A Modular Microscale Granuloma Model for Immune-Microenvironment Signaling Studies in vitro

Tuberculosis (TB) is one of the most potent infectious diseases in the world, causing more deaths than any other single infectious agent. TB infection is caused by inhalation of Mycobacterium tuberculosis (Mtb) and subsequent phagocytosis and migration into the lung tissue by innate immune cells (e.g., alveolar macrophages, neutrophils, and dendritic cells), resulting in the formation of a fused mass of immune cells known as the granuloma. Considered the pathological hallmark of TB, the granuloma is a complex microenvironment that is crucial for pathogen containment as well as pathogen survival. Disruption of the delicate granuloma microenvironment via numerous stimuli, such as variations in cytokine secretions, nutrient availability, and the makeup of immune cell population, can lead to an active infection. Herein, we present a novel in vitro model to examine the soluble factor signaling between a mycobacterial infection and its surrounding environment. Adapting a newly developed suspended microfluidic platform, known as Stacks, we established a modular microscale infection model containing human immune cells and a model mycobacterial strain that can easily integrate with different microenvironmental cues through simple spatial and temporal “stacking” of each module of the platform. We validate the establishment of suspended microscale (4 μL) infection cultures that secrete increased levels of proinflammatory factors IL-6, VEGF, and TNFα upon infection and form 3D aggregates (granuloma model) encapsulating the mycobacteria. As a proof of concept to demonstrate the capability of our platform to examine soluble factor signaling, we cocultured an in vitro angiogenesis model with the granuloma model and quantified morphology changes in endothelial structures as a result of culture conditions (P < 0.05 when comparing infected vs. uninfected coculture systems). We envision our modular in vitro granuloma model can be further expanded and adapted for studies focusing on the complex interplay between granulomatous structures and their surrounding microenvironment, as well as a complementary tool to augment in vivo signaling and mechanistic studies.

and M. bovis BCG (mCherry) form these aggregates within a 3D collagen plug. Images were obtained using standard fluorescence microscopy. CellTracker Green staining of the MDMs enabled visualization of these aggregates with live cells (no fixing), as CellTracker Green stains live cells and doesn't require additional staining steps. Scale bars: 100 μm.  N/A  3  3  Experiment 2  1  2  2  3  3  Experiment 3  1  2  2  3  3  Experiment 4  1  2  2  3 3 1: no aggregate formation observed in Stacks infection culture wells 2: aggregate formation observed in minority of wells/progressive formation of aggregates 3: aggregate formation observed in majority of wells in device N/A: no observations noted in lab notebook p.i.: post infection At Day 1 p.i., no aggregates were observed within the infected (+BCG) culture wells within the Stacks device. Starting Day 2 p.i., aggregate formation was observed in a minority of the wells, with progressively more aggregates forming on Day 3 p.i. By Day 4 p.i., aggregates were observed in the majority of wells. Due to the microscale volume of media used to feed each well (8 μL) and the open characteristics of the platform, changes in humidity and temperature can have significant effects on evaporation and culture temperature; therefore, wells were not imaged daily so as to minimize the amount of time the cultures were out of the humidified cell culture incubator; qualitative observations are provided here to illustrate an overview of the aggregate formation timeline for future users. Figure 4: Model granuloma layers secrete proinflammatory cytokines following infection (Day 1 p.i.). Individual cytokine profiles from each independent experiment across all days are included to demonstrate the model's signaling ability over time across replicates. P < 0.05; ratio paired t-test; "n/a" indicates that the ratio paired t-test could not be run as the sample contained values equal to zero (signal equal to or below blank standards).These data are the expanded results from the average data shown in Figure 3.

