An Artificial Leukemic Niche for in vitro Potency Testing of Genetically Engineered Living Drugs

Abstract

3D cell culture has been proven to be superior to 2D culture approaches in numerous applications. Some of the advantages of threedimensional culture systems comprise the ability of the cells to form cell-cell-contacts, to establish a cellular polarity, and the formation of oxygen and nutrient gradients, among others. Moreover, co-cultures in 3D are much more physiologically relevant compared to their two-dimensional counterparts. With this in mind, we developed an in vitro-model of the leukemic stem cell niche as a test platform for third generation α-CD19 chimeric antigen receptor-T-cell (CAR-T) potency with the CAR-T-cells being one member of the so-called class of living drugs. This is of clinical relevance since leukemia initiating cells (LIC) in the bone marrow hide from circulating immune cells and thus are made responsible for the primary disease and patient relapses. The in vitro-model is based on microcavity arrays that have proven to support a plethora of cells with regard to growth, differentiation, and maintenance of stem cell characteristics. In this article, we characterize the third generation CAR-T-cell response to a CD19 positive leukemic cell line (NALM-6) in co-culture with mesenchymal stromal cells as a leukemic niche surrogate. For this, confocal microscopy data were analyzed with respect to leukemic cell count prior and after CAR-T-cell infusion into a bioreactor system as a potency assessment, distribution of leukemic cells within the niche, cellular mobility, Caspase 3/7-, and Interferon ϒ-release. We could show that this 3D model of the leukemic bone marrow niche might be of use to assess CAR-T-cell potency in a bone-marrow-like environment and thus adds additional information to classical assays such as the 51Cr-release assay of suspension cells.

Keywords: Microcavity Arrays; Leukemic Niche; Living Drugs; Potency Test

Abbreviations: CAR-T cell: Chimeric Antigen Receptor T-cell; PBS+/+: Phosphate Buffered Saline solution with Calcium and Magnesium; PBS-/-: Phosphate Buffered Saline without Calcium and Magnesium; pMSC: primary Mesenchymal Stromal Cells; LIC: Leukemia Initiating Cells.

Introduction

When Claude Bernard compiled his experimental work in 1878 in his book “La Science Experimental” [1], he likely laid the foundation for modern cell culture. The significance of his discovery — that physiological systems of an organism can be maintained even after its death — was probably not fully realized at the time. Although Carl Friedrich Wilhelm Ludwig had already recognized this in 1856, Bernard’s findings eventually led to in vitro biology. His descriptions marked the beginning of a development that today aims not only at keeping entire organs alive but also at growing them entirely three-dimensionally in vitro. However, 2D cell culture is still very popular since it is easy to learn, large cell numbers can be generated with little effort, and the consumables are inexpensive which is why 3D approaches only in the last 10 years became more and more popular. Because only a minute number of cell types in vivo grow in monolayers, such as endothelial cells, organotypic behaviour can only be achieved in 3D and in co-cultures of different cell types. This approach has ultimately led to organoids, introduced by Sasai [2] and Clevers [3] as a fundamental tool in many research fields. Organoids can be defined as complex 3D structures that display architectures and functionalities similar to in vivo organs and that develop from stem cells or organspecific progenitors through a self-organization process [4]. Although this 3D culture type is the state of the art approach for organotypic function and morphology, less complex structures such as co-culture spheroids comprising 2 or 3 cell types remain valid models and borders between them and organoids may be fluid. We have previously shown that such co-cultures of only 2 cell types, primary mesenchymal stromal cells and CD34+ hematopoietic progenitor cells, used as an artificial hematopoietic stem cell niche in a microcavity array-based bioreactor, can maintain the stemness of CD34+ cells already to a certain extent [5]. In this paper, we used the work of the artificial hematopoietic stem cell niche as a basis for an artificial leukemic niche. Therefore, the CD34+ hematopoietic progenitor cells from cord blood were exchanged against a CD19-positive acute lymphocytic cell line called NALM-6. Due to the biological diversity in primary mesenchymal stromal cells and to facilitate standardization, we also exchanged the primary mesenchymal stromal cells against the mesenchymal line HS-5. This bone marrow stromal cell line still expresses relevant stroma markers and supports proliferation of hematopoietic progenitor cells when co-cultured in serum-deprived media with no exogenous factors, making them a suitable cell type for a leukemic niche as well [6]. This microcavity arraybased artificial leukemic niche may be of use in potency testing of living drugs, such as chimeric antigen receptor T-cells (CAR-T) or T-cell receptor engineered T-cells (TCR) since it delivers additional information and circumvents some of the disadvantages of classical assays. For example, the 51Cr release assay [7] is read out 4 h after effector cell addition and is of limited use for longer analysis times due to the fact that spontaneously released 51Cr into the medium increases, thereby decreasing the signalto-noise ratio of the assay, not to mention the need for the prerequisites of handling radioactive materials [8]. For the BLI luciferase assay, target cells have to be transduced with a luciferase reporter gene so that only live target cells deliver a signal. In case effector cells are active, target cells are killed, leading to a decrease in the signal [8]. Although this assay is a non-radioactive one, it harbors some disadvantages such as that target cells have to be transduced prior to the assay, which might not be possible for all target cells. Moreover, the used firefly luciferase, with its molecular weight of 62 kDa, might not efficiently leak into the medium upon effector-cell mediated pore formation in the target cell [9]. The only assay for adherent cells, the impedance-based assay, uses 2D cultured cells on top of electrodes incorporated into a well plate. Effector cells decrease the impedance upon lysis of adherent target cells. However, only 2D cultured target cells are measured, delivering no information about the potency in 3D. Although flow assays allow for analysis and quantification of the cytotoxicity of subpopulations in heterogeneous cell mixtures, this type of assay is laborious [8] and requires dedicated instruments. The abovementioned test systems do not take into account that leukemia initiating cells (LIC) hide in a three-dimensional environment - the bone marrow. Here, they are protected from immune cells and are made responsible for the primary disease as well as for patient relapses. This is why we tried to create a 3D test system for living drugs based on microscopy data that can be analysed with open source software and as well as being able to deliver data of released cellular constituents that can easily be measured by e.g. ELISA and that recapitulates the three-dimensionality of the bone marrow microenvironment.

