Nowadays, as a result of the vast development of therapeutic intervention in lung cancer, including targeted therapy, thousands of patients have a longer overall survival and an improved quality of life. Although lung cancer remains the leading cause of cancer-associated mortality worldwide, the mortality of the malignant tumor has reduced slightly (1,2). Notably, metastasis is responsible for as much as 90% of cancer-associated mortality. Lung cancer exhibits a strong predilection to develop to bone metastasis, which would bring not only physical torment, including pain, broken bones and spinal compression, but also the mental affliction to the patients, which greatly reduce the quality of life and overall survival of the patients (3,4). Despite the fact that the medical management of bone metastases in lung cancer has made great progress, the effective curative clinical therapy is limited (5). Additionally, the mechanism regarding bone metastasis remains poorly understood in the pathogenesis of lung cancer. In order to have a general understanding of the mechanisms of bone metastasis in lung cancer, the present review outlined the relative previous studies and suggested the possible mechanisms involved.

According to accumulating evidence, the bone marrow is one of the most common places for the lung cancer cells to metastasize (6). With the bone organophilic phenomenon, Paget's ‘seed and soil’ hypothesis may provide certain reasonable explanation (7). However, contrary to the passive course of the seed planted into the soil, it is an active process for the lung cancer cells colonize (‘seed’) in the bone marrow (‘soil’), under the driving effect of multiple molecules, signaling pathways and cells. Bone metastasis in lung cancer is the completion of a complex succession of cell-biological events, in which there are multiple interactions of cancer cells with host cellular and extracellular microenvironments being involved, as well as different types of molecules, including cytokines (8), adhesion molecules, hormones (9) and chemokines (10,11). It was demonstrated that three types of lesions exist in bone metastases, osteoblast-mediated bone formation, osteoclast-mediated bone resorption and a mixture of each (12). The osteolytic lesion is predominant in the bone metastasis of lung cancer. The present review aimed to introduce the functions of various molecules involved in different processes of bone metastasis in lung cancer. Considering of the ‘seed and soil’ hypothesis, the present review will begin with the ‘seed’, the lung cancer cell, and discuss its metastatic potential.

A large quantity of experimental evidence supports there being a group of cells in the tumor exhibiting the ability of self-renewal, multilineage differentiation and superior levels of malignancy. These cells may be termed cancer stem cells (CSCs) (13,14). It has been experimentally demonstrated that the CSCs exist in an unperturbed solid tumor (15). CSCs have been identified in breast (16), colon (17,18), pancreatic (19), prostate (20–22) and brain cancer (23). Additionally, CSCs have been identified and isolated from lung cancer, and it was demonstrated that the lung CSCs express tissue-specific cell surface markers, including cluster of differentiation (CD)133+ (24). In addition, these lung CSCs are thought to have high drug resistance and tumorigenicity, owing to tumor regeneration following chemotherapy (25,26). For this reason, certain scholars have selected the lung CSCs which expressed tissue-specific cell surface markers, including CD133, CD117 and nuclear β-catenin, and do not express differentiation markers, including cytokeratin (CK)8/18. It was revealed that these CSCs, through an efficient cytokine network production, have high tumorigenic and metastatic potentials (27).

However, with the exception of intrinsic CSCs, the neighboring cancer cells may acquire CSC-like characteristics under the interaction with the active stroma (28). The epithelial-to-mesenchymal transition (EMT) is a cell-biological program, in which the epithelial cells change to exhibit the traits of cells disseminating from primary tumors and seeding metastases (29–31). Experimental data has revealed that the non-CSCs can be induced to enter into a CSC-like state via the EMT (32,33). However, the induced CSCs may well lapse back to a fully epithelial state without CSC function (34). To a certain extent, the transformation of the tumor cells may not completely allow the cells to obtain the function of CSCs, therefore, they may be in the stage between the epithelial cell and induced CSCs (35). In order to study the biology of migrating CSCs, the epithelial-specific and cell-surface markers involved in the EMT are used to detect the circulation tumor cells (CTCs) (36–38). Using a blood filtration approach in patients with lung cancer, Hou et al (39) observed the CTCs lost the expression of E-cadherin, while obtained the expressing of the vimentin, indicating that the CTCs performed a feature of EMT, when the CSCs or induced CSCs enter into the circulation (39).

