State-of-the-art techniques for imaging the vascular microenvironment in craniofacial bone tissue engineering applications

Abstract

Vascularization is a crucial step during musculoskeletal tissue regeneration via bioengineered constructs or grafts. Functional vasculature provides oxygen and nutrients to the graft microenvironment, facilitates wound healing, enhances graft integration with host tissue, and ensures the long-term survival of regenerating tissue. Therefore, imaging de novo vascularization (i.e., angiogenesis), changes in microvascular morphology, and the establishment and maintenance of perfusion within the graft site (i.e., vascular microenvironment or VME) can provide essential insights into engraftment, wound healing, as well as inform the design of tissue engineering (TE) constructs. In this review, we focus on state-of-the-art imaging approaches for monitoring the VME in craniofacial TE applications, as well as future advances in this field. We describe how cutting-edge in vivo and ex vivo imaging methods can yield invaluable information regarding VME parameters that can help characterize the effectiveness of different TE constructs and iteratively inform their design for enhanced craniofacial bone regeneration. Finally, we explicate how the integration of novel TE constructs, preclinical model systems, imaging techniques, and systems biology approaches could usher in an era of “image-based tissue engineering.”

INTRODUCTION

Craniofacial disorders, which are malformations in the skull or the face, result in serious clinical problems. It was reported that 75% of birth defects in the United States are cranial defects, which affect around 22,500 children annually (1). According to a recent review (2), various disorders can cause craniofacial defects, including congenital malformations, tumor resection, infection, severe trauma, and accidents. Many patients suffer from critical-sized craniofacial bone defects, i.e., defects that are too large for the bone to heal spontaneously. Treating such defects requires reconstructive surgery that is often followed by revision surgeries because the bone graft used did not fit the defect well, or integrated poorly with the intact bone, resulting in eventual resorption and infection (3). Current materials used in lieu of a bone graft include titanium, methylmethacrylate (MMA), polyetheretherketone (PEEK), and hydroxyapatite (HA) (3). However, these materials pose several challenges including poor biomechanical properties, lack of osteointegration, and high cost (3). In addition, they can also result in complications such as chronic pain, dysesthesia, or infections (4). Recently, the concept of reengineering bone from a combination of scaffolds, cells, and growth factors has emerged as a promising solution for successful craniofacial bone regeneration. However, several challenges remain before this becomes a clinical reality including designing materials with appropriate mechanical and biological properties suitable for the bone defect healing environment and vascularization in large tissue engineering (TE) constructs (5). Therefore, elucidating the biology underlying craniofacial bone healing with imaging approaches could provide invaluable insights for developing novel TE materials and techniques.

The two major physiological processes involved in craniofacial bone healing are osteogenesis and angiogenesis (6). Osteogenesis or bone growth is the primary parameter being evaluated during craniofacial bone healing applications. Analogously, angiogenesis is the process of de novo vessel formation that is essential for the supply of oxygen and nutrients to the healing bone. The interplay between these processes involves complex cellular mechanisms and signaling pathways (7). At the cellular level, osteoblasts that differentiate from mesenchymal stem cells (MSCs) are the primary contributors to osteogenesis (8). The main cell types involved in angiogenesis are endothelial cells (ECs) and perivascular cells including pericytes, smooth muscle cells, and adventitial fibroblasts (9). For the sake of clarity, in Fig. 1, we have illustrated the endothelial cells that sense cytokines in the tissue microenvironment, sprout, proliferate, and anastomose to extend and maintain the blood vessel network during bone healing (7). In addition, multiple signaling pathways are involved in osteogenesis and angiogenesis, including the bone morphogenetic protein (BMP)/transforming growth factor (TGF) β pathway, Ca²⁺ signaling, the Wnt pathway, the fibroblast growth factor (FGF) pathway, insulin-like growth factor (IGF) signaling, the platelet-derived growth factor (PDGF) pathway, vascular endothelial growth factor (VEGF) pathways, and the Notch pathway, to name a few (10, 11). Understanding these processes and their surrounding microenvironment with imaging can help researchers improve outcomes during craniofacial bone healing. Among the challenges of using TE methods to repair injured craniofacial bone tissue, successfully modulating the vascular microenvironment (VME) has posed a major challenge. This VME comprises macro- and microvessels, immune cells, blood cells, osteoblasts, growth factors, and metabolites such as oxygen. For example, the lack of oxygen (i.e., hypoxia) hinders the successful integration of many tissue-engineered constructs (12). Different approaches in TE have been developed to overcome this, including using oxygen-releasing/generating biomaterials (13, 14) and integrating oxygen-releasing micro-tanks into scaffolds (15). In addition, as the diffusion distance of oxygen from a capillary is limited to ∼100–200 μm (16), the establishment of new vascularization (i.e., angiogenesis) is necessary for the sustained health and integration of any tissue-engineered construct. Although the body can spontaneously initiate angiogenesis as a part of the wound healing cascade or in response to hypoxic signals within the graft microenvironment, the de novo growth of new blood vessels first appears in the wound bed 3–5 days after injury (17) and is often insufficient to continuously supply adequate oxygen and nutrients to grafts in craniofacial tissues. The cessation of angiogenesis after graft implantation or wound healing can further limit oxygen availability within the graft microenvironment (18, 19). Inducing angiogenesis and sustaining a patent microcirculation after graft implantation can thus play a crucial role during craniofacial tissue regeneration. However, there is a dearth of in vivo data on vascular formation and remodeling within the graft. This includes information on in vivo changes in vascular morphology and function (e.g., perfusion) that contribute to the overall vascular microenvironment (VME) of the graft site.

Figure 1.

Imaging the vascular microenvironment (VME) during bone healing in preclinical models of craniofacial tissue engineering (TE). A: various parameters within the VME ranging from the macro- and microvasculature to cell types, growth factors (GFs), and oxygen can be imaged and quantified using an array of in vivo and ex vivo imaging modalities (B). C: a schematic illustrating the widely used calvarial defect model and the VME within it (inset), which indicates the factors involved in bone-healing associated angiogenesis and osteogenesis. BLI, bioluminescence imaging; CT, computed tomography; FL, fluorescent microscopy; IOS, intrinsic optical signal imaging; MPLSM, multiphoton laser scanning microscopy; MRI, magnetic resonance imaging; OCT, optical coherence tomography; PAI, photoacoustic imaging; PET, positron emission tomography; 2P-IVM, two-photon intravital microscopy; 2PLM, two-photon phosphorescence lifetime microscopy; SPECT, single-photon emission computerized tomography. Schematics created using BioRender.com.

