Confocal Microscopy and Imaging Flow Cytometry – Tools for Selection of Antibodies to be Developed as Therapeutic ADCs

Research article

Antibody-drug Conjugates (ADCs) are a growing class of targeted anti-cancer therapies, that combine the specific binding of an antibody with the toxic effects of chemotherapy. In developing new ADCs, one should take into consideration not only the affinity and binding properties of the antibody but also its ability to internalize into the lysosome, where the conjugated drug is released. Here we describe how the use of confocal microscopy and imaging flow cytometry (IFC) analyses assist in the process of ADC development. We show that the use of the Bright Detail Intensity (BDI) vs. the Bright Detail Similarity (BDS) features enables better selection of what we define as “best candidate to be used as an ADC” and discuss the advantages of using both confocal and IFC in combination, rather than each of these methods alone.

Authors: Mira Toister-Achituv ♦, Ziv Porat,* Doron Kalimi,*  Ohad Tarcic, Nir Berger, Achim Doerner, Arie Zauberman [Author Information]

Key terms: Confocal Microscopy, ImageStream, IFC, Flow cytometry, Antibody-drug conjugate (ADC), internalization, colocalization, Lysosome, BDI, BDS.

Credits for research & footnotes
This work was supported by the Discovery Development Technologies (DDTech) and the ADC departments at Merck KGaA, Darmstadt.


ADCs are antibody-based anti-cancer therapies, that aim to confer significant improvements to current antibody treatments or chemotherapies alone. [1]. They combine the antigenic specificity of an antibody with anti-proliferative cytotoxic compounds which are linked by specific linkers [2][3][4][5]. This combination enables potent chemotherapies to target the tumor in a selective manner and unload their cytotoxic payloads inside the tumor cells. The approach of antibody-mediated drug delivery holds the promise of widening the therapeutic window of a toxic drug, by increasing its maximal tolerated dose and lowering the minimal efficacious dose of the treatment. To date, Kadcyla® [5] and Adcetris® [6] are the only ADCs approved and marketed, but more than 40 ADCs are currently in clinical trials combining a variety of linkers and payloads. [7]

In ADCs, specialized linkers are used for conjugation of drugs to the antibody, which is divided into two categories: cleavable and non-cleavable linkers. [8] Peptide cleavable linkers such as Phe-Lys and Val-Cit are frequently used due to their relative stability in human serum and rapid ability to hydrolyze in the presence of lysosomal enzymes within a cell. [9][10]

Non-cleavable linkers are also used, which are highly stable in human sera and can release the active drug within the cell after degradation of the antibody within the lysosome. [11][12] As such, most ADCs to date require internalization into the cancer cell and transport to the lysosome, for the drug to be released and subsequently reach its target destination in the cell: cytoplasmic when using drugs that target e.g. the microtubules such as Dolastatin 10 [13], or nuclear when using DNA disruptors such as Calicheamicin which induces DNA double-strand breaks. [14]

Consequently, when developing new ADCs, the ability to choose the best antibody from a panel of candidates is crucial for all downstream processes to come. This choice cannot depend solely on the affinity and binding properties of the antibody, as is done when the drug is the antibody itself [15][16], but must take into account also its ability to internalize efficiently into cells and be transferred to the lysosome. These capabilities need to be tested on all candidate antibodies, from which the final hit is chosen and must not be disrupted after the antibody is conjugated to the linker and payload of choice.
To date, antibodies that are developed as ADCs are chosen either solely by affinity to their antigen target, or in more advanced cases, by analyzing their internalization abilities by Conventional Flow Cytometry (CFC) and transport to the lysosomes by confocal microscopy. [17] CFC analysis is a high throughput technique, enabling detection of internalization of a fluorescently labeled antibody to many cells in a high throughput manner, but it does not confer spatial resolution of the fluorescent signal within the cells. Thus, CFC analysis can provide internalization levels according to fluorescent intensity but cannot indicate where the internalized antibody is located in the cell. In contrast, confocal microscopy’s high-resolution imaging enables clear detection of fluorescently labeled organelles within a cell. With the aid of fluorescent dyes that stain specifically the internalizing antibodies, it enables detection of localization of the antibodies within these organelles. This technique can be used for primary qualitative screening of a number of candidate antibodies on several cancer cell lines. But the quantification of colocalization or comparison of internalization of several antibodies by confocal microscopy is highly time-consuming and tedious, and only a relatively limited number of cells can be screened in each experiment. Additionally, if the cancer cell lines being analyzed grow in clumps or sheets, methods for analyzing confocal images, which provide digital segmentation to single cells, are non-optimal and may lead to distorted results and errors in calculations.

