Injectable Therapeutic Organoids Using Sacrificial Hydrogels

Summary

Organoids are becoming widespread in drug-screening technologies but have been used sparingly for cell therapy as current approaches for producing self-organized cell clusters lack scalability or reproducibility in size and cellular organization. We introduce a method of using hydrogels as sacrificial scaffolds, which allow cells to form self-organized clusters followed by gentle release, resulting in highly reproducible multicellular structures on a large scale. We demonstrated this strategy for endothelial cells and mesenchymal stem cells to self-organize into blood-vessel units, which were injected into mice, and rapidly formed perfusing vasculature. Moreover, in a mouse model of peripheral artery disease, intramuscular injections of blood-vessel units resulted in rapid restoration of vascular perfusion within seven days. As cell therapy transforms into a new class of therapeutic modality, this simple method—by making use of the dynamic nature of hydrogels—could offer high yields of self-organized multicellular aggregates with reproducible sizes and cellular architectures.

Graphical Abstract

Highlights

• •Therapeutic, prevascularized organoids were formed in a sacrificial scaffold
• •The organoids are highly reproducible and grown in a high-throughput manner
• •The organoids rapidly formed perfusing vasculature in healthy mice
• •Therapeutic potential was assessed in a mouse model of peripheral artery disease

Biotechnology; Bioelectronics; Biomaterials

Introduction

Organoids, such as vascularized organoids or spheroids (de Moor et al., 2018, Wimmer et al., 2019, Mcguigan and Sefton, 2006), are three-dimensional multicellular clusters that mimic the structure and function of native tissues and are useful for on-chip drug screening (Alajati et al., 2008, Nam et al., 2015). For use as a cell therapy, delivery of cells within well-controlled microenvironments, rather than suspensions of isolated cells, could promote and maintain desired cellular functions within dynamic and complex in vivo environments (Takebe et al., 2013, Walser et al., 2013, Yap et al., 2013, Dissanayaka et al., 2014, Verseijden et al., 2010, Meyer et al., 2012). As organoids are increasingly being explored for in vivo studies and therapy, there is increasing recognition of the unmet challenge in generating multicellular aggregates with high reproducibility and control. As one example, even though control over “organoid size, shape, cellular composition and 3D architecture … is essential in order to understand the mechanisms that underlie organoid development in normal and pathological situations, and to use them as targets for manipulation or drug testing,” reproducibility has been cited as “the major bottleneck of current organoid systems” (Huch et al., 2017).

The major current methods for generating organoids include spinner cultures (Sutherland et al., 1971), hanging drops (Tung et al., 2011, Frey et al., 2014), and non-adhesive 96-well plates (Ehsan et al., 2014, Yuhas et al., 1977, Metzger et al., 2011, Wenger et al., 2005) (Table S1), but these methods are difficult to scale or harsh to cells. (Alternatively, microtissues that are “cells in gels” (Chung and Park, 2009, Griffin et al., 2015, Huebsch et al., 2015, Li et al., 2014) typically feature cells moving to pre-formed pores within a hydrogel scaffold, but the cells are limited in their ability to self-organize into desired structures (Ovsianikov et al., 2018), and the resultant gels exhibit variable structures and sizes dependent on the pores and may be undesired in the implanted site due to potential immunogenicity.) More recently, methods to fabricate organoids based on micro-sized wells have faced challenges of either high adsorption (of steroid hormones, small molecules, and drugs (Li et al., 2009, Toepke and Beebe, 2006) for PDMS-based wells) or inefficient and harsh processes, usually involving vigorous pipetting or high-speed centrifugation, to separate and remove the cellular clusters from the microwells. Such procedures produce cellular clusters at a low yield and could damage cellular structures and function. Recognizing this limitation, other studies have proposed more complex methods to actively release cellular clusters from microwells (Shimizu et al., 2013, Tekin et al., 2010, Anada et al., 2010). As such, there still lacks reliable methods to generate organoids at high yield and with reproducibility and control over aggregate size and cellular organization.

