This protocol presents an automated, image-based high-throughput technique to identify compounds modulating natural killer cell-mediated breast cancer cell killing in the presence of a therapeutic anti-HER-2 antibody.
Immunotherapy with antigen-specific antibodies or immune checkpoint inhibitors has revolutionized the therapy of breast cancer. Breast cancer cells expressing the epidermal growth factor receptor HER2 can be targeted by the anti-HER-2 antibody trastuzumab. Antibody-dependent cellular cytotoxicity (ADCC) is an important mechanism implicated in the antitumor action of HER-2. Trastuzumab bound to cancer cells can be recognized by the Fc receptors of ADCC effector cells (e.g., natural killer (NK) cells, macrophages, and granulocytes), triggering the cytotoxic activity of these immune cells leading to cancer cell death. We set out to develop an image-based assay for the quantification of ADCC to identify novel ADCC modulator compounds by high-content screening. In the assay, HER2 overexpressing JIMT-1 breast cancer cells are co-cultured with NK-92 cells in the presence of trastuzumab, and target cell death is quantified by automated microscopy and quantitative image analysis. Target cells are distinguished from effector cells based on their EGFP fluorescence. We show how compound libraries can be tested in the assay to identify ADCC modulator drugs. For this purpose, a compound library test plate was set up using randomly selected fine chemicals off the lab shelf. Three microtubule destabilizing compounds (colchicine, vincristine, podophyllotoxin) expected to interfere with NK cell migration and degranulation were also included in the test library. The test screen identified all three positive control compounds as hits proving the suitability of the method to identify ADCC-modifying drugs in a chemical library. With this assay, compound library screens can be performed to identify ADCC-enhancing compounds that could be used as adjuvant therapeutic agents for the treatment of patients receiving anticancer immunotherapies. In addition, the method can also be used to identify any undesirable ADCC-inhibiting side effects of therapeutic drugs taken by cancer patients for different indications.
Immunotherapy with anticancer antibodies, immune checkpoint inhibitors, or chimeric antigen receptor-expressing T (CAR-T) cells represents a powerful approach to cancer treatment1,2,3. Trastuzumab is a humanized monoclonal anti-HER-2 (human epidermal growth factor receptor 2) antibody used for treating HER-2 positive early stage or metastatic breast cancer, as well as HER-2 positive metastatic gastric cancer4,5,6. It primarily acts by inhibiting the proliferation stimulating effect of the epidermal growth factor4. It has been reported, however, that trastuzumab efficiently triggers cancer cell death even if the cancer cells have lost their responsiveness to HER-2 stimulation7. This paradoxical effect of the antibody is due to antibody-dependent cell-mediated cytotoxicity (ADCC)7. ADCC can be mediated by natural killer (NK) cells, granulocytes, and macrophages collectively known as the effector cells of ADCC8,9. If an antibody, such as trastuzumab, binds to tumor cells, then these effector cells use their Fc receptors to bind the constant (Fc) region of the antibody. The antibody bridges the tumor cells and the Fc receptor-bearing effector cells, triggering the release of their cytotoxic mediators10. Natural killer cells release the cytotoxic cargo of their granules containing perforin to generate pores in the target cell membrane and granzyme (triggering cell death signaling pathways) into the immune synapse leading to apoptosis of the cancer cells (see Figure 1).
Figure 1: Effector and target cell interactions in ADCC. The cell surface Fcγ receptor of the effector NK cell recognizes the Fc region of the anti-HER2 trastuzumab antibody specific for the HER2 molecule expressed on the surface of the tumor cell. Thus, the so-called immunological synapse is established between the two cells, inducing the directed exocytosis of cytotoxic granules of the effector cell. The released perforin and granzyme molecules eventually result in apoptosis of the target cell. Please click here to view a larger version of this figure.