Imaging Flow Cytometry (IFC) combines the high-throughput and quantification abilities of flow cytometry with the detailed imagery of microscopy. Briefly, cells in suspension pass in a single file through a set of lasers and are imaged by a sensitive CCD camera in time delay integration mode, enabling simultaneous acquisition of bright-field, dark-field as well as up to 10 different fluorescent channels. [18] There are several advantages for IFC that make it ideal for high throughput analysis: cells pass one by one through the flow chamber and are imaged in a uniform manner, reducing the variability between cells and facilitating consistent analysis. As cells are in suspension, segmentation is not an issue and analysis is carried out on single cells. Using the manufacturer dedicated software, ImageStream Data Exploration and Analysis Software (IDEAS®), automated, unbiased quantification of morphological features as internalization and colocalization is carried out on a population of cells and applied unequivocally to all samples. Ideally, this technique could suffice to determine internalization and colocalization comparisons, yet the ImageStream® apparatus is much less abundant than confocal microscopes, in many cases situated within a general facility, rendering it much less accessible for wide use. Additionally, though the imagery abilities of IFC are of high resolution, they do not reach the explicit analysis obtained by confocal microscopy.

In this article, we show how the combined use of confocal microscopy and IFC analyses support each other in the process of ADC development. The analysis is initiated by confocal microscopy on a number of antibodies providing unequivocal information regarding internalization and the localization of the antibodies to chosen internal organelles. Though this data is qualitative, it suffices for discarding antibodies that clearly do not internalize into the cells or localize to the lysosomes. Antibodies are then chosen for IFC analysis, for comparative quantification of internalization and colocalization in tens of thousands of cells.

We show how combining two features of the IFC analysis software, takes into account not only the colocalization factor but the amount of antibody that internalizes into the cell. This results in selecting an antibody with enhanced internalization and lysosomal targeting which should yield a good backbone for the new ADC. [2] Finally, we show how these techniques to aid in more advanced stages of ADC development such as comparison to other therapeutic antibodies that bind the same target and choosing the final antibody format to be used as the therapeutic ADC.

Materials and Methods

Cell cultures and growth conditions
MDA-MB231 breast cancer cell line (ATCC® HTB-26™) and A431 epidermoid carcinoma cells (ATCC® CRL-1555™) were cultured in DMEM medium at 37°C in 10% CO2. MDA-MB468 breast cancer cell line (ATCC® HTB-132™) was cultured in RPMI 1640 medium at 37°C in 5% CO2. All cells were passaged every 3-4 days. A detachment of cells was carried out with TrypLE Express (Gibco cat:12604-013) or EDTA 0.5mM EDTA (03710 SIGMA).

Confocal microscopy internalization assays
A day before the experiment, sterile coverslips (Marienfled 5-CG18) were placed in 12 well plates (Greiner Cat:665180) and coated with 0.4ml Fibronectin (SIGMA: F1141) for 1hr at 37°C. Cells in exponential growth phase were harvested using TrypLE Express (Gibco cat:12604-013), seeded at a concentration of 0.25X106 cells/well, and incubated over-night at 37°C, 5% CO2.

On the day of the experiment tested antibodies were diluted in cell growth medium to a final concentration of 5µg/ml and 0.5ml was applied on the cells. All plates were incubated on ice for 1hr binding (Time 0). Time 0 plates were then fixed. Plates for internalization were transferred to 37°C, 5% CO2, and incubated for 1.5-5hr, depending on the antibody used, and then fixed (for specific times for each antibody see results section and figure legends).

Fixation was carried out with 0.5ml of 4% paraformaldehyde (EMS cat:15710) for 20 minutes at room temperature, followed by 3 x 2ml PBSX1 washing and 10 minutes neutralization with 0.5ml of 0.1M glycine (Merck 1.04201).
Permeabilization was performed by adding onto the wells 0.3 ml of 0.3% Triton X-100 (Merck Cat#648466) for 2 minutes. The samples were blocked in 3% BSA (Bovo Star BSAS 0.10) for 1hr. Staining was carried out in a dark and humid environment for 40 minutes using a 0.3ml mixture of secondary antibody (Anti-human FC Alexa Fluor488 Jackson Cat# 109-546-170 final concentration 7.5μg/ml), lysosome marker (Anti LAMP2 Ab Bioss bs-2379R-A555, final concentration 3.3 μg/ml), Late endosome marker (anti-Rab7 Abcam AF647 ab198337, final concentration 0.7 μg/ml) and DAPI (SIGMA Cat# Dg542, final concentration 1μg/ml) in 1% BSA. After two washing steps in PBSx1 the cover slips were mounted (Dako S3032) on microscope slides (Thermo-Scientific Cat # J1800AMNZ) and left to dry over-night before confocal microscopy analysis.

Confocal microscopy and analysis
Confocal microscopy was carried out using an Olympus FluoView FV1000 IX81 confocal laser-scanning microscope. Pinhole size was set to 1 Airy unit at pixel size 0.1µm, using an UPLSAPO 60x/1.35 NA oil-immersion objective. Lasers suitable for DAPI (excitation: 405nm laser intensity 1%, emission: 427-471nm), Alexa Fluor® 488 (excitation 488nm at laser intensity 0.2%, emission: 501-545nm), Alexa Fluor® 555 (excitation: 559nm laser intensity 10%, emission: 575-675nm) and Alexa Fluor® 647 (excitation: 635nm laser intensity 10%, emission: 655-755nm) were used. Data was collected sequentially. Analysis was done using the FIJI free software.
Internalization for IFC analysis

For internalization analysis cells were harvested using 0.5mM EDTA (03710 SIGMA). For each time-point, 3ml of the medium containing 5X106 cells and tested antibodies (1.7µg) were seeded into 6 well plates.