Many hydrogels are biocompatible and have been used as a dynamically responsive biomaterial (such as microfluidic valve (Beebe et al., 2000), changing cellular microenvironment (Gillette et al., 2010), and stimuli-responsive drug release (Zhao et al., 2011)). We hypothesize that a dynamic change in the cross-linking state of hydrogels could gently release organoids and sought to demonstrate the strategy for producing large numbers of vascularized organoids to treat a mouse model of peripheral artery disease (Novosel et al., 2011, Dimmeler et al., 2014, Kim et al., 2016, Sun et al., 2016, Lovett et al., 2009), the most severe form of which is critical limb ischemia (CLI), which can lead to amputations (Davies, 2012). Thus far, attempts of cell therapy for treating CLI (including tens of clinical trials) have featured mesenchymal stem cells (MSCs) (Chen et al., 2008, Davies, 2012, Lawall et al., 2011, Raval and Losordo, 2013, Tongers et al., 2008) (which have not yet shown significant new re-perfusion) or endothelial cells (ECs) (which could die from prolonged deprivation of oxygen and nutrients before they form the desired blood vessels and anastomose with host vasculature (Benoit et al., 2013, Botham et al., 2013, Lawall et al., 2011)). We sought to demonstrate our approach to produce organoids with high reproducibility and scalability, as well as the ability to retain functionality after passing through needles to obviate invasive surgery (Kusamori et al., 2014, Lee et al., 2011, Mirabella et al., 2017) of ischemic sites with impaired wound healing. We also assessed the ability of the pre-formed blood-vessel units, after injection, to rapidly integrate with the host's vascular network in a healthy mouse model and to restore perfusion in an ischemic hindlimb mouse model.

Results

Hydrogels as a Sacrificial Scaffold and as a Gentle and Scalable Method for Producing and Harvesting Organoids

Sacrificial materials are widely used in micromachining of microelectromechanical systems (MEMS) to release patterned metals or semiconductors from a substrate (Figure 1A). To cellular structures, some hydrogels (such as agarose or poly(ethylene)glycol as previously demonstrated) can be non-adhesive and thereby promote cells to interact with one another and contract into microtissues and organoids (de Moor et al., 2018, Lee et al., 2018). We hypothesized that a dynamic change in the cross-linking state of alginate, which can be achieved by adding calcium or a chelator and has been demonstrated for other purposes (Chen et al., 2017, Gillette et al., 2010, Gillette et al., 2011, Wisdom et al., 2018) could similarly release cell-based structures from a surface without significantly disrupting the organoid structures or underlying cell function (Gillette et al., 2008). Specifically, we deposit the sacrificial material, create the sacrificial structure by cross-linking the alginate in its patterned state, deposit cells on top to allow cellular self-organization to take place, and remove the sacrificial layer by adding a chelator (5% w/v sodium citrate) (Figure 1A; see Figure S1 for fabrication details). The alginate is uncrosslinked within ∼12 min (Figure 1B), to gently release a large number of organoids floating in solution (Figures 1C and 1D; Video S1). The resulting organoid solution could be gently pipetted into tubes, centrifuged and resuspended in a culture medium suitable for downstream manipulation or direct cell delivery. Each step is simple, can be conducted with sterile liquid handling, and can be automated.

Figure 1.

Schematic Diagram of Method of Using Sacrificial Hydrogels to Produce Therapeutic Organoids

(A) Schematic demonstrating the parallels between the surface micromachining method to fabricate MEMS devices such as a microcantilever (top) and the use of sacrificial alginate microwells to fabricate organoids (bottom). Both methods involve the use of a sacrificial layer (blue) to fabricate the final structure (red).

(B) Time required to completely uncrosslink alginate microwells following incubation with different concentrations of a chelator (sodium citrate) by measuring the percent change in mass over time (n = 3, data are represented as mean ± standard deviation).

(C) Schematic diagrams (top) and corresponding experimental images (bottom) showing the steps of organoid fabrication and in vivo perfusion. Experimental data were collected using GFP-labeled HUVECs and RFP-labeled mouse MSCs. First, a co-culture of endothelial cells (green) and mesenchymal stem cells (red) is seeded on dissolvable alginate microwells. Second, after being cultured in maintenance medium without growth factors for 3 to 4 days, cells self-organize into organoids with an endothelial core. A switch into culture medium with vasculogenic growth factors for an additional four days promoted formation of vessels within the organoids. Third, alginate microwells were dissolved with 5% sodium citrate to release organoids. Fourth, suspension of organoids could be centrifuged and assembled into a macro-tissue in vitro to study vascular formation or injected into the subdermis or ischemic hindlimb of a mouse to demonstrate engraftment in vivo. Fifth, injected organoids rapidly connected to form perfused microvasculature in vivo. Scale bar: 100 μm.

(D) The liquid handling steps in the process: (1) seeding the co-culture of ECs (green) and MSCs (red) by pipetting cells onto alginate microwell construct, (2) adding maintenance media once the cells have settled to the bottom of the microwells (approx. 30 minutes), (3) switching to vasculogenic media once an endothelial core has formed, and (4) gently dissolving the alginate microwells (approx. 12 minutes) to harvest organoids (the organoids can be gently washed prior to injection).