Despite this, no definite evidence exists to affirm that it is the CSCs that launch the distant metastasis. Considering the ability of self-renewal, multilineage differentiation and superior levels of malignancy, it is generally thought the CSCs are the ‘seed’ to plant into the distant ‘soil’. Whether certain lung CSCs perform the bone organophilic property requires further investigation.

According to the anatomy and pathology, the tumor cells would be arrested in the bone marrow with the bloodstream in the bone metastasis, then they have to egress from the central sinus of the bone marrow. Following the attachment to bone surfaces, the tumor cells induce osteolytic bone destruction in conjunction within the bone microenvironment, and colonize in the bone (84). The bone metastasis of lung cancer predominantly occurs at the spinal bone, ribs, sternum, and the tips of the long bone where there are special micro-sinuses. These sinuses provide good access for the tumor cells to exude from the vessels. The wide-diameter and slow bloodstream of the sinus are better for the tumor cells to travel though and adhere to the endothelial cells. The sinus has a special structure, which is composed of endothelial cells with broad intercellular space, and the tumor cells can easily egress from the sinus and then attach to the bone.

The tumor cell attachment and lodgment in the bone are complex processes, which are actively driven by specific interactions between tumor and normal cells, and with the ECM components of the osseous milieu. Prior to the tumor cells to joining in the osseous milieu, the attachment to the perimyelis is the first step. Certain adhesion molecules are involved in this step, including vascular cell adhesion protein-1, α4β1 integrin and cadherin-11 (84).

Notbaly, MGr1-Ag demonstrates high laminin-binding activity and it was observed that compared with the expression in cells without bone-metastatic ability (SBC-3 cell line), MGr1-Ag was highly expressed in bone-metastatic SCLC cells (SBC-5 cell line). In addition, MGr1-Ag promotes SCLC cell invasion and bone metastasis both in vitro and in vivo via the EMT pathway (85). Discoidin domain receptor-1 (DDR1) is a collagen receptor highly expressed in invading tumor cells and mediates tumor cell survival in bone metastasis (86). Karmele et al (87) found that disruption of DDR1 hampers lung cancer cell survival, leading to impaired early tumor-bone engagement during skeletal homing, and crucially altering bone colonization. PDGF is secreted from the bone stroma. Previous studies have found that the disruption of the PDGF receptor in the bone marrow stroma prevents efficient engagement required for bone homing and osseous colonization, by altering heterotypic tumor-stromal and tumor-matrix interactions (88).

Once the tumor cells land on the bone successfully, the tumor cells join with the bone cells and began to interact with the local bone microenvironment with numerous different types of cells, including osteoclasts, osteoblasts, osteocytes, macrophages and adipocytes. Under the interaction with several molecules and signaling pathways, the tumor cells can construct a hospitable environment for survival and proliferation.

In the pathophysiology of bone metastasis, multiple evidence has demonstrated that it is the osteoclast that destroys the bone, rather than the tumor cells (89,90). As with the bone metastasis of lung cancer, the destruction of the bone is primarily the soluble osseous, in which the osteoclast is the arch-criminal. In this osteolytic lesion, the tumor cells promote osteoclast formation by continuingly secreting pro-osteoclastogenic factors, including parathyroid hormone-related protein (PTHrP), receptor activator of nuclear factor-κB (RANK)L and macrophage colony-stimulating factor. Simultaneously, the excessive generation of abnormally activated osteoclasts elevate the bone resorption, in which the growth factors, including TGF-β, insulin-like growth factors (IGFs), PDGFs and bone morphogenetic proteins, are released from the bone matrix. As a result of that, the abundant growth factors provide a rich ‘soil’ for the tumor cells to proliferate. Additionally, the destruction of the bone matrix supply provides more room for the tumor cells to expand. Therefore, this tumor-osteoclast cooperation forms a vicious cycle in the bone microenvironment, accelerating the bone injury, as well as the pain (91). Therefore, the metabolism of the osteoclast is becoming the critical point to research the development of the osteolytic lesion. Notably, multiple molecules and signaling pathways are involved in this process in conjunction with tumor cells, osteoblasts and other cell types in the bone microenvironment, while the key signaling pathway is the osteoprotegerin (OPG)/RANK/RANKL axis.