Samples for time 0 were incubated for 1hr on ice with shaking. Samples for internalization were incubated at 37°C, in a 5% CO2 incubator at 125 RPM for the time required for internalization, depending on the antibody used (exact time for each antibody is detailed in the Results section). At the end of each time point, samples were transferred to 15ml tubes and centrifuged at 1400 RPM for 5 minutes and the supernatant was aspirated.

Stripping of external antibody (membrane-bound, non-internalized antibody) was carried out by adding 0.75ml of PBS pH 2.5 for 2.5 minutes which was then neutralized with 1ml PBS pH 8.9. Samples were centrifuged at 1400 RPM for 5 minutes and the supernatant was aspirated.

Fixation and staining for IFC analysis
Fixation was carried out by adding 3.3ml ice-cold methanol (Merck 1.06009) to a final concentration of 70% and incubating over-night at -20°C. The following day 10ml of PBS pH 7.4 was added to each sample, the samples were centrifuged and washed twice. Permeabilization was carried out by adding 0.05% Triton X-100 (Merck Cat#648466) in PBS at room temperature for 15 minutes after which samples were centrifuged and washed twice in PBS. Blocking was carried out using 3% BSA (Bovo Star BSAS 0.10) in PBS for 1hr. Staining was carried out using a 0.3ml mixture of secondary antibody (Anti-human FC Alexa Fluor 488, Jackson Lab. Cat# 109-546-170 final concentration 7.5 μg/ml), lysosome marker (Anti LAMP2 Ab Bioss bs-2379R-A555, final concentration 10μg/ml), late endosome marker (Anti–RAB7 EPR7589 Abcam Cat: ab198337, final concentration 0.35μg/ml) and DAPI (SIGMA Cat: Dg542, final concentration 1μg/ml) in 1% BSA, for 40 minutes in the dark. After several washes, samples were resuspended in 5ml of PBS+5mM EDTA (SIGMA Cat #03710) and kept at 4°C. On the day of analysis, samples were centrifuged and resuspended in 50µl of PBS + 5mM EDTA (03710 SIGMA).

Multispectral imaging flow-cytometry (ImageStreamX®) analysis

Data collection
Cells were imaged using multispectral imaging flow cytometry (ImageStreamX mark II imaging flow-cytometer; Amnis Corp, Seattle, WA, Part of EMD Millipore). A 40X lense was used (NA=0.75) and excitation lasers: 405nm for DAPI (Channel 7), 488nm for ADC-AF488 (Channel 2), 561nm for AF555 (channel 3), and 642nm for AF647 (Channel 11). Cells were gated for single cells, using the Area (the number of pixels covered by the image mask, in square microns) and Aspect ratio (the Minor Axis divided by the Major Axis of the best-fit ellipse) features, and for focused cells, using the Gradient RMS feature (measures the sharpness quality of an image by detecting large changes of pixel values in the image). From each sample 1-3×104 cells were collected, at a rate of 50 cells/sec enabling analysis of 30,000 in approximately 10 minutes.

Data analysis
Data were analyzed using the manufacturer image analysis software (IDEAS® 6.2; Amnis Corp). Images were compensated for fluorescent dye overlap by using single-stain controls. Acquisition gating was verified again by plotting the Area and Aspect ratio features of the brightfield for single cells. To further exclude debris and aggregates the area of the DAPI staining (channel 7) was plotted against its intensity. To include only focused cells, the Gradient RMS and contrast features were used. Cropped cells were further eliminated by plotting the cell area of the bright field image against the Centroid X feature (the number of pixels in the horizontal axis from the left corner of the image to the center of the cell mask). Cells positively stained for the examined antibody were selected according to their Intensity (the sum of the background-subtracted pixel values within the masked area of the image) and the Max Pixel feature (the largest value of the background-subtracted pixels contained in the input mask). Gating was set according to a sample stained with all dyes except the primary antibody for the ADC. Images of the gating strategy can be seen in supplement figure 1.

Internalization was calculated using the Internalization feature of the IDEAS® software. This calculates the ratio of the intensity inside the cell to the intensity of the entire cell, background-subtracted and mapped to a log scale (the higher the score, the greater the concentration of intensity inside the cell).

To identify the internal part of the cell, initially, an Object mask was applied. This mask digitally identifies the whole cell by segmentation of images to an area corresponding to the morphology of the cell. Then an Erode mask was applied on top of the Object mask which removes the selected number of pixels from all edges of the starting mask (removes the membrane) thus enabling to detect only the inner part of the cell.

The internalization score difference was calculated by subtracting the calculated internalization at time 0 (background) from the calculated internalization at time points stated for each antibody (i.e. calculated internalization at 1.5 hr after external Ab stripping was subtracted from calculated internalization detected at time 0 after external Ab stripping).

