Homogeneous plate based antibody internalization assay using pH sensor ﬂ uorescent dye

Receptor-mediated antibody internalization is a key mechanism underlying several anti-cancer antibody therapeutics. Delivering highly toxic drugs to cancer cells, as in the case of antibody drug conjugates (ADCs), ef ﬁ cient removal of surface receptors from cancer cells and changing the pharmacokinetics pro ﬁ le of the antibody drugs are some of key ways that internalization impacts the therapeutic ef ﬁ cacy of the antibodies. Over the years, several techniques have been used to study antibody internalization including radiolabels, ﬂ uorescent microscopy, ﬂ ow cytometry and cellular toxicity assays. While these methods allow analysis of internalization, they have limitations including a multistep process and limited throughput and are generally endpoint assays. Here, we present a new homogeneous method that enables time and concentration dependent measurements of antibody internalization. The method uses a new hydrophilic and bright pH sensor dye (pHAb dye), which is not ﬂ uorescent at neutral pH but becomes highly ﬂ uorescent at acidic pH. For receptor mediated antibody internalization studies, antibodies against receptors are conjugated with the pHAb dye and incubated with the cells expressing the receptors. Upon binding to the receptor, the dyes conjugated to the antibody are not ﬂ uorescent because of the neutral pH of the media, but upon internalization and traf ﬁ cking into endosomal and lysosomal vesicles the pH drops and dyes become ﬂ uorescent. The enabling attributes of the pHAb dyes are the hydrophilic nature to minimize antibody aggregation and bright ﬂ uorescence at acidic pH which allows developmentofsimpleplatebasedassaysusinga ﬂ uorescentreader.Usingtwodifferenttherapeuticantibodies – Trastuzumab (anti-HER2) and Cetuximab (anti-EGFR) – we show labeling with pHAb dye using amine and thiol chemistries and impact of chemistry and dye to antibody ration on internalization. We ﬁ nally present two new approachesusingthepHAbdye,whichwillbebene ﬁ cialforscreeningalargenumberofantibodysamplesduring early monoclonal development phase. © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).


