Near infrared fluorescence imaging of EGFR expression in vivo using IRDye800CW-nimotuzumab

Nimotuzumab is a humanized anti-epidermal growth factor receptor (EGFR) monoclonal antibody that is approved in many countries for the treatment of EGFR-positive cancers. Near infrared (NIR) fluorescent dye-labeled antibodies represent an attractive class of image-guided surgical probes because of their high specificity, tumor uptake, and low dissociation from tumor cells that express the antigen. In this study, we developed a NIR fluorescent dye-labeled nimotuzumab immunoconjugate, IRDye800CW-nimotuzumab, and evaluated in vitro binding with EGFR-positive cells, in vivo tumor uptake by NIR fluorescent imaging, and ex vivo biodistribution. There was no difference in binding between nimotuzumab and IRDye800CW-nimotuzumab to EGFR-positive cells. In mice bearing EGFR-positive xenografts, IRDye800CW-nimotuzumab uptake peaked at 4 days post injection and slowly decreased thereafter with high levels of accumulation still observed at 28 days post injection. In EGFR-positive xenografts, IRDye800CW-nimotuzumab showed more than 2-fold higher uptake in tumors compared to IRDye800CW-cetuximab. In addition, liver uptake of IRDye800CW-nimotuzumab was two-fold lower than cetuximab. The lower liver uptake of IRDye800CW-nimotuzumab could have implications on the selected dose for clinical trials of the immunoconjugate. In summary, this study shows that nimotuzumab is a good candidate for NIR fluorescent imaging and image-guided surgery.

Currently, resection of tumor tissue relies on images taken before surgery, either by computed tomography (CT), positron emission tomography (PET), or magnetic resonance imaging (MRI), followed by unassisted surgery using white light illumination and visual inspection. This leads to a high risk of tumor recurrence due to ill-defined tumor margins and inadequate resection [8]. Use of new Research Paper technologies such as neuronavigation and intraoperative CT or MRI have increased in recent years, however, evidence of the advantage of these technologies over conventional techniques remains scarce. In addition, these tools are not antigen specific, do not significantly improve tumor visualization, and are not performed in real time. Intraoperative fluorescence probes offer improved sensitivity and tumor specificity and can be used during surgery thereby improving surgical outcomes [9].
Image-guided surgery is a recently developed technique that is used to specifically target tumors with fluorescent labels to more precisely identify cancer cells, aiding surgeons in defining tumor margins and completely removing tumors during resection [10]. In a multicenter randomized phase III trial, Stummer et al., [11] compared two types of surgical resection for glioma followed by standard radiotherapy. Patients with malignant tumors were randomly allocated to receive oral 5-aminolevulinic acid (in situ fluorescent dye) before surgery and resection of fluorescent tissue or to conventional microsurgery with white light. Fluorescence-guided surgery led to a higher number of completely resected tumors than whitelight guided surgery (65% vs 36%) and the 6-month progression-free survival was two-fold higher using fluorescence (41%) compared to white light (21%). These results show the importance of delineating tumor margins during resection, however, 5-aminolevulinic acid relies on the high metabolism of cancer cells to produce the fluorescence. Targeted approaches using fluorescent molecules that specifically bind cancer cells are needed to determine tumor margins and provide quantitative information about EGFR expression.
There are two major obstacles in using fluorophores for imaging: auto-fluorescence from human tissue and limited tissue penetration [13]. These issues can be overcome by using NIR fluorescent dyes that have emission wavelengths with reduced auto-fluorescence [13] and have tissue penetration of < 1 cm for fluorescence reflectance imaging (FRI) and < 10 cm for fluorescence molecular tomography (FMT), [13]. NIR dyes are also useful in real-time endoscopy or during surgery where the wound bed is exposed.
IRDye800 is an NIR fluorophore [12] that has been conjugated to antibodies to target tumors, including IRDye800CW-cetuximab [14] and IRDye800CWpanitumumab [15] for head and neck cancer and IRDye800CW-trastuzumab for breast cancer [16]. IRDye800CW-cetuximab is currently in clinical trials for imaging and image-guided surgery for head and neck cancer, malignant glioma, and pancreatic cancer. IRDye800CW-Bevacizumab is clinical trials for imaging and image-guided surgery of breast cancer and colorectal cancer [17].
Nimotuzumab is an anti-EGFR antibody with binding characteristics that make it desirable as a molecular-targeted imaging probe. First, it has minimal transient binding to low EGFR-expressing healthy tissues such as the skin [18]. Garrido et al., [18] showed that nimotuzumab requires bivalent binding for stable attachment to the cell surface and that the low skin toxicity of nimotuzumab is attributed to its transient monovalent binding in low-EGFR expressing tissues. This low transient monovalent binding is due to a 10-fold lower affinity of nimotuzumab for EGFR compared to cetuximab or panitumumab [18]. Second, nimotuzumab has limited toxicity compared to EGFR antibodies cetuximab and panitumumab, which have significant cutaneious toxicity [19]. Third, nimotuzumab is approved for head and neck cancer, glioma, and nasopharageal cancer in several countries [20]. There are also more than 60 clinical trials ongoing or completed using nimotuzumab for cancer therapy [17].
In this study, we evaluated IRDye-800CWnimotuzumab as a NIR fluorescent imaging probe and compared it to IRDye-800CW-cetuximab. We showed that IRDye-800CW-nimotuzumab binds to a variety of EGFR positive human cancer cell lines in vitro and in vivo. Biodistribution studies showed very little nonspecific IRDye-800CW-nimotuzumab organ uptake and IRDye-800CW-nimotuzumab accumulated primarily in the tumor after 48 hours post injection (hpi). Tumor uptake of IRDye-800CW-nimotuzumab was comparable to IRDye800CW-cetuximab, which is currently in clinical trials for image-guided surgery for head and neck cancer, pancreatic cancer, and glioma.

