Photoacoustic and Fluorescence Imaging of Cutaneous Squamous Cell Carcinoma in Living Subjects Using a Probe Targeting Integrin αvβ6

Cutaneous squamous cell carcinoma (cSCC) is the second most common non-melanoma skin cancer worldwide. Today, cSCC is diagnosed by visual inspection followed by invasive skin biopsy. There is a need to develop non-invasive diagnostic tools to achieve early and accurate detection. Photoacoustic imaging (PAI) possesses high ultrasonic resolution and strong optical contrast at new depths (<1–5 cm). Together with exogenous contrast agents, PAI has found promising use in various tumors in living subjects. The expression of integrin αvβ6 is significantly up-regulated in cSCC. We fabricated an anti-integrin αvβ6 antibody and labeled it with indocyanine green (ICG) to form an ICG-αvβ6 antibody. The results showed that the ICG-αvβ6 antibody probe could be used to detect cSCC with high specificity (3-fold over the control by PAI) and deep penetration (approximately 1 cm) by PAI. This suggests that the ICG-αvβ6 antibody is a promising probe targeting the integrin αvβ6 for detection of cSCC tumors by PAI and fluorescence imaging. It may find clinical application in the early diagnosis of cSCC as well as in intraoperative navigation.

human use, the rest of the imaging agents are not yet approved. In this study, we selected a near-infrared fluorescent dye, indocyanine green (ICG), to serve as a multimodal fluorescence and photoacoustic contrast agent. ICG has been in clinical use for decades for retinal angiography and liver function studies. Recently, it is expected to achieve sensitive fluorescence and photoacoustic signals; having been developed for tumor imaging, it is potentially applicable in clinical photoacoustic imaging 6 .
Integrins are a family of heterodimeric cell surface receptors. Integrin α v β 3 has been shown to be expressed on the surface of cancer cells and the tumor neovasculature. However, in certain cancers, integrin α v β 6 becomes highly overexpressed on cell surfaces and is undetectable in most normal adult tissues 7 . The expression of α v β 6 is significantly up-regulated in cSCC 8,9 . In our previous study 10 , the probe A740-R 0 1, a peptide labeled with the fluorescent dye Atto 740, was able to detect α v β 6 -positive tumors in living subjects, but the signals were relatively weak. Moreover, Atto 740 is still not approved for human use.
In this study, we plan to fabricate an anti-α v β 6 antibody, label it with ICG and evaluate the ability of the ICG labeled antibody to detect cSCC tumors by PAI and fluorescence imaging.

Results
Production of anti-α v β 6 antibody. Expression yields for anti-α v β 6 antibody from FreeStyle 293 F cells were approximately 33.2 mg/L ( Supplementary Fig. S1) following NAb Protein G Spin column purification. Under non-reducing conditions, SDS-PAGE revealed a single band present at ~150 kDa. Reducing SDS-PAGE conditions for the antibody revealed two bands at ~51 kDa and 23 kDa, representing the heavy chain and light chain of the antibody (Fig. 1a, Supplementary Fig. S2a). Mass spectrometry further confirmed the molecular weight of the antibody. A clear peak corresponding to the entire antibody was visible at 150,319 m/z ( Supplementary Fig. S3a). Under reducing conditions, peaks of the heavy chain and light chain were visible at 52,609 m/z and 22,419 m/z ( Supplementary Fig. S3b).
Synthesis and characterization of ICG-conjugated anti-α v β 6 antibody. The covalently bound ICG on the ICG-α v β 6 antibody and ICG-ScrIgG could be observed on reducing and non-reducing SDS-PAGE by the fluorescence imaging system (Fig. 1b, Supplementary Fig. S2b). The conjugation ratio of dye to antibody was 5.1 and 5.2 for ICG-α v β 6 antibody and ICG-ScrIgG, respectively, as calculated from the optical spectrum ( Supplementary Fig. S4).
The maximum absorption wavelength of the ICG-α v β 6 antibody was 710 nm (Fig. 2a, Supplementary Fig. S4). The strongest signal from the ICG-α v β 6 antibody can be detected by photoacoustic instrument at laser light 710 nm, while a limited photoacoustic signal could be detected at 900 nm. On the other hand, for mouse blood, photoacoustic signals at 710 nm were a little bit weaker than those at 900 nm. Subtraction of the images obtained at 710 nm and 900 nm could eliminate the signals from blood and highlight the signals from the ICG-α v β 6 antibody (Fig. 2b,c).