Introduction
Receptor-mediated antibody internalization plays an important role in efficacy and dosage of therapeutic antibodies.For example, efficacy of antibody drug conjugates (ADCs) (Casi and Neri, 2012;Drachman and Senter, 2013;Panowksi et al., 2014;Ritchie et al., 2013) is to a large extent driven by the efficiency of antibody internalization followed by trafficking into the lysosomal compartment and release of the toxic drug molecule leading to cell death (Flygare et al., 2013).More recently, it has been reported that in the case of epidermal growth factor receptor (EGFR) induced cancers, a superior response is observed if the EGFR is efficiently removed from the surface by inducing receptor internalization and degradation (Pedersen et al., 2010;Ben-Kasus et al., 2009;Friedman et al., 2005;Kontermann, 2012).In this case, improved efficacy of the treatment was observed when a mixture of antibodies targeting different epitopes of EGFR led to enhanced receptor internalization.This approach is different from the current approach of targeting EGFR with a single antibody like Cetuximab or Panitumumab, which works mainly by inhibiting EGF binding and receptor dimerization and inducing ADCC (Antibody Dependent Cell Cytotoxicity).While receptor-mediated antibody internalization improves the efficacy of antibody therapeutics, it can also have an opposite effect because internalization can lead to rapid clearance of the drug in vivo and hence may require high antibody dosage or frequent dosing (Chaparro-Riggers et al., 2012;Amano et al., 2010;Lammerts van Bueren et al., 2006).
In spite of increasing awareness that internalization is a critical antibody attribute, antibodies continue to be screened only for affinity binding using Enzyme Linked Immunosorbent Assay (ELISA) during early monoclonal development phase.This approach may not be optimal since there is some evidence that high affinity may not always correlate with good internalization (Lyon et al., 2012).Hence there is a need for high throughput methods to screen antibodies specifically for internalization properties.
Several methods have been used to study receptor-mediated antibody internalization (Casalini et al., 1993;Isa et al., 2014;Liao-Chan et al., 2015;Hazin et al., 2015).Confocal imaging using fluorescence labeled antibodies is perhaps the most frequently employed and has the ability to localize antibodies upon internalization into specific subcellular compartments like the endosome and lysosome.The imaging based methods are low throughput and offer mostly qualitative data.For quantitative data, antibodies labeled with radioisotopes or fluorescent molecules are incubated with the cells and allowed to internalize.After removing membrane associated antibodies using a low pH (pH ~2.5) wash, the internalized antibodies are detected by measuring radioactivity or fluorescence intensity of the cells.The acid wash methods are endpoint assays, have deleterious effects on the cells and have low signal over background ratios if the washing is not efficient.A variation of this approach uses a secondary antibody targeted against the fluorescent tag to quench the signal on the cell membrane (Liao-Chan et al., 2015) and have the advantage that it avoids the cell apoptosis resulting from low pH wash.The method is an endpoint assay with multiple incubation and washing steps.Homogeneous internalization assays have been reported including one using a dual labeled antibody (Gong and Urlacher, 2015) and also using antibodies labeled with pH sensitive dyes (Berguig et al., 2012;Ogawa et al., 2010;Gostring et al., 2010;Nordberg et al., 2008).The pH sensitive dyes are not fluorescent at neutral pH but become fluorescent when they internalize and traffic into low pH vesicles like endosome and lysosome.However, these dyes have not been routinely used for antibody screening or monitoring kinetics of internalization probably due to a relatively small increase in fluorescence on pH change (Isa et al., 2014;Liao-Chan et al., 2015).All of the methods described above are also limited in their throughput or lack sufficient sensitivity to be used as a screening tool (see note at the end of the Conclusion section).
In this paper, we describe a method to study antibody internalization, which is sensitive, specific and can be used in a simple 96-well plate based format for screening purposes.The method relies on a new pH sensor dye (pHAb dye) comprising a rhodamine-class core structure and two substituted piperazine groups (Fig. 1).The dye shows very low fluorescence at pH ≥ 7.5 and a dramatic increase in the fluorescence at pH b 6.The dye has excitation and emission maximum of 532 nm and 560 nm, is photostable in solution and displays no loss of signal with multiple cycles of acid, base titration.
We have recently reported the use of the dye to detect ligand mediated receptor internalization (Robers et al., 2015).For antibody internalization studies, two reactive versions of the dyean amine and a thiol reactive formwere made.The amine reactive dye has a succinimidyl ester group for binding to a lysine amino acid and the thiol reactive version has a maleimide reactive group for binding to reduced cysteine in the antibody hinge region.To make the dyes compatible for protein conjugation, they were also sulfonated, which makes them hydrophilic and less prone to precipitating the antibody.Antibodies labeled with pHAb dyes upon binding to receptors on the cell membrane will not be fluorescent due to the neutral pH of the media.However, upon receptor-mediated internalization, antibodies will traffic into early endosomes and lysosomes where the pH is acidic and the dye will fluoresce allowing visualization of the dye labeled antibodies.A significant increase in fluorescence of pHAb dyes in an acidic environment combined with the hydrophilic nature of the dye makes them uniquely suitable for monitoring antibody internalization.
To demonstrate the broad utility of the pHAb dye for receptormediated antibody internalization, we successfully used it to develop an internalization assay for two therapeutic antibodies, Trastuzumab and Cetuximab, which bind to HER2 and EGFR respectively.Both the antibodies, which are known to internalize (Panowksi et al., 2014;Pedersen et al., 2010), were labeled with pHAb dyes using amine or thiol chemistry and used for time and concentration dependent internalization studies and read on a 96-well fluorescent plate reader.The assay is homogeneous and internalization can be monitored in real time without removing excess antibody.To further simplify the assay to make it suitable for screening applications we show that: (a) antibodies can be labeled with pHAb dyes and used without removing excess dye and (b) secondary antibodies labeled with pHAb can be used instead of labeling primary antibodies.

pHAb conjugation to antibody
Antibodies were conjugated with pHAb dye either using amine chemistry or thiol chemistry according to a manufacturer (Promega) recommended protocol.Briefly, for conjugating antibody with pHAb dye at lysine amino acids, antibody is reacted with an excess of amine reactive pHAb dye (5-20 molar excess) for 1 h.Free dye was subsequently removed using a Zeba desalting column (Thermo-Scientific) and the concentration of dye-labeled antibody and dye to antibody ratio (DAR) was calculated as recommended by the manufacturer.For labeling antibody with thiol reactive pHAb dye, disulfide bonds present in the antibody hinge region were first reduced to free thiols by incubating for 1 h with 2.5 mM DTT. Excess DTT is removed using a Zeba desalting column followed by a 1 h reaction with excess of thiol reactive pHAb dye (5-20 M excess).Excess dye was removed as before and the concentration of pHAb labeled antibody as well as DAR was calculated according to the manufacturer's protocol (Promega).

Cell based ELISA
SKBR3 cells overexpressing HER2 and A431 cells overexpressing EGFR were plated at 15,000 cells per well in a 96-well polystyrene plate and grown to confluence by overnight incubation at 37 °C, then fixed using 4% paraformaldehyde.Dilution series of Trastuzumab and Cetuximab labeled with pHAb dye were made in PBS containing 10 mg/ml BSA (PBSB).50 μl samples were added to the respective plates in triplicate and incubated for 1 h.After washing with PBS containing 0.05% Tween 20 (PBST), plates were incubated for 1 h with anti-Human-IgG (H + L)-HRP (horseradish peroxidase) conjugate (Promega) diluted 1:5000-fold in PBSB.After washing with PBST, TMB (3,3´, 5,5´-tetramethylbenzidine) (Promega) was used as the HRP substrate.The colorimetric reaction was stopped by adding 1N HCl and plates were read at 450 nm.ELISA for anti-EGFR mouse antibodies was performed using A431 cells as above except that anti-mouse-IgG-HRP conjugate (Novex) was used at 1:5000 dilution.