Currently, in clinical settings, mostly nonoptical imaging methods such as X-ray, computed tomography (CT), and magnetic resonance imaging (MRI) are used to evaluate the integration of TE constructs and the structural changes that occur during craniofacial bone healing. For example, Policicchio et al. (20) performed a clinical study using CT imaging to compare bone healing results with two different titanium cranioplasties using standard precurved mesh versus custom-made prostheses. They observed that the customized cranioplasty gave better cosmetic results, shorter surgical time, and fewer fixations between the two. Due to limitations such as imaging costs and availability of the appropriate imaging modality, clinical studies may not acquire functional data such as angiogenesis, blood flow, cellular distribution, etc., to complement the anatomical data characterizing craniofacial bone remodeling. Consequently, researchers have turned to preclinical models to investigate functional changes in the VME that accompany healing of a craniofacial bone defect. Previously, the evaluation of graft vascularization was limited to bright-field imaging of histological stains [e.g., the Von Willebrand factor (VWF)] or fluorescence (FL) microscopy of CD31-labeled endothelia and the erythrocytes within (21, 22). Such optical microscopy approaches can only provide a static, two-dimensional (2-D) snapshot of vascular structure without any information on vascular function. Recent innovations in imaging have greatly facilitated the in vivo evaluation of the VME in these preclinical models, informed the design of TE constructs, and aided the evaluation of their efficacy in the regeneration of craniofacial tissue (23). These advances in preclinical imaging include the simultaneous characterization of the structure [i.e., 2-D/three-dimensional (3-D) qualitative and quantitative assessments], function (i.e., intravascular oxygenation, perfusion), endothelial cell phenotype (24), and its spatial relationship with other cells within the VME. In addition, enhancements in spatial and temporal resolution, contrast-to-noise ratio (CNR), tissue penetration, and longitudinal imaging capabilities have now made it feasible to monitor the efficacy of TE constructs embedded deep within tissue in vivo (23). Figure 1 illustrates the range of imaging techniques that have been used for interrogating VME parameters in craniofacial bone TE research. We review these state-of-the-art imaging techniques in the ensuing sections.

Imaging Vascular Morphology, Function, and Phenotype in Craniofacial Defects

The vasculature crucially impacts the osteointegration of TE constructs, due to its role in providing oxygen and nutrients and creating an appropriate microenvironment for tissue regeneration (25). Multiple methods have been developed for facilitating vascularization within TE constructs, including porous scaffolds, vascular grafts, and delivering growth factors (26). However, to evaluate these TE designs and the VME around them in vivo is challenging for the following reasons. First, in preclinical small animal models such as the mouse, the diameter of the vasculature can vary from ∼250 μm for veins and ∼150 μm for arteries, down to 4 μm for capillaries (27). Thus, to investigate these microvascular structures, the imaging platform should be capable of acquiring images with high spatial resolution [at least 100 μm for visualizing macrovascular structures and <10 μm for microvasculature (28)]. To investigate the vasculature within a craniofacial bone or around a cranial defect, an optical imaging modality needs to have optical access (i.e., a light path) through an aperture or window. Also, the imaging method must be capable of acquiring images with sufficient CNR to enable one to distinguish vascular structures from the background tissue for subsequent characterization. Moreover, platforms aiming to provide long-term imaging of changes in VME after implantation of a TE construct or during defect healing (e.g., over days or weeks) should not impede the normal functioning of the animal or interfere with the graft site and healing process. In addition to conventional structural or anatomical imaging, some TE applications may require the simultaneous acquisition of functional data such as blood flow within individual blood vessels or overall tissue perfusion, which may require the use of a combination of imaging methods (23).

In clinical settings, nonoptical methods are often used for vascular evaluations in vivo, including X-ray, CT, MRI, ultrasound, and positron emission topography (PET). This is because they have excellent tissue penetration and coverage (∼1 cm–1 m) albeit with lower spatial resolution (∼100 μm–1 cm) (29), which is more suitable for the evaluation of relatively larger diameter vessels (∼100–1,000 μm) in a patient (23). Preclinical imaging of TE constructs and the surrounding vasculature requires much higher spatial resolution because the blood vessels in small animals (e.g., mice) are 10 times smaller than that in humans (27). Thus, studying TE applications in the VME using the clinical nonoptical methods listed above is very limited in the preclinical space due to their low spatial resolution. Recent advances in these imaging techniques have enhanced their potential for evaluating microvessel structures (< 100 μm) in preclinical models. For example, the development of ultrafast ultrasound localization microscopy allows structural and functional imaging of rodent cerebral microvessels at depths of 10 mm below a thinned skull (30). Apart from these new advancements, micro-CT is the most widely used nonoptical imaging method in the preclinical space, as it can achieve submicron spatial resolutions for evaluating microvascular structures (31). For example, using the “VascuViz” imaging workflow recently developed by our laboratory (32), we perfused the murine vasculature at 1, 2, 3, and 4 wk after the creation of a 2-mm cranial defect (Fig. 2A) and performed high resolution (10 μm) ex vivo micro-CT imaging (33). This allowed us to simultaneously assess 3-D changes in vascular structure (Fig. 2B) and bone growth within the defect (Fig. 2C). Using this workflow, we observed that the blood vessel density peaked by the end of week 2 and plateaued during the next 2 wk, which correlated with the high growth rate of bone during the first 2 wk of healing and smaller growth during the subsequent 2 wk (Fig 2D). A limitation of micro-CT imaging in this application was that the vasculature needed to be perfused with a contrast agent, and the imaging was conducted ex vivo. Nevertheless, the VascuViz protocol permits the combination of ex vivo data with in vivo functional data (i.e., blood flow assessed with laser speckle imaging) to yield complementary information on VME changes (32).