No significant decrease of photoacoustic signal from the ICG-α v β 6 antibody was observed after irradiation with 710 nm and 900 nm laser light for 30 min. Binding affinity. The binding affinity of the anti-α v β 6 antibody and ICG-α v β 6 antibody against immobilized integrin α v β 6 (n = 3 each) was demonstrated (Fig. 3a). Sigmoidal curves were fitted to the absorbance data using the software GraphPad Prism (GraphPad, La Jolla, CA). The IC 50 values of the anti-α v β 6 antibody and ICG-α v β 6 antibody towards integrin α v β 6 were 6.5 nM and 28.4 nM, respectively. Competitive binding ELISAs against integrin α v β 3 and α v β 5 were used to confirm the specificity of the antibody towards α v β 6 ( Supplementary Fig. S5a,b).
In microscopic study, higher fluorescence signals within integrin α v β 6 positive A431 cells could be observed after the cells were incubated with ICG-α v β 6 antibody than with ICG-ScrIgG (Fig. 3b). Much lower fluorescence  (a) Determination of binding affinity by competitive binding ELISA for the anti-α v β 6 antibody (green, IC 50 = 6.5 nM) or ICG-α v β 6 antibody (brownish red, IC 50 = 28.4 nM) following incubation with fibronectin against immobilized integrin α v β 6 . Each data point represents the mean of triplicate experiments and errors bars indicate standard deviation. (b) Microscopic fluorescence images of cSCC A431 cells (integrin α v β 6 positive) incubated with ICG-α v β 6 antibody or ICG-ScrIgG for 4 h at 37 °C. Higher signals were observed after incubation with ICG-α v β 6 antibody. signals within A431 cells were observed when they were blocked with excess unlabeled antibody before the addition of the ICG-α v β 6 antibody ( Supplementary Fig. S5c). A limited fluorescence signal could be observed within integrin α v β 6 negative 293 T cells ( Supplementary Fig. S5d). These results confirm the specific binding of the ICG-α v β 6 antibody to integrin α v β 6 -expressing cells.
Sensitivity Study. The minimum detectable number of A431 cells incubated with ICG-α v β 6 antibody was determined to be 0.06 million by PAI (P = 0.03 when compared with blank) and 0.03 million by fluorescence imaging (P = 0.04 when compared with blank) (Fig. 4).
Small Animal Imaging and Tissue Biodistribution. The efficacy of the ICG-α v β 6 antibody at detecting cSCC tumors in vivo was studied in nude mice (Fig. 5, Supplementary Fig. S6). All photoacoustic images were collected at 710 nm and 900 nm. The images were subtracted to un-mix the probe signals ( Supplementary Fig. S6). The un-mixed signals were then fused with images at 900 nm to demonstrate the vasculature of the tumor and signals of the probes simultaneously ( Fig. 5a,b,c).
The photoacoustic signals increased dramatically 1 min post injection of probes and then decreased until the 1 h time point. No significant difference of photoacoustic signal could be observed after tail vein injection of ICG-α v β 6 antibody (3.96 AU ± 0.44) and ICG-ScrIgG (4.29 AU ± 0.55; P > 0.05) at the 1 min time point or at the 1 h time point (2.22 AU ± 0.43 and 1.86 AU ± 0.47, respectively; P > 0.05). Higher photoacoustic and fluorescence signals were observed at all the other time points (2 h, 4 h, 24 h and 48 h) post injection of ICG-α v β 6 antibody compared to ICG-ScrIgG (P < 0.05) (Fig. 5d). In vivo photoacoustic imaging of cSCC tumors showed a 3-fold higher signal (3.36 AU ± 0.56) 24 h post injection of ICG-α v β 6 antibody than that of ICG-ScrIgG (1.03 AU ± 0.21; P = 0.002). The signals could be inhibited by administration of an excess of unlabeled anti-α v β 6 antibody 10 min before injection of ICG-α v β 6 antibody (1.43 AU ± 0.29; P = 0.006).
The biodistribution of probes in various organs and tumors was evaluated ex vivo by fluorescence imaging (Fig. 6a). Higher maximum radiance in cSCC tumors was observed 48 h post injection when ICG-α v β 6 antibody was injected compared to ICG-ScrIgG. No significant difference was observed in other organs between the two groups. Higher maximum radiance was observed in the liver, followed by the spleen, intestine and kidney (Fig. 6b) in both groups of mice. Microscope images confirmed the presence of signals from the probe in the tumor tissue of the ICG-α v β 6 antibody injected mouse but not in that of the ICG-ScrIgG injected mouse ( Supplementary Fig. S7).