MATERIALS AND METHOD Cell Culture

Primary Mesenchymal Stromal Cells: Primary mesenchymal stromal cells (pMSC) were kindly provided by Patrick Wuchter (Medical Faculty Mannheim of the Heidelberg University, Heidelberg, Germany) and were isolated from healthy voluntary donors after obtaining informed written consent according to the guidelines approved by the Ethics Committee of the Medical Faculty of Heidelberg University. The culture protocol is described in detail in [10]. Briefly, bone marrow aspirates (10 - 30 ml) were collected in a syringe containing 10,000 IU heparin to prevent coagulation. The MNC fraction was isolated by density gradient centrifugation on Ficoll-Hypaque (d=1.077 g/cm3 ; Biochrom, Germany) and seeded in tissue culture flasks at a density of 1x106 cells/cm2 (Nunc®- flasks with 75 cm2 area; Nalge Nunc, Naperville, IL, USA) for 2 days. The pMSCs were expanded in T75 culture flasks using commercially available Poietics Human Mesenchymal Stem Cell Basal Medium (PT-3001; LONZA, Walkersville, MD, USA) following the manufacturer’s instructions. For this step, 5,000 cells/cm2 were plated in tissue culture flasks without any precoating. The culture medium was changed twice per week. After reaching 80% confluence, the pMSCs were trypsinized, counted with a Neubauer counting chamber (Brand, Wertheim, Germany) and re-seeded at 104 cells/cm2 for further expansion.

Mesenchymal Stromal Cell line HS-5: HS-5 cells were purchased from ATCC (CRL-3611, Manassas, USA), adapted to 2.5% human platelet lysate (ELAREM Perform, PL BioScience, Aachen, Germany), cultivated in T75 cell culture flasks, and passaged once a week. For this, the culture medium was aspirated and the cells were washed with 2 ml PBS -/- to remove dead cells, metabolic products, and residual culture medium. In the next step, 2-3 ml of 0.25% (w/v) trypsin - EDTA was added to the flask and the cells were incubated for 5 min. Thereafter, 8 ml of culture medium was added to stop the enzyme reaction. The suspension was removed by careful pipetting and transferred to a 15 ml Falcon tube. Then, the cell suspension was centrifuged at 300 g for 5 min, the supernatant was aspirated, and the cell pellet was resuspended in culture medium to seed the desired number of cells in 12 ml culture medium.

NALM-6 Cells: We used the NALM-6 cell line (ATCC, Manassas, USA) as target cells since this Acute Lymphocytic Leukemia (ALL) line expresses CD19 and thus can be detected by the used α-CD19-CAR-T-cells. NALM6 cells were initially cultivated in IMD medium supplemented with 10% heat-inactivated FBS, 1% Sodium-Pyruvate, 1% non-essential amino acids (NEAA), 1.5% GlutaMax, and 0.5% Pen/Strep in T25 cell culture flasks for suspension cells and passaged every two to three days. The cells were then adapted to 2.5% human platelet lysate for better adaption to the co-culture with HS-5 cells. For passaging, 4 ml of pre-heated culture medium (37°C) was placed in a new flask and 1 ml of cell suspension was carried over into the new flask. The culture medium was supplemented with 2.5% human platelet lysate and heparin.

NALM-6-GFP Cells: NALM-6-GFP cells were purchased from Imanis Life Sciences (CL138, Rochester, Minnesota, USA) and generated from the parental NALM-6 cells (CRL-3273) by transduction with a lentivirus construct containing eGFP (LV-eGFP-PGK-Puro). The cells were cultivated in IMD medium supplemented with 10% heat-inactivated FBS, 1% Sodium-Pyruvate, 1% non-essential amino acids (NEAA), 1.5% GlutaMax, and 0.5% Pen/Strep in T25 cell culture flasks for suspension cells and passaged every two to three days.

HL-60 Cells: HL-60 cells (ATCC, Manassas, USA) were cultivated in RPMI 1640 medium, supplemented with 15% heat-inactivated FBS, 1.5% GlutaMax, 0.5% Pen/Strep, 1% Sodium-Pyruvate, and 1% non-essential amino acids (NEAA). The cells required splitting before exceeding a concentration of 106 cells/ml. For splitting, 0.5 ml of the cell suspension was transferred in a new flask containing 9.5 ml fresh medium. The cells were then adapted to 2.5% human platelet lysate for better adaption to the bioreactor co-culture.

CAR-T-Cells: Third generation CAR-T-cells were kindly provided by Michael Schmitt (Department of Internal Medicine V, University of Heidelberg, Heidelberg, Gemany) either as a fresh, cultured batch or frozen. Briefly, patient-derived T cells were transduced with a retroviral third generation CAR vector (RV-SFG.CD19.CD28.4-1BBzeta) comprising CD137 (4-1BB) and CD28 as costimulatory domains [11]. The cells were cultivated in a mixture of RPMI (45%) and Click’s medium (45%) supplemented with FBS (10%) and glutamine. CAR-T-cells were used in an effector:target (CAR-T: NALM-6) ratio of 15:1 after initial effector:targetratio tests.

Microcavity Arrays: Microcavity array manufacturing has extensively been described earlier [12, 13]. Basically, 50 µm thin heavy ion irradiated polycarbonate films (it4ip, Louvain-la-Neuve, Belgium) were heated to the glass transition temperature of 154°C with the help of a custommade microthermoforming machine. After reaching the glass transition temperature, nitrogen gas was used to press the polymer film into the microcavity array mold. Immediately after reaching the forming pressure of 20 bars, the setup was cooled to 40°C, after which the microcavity array was demolded. Next, pore formation was realized by etching the polymer film with a 2 M sodium hydroxide solution for 2 hours. The pore size after this etching time was 3 µm, preventing cells from leaving the microcavities but insuring medium supply from below. The molded, etched, and quality-controlled microcavity arrays were cut to the size of 20 mm x 20 mm using a stencil and a scalpel. In order to sterilize and hydrophilize the microcavity array, a descending isopropanol series was performed. For this purpose, 4 Petri dishes were filled with 100%, 70%, 50% or 30% isopropanol, respectively, and 2 Petri dishes with sterile water. It was sufficient to pull the microcavity array for 5-10 seconds in the direction of descending isopropanol concentration so that the surface of the array was completely wetted. The third step is the functionalization of the surface with collagen I to achieve an improved adhesion for adherently growing cells. Rat tail collagen I was dissolved in 0.2% acetic acid and diluted to 2 mg/ml. 18 µl of this dilution was mixed with 132 µl sterile water in an Eppendorf tube, resulting in a concentration of about 10 µg/cm2 collagen on top of the microcavity array. The array was placed in a new Petri dish with the opening of the cavities facing upwards and 150 µl of the collagen-containing working solution was pipetted onto the microcavities. The array was incubated at 4°C over night. Afterwards, the collagen solution was removed and the array washed twice with PBS+/+.