Figure 1.0: Confocal microscopy analysis of candidate antibodies. Five candidate antibodies (Ab-1 – Ab-5) were internalized into MDA-MB231 breast cancer cell line for 1.5hr as detailed in Materials and Methods. Antibodies (green) can be detected as specks with in the cells. Colocalization of the antibodies to late endosomes (LE – Magenta) is depicted by white specks and some are accentuated by arrows. Colocalization to lysosomes (Lyso – Red) is depicted by yellow specks and accentuated by arrows. Cell nuclei are depicted in blue. All pictures were taken at a magnification of x60. Scale bar = 10um. Separate staining and overlays can be seen in the supplement figures 2 and 3. Similar results obtained in at least three independent experiments. Click here to enlarge.

To quantify the colocalization of the ADC with lysosomes / late endosomes, the Bright Detail Similarity (BDS) feature was used for the corresponding channels. This feature calculates the log-transformed Pearson’s correlation coefficient of the localized bright spots with a radius of 3 pixels or less within the masked area and is used to compare the small bright image detail of two images. The coefficient is log-transformed to increase the dynamic range between {0, inf}. This feature will calculate e.g. the amount of colocalization between an antibody and a lysosome, being 0 for not colocalized and 1 for fully colocalized. Yet, it does not consider how much antibody (how many spots) are detected within a cell. Therefore, to further quantify the amount of signal within the cell, which correlates to the amount of antibody internalized into the cell, the Bright Detail Intensity R3 (BDI) was calculated. This feature computes the intensity of localized bright spots that are 3 pixels in radius or less within the masked area in the image, subtracted for local background around the spots. This was calculated on the eroded mask described above, to take into account only staining within the cell. Graphs in this article depict the percent of cells that were high in both BDI and BDS (hence, have a lot of internalized antibody which most is colocalized to the organelle tested) subtracted from the background obtained at time 0.


Choosing the candidate antibody
To develop a therapeutic ADC targeting a membrane receptor (protein X), a naïve antibody phage display library was screened, and clones were selected by conventional flow cytometry (CFC) according to their binding affinity to target X (not shown in this article). These clones were then analyzed for their ability to internalize into cells and colocalize to the lysosomes.

Initially, confocal microscopy was carried out to qualitatively detect internalization and colocalization to lysosomes. Proteins that internalize into lysosomes are quickly degraded, which may result in a reduced ability to detect their fluorescence in this organelle. Therefore, we also analyzed for colocalization to late endosomes (LE), as many of the proteins that reach the LE will eventually be transported to the lysosomes [19][20]. In the example shown in Figure 1.0, all five antibodies (Ab-1 to Ab-5) displayed internalization into the cells as seen by the green fluorescent specks inside the cells (Figure 1.0 and supplementary Figure 2.0 and Figure 3.0). All antibodies also displayed a certain degree of colocalization to early endosomes and lysosomes.

Figure 2.0 – IFC analysis of candidate antibodies. Comparison of internalization and colocalization to internal organelles of selected ADCs in MDA-MB231 cells was carried out by IFC analysis. (A) The graph indicates internalization as calculated by the subtraction of internalization after 1.5hr from internalization at time 0 (+/- SE). Internalization was calculated using the “Internalization feature” of the IDEAS® software. (B) A BDI vs. BDS plot and representing pictures of cells from four quarters: purple – depicting cells where the antibody is not internalized; green – cells containing internalized antibody that is not colocalized to the organelle tested (lysosomes in this case); red – cells containing highly internalized and colocalized antibodies. The white quarter depicts dead cells and artifacts. The percent of gated cells in each quarter is stated. Rab7 – late endosomal marker; LAMP2 – lysosomal marker. (C) A BDI plot of internalized antibody at time 0 (yellow) and after internalization (green), showing how the gate (R2) for the dot plot was set. In this case R2=1000. Click here to enlarge.

To quantify the internalization and colocalization of the different antibodies to the late endosomes and lysosomes in an automated, high-throughput manner we utilized Imaging Flow Cytometry (IFC). Internalization was carried out and at least 10,000 cells were quantified using the internalization feature of the IDEAS® software.

Figure 2.0 A represents the quantification of internalization for each antibody subtracted for the background values of the internalization at time 0 (additional details in Materials and Methods). Of the five candidates, antibodies 1, 4 and 5 showed higher internalization than the other two candidates.

To analyze the comparative amount of transport of the antibodies to the late endosomes and lysosomes, cells were also stained with an antibody that specifically stains these organelles (anti-Rab7 and anti-LAMP2 respectively).

