Bioengineering of Humanized Bone Marrow Microenvironments in Mouse and Their Visualization by Live Imaging

Human hematopoietic stem cells (HSCs) reside in the bone marrow (BM) niche, an intricate, multifactorial network of components producing cytokines, growth factors, and extracellular matrix. The ability of HSCs to remain quiescent, self-renew or differentiate, and acquire mutations and become malignant depends upon the complex interactions they establish with different stromal components. To observe the crosstalk between human HSCs and the human BM niche in physiological and pathological conditions, we designed a protocol to ectopically model and image a humanized BM niche in immunodeficient mice. We show that the use of different cellular components allows for the formation of humanized structures and the opportunity to sustain long-term human hematopoietic engraftment. Using two-photon microscopy, we can live-image these structures in situ at the single-cell resolution, providing a powerful new tool for the functional characterization of the human BM microenvironment and its role in regulating normal and malignant hematopoiesis.


Introduction
Cell fate decisions observed in stem cell compartments are tightly regulated by both intrinsic and extrinsic factors. In particular, it is now widely recognized that the BM microenvironment plays a fundamental role in controlling the switch in HSCs from a quiescent to an active state, as well as in their self-renewal or differentiation fate decision 1 . Moreover, recent findings indicate that hematological malignancies affect the function of the BM microenvironment, pointing to the existence of active crosstalk between the two compartments 2,3,4,5,6 . Despite recent advances, many key questions remain about how the activity of specific BM-niche components contribute to HSC behavior and malignant transformation.
The BM microenvironment is a highly heterogeneous and complex mixture of many different cell types, each with specialized functions. The abundant endothelial (EC) and vascular component regulate nutrient and metabolite turnover, the ingress and egress of different cells to and from the BM, and several HSC functions 7,8 . Mesenchymal stromal cells (MSCs), a heterogeneous population of undifferentiated stem cells and progenitors committed through three different lineages (i.e., osteogenic, chondrogenic, an adipogenic), are another fundamental component of the BM niche. These MSCs localize both in central areas of the BM and in the proximity to the endosteal region. They may be associated with vascular structures and are implicated in the regulation of HSC function 9,10,11,12,13,14,15 . Many reports suggest that HSCs reside in various defined sites within the marrow and that their function might depend upon their precise localization. Most of the present knowledge regarding HSCs and their interaction with the BM microenvironment derives from murine studies 1 .
The use of xenograft models has extended this knowledge to human normal and malignant HSCs, engrafting within the murine BM of immunodeficient mice cone supplying 1.5% isoflurane and 2 L/min O 2 . Keep the mouse under anesthesia during the surgical procedure and frequently check the animal state. 3. While it is under anesthesia, use ophthalmic gel on the eyes of the mouse to prevent dryness, and keep the mouse at 37 °C. 4. Shave the surgical area on the back of the mouse using an electric trimmer. To sterilize the skin, dip a cotton tip in diluted clorexidine (diluted 1:10 in PBS) and use this tip to clean the skin surface. Repeat this procedure twice. 5. Using sterile forceps and a scalpel (or scissors), make a 0.5-to 0.7-cm anterior-to-posterior full incision of the skin. Use the forceps inserted under the subcutaneous tissue to make a pocket. 6. Insert the scaffold subcutaneously, making sure that it is placed deep within the pocket (Figure 1G and H). Close the incision with surgical glue (Figure 1I and J). 7. To treat post-surgical pain, subcutaneously administer buprenorphine (0.1 mg/kg bodyweight). 8. During recovery, place the animal on its side in a pre-warmed cage and monitor recovery until normal behavior is observed. 9. Dilute pain medication in water (carprofen, 0.1 mg/mL of water) and provide it to animals as drinking water for 4 days after surgery. 10. Check the animal and the wound frequently during the 48-h post-surgery period for possible adverse effects.

Mouse Treatments, Euthanasia, and Sample Retrieval for Imaging
NOTE: Analysis of the scaffolds is performed between 8 and 24 weeks post-implantation.
1. 60 min before imaging, keep the implanted mouse warm in a heating box at 37 °C and intravenously administer 100 µL of human immunoglobulin to block unspecific sites. 2. 30 min before imaging, intravenously administer 10 µg (per mouse) of specific antibodies to label the cells of interest. 3. 5 min before imaging, intravenously administer 15 µL (diluted in 100 µL of PBS) of 655-nm fluorophore-labeled, vessel-pooling agent (655-VPA) to visualize vascular structures. 4. Euthanize the mouse via cervical dislocation. 5. Using sharp scissors, make a longitudinal skin incision on the back of the mouse, near the original implantation site. 6. With the help of tweezers and scissors, carefully separate the skin from the subcutaneous pocket where the scaffold has been implanted. 7. Hold the scaffold with tweezers and gently explant it from the skin by cutting the residual membrane and tissue surrounding the scaffold using scissors. See examples of scaffolds to be recovered in Figure 2. 8. Secure the scaffold with fast-acting adhesive glue to an imaging plate (a 35 mm x 10 mm Petri dish) and fill with saline solution (PBS) at room temperature. 9. For BMP scaffolds, before filling the plate with PBS, use a surgical microdrill to thin the bone surface under a microsurgical microscope; this allows fluorophore visualization and high-resolution image capture. Use either the 1.2-or 1.6-mm burrs, depending upon the size of the scaffold. NOTE: The user will realize how much to drill depending upon the thickness of the bone. In general, as the BMP scaffolds are vascularized, the bone will slightly change color and become more red when approaching the correct thickness for imaging. 10. Insert the plate onto the stage of the confocal microscope.