Inoculation of Microcavity Arrays: A cell count of 200,000 NALM6 and 500,000 - 1,000,000 HS-5 cells or 200,000 of the larger pMSC were used to inoculate a microcavity array. The cell suspensions were combined and gently mixed in a 0.5 ml Eppendorf tube so that in 150 µl medium both cell types were present in the desired number. The collagenized microcavity array was taken out of the PBS+/+ solution, any remaining droplets were carefully aspirated from the edge, and the array was placed in a fresh Petri dish. In the next step, the 150 µl suspension containing the NALM-6 and pMSC/HS-5 cell mixture was added on top of the collagenized cavities using an Eppendorf pipette. The array was then placed in the incubator for 2 h using a culture medium consisting of a mixture of NALM-6 and pMSC/HS-5 cell culture medium in a 1:1 ratio.

Bioreactor Setup: The bioreactor is a custom-made microbioreactor for microcavity arrays that has been extensively described elsewhere [10-15]. 3D bioreactor culture was performed in microcavity arrays which has also already been described extensively elsewhere [10-16]. In essence, the bioreactor is the size of a 5 cm petri dish and can house up to 2 microcavity arrays. However, all experiments have been performed with only one microcavity array inside. A superfusion setup was chosen for the medium flow due to the fact that the bioreactor was modified for CAR-T-cell infusion into the bioreactor directly on top of the microcavity array (Figure 1). Besides the bioreactor, the complete setup also comprised a cassette pump and a medium reservoir. The inoculated microcavity array was inserted into the bioreactor as depicted in figure 1.

Figure 1: Left: Bioreactor body setup, right: Microcavity array details.

For this, the bioreactor compartment was carefully opened and flipped over, the lower lid was carefully lifted off and the array was inserted into the lower compartment. Then, the reactor was closed and the bioreactor medium flow established. With the intention to inject cells and medium during the experiment, it was important not to fill up the central bioreactor compartment completely. At the same time, it was ensured that all cells were sufficiently covered with medium. The bioreactor was driven in superfusion mode, meaning that the medium flow was parallel to the microcavity array surface. A flow rate of 400 μl/min was established to provide a sufficient nutrient and gas exchange. The seal of the Falcon tube containing the culture medium was opened in the incubator to ensure a sufficient gas exchange. A designed holder allowed the cap to be lifted easily and at the same time preventing it from falling off. 24 h prior to the experiment (-24 h), the microcavity arrays were inoculated and after 2 h of incubation, mounted into the bioreactor system. Afterwards, the CAR-T cells were injected directly into the injection chamber above the microcavity array via the recessed cannula (0 h) (Figure 2).

Figure 2: Modified bioreactor for CAR-T-cell infusion.

To ensure a quantitative CAR-T-cell infusion, the stained CAR-T-cells were resuspended in 1 ml of bioreactor medium. The cell suspension was drawn up in a 1 ml syringe and stored vertically in the incubator. This is to allow time for the CAR-T-cells to sediment and pool in the syringe outlet before use. Upon infusion, the pump was turned off to prevent the CAR-T-cells from being flushed out before reaching the microcavity array surface. Thereafter, the cultivation was performed for another two days. Control experiments were carried out using HL-60 cells as non-effector cells instead of the CAR-T-cells.

Experimental Protocol: The experimental protocol was designed such that the inoculated microcavity array was subjected a pre-culture period in the bioreactor for 24 h. Afterwards, CAR-T- or control cells were infused. Image stacks and samples were taken or drawn, respectively at -24 h, 0 h, 4 h, 24 h, 48 h, and 72 h (Figure 3).

Figure 3: Experimental protocol. Microscope, syringe, and plate reader illustrations from NIAID NIH BIOART Source (bioart.niaid.nih.gov/ bioart/000505 & 000086 & 000161) [47].

Cell Staining: Three different fluorescent dyes were used to monitor living cells with regard to their movement, spatial arrangement, and proliferation behavior: CellTrackerTM Green CMFDA (C2925, Life Technologies, Darmstadt, Germany), CellTrackerTM Blue CMAC (C2110, Life Technologies, Darmstadt, Germany), and CellTrackerTM Red CMTPX (C34552, Life Technologies, Darmstadt, Germany). Cell TrackerTM Green was used for the ALL-line NALM-6, CellTrackerTM Blue was used to stain the mesenchymal stromal cell line HS-5 or pMSC, respectively, and CellTrackerTM Red was used to stain CAR-T-cells. Staining was performed according to the manufacturer’s protocol. Modified Cell Titer Glo®-Assay: The CellTiter-Glo® 2.0 Cell Viability Assay (Promega, Walldorf, Germany) determines the number of cells in a culture by converting the released ATP into a luminescence signal via an enzymatic reaction cascade. In this reaction, ATP from lysed cells fuels the Ultra-Glo recombinant luciferin reaction to oxyluciferin creating a luminescence signal. The detected luminescence is proportional to the cell count. However, we used the chemicals of the kit for the realtime detection of the luminescence signal in the bioreactor circulation, indicating CAR-T-cell mediated target cell killing. For this, 1 ml of the CellTiter-Glo® substrate was injected into the bioreactor circulation prior and after the infusion of CAR-T-cells. The signal was recorded via a confocal laser scanning microscope (Leica TCS SP5, Leica, Mannheim, Germany) without active lasers and using the high-definition photomultiplier in the wavelength range of 490 to 700 nm as a detector.