Confocal microscopy for antibody internalization
HER2 positive SKBR3 cells were used to study internalization of pHAb conjugated Trastuzumab.MCF7 cells with no HER2 on their cell membrane were used as a negative control.SKBR3 cells were grown in McCoy's media with 10% fetal bovine serum and MCF7 cells were grown in Eagle's minimal essential media (MEM) with 10% fetal bovine serum and 0.01 mg/ml human recombinant insulin.Cells were plated at 40 K per well in a Lab Tek II chambered #1.5 German coverglass system (Nunc) overnight.Media was changed to OptiMEM/4% fetal bovine serum and cells were treated with 3 nM Trastuzumab-pHAb.Images were captured on a Nikon C2si confocal microscope (Ex/Em: 561/ 580LP nm) every hour for 16 h.The 561 nm diode laser was set at 10%, PMT = 130 V and confocal aperture was 40 μM.To visualize antibody-pHAb receptor binding, cells were treated with 3.0 nM Trastuzumab-pHAb for 1 h at 4 °C.Media was removed, washed and 200 μl of OptiMEM/FBS was added.Buffer pH was adjusted by adding 50 μl of 0.1 M Citrate buffer, pH 5.0 or 0.1 M phosphate buffer, pH 8.0.

Plate based internalization assay
Three different cell lines were used throughout these experiments: (1) SKBR3 cells overexpressing HER2, for internalization of pHAb labeled Trastuzumab; (2) A431 overexpressing EGFR for internalization of pHAb labeled Cetuximab; and (3) MCF7 cells without HER2 or EGFR on their cell membrane as negative control.SKBR3 cells were grown in McCoy's media with 10% fetal bovine serum, A431 cells were grown in DMEM with 10% fetal bovine serum and MCF7 cells using Eagle's minimal essential media (MEM) with 10% fetal bovine serum and 0.01 mg/ ml human recombinant insulin.Cells were grown in T75 cm 2 flask overnight till they reached confluence.Cells were trypsinized and plated in a 96-well black, clear-bottom plate (Costar 3603) or black plate (Thermo Scientific #165305) at the density of 20 K per 90 μl per well.A clear bottom plate has the advantage that it allows visualization of antibody internalization using a confocal microscope or EVOS cell imaging system (Life Technologies).Plates were incubated for 20-24 h before treatment with pHAb labeled antibodies.
For internalization, pHAb conjugated antibodies were added to the cells at the desired concentration and mixed gently for 1-2 min on a plate mixer and then incubated overnight to allow internalization (internalization can be detected in a few hours).Plates are read on a fluorescent plate reader at Ex/Em: 532 nm/560 nm on a Tecan Infinity M1000 Pro.To achieve higher sensitivity, media is replaced by PBS before reading the plate.Control wells are treated just with PBS and fluorescence signal is used as background coming just from the cells and are subtracted from positive wells.

