Genetic variants of FGFR family associated with height, hypertension, and osteoporosis

Abstract Background Hypertension and osteoporosis are the most common types of health problems. A recent study suggested that the fibroblast growth factor receptor-like protein 1 (FGFRL1) gene in giraffes is the most promising candidate gene that may have direct effects on both the skeleton and the cardiovascular system. Aim Our study purposed to replicate the finding that the FGFR5 gene is related to giraffe-related characteristics (height, hypertension, and osteoporosis), and to assess the associations between genetic variants of the FGFR family and three phenotypes. Subjects and methods An association study was performed to confirm the connections between hypertension, osteoporosis, and height and the FGFR family proteins (FGFR1 to FGFR5). Results We identified a total of 192 genetic variants in the FGFR family and found six SNVs in the FGFR2, FGFR3, and FGFR4 genes that were associated with two phenotypes simultaneously. Also, the FGFR family was found to be involved in calcium signalling, and three genetic variants of the FGFR3 gene showed significant signals in the pituitary and hypothalamus. Conclusion Taken together, these findings suggest that FGFR genes are associated with hypertension, height, and osteoporosis. In particular, the present study highlights the FGFR3 gene, which influences two fundamental regulators of bone remodelling.


Introduction
Hypertension and osteoporosis are the most common types of endocrine dysfunction, which together affect almost half of all adults (Sözen et al. 2017;Al-Makki et al. 2022).It is now clear that these two major metabolic disorders contribute to each other through diverse pathophysiological mechanisms, including inflammation and immune activation (Do Carmo and Harrison 2020).For instance, the renin-angiotensin system (RAS) is a part of the endocrine system that regulates fluid balance and blood pressure, and dysfunction of RAS leads to hypertension (Papaioannou et al. 2016).In addition, a previous study demonstrated that RAS of the bone tissue is intimately involved with bone metabolism (Rachid et al. 2019).Like RAS, the calcium signalling pathway is also recognised as a representative metabolic control system (Alves-Lopes et al. 2020).
A recent study suggested that the fibroblast growth factor receptor (FGFR)-like protein 1 (FGFRL1) gene in giraffes is the most promising candidate gene that may have direct effects on both the skeleton and the cardiovascular system (Liu et al. 2021).Likewise, in both humans and mice, FGFRL1 has been shown to play a significant role in the development of the cardiovascular and normal skeletal systems (Catela et al. 2009;Rieckmann et al. 2009;Kähkönen et al. 2018;Wang et al. 2022).FGFRL1 is primarily expressed in kidney and musculoskeletal tissues, such as cartilage and muscles (Gerber et al. 2020).In addition, FGFRL1 activation in humans is associated with overgrowth via inhibition of FGF signalling (Matoso et al. 2014).
FGFRL1 was identified as an additional FGFR in the year 2000, making it the most recently discovered member of the FGFR family; it is now called FGFRL1 or FGFR5 (Wiedemann and Trueb 2000;Trueb 2011).The FGFRs are composed of five transmembrane receptors (Ornitz and Itoh 2001), including three extracellular immunoglobulin (Kanehisa et al.) domains, an intracellular tyrosine kinase domain, and a transmembrane domain.They have roles in cellular migration, apoptosis, proliferation, and differentiation.Interactions with FGFRs and FGFs trigger the phosphorylation of specific cytoplasmic residues, followed by activation of various cytoplasmic signal transduction pathways, such as the RAS-MAP kinase pathway (Yang et al. 2016).
Meanwhile, FGFRL1 (FGFR5) has three extracellular Ig domains like the four classical receptors (FGFR1 to FGFR4), but only a single transmembrane domain (Goetz and Mohammadi 2013;Ornitz and Itoh 2015).In contrast to the classical receptors, FGFRL1 does not function to activate cytoplasmic signal transduction pathways.As it binds FGF ligands but lacks the transmembrane domain, FGFRL1 has been considered a decoy receptor with negative effects on cell proliferation (Wiedemann and Trueb 2000;Yang et al. 2016).Given that FGFRL1 is a member of the FGFR family with a proven effect on the skeleton and the cardiovascular system, we hypothesised that other FGFR members (FGFR1, FGFR2, FGFR3, and FGFR4) are also associated with the skeletal and cardiovascular systems.
To the best of our knowledge, this is the first study to demonstrate a genetic association between FGFR family proteins and the cardiovascular and skeletal systems.This study was inspired by previous research that demonstrated the molecular mechanism of FGFR5 (Liu et al. 2021) and the fact that there is a structural difference between FGFR5 and classical FGFRs.Besides, the previous study focussed on giraffes, which have representative biological characteristics of long neck and legs.Thus, we analysed associations of the genetic variants in FGFRs that corresponded to humans.We performed an association study to confirm the interrelation of hypertension, osteoporosis, and height with not only the FGFR5 gene but also the classical FGFRs (FGFR1 to FGFR4).This study is aimed to replicate the previous finding that the FGFR5 gene is related to giraffe-related characteristics (height, hypertension, and osteoporosis), and to analyse the associations between genetic variants of the FGFR family and these three phenotypes.