Caspase 3/7-Assay: To be able to correlate the microscopic and Cell Titer Glo®-assay data with another parameter, we quantified the activity of released Caspase 3/7 from killed NALM-6 cells. The Caspase Glo-® 3/7 assay (G8090, Promega, Walldorf, Germany) was used according to the manufacturer’s protocol. For the assay, inoculated microcavity arrays were placed in Petri dishes with medium. At indicated time points, 100 µl medium samples from directly above the microcavity array and from the margin of the Petri dishes were drawn. Caspase 3/7-activity was then determined by transfer of the sample volume to an MTP that was measured with the help of a luminometer every 20 min for 2 h.

nterferon-ϒ-Assay: Interferon-ϒ (IFN-ϒ) release is an indicator of CAR-T-cell activation. Therefore, we quantified the IFN-ϒ release into the bioreactor circulation. The human IFN-ϒ ELISA kit (ThermoFisher Scientific, Darmstadt, Germany) was used according to the manufacturer’s protocol. Briefly, 50 µl of biotinylated antibody was pipetted to each well. Afterwards, samples and standards were pipetted to the wells and incubated at room temperature for 2 h. The plate was washed three times, the Streptavidin-HRP-solution was added and incubated for 30 min at room temperature. The plate was washed again three times, 100 µl of TMB solution was added to each well and incubated for 30 min in the dark at room temperature. At the end, 100 µl of stop solution was added and the absorbance measured at 450 nm with background subtraction at 550 nm. CAR-T-cell counts: For each experiment, the NALM-6 cells of 3 microcavities for each time point were counted. “Total cell numbers” refers to NALM-6 cell numbers assuming that the cells are equally distributed over the entire microcavity array with 634 microcavities and therefore, the number of detected NALM-6 cells in 3 microcavities has been multiplied by factor of 211. CAR-T-Cell Potency

Calculation: For the quantification of the CAR-T-cell potency based on NALM-6 cell counts and considering also the growth 

R-T Number of injected CAR-T-cells Software: For 3D reconstruction, single cell detection in microcavities of confocal image series, counting, and single cell tracking, the Fiji software [17] was used. The Fiji plugin “3D objects counter” counted the cells, the plugin “TrackMate” tracked the cells. The three-dimensional distribution of NALM-6 cells in microcavities was reconstructed via MATLAB 2021a. The whole protocol for 3D reconstruction, detection, counting, and tracking of NALM-6 and CAR-T-cells is made available upon request.

Statistics: All experiments were carried out at least with n = 3, unless otherwise stated. Error bars represent the standard error of the mean. The data were subjected a Mann-Whitney-U-Test to compare the arithmetic means of two samples with independently measured data. The significance level was set to 5% and consequently, all results with p values ≤ 0.05 were considered as being statistically significantly different, marked with a*.

RESULTS

MSC and NALM-6 bioreactor co-culture appearance

First, we imaged the microcavities at wavelengths for the used dyes and merged the fluorescence images with the brightfield ones to get an impression of the bioreactor co-culture comprising MSCs and NALM-6 cells (Figure 4).

Figure 4: Appearance of the pMSC/NALM-6-bioreactor co-culture at -24 h, 0 h, and 72 h showing the NALM-6 only (green), the MSC only (blue), merged image (green/blue), and merged image with overlay of the brightfield image.

As can be seen, the NALM-6 cells distribute evenly within the cellular pMSC network and proliferate. pMSC do not seem to proliferate as much as the NALM-6 cells but this can be attributed to the longer doubling time (~50 h compared to ~24 h, respectively). Pilot Experiment.

Pilot Experiment with CAR-T-Cell Infusion

We next checked whether the CAR-T-cells can successfully be infused into the running bioreactor setup and kill the target cells. After the 24 h pre-culture period (Figure 3), 3 Mio. CAR-T-cells were infused into the bioreactor which was then imaged by recording z-stacks with a confocal microscope after 0 h (Figure 5), 15 min, 24 h, 48 h, and 72 h (Figure 6).

Figure 5: Appearance of the NALM-6/pMSC after 24 h of co-culture. Left part: inclined sight on top of the microcavities and cross sections of the left side images for the highlighted areas. Right: reconstructed positions of the NALM-6 cells in 3 representative microcavities using a MATLAB script.

Figure 6: Microscopy images of the co-culture. Top view and three-dimensional representation of the NALM-6 cell count and distribution after a 24 h pre-culture period, after 15 min, 4 h, 24 h, 48 h, and 72 h after CAR-T-cell infusion. White arrows: NALM-6 depleted areas in the pMSC spheroid.

In this experiment, we noticed in the cross section images that the NALM-6-GFP cells tend to stay in the upper region of the microcavities although during inoculation of the mixture there was an even distribution of the cells. After CAR-T-cell infusion, there was a noticeable decrease of NALM-6 cells in the microcavity co-culture with the used effector:target ratio of 15:1 which could be quantified using the Fiji plugin “3D objects counter”. From 90,873 cells in the beginning of the experiment to 111,161 cells at the time point of CAR-T-cell administration, only 13,208 could be detected after 72 h which corresponds to a decrease of 88% by the CART-cell attack (Figure 7).

Figure 7: Quantification of the NALM-6 cell count in the microcavity array after CAR-T-cell infusion in an effector:target ratio of 15:1.

Figure 8: Determination of the CAR-T-cell potency of three different batches in the leukemic niche model of the microcavity array platform. * p ≤ 0.05.

By having shown that it is in principle feasible to detect CAR-T-cell mediated killing of NALM-6-GFP cells in the microcavity array platform, we conducted a series of experiments with different batches of CAR-Tcells to determine whether there might be differences in the potency of CAR-T-cells. For all following experiments, NALM-6 instead of NALM-6- GFP cells were used due to the possibility of a better media adaptation for the co-culture. Figure 8 shows the results of NALM-6 cell counts for three different batches of CAR-T-cells. As obvious, there seem to be large differences in the potency of the CAR-T-cells, although not significantly different in this series of 3 experiments. These data translate into the CAR-T-cell potency according to formula [1] as shown in Figure 9.

Figure 9: CAR-T-cell potency of three CAR-T-cell batches.

Control Experiment with HL-60 Cell infusion

As a control for the CAR-T-cell series, two experiments were conducted in which the CAR-T-cells were substituted by HL-60 cells. The HL-60 suspension cell lineage was treated as if they were CAR-T-cells, meaning that they were stained with CTR and infused into the bioreactor at 0 h. The green-stained leukemic cell line was counted in the microcavities as before; the change in cell count is illustrated in comparison to the NALM6 cell count due to infusion of CAR-T-cells (Figure 10). The HL-60 cell results were compared with the data from batch 1 CAR-T-cells with the highest observed potency.