The step of cellular self-organization can be adjusted depending on the organoid of interest. For vascularized organoids (Figure 1C), we seeded a co-culture containing ECs and MSCs (of either mouse or human origin) into the dissolvable alginate microwells. We cultured the cells in media without growth factors (“maintenance” medium) to induce cellular self-organization, followed by a vasculogenic medium with growth factors to induce cell-cell interactions including sprouting of blood-vessel-like structures. The organoids, now containing blood-vessel-like structures, are gently released by dissolving the alginate microwells.

The same small number of steps (Figures 1C and 1D) can harvest a large number of organoids by using alginate templates with large numbers of microwells. As an example, we demonstrated three different sizes of alginate microwell inserts for culture dishes (Figure 2A): 15.6-mm diameter inserts containing >1,000 microwells (yielding >24,000 organoids on a 24-well plate), 22.1-mm inserts containing >3,000 microwells (yielding >36,000 organoids on a 12-well plate), and a 60-mm diameter insert containing >30,000 microwells in a 60-mm culture dish. If desired, the inserts can be stacked to increase the number of organoids produced in the same area with additional media changes. Furthermore, the alginate microwells can be stored for up to one month (in the presence of 1.8 mM calcium chloride) without signs of degradation.

Figure 2.

Production of Organoids at Large Scale and Functionality of Organoids to Form Macrotissue

(A) Pictures of three alginate microwells constructs for inserts into 24-well plates, 12-well plates, or 60-mm dishes with the capacity to produce 24 ×1,000; 12 ×3,000; or 30,000 organoids respectively.

(B) Picture of 250 million cells for seeding into alginate microwells. Cells in this figure are GFP-labeled HUVECs and RFP-labeled mouse MSCs.

(C) Stitched brightfield image of cells seeded in a 60-mm construct with 30,000 wells to create 30,000 organoids. Scale bar: 1 cm.

(D) Picture of a 1-mm-thick macrotissues with an area of 1 cm² assembled in vitro by collecting the 30,000 mature prevascularized organoids produced with the alginate microwell (A and B) construct in a 60-mm dish. Scale bar: 1 cm.

(E) Fluorescence images of the macrotissue in (D) with a close-up of the closely packed organoids with endothelial cores (green). Scale bars: 1 mm (left) and 500 μm (right).

We demonstrated this massive parallel production of more than 30,000 organoids by seeding a quarter of a billion cells (Figure 2B) in one 60-mm dish insert (Figure 2C). We also tested the ability of the organoids to assemble in vitro into a macrotissue primed to form a microvascular network (Figure 2D). We co-cultured RFP-labeled MSCs and GFP-labeled ECs for four days, gently harvested the organoids, and assembled them into a macroscopic tissue with surface area of 1 cm² and a height of 1 mm (Figure 2D). We performed fluorescence imaging of this macroscopic tissue (Figure 2E). The organoids were densely packed, and exhibited distinct endothelial core structures, confirming that the gentle harvest and assembly did not disturb the internal architecture of the organoids. The assembled macrotissue, consisting of fully contracted organoids, did not visibly contract during subsequent in vitro culture.

Production of Organoids with Reproducible Size and Structure

Next, we studied whether the sizes and internal architectures of organoids could be controlled reproducibly. In the absence of exogenous growth factors, we observed that GFP-labeled human umbilical vein endothelial cells (HUVECs), which were initially randomly distributed alongside RFP-labeled mouse MSCs, migrate to the center of the organoids and form endothelial cores after culture in the “maintenance” medium for three days (Figure 3A, top panel, and Video S2). Similarly, ECs also formed endothelial cores when co-cultured with another cell type (fibroblasts) in a medium without growth factors (Figure S2), and the endothelial cores were more pronounced than in a previous observation (Wenger et al., 2005). By contrast, ECs did not migrate to the center when the organoids were initially cultured in a vasculogenic medium containing 40 ng/mL VEGF and 40 ng/mL bFGF (Figure 3A, bottom panel), consistent with a previous observation (Dissanayaka et al., 2014). Overall, the data showed the organoids to exhibit reproducible internal architectures containing endothelial cores.

Figure 3.