Figure 2.

Correlating osteogenesis and angiogenesis in a calvarial defect model with multimodality imaging. A: representative image of a 2 mm mouse calvarial defect (red hatched circle) in an excised sample prepared for multicontrast ex vivo CT imaging using the VascuViz protocol (32). B: three-dimensional (3-D) visualization created from concurrently acquired vascular (red channel) and bone (gray channel) CT volumetric data. Revascularization of the defect site over 28 days is clearly visible in the red channel. C: 3-D bone volume within the defect area segmented from 3-D ex vivo CT data acquired from four animals euthanized at days 7, 14, 21, and 28 postdefect creation, illustrates robust osteogenesis within the calvarial defect. D: plots illustrating the correlation between bone growth measured with ex vivo CT and microvascular density measured with in vivo IOS imaging during bone healing over 4 wk. One can see a significant increase in angiogenesis until day 8 followed by a plateau that correlated with osteogenesis over the same period. Each data point is the means ± SE (preliminary analysis, n = 3 mice., *P < 0.05). Statistical significance was determined using one-way ANOVA. Adapted from Ref. 33. CT, computed tomography; IOS, intrinsic optical signal imaging.

For preclinical TE applications, optical imaging techniques are more widely used due to their adaptability, ease of use, and higher spatial resolution (ranges from <100 μm to <1 μm) (29). Popular techniques include optical coherence tomography (OCT), FL microscopy, photoacoustic imaging (PAI), confocal microscopy, multiphoton microscopy (MPM), and laser speckle contrast (LSC) imaging (23). For the evaluation of craniofacial bone, optical imaging methods can be used to perform ex vivo and in vivo structural and functional imaging of microvasculature. To get optical access to the vasculature inside the craniofacial bone, optical clearing is often used, which allows ex vivo quantification of the vascular distribution and its relationship with bone growth. Luo et al. (34) developed a poly(ethylene glycol) associated solvent system (PEGASOS) tissue clearing method that resulted in complete transparency of craniofacial bone samples. They used Cdh5-Cre; Ai14 mice in which the vascular endothelial cells expressed vascular endothelial-cadherin (Cdh5) and intraperitoneally injected calcein green to evaluate osteogenesis. Two-photon microscopy (2PM) was used to acquire 3-D images of craniofacial bone vasculature and osteogenesis to elucidate their spatial relationship during postnatal craniofacial development (Fig. 3A). The primary advantage of 2PM over conventional fluorescence imaging for these applications is that it allows the acquisition of clearer images (i.e., less background scatter) from different tissue depths to compose a 3-D image volume or depth-resolved image stack (36). This group found that osteogenesis occurred over the entire calvarium during early postnatal growth and was only found in the suture region and bone marrow space after 3–4 wk. The trend was similar for vascular growth, and there was a strong spatial association between craniofacial bone vascularization and osteogenesis during the entire period of postnatal development. In addition, they investigated the association of skeletal stem cells (SSCs) with craniofacial vasculature and examined their role in osteogenesis. They labeled the Gli1+ MSCs by inducing 6–8-wk-old Gli1-CreERT2; Ai14 mice with tamoxifen and performed optical clearing of the calvarial bone and imaged these samples with 2PM to track these cells. They found that the Gli1+ cells exclusively localized in the cranial suture, and osteogenesis made visible by calcein green, which was incorporated into the matrix of osteogenic sites through chelation, appeared adjacent to these cells. In addition, the vasculature around the suture, which was labeled by laminin whole mount staining, was also surrounded by Gli1+ cells. These findings illustrated that there was a strong spatial association of Gli1+ cells with vasculature and osteogenic activity, which indicated the crucial role of this stem cell in craniofacial bone development and regeneration.

Figure 3.

Imaging vasculature in the craniofacial bone ex vivo using tissue clearing. A: two-photon microscopy combined with the poly(ethylene glycol) associated solvent system (PEGASOS) tissue clearing method enabled three-dimensional (3-D) imaging of the craniofacial bone vasculature (red channel) and osteogenesis (green channel). Here, the blood vessels were labeled with vascular endothelial-cadherin (Cdh5) and bone with calcein green. Adapted with permission from Ref. 34. B: light-sheet microscopy of blood vessels labeled with endomucin and CD31 (i.e., red and green channels) and osterix-labeled skeletal progenitors (gray channel) in the parietal and posterior frontal bones of the calvarium. C: images from the boxed region of interest in B, demonstrating the high-resolution and signal quality obtained in each channel with different stains: F555 for endomucin (red channel; C); AF647 plus for osterix (gray channel; D); AF800 plus for CD31 (green channel; E); merged channels (F). Adapted from Ref. 35.

Recently, a finding from Kusumbe et al. (24) illustrated the importance of specific vascular phenotypes in bone growth. In this study, immunostaining was performed on decalcified mouse tibia samples with two different vascular endothelial markers, CD31 and Emcn, and confocal imaging was used to assess the distribution of these markers. It was found that CD31⁺ Emcn⁺ or “type H” vessels exhibited close spatial relationships with osteoprogenitor cells, and the endothelium of this type of vessel mediated neo-angiogenesis in bone. Several ensuing studies have used advanced imaging techniques such as light-sheet microscopy (LSM) and MPM to further investigate the spatiotemporal distribution of these blood vessel phenotypes in craniofacial bone TE applications. Rindone et al. (35) recently developed a quantitative 3-D imaging platform capable of characterizing the distribution of vessel subtypes and their relationship to osteoprogenitors. In this study, the authors were able to optically clear and stain the frontoparietal bones of the mouse calvarium and conduct single-cell (i.e., 1.3 μm x–y pixel size and 2.5 μm z step size) resolution 3-D imaging using LSM (Fig 3, B–F). They found a differential distribution of various vessel phenotypes within the calvarium, where most CD31^hiEmcn⁻ arteries and arterioles were found in the periosteum and dura mater, whereas CD31^loEmcn^hi/lo sinusoids were only present in the marrow cavities. In addition, CD31^hiEmcn^hi capillaries exhibited the most intense expression at the transcortical canals where they connected arterioles in the periosteum and dura mater to the venous sinusoids. Rindone et al. further looked into the spatial relationship of bone progenitor cells with different vascular phenotypes in postnatal and healing calvarium. They found a preferential association of CD31⁺ Emcn⁺ (type H) blood vessels with skeletal progenitors (i.e., Osterix+ or Gli1+ cells) compared with other blood vessel phenotypes. In addition, they observed infiltration by this blood vessel type and its associated bone progenitor cells into the calvarial defect following injury. This high-resolution 3-D map of the cranial VME provides a powerful tool for understanding the role of blood vessel phenotypes in cranial bone regeneration. One could envision leveraging this method to design new TE constructs that directly target osteogenesis-promoting (i.e., type H) vessels for craniofacial defect treatment.