Figure 10: Changes in NALM-6 cell count due to CAR-T- (batch 1) or HL-60 cell infusion, repectively. The cell count is determined for the time points -24 h, 0 h, 4 h, 24 h, and 48 h. After the inspection at 0 h, 3,000,000 CAR-T-cells or HL-60 cells, respectively, were infused.

Single Cell Killing Attack Observation

We were interested whether it was possible to observe single cell killing events in the leukemic niche and identified these from the z-stack images. Figure 11 shows a sequence of images capturing such an event. The process of CAR-T-cell docking to the NALM-6 cell and concomitant lysis lasted about 7 min. after which the CellTrackerTM Green signal of the NALM-6 cell disappeared. Since this approach is very time-consuming and random, we tried to identify similar events with two other approaches. The first approach was based on the fact that lysed cells release their intracellular ATP into the medium. This ATP can in principle catalyze a luminescence reaction such as that of luciferin. We, therefore, added recombinant Luciferase and Luciferin into the bioreactor medium prior to CAR-T-cell infusion and recorded the luminescence signal as described in the Material and Methods section. This approach was successful (Figure 12), however, the signals generated this way were very weak due to the low ATP content of a single cell and/or the enzyme/substrate concentration in the medium.

Figure 11: Image sequence of CAR-T-cell mediated NALM-6 cell lysis. In a 7 min process, the CAR-T-cell is docking to a NALM-6 cell and is lysing it.

Figure 12: Detection of a CAR-T-cell mediated killing event of a NALM-6 cell by detecting the luminescence based on the conversion of added recombinant luciferin by ATP release from the NALM-6 cell.

We then tried to show the typical IFN-ϒ release of activated CAR-T-cells. We sampled the bioreactor during pre-culture (-24 h), at time 0 h, 4 h, 24 h, and 48 h after CAR-T-cell infusion. We showed that there is a tendency for a higher IFN-ϒ release using fresh CAR-T-cells compared to frozen and thawed ones (Figure 13). However, due to the limited number of experiments and the large error bars, no statistical difference could be determined.

Figure 13: Cumulated IFN-ϒ release of fresh or frozen/thawed CAR-T-cells into the bioreactor medium (n = 4 for fresh CAR-T cells, n= 2 for frozen/ thawed CAR-T-cells), * p ≤ 0.05.

Caspase 3/7-Assay revealed CAR-T-cell activity

In case, CAR-T-cells kill target cells, Caspase 3 and 7 are released into the bioreactor medium. These can be detected by a Caspase 3/7-assay. Figure 14 shows that compared to a control without CAR-T-cell infusion, the 24 h and 48 h values of Caspase 3/7 tend to increase as obvious from the increase in activity of the medium directly above the array and the periphery of the culture dish and the reduced overall activity of the remaining cells. This experiment could only be performed once.

Figure 14: Caspase 3/7-assay of control without CAR-T-cell (left), after 24 h (middle), and 48 h after CAR-T cell infusion. * p ≤ 0.05.

Single Cell Tracking of NALM-6 and CAR-T-cells

The microcavities are confined compartments that house the cellular leukemic niche constituents. During pre-culture and after CAR T-cell infusion, there was a noticeable movement of the cells within the microcavities. We analysed the tracks of 15 NALM-6 and CAR-T-cells over a time period of 45 min. shown in Figure 15. Whereas the NALM-6 cells’ mobility was limited within the 45 min observation window, the CAR-T-cells showed a significant movement towards the center of the spheroid. This becomes all the clearer when all cell tracks are superimposed (Figure 16). When comparing the track length at the outer regions of the spheroids, longer tracks of CAR-T cells can be found compared to the NALM-6 cells, indicating a higher mobility of the CAR-T-cells.

Figure 15: NALM-6 (upper row) and CAR-T-cell (lower row) movement inside the microcavities was tracked over a period of 45 min. A: 15 min, B: 30 min, and C: 45 min after CAR-T-cell infusion. Upper row: NALM-6 cells, Lower row: CAR-T cells.

Figure 16: Superimposition of all CAR-T-cell (A) and NALM-6 cell tracks (B). The violet blobs show the spots of detected NALM-6 cells. Colors are coding for shorter tracks that upon finding by the software are designated with new colors.

DISCUSSION

In this paper, we described a platform that may provide additional information of the potency of living drugs by using a co-culture of pMSCs or the MSC line HS-5 and NALM-6 as a leukemic niche surrogate. Within the plethora of approaches to recapitulate at least some hematopoietic/ leukemic stem cell niche characteristics, the use of co-cultures is just one of them. Unfortunately, the complex nature of the stromal environment in lymph nodes, bone marrow, and secondary lymphoid also produce stem organs, characterized by a broad variety of cell populations in different differentiation stages [18, 19], as well as the characteristics of the tumor cell populations (e.g., suspension growth), have limited the development of 3D models of blood cancers [20], with only a few attempts reported in the literature [21].

For the potency assessment of the 31 FDA approved cell therapy products, 104 potency tests could be identified of which 33 are redacted, leaving 71 non-redacted potency tests. Of these, 37 assays used measurements of cell viability or cell count as a potency test [22]. However, since these data do not specify whether the test system is a 3D one, the search for published literature with the keywords “3D potency assay CAR-T-cell” is more realiable. 3D in vitro leukemia models comprise 3D hydrogels modified with extracellular matrix proteins [23-25], a hanging spheroid plate approach [26], and more recently, a leukemiaon-a-chip-model using Spongostan sheets was reported [21, 27].

Among the results are also studies using thermo-responsive scaffolds as a 3D synthetic matrix against glioblastoma spheroids [28], cancer spheroid models [29, 30], as well as a NALM-6 based system simulating the tumor suppressing microenvironment with the help of the AVATAR system as well as an acute B cell lymphoblastic leukemia mouse model that quantified the bioluminescence image monitoring method [31]. With our co-culture model of MSC/NALM-6 cells, we could show that principally CAR-T-cells are able to attack cells of the CD19+ acute lymphoblastic line NALM-6 in an effector:target-ratio of 15:1 in the microcavities very efficiently - depending on the CAR-T-cell batch - and resulting in a decreased number of ALL cells. Although simpler than organoids, the mesenchymal stromal cell spheroid system as an artificial leukemic niche has previously been shown to support B-ALL cells in vitro [32] indicating the principal suitability of the approach. Major reasons for this seem to be the existence of specialised mesenchymal stromal cells, called CXCL-12-abundant reticular (CAR) cells, that, besides CXCL-12,cell factor (SCF) and IL-7, both of which are required for the maintenance of lymphoid progenitors and mature B-cells [33, 34]. Moreover, NALM6 cells have previously been used as target cells in α-CD19-CAR-T-cell assays [31, 35].

