1α-hydroxylase (Lips, 2006), which is, among other tissues, expressed by human bone cells (Van Driel et al., 2006; Van der Meijden et al., 2014). The regulation of activity of 1α-hydroxylase within bone is yet to be determined. Van Driel et al. (2006) demonstrated that in cultured bone cells 25(OH)D3, PTH and calcium did not affect CYP27b1 mRNA expression and activity. In contrast, treatment of primary human osteoblasts with high doses of calcium did increase CYP27b1 mRNA expression (Van der Meijden et al., 2016c). Pulsatile fluid flow, an in vitro model for mechanical loading, increased CYP27b1 expression in primary human osteoblasts  (Van der Meijden et al., 2016a), suggesting that 1α-hydroxylase within bone is at least partly regulated by mechanical loading. Among healthy, young women, serum 25(OH)D3 did not differ between different training groups (Lester et al., 2009). However, among rats an increase in serum 1,25(OH)2D3 after endurance training compared to controls was observed (Iwamoto et al., 2004). Whether vitamin D signalling within bone is affected by mechanical loading in vivo remains to be elucidated.

Osteocytes may respond to mechanical loading by altering the expression of FGF23 and CYP27b1. As the response of osteocytes to mechanical loading is determined by the type of loading, FGF23 production and 1α-hydroxylation may be differentially changed according to the type of loading. We hypothesize that mechanical loading induced by physical exercise influences FGF23 production and 1α-hydroxylation in bone, with the greatest response after a combination of both peak power training and endurance training. Therefore, the aim of this study is to investigate how different kinds of training modalities affect the production of phosphate regulating proteins and 1α-hydroxylation by rat bone tissue in vivo.

Materials and Methods

Results

FGF23 protein expression in tibiae

As shown in Fig. 1, immunohistochemical analysis of FGF23 protein expression by cortical osteocytes in the tibiae did not reveal any differences in the number of stained cells between the four groups. FGF23 staining in trabecular bone could unfortunately not be quantified, due to low contrast between positive osteocytes and aspecific staining of the marrow.

mRNA expression in tibiae

Figure 2A shows that FGF23 mRNA expression levels did not differ amongst the groups. mRNA levels of osteocyte maturation markers SOST, DMP1, PHEX and MEPE were significantly lower after PT compared to those after ET and PET, as illustrated by Figs. 2B–2E, respectively. Furthermore, Figs. 2F and 2G show that CYP27b1 and VDR mRNA expression levels were lower after PT compared to those after ET. Expression of CYP24 mRNA was below detection limit.

mRNA expression in kidneys

Assuming that FGF23 production would be altered by physical exercise, we investigated mRNA expression levels of FGF23 responsive genes in the kidney to assess whether this would have systemic influences. Figure 3 reveals that there were no significant differences in CYP27b1, CYP24, VDR, NPT2a and NPT2c mRNA expression levels. Therefore, in kidney tissue physical exercise did not induce any changes in mRNA expression levels of FGF23-responsive genes and VDR.

Serum biochemical analysis

Figure 4 shows there were no significant differences in plasma c-term FGF23 concentration between groups. However, we did see an increase in serum PTH concentration in the PT-group compared to the controls. Serum 25(OH)D3, calcium and phosphate levels were not significantly different between groups.

Discussion

This study aimed to investigate whether different kinds of training modalities affect phosphate homeostasis and 1α-hydroxylation in rat bone in vivo.

In contrast to our hypothesis, we observed no differences between groups in FGF23 expression in both bone and serum. In line with our results, FGF23 serum levels were not affected by either submaximal exercise or high intensity exercise with a bicycle ergometer among young men (Emrich et al., 2018). Yet, Lombardi et al. (2014) described increased FGF23 serum concentrations among participants of the multiple-stage bicycle race Giro d’Italia. However, confounding factors in that study could be the high dietary intake as well as weight loss, suggesting the observed changes might be an effect of altered metabolic state rather than mechanical loading of bone cells. Moreover, these cyclists showed signs of induced bone resorption (Lombardi et al., 2012), possibly as a consequence of the heavy metabolic stress in the absence of mechanical loading, whereas we assume a situation of bone formation after mechanical loading in our model. Our findings also stand in contrast with those of Li et al. (2016), who reported increased serum FGF23 after acute exercise, exhaustive exercise, and chronic exercise among mice. However, these mice trained for only one week, whereas the rats in our model performed exercise for up to 6 weeks. As Li et al. also described that FGF23 promotes exercise endurance, it is possible the rats in our study at first responded to training with FGF23-upregulation to adapt to the exercise, but that after 6 weeks the rats were adapted and thus no longer showed increased FGF23 at the time of measurement.

Quantification of FGF23 substrates, such as genes involved in vitamin D metabolism and sodium-phosphate co-transporters in kidney tissue, ruled out that post-translational modification differed amongst groups. As FGF23 is considered the major phosphate regulating hormone (Bonewald & Wacker, 2013), it is not surprising that the lack of change in FGF23 or FGF23 substrates was paralleled with unchanged phosphate concentrations in serum after 6 weeks of training. Therefore, these results indicate that FGF23 production in bone and systemic phosphate homeostasis were not affected after 6 weeks of treadmill running.