The investigation of these blood vessel phenotypes within the TE context can also be conducted in vivo. Zhai et al. (37) used electrospun nanofibers to deliver bone marrow stromal cells (BMSCs) to critical-sized mouse calvarial defects, which induced robust angiogenesis and osteogenesis. They then imaged the response of different blood vessel phenotypes during this regenerative process and compared it with vessels in defects treated with an acellular construct (Fig. 4). Using multiphoton laser scanning microscopy (MPLSM), a high-resolution imaging technique capable of better tissue penetration (∼400–1,000 μm) compared with confocal microscopy (∼50–100 μm) (38), they performed longitudinal in vivo imaging and showed that CD31⁺ Emcn⁻ vessels exhibited smaller mean diameters, higher blood flow and oxygen tension (pO₂) than CD31⁺ Emcn⁺ vessels during bone healing (Fig. 4, G and I). In addition, they found different distribution patterns for each vessel phenotype within the nonbone-forming and bone-forming regions of the defect. It was observed that CD31⁺Emcn⁺ vessels were mostly located at the central region of a healing calvarial defect and surrounded by osteoblastic cells. These vessels extended into arterial CD31⁺Emcn⁻ vessels, which were found at the periphery of the defect. Together these two types of vessels formed a functional blood vessel network that supplied nutrient and oxygen for bone regeneration. Collectively, these results indicate that different vascular phenotypes exhibited different potentials in terms of their ability to stimulate an osteogenic response. These findings could be used to develop angiogenesis-inducing TE designs that target specific vascular phenotypes for more efficient bone regeneration.

Figure 4.

Different vascular phenotypes in a calvarial defect model and their structural/functional characteristics as assessed with in vivo multiphoton imaging. A: representative image of a 2 mm mouse calvarial defect perfused with an oxygen-sensitive phosphorescence probe (PtP–C343) mixed with an intravascular dye (rhodamine dextran) at 3 wk after defect creation. B: intravital multi-photon laser scanning microscopy (MPLSM) image showing the bone (gray channel) and dextran-rhodamine labeled vasculature (red channel) at the defect site. C: 2PLM image showing pO₂ measurements (yellow dots) made with PtP–C343 within the rhodamine-dextran-labeled blood vessels, from the ROI in the boxed region in B. D: CD31 (red channel) and endomucin or Emcn (green channel) staining reveal CD31⁺Emcn⁻ vessels (red) and CD31⁺Emcn⁺ (green) vessel phenotypes, respectively, in the same region as in C. E: line scans acquired with MPLSM indicate RBC velocity within vessel segments (red channel). One can see the distribution of blood flow within different vascular segments, which can be quantified for different vascular phenotypes. F: CD31 and Emcn staining revealed the vessel phenotypes within the region shown in E. Measurements of vessel diameter (G), RBC velocity (H), and pO₂ (I) within CD31⁺Emcn⁻ vessels (red) and CD31⁺Emcn⁺ vessels (green). These data indicated that CD31⁺Emcn⁺ vessels exhibited smaller diameters, higher blood flows. and higher pO₂ values than CD31⁺Emcn⁻ vessels. (*P < 0.05, ***P < 0.001, n = 150 vessels in 6 mice; Data were statistically analyzed using one-way ANOVA. Scale bar = 100 μm). Adapted with permission from Ref. 37. 2PLM, two-photon phosphorescence lifetime microscopy; RBC, red blood cell.

In addition, to thoroughly characterize the VME in a tissue regeneration study, researchers can harness the advantages of a multimodal imaging system, i.e., one which consists of different imaging techniques or contrast mechanisms to simultaneously assess vascular morphology and function. Recently, Mendez et al. (39) developed a novel multicontrast imaging platform that included intrinsic optical signals (IOSs), LSC, and FL imaging for monitoring vascular changes in a critical-sized calvarial defect model with TE constructs. IOS relies on the wavelength-dependent absorption of oxygenated (HbO₂) and deoxygenated (HbR) hemoglobin, to distinguish blood vessels from the background, as the hemoglobin-rich blood vessels absorb the incident light and appear dark (40). LSC permitted dynamic changes in blood flow to be imaged in vivo due to the interference pattern that arises from moving red blood cells (RBCs) under coherent laser illumination (41). FL imaging allowed quantifications of changes in the angiogenic vascular bed within a tissue-engineered scaffold. This novel method showed the feasibility of correlating the microvascular morphology and hemodynamic parameters with the microvessel maturity in a preclinical calvarial defect model. In addition, the structural and functional data acquired with such imaging systems can be leveraged by tissue engineers to improve the design of their TE constructs. For example, if blood flow measurements indicated insufficient perfusion that can potentially lead to a hypoxic VME, one could envision the use of oxygen-generating scaffold materials. Alternately, proangiogenic growth factors could be added to the TE construct to promote angiogenesis. Recently, using an improved version of this multimodality imaging platform, we successfully conducted longitudinal in vivo functional imaging in a mouse calvarial defect with a cranial window setup (33). Using this platform, we were able to assess in vivo changes in microvascular morphology and bone formation using IOS imaging (Fig. 5A), changes in blood flow with LSC imaging (Fig. 5B), and bone formation with bright-field imaging at high spatial (5 µm) and temporal (200 ms) resolution. Our initial results indicate a strong correlation between increases in vascular density, blood flow, and bone growth (Fig. 5, A and B). Together, these data help us to better understand the healing cascade during craniofacial bone repair. Our novel imaging framework is well-suited for characterizing the VME within the calvarial defect and we are now planning to leverage these in vivo structural and functional data to enhance the design of TE constructs for craniofacial bone regeneration.