Production of Vascularized Organoids with High Reproducibility in Size and Structure

(A) Confocal fluorescence images of co-culture organoids of GFP-labeled HUVECs (green) and RFP-labeled mouse MSCs (red) over the first three days in maintenance medium without growth factors (top) or in vasculogenic medium with 40 ng/mL VEGF and 40 ng/mL bFGF (bottom). The cells self-organize by migration and either formed endothelial cores when cultured in media without growth factors (top) or had endothelial cells randomly distributed near the surface of the organoid and did not form endothelial cores when cultured in media with growth factors (bottom).

(B) Overlay of fluorescent and transmitted images showing parallel production of organoids in arrays of different sizes of microwells (with either 100, 200, or 400 μm diameter) and different co-culture ratios (1 EC: 3 MSC, 1 EC: 1 MSC, or 3 EC: 1 MSC). Different sizes of microwells yield different sizes of organoids, either unvascularized with only MSCs or prevascularized with a co-culture of ECs and MSCs, and different co-culture ratios yield different endothelial core sizes. Scale bar: 100 μm.

(C) Quantitative analysis of cell aggregation into organoids and the formation of an endothelial core over time in 200 μm microwells, as measured by the radius of the smallest circle that can contain all MSCs (red) or all ECs (green) (n> 20). Data are represented as mean ± standard deviation.

(D) Barplot showing the size of fully contracted organoids (red) and the size of the endothelial cores (green) for all tested microwell sizes and co-culture ratios. Data are represented as mean ± standard deviation.

(E) Reproducibility of endothelial cores; the number of organoids produced in 1 mm² (dark gray) and the number of organoids containing and endothelial core (light gray) for all tested microwell sizes and co-culture ratios.

See also Figures S2–S6.

Video S2. 3D Formation (Self-Assembly and Self-Sorting) of Organoids over the First 20 Hours, Related to Figure 3

Subsets of data are shown in Figures 3B and S6A

We characterized the reproducibility of the method in controlling the size of the prevascularized organoid. By varying the microwell sizes and the co-culture ratios of cell types (Figure 3B), we controlled the number of cells that could aggregate into a single organoid. For example, microwells of three different sizes (100, 200, and 400 μm diameter) yielded organoids of three different sizes (39 ± 3 μm, 71 ± 5 μm, and 82 ± 7 μm diameter, respectively, all at the same cell-seeding concentration) (Figure 3B). The well size was chosen to be large enough to hold all the cells at the initial seeding concentration but small enough to ensure sufficient cell-cell contact to form a single organoid rather than multiple organoids. The cells aggregated into compact organoids within the first two days of in vitro culture, as seen by the decreasing radius of the smallest circle to include all cells (Figure 3C), with the main contraction happening in the first day and no further contraction after three days. The fully contracted organoids had a regular size distribution (Figure S3). We also observed that the size of the fully contracted organoids (at day 2 and after) correlated to the number of cells in the organoid as expected; the diameter of the organoids' cross-sections related to the number of cells in the organoid and the cells’ typical volume as r_organoid= (6/π·v_cell·n_cell)^1/3/2 (Figure S4).

Also, we quantitatively analyzed the formation of organoids for cultures containing only MSCs and co-cultures with endothelial cell: mesenchymal stem cell (EC:MSC) ratios of 1:3, 1:1, and 3:1 (Figures 3B–3E). In 200-μm microwells, over three days, cells contracted into an organoid and ECs migrated toward the center (Figure 3C), and co-cultures in 400 μm microwells showed similar trends in organoid contraction and EC migration (Figure S5). Co-cultures in 100-μm microwells, however, did not contain enough cells (fewer than 150 cells in total) to form a distinct center (Figure S6). We also observed that the organoids per unit area and the number of organoids containing defined internal architectures could be controlled by varying microwell sizes and ratios of cell types (Figure 3E). (In subsequent in vivo studies, we have used 200-μm microwells with ratios of MSC only, 1 EC:3 MSC, and 1 EC:1 MSC, as these conditions showed aggregation involving almost all the cells within the microwells.) Overall, the data showed the method can produce organoids with internal architectures at high throughput and different sizes controllably.

Production of Prevascularized Human Organoids with Reproducible Size and Structure

We examined the effectiveness of this method for producing prevascularized organoids containing human adipose-derived MSCs (hAMSCs) with GFP-labeled human umbilical vein endothelial cells (HUVECs), in ratios of MSCs only, 1 EC:3 MSC, and 1 EC:1 MSC. We examined the maturation of organoids over eight days, where organoids were first grown in maintenance medium over three days to form endothelial cores, and then switched to vasculogenic medium containing exogenous growth factors for five days (Figure 4A). By day 8, vessel-like structures, such as lumens within the center of the organoid, with sprouting and maturation of vessels toward the surface were observed (especially evident in the larger organoids of the 400-μm wells). The initial migration of ECs was apparent after 20 h (Figures 4B and S7). In addition, we placed multiple prevascularized organoids inside 400-μm alginate wells that were collagen-doped, to mimic the adhesiveness of native tissues. Within 24 h, organoids attached to each other and contracted to form a larger, compact mesotissue (aggregation of multiple organoids) with a smooth outer border (Figure 4C, with additional time points in Figure S8 and Video S3). Hence, this method produced organoids containing human ECs and MSCs, with control over sizes and spatial architectures, and confirming the ability to form a prevascularized mesotissue.