A balance exists between the osteoclastic bone resorption regulated by the osteoclast, and the osteoblastic bone formation dominated by the osteoblast in the bone metabolism. Additionally, it is widely hypothesized that the OPG/RANK/RANKL axis is the primary factor to regulate this balance. The osteolytic metastatic lesion is the result of the fact that the osteoclastic bone resorption supersedes the osteoblastic bone formation, indicating that the axis is out of balance. It has been previously reported that the soluble RANKL and OPG in the serum are elevated in lung cancer patients with bone metastases (92), indicating that the axis serves a role in the bone metastasis of lung cancer. In samples from patients with bone metastasis of NSCLC, it was demonstrated that the upregulation of OPG, RANK and RANKL, and the increase in the ratio of RANKL/OPG occurred. In vitro, the NSCLC cells with transfection of RANKL cDNA exhibited higher motility and invasion, while the ability reduced following the addition of OPG (93). When inhibiting the RANK-RANKL interaction, the osteolytic lesion induced by the NSCLC cell line A549 in SCID mice was limited (94).

Notably, an abundance of molecules have been revealed to take part in the regulation of the axis, including prostaglandin E, parathyroid hormone, TGF-β, interleukin-1, tumor necrosis factor-α and PTHrP. Of these, PTHrP has been identified to serve a vital role in the bone metastasis of breast and prostate cancer (95,96).

In addition, there is popular belief that PTHrP may be one of the unique regulatory factors involved in the bone metastasis of lung cancer (97). Positive expression of PTHrP in lung cancer indicated a higher chance for the development of the bone metastasis, and researchers also revealed that the PTHrP produced by the lung cancer cells may induce the osteoblasts to express the RANKL, as well as reduce the generation of OPG, leading to osteoclast maturation and activation (98). Miki et al (9) repeatedly injected the neutralizing antibody of PTHrP into the SCLC model mice and found that the bone metastasis was markedly suppressed (9). It was revealed that a highly bone metastatic lung squamous cell carcinoma cell line (HARA) overexpressed PTHrP, and that the treatment of nude mice with anti-PTHrP antibody inhibited the formation of bone metastasis (99). The TGF-β can also stimulate the lung cancer cells to overexpressing PTHrP and if the TGF-β signaling pathway was suppressed, the PTHrP expression was also reduced (100). Therefore, it is clear that the different cytokines can interact or influence each other, and their complex association constitutes a web to regulate the bone metastasis microenvironment. Deng et al (82) analyzed the gene expression levels in the HARA human lung squamous cancer cell line, which exhibits a high bone metastatic tendency, and demonstrated that besides PTHrP, ezrin was also expressed at a higher rate in the bone metastasis lesion (82).

Apart from the osteoclasts, researchers also pay increasing attention to the macrophages in the tumor microenvironment, as well as in the bone metastasis (101). Macrophages are derived from myeloid progenitors and are an important component of the bone marrow. This cell type can be classified into two main subsets: M1 macrophages, generally promoting inflammation, and M2 macrophages, typically suppressing inflammation and assisting tissue repair (102). Within the tumor microenvironment, M1 macrophages have been proposed to be antitumorigenic and kill cancer cells. By contrast, the M2 macrophages are generally considered to be pro-tumorigenic by releasing a variety of growth factors, including fibroblast growth factors and VEGF, to promote tumor growth and invasion (103). Hiraoka et al (104) broadly assessed the effects of monocytes and macrophages in a metastatic lung cancer cell cardiac injection mouse model. When they depleted monocytes and macrophages with clodronate-packaged liposomes, they found that targeting both macrophages and osteoclasts revealed a more pronounced reduction in the number and size of bone metastatic lesions compared with the sole osteoclast-targeting agent, reveromycin A (104).