Study participants
The data used in the present study were collected from the Health Examinee (HEXA) study, a large-scale genomic cohort study that is part of the Korea Genome and Epidemiology Study (KoGES).A total of 173,357 participants who visited health examination centres and training hospitals regularly from 2004 to 2013 and were between the ages of 40 and 79 were recruited from eight regions of Korea for the HEXA cohort study.More information regarding the HEXA study is presented elsewhere (Group 2015).Single nucleotide variation (SNV) information was available for 58,698 participants.A trained nurse measured the height (cm), weight (Kähkönen et al.), systolic blood pressure (SBP), and diastolic blood pressure (DBP) of each participant.Each diastolic and systolic blood pressure was measured three times at intervals of more than 5 min by a mercury sphygmomanometer in a seated position, and the average value was used (Kim et al. 2021).Body mass index (BMI) was calculated in kilograms per square metre (kg/m 2 ) using the measured height and weight values.Characteristics including age, the number of participants, height, weight, BMI, SBP, and DBP are listed in Table 1 according to disease status.
Hypertension was defined as SBP ≥ 140 mmHg or DBP ≥ 90 mmHg, taking anti-hypertensive medication, or having a medical history of hypertension (n = 17,086).Participants with SBP < 120 mmHg and DBP < 80 mmHg, no history of taking anti-hypertensive drugs, and no history of having hypertension, were identified as the control group for hypertension (n = 31,440).Osteoporosis was diagnosed by a medical doctor based on bone indices measured through whole-body dual-energy X-ray absorptiometry (DXA).Participants were diagnosed with osteoporosis with a T score of −2.5 or less according to the World Health Organization's standard criterion.A total of 3,074 participants were identified as the osteoporosis group, and the remaining 55,535 participants were defined as the control group for osteoporosis.

Genotyping
Genotype data were obtained from the Centre for Genome Science, Korea National Institute of Health (KNIH).Peripheral blood was extracted for DNA sampling and genotyped using the Axiom ® 2.0 Reagent Kit (Affymetrix Axiom ® 2.0 Assay, Waltham, MA).SNV genotyping was carried out for the HEXA study using the KoreanChip (KCHIP) designed by the Centre for Genome Science at the KNIH.The location of the identified genes was based on the National Centre for Biotechnology Information (NCBI) Human Genome Build 37 (hg19) assembly, and all of the gene regions were expanded by 5 kb at both ends of the transcript.Samples with a genotyping missing call rate > 0.05, excessive heterozygosity, and gender inconsistencies were removed from the sample pool.Markers with a Hardy-Weinberg equilibrium (HWE) test result of p < 1 × 10 −4 and a minor allele frequency (MAF) < 0.01 were eliminated by the quality control process.Details of quality control for the genotyping data have been described previously (Kong and Cho 2019;Ahn et al. 2020).Abbreviations: M: mean value; SD: standard deviation; Bmi: body mass index; sBP: systolic blood pressure; DBP: diastolic blood pressure.*significant differences in characteristics between the controls and cases were determined by the two-tailed student's t-test.