Figure 4.

Production of Vascularized Organoids with Human Cells

(A) Maturation of endothelial cores with dynamic culture conditions for two co-culture ratios; 1 GFP-HUVEC: 3 hAMSC (left) and 1 GFP-HUVEC: 1 hAMSC (right). The cells are seeded (day 0) and initially cultured in maintenance medium without growth factors to form endothelial cores. After three days the organoids were cultured in vasculogenic medium with 40 ng/mL VEGF and 40 ng/mL bFGF and the endothelial cores matured into vessels with discernable lumens (red arrows) and sprouts (white arrows). Scale bar: 200 μm.

(B) Epifluorescence, brightfield, and overlay images showing early self-organization of prevascularized organoids over the first 20 h, with a 1 GFP-HUVEC: 1 hAMSC co-culture in 400 μm microwells. Scale bar: 200 μm.

(C) Epifluorescence, brightfield, and overlay images showing fusion of prevascularized organoids (same conditions as in right A and B) into mesotissues over the first 24 h of the fusion process within a 400-μm collagen-doped alginate microwell. Scale bar: 200 μm.

See also Figures S7 and S8.

Video S3. Fusion of Organoids of Different EC:MSC Ratios Shown Are Organoid Ratios 1:1, 1:3 and MSC Only, Related to Figure 4

Subsets of data are shown in Figures 4C and S8

Rapid Host Perfusion of Prevascularized Organoids in Mouse Model

Next, we assessed the effectiveness of the prevascularized organoids to self-organize to form a vascular network in vivo, anastomose to native host vasculature, and be perfused with host blood in a mouse model (Figure 5A). To facilitate real-time visualization, we performed surgery to place a window chamber (Figure S9) to permit brightfield, epifluorescence, and confocal imaging. We used organoids formed in 200-μm wells yielding organoids approximately 70 μm in diameter, which is also within the diffusion limit of oxygen (Lovett et al., 2009). We produced and harvested prevascularized organoids made of human cells (GFP-labeled HUVECs and hAMSCs), which we injected into SCID mice, a well-established animal model for studying integration of xenografts made of human cells (Steffens et al., 2009). We could inject and monitor the vascular formation for multiple different conditions (e.g., 1 HUVEC: 1 hAMSC and hAMSC only) in the same mouse, by utilizing the strong bond between the fascia and the subdermis. We injected the organoids through the fascia and into the space between the fascia and the subdermis, leaving the subcutaneous tissue intact between injection sites to create a barrier (Figure 5B). The organoids held up intact to the shear stress of injection through a syringe and needle (Figure S10). Interestingly, the shell of MSCs shielded the central blood-vessel building block against shear and preserved the organoids' architectural integrity after they passed through the needle. We also demonstrated the organoids could be injected directly into adipose tissue (Figure S11) and muscle tissue (Figure 6) with good integration.

Figure 5.

Rapid In Vivo Vascularization in Healthy Mice upon Injection of Organoids, as Observed in Real Time via a Window Chamber

(A) Schematic diagram of experimental setup for observing vascular formation and integration with host vasculature in vivo in real time via a window chamber. Organoids (from human cells formed under dynamic culture conditions in 200-μm microwells yielding organoids 71 ± 5 μm in diameter) were injected into a window chamber implant in an SCID mouse.

(B) Real-time in vivo stereoscopic images of prevascularized microtissues with 1 GFP-HUVEC: 1 hAMSC (top row) and unvascularized organoids with hAMSC only (bottom row) through window chamber at different time points. In the top row, newly formed vessels are apparent within 4 days, and blood-filled vessels observed by day 7. In the bottom row, the dashed white line indicates the area of organoids implant and no neo-vascularization was observed. Scale bar: 500μm.