Characterization of pHAb labeled antibodies
Functionality of the labeled antibody is dependent on the chemistry used for labeling and average number of labels per antibody (Acchione et al., 2012;Shrestha et al., 2012;Vira et al., 2010).Hence, Trastuzumab and Cetuximab were labeled with pHAb dyes using both amine and thiol chemistries, in the presence of 5, 10 and 20 M excess of dye to get a range of dye to antibody ratios (DARs) (Table 1).DAR of 2.7 to 9 was obtained with amine chemistry whereas for thiol based labeling method, DAR had a narrower range from 1.7-3.9.DAR of 3.5 and 2.8 for Trastuzumab and Cetuximab respectively was common to both the chemistries and allowed us to evaluate the impact of dye position on antigen-antibody binding and internalization.pHAb labeled antibody preparations (n = 12) were evaluated for antigen-antibody binding activity using ELISA, and for fluorescence response to pH change.For clarity, we used Trastuzumab-amine-pHAb and Cetuximab-amine-pHAb to indicate antibodies conjugated using amine chemistry; and for thiol chemistry, we used Trastuzumab-thiol-pHAb and Cetuximabthiol-pHAb.
ELISA results show that the most significant impact on antibody-antigen binding happened with Trastuzumab-amine-pHAb (Fig. 2A).A steady decrease in binding activity of Trastuzumab, as indicated by the rightward shift of the binding curve, was seen with increase in DAR.However, when pHAb was conjugated to Trastuzumab using thiol chemistry (Trastuzumab-thiol-pHAb) the impact of antigen-antibody binding was minimal (Fig. 2B).This was true even when DARs were the same for two chemistries indicating that amine chemistry rather than DAR is interfering with antibody-antigen binding.For Cetuximab, pHAb labeling using either amine or thiol chemistry had only a slight impact on binding to EGFR, even at a high DAR of 8.5 (Fig. 2C and D).
Our results indicate that both the labeling chemistry and antibody characteristics are important considerations for antibody labeling.
Thiol chemistry places the dyes in the hinge region of the antibodies, away from the antigen binding site hence minimizing any loss of antibody-antigen activity.On the other hand, labeling antibodies with pHAb dye using amine chemistry can place dyes randomly throughout the antibody surface including near CDR region, which will have significant influence on antibody-antigen binding.Our results also point to the difficulty in deciding a priori which chemistry and DAR will work best for a specific antibody.Although not seen here, we occasionally observe some precipitation of labeled antibodies, which is antibody specific and may represent an antibody population that is heavily labeled.In such cases, a brief centrifugation is used to remove aggregated antibodies and the rest of the sample is very stable.
In addition to the impact of pHAb dye on antibody-antigen binding, the pH response of the pHAb dye after conjugation to antibodies is very important.Trastuzumab-amine-pHAb and Trastuzumab-thiol-pHAb along with the respective free dyes were diluted into buffers of different pH and fluorescence was measured (Fig. 3).For Trastuzumab-amine-pHAb, the increase in fluorescence with decreasing pH was similar to that of free dye (pHAb-SE) even at a high DAR (Fig. 3A) and had a half maximal response at around pH 6.5.However, for Trastuzumab-thiol-pHAb the fluorescence response to pH change was different than that of free dye with half maximal response dropping to pH 6.0 (Fig. 3B).Moreover, for thiol chemistry there still was some fluorescence (~10% of the maximum signal at pH-4.0) at pH ~8.0 which resulted in fold increase in fluorescence between pH 8.0 and pH 4.0 to be lower for Trastuzumab-thiol-pHAb (10 fold) compared to that for free dyes or Trastuzumab-amine-pHAb (~50 fold).A possible reason for unusual response with thiol chemistry may be that pHAb dyes are restricted into a very tiny space in the antibody hinge region causing local pH to  change or to have some quenching effect.Similar results were seen for Cetuximab (not shown).Our results indicate that pHAb conjugated to antibodies maintains its pH sensing properties and can be used for internalization studies.
Moving forward, we decided to focus on four different antibody-pHAb conjugates: Trastuzumab-amine-pHAb (DAR 3.5), Trastuzumabthiol-pHAb (DAR 3.4); Cetuximab-amine-pHAb (DAR 2.8) and Cetuximab-THIOL-pHAb (DAR 2.8).Our decision was based on the results from ELISA test and also on the fact that DARs between 2 and 4 have been reported (Vira et al., 2010) to be the best for labeling antibodies and two FDA approved ADCs have a DAR of ~3.5 (Flygare et al., 2013).

pHAb conjugated antibodies enable real-time and homogeneous internalization assay
To demonstrate the use of pHAb labeled antibodies for internalization studies, we selected Trastuzumab-amine-pHAb (DAR 3.4) and monitored its internalization in HER2 overexpressing SKBR3 cells using confocal microscopy.3.0 nM of Trastuzumab-amine-pHAb (DAR 3.4) was incubated with SKBR3 cells at 4 °C for 1 h to allow binding but prevent any internalization.Cell media containing excess antibody was replaced with phosphate buffer (pH 8.0) and imaged (Fig. 4A).Since dye is not fluorescent at pH 8.0, the cells are not visible.However to make sure antibody is indeed bound to the receptors on the cell membrane, the buffer was replaced with citrate buffer (pH 5.0) to make the dye fluorescent and upon imaging (Fig. 4B) fluorescence on the cell membrane was clearly visible.Buffer was subsequently replaced with DMEM media and cells were incubated at 37 °C for 24 h to initiate internalization.Upon internalization, Trastuzumab-amine-pHAb trafficked into acidic vesicles like endosome and lysosomes and pHAb dye became fluorescent and was visible as punctate structures (Fig. 4C).Binding and internalization is specific since no fluorescence was seen when MCF7 cells (HER2 negative) were used instead of SKBR3 cells (Fig. 4D).To further ensure specificity, we incubated a pHAb labeled nonspecific antibody, which should not bind to HER2 with SKBR3 cells, and saw no binding (result not shown).
Because fluorescence is visible only when antibody is internalized, we reasoned that it should be possible to monitor internalization in real time without even removing the excess unbound antibody.To the best of our knowledge, running internalization in real time in the presence of excess antibody hasn't been yet reported so far.Trastuzumabamine-pHAb (3.0 nM) was added to the cells and cells were imaged every hour for 19 h (Fig. 5 shows the presence of large number of punctate structures (Fig. 6).An interesting observation was the differences between Figs. 4C and 6, which should be due to differences in internalization in the absence and presence of excess antibody respectively and may be worth pursuing in later studies.