We tested three batches of CAR-T-cells with an apparent potency difference in our assay. Whereas one batch showed a tendency to reduce the NALM-6 cell numbers, one batch showed a tendency for a moderate change, and the last one displayed a tendency to no change so that even the proliferation of the NALM-6 cells in the model could be followed. This may reflect a difference in potency of the different batches that could have made a significant difference in a clinical setting. However, for this pilot experimental series with only 3 batches of CART-cells, the results may also be biased by other factors. These include the large differences in cell counts (error bars) e.g., for the experiment in which no CAR-T-cell effect was seen.

This batch of CAR-T-cells was pre-cultivated shorter than the other batches and only 2 Mio. cells were infused due to cell shortage of this batch. Looking at the lower limit of the error bars, at least a moderate potency might also be reasonable. Re-calculation of the potency, corrected for the 33% lower cell numbers, would most likely deliver a more distinct decrease in NALM-6 cells. The emergence of the larger error bars may be linked to the threshold value of the segmentation/cell detection algorithm of Fiji which sometimes does not lead to a reliable separation of two neighbouring cells and thus needed manual adjustment.

Depending on the NALM-6 cell distribution in the microcavity, this may have led to inconsistent counting results so that an adjustment of the target cell number can probably improve the counting results since detection of single cells is easier due to a better spatial separation. Moreover, only 3 microcavities out of 634 in total could be counted at a time because of the large amount of data collected during z-stack recordings which might also be a possible source in different cell counts because NALM-6 cells obviously distribute not as homogeneously as necessary and expected. In addition to the Fiji algorithm, there are methods for improving the 3D segmentation and tailoring it to exact microcavities by using the plugin StarDist [36, 37]. An even easier method is provided via ZeroCostsDL4Mic which guides the user and provides a user interface [38, 39]. The algorithm is based on star convex shape finding, but even more complicated to detect stromal cells can easily be analyzed by annotating them correctly.

The control experiment with HL-60 cells showed that upon infusion, the number of NALM-6 cells decreases by factor of two until the end of the experiment. During the pre-culture period, we sometimes noticed a decrease in NALM-6 cell numbers priot to CAR-T-cell infusion. Looking at possible reasons, we found that the CellTrackerTM Green stain in the used concentration leads to an inhibition of NALM-6 cell proliferation and even to cell killing which negatively influences the CAR-T-cell potency calculation. As an alternative approach, the CellTrackerTM Green concentration could be decreased, the assay switched to a different dye, or NALM-6 cells expressing a reporter gene could be used, just like in the initital experiments with CAR-T-cells. To be better able to judge the data, we performed additional assays for intracellular markers released from effector and target cells, respectively.

The parameters comprised Interferon-ϒ and Caspase 3/7 release. IFN-ϒ is released from activated CAR-T-cells and leads to cell death in target cells via an increase in antigen and MHC class I molecule presentation which increases the sensitivity to T cell-mediated direct killing, and indirectly contribute to cytotoxicity through the upregulation of the cytokine TNF-α, a member of the TNF superfamily [35-41]. Moreover, IFN-ϒ reduces tumor-associated angiogenesis and induces chemokine and cytokine secretion, as well as exerts a direct anti-proliferative effect in some cancer types [41] that ultimately lead to cell death. For the IFN-ϒ release assay, we compared differences in the CAR-T-cell activation potential of fresh and frozen/thawed CAR-T-cells upon incubation in the bioreactor. We could show that fresh CAR-T-cells showed a tendency to release more IFN-ϒ upon activation whereas the IFN-ϒ level of frozen/thawed cells was hardly detectable.

A significant increase could not be detected because the data of different CAR-T-cell batches have been pooled, resulting in large error bars that may be due to a different potency of the batches as discussed above. But less IFN-ϒ release is not unlikely since it has been shown that frozen/thawed CAR-T-cell products display an elevated expression of mitochondrial dysfunction, cell cycle damage pathways, and apoptosis signaling [42]. The results point to the fact that CAR-Tcells become activated in the test system environment. This is supported by the Caspase 3/7 activity released from the target cells. Although this assay was performed within a static leukemic niche (microcavity array was placed in a Petri dish and overlayed with culture medium) and was only performed once, the tendency to an increase in Caspase 3/7 activity over the experimental time period of 48 hours support the cell count and IFN-ϒ release data.

The latter have been shown to occur in the T-cell mediated extrinsic apoptosis pathway via T-cell binding to TNF receptor 1 (TNFR1), FAS (CD95) and Death Receptor 4 upon which Caspase 8 is activated, and that in turn activates Caspase 3/7, resulting in the induction of apoptotic cancer cell death [43]. However, more experiments are needed to substantiate these findings. To complement the assay data for intracellular markers of effector cell activation, CD107a could be determined. CD107a is expressed by effector cell degranulation and correlates with cell lysis [44, 45]. In an attempt to detect single cell killing events we were able to identify these from image stacks. However, these analyses were time consuming which is why we designed an experiment in which a bioluminescence signal upon ATP-release from killed cells should be detected.

By spiking luciferase and its substrate into the bioreactor medium, we showed that single cell killing events can be detected that way. The signals were very weak and signal intensity can be improved. In the last series of experiments, we characterised the movements of NALM-6 and CAR-T-cells within the microcavities as this may be an additional parameter to look at for the characterization of CAR-T-cell potency. In the experiments performed, it became obvious that in the coculture of NALM-6 with HS-5 cells, the distribution of the cells within the microcavity differed from the pilot experiment performed with NALM-6- GFP and pMSC. In the latter, the NALM-6-GFP cells were located in the lower region of the microcavity whereas the NALM-6 cells were located in the upper part of the microcavity in the HS-5 co-culture. Therefore, it may be a characteristic of the co-culture instead of only depending on the migration behaviour of the NALM-6 cell line.

The migration behavior shifted as soon as the CAR-T-cells came close to the NALM-6 cells, and this association of cells moved within the microcavities. A particularly noticeable aspect is the movement of CAR-T-cells only at the outer cell layers of the co-culture cell cluster with no entering of the spheroid duringthe first 1.5 h after CAR-T-cell infusion. Thus, it is assumed that the HS-5 form very tight aggregates in contrast to pMSC. In addition, NALM-6 cell movement increased as more CAR-T-cells entered the cluster.