Initially we determined the amount of colocalization by using the Bright Detail Similarity (BDS) feature in the IDEAS® software. BDS calculates the degree of overlap between two fluorescent masks on a pixel by pixel basis. As bright spots in two images can be totally correlated or not correlated at all, the coefficient varies between 0 (uncorrelated) to 1 (totally correlated), mapped to a log scale. However, BDS detects each spot as one entity and the result does not take into account the amount of antibody that internalized into the cell. This means that if an antibody internalized poorly into the cell and is detectable as only one speck that is totally colocalized to a lysosome, the BSD feature will calculate relatively the same result as with a good internalizing antibody, showing many spots in the cell, that all colocalize to the lysosome. When assessing which antibody to choose as the backbone for a potent ADC, it is important to address not only the ability of the antibody to reach the lysosome (manifested by colocalization, as quantified by BDS) but also one that internalizes to a high extent. For this, we needed to employ an additional feature, that would provide an estimation of amount of antibody that internalized into the cell, and plot that together with the colocalization analysis.

The Bright Detail Intensity (BDI R3) feature calculates the intensity of bright spots that are 3 pixels in radius or less within the mask. To enable detection of spots of internalized antibody and not ones on the membrane, BDI was calculated within a mask that contains only the cytoplasm but excludes the membrane (for details see Materials and Methods). The BDI increases as more spots are visible in the cell or when their intensity is higher. As such, it can provide a good estimation of amount of antibody that internalized into a cell.
Plotting BDI to BDS (Figure 2.0 B) creates a dot plot where each quarter represents a different population of cells: high BDI–high BDS (red quarter) represents the cells containing many internalized antibodies with high amount of colocalization. High BDI low BDS (green) are cells that contain a lot of internalized antibody but not colocalized to the lysosomes. Cells in the Low BDI and BSD (purple) quarter exhibit low antibody staining inside the cell, or just on the mask itself (close to the inner side of the membrane). Due to strong fluorescence of membrane-bound antibodies, some background staining is detected by the BDI but these are not colocalized. For our analysis we considered these as ‘outside’ though they may be internalized but too close to the membrane to decipher as internalized antibodies. Low BDI and high BDS (white) are dead cells or artifacts that are stained by both fluorophores and therefore seem to be colocalized.

Figure 3.0 – Colocalization of candidate antibodies to internal organelles. To analyze the extent of colocalization of the candidate antibodies to late endosomes and lysosomes in MDA-MB231 cells, BDI and BDS were calculated using the IDEAS® software and plotted against each other. Percent of cells containing highly internalized and colocalized antibodies (high BDI high BDS) was assessed and subtracted from percent at time 0 (background). Click here to enlarge.

The BDS calculation is such that two fluorophores are considered colocalized if their BDS value is above 1, as explained in the Materials and Methods section and also in Pugsley et al [21][22]. For determination of gating, BDI of time 0 vs. BDI of internalized antibody was plotted and the gate was set as seen in Figure 2.0 C. The R2 value is then set as the gate of the BDI in the dot plot and thus the quarters were obtained.

When comparing all 5 antibodies for colocalization to internal organelles (Figure 3.0) both Ab-4 and Ab-5 showed the highest colocalization to lysosomes. Analyzing the percent gated values (Y axis), we found it striking to see a very low percentage of colocalization after internalization, in comparison to preliminary qualitative data by confocal microscopy. The reason for these low values was found to be due to high background from membrane bound antibody. In all subsequent experiments, this issue was solved by external antibody stripping, which decreased the fluorescent signal from membrane bound antibody and enabled improved intracellular signal analysis.

Taking the internalization and colocalization into account and additional information such as ability of the antibody to kill cells when loaded with a cytotoxic drug (not shown in this article), it was decided to proceed with Ab-5 for conjugation to a linker-payload and for additional evaluations.

Figure 4.0 – Comparison of the chosen candidate antibody to its ADC. To verify that addition of a linker and payload to the chosen antibody did not hinder its internalization or trafficking ability, a comparison was done by IFC (A) Internalization to MDA-MB231 cells after 1hr was calculated using the IDEAS® Internalization feature and was subtracted from the internalization detected at time 0 (background; +/- SE) (B) Colocalization was calculated as percent of cells containing highly internalized and colocalized antibodies (high BDI high BDS) subtracted from percent at time 0 (background). Dot plots of BDI vs. BDS to lysosomes are shown in supplementary Fig 4.0 A. Similar results obtained in two separate experiments and one additional experiment in a different cell line as can be seen in supplementary Figure 4.0 B. Click here to enlarge.

Analysis of the ADC
Conjugation of a linker-payload to an antibody can affect, or even disrupt, internalization and/or localization to different compartments in the cells.

To analyze whether the conjugation of Ab-5 changed its ability to internalize or be transported to the lysosome, IFC analysis was carried out with the naked antibody and its corresponding conjugated counterpart (ADC-5). Analysis showed that internalization of the ADC was slightly impaired in comparison to the naked antibody (Figure 4.0 A).

Colocalization to late endosomes was similar for both antibodies, but there was a big increase in transfer of the ADC to the lysosomes in comparison to the naked antibody (Figure 4.0 B).

Comparison of a novel ADC to similar antibodies binding the same target antigen

When assessing novel therapeutic antibodies to function as ADCs, comparison to other antibodies that bind the same target is of great importance. Specifically, if other companies are developing antibodies to the same target, it is essential to show that the chosen antibody is superior, or at least functions as well as, other antibodies in the field.