Time and concentration dependent internalization of pHAb conjugated antibodies in 96-well plate format
The ability to study real-time internalization kinetics of antibodies will be very important in selecting antibodies with maximum therapeutic efficacy.However, due to limited throughput and qualitative data it is not ideal as an antibody screening tool.We reasoned that due to the dramatic increase in fluorescence intensity of pHAb dyes at low pH, homogeneous internalization assays could be done in a 96-well plate format using a fluorescent plate reader.To test the concept, 30 nM of Trastuzumab and Cetuximab labeled with pHAb dye (using both thiol and amine chemistries) were added at different time points to SKBR3 and A431 cells, respectively, and incubated at 37 °C.Media was replaced with phosphate buffer (pH 7.2) and fluorescence signal was measured (Fig. 7).Internalization for all four antibodies could be detected within 1 h and fluorescence signal increased over time and was not saturated even after 24 h.Internalization was specific as no signal was seen when these antibodies were incubated with MCF7 cells that do not express either EGFR or HER2 receptors.Absolute fluorescence signal at 24 h was lower for Trastuzumab-amine-pHAb antibodies compared to Trastuzumab-thiol-pHAb and correlates with lower binding affinity observed using ELISA (Fig. 2).Lower fluorescence signal upon internalization was also observed for Cetuximab-amine-pHAb even though  ELISA results showed no major difference in affinity between amine and thiol conjugated antibodies.
In addition to providing kinetic information, it is desirable to have a sensitive internalization assay.A sensitive assay is important for identifying internalizing antibodies during early antibody development phase where the amount of antibody available is small and antibody may have low affinities.To determine the sensitivity of the plate-based assay, a dilution series of Trastuzumab and Cetuximab labeled with pHAb dye (using both thiol and amine chemistries) were added to SKBR3 and A431 cells respectively and incubated at 37 °C for 24 h (Fig. 8).Assays were sensitive and we were easily able to detect internalization of 0.3 nM (~30.0 pg/ml) of both Cetuximab and Trastuzumab.Sensitivity obtained using a plate-based method was similar to that reported using fluorescent imaging and cell toxicity based methods (Pedersen et al., 2010;Klussman et al., 2004;Lewis Phillips et al., 2008).No fluorescence was detected when Trastuzumab-pHAb dye was added to MCF7 cells containing no HER2 or EGFR receptors indicating the specificity of internalization.It is also worth noting that absolute fluorescence signal from internalization was lower for amine-conjugated antibodies.

Performance of antibody-pHAb dye conjugate without removing excess free dye
Results so far, clearly show the utility of pHAb labeled antibodies for internalization studies.However, step involving removal of free dye after labeling is a major roadblock in cases where a large number of samples is available and the amount of sample is limited.It is possible to use 96 well dialysis setups that are commercially available but to really simplify the labeling we wanted to know if removal of excess free dye after conjugation is even necessary.We looked at the internalization of Cetuximab labeled with pHAb dye with or without dialysis step to remove excess dye (Fig. 9).Internalization profile is similar for dialyzed and un-dialyzed sample and is due to the fact that free excess dye is not fluorescent in the media and does not get internalized into the cells indicating that dialysis step to remove excess dye is not necessary.This approach may be difficult to implement with thiol based labeling chemistry since DTT used for reducing disulfide bonds will still have to dialyzed otherwise it will interfere with the maleimide chemistry.Replacing DTT with TCEP may be an option since TCEP doesn't react readily with maleimide (Getz et al., 1999) but was not tested in this report.

Use of pHAb dye labeled secondary antibodies for screening antibodies from cell media
Data presented so far have highlighted the advantages of using pHAb dyes for time and concentration dependent antibodies internalization studies relevant during designing of therapeutic drug.These studies used purified antibodies and hence will have limited utility as screening tool in early stages of antibody screening where antibodies are present in the cell media.We recently reported an on-bead antibody labeling and purification method (Nath et al., 2015) that uses magnetic affinity beads to capture antibodies from the cell media followed by antibody labeling and purification.However, the method still involves multiple steps of washing and incubation.We therefore investigated whether pHAb labeled secondary antibodies could be used instead of labeling primary antibodies.The advantage for screening will be that pHAb  labeled secondary antibodies can be added to the media expressing antibodies and then directly used for internalization.
25 nM of Goat anti-Human IgG labeled with pHAb dye using amine chemistry (DAR = 2.5) was first mixed with 1.0 nM and 10 nM of Cetuximab and added to the A431 cells and read after 24 h incubation (Fig. 10).Internalization of both 1.0 nM and 10.0 nM antibody concentrations were detected however absolute fluorescent signal as well as signal over background were lower compared to the method where Cetuximab was directly labeled with pHAb dyes.We noticed that background was slightly higher with 25 nM of secondary antibody labeled pHAb dye and fluorescence signal was low.We attribute low fluorescence signal to possible formation of large complexes, when multiple secondary antibodies bind to the primary antibody (Harlow and Lane, 1998), resulting in inefficient internalization.In spite of this limitation the approach may be useful in primary screening method of antibodies from cell media samples where antibody concentrations are typically in the range of 10-50 μg/ml (~100-500 nM) and well within the sensitivity of the assay.
To test the applicability of the assay for screening of antibodies, a panel of commercial anti EGFR mouse monoclonal antibodies (Table 2) was screened for internalization using anti-mouse conjugated with pHAb dye using amine chemistry (Fig. 11).In parallel, cell based ELISA was used to evaluate the binding of the antibodies to the EGFR (Fig. 12).ELISA data indicates that antibodies raised using full length EGFR from A431 are the best binders which is in agreement with many published reports (Dreyer et al., 2010;Lim et al., 2011;Butler et al., 1997) that antibodies screened against folded epitopes have a better chance of binding a fully folded functional protein.However, of the antibodies that bind to the EGFR, only four (Abcam 30, LifeBio -LS-C87999 & C88141, and MS268) showed significant internalization.Our observation is in line with the reports that receptor-mediated antibody internalization is epitope dependent (Pedersen et al., 2010;Beers et al., 2010;Rudnick et al., 2011) and demonstrates the importance of screening directly for internalization.
Real cell media samples were not available to us but the fact that 10 different samples could be easily tested in multiple replicates simultaneously, points to the utility of the method for screening purposes.