Statistical analysis
To extend the coverage of genetic variants, SNV imputation was executed using the IMPuTE2 program with 1000 genome phase 3 as the reference panel.The associations of genetic variants with hypertension, hypertension component traits (SBP and DBP), and osteoporosis were analysed with adjustment for age, sex, and BMI.The association of genetic variants with height was analysed with adjustment for only age and sex.using PLINK version 1.90 beta (https://www.cog-genomics.org/plink2(accessed on 27 May 2021)), a logistic regression model (for hypertension and osteoporosis) and a linear regression model (for SBP, DBP, and height) were constructed via an additive model with the necessary adjustments according to phenotype.Values of p < 0.05 were considered to imply statistical significance.Also, there were a total of 192 tag SNVs in the FGFR family, the criterion threshold was set to p = 2.60 × 10 −4 considering the Bonferroni multiple corrections.Linkage disequilibrium (Al-Makki et al.) was defined using the Haploview 4.1 program following the approach presented by Gabriel et al. (Gabriel et al. 2002).
We sorted out tag SNV, the representative SNV in each genome region with high LD, from the haplotype block under the condition that the LD measure r 2 ≥ 0.8.1).There were 15 and 14 significant SNVs (p < 0.05) associated with hypertension and osteoporosis, respectively (Supplementary Table 1).Also, linear regression analysis for height and systolic/diastolic blood pressure revealed 27, 7, and 9 SNVs with p < 0.05, respectively.Regional plots were generated to show the association results in each gene (Figure 1).Among the three phenotypes (SNVs related to systolic/diastolic blood pressure were considered to be associated with hypertension), genes that contained SNVs that were simultaneously associated with two of the three phenotypes were selected (Table 2).

Selection of genes and genetic variants
In the FGFR1 gene, rs10958682 was associated with hypertension and blood pressure, and rs147849069 was connected to height.Additionally, the FGFR5 gene independently showed a correlation between height (rs117864192; p = 4.17 × 10 −3 ) and osteoporosis (rs74921869; p = 4.35 × 10 −3 ) with a high level of significance (Supplementary Table 1).There was no genetic variant in the FGFR1 and FGFR5 genes that was significantly associated with two or more phenotypes.Thus, we focussed on FGFR2, FGFR3, and FGFR4.

Association of the FGFR genes with height, hypertension, and osteoporosis
A total of 6 SNVs showed associations with two or more phenotypes at the same time.These six SNVs were located in three genes (Table 2).Rs199545667 (FGFR2) was associated with hypertension and osteoporosis at the same time.Both rs4752570 (FGFR2) and rs351855 (FGFR4) were associated with both hypertension and height.The three genetic variants belonging to the FGFR3 gene (rs1530587, rs116721553, rs10212860) were all linked to bone-related phenotypes, and were simultaneously associated with osteoporosis and height.Among them, rs116721553 satisfied the significance level of the Bonferroni-corrected test in the analysis of its association with height (Table 2).In addition, rs351855 in the FGFR4 gene was a missense SNV.
Rs199545667 (FGFR2) was found to lower the risk of hypertension and osteoporosis, but we found no correlation with blood pressure, which is hypertension-related.On the other hand, a minor allele of rs4752570 was associated with increased height and reduced SBP and DBP (Table 2).However, it was not directly correlated with hypertension.All three SNVs of the FGFR3 gene presented the same tendency, which was to increase both height and the risk of osteoporosis.Finally, rs351855 (FGFR4) was associated with a decrease in the risk of hypertension and a decrease in DBP (p = 0.023), and showed the same trend concerning blood pressure (Table 2).

Functional analysis
KEGG pathway analysis of the FGFR family identified various biological processes and pathways, including the MAPK, RAS, and calcium signalling pathways (Figure 2) (Kanehisa et al. 2022).Among them, we searched the KEGG database for the calcium signalling pathway, which is closely connected with blood pressure, osteoporosis, and height.The FGFR family was affiliated with receptor tyrosine kinases (RTKs).RTKs are the high-affinity cell surface receptors for many polypeptide growth factors, cytokines, and hormones and consist of the epidermal growth factor (EGF) receptor family, insulin receptor family, and others (Robinson et al. 2000;Zwick et al. 2001).
Among four genetic variants (rs199545667, rs1530587, rs10212860, and rs351855) which were predicted motif changes, rs1530587 and rs10212860 in the FGFR3 gene had a RegulomeDB score of 2b and rs351855 in the FGFR4 gene s.e: standard error; oR: odds ratio; Ci: confidence interval.
15 expression quantitative trait loci (eQTL) hits (Table 3).Expression of FGFR2, FGFR3, and FGFR4, which showed association with two of the three phenotypes in bone-related cells were confirmed and present roles in bone growth (Supplementary Figure 1).Also, the GTEx database analysis produced single-tissue QTLs of genetic variants in the FGFR3 gene (Figure 3).Three genetic signals in FGFR3 (rs1530587, rs116721553, and rs10212860) were identified as significant signals for the FGFR3 gene in the pituitary and hypothalamus, at p < 1 × 10 −12 (Figure 3).