Conjugation and characterization of IRDye800CW-labeled antibody conjugates
IRDye800CW was conjugated to nimotuzumab (anti-EGFR) and a control antibody (anti-maltose binding protein) non-specifically through free primary amines with NHS chemistry. Conjugated and unconjugated antibodies were analyzed for purity, size, and binding to recombinant EGFR. Nimotuzumab, the control antibody, and the IRDye800CW-immunoconjugates were ≥ 95% pure and had the expected molecular weight (Supplementary Figure 1). IRDye800CW-nimotuzumab but not the control antibody bound to recombinant human EGFR with a K D of 20.1 ± 0.8 nM, which was not significantly different than the K D of unlabeled nimotuzumab, 22.3 ± 2.3 nM (p value > 0.05) (Supplementary Figure 2).

In vitro characterization and serum stability of nimotuzumab and IRDye800CW-nimotuzumab
We characterized the binding of nimotuzumab and IRDye800CW-labeled nimotuzumab to the EGFRpositive cell lines DLD-1 or A431 cells using flow cytometry ( Figure 1A, 1B, Supplementary Figure 3). A slight increase in the K D was observed for IRDye800CWnimotuzumab relative to unlabeled nimotuzumab (p value > 0.05). IRDye800CW-nimotuzumab had a K D of 9.7 ± 1.6 nM and 10.4 ± 1.3 nM on DLD-1 and A431 cells, respectively, whereas nimotuzumab had a K D of 4.5 ± 2.4 nM and 9.8 ± 3.5 nM on DLD-1 and A431 cells, respectively. The control antibody had negligible cell binding ( < 5%) ( Figure 1B).
Next, we characterized the affinity and the relative amount of EGFR per cell on various cancer cell lines. The level of nimotuzumab binding in these cell lines was used to approximate EGFR expression. Using flow cytometry, we measured the change in mean fluorescence intensity (ΔMFI) of nimotuzumab to classify cell lines into four groups based on EGFR expression: High -A-431 and FaDu; medium -HT-29, DLD-1 and MDA-MB-468; low -MCF7 and T-47D; and EGFR-negative -MDA-MB-435. There was no statistical significance in EGFR expression between different cell lines within each group but there was significant statistical difference between each group (p-value < 0.0001) ( Figure 1C).
The K D of nimotuzumab binding to each cell line was determined using flow cytometry ( Figure 1D). Despite having the highest EGFR expression, A-431 cells had a higher K D than the medium EGFR expressing cell lines ( Figure 1D). Two cell lines MCF7 and T47D with lower EGFR expression showed significantly weaker binding. The K D for nimotuzumab binding to A-431 was significantly higher (p < 0.05) when compared to FaDu, MDA-MB-468, DLD-1, and HT-29. K D values of nimotuzumab binding to MCF7 and T-47D were statistically significantly higher (p < 0.01) than FaDu, MDA-MB-468, DLD-1 and HT-29. These results showed that affinities of nimotuzumab to cell lines were correlated with receptor density, except when EGFR expression was extremely high as was seen in A-431 cells.
The serum stability of IRDye800CWnimotuzumab was measured by incubating IRDye800CWnimotuzumab in mouse serum for one week. 89±10% of IRDye800CW-nimotuzumab remained intact at 7-days ( Figure 1E, Supplementary Figure 4).