Discussion
Receptor-mediated antibody internalization is a key mechanism driving the efficacy of therapeutic antibodies but the current tools to study internalization are cumbersome multistep processes and require    expensive instrumentation.Here, we present a 96-well plate based homogeneous assay for studying antibody internalization which, is simple, robust and will be useful as a screening tool for selecting antibodies with good internalization properties and also for understanding the mechanism of internalization for a specific antibody.
The internalization assay presented here relies on a pH sensitive pHAb dye with low or no fluorescence at pH N 7, which becomes highly fluorescent at the acidic pH typically present in early endosomes and lysosomes.For the assay to be useful for antibody internalization, pHAb dye was modified to have amine and thiol reactive groups for labeling antibodies at lysines amino acids present at the antibody surface or in the antibody hinge region where free thiol groups can be generated by reducing disulfide binds between cysteine groups.In an attempt to minimize the precipitation problem, pHAb dyes were synthesized with two negatively charged sulfonate groups that make the dye hydrophilic (Panchuk-Voloshina et al., 1999) and as a result high DAR can be achieved with various antibodies without precipitating the antibodies (Table 1).However, stability of antibody-small molecule conjugates is a complex phenomenon dependent on several factors (Acchione et al., 2012;Shen et al., 2012) and it is possible that some antibodies could still precipitate upon conjugation with pHAb dyes.
In addition to the stability of antibody-pHAb conjugate, preserving antibody-antigen binding is also critical and was tested for various antibody-pHAb conjugates (Fig. 1).Labeling antibodies using amine chemistry is frequently used because there are around 80 lysine groups in an antibody of which around 40 are accessible for modification (Flygare et al., 2013).When two antibodies were labeled with pHAb dyes using amine chemistry, we saw a significant impact on binding affinity in case of Trastuzumab but minimal impact on Cetuximab even at a high DAR of 8.5.Loss of activity after labeling may be attributed to the presence of modified lysines close to the antibody CDR region which will have a detrimental impact on the antigen antibody binding and in fact Trastuzumab has one lysine group in its CDR region, whereas Cetuximab has none (Magdelaine-Beuzelin et al., 2007).Unlike amine chemistry, thiol chemistry places the dye at the defined position in the hinge region away from the antibody CDR region and may be the reason for minimal impact on the antibody binding affinity seen with two antibodies and reported in the literature (Hamblett et al., 2004).The relationship between DAR, labeling chemistry and antibody activity is complex and there are conflicting report in the literature with some recommending a DAR of 2-4 as ideal (Vira et al., 2010), and in fact two the FDA approved ADCs have a DAR of ~3.5 (Flygare et al., 2013); whereas other reports indicate no strict correlation between DAR and antibody activity (Shrestha et al., 2012).Finally, we compared the pH response of free pHAb dyes with pHAb dyes conjugated to antibodies and found a small shift in pH response for pHAb dyes only when conjugated through thiol groups.Fluorescent dyes are known to affect the fluorescence yield when in close proximity (Ogawa et al., 2010;Panchuk-Voloshina et al., 1999) and hence a shift for the pHAb dye is not surprising; however, the pH response for the dye is still in the useful range to detect the presence of pHAb labeled antibodies in the acidic vesicles.
An enabling aspect of pHAb dye for internalization is that it will fluoresce only when internalized into the acidic endosomal and lysosomal vesicles.When cells overexpressing HER2 were incubated with Trastuzumab-amine-pHAb, no fluorescence was visible on the cell membrane but imaging the cells at pH 5.0 showed the fluorescence on the outer membrane, confirming that antibodies are indeed bound to the receptors on the surface.After 24 h of incubation, highly fluorescent punctate structures indicative of internalization were clearly visible.Under similar conditions, no internalization was seen with HER2 negative MCF7 cells indicating that binding and internalization were specific.Confocal microscopy has been extensively used for antibody internalization studies but most of these reports are multistep endpoint assays involving either stripping of membrane bound fluorescent antibodies using an acidic buffer or quenching of fluorescence using anti-fluorophore antibodies (Casalini et al., 1993;Liao-Chan et al., 2015).Moreover, most of these current assays involve incubating cells with antibodies then removing the unbound antibody followed by internalization over a period.There are three problems with this approach: first, this approach is not typical of in vivo scenario where antibodies will be present around cancer cells for an extended period of time; second, antibodies with higher off rates will dissociate from the surface during incubation and will not be available for internalization; third, newly synthesized receptors will remain unlabeled and hence will not be visible upon internalization.For example, EGFR is turned over with typical half-lives ranging from 8 to 24 h depending on cell type and the level of EGFR expression (Sorkin and Duex, 2010).