Discussion
Our study revealed the associations and signals of genetic variants in the FGFR family.FGFR5, which was previously suggested to be the most promising candidate for skeletal and cardiovascular system development, was replicated in the human cohort.We identified a total of 192 genetic variants in the FGFR family and found six SNVs in the FGFR2, FGFR3, and FGFR4 genes which were associated with two phenotypes simultaneously.Also, the FGFR family was implicated in calcium signalling and three genetic variants of the FGFR3 gene revealed significant signals in the pituitary and hypothalamus.
Evidence suggesting an inverse association between blood pressure and height has been established in various population studies (Cochran et al. 2021).One study of 33,197 Chinese individuals found that height is related to lower SBP and pulse pressure (PP) (Song et al. 2016).Another study using the 534 Helsinki cohort revealed that height is correlated with blood pressure (Korhonen et al. 2017).Subsequently, a cross-sectional study of Nepalese adults also suggested that hypertension has a negative relationship with both SBP and DBP (Das Gupta et al. 2019).On the other  The lower the Regulome DB score, the greater the effect on snV.
hand, some showed no remarkable association between and blood pressure.For instance, Smith et al. and Smulyan et al. did not find any statistically significant connection between height and blood pressure (Smulyan et al. 1998;Smith et al. 2001).Referring to these studies on the association between height and blood pressure, this study was able to contemplate the effects of SNV on height and blood pressure.In the present study, although rs199545667 was significantly associated only with the nominal variable (hypertension) and rs4752570 showed the opposite result, genetic variants in FGFR2 revealed an inverse relationship between hypertension (or hypertension-related traits) and height, in line with several previous studies.However, rs351855 in the FGFR4 gene, which did not show a negative relationship between the two phenotypes, seemed to influence both the prevalence of hypertension and height independently.
Our findings on the associations of height with blood pressure were somewhat paradoxical depending on the gene, but, as in the various studies mentioned above, the mechanism underlying the association between height and hypertension is still unclear (Song et al. 2016).The conflicting results may be due to inter-individual differences in factors that impact blood pressure or height, such as biological and physiological factors including nutritional status and age (Vokonas et al. 1988;DeBoer et al. 2012).However, these potential shortcomings are also opportunities for future studies.More studies are needed to clarify the association between height and blood pressure.
Meanwhile, three SNVs (rs1530587, rs116721553, and rs10212860) in the FGFR3 gene were found to be related to the skeletal system.FGFR3 is an essential regulatory molecule that is expressed at all stages of chondrogenesis and regulates various downstream signalling pathways (activating p38, STAT1, and ERK1/2, increasing expression of the cell cycle inhibitor, and decreasing AKT phosphorylation) (Ornitz and Marie 2015;Ornitz and Legeai-Mallet 2017); it also directly influences the multiplication and differentiation of immature chondrocytes (Iwata et al. 2000;2001;Havens et al. 2008).After the generation of a growth plate, FGFR3 restrains differentiation into hypertrophic chondrocytes and chondrocyte multiplication and ultimately suppresses chondrogenesis (Colvin et al. 1996;Deng et al. 1996).In summary, although FGFR3 promotes proliferation of immature chondrocytes, it also inhibits chondrocyte proliferation following generation of a growth plate, thus preventing overgrowth.
Indeed, one study demonstrated that growth plate activity is deregulated in mice lacking Fgfr3, leading to overgrowth of the appendicular skeleton (Colvin et al. 1996;Deng et al. 1996;Eswarakumar and Schlessinger 2007).Researchers expected functional mutations might occur in FGFR3 since mice lacking FGFR3 are viable, and this expectation was validated by the discovery of a family with dominantly inherited camptodactyly, tall stature, and hearing loss (CATSHL) syndrome (Makrythanasis et al. 2014;Escobar et al. 2016).CATSHL syndrome results from a missense mutation in the human FGFR3 gene that can cause anomalies by inhibiting negative regulation of bone growth (Escobar et al. 2016).In the present study, we found that the minor allele of the genetic variant in FGFR3 increases both height and the risk of osteoporosis.Although FGFR3 mutations do not cause critical disability, they have an effect on the skeletal phenotype within the normal range.Notably, our finding that taller height is associated with minor alleles in three genetic variants of the FGFR3 gene is supported by the results of these other studies.Together, these findings highlight the potential value of the FGFR3 gene in the skeletal system.
Neurological and skeletal conditions are intimately interrelated, so much so that the new field of neuroendocrinology arose based on the evidence that pituitary hormones directly control metabolism and bone remodelling (Mazziotti et al. 2018).For example, activation of the thyroid is under the control of TSH-releasing hormone (TRH) from the hypothalamus and thyroid-stimulating hormone (TSH) from the pituitary (Liyanarachchi and Debono 2017).Accordingly, the thyroid hormone level is set based on the interactions of different components of the hypothalamus-pituitary thyroid (HPT)-bone axis (Sharan and Yadav 2014).TSH produced by the pituitary gland acts on the TSH receptor (TSHR), and overexpression of TSHR inhibits osteoclastogenesis (Mazziotti et al. 2018).On the contrary, an increase in osteoclast development was observed in tshr-null mice lacking exon 1 of the tshr gene, suggesting that TSH may have osteoprotective effects (Marians et al. 2002).Referring to the GTEx portal, the present study verified statistically significant single-tissue sQTLs of genetic signals of the FGFR3 gene in the pituitary and hypothalamus.Given the overall endocrinological and physiological function of the hypothalamus-pituitary axis, our study supports the potential value of the FGFR3 gene in the bone remodelling process via the hypothalamus-pituitary axis.
Additionally, FGFRs were involved in the calcium signalling pathway, which is essential for endothelial control of cardiovascular and osteoclast activity.Calcium signals at myoendothelial projections are a fundamental factor in endothelial control of vascular tone (Wilson et al. 2019).One study demonstrated disruption in calcium signalling mediated by inositol trisphosphate (IP3), which results in release of calcium and creates microdomains that are essential to the balance of vascular function, in hypertensive rats (Yuan et al. 2016;Lin et al. 2019;Wilson et al. 2019).In addition, from the bone homeostasis point of view, it is well-known that calcium is essential for activating proliferation of osteoclast precursors and suppressing the resorption of mature osteoclasts (Kajiya 2012).Calcium homeostasis in bone is monitored by parathyroid hormone (PTH).The parathyroid gland has calcium-sensing membrane receptors, and a low level of calcium releases PTH, followed by stimulation of osteoclasts (Teitelbaum 2000).
This study replicated that the FGFR5 gene is associated with giraffe-related characteristics (height, hypertension, and osteoporosis) using a human cohort, and verified the associations between genetic variants of the FGFR family and three phenotypes.Six SNVs in FGFR2, FGFR3, and FGFR4 showed association with two of the three phenotypes (height, hypertension, and osteoporosis) simultaneously.Furthermore, conjugating public datasets resources, our study supported that FGFR family members have pleiotropic effects on calcium signalling pathways that are ultimately linked to bone growth and hypertension, and indicated the potential impact of FGFRs on vascular and skeletal systems.