In vivo near infrared imaging
IRDye800CW-nimotuzumab and IRDye800CWcontrol antibody were used to image mice bearing A-431, DLD-1, and MDA-MB-435 xenografts, which are high, medium, and low EGFR expressing cell lines, respectively (Figure 2A-2E). A-431 is an epidermoid carcinoma with the highest EGFR expression where DLD-1 is a Duke's type C. colorectal adenocarinoma with intermediate EGFR expression. MDA-MB-435 cells were used as a negative control as they had no detectable EGFR expression ( Figure  2C). The control antibody was tested in A-431 and DLD-1 cell lines ( Figure 2D, 2E).
Mice were imaged in dorsal and ventral positions. Dorsal images showed minimal kidney uptake of IRDye800CW-nimotuzumab and long residence time in the xenograft in both A-431 and DLD-1 xenografts with tumor accumulation observed up to 21 days (Figure 2A, 2B). Ventral images showed some clearance through the bladder and fast liver clearance of IRDye800CW-nimotuzumab in the first few hpi, with a significant decrease after 24 hpi. In A431 xenografts the fluorescent signal of IRDye800CW-nimotuzumab was significantly higher than the IRDye800CW-control IgG at all time points (p < 0.01 for 1 hour, p < 0.001 for other time points) ( Figure 2F). In DLD-1 xenografts there was no significant difference in the fluorescent signal between IRDye800CW-nimotuzumab and the IRDye800CW-control IgG up to 24 hpi (p > 0.05) ( Figure 2F). By 72 hpi IRDye800CW-nimotuzumab was significantly higher in DLD-1 xenografts compared to the IRDye800CW-control antibody (p < 0.05).
When different xenografts treated with IRDye800CWnimotuzumab were compared, accumulation of IRDye800CW-nimotuzumab in A-431 (high EGFR expression) and DLD-1 (intermediate EGFR expression) cells was not significantly different at any of the imaging time points (p > 0.05) ( Figure 2G). However, IRDye800CWnimotuzumab accumulation in A-431 and DLD-1 cells was significantly higher than accumulation in EGFR negative (MDA-MB-435) xenografts after 72 hpi (p < 0.05 at 72 hpi for both A-431 and DLD-1, p < 0.01 at 168 hpi for DLD-1 and p < 0.001 at 168 for A-431) ( Figure 2F). These results showed that IRDye800CW-nimotuzumab was specific for EGFR expression, but EGFR expression alone did not correlate with increased signal.
The serum stability of IRDye800CW-nimotuzumab indicated that it would be stable in vivo. To confirm that the fluorescence observed in the xenograft image was not due to free IRDye800CW, we imaged A-431 xenografted mice injected with quenched IRDye800CW (Supplementary Figure 5). IRDye800CW distributed quickly throughout the mice within 1 hpi and by 6 hpi the dye cleared through the bladder in ventral images. At 24 hpi, only a very faint signal was observed in the xenograft.
During imaging experiments, we did not observe any adverse effects on the mice from administering www.impactjournals.com/oncotarget  IRDye800CW-nimotuzumab. In these experiments, no mice died or showed signs of significant weight loss (Supplementary Figure 6), or other gross abnormalities.