Live-imaging Using Two-photon Microscopy
NOTE: When using non-descanned detectors (NDD), always use the NDD slider for imaging to direct the fluorescence to the NDD. The microscope configuration is provided in Figure 3.
1. Switch on the microscope and the computer, start the software by clicking "Start System," and go to the "Acquisition" mode. 2. Tick the "Show manual tools" box. In the "Laser" menu switch "on" the two-photon laser and allow it to warm up and stabilize. 3. In the "Imaging Setup" menu, simultaneously activate "Channel Mode" and "Switch track every Frame." In the "Light Path" menu, select "Non Descanned'" and "Main Beam Splitter MBS 760+." Tick to activate the four NDDs and set the configuration as illustrated in Figure 3. NOTE: With this configuration, the collagen signal from bone structures (second harmonic generation, SHG) is collected at 380 -485 nm, FITC-hCD31+ human endothelial cells and AF488-hCD45+ human hematopoietic cells at 500-550 nm, and 655-VPA at 640 -690 nm. 4. In the "Channels" menu, set the "laser wavelength" to 890 nm and the power to 50%. Set the "Gain (Master)" to 500-600, the "Digital Offset" to 0, and the "Digital Gain" to 15 for each channel. Adjust these values once the acquisition has started. 5. In "Acquisition Mode," set up the required parameters to obtain high-resolution images without damaging the tissue and bleaching the fluorophores. Set "Scan Mode" to "Frame," "Frame Size" to "x512 y512," Line Step" to 1, "Speed" to 9, "Averaging number" to 8, "Bit Depth" to "8 Bit," "Mode" to "line," "Direction" to "bidirectional," and "Method" to "mean." 1. Set the "Zoom" to 1 for the initial scan of the image, and increase it if required to focus on particular areas.
6. Place the plate containing the scaffold on the microscope stage under the 20X, 1.0 NA water-immersion lens and lower the lens until it touches the saline solution. Set the focus of the lens on the scaffold using the microscope eyepieces, using a lamp as the light source. 7. Activate the "Z-stack" menu, select the "First/Last" function, and set the required interval between two sub-sequential slices (e.g., 2-µm Zstack). Keep the intervals constant within the Z-stack. 8. Select "Live" to image a live scan of the sample and adjust the "Digital Gain" and "Digital Offset" for optimal exposure. To visualize multiple channels at the same time, select the "Split" function. 9. In the "Stage" menu, while in "Live" mode, scan the image and "Mark" regions of interest (ROI), such as the location of human hematopoietic cells and vascular structures. When the scan of the sample is complete, move to the first ROI to start imaging.

Sample Processing for Histology and Immunostaining
NOTE: Samples are processed according to the protocol described in the JoVE general laboratory techniques 54 describing sample fixation, embedding, and sectioning processes. Bone-forming samples should be treated for 7 days in an EDTA-based decalcifying agent between the fixation and embedding processes. The blocking/permeabilization solution is 10 mM PBS pH 7.4 buffer with 1% Triton X-100, 1% bovine serum albumin (BSA), and 10% normal goat serum (NGS).
1. Put the slices in xylene for 10 min), xylene for 5 min, 100% ethanol for 5 min, 70% ethanol for 5 min, 50% ethanol for 5 min, and H 2 O for 5 min. 2. Transfer the slices to a citrate-based antigen unmasking working solution.