However, it was difficult to analyze the tracks because the cell movement is three dimensional and the movement was only recorded in one x/y-plane. Nevertheless, to have comparability between CAR-T- and NALM-6 cells, only two gaps in the tracks were allowed, and the distance traveled from one frame to the other was limited to twice the cell diameter. We could show that the movement of CAR-T-cells is again more pronounced than that of the NALM-6 cells. Likewise, the velocity was calculated to be twice as high for CAR-T- compared to the NALM-6 cells. However, even under the described circumstances, the CAR-T-cell action could be demonstrated. Whether this image-based approach can be of further value remains to be determined but is an additional approach to e.g., image-based T-cell assays for monitoring apoptosis and CAR expression [46].

AUTHOR CONTRIBUTIONS

Laura Theile, Maximilian Ell, Fabian Lansche, Clara Schmedt, and Cordula Nies performed the experiments. Christoph Grün proofread the manuscript and designed some of the figures. Eric Gottwald is the principal investigator and wrote the manuscript.

ACKNOWLEDGMENT

We thank Patrick Wuchter (Medical Faculty Mannheim of the Heidelberg University, Heidelberg, Germany) for providing the pMSC and Michael Schmitt (Department of Internal Medicine V, University of Heidelberg, Heidelberg, Gemany) for providing the CAR-T-cells. We also acknowledge the support by the KIT-Publication Fund of the Karlsruhe Institute of Technology