Ex vivo imaging and biodistribution
We performed biodistribution analysis of IRDye800CW-nimotuzumab at 24, 72, and 168 hpi in mice bearing DLD-1 xenografts (Figure 3). Biodistribution was also quantified at 72 hpi in EGFR-negative xenograft MDA-MB-435 and for the IRDye800CW-control antibody in DLD-1 xenografts. In vivo NIR images were taken of mice ( Figure 3A) and ex vivo white light and NIR images of dissected organs ( Figure 3B, 3C). Consistent with the in vivo imaging, fluorescence is mostly visible in the liver, tumor, and kidney at 24 hpi. Additionally, there was low uptake in the skin, ovaries/testes, lungs, heart, spleen, pancreas, and stomach. After 72 hpi, the fluorescence was cleared from most organs except for the tumor and to a lesser extent the liver, skin, and kidneys. At 168 hpi fluorescence was only visible in the tumor and to a lesser extent in the liver.
NIR ex vivo images of organs collected from five mice bearing DLD-1 xenografts injected with IRDye800CWnimotuzumab were used to calculate the signal per area for each organ at 24, 72, and 168 hpi ( Figure 3D). At all time points, the signal was highest in the tumor with average fluorescent values of 358 ± 45, 356 ± 32, 368 ± 39 AU, at 24, 72, and 168 hpi, respectively. Low uptake was observed in the liver, skin, kidneys, and lungs and these decreased to background values at 168 hpi. At 72 hpi the highest uptake was observed for IRDye800CW-nimotuzumab in EGFR-positive DLD-1 xenograft (356 ± 32 AU) compared to EGFR-negative MDA-MB-435 (177 ± 23 AU) and IRDye800CW-control antibody in DLD-1 (84 ± 14 AU) ( Figure 3E).
Organs collected for biodistribution were homogenized to analyze the amount of IRDye800CWnimotuzumab per organ weight ( Figure 4A). The blood, xenograft, and liver had the highest signal in the 800 nm channel, which was consistent with the ex vivo analysis ( Figure 3D, 3E) for the xenograft and liver. Signal per gram was calculated ( Figure 4B) based on the fluorescence signal measured from the homogenized organs and tissues. Blood had the highest signal at 24 hpi followed by the liver and xenograft. After 72 hpi the fluorescence signal in the blood and liver decreased whereas the xenograft increased. After 168 hpi the blood and liver further decreased to near background and the xenograft remained high. All other organs and tissues tested had fluorescence signal at background levels ( Figure 4B). This analysis gave similar results to what was observed in the ex vivo analysis ( Figure 3D) and were consistent with the in vivo imaging showing IRDye800CW-nimotuzumab accumulating in the tumor with low background accumulation in other organs.

Comparison of IRDye800CW-nimotuzumab to IRDye800CW-cetuximab
We next compared EGFR-binding and xenograft imaging of IRDye800CW-nimotuzumab to IRDye800CWcetuximab, an anti-EGFR fluorescent imaging probe currently in clinical trials. Cetuximab was conjugated to IRDye800CW and binding was measured using flow cytometry and compared to IRDye800CWnimotuzumab and unlabeled antibodies ( Figure 5A). There was no significant difference in binding between the IRDye800CW-cetuximab and unlabeled cetuximab (82 ± 8 pM vs 100 ± 34 pM, p > 0.5). Cetuximab had a significantly lower K D ~ 80 pM compared with nimotuzumab (4-10 nM, p < 0.0001). These results are similar to previously published K D values for cetuximab of 100 pM [21].
A comparison of all antibodies and cell-lines used in this study is shown in Figure 6. In the xenograft, controls (IRDye800CW-MBP IgG and IRDye800CWnimotuzumab in MDA-MB-435) were found clustered together, IRDye800CW-nimotuzumab in the A-431 and DLD-1 cells were found clustered together ( Figure 6A). IRDye800CW-cetuximab was between controls and IRDye800CW-nimotuzumab. In the liver, there was some accumulation of the control IgG and of IRDye800CWcetuximab to a much lesser extent. The fluorescent signal    of IRDye800CW-nimotuzumab peaks later (T max = 96 ± 14 hpi) than IRDye800CW-cetuximab (T max = 40 ± 8 hpi) and has a higher maximum accumulation (S max ) in the xenograft (Figure 6A, 6B). IRDye800CW-cetuximab has previously been shown to peak at 48 hpi [14]. The difference between IRDye800CW-nimotuzumab and IRDye800CW-cetuximab in the maximum peak and time could be explained by their differences in clearance. IRDye800CW-cetuximab was found to clear through the liver more quickly than IRDye800CW-nimotuzumab, at early time-points up to 1.5-fold more IRDye800CWcetuximab was in the liver than IRDye800CWnimotuzumab ( Figure 6C). This also reduced the circulating time of IRDye800CW-cetuximab, where there was more IRDye800CW-nimotuzumab in circulation (contralateral) then cetuximab after 48 hpi ( Figure 6C). This delayed clearance resulted in significant signal in the tumor for up to 28 days, which allows the tumor to be imaged over an extended period of time if necessary. This also gives surgeons more time to perform resection of the tumor.