(C) Quantification of neo-vascularization of the prevascularized organoids as the total length of vasculature within three ROIs of 800-by-800 μm containing up to 60 blood-filled vessel branches. The total length of vasculature increases substantially after day 7 for prevascularized organoids. There is no substantial difference in total length of the vasculature for the unvascularized organoids. Note that the blood-filled vessel intersections have been interpreted as branches and not as overpasses (which are two separate vessels, one over the other); however, further work is required to verify this. Data are represented as mean ± standard deviation.

(D) Distributions of branching length in the newly formed microvasculature (B and C) at day 7, 9, and 11. Lines above histogram indicate the mean branch length and standard deviation for day 7, day 9, and day 11 as 93 ± 39 μm, 86 ± 29 μm, and 93 ± 44 μm, respectively.

(E) Real-time in vivo images of prevascularized organoids with endothelial cells in green. The confining pressure of the intact fascia likely causes the spheroids to be organized in a two-dimensional grouping and was imaged at the plane of the endothelial cores. Red arrow heads point to luminous, blood-filled vessels (as indicated by dark lines in fluorescence images and dark areas of brightfield images). Scale bar: 250 μm.

See also Figures S9–S12.

Figure 6.

Rapid In Vivo Restoration of Perfusion and Muscle Fiber Regeneration upon Injection of Organoids in Ischemic Hindlimb

(A) Representative images of blood perfusion in the hindlimbs measured with laser speckle contrast imaging (LSCI) for the two experimental groups. The superficial femoral artery was isolated from the femoral vein and nerve bundle along the length of the thigh, ligated and excised. The mouse then received four injections along the length of the thigh of either 25 μL PBS (control) or organoids (corresponding to 0.5 × 10⁶ cells) injected at each site. The organoids were formed in 200-μm microwells with maintenance media and a 1 mEC: 1 mMSC co-culture ratio yielding vascularized organoids 71 ± 5 μm in diameter with endothelial cores. The perfusion of each limb was measured as the average LSCI intensity of the planar surface of the paw (dashed white outline).

(B) Quantification of perfusion in the hindlimbs as the perfusion ratio (R/L) between the naive left (L) hindlimb and the ischemic right (R) hindlimb (n=3 mice for each condition) with the control mice in blue and the organoid treatment mice in green. Data are represented as mean ± standard deviation. The best-fit (dashed) and 95% confidence interval (dotted) lines are shown (from day 0–9 for organoids, from day 0–14 for control).

(C) Histology of the gastrocnemius muscle on day 14 with an H&E stain. White arrows indicate centralized nuclei of regenerating muscle fibers. Scale bar: 50 μm.

(D) Percentage of myofibers characterized as necrotic because of hyalinization (pink) or as regenerating, viable myofibers assessed by centralized nuclei (purple) (n=3).(4 ROIs of 587 × 440 μm² for each condition). Data are represented as mean ± standard deviation; ∗ indicates significantly more regenerating fibers with p < 0.05.

See also Figures S10 and S13.

We followed the formation of new vasculature by taking epiflourescent and stereoscopic images through the window chamber. Stereoscopic imaging (e.g., of the 1 HUVEC: 1 hAMSC conditions) showed vessel formation between day 4 and 7 (Figure 5B, top rows), with the implanted vasculature connecting to the host vasculature and becoming perfused (Figure 5B, top rows). After just seven days, host perfusion of the implanted vasculature was prominent and intense. The vessels were functional for the remaining 16 days of the 23-day in vivo studies. Quantitatively, we measured the length and number of the branched vessels and the total length of perfused vasculature in three regions of interest (ROIs) within the area of injected organoids (Figures 5C and S12). At day 7, areas injected with prevascularized organoids showed significant formation of new perfused vasculature, whereas areas injected with organoids consisting of MSCs only showed no increase in perfused vasculature (Figure 5B bottom row and Figure 5C). For all four mice tested (each with multiple conditions in the window chamber), all conditions with EC-containing organoids showed rapid vascularization of the injected organoids.

We also explored whether this self-organizing, “micro-to-macro” strategy could provide a limited but reproducible level of architectural control in the overall branching length of implanted, perfused microvasculature. Specifically, we hypothesized that average distances between endothelial cores could be related to diameters of organoids. The mean length of the perfused branches for 1 EC:1 MSC at day 7 was 93 ± 39 μm, with minimal changes by day 9–11, when the mean branch length was 86 ± 29 μm and 93 ± 44 μm, respectively (Figures 5D and S12). Indeed, the length of the newly formed vasculature's branches reflected the core-to-core distances between the densely packed, injected organoids with diameters of 71 ± 5 μm.