REFERENCES

1. Bernard C. La Science Experimental. Paris: J.B. Baillère & Fils; 1878.
2. Eiraku M, Takata N, Ishibashi H, Kawada M, Sakakura E, Okuda S, et al. Self-organizing optic-cup morphogenesis in three-dimensional culture. Nature. 2011; 472: 51-6.
3. Sato T, Vries RG, Snippert HJ, van de Wetering M, Barker N, Stange DE, et al. Single Lgr5 stem cells build crypt-villus structures in vitro without a mesenchymal niche. Nature. 2009; 459: 262-5.
4. Rossi G, Manfrin A, Lutolf MP. Progress and potential in organoid research. Nat Rev Genet. 2018; 19: 671-87.
5. Wuchter P, Saffrich R, Giselbrecht S, Nies C, Lorig H, Kolb S, et al. Microcavity arrays as an in vitro model system of the bone marrow niche for hematopoietic stem cells. Cell and Tissue Research. 2016; 364: 573-84.
6. Detailed product information HS-5 [Internet]. ATCC. 2025.
7. Brunner KT, Mauel J, Cerottini JC, Chapuis B. Quantitative assay of the lytic action of immune lymphoid cells on 51-Cr-labelled allogeneic target cells in vitro; inhibition by isoantibody and by drugs. Immunology. 1968; 14: 181-96.
8. Kiesgen S, Messinger JC, Chintala NK, Tano Z, Adusumilli PS. Comparative analysis of assays to measure CAR T-cell-mediated cytotoxicity. Nat Protoc. 2021; 16: 1331-42.
9. Fu X, Tao L, Rivera A, Williamson S, Song XT, Ahmed N, et al. A simple and sensitive method for measuring tumor-specific T cell cytotoxicity. PLoS One. 2010; 5: e11867.
10. Wuchter P, Saffrich R, Giselbrecht S, Nies C, Lorig H, Kolb S, et al.Microcavity arrays as an in vitro model system of the bone marrow niche for hematopoietic stem cells. Cell Tissue Res. 2016; 364: 573- 84.
11. Schubert ML, Dietrich S, Stilgenbauer S, Schmitt A, Pavel P, Kunz A, et al. Feasibility and Safety of CD19 Chimeric Antigen Receptor T Cell Treatment for B Cell Lymphoma Relapse after Allogeneic Hematopoietic Stem Cell Transplantation. Biol Blood Marrow Transplant. 2020; 26: 1575-80.
12. Giselbrecht S, Gietzelt T, Gottwald E, Trautmann C, Truckenmüller R, Weibezahn KF et al. 3D tissue culture substrates produced by microthermoforming of pre-processed polymer films. Biomed Microdev. 2006; 8: 191-9.
13. Truckenmuller R, Giselbrecht S, Rivron N, Gottwald E, Saile V, van den Berg A, et al. Thermoforming of film-based biomedical microdevices. Adv Mater. 2011; 23: 1311-29.
14. Nies C, Gottwald E. Artificial hematopoietic stem cell niches- dimensionality matters. Adv Tissue Eng Regen Med Open Access. 2017; 2: 236-247.
15. Gottwald E, Lahni B, Thiele D, Giselbrecht S, Welle A, Weibezahn KF. Chip-based three-dimensional cell culture in perfused micro- bioreactors. J Vis Exp. 2008; 15: 564.
16. Giselbrecht S, Gietzelt T, Gottwald E, Trautmann C, Truckenmuller R, Weibezahn KF, et al. 3D tissue culture substrates produced by microthermoforming of pre-processed polymer films. Biomed Microdevices. 2006; 8: 191-9.
17. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T, et al. Fiji: an open-source platform for biological-image analysis. Nature Methods. 2012; 9(7): 676-82.
18. Raic A, Naolou T, Mohra A, Chatterjee C, Lee-Thedieck C. 3D models of the bone marrow in health and disease: yesterday, today and tomorrow. MRS Commun. 2019; 9(1): 37-52.
19. Cupedo T, Stroock A, Coles M. Application of tissue engineering to the immune system: development of artificial lymph nodes. Front Immunol. 2012; 3: 343.
20. Rossi M, Alviano F, Righi S, Sabattini E, Agostinelli C. Three- dimensional models: a novel approach for lymphoma research. J Cancer Res Clin Oncol. 2022; 148: 753-65.
21. Mattioda C, Voena C, Ciardelli G, Mattu C. In Vitro 3D Models of Haematological Malignancies: Current Trends and the Road Ahead?. Cells. 2025; 14: 38.
22. Simon CG, Bozenhardt EH, Celluzzi CM, Dobnik D, Grant ML, Lakshmipathy U, et al. Analysis of the measurements used as potency tests for the 31 US FDA-approved cell therapy products. Journal of Translational Medicine. 2025; 23: 259.
23. Jacamo R, Chen Y, Wang Z, Ma W, Zhang M, Spaeth EL, et al. Reciprocal leukemia-stroma VCAM-1/VLA-4-dependent activation of NF-κB mediates chemoresistance. Blood. 2014; 123: 2691-702.
24. Torres-Barrera P, Moreno-Lorenzana D, Alvarado-Moreno JA, García- Ruiz E, Lagunas C, Mayani H, et al. Cell Contact with Endothelial Cells Favors the In Vitro Maintenance of Human Chronic Myeloid Leukemia Stem and Progenitor Cells. Int J Mol Sci. 2022; 23: 10326.
25. Barbaglio F, Belloni D, Scarfò L, Sbrana FV, Ponzoni M, Bongiovanni L, et al. Three-dimensional co-culture model of chronic lymphocytic leukemia bone marrow microenvironment predicts patient-specific response to mobilizing agents. Haematologica. 2021; 106: 2334-44.
26. Chen Z, Han S, Sanny A, Chan DL-K, van Noort D, Lim W, et al. 3D hanging spheroid plate for high-throughput CAR T cell cytotoxicity assay. Journal of Nanobiotechnology. 2022; 20: 30.
27. Ma C, Witkowski MT, Harris J, Dolgalev I, Sreeram S, Qian W, et al. Leukemia-on-a-chip: Dissecting the chemoresistance mechanisms in B cell acute lymphoblastic leukemia bone marrow niche. Science Advances. 2020; 6: eaba5536.
28. Lizana-Vasquez GD, Ramasubramanian S, Davarzani A, Cappabianca D, Saha K, Karumbaiah L, et al. In Vitro Assessment of Thermo- Responsive Scaffold as a 3D Synthetic Matrix for CAR-T Potency Testing Against Glioblastoma Spheroids. J Biomed Mater Res A. 2025; 113: e37823.
29. Weems O, D. S, Millard D. CAR T Cell potency assessment with 3D cancer spheroid models 2022.
30. Zurowski D, Patel S, Hui D, Ka M, Hernandez C, Love AC, et al. High- throughput method to analyze the cytotoxicity of CAR-T Cells in a 3D tumor spheroid model using image cytometry. SLAS Discovery. 2023; 28: 65-72.
31. Bronevetsky Y, Massi E, Garcia C, Liu N, Xing Y, Czeryba N, et al. editors. Functional potency assay predicts CAR-T effectiveness in tumor microenvironment. Cancer Research. 2023; 87: 1782.
32. Balandrán JC, Dávila-Velderrain J, Sandoval-Cabrera A, Zamora- Herrera G, Terán-Cerqueda V, García-Stivalet LA, et al. Patient- Derived Bone Marrow Spheroids Reveal Leukemia-Initiating Cells Supported by Mesenchymal Hypoxic Niches in Pediatric B-ALL. Front Immunol. 2021; 12: 746492.
33. Pinho S, Frenette PS. Haematopoietic stem cell activity and interactions with the niche. Nat Rev Mol Cell Biol. 2019; 20: 303-20.
34. Cordeiro Gomes A, Hara T, Lim VY, Herndler-Brandstetter D, Nevius E, Sugiyama T et al. Hematopoietic Stem Cell Niches Produce Lineage-Instructive Signals to Control Multipotent Progenitor Differentiation. Immunity. 2016; 45: 1219-31.
35. Montalvo MJ, Bandey IN, Rezvan A, Wu K-L, Saeedi A, Kulkarni R et al. Decoding the mechanisms of chimeric antigen receptor (CAR) T cell-mediated killing of tumors: insights from granzyme and Fas inhibition. Cell Death & Disease. 2024; 15: 109.
36. Schmidt U, Weigert M, Broaddus C, Myers G. Cell Detection with Star-convex Polygons. International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI); 2018; 265-273.
37. Weigert M, Schmidt U, Haase R, Sugarawa K, Myers G. Star-convex Polyhedra for 3D Object Detection and Segmentation in Microscopy. EEE Winter Conference on Applications of Computer Vision (WACV). 2020; 3655-62.
38. Laine RF, guijacquemnt, Henriques R, Kretzel D, Quynag W, esgomezm, et al. HenriquesLab/ZeroCostDK4Mic: v. 1.13. Zenodo; 2021.
39. von Chamier L, Laine RF, Jukkala J, Spahn C, Krentzel D, Nehme E, et al. Democratising deep learning for microscopy with ZeroCostDL4Mic. Nat Commun. 2021; 12: 2276.
40. Castro F, Cardoso AP, Gonçalves RM, Serre K, Oliveira MJ. Interferon- Gamma at the Crossroads of Tumor Immune Surveillance or Evasion. Front Immunol. 2018; 9: 847.
41. Shen J, Xiao Z, Zhao Q, Li M, Wu X, Zhang L, et al. Anti-cancer therapy with TNFα and IFNγ: A comprehensive review. Cell Prolif. 2018; 51: e12441.
42. Panch SR, Srivastava SK, Elavia N, McManus A, Liu S, Jin P, et al. Effect of Cryopreservation on Autologous Chimeric Antigen Receptor T Cell Characteristics. Mol Ther. 2019; 27: 1275-85.
43. Lee YG, Yang N, Chun I, Porazzi P, Carturan A, Paruzzo L, et al. Apoptosis: a Janus bifrons in T-cell immunotherapy. J Immunother Cancer. 2023; 11: e005967.
44. Alter G, Malenfant JM, Altfeld M. CD107a as a functional marker for the identification of natural killer cell activity. J Immunol Methods. 2004; 294: 15-22.
45. Aktas E, Kucuksezer UC, Bilgic S, Erten G, Deniz G. Relationship between CD107a expression and cytotoxic activity. Cell Immunol. 2009; 254: 149-54.
46. Huang Y, Guo N, Pierce M, Chan L, Lin B, Yang Y. Developing image- based T-cell assays for monitoring apoptosis and CAR expression. ELRIG Drug Discovery 2024.
47. SOURCE NB. Confocal Microscope. In: BIOART N, editor. 2024.

Citation

Theile L, Ell M, Lansche F, Schmedt C, Gottwald E et al, (2025) An Artificial Leukemic Niche for in vitro Potency Testing of Genetically Engineered Living Drugs. SM J Biomed Eng 7(1): 13.