Figure 5.0 – Comparison of an in-house antibody to antibodies that bind the same target. An in-house antibody to protein Y was compared by IFC analysis to antibodies that bind the same target. Internalization was carried out for 2hr into A431 epidermoid carcinoma cells (A) Internalization was calculated using the IDEAS® Internalization feature. Background (internalization at time 0) was subtracted. Bars indicate internalization difference (+/- SE) (B) Colocalization was calculated as percent of cells containing highly internalized and colocalized antibodies (high BDI high BDS) subtracted from percent colocalization detected at time 0 (background). Similar results obtained in one additional cell line. Click here to enlarge.

To test this, we compared the internalization and internal organelle localization of an antibody we developed against protein Y, to two antibodies that bind the same target antigen. Figure 5.0 A and Figure 5.0 B show that internalization as well as colocalization to both late endosome and lysosomes is similar for all antibodies.

Decision on the therapeutic antibody format – use of two antibodies or one biparatopic ADC

In certain cases of therapeutic antibody or ADC treatments, it is beneficial to use two antibodies to target one specific protein, or two different antibodies targeting two proteins in the same pathway. These two can be given as separate antibodies, where one or both can be conjugated as an ADC and given in combination as a therapeutic drug. Alternatively, they could be combined in one antibody or ADC, where each arm binds a different protein (a bispecific) or each arm is directed to a different epitope in the same protein (a biparatopic).[23][24]

To evaluate the potential of a biparatopic ADC, we compared its ability to internalize and colocalize to lysosomes to its two single IgG counterparts. Figure 6A (and supplementary Figure 5) clearly shows by confocal microscopy, the superiority of internalization and colocalization of the biparatopic antibody in comparison to the two single IgGs.

Figure 6.0 – Antibody format for improved internalization and delivery to the lysosome. Superiority of internalization of a biparatopic antibody over its two IgG counterparts detected by confocal microscopy and IFC analysis. (A) Two IgGs and one biparatopic antibody were internalized for 5hr into MDA-MB468 breast cancer cell line as detailed in Materials and Methods. Antibodies (green) can be detected as specks in the cells. Colocalization to lysosomes (red) is depicted by yellow specks and accentuated by arrows. All pictures were taken at a magnification of x60. Separate staining and overlays can be seen in the supplement figure 5. (B) Internalization after 5hr was calculated using the IDEAS® internalization feature and was subtracted from the internalization detected at time 0 (background; +/- SE) (C) Colocalization was calculated as percent of cells containing highly internalized and colocalized antibodies (high BDI high BDS) subtracted from percent at time 0 (background). Similar trends seen in two additional experiments. Click here to enlarge.

After 1.5 hours of internalization, nearly all the biparatopic antibody was internalized, whereas most of the two IgGs were still detected on the cell membrane. Though internalization seems relatively high for Anti-A and Anti-B when observing single cells by confocal microscopy, examination revealed that both antibodies internalize only into to part of the population, as depicted in e.g. Figure 6.0 A middle panel, whereas the biparatopic antibody internalized to all the cells. Additionally, massive colocalization was detected for the biparatopic antibody as depicted by large yellow spots in figure 6.0 A. IFC analysis on a large population of cells (30,000) reinforced the confocal results, showing that internalization of the biparatopic antibody is slightly improved in comparison to Anti-A and highly improved in comparison to Anti-B (Figure 6.0 B) and colocalization to lysosomes is also greatly enhanced (Figure 6.0 C).

When a therapeutic agent to be developed is an ADC, it is essential to select for an antibody that will internalize efficiently into cells and release its toxic payload within them. Such an antibody, when linked to a toxin, will ideally have low off-target toxicity, as the drug will function to kill the cell from within and not spread the toxic payload in the circulation or surrounding environment of the tumor it is targeting. As most ADCs to date utilize linkers that require cleavage or degradation within the lysosome, [4] a selection method that will not only test for binding and internalization, but also quantify the ability of the antibodies to reach this organelle, is highly advantageous.

To date, CFC and confocal microscopy are being used for the selection of antibodies for ADCs. While CFC enables to evaluate internalization rate, it does not have the ability to assess antibody localization within the cell. In comparison, confocal microscopy provides unequivocal detection of localization of the tested antibody in the cell, yet it is not the best tool for quantification of the fluorescent signal, as analysis is time consuming and may not be accurate when digital segmentation of cells is required. IFC, coupled with high throughput, unbiased quantitative analysis, makes it ideal for such a test, as it enables to analyze a population of tens of thousands of cells within a few minutes, in a uniform and controlled manner. The benefit of analysis on a large population can be seen clearly in the comparison of the internalization of the biparatopic antibody to its two counterparts (Figure 6). As IFC considers not only the amount of internalization per cell but on the whole population, low internalization may be obtained due to a poorly internalizing antibody to all cells or an antibody that internalizes only to few cells of a population. In the case of Anti-B in Figure 6A, one cell displays a high amount of internalization while other cells show low or no internalization at all, and IFC detected lower internalization of this antibody in comparison to the biparatopic antibody which internalized massively to all the cells.

