Introduction
Roughly three in four patients with advanced breast cancer develop incurable bone metastases (Siegel, Miller, & Jemal, 2018). Once bone metastasis occurs, the lesions are overwhelmingly osteolytic, putting patients at great risk of suffering skeletal related events (SREs), including severe bone pain and fracture (Coleman, 2005). The mechanical environment in the skeleton is well-known to control bone tissue homeostasis, with increased loading preventing or reversing osteoporotic bone loss and associated fracture risk (von Stengel, Kemmler, Kalender, Engelke, & Lauber, 2007). Similarly, mechanical signals are emerging as a critical factor in bone metastasis and tumor-induced bone disease (TIBD) (Beaton et al., 2009; Sheill, Guinan, Peat, & Hussey, 2018). A few promising reports showed that in patients with existing bone metastases, physical therapy resulted in multiple positive physical outcomes (e.g. increased muscle strength) without SREs (Cormie et al., 2013; Galvao et al., 2018), although changes to bone mass and architecture were not explicitly evaluated. In preclinical models, increased loading was protective against TIBD, such as tibial compression (Fan et al., 2020; Lynch et al., 2013), low intensity vibrations (Pagnotti et al., 2016), and ankle loading (Yang et al., 2019). But, the impacts of loading on tumor cell function may be dose-dependent, where high damage-causing forces may reverse the protective effects of loading against TIBD (Fan et al., 2020). Clearly, more work is needed to clarify the impacts of anabolic loading on TIBD before successful translation to the clinic.
When forces are applied to the skeleton during physical activity, the deforming tissue pressurizes the interstitial fluid, resulting in fluid flow from high to low pressure. These interdependent physical signals are then translated into intracellular biochemical signals, thereby stimulating bone formation when increased (Robling & Turner, 2009; Thompson, Rubin, & Rubin, 2012). Thus, elucidating the mechanisms of how each of these mechanical signals is translated into stimuli in cancer cells within a bone microenvironment is needed to understand how anabolic loading is anti-tumorigenic. To this end, tissue engineering-based approaches have been an indispensable tool. Loading-induced interstitial fluid flow is widely recognized as an anabolic signal for stimulating osteogenesis and bone formation (McCoy & O’Brien, 2010), and can be generated in scaffolds via directly-applied perfusion or compression of a hydrated scaffold. Recently, these approaches have been leveraged to help uncover how skeletal mechanical signals impact cancer cells in bone. Dynamic compression applied to breast cancer cells in a mineral-containing polymeric scaffold downregulated expression of osteolytic genes (Lynch et al., 2013). In contrast, dynamic compression applied to Ewing Sarcoma cells in a hydrogel increased their drug resistance (Marturano-Kruik et al., 2018; Santoro, Lamhamedi-Cherradi, Menegaz, Ludwig, & Mikos, 2015), underscoring that cancer type and microenvironment are important factors in bone cancer mechanobiology. Mechanical signals also regulate tumor cell interactions with resident bone cells. For example, mesenchymal stem cells (MSCs) increased their osteopontin production upon exposure to breast cancer-derived soluble factors during compression-induced osteogenic differentiation (Lynch et al., 2016), indicating that tumor cells may stimulate bone cells to secrete proteins that promote tumor cell adhesion. Further, mechanically loaded, paclitaxel-releasing MSCs inhibited the growth of multiple myeloma cells in the Rotary Cell Culture System (Bonomi et al., 2017; Ferrarini et al., 2013).
While fluid flow is clearly important for bone homeostasis, the impact of matrix deformations is much less understood. Matrix deformations and fluid flow occur together (Robling & Turner, 2009), thus delineating the individual roles of each signal in cancer cell mechanobiology studies is challenging. One approach is to apply perfusion and compression in combination and isolation, and several recent studies using this approach report that fluid flow- and compression-induced signals together enhance bone anabolism (Ramani-Mohan et al., 2018; Zhao, Mc Garrigle, Vaughan, & McNamara, 2018; Zhao, Vaughan, & McNamara, 2015). For example, multiscale computational modeling of a hydrogel scaffold undergoing perfusion, compression, or both predicted that the combination of low magnitude (0.5% peak strain) compression and pore pressure (10 kPa) would induce more osteogenic differentiation and bone mass (Zhao et al., 2018), perhaps as a result of greater cell deformation (Zhao et al., 2015). Further, when computational approaches were combined with combinatorial experiments involving MSCs with an AP-1 (an intracellular strain sensor) luciferase reporter, applied compression resulted in the greatest cellular deformation and osteogenesis, suggesting physical strain is the main driver of bone anabolism rather than fluid flow alone (Ramani-Mohan et al., 2018). Thus, both mechanical signals should be considered when investigating the role of anabolic loading on tumor cells in the skeleton. Further, computational simulations may help shed light on their respective roles.
Here, we report on multiphysics computational simulations of multi-modal loading bioreactor experiments in advance of the metastatic breast cancer cellular experiments to ensure a mechanical environment in the anabolic range. Our bioreactor delivers compression and perfusion individually or in combination to multiple bone-mimetic 3D constructs (Fig. 1A). We have previously determined the necessary inlet flow velocity to engender physiological and anabolic wall shear stresses (WSSs) in our bone-mimetic scaffold (Liu, Han, Hedrick, Modarres-Sadeghi, & Lynch, 2018), and we also determined an osteogenic level of dynamic compression that also impacted bone cell-tumor cell interactions (Lynch et al., 2016). Here, we extend our previous work to include simulations of multiple magnitudes as well as dynamic compression, and we anticipate that the combination of applied compression and perfusion will synergize to produce the greatest mechanical signals (Zhao et al., 2018).
Materials and Methods
Overview of Experimental Setup used for Simulations
Our experimental setup includes a multi-modal loading bioreactor and bone-mimetic scaffold. With our bioreactor (Bangalore integrated System Solutions), we can apply compression and perfusion, which are independently controlled, either in isolation or in combination. Mechanical stimuli is applied to breast cancer cells seeded in highly porous scaffolds fabricated from poly(lactide-co-glycolide) (PLGA) microspheres and hydroxyapatite (HA) scaffolds (Lynch et al., 2013; Pathi, Kowalczewski, Tadipatri, & Fischbach, 2010). We simulated eight experiments with dynamic compression and steady perfusion in various combinations with a static, nonloaded control (Fig. 1B). The compression conditions included no compression (C-), low compression (5% peak bulk strain, C+), and high compression (10% peak bulk strain, C++) applied at 1 Hz. The high compression value (10% peak strain) previously inhibited expression of resorption genes in breast cancer cells within the same PLGA-HA scaffold modeled here (Lynch et al., 2013), and modulated their interactions with bone marrow mesenchymal stems during osteogenic differentiation (Lynch et al., 2016). The low compression value (5% peak strain) serves as a lower stimulus, but is within the range of commonly used strains in experiments to stimulate osteogenic responses (Bhatt et al., 2007).
The inlet perfusion velocity conditions included no perfusion (P-), steady low perfusion (0.3 mL/min, P+), and steady high perfusion (0.6 mL/min, P++). Our rationale for applying steady, as opposed to dynamic, perfusion is that no clear consensus currently exists as to which profile is better for osteoinductive responses in bone cells in tissue engineered constructs (McCoy & O’Brien, 2010). The low perfusion rate (0.3 mL/min) was previously simulated via CFD, and resulted in internal shear stresses that are in the osteogenic range and flow velocities similar to intracanalicular flow velocities during in vivo tibial loading (Liu et al., 2018).