Representative Results
In Figure 1, representative images of the scaffold cell seeding and implantation processes are shown. In Figure 1C, note that cells are injected directly into the scaffold. In Figure 1G, note that an incision is made in the back of the mouse, where the subcutaneous pocket is created and the scaffold is implanted. Figure 2 shows the gross morphology of different scaffolds implanted in NSG mice and retrieved after 8 weeks. Note the slight vascularization in hMSC seeded scaffolds (Figure 2A). The co-seeding of human ECs with hMSCs in the scaffold allows for the formation of more relevant vasculature in scaffolds ( Figure 2B). Finally, the presence of rhBMP-2 induces bone formation. The retrieved scaffolds are bigger in this case, and they are constituted by bone-resembling hard tissue. Figure 3 shows the channel configuration setup on the microscope for live-imaging with NDD (details in the figure legend). Figure 4 and Video 1 show human hematopoietic cells in hMSC-coated scaffolds. Scaffolds were explanted 8 weeks post-implantation and after the intravenous inoculation of AF488-hCD45 antibody and 655-VPA. This procedure allows for the visualization of implanted human hematopoietic cells and the vascular structure by two-photon confocal microscopy. In this case, the images show blood vessels (655-VPA) in scaffolds and the long-term engraftment of human hematopoietic cells (AF488-hCD45) in the scaffold parenchyma. Figure 5 and Video 2 correspond to human scaffolds seeded with hECs and hMSCs. 8 weeks after surgery, scaffolds were explanted after the intravenous inoculation of FITC-hCD31 antibody and 655-VPA, and images were acquired with a two-photon confocal microscope, as mentioned before. Images show the participation of hECs in vessel formation in the scaffold, resulting in a murine-human chimeric vasculature. Figure 6A shows representative data of the approach used to stimulate bone formation in MSC scaffolds. Similar to previous figures, 8 weeks after implantation, the intravenous inoculation of 655-VPA was performed, scaffolds were retrieved, and images were acquired with two-photon confocal microscopy. rhBMP-2-stimulated scaffolds induce the formation of bone tissue, which could be visualized due to the SHG (cyan color in the images) provided by the calcium in the bone. The provided images also show the formation of cavities and vascularized endosteal tissue, which highly resemble the BM endosteal tissue. In Figure 6B and Video 3, hECs were co-implanted with hMSCs. Scaffolds were retrieved after the intravenous inoculation of FITC-hCD31 antibody and 655-VPA, and two-photon confocal microscopy images show the participation of hECs in the neovascularization in a bone-forming scaffold. Figure 7 shows representative images of histology, a procedure performed to corroborate previously described results. Immunofluorescent images show mouse vasculature, hECs, hMSCs, and long-term engrafted human hematopoietic cells in the scaffold structures. Scaffolds retrieved from mice were fixed and used for immunofluorescence. In the rhBMP-2 carrier bone-forming scaffolds (Figure 7D-F), note the morphology of the tissue, resembling mature bone marrow with adipose tissue. In this bone-forming scaffold, we show that hMSCs are fibroblasts, which would indicate that they contribute to newly formed tissue as stromal cells. We also show human adipocyte marker expression, which would indicate that hMSCs also contribute to adipose tissue formation.

Significance with Respect to Existing Methods:
In this protocol, we described a method to generate different humanized microenvironments in mice and to visualize their architecture via twophoton microscopy and histology. The representative data provided shows the feasibility of the approach, using different stromal cells to engineer humanized tissues. The protocol has specific applications to the study of human hematopoietic cells and bone marrow niche-derived cells in normal and pathological conditions. These applications include the study of clonal evolution, drug screening, and crosstalk between human HSCs and stromal components. In the emerging field of tissue engineering, several alternative approaches have been proposed. Approaches of note include the development of 3D humanized BM structures in vitro 55,56,57,58,59,60,61,62,63 and the orthotopic graft of humanized BM scaffolds in mice 64 . Our approach has the advantage of combining both the complexity of the in vivo system with the easy anatomical accessibility of the humanized tissue graft.

Modifications and Troubleshooting:
A source of variability in this protocol can be found in the selection of cells used to seed the scaffolds. In our work, we used BM-derived hMSCs. However, mesenchymal cells can be obtained from several tissues, which may show distinctive properties depending upon the origin. Therefore, the use of hMSCs derived from different organs can be considered. However, their ability to form bone tissue in vivo should be tested prior to use in this protocol.This protocol uses a commercially available human endothelial cell source (i.e., E4ORF1-transduced HUVEC). Recently, the use of organ-specific endothelial cells for different purposes has been reported 65,66 . Furthermore, the use of primary hECs derived from the BM could represent an interesting improvement to the protocol. Therefore, the use of different sources of endothelial cells may produce different in vivo outcomes.
We used NSG immunocompromised recipient mice to favor the implantation of humanized scaffolds and to avoid tissue rejection. We do not exclude the possibility of using this protocol to engineer ectopic bone marrow tissues in other mouse strains. Indeed, rhBMP-2 can induce bone formation in different mammalian models 47,48,49,50,52 . However, differences in cell viability and long-term transplantation are likely to be observed using different strains/models. The timing of scaffold recovery can also be flexible, depending upon the final purpose of the experiment. In the presented protocol, we recover samples at 8 -12 weeks after implantation to assess long-term hematopoietic engraftment. To study early steps of the human BM niche formation (e.g., osteochondral tissue formation 47 or vascular development), different time points can be chosen.
The live-imaging technique we described in this protocol is indicated for short-term imaging of explants. The use of an equilibrated chamber for maintaining physiological temperature, oxygen tension, and CO 2 concentration should be considered in cases of long-term imaging, such as to study motility behaviors.

Critical Steps within the Protocol:
Among the challenges related to the protocol, we would highlight the technical skills required for some steps. Mesenchymal and endothelial cells should be used at low cell passage numbers; otherwise, they will not be able to support human hematopoietic cell engraftment in vivo or to participate in de novo vasculature and bone formation in vivo. We recommend the use of hMSCs and hECs at passages 1 -5. Scaffold preparation and cell-seeding steps require basic cell-culture skills and knowledge of the properties of the specific cells used in the procedure. The surgery protocol is quite straightforward but requires some practice. Maintenance of an aseptic environment to avoid contamination of the implanted scaffolds in immunodeficient mice is crucial to ensure the success of the experiment. Sample explant and live imaging require surgical