We also used epiflourescence and confocal microscopy to characterize the formation and integration of the new vasculature. Observing the GFP-labeled HUVECs through the window chamber (Figure 5E), we noticed the endothelial cores connecting with each other over time: the ECs initially appeared as discrete cores (day 0), then sprouted toward neighboring cores (day 4), connected with the host vasculature and became perfused (day 7), and stabilized as the perfused vascular network matured (day 9, 12, and 23). Between days 4 and 7, the network matured to form lumens (Figure 5E, red arrows). (We further confirmed the luminous structure of the newly formed, perfused network on day 11 via confocal microscopy on day 11; Video S4.) Moreover, we observed that areas indicating newly formed lumens (consisting of GFP-labeled HUVECs) co-localized with areas indicating host blood perfusion, further confirming that it was the newly formed luminous vasculature that was perfused, rather than angiogenesis from the host into the implanted tissue.

Video S4. Spatially Controlled Organoids In Vivo at Day 11, Related to Figure 5

Z-stack video of organoids in window chamber with 9-well insert at day 11 (maximum intensity projection is shown in Figure 1C)

Rapid Re-vascularization and Restoration of Perfusion in a Hindlimb Ischemia Mouse Model

Finally, we assessed the effectiveness of this approach to treat ischemic conditions that are manifested in peripheral artery disease (PAD). We induced hindlimb ischemia by high femoral ligation with complete excision of the superficial femoral artery in C57BL/6 mice (Brenes et al., 2012) (Figure S13). We chose this surgical model because it consistently achieved reduced perfusion in the distal hindlimb but was more reproducible and better represented chronic manifestations of atherosclerotic disease than more severe hindlimb ischemia models, where all side branches were severed but resulted in less reproducible symptoms (Goto et al., 2006). We visualized and measured the perfusion of blood vessels close to the planar surface of the paw using laser speckle contrast imaging (LSCI) (Briers, 2001). Over a defined region, LSCI can image perfusion in microvasculature within 300 μm of the skin surface and provide accurate relative measurements of velocity of blood flow (Briers, 2001). Complete excision of the superficial femoral artery in the right hindlimb induced ischemia and limited perfusion, as confirmed by LSCI (Figure 6A, shown are days 1, 7, 9, and 14 post-ligation). In the saline-treated control group, the hindlimb recovered by approximately days 14–19 (Figures 6A and 6B), consistent with previous results for this femoral ligation model (Brenes et al., 2012).

For the cell-treatment group, we injected organoids into the hindlimb at four sites, taking advantage of the robustness of the organoids to shear stress (Figure S10) and thereby obviating invasive surgery (Mirabella et al., 2017). The organoids contained 2 million cells across the sites, in the range of previous studies (Brenes et al., 2012). Although mice in both the control and organoids groups exhibited a similar level of depressed perfusion in the injured hindlimb two days post-ligation, mice injected with organoids rapidly regained perfusion in the ischemic leg after 7 to 9 days (Figures 6A and 6B), a week quicker than the untreated control group.

Histological analysis of the gastrocnemius muscle from mice confirmed the ischemic limb of the control group displayed only few regenerating myofibers (Figure 6C), at levels indistinguishable from that of naive limbs (Figure 6D). Also, they showed signs of tissue necrosis, as indicated by a fragmented collection of short hypereosinophilic and swollen pale eosinophilic myofibers (Figure 6C). By contrast, histological cross-sections from mice injected with organoids showed centralized nuclei characteristic of regenerating myofibers (Figure 6C), with significantly more regenerating fibers than those of naive hindlimbs and ischemic limb of control group (Figure 6D). Moreover, cross-striations, which are characteristic of viable myofibers, were also more apparent in the hindlimbs of mice treated with organoids compared with control group (Figure 6D). The mice treated with injections of prevascularized organoids regained perfusion of the ischemic limb in just nine days, had more viable myofibers, and exhibited significantly more regenerating myofibers.

Discussion

Using Sacrificial Hydrogels to Produce Organoids with High Reproducibility and Scalability

Like the development of micromachining techniques for producing MEMS structures reproducibly and on a large scale, we have developed a technique to use sacrificial hydrogels to produce clusters of self-organized cell-based structures with high reproducibility and scalability. Previously, we and other groups have shown the use of microfabricated hydrogels, including sacrificial techniques, to form in vitro microvascular networks (Gillette et al., 2008, Gillette et al., 2010, Miller et al., 2012, Bertassoni et al., 2014). This paper demonstrates that the dynamic structure of hydrogels can also be exploited to produce and gently release organoids for cell therapy.

