In this study, we used the genomic-wide based pathway approach to explore the joint effects of different gene variants in common pathways on FN bone geometry variations. Two pathways were identified as the most statistical promising pathways for FN Z after multiple testing adjustments, 1- and 2-Methylnaphthalene degradation pathway and EphrinA-EphR signaling cascade.
As we known, 1- and 2-Methylnaphthalene are components of polycyclic aromatic hydrocarbons (PAHs) present in cigarette smoke. A recent in vivo
study showed PAHs may function as ligand for aryl-hydrocarbon receptor (AhR) a transcription factor that regulates gene expression, to cause the loss of bone mass and strength [23
]. However, there has been little direct research into the effects of methylnaphthalene and its metabolites on bone and no biological correlation has been made thereof. In the present study, validation analysis in the Framingham sample demonstrated a significant but modest effect of 1- and 2-Methylnaphthalene degradation pathway genes (R-square=1.24%).
EphR and their cell-surface-bound ephrin ligands, including A- and B-subclass, constitute the largest subfamily of receptor protein-tyrosine kinases. EphrinA can activate EphA receptors and their downstream molecules (as reviewed in [24
]). EphrinA-EphR pathway is known to function in reorganization of the actin cytoskeleton through Rho family GTPases. Rho family GTPases, including RhoA, Rac1 and Cdc42, have been implicated in the contraction of actin cytoskeletal systems, promoting the formation and elongation of lamellipodia and filopodia, respectively [25
]. Since actin cytoskeleton including lamellipodia and filopodia are involved in the biological processes of osteoblastic adhesion and migration that are closely related to the bone geometry qualities, it is reasonable to hypothesize that EphrinA-EphR pathway may contribute to the genetic architecture underlying the variation of bone geometry.
In addition, the potential role of EphrinA-EphR pathway was revealed in the skeletal patterning (as reviewed in [26
]). Several molecules such as ephrin A5 and EphA2-5 were expressed on the surface of osteoblasts or pre-osteoblasts [27
]. Besides, the functions of EFNB2 in osteoclasts and EphB4 in osteoblasts were found to associate with the switch from bone resorption to bone formation [30
]. In previous studies, crosstalks were detected between EFNB2 with both EphA3 and EphA4 [31
], and between EphB2 and ephrinA5 [34
]. These cross-talks may bridge A- and B-subclass Eph receptors/ephrins signaling, which suggested more complex regulation of EphrinA-EphR pathway in normal and pathological bone remodeling.
In this study, the pathway-based association analysis in U.S. whites demonstrated that 21 genes in this pathway were associated with FN Z at the nominal significance level (p
value < 0.05). These genes included an ephrin (EPNA5) gene and 5 EphR genes (EphA1, EphA3-5 and EphA7), 2 Rho family member genes (Rac1 and cdc42) and their downstream p21-activated kinase genes (PAK1 and PAK7), 4 RhoA downstream genes (ROCK2, LIMK2, CFL1 and ACTG2), 3 adaptor molecular of EphR genes (PIK3CG, FAK1 and PTPN11), and others. Some of the identified pathway genes have been known to be important for bone, especially for osteoblasts. For example, RhoA-ROCK cascade was found to mediate the differentiation of human mesenchymal stem cells (hMSCs) to osteoblast and then promote osteogenesis, specifically via its effects on cytoskeletal tension [35
]. GTPases Cdc42 and Rac were also suggested to communicate with receptor of advanced glycation end products to regulate the osteoblast motility [36
]. PIK3CG, as an important modulator of extracellular signals, was found to be important for osteoblastic differentiation, recruitment, migration and survival [37
]. In addition, activated EphA receptors can transmit signals to focal adhesion kinase (FAK), which is protein tyrosine kinase acting as a regulator of the integrin signaling cascade. This kinase was found to be important for mechanotransduction in osteoblasts [41
]. Subsequent validation study confirmed our findings were by demonstrating the significant genetic effects of the same EphrinA-EphR pathway genes (EphA7, PAK1, PIK3CG and LIMK2) and the whole pathway on FN Z in the Framingham sample ().
Results for the multiple step-wise regression analysis for EphrinA-EphR pathway in the Framingham sample
On the whole, the evidences from multiple aspects (pathway-base GWA analyses, validation analyses and functional relevance to bone) strongly supported the importance of EphrinA-EphR pathway for FN Z. Nevertheless, this pathway was associated with FN Z but not with either the other three FN bone geometry parameters or FN BMD in the total U.S. whites. Besides, in the sex-stratified analysis, this pathway was shown to be associated with FN Z only in the male subgroup at a nominal significance level. These results may reflect site-specific effect due to genetic heterogeneity at different skeletal sites and sex-specific effect only present in male subjects. On the other hand, they might also be treated with caution because of the inflated false-positive and/or false-negative rates caused by increased multiple comparisons and insufficient power in individual subgroups. In this study, efforts like permutation-based adjustments have been taken to alleviate it.
Notably, no one EphrinA-EphR pathway gene attained GWA significance (4.2×10−7
) in our conventional GWA studies on FN bone geometry [42
]. That is, pathway-based GWA approach may incorporate information from markers with moderate significance levels so as to detect novel genetic risk factors, which can serve as candidates for further replication studies. This approach has the merit of high-throughput in comparison with traditional candidate pathway association strategy, which may only figure out a very small fraction of the whole genetic architecture of complex diseases/traits. Besides, this approach provides an opportunity to systematically study a large number of pathways without assumptions about the causal pathways, and thus may have the higher power to identify new candidate genetic factors. However, a potential limitation of this study is that some of pathway genes were not covered by our genotype platform. For example, the genome coverage for EphrinA-EphR pathway was only 87.5%.
In summary, our results suggested that the polymorphisms in EphrinA-EphR pathway genes may affect FN bone geometry variations. Future molecular studies as well as validation studies with customized pathway array are necessary to further investigate this pathway in order to clarify its role in bone biology.