Supplementary Information 1: Polypropylene (PP) Device Flattening Protocol
Protocol is for use with a Carver Bench Top Standard Heated Press (Model #4386). This protocol can be adapted for use with alternative heated presses with model-specific changes.
1. Preheat platens to 110°C (this takes ≈ 20 min to heat up and stabilize, as the temperature fluctuates). 2. Prepare a stack of devices (4-5 devices) and carefully align the stack to avoid deformation. 3. Place Kapton® polyimide film (#2271K2, McMaster Carr) on the top and bottom platens and place the stack of devices between the films (to avoid direct contact between the devices and the platens). 4. Turn on the pressure sensor. 5. Pump the hot press handle until contact is just made and the devices are flattened.
-The devices stacks should not be compressed at all (i.e., the pressure sensor should read "0 psi" or fluctuate between "0-10" psi), but the devices should be in contact with the platens and flattened. 6. Let the stacks flatten for 60 min at the above temperature and pressure. 7. After 60 min, turn off the platen temperature and let the temperature drop down to ≈30°C.
-Do not use coolant to lower the temperature as this causes the temperature to drop too rapidly and can cause deformation. This cooling process takes 1.5-2 h. 8. Once the temperature has dropped below 30°C, turn on the coolant circulator and run the coolant for 5 min to completely cool the platens. 9. Turn off the coolant, release the pressure, and remove the devices from the platens. Optimization of the Tween-80 concentration for culture of M. bovis BCG was required for compatibility with the Stacks platform. Traditionally, mycobacteria are cultured in varying concentrations of surfactant (commonly Tween-80) to prevent clumping of mycobacteria during culture. However, the presence of surfactant within the Stacks platform interferes with the capillary pinning necessary for maintenance of cell cultures and stacking of multiple layers. This is due to the decrease in interfacial tension between the culture media and the device surface caused by the presence of surfactant, effectively decreasing the contact angle of the media on the surface. In order to alleviate the effects of surfactant on the capillary pinning, the concentration of surfactant must be well below the critical micelle concentration (CMC), which is 0.0013% w/v 5 , in the final media. Therefore, we cultured the BCG at a concentration of 0.003% w/v Tween-80, which yields a final concentration (after all dilutions and mixing) of 0.0000025% w/v Tween-80 in the collagen plug; this concentration was selected as it was the highest concentration of Tween-80 we could use without loss of capillary pinning on the surface. However, to decrease the aggregation of BCG within our model system, the BCG were vortexed, vigorously pipetted, passed through a 27G needle to disperse aggregates, and then allowed to settle for ≈ 1 min before aliquoting from the top of the culture for use.

Supplementary Information 3: Culture Optimization of Endothelial Layer
To miniaturize the in vitro angiogenesis model [6][7][8] we adapted for this layer and to ensure functionality, we optimized certain components of the culture system within the Stacks platform. The first condition optimized was the seeding density; too high of a seeding density results in formation of a confluent monolayer or islands of cells on the surface of the Matrigel, whereas too low of a seeding density results in a dispersed culture that does not form endothelial connections. Therefore, the seeding density was calculated from previously established protocols using 48 well plates 7 to obtain a similar ratio of cells to area in the Stacks well (1,650 cells/well of the Stacks layer). We observed that concentrations greater or less than 1,650 cells/well resulted in inconsistent tubule or network formation.
Additionally, the volume of Matrigel within each well was optimized. The final volume of 3 μL was selected to allow complete and reproducible coating of the bottom of each well and to limit the effect of the meniscus on cell localization (i.e., cells aggregating towards the middle of the well/bottom of the meniscus) and cell visualization. Lower volumes of Matrigel (< 3 μL) resulted in uneven coating of the surface of the well floor, causing a non-uniform surface wherein cells adhered to both Matrigel and exposed polystyrene on the floor of the well, leading to altered morphology. Higher volumes of Matrigel (> 3 μL) caused an exaggerated meniscus effect, resulting in cell aggregation in the center of the well. In cases where the volume of the well was completely filled with Matrigel (> 3.6 μL), we observed cell growth both on the Matrigel and on the polystyrene pinnng ridge surrounding the well, resulting in varying surfacedependent morphologies (i.e., cells displayed different morphologies on the polystyrene and the Matrigel).  ImageJ Macro Image Preparation and Analysis.txt The file for the injection molded device was reproduced from Yu et al. 1