The IDEAS® analysis IFC software contains functions that support analysis and quantification, yet application of only one function to analyze the data does not always deliver the full analysis required. As has been shown by both Rajan and Pugsley [21][25] when quantifying autophagosomes and autophagosome flux, using only the BDS function, that evaluates colocalization of two fluorophores to each other, does not take into account the number of spots within the cell, thus rendering the analysis lacking. In her detailed analysis [25] Pugsley shows several methods that can be used to analyze the same data. But only the application of a combination of functions: BDS and spot count in her example, enabled to obtain a more appropriate analysis. Similarly, we show that use of a combination of functions, BDI and BDS in our case, enabled to quantify the amount of colocalization of the tested antibodies to lysosomes and took into account also the amount of total internalization of the antibodies into the cell. This is a crucial factor in ADC analysis, as only the combination of a highly internalizing antibody that is trafficked efficiently to the lysosome will yield a good base for a potent ADC.

By initially qualitatively screening of antibodies by confocal microscopy after-which antibodies are quantitively analyzed by IFC, we can select an antibody with high internalizing and trafficking capabilities. Though this is not the only selection that the antibodies undergo, in the choice of becoming a potent ADC, it enables to sieve out, in an efficient manner, antibodies that do not internalize or traffic poorly to the lysosomes, and to focus on the ones with most potential. To our knowledge, no report detailing the use of IFC and confocal microscopy for selection of antibodies has been published to date.

As IFC is a highly efficacious method for comparison analysis, it is a great tool for additional analyses that are required during ADC development, as shown here: comparison between the chosen antibody and other antibodies that bind the same target antigen or for defining the final format to be used as the ADC drug (i.e. IgG or biparatopic).

As many of the toxins used in ADCs are hydrophobic, their conjugation to the antibody may change the biophysical properties of the ADC in such a way that can hamper internalization and/or trafficking [26][27][28]. The analysis shown here, enables to detect such cases early in the development stage and change or re-model the ADC to restore its activity. Analysis can be carried out to evaluate changes in internalization and trafficking capabilities of the same antibody linked to different linkers and/or payloads, or to compare the same linker-payload that is conjugated to different FC variants. These types of analyses, run over time on many different antibodies and ADCs, create a wealth of knowledge and broader understanding of ADC technologies and may help in the development of more efficacious drugs for the benefit of patients in the future.

Supplemental Figures
Supplemental Figure 1.0 – Gating strategy for IFC analysis. [Figure]
Supplemental Figure 2.0 – Colocalization of candidate antibodies to late endosomes – confocal analysis. [Figure]
Supplemental Figure 3.0 – Confocal and IFC Analysis for ADC Selection [Figure]
Supplemental Figure 4.0 – Confocal and IFC Analysis for ADC Selection [Figure]
Supplemental Figure 5.0 – Confocal and IFC Analysis for ADC Selection [Figure]