DISCUSSION
Nimotuzumab is approved in a number of countries for squamous cell carcinoma of the head and neck, glioma, and nasopharyngeal cancer [20]. There are 65 clinical trials with nimotuzumab on clinicaltrials.gov that cover a number of cancers, including esophageal, head and neck, lung, cervical, brain, colorectal, stomach, and pancreas. There is a need for fluorescent imaging probes for these and other EGFR positive cancers that can be used for image-guided surgery. In this study, we showed that IRDye800CW-nimotuzumab binds to EGFR positive cell lines in vitro and in vivo and showed fluorescent imaging properties that make it an excellent candidate for repurposing as an imaging probe. A benefit of using therapeutic molecules for imaging is that many of the side effects and toxicities are already established. Cetuximab, for example has been repurposed for fluorescent imaging and is in phase I and phase II clinical trials for malignant glioma and pancreatic cancer imaging and fluorescenceguided surgery with IRDye-800CW [17]. Nimotuzumab is safe and well tolerated in the clinical setting and should show less side effects for imaging as the imaging dose is significantly lower than the therapeutic dose.
There are three FDA approved EGFR monoclonal antibodies, cetuximab, panitumumab, and necitumumab. Necitumumab and panitumumab are approved for treatment of non-small-cell lung carcinoma and metastatic colorectal cancer [22,23]. Cetuximab is approved for head and neck cancer and colorectal cancer [19,23]. Necitumumab is associated with infusion reactions, hypomagnesemia, diarrhea, and dermatological toxicities [22]. Panitumumab has grade 3 and grade 4 adverse events and a high incidence (90-100%) of acneiform rash [23] similar to cetuximab. In contrast, nimotuzumab has a much lower incidence (3%) of this type of rash and an absence of diarrhea or anaphylaxis, which one or both have been associated with EGFR targeting agents [20,24].
The decreased toxicity of nimotuzumab is proposed to be due to the relatively high K D value of nimotuzumab, which requires bivalent binding to accumulate on EGFR expressing cells. Normal tissues such as skin with low levels of EGFR expression are bound monovalently by nimotuzumab and thus are retained transiently. Tumors overexpressing EGFR are bound bivalently by nimotuzumab and this avidity causes it to accumulate in EGFR overexpressing tumors [18]. We observed a correlation between IRDye800CW-nimotuzumab affinity and EGFR expression for low and intermediate EGFR expressing cell lines, however, at higher EGFR expression levels a threshold was reached where no further increase in IRDye800CW-nimotuzumab binding was observed. We did not see a similar correlation between receptor density and in vivo IRDye800CWnimotuzumab binding. DLD-1, which has intermediate EGFR expression had similar IRDye800CWnimotuzumab uptake compared to A-431, which had 15-fold higher EGFR expression. Similar results were reported with IRDye800CW-cetuximab accumulation in intermediate and high EGFR expressing cell lines [25]. These results could be due to physiologic conditions unrelated to EGFR expression, such as differences in vasculature, perfusion, and necrosis, which will ultimately affect delivery of the antibody [26,27].
The dose of IRDye800CW-nimotuzumab was chosen for imaging and no therapeutic or toxic effects were specifically examined or expected. IRDye800CW has a drug master file with the FDA and is safe (NOAEL) at doses of 20 mg/kg [28], which is > 500-fold higher than the conjugated dye concentration used in this study. A therapeutic dose of nimotuzumab (1200 mg per patient) has no significant toxicities [29]. In this study, we used a human equivalent dose [30] > 80-fold lower than the human therapeutic dose and no adverse effects on the mice were observed.
In summary, the preclinical imaging properties of IRDye800CW-nimotuzumab combined with its low clinical toxicity, make it a promising antibody for repurposing as a NIR fluorescent imaging probe.