Surprisingly, PT caused a decrease in both CYP27b1 and VDR mRNA levels in tibial bone compared to mRNA levels after ET, suggesting decreased vitamin D signalling in bone in the PT group. We had expected to observe an increase in local vitamin D signalling after PT compared to ET. Possibly, PT might have been too strenuous, causing fatigue damage of the bone leading to apoptosis of osteocytes (Jilka, Noble & Weinstein, 2013). This process could also have accounted for the observed decreased vitamin D signalling after PT. In line with other studies (Lombardi et al., 2014; Lester et al., 2009), no differences between groups were observed for serum 25(OH)D concentrations, suggesting the observed decreased vitamin D signalling is rather a local process within bone than a systemic effect of exercise.

We hypothesized that alterations in FGF23 expression and 1α-hydroxylation in bone tissue in response to mechanical loading are related; both FGF23 expression and 1α-hydroxylation are likely to be altered by mechanical loading and both processes may be involved in regulation of each other. As we did not observe differences in FGF23 production but did observe decreased vitamin D signalling in bone tissue, we cannot confirm this hypothesis. Previous studies show contradicting results regarding this issue. In cultured rat osteoblasts FGF23 mRNA expression levels appeared to be enhanced by “autocrine/paracrine action of osteoblast-derived 1α, 25(OH)2D” (Tang et al., 2010). In contrast, incubating cultured primary human osteoblasts with FGF23 did not cause changes in mRNA expression of CYP27b1 and VDR (Van der Meijden et al., 2016c). Moreover, in human bone samples an association between FGF23 and CYP27b1, CYP24 or VDR mRNA expression could not be shown (Ormsby et al., 2013). Possibly, in our rat model an acute change in FGF23 expression and 1α-hydroxylation in bone tissue may have occurred, but a new bone balance between the two was established at the time of investigation.

Furthermore, PTH levels were higher after PT compared to controls 22 h after the last training. It is likely that the observed PTH concentrations after PT reflect an increased basal level of PTH in this group after six weeks of training. In general, a transient increase of PTH is observed during and directly after exercise, depending on the type of exercise (Maimoun & Sultan, 2009), and most studies report normalization of PTH serum concentrations shortly after exercise (Maimoun & Sultan, 2009; Scott et al., 2011; Gardinier, Mohamed & Kohn, 2015), but some studies observed a prolonged increase in PTH 24 h after a single exercise  (Brahm, Piehl-Aulin & Ljunghall, 1997; Bouassida et al., 2003; Thorsen et al., 1997). Low basal PTH concentrations are associated with higher physical fitness (Brahm, Piehl-Aulin & Ljunghall, 1997).

Another interesting observation is that PT compared to ET and PET showed lower expression of osteocyte maturation markers SOST, DMP1, PHEX and MEPE. Therefore, it seems that PT caused a suppression of mature osteocyte function, which was reversed by additional endurance training. In general, DMP1 and MEPE are known to increase after mechanical loading (Gluhak-Heinrich et al., 2003; Yang et al., 2005; Reijnders et al., 2013), whereas SOST is downregulated by mechanical loading (Robling et al., 2008; Gardinier et al., 2016). DMP1, MEPE and PHEX are not only markers of osteocyte maturation, but are functionally involved in phosphate regulation and bone mineralization (Rowe, 2012). DMP1 and PHEX are known to be involved in inhibition of FGF23 production (Martin et al., 2011). Therefore, it was expected that reduction of these factors would lead to an increase in FGF23 production in the PT group in our rat model. However, after 6 weeks of peak training FGF23 production and function in systemic phosphate homeostasis appeared unchanged, which does not exclude that local changes in phosphate regulation in bone tissue may have occurred after PT. Moreover, it has been suggested that DMP1 and MEPE cause local changes in stiffness and mineralization of canaliculi and lacunae, thereby altering the osteocytic response to mechanical loading  (Harris et al., 2007). Possibly a new balance has been established between DMP1, MEPE and FGF23 after 6 weeks of peak training, in which FGF23 carries out its systemic functions whereas DMP1 and MEPE have a local function in lacunar remodelling.

Local mechano-response in bone has the potential to lead to systemic effects, because bone is an endocrine organ (Han et al., 2018), and changes in bone resorption and formation can lead to alterations in systemic calcium and phosphate. However, in comparison to the many local mechano-responsive changes in tibial bone in this rat model, the only observed systemic effect was an alteration of PTH concentrations after PT. PTH concentrations respond to many other stimuli than just mechanical loading, such as changes in the adrenergic system, pH and or lactic acid (Bouassida et al., 2006). As neither systemic FGF23, 25(OH)D3, calcium and phosphate, nor any of the genes measured in kidney tissue differed between groups, it seems that the effects of mechanical loading in this model did not lead to major systemic differences between groups.

Remarkably, we did not observe any statistically significant effect of PET compared to controls or ET. Therefore, it seems that the observed changes in gene and protein expression after PT have been reversed by additional ET. These results should be interpreted with caution; the rats performing PET had more training than the other groups. By performing both training programs, they underwent a higher amount of strain, which might be a confounding factor.

In this study organs were harvested 22 h after the last training session in order to study contractile muscle force characteristics in situ (Furrer et al., 2013). Therefore, acute effects of mechanical loading on mRNA concentrations might have already disappeared at the time of investigation. The observed differences after PT might indicate a possible new steady state.

Another limiting factor to our study is the lack of accurate FGF23-staining in trabecular bone. Trabecular bone is generally considered to be metabolic active and possible effects of exercise on FGF23 protein expression within bone marrow could unfortunately not be quantified.

Conclusion

In conclusion, six weeks of physical exercise did not cause significant differences in phosphate homeostasis on a systemic level. However, mRNA expression levels of genes involved in local phosphate regulation and vitamin D signalling within bone tissue differed among groups, suggesting a local mechano-response rather than a systemic response in this rat model.

Supplemental Information