Figure 5.

Multicontrast in vivo imaging in a mouse calvarial defect model showing vascular and bone growth over 4 wk. Longitudinal IOS (A) and LSC (B) images from a mouse calvarial defect showing the changes in vascular morphology and perfusion that were concomitant with osteogenesis (black arrows in the IOS images). The red hatched circles in the IOS images indicate the 2 mm defect area. The red arrows in A and B indicate regions exhibiting angiogenesis, whereas the black arrows indicate regions exhibiting osteogenesis. To derive the intravascular oxygen saturation (sO₂) within the cranial window, images were acquired by IOS using 570 nm (C) and 600 nm (D) illumination at D26 postdefect creation. From these images, an sO₂ map (E) was computed using a vascular-specific multiwavelength method (unpublished work). Adapted from Ref. 33. IOS, intrinsic optical signal imaging; LSC, laser speckle contrast.

Imaging Oxygenation within the Craniofacial VME

One of the most important factors that determine the successful integration of TE constructs and eventual tissue regeneration is the local availability of oxygen within the VME (12). As oxygen is a crucial metabolic substrate and signaling molecule for cell survival, multiple TE methods have been developed to enhance the local availability of oxygen within the graft microenvironment including the use of oxygen-releasing or oxygen-generating biomaterials such as peroxide, perfluorocarbons, and micro-tanks (42). However, the efficacy of these methods has only been evaluated based on the resulting tissue or bone growth without direct in vivo assessments of tissue oxygenation. Recently, direct monitoring of oxygenation in vivo has become feasible with two-photon phosphorescence lifetime microscopy (2PLM), in which phosphorescence quenching is combined with two-photon laser scanning microscopy (2PLSM) (43). This method requires the incorporation of specific oxygen probes like PtP-C343, whose light emission in the excited triplet state can be efficiently quenched by molecular oxygen in the tissue. The oxygen partial pressure (pO₂) can then be derived from the measurement of the triplet decay time. The combination of this oxygen monitoring method with 2PLSM enables high spatial resolution oxygenation imaging. In the study by Schilling et al. (44), PtP-C343 was incorporated into electrospun PCL fibers that were implanted into a critical-sized mouse calvarial defect. Using the principles described earlier, they used 2PLM combined with 2PLSM to measure the pO₂ at the defect site. They observed decreased in vivo pO₂ 7 days after defect creation, indicating the development of hypoxic conditions within the VME. At 19 days postsurgery, when revascularization occurred at the edge of the defect and promoted bone healing, the in vivo pO₂ levels increased within these marginal areas but remained lower in the middle of the defect where less angiogenesis was observed. These findings demonstrated how mapping in vivo tissue oxygenation facilitated a deeper understanding of bone defect healing, which in turn helped to advance the design of oxygen-delivering biomaterials. Using the same method, they also evaluated the pO₂ in different vascular phenotypes within a calvarial defect and found that CD31⁺Emcn⁻ had higher pO₂ levels compared with other phenotypes, which indicated the role of this type of vessel in delivering oxygen to the bone defect site (Fig. 4, C and I) (37). In addition, the in vivo oxygenation within the newly formed blood vessels in a mouse calvarial bone defect can be mapped (Fig. 5, C–E) using high spatial resolution (5 μm) data from multi-wavelength IOS. This method used the differences in absorption of HbO₂ and HbR at different incident wavelengths. For example, under 570 nm illumination, HbO₂ and HbR exhibit similar absorption, resulting in the intensity of the IOS image being inversely related to the total hemoglobin concentration (HbT) (40). Therefore, assuming a constant hematocrit, HbT can be used as a surrogate of the total blood volume and for imaging the microvascular structure of perfused vessels (45). Under 600 nm illumination, HbR exhibits high absorption relative to that of HbO₂, which enables one to estimate the HbR concentration. As the HbO₂ can be estimated by subtracting HbR from HbT, one can then derive the oxygen saturation (sO₂), which is the fraction of HbO₂ relative to HbT (40). One could clearly visualize the sO₂ distribution within angiogenic vessels in osteogenic areas, indicating those vessels that were functionally supplying oxygenation for bone growth. This difference in the spectral absorption of HbO₂ and HbR has also been exploited in PAI for sO₂ mapping. PAI uses a narrow-pulse-width laser to illuminate the tissue and generate an acoustic signal directly related to the optical absorption properties of the tissue (46). As the scattering of this acoustic signal in the tissue is less than optical scattering, PAI overcomes the limitation of poor tissue penetration of other optical imaging techniques. Similar to the principles described for IOS imaging earlier, PAI can be performed at multiple wavelengths and the relative concentrations of hemoglobin species are calculated by linear unmixing of the multispectral PAI data (46). PAI has been implemented at a wide range of spatial resolutions and penetration depths ranging from submillimeter resolution at depths of several centimeters for photoacoustic tomography (PAT) to submicron resolution at a depth of a few hundred microns for photoacoustic microscopy (PAM). As a result, PAI has found utility in clinical and preclinical imaging (47). For example, Zhu et al (48) reported an ultrafast widefield PAI system capable of real-time in vivo imaging of morphological and functional vascular changes through a cranial window. They showed that the system could achieve submicrovessel resolution (∼10 μm) imaging over the entire cortex of an adult mouse brain. They demonstrated that arterial and venous structures reacted differently to hypoxic conditions, with significant sO₂ changes in veins and more vasodilation in the arteries. We believe that this method can also be used as a tool for evaluating and optimizing the design of oxygen-releasing scaffolds for bone regeneration. For example, the oxygenation of vessels can be mapped within bone defect models with and without scaffold implantation, and their comparison would indicate the effectiveness of the scaffold in improving oxygenation inside the wound or defect.