Several assays have previously been developed to quantify cytotoxicity, including ADCC. The gold standard is the radioactive chromium release method, where the target cells are labeled with radioactive 51Cr isotope, and ADCC is quantified by measuring radioactivity from the supernatant of lysed target cells11. Because of the obvious problems due to the strictly regulated handling, storage, and disposal of radioactive pharmacons and wastes, this method has become increasingly non-popular among life scientists. In addition, it is not amenable to high-throughput applications either. Measuring the activity of enzymes (e.g., lactate-dehydrogenase) released from the killed target cells can provide a non-radioactive alternative to the 51Cr assay12. These assays, however, fail to distinguish between target and effector cell deaths. Electric Cell-substrate Impedance Sensing (ECIS) proved suitable for the quantification of ADCC13, but the ECIS equipment is not available in most laboratories, and the technique is not compatible with high-throughput applications/screening. Fluorescently labeled cells represent a popular alternative in many cell biology assays and are often used in flow cytometry or plate reader-based applications14,15,16. However, these assays often contain washing steps or are otherwise incompatible with high-throughput applications (e.g., flow cytometry-based techniques). Some popular cytotoxicity assays, which in theory should be suitable for ADCC quantification, fail to reliably determine ADCC efficiency13. Recently, with the spreading of fluorescent confocal microscopy, image-based, high-content assays are becoming increasingly popular in various areas of life sciences17. On the one hand, cell imaging equipment are now rather ubiquitous, while, on the other hand, virtually endless morphological parameters can be gathered from the acquired images. Therefore, we set out to develop a high-content screening compatible ADCC assay and to demonstrate its suitability for compound library screening.
Here, we present an image-based ADCC assay and demonstrate how this assay can be used for High-Content Screening (HCS) to identify ADCC modulating compounds. The model is based on JIMT-1 breast carcinoma target cells, CD16.176V.NK-92 effector cells and the humanized monoclonal anti-HER2 antibody trastuzumab. With this method, it is possible to identify drugs that can enhance the tumor-killing action of NK cells or to gain insight into the mechanism of NK cell-mediated ADCC by identifying small molecules interfering with ADCC. We suggest that life scientists aiming to quantify cell-mediated cytotoxicity with special regard to ADCC may benefit from using this assay either for the discovery science or drug development. This assay may be an alternative if a laboratory has access to and some experience in fluorescent imaging and quantitative image analysis.
NOTE: Key steps of the assay workflow are presented in Figure 2.
Figure 2: Workflow of the ADCC screen. JIMT-1-EGFP target cells seeded into 96 well HCS plates are treated with drugs of the compound library. In turn, unstained NK (effector) cells and trastuzumab are added, and the plate is imaged at 0 timepoint and after 3 h of incubation. ADCC evaluation is based on the change in the number of viable (surface adherent) target cells. Please click here to view a larger version of this figure.
1. Coating of the HCS plate
2. Seeding of JIMT-1 Enhanced Green Fluorescent Protein (EGFP) cells
NOTE: EGFP-expressing JIMT-1 cells were generated in our previous work18, and the cells were cultured in T25 tissue culture flasks in JIMT-1 media (see composition in step 1.1).
3. Pre-treatment of JIMT-1 EGFP cells with the compound library
4. Starting the ADCC assay by adding the effector cells
NOTE: CD16.176V.NK92 cells (hereafter referred to as NK92 cells) were cultured in α-MEM supplemented with 20% FBS, 1% MEM-NEAA, 1% Na-pyruvate, 1% glutamine, 1% penicillin-streptomycin and 100 IU/mL IL-2.
5. Imaging
NOTE: The plates should be imaged at two time points, first, immediately after the addition of the effector cells to the target cells and second, at 3 h after the addition of NK cells. For imaging, the high-content analyzer and its software or suitable alternatives can be used (see Table of materials).
6. Image analysis
NOTE: To analyze the ADCC efficiency, the viable JIMT-1 cells are counted. Target cells killed by ADCC detach from the surface and move away from the focal plane of the microscope. Therefore, the difference between the number of viable cells at the beginning and at the end of the ADCC reaction corresponds to target cells eliminated by ADCC. To show how to build up the evaluation sequence, a control ADCC well is shown in the video.
To demonstrate how the assay works in real life, we created a test library of 16 compounds selected randomly from the lab shelves (Figure 3). In addition, DMSO was also included as a negative control, and three microtubule polymerization inhibitor compounds (colchicine, vincristine, and podophyllotoxin) as positive controls. The latter were expected to inhibit ADCC by interfering with NK cell migration to the cancer cells and NK cell degranulation. All test compounds and DMSO were placed onto the test library plate in quadruplicates, and DMSO was also added to the first and last columns of the plate (Figure 3).
Figure 3: The compound library used for testing the HCS ADCC assay. (A) The plate map of the compound library is shown. Each compound is present on the library plate in quadruplicates. DMSO was added as a negative control to the first and last column of the plate. Wells containing DMSO and the three microtubule assembly inhibitors (colchicine, vincristine, and podophyllotoxin) are highlighted in color. (B) Names, plate positions, and abbreviated names of test compounds are presented. Please click here to view a larger version of this figure.