Reagents
Nimotuzumab was supplied by the Center of Molecular Immunology (CIM) (Havana, Cuba). IRDye800CW was provided by Li-Cor Biosciences (Lincoln, NE). Cetuximab was obtained from Royal University Hospital (RUH) Saskatoon Pharmacy. Antimaltose binding protein (anti-MBP) Fab was a generous gift from Dr. Sidhu (University of Toronto, Canada) and was sub-cloned into pFUSE-ss-CHIg-hG1and pFUSE-CLIG-hk plasmids (InvivoGen) to produce the IgG1kappa [31]. Anti-MBP IgG was expressed using Expi293 expression system (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer's instructions. The IgG was purified on MabSelect SuRe column (GE Healthcare Life Sciences, Chicago, Il) according to the manufacturers' instructions and dialysed overnight in phosphate buffered saline (PBS).

Antibody labeling with IRDye® 800CW and characterization
One milligram of antibody was labeled in 1X PBS (pH 7.0) with IRDye ® 800CW NHS Ester (Li-Cor Biosciences, Lincoln, NE) at a 1:3 antibody:dye molar ratio at 4°C overnight according to the manufacturer's instructions. After labeling the unconjugated dye was removed using a Zeba 7000 Da molecular weight cut-off desalting column (ThermoFisher Scientific, Waltham, MA). Labeled antibodies were filter sterilized using Ultrafree MC centrifugal filter unit (Millipore, Billerica, MA). The concentration of IRDye800CW-nimotuzumab was measured using the following formula, All labeled and unlabeled proteins were analyzed for purity using 2.5 µg of the protein loaded on an Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA) according to the manufacturer's instructions.
Binding kinetics between the antibodies and recombinant proteins were measured by BLI with ForteBio Octet RED384 (PALL Corporation, Port Washington, NY). Antibodies were immobilized on Anti-human FAB-CH1 sensors (Forte Bio, Menlo Park, CA) according to the manufacture's instructions. After immobilization, antibodies were exposed to 500 nM, 166 nM, and 55 nM concentrations of the target proteins, human EGFR (Sino Biological, Beijing, China). At the same time, empty sensors were exposed to the same concentrations of the target protein to be used for subtraction of nonspecific binding. All reactions were performed at 30°C in 1X kinetics buffer (Forte Bio, Menlo Park, CA). The equilibrium dissociation constant (K D ) was obtained using a 1 to 1 model with global fitting. Data analysis and curve fitting was performed using Data Analysis software 7.1.0.33 (Forte Bio, Menlo Park, CA). A baseline was determined in 1X kinetics buffer for 5 minutes. The sensors were loaded for 5 minutes in 75 nM of antibody or 1X kinetics buffer (for negative subtraction). The second baseline was collected for 5 minutes in 1X kinetics buffer. Association was performed for 3 minutes in 500 nM, 166.7 nM, 55.5 nM of the target protein. Dissociation was determined in 1X kinetics buffer for 10 minutes after which the sensors were regenerated.

Serum stability assay
IRDye800CW-Nimotuzumab IgG was incubated in mouse serum at a final concentration of 0.2 mg IgG per mL of serum at 37°C protected from light. Samples collected at 0, 24, 48, 72, and 168 h were run on sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) and imaged in the 800 channel on the Odyessy CLx NIR scanner (LI-COR, NE) to visualize the IRDye800CW-Nimotuzumab IgG band. Subsequently the gel was Coomassie Blue stained and rescanned with the 800 and 700 channel to visualize the serum proteins in the 700 fluorescence channel to evaluate loading and visualize the protein ladder. The relative fluorescence intensity of each time point IRDye800CW-nimotuzumab IgG band was normalized to the 0 h band intensity. Each sample was done in duplicate and analysis is the average of a minimum of four data points ±SD.