In addition to imaging oxygen-eluting scaffolds, such methods could also be applied to evaluate other TE designs. For example, in a rodent model of traumatic brain injury, Bragin et al. (49) investigated the role of the Drag-reducing Polymer (DRP), a “liquid scaffold,” on microvascular perfusion via a cranial window system. Using in vivo 2PLSM, they measured microvascular blood flow and tissue oxygenation in traumatized rat brains with and without DRP injection. After traumatizing the brain, there was a significant increase in intracranial pressure, which compromised blood flow in functional capillaries and diverted the flow to pathological nonnutritive microvascular shunts (MVSs). It was found that DRP effectively reduced the pathological blood flow in MVS and partially restored perfusion in collapsed capillaries. Overall, this study confirmed the effectiveness of the liquid scaffold DRP in decreasing intracranial pressure (ICP) and tissue hypoxia resulting from brain trauma. Therefore, monitoring oxygenation with advanced imaging technologies could enhance the workflow for testing the efficacy of novel TE constructs.

Imaging Cells and Growth Factors within the Craniofacial VME

Other factors within the VME also play an important role in TE applications. For example, growth factors (GFs) can stimulate different cellular processes during tissue healing, such as cell proliferation, migration, and differentiation (50). TE techniques such as GF-delivering scaffolds and GF-signaling modification can directly modulate vascular growth and remodeling via cytokines, such as VEGF (51), PDGF (52), and TGFβ (53). In TE applications, different cell types such as progenitor cells (37, 54) and immune cells (55, 56) can impact tissue regeneration. Collectively, these factors are crucial to nutrient delivery and successful integration of the tissue-engineered construct and need to be monitored in vivo. In addition, growth factors and cytokines for vascularization have also been imaged in wound healing and TE applications (28, 57). However, it is challenging to monitor these elements in vivo because higher spatial resolution, up to cellular scales (i.e., <5 μm) is required, and combining high-spatial resolution cellular data with functional data from the VME requires specialized multimodal imaging systems. Several research groups have combined innovative TE models with various ex vivo and in vivo imaging techniques to facilitate our understanding of the role of these growth factors.

Longitudinal in vivo imaging of cells in craniofacial TE applications has been realized using an optical window setup combined with high-resolution intravital microscopy. For example, Huang et al. (54) used an osteogenic-specific promoter-driven GFP reporter mouse model (Col2.3GFP) to track osteogenesis and angiogenesis during cranial bone defect healing (Fig. 6). Using this transgenic model and MPLSM, they showed that early-stage donor-dependent bone formation occurred when bone marrow stromal cells (BMSCs) were implanted. Via in vivo imaging over 9 wk, they found that osteoblastic activities were initiated at the leading edge of the bone defect during the early stages of healing and that expansion of osteoblasts was coupled with vigorous angiogenesis. Both, angiogenic and osteoblast coverage decreased after 9 wk. This combination of TE model and MPLSM enabled high-resolution in vivo tracking of the vascular and cellular events during early-stage bone healing and helped elucidate the spatiotemporal regulation of osteogenesis and angiogenesis.

Figure 6.

Longitudinal in vivo multi-photon laser scanning microscopy (MPLSM) imaging of a calvarial defect model. A–F: healing of a 1 mm calvarial defect was tracked over 9 wk using in vivo MPLSM. Progressive defect healing and neovascularization were simultaneously imaged by combining multichannel images of Col2.3GFP(+) osteoblasts (green channel), SHG (+) bone matrix (white channel) and blood vessels (red channel). Here, the scale bar = 100 μm. These data indicated that robust angiogenesis, osteogenesis, and cell recruitment occurred during the 9 wk of healing. Quantitative analyses of osteoblast cell volume (B), bone volume (C), center defect area (D), and cell advancing distance relative to the radius of the original defect indicated a strong correlation between osteoblast distribution and bone growth. Data are expressed as the means ± SE (n = 4 mice). *Significant difference (P < 0.05). Statistical significance was determined using one-way ANOVA and a Tukey’s post hoc test. Adapted with permission from Ref. 54.

Researchers have also developed novel optical window setups to allow the simultaneous visualization of vasculature, bone, and cells surrounding an implant. Khosravi et al. (58) reported a novel cranial implant window chamber (CIWC) they designed that stabilized a titanium implant within a mouse cranial defect and allowed tracking of vascular network, bone growth, and progenitor cells in the peri-implant wound microenvironment. They showed that this model can be combined with both confocal and multiphoton microscopic imaging systems and visualize vascular network by FITC-dextran injection, mesenchymal progenitor cells (MPCs) labeled with tdTomato, and bone collagen matrix in the SHG channel. To quantify the visualization results, they also custom-designed a graphical user interface (GUI) written via the scripting language MATLAB (Mathworks, MA) to process and analyze the changes in vascular and cellular distribution. Using this model, this group has shown that surface topography of a metallic implant in a calvarial defect can greatly affect the pattern of angiogenesis in the wound site (59). In addition, recruitment of cells can also be affected by the implant topography. It was illustrated that a nanotopographically complex implant surface can benefit the recruitment of endothelial cells and mesenchymal progenitor cells, which both contribute to bone repair (60). They found that local cytokine gradient release from platelets can be increased by the initial contact with the implant, which facilitated the migration of these cells. These findings not only elucidated the biology of bone defect healing with an implant but also implicated the benefits of using nanotopographically complex biomaterials in the design of metallic implants.