Using our test library, we ran the assay and evaluated the results as described in the protocol. Since JIMT-1 cells express EGFP, they can be easily distinguished from the effector cells that are non-fluorescent. The assay is based on detecting the change in the number of surface adherent (viable) target cells. Under our assay conditions, NK cells caused approximately 50% JIMT-1 cell death (+NK group), while none of the test compounds caused any toxicity in the absence of NK cells (-NK group) (Figure 4). The assay conditions were previously optimized to achieve this medium level of cytotoxicity18, assuming that this would permit the identification of both ADCC activator and inhibitor compounds. The positive control microtubule assembly inhibitors showed up as "hits" in all quadruplicate positions as expected, indicating the high reliability of the assay (Figure 4).
Figure 4: Testing a compound library in the ADCC model. (A) Images of ADCC reactions were taken 3 h after co-incubation of NK cells and JIMT-1 cells in the presence of DMSO with the 10x objective of the HCS imager. (The scale bar is 200 µm.) One part of the image is magnified, showing the unstained effector NK cells and the EGFP-transduced target JIMT-1 cells. (The magnification of the original image is 4x, the scale bar is 50 µm.) (B) The ADCC assay was performed with EGFP-transduced JIMT-1 breast carcinoma cell line and CD16.176 V.NK-92 cell line. The applied effector to target (E:T) ratio was 2:1. In the case of -NK group, the JIMT-1 EGFP cells were incubated without NK cells and anti-HER2 antibody, while in the +NK group (ADCC) 10 µg/mL trastuzumab was added to the JIMT-1 and NK cell co-cultures. The number of viable JIMT-1-EGFP cells was detected using high content analysis equipment immediately after the addition of NK cells and after 3 h of incubation. Target cell viability in the +NK group was 50% of that in the -NK group. The red dashed line shows the threshold value, which represents samples with ≥70% viability compared to the average of DMSO control (n=20) killing. Please click here to view a larger version of this figure.
The ADCC reaction has been described a relatively long time ago. Key molecular events of the process have also been described19. Methods for measuring ADCC range from the gold standard radioactive chromium release assay, cytoplasmic enzyme release assays to several fluorescence-based flow cytometry or microplate assays20. However, a common limitation of these assays is that they are not amenable to high-throughput applications. Previously, we developed an image-based HCS assay suitable for screening compound libraries for ADCC-modifying drugs18. The previously developed assay, however, had a major drawback i.e., it required a washing step because the target cells had to be pre-stained with calcein-acetoxymethylesther, and the excess dye had to be removed before the addition of the NK cells and trastuzumab. The current method represents an advanced version of the assay in that the use of EGFP-transduced JIMT-1 cells permits straightforward differentiation between effector and target cells without a staining and washing step. Moreover, one must also be aware of the feature of our assay that it cannot distinguish between dead and dying cells (not yet fully detached). Furthermore, it is possible that the test compounds may increase cell adhesion, and, as a result cell debris may remain in the focal plane resulting in a false positive result. In addition, the fluorescent properties of test compounds may cause interference. Therefore, although screens are typically run once, it is highly recommended to validate hits in other types of assays as well. Additional limitations of our assay include the indirect identification of cell death i.e., the method is based on the quantification of the change in the number of viable cells instead of dead cells. However, it is important to note here that a previous study found that ECIS-based detection of viable cells was superior to some of the methods directly measuring cell death13. Therefore, this feature of our method does not necessarily represent a drawback. What may really limit the application of this assay, however, is that it requires at least a basic knowledge in advanced image analysis, a skill that is becoming increasingly common today.
Although so far, a small library (less than 1000 compounds) has been tested with the original version of the assay18. We have not yet identified an ADCC booster drug. The reason for this might be that ADCC activators might be rare, and much bigger libraries need to be screened to identify one. Theoretically, it is also possible that once the ADCC reaction starts, it proceeds without major limiting steps so that the process would be difficult to improve further by a pharmacon. One possible way around this problem might be to suppress ADCC activity (e.g., with corticosteroids) and run the screen looking for compounds restoring full ADCC activity. Nevertheless, we think that identifying ADCC inhibitory effect of known medical drugs may also be important in order to raise awareness of clinicians prescribing these drugs to patients with various indications.