[1] Senter PD, Sievers EL. The discovery and development of brentuximab vedotin for use in relapsed Hodgkin lymphoma and systemic anaplastic large cell lymphoma. Nat. Biotechnol. 2012;30:631–637. Available at: Accessed December 4, 2018.
[2] Carter PJ, Senter PD. Antibody-Drug Conjugates for Cancer Therapy. Cancer J. 2008;14:154–169. Available at: Accessed December 4, 2018.
[3] Thomas A, Teicher BA, Hassan R. Antibody–drug conjugates for cancer therapy. Lancet Oncol. 2016;17:e254–e262. Available at: Accessed December 4, 2018.
[4] Diamantis N, Banerji U. Antibody-drug conjugates—an emerging class of cancer treatment. Br. J. Cancer 2016;114:362–367. Available at: Accessed December 19, 2018.
[5] Venkatesan P. Trastuzumab emtansine for HER2-positive breast cancer. Lancet. Oncol. 2016;17:e528. Available at: Accessed December 4, 2018.
[6] Katz J, Janik JE, Younes A. Brentuximab Vedotin (SGN-35). Clin. Cancer Res. 2011;17:6428–36. Available at: Accessed December 19, 2018.
[7] Polakis P. Antibody Drug Conjugates for Cancer Therapy. Pharmacol. Rev. 2015;68:3–19. Available at: Accessed December 4, 2018.
[8] Ducry L, Stump B. Antibody−Drug Conjugates: Linking Cytotoxic Payloads to Monoclonal Antibodies. Bioconjug. Chem. 2010;21:5–13. Available at: Accessed December 4, 2018.
[9] Doronina SO, Toki BE, Torgov MY, Mendelsohn BA, Cerveny CG, Chace DF, DeBlanc RL, Gearing RP, Bovee TD, Siegall CB, Francisco JA, Wahl AF, Meyer DL, Senter PD. Development of potent monoclonal antibody auristatin conjugates for cancer therapy. Nat. Biotechnol. 2003;21:778–784. Available at: Accessed December 4, 2018.
[10] Dubowchik GM, Mosure K, Knipe JO, Firestone RA. Cathepsin B-sensitive dipeptide prodrugs. 2. Models of anticancer drugs paclitaxel (Taxol), mitomycin C and doxorubicin. Bioorg. Med. Chem. Lett. 1998;8:3347–52. Available at: Accessed December 4, 2018.
[11] Erickson HK, Park PU, Widdison WC, Kovtun Y V., Garrett LM, Hoffman K, Lutz RJ, Goldmacher VS, Blättler WA. Antibody-Maytansinoid Conjugates Are Activated in Targeted Cancer Cells by Lysosomal Degradation and Linker-Dependent Intracellular Processing. Cancer Res. 2006;66:4426–4433. Available at: Accessed December 4, 2018.
[12] Erickson HK, Widdison WC, Mayo MF, Whiteman K, Audette C, Wilhelm SD, Singh R. Tumor Delivery and In Vivo Processing of Disulfide-Linked and Thioether-Linked Antibody−Maytansinoid Conjugates. Bioconjug. Chem. 2010;21:84–92. Available at: Accessed December 4, 2018.
[13] Jordan MA, Wilson L. Microtubules as a target for anticancer drugs. Nat. Rev. Cancer 2004;4:253–265. Available at: Accessed December 4, 2018.
[14] Adrian L. Smith* †,‡ and, K. C. Nicolaou† §. The Enediyne Antibiotics. 1996. Available at: Accessed December 4, 2018.
[15] Presta LG. Selection, design, and engineering of therapeutic antibodies. J. Allergy Clin. Immunol. 2005;116:731–6; quiz 737. Available at: Accessed December 19, 2018.
[16] Tsuruta LR, dos ML, Moro AM. Display Technologies for the Selection of Monoclonal Antibodies for Clinical Use. In: Antib. Eng. InTech; 2018. Available at: Accessed December 19, 2018.
[17] Harper J, Mao S, Strout P, Kamal A. Selecting an Optimal Antibody for Antibody–Drug Conjugate Therapy: Internalization and Intracellular Localization. In: Methods Mol. Biol. Vol 1045. 2013. p 41–49. Available at: Accessed December 4, 2018.
[18] Han Y, Gu Y, Zhang AC, Lo Y-H. Review: imaging technologies for flow cytometry. Lab Chip 2016;16:4639–4647. Available at: Accessed December 4, 2018.
[19] Huotari J, Helenius A. Endosome maturation. EMBO J. 2011;30:3481–500. Available at: Accessed December 22, 2018.
[20] Ritchie M, Tchistiakova L, Scott N. Implications of receptor-mediated endocytosis and intracellular trafficking dynamics in the development of antibody drug conjugates. MAbs 2013;5:13–21. Available at: Accessed December 22, 2018.
[21] Rajan R, Karbowniczek M, Pugsley HR, Sabnani MK, Astrinidis A, La-Beck NM. Quantifying autophagosomes and autolysosomes in cells using imaging flow cytometry. Cytom. Part A 2015;87:451–458. Available at: Accessed December 20, 2018.
[22] Pugsley HR. Assessing Autophagic Flux by Measuring LC3, p62, and LAMP1 Co-localization Using Multispectral Imaging Flow Cytometry. J. Vis. Exp. 2017. Available at: Accessed December 4, 2018.
[23] Comer F, Gao C, Coats S. Bispecific and Biparatopic Antibody Drug Conjugates. In: Humana Press, Cham; 2018. p 267–280. Available at: Accessed December 22, 2018.
[24] Krishnamurthy A, Jimeno A. Bispecific antibodies for cancer therapy: A review. Pharmacol. Ther. 2018;185:122–134. Available at: Accessed December 22, 2018.
[25] Pugsley HR. Quantifying autophagy: Measuring LC3 puncta and autolysosome formation in cells using multispectral imaging flow cytometry. Methods 2017;112:147–156. Available at: Accessed December 4, 2018.
[26] Källsten M, Hartmann R, Artemenko K, Lind SB, Lehmann F, Bergquist J. Qualitative analysis of antibody-drug conjugates (ADCs): an experimental comparison of analytical techniques of cysteine-linked ADCs. Analyst 2018;143:5487–5496. Available at: Accessed December 20, 2018.
[27] Hollander I, Kunz A, Hamann PR. Selection of Reaction Additives Used in the Preparation of Monomeric Antibody−Calicheamicin Conjugates. Bioconjug. Chem. 2008;19:358–361. Available at: Accessed December 20, 2018.
[28] Burke PJ, Senter PD, Meyer DW, Miyamoto JB, Anderson M, Toki BE, Manikumar G, Wani MC, Kroll DJ, Jeffrey SC. Design, Synthesis, and Biological Evaluation of Antibody−Drug Conjugates Comprised of Potent Camptothecin Analogues. Bioconjug. Chem. 2009;20:1242–1250. Available at: Accessed December 20, 2018