Figure 1 .
Figure 1.Regional plot for all of the genetic variants in FGFR family.Association of hypertension related signals in the FGFR1 (a) and FGFR2 (b), and association of height related signals in the FGFR3 (c) and FGFR4 (d) were plotted as -log 10 P-values.The colour of each snV plot shows its linkage disequilibrium (Al-makki et al.) (using the r 2 values) with the novel snVs (purple dismond) within the association locus and the y-axis on the right shows recombination rate using the Hapmap database.

Figure 2 .
Figure 2. The KEgg signalling pathway for calcium signalling pathway of FGFR family.FGFR family found in receptor tyrosine kinases (RTK) according to KEgg is red.

Figure 3 .
Figure 3. Expression of the three genetic variants in the FGFR3 gene.Expression for each genotype in hypothalamus of brain and pituitary gland, reaching under p < 1 × 10 −12 .

Table 1 .
Characteristics of the HEXA study participants.

Table 2 .
genetic variants in FGFR associated with two of the three phenotypes (height, hypertension, and osteoporosis).Age, sex, and body mass index (Bmi) were included as covariants in all genetic models.findings with p < 0.05, which satisfy the suggested replication level, are indicated in bold.The p-value which is satisfied the Bonferroni-corrected significance level (p < 2.60 × 10 −4 ) is indicated in bold and underlined.Abbreviations: snV: single nucleotide variation; Chr: chromosome; mAf: minor allele frequency; β: regression coefficient;