As for the technical details of the assay, we recommend setting up the assay in parallel with an alternative cytotoxicity test. In our hand, ECIS (electric cell-substrate impedance sensing) proved to be the most reliable system for the quantification of ADCC efficiency13. At the start of the project, ECIS was used to adjust critical parameters (time, effector to target cell ratio, trastuzumab concentration) and then transferred the assay to the HCS platform18. We recommend doing the same and fine tune these parameters, as they might change from lab to lab. When setting up the assay, it is important to pay special attention to some of the critical steps of the procedure. Standardizing culture conditions (splitting frequency, cell feeding, plating consistency) for both NK cells and target cells may increase reproducibility of ADCC efficiency. Moreover, suspending NK cell cultures must be performed with care as these cells tend to be sensitive to mechanical stress (Guti et al. unpublished observation). As for the compound library, it is important to make sure that all wells contain equal volumes of test compounds if pin tools are used for the transfer of compounds from the library plate to the assay plate. This may prevent varying amounts of compound solutions from sticking to the outer side of the pins. Furthermore, hit compounds identified in a screen should be validated preferably in a reliable assay based on a different principle. For this, we recommend the ECIS-based method11 (see above in the Introduction).
In conclusion, we set up a JIMT-1 EGFP-based in vitro ADCC assay system suitable for testing compound libraries with automated image analysis. The method was suitable for the identification of pharmacons with known ADCC modifying effects. The method is potentially suitable for the determination of cancer cell death caused by trastuzumab drug conjugates (e.g., recently developed trastuzumab-emtansine21) where the mechanism of toxicity is more complex and is only partially caused by ADCC and partially by the tubulin inhibitor mertansine. In the future, we plan to adopt the technology to quantify ADCC in 3D spheroids, which represent a more accurate reflection of tumor behavior than 2D models.
The authors have nothing to disclose.
LV received funding from National Research, Development and Innovation Office grants GINOP-2.3.2-15-2016-00010 TUMORDNS", GINOP-2.3.2-15-2016-00048-STAYALIVE and OTKA K132193, K147482. CD16.176V.NK-92 cells were obtained from Dr. Kerry S. Campbell (Fox Chase Center, Philapedlphia, PA, on behalf of Brink Biologics, lnc. San Diego, CA), are protected by patents worldwide, and were licensed by Nantkwest, lnc. Authors are thankful to György Vereb and Árpád Szöőr for their help with the use of the NK-92 cell line and for technical advice.
5-fluorouracil | Applichem | A7686 | in compound library |
96-well Cell Carrier Ultra plate | PerkinElmer | LLC 6055302 | |
Betulin | Sigma | B9757 | in compound library |
CD16.176V.NK92 cells | Nankwest Inc. | ||
Cerulenin | ChemCruz | sc-396822 | in compound library |
Cisplatin | Santa Cruz Biotechnology | sc-200896 | in compound library |
Colchicine | Sigma | C9754 | in compound library |
Concanavalin-A | Calbiochem | 234567 | in compound library |
Dexamethasone | Sigma | D4902 | in compound library |
DMEM/F-12 medium | Sigma | D8437 | in JIMT-1 EGFP medium |
DMSO | Sigma | D2650 | in compound library |
Etoposide | Sigma | E1383 | E1383 |
Fetal bovine serum (FBS) | Biosera | FB-1090/500 | JIMT-1 EGFP and NK medium |
Fisetin | Sigma | F4043 | in compound library |
Freedom EVO liquid handling robot | TECAN | ||
Gallotannin | Fluka Chemical Corp. | 16201 | in compound library |
Glutamine | Gibco | 35,050–061 | in NK medium |
Harmony software | PerkinElmer | ||
Humanized anti-HER2 monoclonal antibody (Herzuma) | EGIS Pharmaceuticals, Budapest Hungary | N/A | |
Humulin R (insulin) | Eli Lilly | HI0219 | JIMT-1 EGFP medium |
IL-2 | Novartis Hungária Kft. | PHC0026 | in NK medium |
Isatin | Sigma | 114618 | in compound library |
MEM Non-essential Amino Acids (MEM-NEAA) | Gibco | 11,140–050 | in NK medium |
Na-pyruvate | Lonza | BE13-115E | in NK medium |
Naringenin | Sigma | N5893 | in compound library |
NQDI-1 | Sigma | SML0185 | in compound library |
Opera Phenix High-Content Analysis equipment | PerkinElmer | ||
Penicillin–streptomycin | Biosera | LM-A4118 | JIMT-1 EGFP and NK medium |
Pentoxyfilline | Sigma | P1784 | in compound library |
Phosphate buffered saline (PBS) | Lonza | BE17-517Q | to wash the cells |
Podophyllotoxin | Sigma | P4405 | in compound library |
Quercetin | Sigma | Q4951 | in compound library |
Tannic acid | Sigma | T8406 | in compound library |
Temozolomide | Sigma | T2577 | in compound library |
Trypan blue 0.4% solution | Sigma | T8154 | for cell counting |
Vincristine sulfate | Sigma | V0400000 | in compound library |
α-MEM | Sigma | M8042 | in NK medium |