In contrast, we reasoned that a dramatic increase in the fluorescence of pHAb dyes on internalization and relatively low background fluorescence from membrane bound dyes would allow us to monitor the internalization in real time without washing off the excess antibody.Indeed, we were able to incubate SKBR3 cells in presence of 3 nM of Trastuzumab-amine-pHAb (DAR 3.5) and image it continuously for 19 h.Minimal fluorescence was seen at the initial time point and fluorescent intensity increased with time as evident by punctate structures.Unlike the image in Fig. 3 where punctate structures were located within the cells, the real time images show these structures to be around the inside of the cell membranes as well as within the cells and may indicate the dynamic process of antibody in early endosomes, late endosomes and lysosomes.To the best of our knowledge, this is the first report of real-time monitoring of antibody internalization kinetics and shows that additional information can be gleaned regarding the mechanism of internalization.
Encouraged by the results from confocal microscopy, we expected the assay be suitable in a 96-well plate format and that quantitative data can be generated.Results obtained with Trastuzumab and Cetuximab show that internalization signal can indeed be detected within one hour of incubation and has not reached saturation even after 24 h.Absolute fluorescence signal from Trastuzumab-amine-pHAb is lower compared to others and can be attributed to the loss of antibody activity.The fact that the fluorescent signal from internalization is increasing even after 24 h may be result of several factors including: (a) slow antibody internalization.For example, EGFR is known to rapidly internalize in presence of EGF (Sorkin and Duex, 2010) whereas when bound to antibody, receptor undergoes slow internalization (Pedersen et al., 2010); (b) antibody-pHAb dye translocating into lysosomes where antibody is degraded and pHAb dyes may be accumulating in the lysosomes; (c) as internalization is done in the presence of excess antibody, new receptor may be getting expressed, labeled and internalized leading to signal increase.Relatively few studies have looked at the long-term antibody internalization (Pedersen et al., 2010;Gong and Urlacher, 2015;Klussman et al., 2004;Lewis Phillips et al., 2008;Austin et al., 2004) using tedious imaging based techniques, radioactivity measurement or cell cytotoxicity.Our approach significantly simplifies the process and should enable better understanding of the internalization process at a higher throughput.Although not done here, it will be interesting to see at what point the fluorescence reaches saturation and the ultimate fate of the dye within the cell.
Previous experiments were done at 30 nM but bright fluorescence of the pHAb dye allowed detection of internalization even at a very low antibody concentration (0.3 nM).To achieve maximum sensitivity in our assays, we typically replace the media with a pH 7.5-8.0phosphate buffer, which reduces the small amount of pHAb fluorescence present at neutral pH and also reduces some quenching of signal from FBS containing media.A sensitive assay will allow a wide array of antibodies with different affinities to be tested, which will typically be the case during early antibody discovery phase.Moreover, in our case, the number of receptors per cell is in the range of 1 to 2 million (Rusnak et al., 2007), however, many cases will have a lower level of assays that will require sensitive assays.
We have demonstrated the advantages of pHAb dyes in 96-well plate based homogeneous assays for receptor-mediated antibody internalization; however, the need to label antibodies with pHAb dyes can be a challenge for wider applicability of this technology as screening tool.One challenge during early antibody screening stage is the limited amount of antibodies available for research.Hence dialysis steps involved during labeling can be a major challenge.With amine conjugation chemistry, we are able to demonstrate that presence of excess dye does not interfere with the internalization or with the assay sensitivity.It can easily be envisioned that if a large panel of purified antibodies is available in limited quantity, simply incubating with 5-20 molar excess of amine reactive pHAb dye followed by internalization experiment will allow the selection of promising lead candidates.This approach is especially useful as purification of antibodies at early monoclonal screening stage using automated approaches becomes simpler.In a complimentary approach, we also showed that a secondary antibody labeled with pHAb dye can be used instead of labeling primary antibodies.The use of secondary antibodies for internalization experiments has been reported before (Klussman et al., 2004) but the effect of large primary and secondary antibody complexes on internalization needs to be taken into account.