For purposes of cell therapy, it is critical for clinical efficacy, process control, and regulatory approval that cells introduced into the body are generated via tightly controlled processes and exhibit reproducible origin, size, and structure. Previous studies have observed that a “lack of control over the process is likely to underpin the variability in systems and experiments that, with few exceptions, does not allow [organoids] to yield their full potential,” and the importance of achieving reproducible “organoid size, shape, cellular composition, and 3D architecture” in future research on organoids as well as use for therapeutic purposes (Huch et al., 2017). Compared with current organoid systems, our method can generate self-organized multicellular aggregates with both high yield (Table S1) and high reproducibility over aggregate size and cellular organization (Table S2). Moreover, the aggregate size and features of cellular organization can be tuned (Table S2), as our method bears similarities to MEMS fabrication technologies (in contrast to “cells in gels” systems that feature a distribution of pore sizes). In this study, sizes and internal architectures of the organoids were reproducible for different types of cells (MSCs and ECs of mouse and human origin), cell ratios, and overall size of microwells that determined the diameter of the contracted organoids. Even at the tissue level in vivo, branching lengths of the vascular network were reproducible (by contrast, microtissues with ECs had previously yielded non-uniform branching lengths (Ehsan et al., 2014; Rouwkema et al., 2006, Walser et al., 2013)).

Also, an ideal method for generating organoids should be scalable and gentle. In the common hanging-drop method, 384 organoids could be produced in the area of an overall standard well plate (with the overall scalability limited by the number of wells (Tung et al., 2011)), whereas the smallest construct shown in Figure 2 produces 24,000 organoids in the same area with fewer steps needed (e.g., media-changing steps, one alginate dissolving step), all of which could be automated by liquid handling. The release of organoids is gentle even at a large scale, in contrast to vigorous pipetting or high-speed centrifugation for current microwell procedures. For cell therapy, it is important that the integrity of the cells be preserved (e.g., an FDA guidance document points to the need “to preserve integrity and function so that the products will work as they are intended” (Administration, 2017)). Beyond cell therapy, large-scale and effective production of organoids (beyond the quarter billion cells demonstrated) could also support studies in developmental biology, cancer cell intravasation (Ehsan et al., 2014), and organ printing.

An Advanced, Controlled Form of Cell Therapy for Disease Such as Ischemic Conditions

To date, more than 50 cell-therapy trials are at clinical stages for treating CLI. Many trials involve injecting MSCs (Davies, 2012, Tongers et al., 2008, Raval and Losordo, 2013, Lawall et al., 2011, Chen et al., 2008) or ECs (such as MarrowStim) (Lawall et al., 2011, Botham et al., 2013, Benoit et al., 2013), but the cells could die from deprivation of oxygen and nutrients before they are able to assemble into vascular networks in vivo and anastomose with host vasculature. In this study, after injection of organoids, we observed rapid revascularization and reperfusion (within 4–7 days in a mouse hindlimb ischemia model, compared with several weeks typical of implanted materials or tissues (Kang et al., 2016, Lokmic et al., 2007, White et al., 2014, Walser et al., 2013)), as well as low muscle necrosis. In a clinical scenario, such an approach could be especially attractive for “no-option” patients on the verge of amputation with subsequently poor mortality outcomes (60% within five years of surgery (Davies, 2012)).

In past studies, needle injection (and organ printing) with unilaminar vascular organoids (Fleming et al., 2010, Mironov et al., 2009) had been challenging due to shear stress formation. It would be advantageous in cell therapy to be able to deliver the cells via minimally invasive injection rather than invasive surgery. Our method produced organoids that held up intact to shear stress during injections, even with high-gauge (25–30) needles (Figure S10). This behavior may partially have been due to the shell of MSCs that protected the endothelial structure; interestingly, previous studies have also shown that the MSCs could act as an immune-suppressive shield for cell therapy in addition to providing angiogenic signaling (Huang et al., 2013, Iwase et al., 2005).

Limitations of the Study

Although we have demonstrated functionality of vascularized organoids composed of ECs and MSCs in vivo (even after minimally invasive injection through a needle), further in vitro work will be required to fully characterize and quantify the capillary-like vascular sprouts within the organoids. Not all combinations of ECs and other cells (including MSCs) may result in the endothelial core structures observed here, although the results of our method suggest cell-cell structures obtained will be reproducible. Moreover, before the vascularized organoids can be tested clinically, more work will have to be done to tailor dosages and study long-term preclinical outcomes compared with current cell therapy of suspensions of single cells. It will also be interesting to explore this technique as a platform for delivery of other types of stem and somatic cells, beyond those studied here.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: May 22, 2020