Mouse xenograft models
CD-1 nude female mice (Charles River Laboratories, Hartford, CT), aged 4-6 weeks, were obtained and housed in accordance with the University Animal Care Committee (UACC) guidelines (protocol # 20150048). All experiments and euthanasia were performed in accordance with UACC guidelines. For xenograft preparation, 10 × 10 6 cells A-431, DLD-1 or MDA-MB-435 were collected and washed with growth media without FBS. Cells were suspended in 50 µL of growth media without FBS plus 50 µL of matrigel membrane matrix (Corning, Corning NY) and cells suspended in 100 µL was injected subcutaneously into the right hind flank of the mouse. Tumor size was monitored weekly until they reached 150-300 mm 3 . The mice were then intravenously injected through the tail vein with 0.5 nmoles (75 µg) of IRDye-800CW-nimotuzumab, IRDye-800CW-cetuximab, IRDye800CW-control IgG or 1 nmole of quenched IRDye800CW free dye.

Fluorescence imaging and ex vivo Biodistribution
Tumor bearing mice (n ≥ 3 per group) were injected via tail vein with 0.5 nmoles of IRDye800CWnimotuzumab or the control IgG (75 µg IgG) antibodies formulated in PBS and imaged using a Pearl Impulse small animal imaging system (LI-COR, NE) at 1, 6, 24, 48, 72, 96, 168, and 504 hpi. Images of mice were analyzed using Image Studio Software (version 3.1). Regions of interest (ROI) for xenografts, liver, kidneys, contralateral side, and background were selected from equivalent-sized areas containing the same number of pixels. Quantification of uptake was done by drawing ROIs (three per organ) for each organ and mean uptake in the organs was determined using the following equation: Normalized fluorescence (au) = Signal xenograft -Signal contralateral / Labeling ratio Where Signal xenograft is the mean fluorescence from the three ROIs from the xenograft, the Signal contralateral is the mean fluorescence from the three ROIs from the contralateral side and the labeling ratio is the labeling ratio of the protein. For biodistribution studies, at least three mice per group were injected via tail vein with 0.5 nmole each of IRDye800CW-nimotuzumab or the control IgG. Animals were imaged at different time intervals after injection and immediately before dissection and organ collection. Immediately after imaging, mice were weighed (Supplementary Figure 6) euthanized and all organs were collected and stored on ice. Dissected organs were immediately weighed in tared tubes and imaged using the Pearl Impulse small animal imaging system to obtain the mean (mean of two ROIs) MFI for the organ. Fluorescence in organs was expressed as mean fluorescence or signal/ area using arbitrary units (AU). After imaging organs were weighed (Supplementary Figure 6) and stored at-80°C for homogenization and subsequent analysis.
Homogenization was performed as described by Oliveira et al. [32] with some modifications. Briefly, 40-60 mg of each organ was added to lysis matrix A (MP Biomedicals Santa Ana, CA) with 1 mL of RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Igepal (NP40), 0.5% Na-deoxycholate, 0.1% SDS and 1% protease inhibitor cocktail (Roche). Samples were homogenized using a FastPrep120 pulsed for 30 seconds on, 5 minutes off three times and centrifuged for 5 minutes at 12,000 × g. 200 µL of homogenate was aliquoted into a black 96-well optical bottom plate (Fisher) and serial diluted in half with RIPA buffer 6 times. The plate was centrifuged for 10 minutes at 1000 × g and scanned on a Li-Cor Odyssey scanner and expressed as mean fluorescence (signal) per gram of tissue.

Statistical analysis
To compare the K D values from flow cytometry and Octet and ΔMFI from flow cytometry and nimotuzumab fluorescence in different cell lines we used one-way analysis of variance (ANOVA) with multiple comparisons using Prism6 version 6. The calculated K D s for labeled and unlabeled nimotuzumab and nimotuzumab compared to the control IgG in fluorescence imaging were compared with a parametric unpaired t-test using Prism6 version 6. All error bars are standard error of the mean (sem) unless otherwise noted.