Conclusion
pHAb dyes are pH sensitive fluorescent dyes that enable the study of receptor-mediated antibody internalization.Internalization assays can be performed in a plate-based homogeneous format and allow endpoint assays as well as real-time monitoring of internalization.We further show that internalization can be monitored even at a very low amount of antibody which is very important during the early monoclonal antibody development phase when the amount of sample is limited and the antibody concentration in the samples is low.Plate-based sensitive internalization assays described here will enable screening of a large library of antibodies for their internalization properties and may result in more efficient identification of lead antibody candidates with superior anti cancer efficacy.
Note: During review process a paper similar to ours was published and will be useful to the readers Riedl, T., et al., High-throughput screening for internalizing antibodies by homogeneous fluorescence imaging of a pH-activated probe.J Biomol Screen, 2016.21(1): p. 12-23.

Fig. 3 .
Fig. 3. Change in fluorescence as a function of pH measured for (A) Trastuzumab-amine-pHAb and (B) Trastuzumab-thiol-pHAb. Responses of free amine reactive dye (pHAb-SE) and thiol reactive dye (pHAb-ME) are also included.Free dye and dye labeled antibody were diluted into buffer of desired pH in a 96 well plate and fluorescence was measured using a 96 well fluorescent plate reader.Data was normalized to the fluorescence at pH 4.0.Mean and standard deviation from triplicate readings were plotted.
Fig. 4. Internalization of Trastuzumab-amine-pHAb monitored using confocal-microscopy.(A) SKBR3 cells were incubated for 1 h at 4 °C with 3.0 nM of Trastuzumab-amine-pHAb. Excess antibody was removed and replaced with pH 8.0 buffer and imaged on a Nikon C2si confocal microscope (Ex/Em: 561/580LP nm).(B) pH of the media was changed to pH 5.0 by replacing the buffer with citrate buffer (pH 5.0) and cells were imaged again to see the fluorescent antibody bound to the cell surface.(C) Citrate buffer was removed and replaced with DMEM media and incubated at 37 °C for 18 h and imaged again.Punctate structures indicate internalized antibody.(D) No internalization was seen when HER2 negative MCF7 cells were incubated with the antibody.

Fig. 6 .
Fig.6.Image of cells collected at 17 h in Fig.5is enlarged for clarity and to better highlight the fluorescent punctate structures.

Fig. 7 .
Fig. 7. Amine and thiol conjugated Trastuzumab and Cetuximab (30 nM) was added to the SKBR3 and A431 cells respectively and incubated for various time points.Cell media was replaced with PBS and plates read on a Tecan Infinity M1000 plate reader at Ex/Em of 532 nm/560 nm.MCF7 cells were used as negative controls.Mean and standard deviation from triplicate readings are plotted.

Fig. 9 .
Fig. 9. Comparison of fluorescence signal upon internalization obtained with and without removing excess dyes after conjugation.Replicate samples of Cetuximab were conjugated with 20 M excess of pHAb dyes using amine chemistry and one sample was dialyzed to remove excess dye whereas other sample was used as such.Two samples at 30 nM were incubated with the A431 cells (EGFR positive) and MCF7 cells (EGFR negative) for internalization and plates read.Average and standard deviation of 4 replicate samples are plotted.

Fig. 10 .
Fig. 10.Internalization of Cetuximab was performed using anti Human IgG labeled with amine reactive pHAb dye. 25 nM of anti Human IgG-amine-pHAb was added to Cetuximab at 0, 1.0 and 10.0 nM and added to A431 and MCF7 cells in a 96 well plate.Cells were incubated for 24 h at 37 °C to allow internalization before reading fluorescence.Average and standard deviation of four replicate samples are plotted.

Fig. 11 .
Fig. 11.Fluorescence signal upon internalization obtained with commercially available anti EGFR antibodies.A431 cells were incubated for 24 h with a mixture of 10 nM of mouse anti EGFR monoclonal antibody and 25 nM of anti mouse antibody labeled with pHAb dyes before reading the plate.Average and standard deviation of four replicate samples are plotted.

Fig. 12 .
Fig. 12. Cell based ELISA of commercial anti EGFR antibodies.Mean and standard deviation from triplicate readings are plotted.Data set was fitted to a four parametric sigmoidal equation using Graphpad.

Table 1
Dye to antibody ratio (DAR) of Trastuzumab and Cetuximab labeled with pHAb dye using amine and thiol chemistries.5, 10, and 20 M excess of dye relative to antibody was used to get a range of DAR.

Table 2
List of commercially available anti EGFR antibodies used for internalization studies using anti-mouse antibody conjugated with pHAb dyes using amine chemistry.