Recently, efforts have also been made to investigate the functions of growth factors (GFs) in vivo. Several growth factors have been shown to promote craniofacial bone healing, including VEGF, BMP-2, and PDGF, to name a few (61, 62). Previously, radiography or micro-CT was used to evaluate changes in defect size following the delivery of growth factors to the defect site to promote healing (61, 62). Recently, with the advent of molecular imaging technologies, simultaneous tracking of growth factors and bone mineralization in the cranial bone has become possible. Molecular imaging methods including in vivo FL imaging (63), in vivo bioluminescence imaging (BLI) (64), PET (65), and single-photon emission computerized tomography (SPECT) (66) have been used to track the expression of growth factors and their effects on craniofacial bone regeneration. Nisarg et al. (63) incorporated BMP-2 and PDGF-BB in a nanolayer coating on the PLGA polymer scaffold, which was then implanted into a critical-sized rat calvarial defect. The release of PDGF-BB and BMP-2 was tracked by near-IR fluorescent dyes and in vivo fluorescence imaging, whereas bone healing was monitored with micro-CT. The tracking of growth factors confirmed their sustained release for ∼11 days for PDGF and ∼20 days for BMP-2, respectively. Micro-CT imaging of the defect volume indicated that either individual or combined release of these two growth factors could enhance craniofacial bone healing. In addition to evaluating their effect on defect size, the direct biological effects of growth factors can also be monitored with molecular imaging. PET imaging has been used to noninvasively evaluate the osteogenic potential of a BMP-2-releasing calcium phosphate cement (CPC) bone substitute implanted in a rat critical-sized calvarial defect (65). ¹⁸F-Fluoride uptake around the implant was monitored, as previous studies have shown that increased ¹⁸F-Fluoride uptake signals in bone indicated osteoblast activity and osteoid production (67). It was found that there was a significant increase of ¹⁸F PET signal around the BMP-2-releasing scaffold, and it was positively correlated with the bone volume increase observed with micro-CT. Similarly, SPECT has been combined with micro-CT to longitudinally evaluate bone mineral formation around a BMP-2-releasing hydrogel implant inside a mouse critical-sized defect (66). In this study, the accumulation of bone mineral was monitored by SPECT through injection of 99mTc-MDP, which chelated the bone minerals, and its distribution was noninvasively monitored by SPECT, which showed higher osteoblastic activities at the defect site with BMP-releasing hydrogels. Overall, tracking cells and GFs within the VME can further improve our understanding of the integration of a TE construct at the host defect site and is especially useful for evaluating the efficacy of cell/GF-delivering TE constructs.

Systems Biology for Craniofacial TE Applications

Although the VME in TE applications can be monitored and evaluated ex vivo or in vivo using advanced imaging techniques, each evaluation and improvement can require several experiments or iterations, which prove challenging when animal models are involved. Moreover, the use of a large number of animals or imaging modalities such as MRI, PET, etc., can prove prohibitively expensive. Therefore, researchers can turn to the in silico space or computational models to obviate some of these concerns. Such mathematical models or simulations could be used to test the performance of a TE construct within a given VME, or to model the effect of different VME cues on the efficacy of the TE construct. Recently, we and others have combined systems biology, imaging, and computational techniques to model the VME or evaluate TE designs (32, 68, 69). A recent example is the use of “network analysis,” which is a systems biology approach, to identify growth factors that can be used in a TE application for tissue regeneration (70). Beheshtizadeh et al. (70) used this method to analyze growth factors contributing to angiogenesis or osteogenesis in bone TE based on in vitro and in vivo data. They identified three significant proteins, including prostaglandin-endoperoxide synthase 2 (PTGS2), TEK receptor tyrosine kinase (TEK), and FGF18 for incorporation in a TE design that would upregulate osteogenesis and angiogenesis. In addition, convenient toolboxes have been developed that use systems biology techniques for visualizing and quantifying different VME parameters in vivo. Such tools could hugely benefit TE researchers by enabling them to characterize the VME more holistically in TE applications. An example of such a toolbox is HemoSYS, which quantifies several hemodynamic parameters within the VME using multicontrast optical imaging data (Fig. 7) (69). HemoSYS is a freely available MATLAB-based interactive toolbox that can yield powerful image-processing results without requiring any programming expertise to operate. When multicontrast in vivo optical images from a VME are imported into the toolbox, it can process them and quantify several VME parameters such a hypoxia, blood flow regulation, hemodynamic response, and spectral heterogeneity of different hemodynamic variables (Fig. 7A). Here, we demonstrate the utility of this toolbox in a craniofacial bone healing study, by using the “coupling analysis” module of HemoSYS. This module quantifies the correlation or “coupling” between blood flow (BF) and blood volume (BV) during the cranial bone healing cascade. In this experiment, a 2 mm cranial bone defect was created on a mouse and was allowed to heal for 4 wk. At day 26, after defect creation, the mouse was administered carbogen gas (i.e., a mixture of 95% O₂ and 5% CO₂) for 3 min after initially breathing room air. This resulted in vasodilation of mature blood vessels that was imaged with IOS (i.e., BV) and LSC (i.e., BF) imaging. Next, these time-series data were processed using the HemoSYS pipeline for “coupling analysis.” The coupling between BF and BV was calculated as the correlation coefficient (r) between their time series acquired during the carbogen gas administration. This can be used to characterize the degree of BF regulation via local vasodilation or vasoconstriction, where a higher “r” value is indicative of vessel maturity. It was observed that in some regions (Fig. 7D), angiogenic vessels exhibited similar levels of coupling compared with mature cortical vessels, which could be observed at the time of defect creation (Fig. 7C), indicating that they were well-perfused and supported bone healing. In contrast, in other regions (e.g., region 3, Fig. 7E), sprouting angiogenic vessels exhibited much less coupling between BV and BF, indicating that their supplying function was not yet well-established. This tool could potentially help tissue engineers investigate changes in microvascular hemodynamics associated with any TE construct or application. Recently, techniques have also been developed for the simultaneous imaging of vasculature with multimodality imaging platforms, which can then be integrated for systems biology characterizations of the VME. For example, we developed a multimodality workflow named “VascuViz” for simultaneous 3-D imaging and visualization of the VME using MRI, CT, and optical microscopy in the same tissue sample (32). Our demonstration of the utility of VascuViz in several systems biology applications also indicated its adaptability for TE applications. For example, the VascuViz workflow was compatible with common TE evaluation methods including tissue clearing and standard histopathology, and could also be used to characterize the VME from the whole tissue to cellular scale. Furthermore, it could be used as the basis for hemodynamic simulations and integrated with cellular scale maps for correlative investigations of the graft site. In addition, this method could potentially be used for the in vivo characterization of the VME associated with any TE design, since their study showed that VascuViz enabled the integration of in vivo blood flow data with ex vivo 3-D vasculature and bone data acquired using micro-CT from a murine calvarium. Collectively, we believe that such “image-based” systems biology tools have great potential in aiding the design, development, and testing of novel TE constructs.

Figure 7.