In the bone microenvironment, besides the osteoclasts and macrophages, other cells (osteoblasts and adipocytes), are also in conjunction with the tumor cells to develop the osteolytic lesion (105), as well as several different approaches and molecules. Despite of the complex process in the bone metastasis of lung cancer, numerous researchers are investigating a specific critical cell or molecule involved in this. For instance, receptor of activated protein C, histone deacetylase 4, paired-like homeodomain 1, roundabout axon guidance receptor homolog 1 and Wnt/T cell factor signaling perform a negative function in bone metastasis of lung adenocarcinoma (106–109). Additionally, increasing attention is being paid to the miR involved in the bone metastasis of lung cancer. Valencia et al (110) found that miR-326 was a relevant biomarker representative of osteolysis in the model of bone metastasis in lung cancer. It was observed that systemic delivery of miR-192 can decrease osteolytic lesions in a lung cancer mouse model (111). In SBC-5 cells, the loss of miR335 promoted SCLC metastatic bone lesion by reducing the expression levels of RANKL and IGF-IR (109).

Compared with the osteolytic bone metastasis in lung cancer, the osteoblastic bone metastasis is seldom observed clinically, therfore, relative research is limited. Although the osteolytic lesion is most frequently observed clinically, osteoblastic bone metastasis and the mixed bone metastasis occur in patients with lung cancer (112). In addition, the bone lesion in the patients with NSCLC is predominantly osteolytic (105). By contrast, SCLC is mostly osteoblastic. However, the mouse model of bone metastasis in lung cancer was used to assess pathological changes in osteoclstic lesion. Therefore, attention must be paid to select the proper models to analyze the bone metastasis in lung cancer.

The process of the bone metastasis in lung cancer can be divided into three steps: i) Escaping from the primary tumor; ii) moving in the circulation; iii) colonizing in the bone. In each of these mechanisms, this process is continuous and progressive, coordinating and complex. Certain molecules will be involved in the different steps with different effects. For instance, when the lung cancer cells leave the primary lesion, the decrease of the adhesion molecules contribute to its dissociation with the neighbor cells. Additionally, the tumor cells moving in the circulation also demand adhesion and dissociation with the endothelial cells. In addition, when attaching to the bone, the tumor cells require the assistance of the adhesion molecules. Therefore, the adhesion and dissociation regulated by the adhesion molecules perform an essential rold during the entire metastatic process. Notably, with the exception of the adhesion molecules, numerous other molecules are cooperating to accomplish each step. It is similar for the lung cancer cells to metastasize to other target organs in the former two steps. Although there are chemokines leading to the bone metastasis in the early steps, they reveal no specificity of bone metastasis. Therefore, the most specific and unique is the third step in bone metastasis of lung cancer. In this process, under the interaction with tumor cells, the osteoclast regulates the osteolytic lesion. In this stage, numerous signaling pathways and molecules are involved. The most important and critical pathway is the RANK/RANKL axis. This has been studied in depth in the field of bone metastasis and diphosphonate treatment targeting this axis has been clinically applied. However, there is no way to delay or stop the occurrence of the bone metastasis and the side-effect of diphosphonate, mandible necrosis, has no effective treatment. In addition, the exact molecules and mechanism of this axis in bone metastasis of lung cancer requires more investigation.

As shown in Fig. 1, from the primary tumor to the circulation and finally, the settling down in the bone, the tumor cells are living in a microenvironment in each site, therefore the tumor cells can generate, invade, escape from the immune system, and obtain more nutrition and space to expand. Therefore, the smart ‘seed’ applies the collective strength from the neighbor cells to trudge to the rich ‘soil’. This indicates that the enemy is more than the tumor cells in the battle to fight with the bone metastasis. Thus, more attention should be paid to the entire microenvironment. Nowadays, with the improvement of the therapy in lung cancer, patients with lung cancer have a longer survival time. Therefore, how to prevent the bone metastasis prior to the late stage is becoming very important. Further investigations and in depth studies must be performed to determine how to prevent the bone metastasis in the earlier stages of lung cancer.