Utility of the HemoSYS toolkit to characterize blood flow regulation within the VME of a calvarial defect model. A: schematic of the multicontrast optical imaging system for acquiring in vivo images of the different hemodynamic parameters to be analyzed with “HemoSYS.” These contrast mechanisms include fluorescence (FL), intrinsic optical signal (IOS), and laser speckle (LS) imaging. As an example, “coupling analysis” can be performed by correlating the blood volume (BV) time-series acquired with IOS and the blood flow (BF) time-series acquired with LS on a pixelwise basis. Adapted from Ref. 69. B: the HemoSYS coupling analysis pipeline was applied to a 4-wk-long calvarial defect healing study. The changes in the coupling (i.e., temporal correlation) between BV and BF during a vasodilation experiment were assessed in the following vessels: cortical vasculature (red box); anastomosed angiogenic vessels (green box); and sprouting angiogenic vessels (blue box). It was observed that the anastomosed angiogenic vessels (D) exhibited similar levels of coupling to the cortical vessels (C), indicating that they were well-perfused to support bone growth. In contrast, sprouting angiogenic vessels (E) exhibited lower correlations between BV and BF relative to cortical and anastomosed angiogenic vessel, which might be indicative of poor or yet to be established perfusion. VME, vascular microenvironment.

Future Directions: Point-of-Care Approaches and Rapid Prototyping/Testing/Design

As explained in the previous sections, multiple important VME parameters, including vascular structure, function, phenotypes, oxygenation, the surrounding cells, and GFs can be evaluated by various imaging modalities. If these VME parameters are evaluated before the implantation of a TE construct, the resulting data can indicate specific problems within the wound/defect site, which can then inform the design of the targeting TE construct to address the problems. For example, if the vascular density information extracted from imaging results showed lower-than-expected angiogenesis, one could consider adding angiogenic factors to the TE construct to enhance angiogenesis at the defect site. Another example would be if the in vivo oxygenation map from imaging indicated the presence of hypoxia, the TE construct could integrate oxygen-generating materials to improve oxygenation at the defect site. Similar concepts can be applied to analyze other VME parameters in a specific defect or wound site, and systems biology modeling could then be used to simulate the effectiveness of a TE design. After the TE constructs have been designed based on these imaging data and tested in vitro, they could be implanted into a preclinical model for in vivo evaluations. Here, imaging technologies could again be used to evaluate the effects of the VME on the TE construct, and vice versa. The in vivo imaging evaluations could also be compared with predictive results from computational or systems biology and serve as the ground truth for future simulations or a new iteration of the construct’s design. This feedback loop could be iterated to guide adjustments that optimize the TE design, which constitutes the basis of our concept of “image-informed TE design” (Fig. 8). We envision that in the near future, the image-informed TE design could be point-of-care (POC) treatments (i.e., “as needed” treatments outside the hospital setting, wherever the patient is situated). Using imaging results from the VME in a patient, computational modeling methods could be used to perform rapid prototyping of the TE construct and determine a customized design for that patient. Moreover, researchers are exploring the use of new computer science technologies such as machine learning (ML) and artificial intelligence (AI) for optimizing TE designs, including ML or AI-guided 3-D printing of TE scaffolds (71, 72). Although in its infancy, several recent perspective papers have pointed out the potential of AI in tissue engineering applications and workflows (73, 74). We believe that together, these technologies would help realize rapid point-of-care TE designs for clinical applications.

Figure 8.

Schematic illustrating the “image-based tissue engineering” workflow. The VME of a tissue engineered site (e.g., bone defect, wound, etc.) can first be characterized by selecting a range of appropriate imaging techniques. Next, quantitative VME parameters acquired from imaging can be used to inform the design of the TE construct. The efficacy of the TE construct can then be evaluated in vitro, ex vivo, or in vivo with imaging, including its effect on the VME. Based on these results, tissue engineers can iteratively modify the design of the TE construct and repeat this process until an optimal design is achieved. Finally, image-based computational modeling and simulations can also be conducted in parallel to predict the performance of the TE construct, or identify relevant VME parameters to be modulated for optimal integration and healing. Schematics created using BioRender.com. TE, tissue engineering; VME, vascular microenvironment.

CONCLUSIONS

The vascular microenvironment (VME) is an important factor to consider in craniofacial TE, and there exist various imaging techniques that can help characterize it from the whole organ to cellular scales. Parameters such as vascular morphology, function, phenotype, oxygenation, and surrounding cell types within the VME can be monitored in vivo by advanced imaging techniques. Monitoring these parameters can require multiple imaging techniques with distinct advantages. First, high-resolution and high contrast-to-noise ratio imaging is needed to clearly visualize microvascular structures and GFs and cells (28). Second, since the sprouting microvascular network is often embedded in tissue, the imaging technique must be capable of “seeing” through regenerating tissue (75). Third, the imaging method should be able to yield quantifiable physiological parameters that are relevant to the graft microenvironment, such as blood flow and intravascular oxygenation (76). Multiple imaging techniques and preclinical models have been developed to satisfy these requirements and overcome these limitations, as discussed in the preceding sections. We have also illustrated how the combination of imaging with systems biology approaches can further improve the evaluation and design of TE constructs. We believe that the continuing development of novel imaging modalities, various types of TE models, and new systems biology methods will usher in an era of “image-informed TE design.” Collectively, these advances would accelerate successful point-of-care therapies tailormade to each patient.

GRANTS

This work was supported by National Cancer Institute Grants 5R01CA237597-03 and 5R01CA196701-05, National Institute of Dental and Craniofacial Research Grant 5R01DE027957-04, and National Institute of Arthritis and Musculoskeletal and Skin Diseases 5R01AR077581-02.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Y.R., W.L.G., and A.P.P. conceived and designed research; Y.R., J.S., and A.P.P. performed experiments; Y.R., J.S., and A.P.P. analyzed data; Y.R., J.S., W.L.G., and A.P.P. interpreted results of experiments; Y.R., J.S., and A.P.P. prepared figures; Y.R. and A.P.P. drafted manuscript; Y.R., J.S., W.L.G., and A.P.P. edited and revised manuscript; Y.R., J.S., W.L.G., and A.P.P. approved final version of manuscript.