Discussion
The IC 50 values of the fabricated antibody and ICG labeled antibody towards integrin α v β 6 were 6.5 nM and 28.4 nM, respectively. The almost 5-fold decrease of IC 50 value may be caused by the combination of ICG on lysine residues present within the variable domain of HC. This combination can potentially have adverse effects on antigen recognition. Decreasing the labeling ratio of ICG or site-specific binding of ICG may reduce this impact. The low inhibition rate of the fabricated antibody against integrin α v β 3 and α v β 5 in competitive binding ELISAs confirms the specificity of the antibody towards integrin α v β 6. The lower fluorescence signal within A431 cells (α v β 6 positive) in the blocking study and limited signals within 293 T cells (α v β 6 negative) further confirm the specific binding of the ICG-α v β 6 antibody to integrin α v β 6 .
Our in vitro study also showed that 0.03 million cells can be detected by fluorescence imaging. If 10 million cells are assumed to be in a 100 mm 3 tumor 11 , the ICG-α v β 6 antibody could theoretically detect lesions as small as 0.3 mm 3 by fluorescence imaging. In vivo study showed xenograft cSCC tumors to be clearly visible by PAI and fluorescence imaging post injection of ICG-α v β 6 antibody. All these indicate that ICG-α v β 6 antibody is potentially useful in diagnosis of cSCC, including special types of SCC such as subungual squamous cell carcinoma 12 , because of the deep detection ability of PAI. Moreover, because high levels of integrin α v β 6 expression have been found at the invasive edge of carcinomas 13 , the ICG-α v β 6 antibody may increase the ability to obtain negative resection margins and visualize residual tumor during surgery by PAI and fluorescence imaging.
A dramatically increased photoacoustic signal 1 min after injection of probes demonstrates the arterial phase of the probes in blood circulation. The similar level of photoacoustic signal enhancement confirms that the same dose of probes (targeted ICG-α v β 6 antibody or untargeted ICG-ScrIgG) was injected into the subjects. A higher signal at other time points after injection with ICG-α v β 6 antibody compared to the ICG-ScrIgG and blocking groups confirms the targeted specificity of the ICG-α v β 6 antibody against tumor.
In a previous study 10 , a knot peptide of approximately 3.8 kDa was labeled with another organic dye, Atto 740, for targeted photoacoustic imaging of tumors in vivo. A 2-fold increase over the value before injection was observed. The ICG labeled antibody fabricated in this study with 150 kDa molecular weight achieved a 3-fold increase in photoacoustic signal. There exists similar binding affinity between the ICG-α v β 6 antibody (28.4 nM) and the labeled peptide A740-R 0 1 (39.4 nM). The relatively small size of the peptide and the resulting short circulation time in living bodies limits the signal enhancement. The slow-clearing antibody with long half-life exhibits high tumor accumulation 14,15 . This enables the ICG-α v β 6 antibody to be a robust probe for targeted imaging of integrin α v β 6 positive cSCC tumors by PAI and fluorescence imaging with high specificity.
There exists a difference between the signal enhancement rate by fluorescence imaging and PAI. Six-fold enhancement was achieved 24 h post-injection of the ICG-α v β 6 antibody in fluorescence imaging. In contrast, PAI showed a 3-fold change. This phenomenon may be attributed to the difference in the efficiency of energy transfer through living tissue via the different modalities and is consistent with previous research 10 .
ICG was selected in this study to label the antibody. Photobleaching may occur after ICG is exposed to light for 1 day 11 . On the other hand, under the irradiation of laser light at 710 nm and 900 nm for 30 min, no significant decrease of photoacoustic signal was observed. This indicates that ICG is relatively stable under certain scanning situations. The ICG labeled ICG-α v β 6 antibody probe will be prone to clinical transformation.
PAI with ICG-α v β 6 antibody allows for targeted imaging of cSCC tumors deep in tissues with high spatial resolution; however, high background signal from blood can limit the achievable detection sensitivity 16 . There are a variety of spectral processing methods to un-mix the signal of probe from blood 17 . In this study, a high photoacoustic signal at 710 nm and a very weak signal at 900 nm could be detected from the ICG-α v β 6 antibody. On the other hand, a relatively weak photoacoustic signal was detected from hemoglobin at both of these two wavelengths. A simple subtraction of images at these two different wavelengths can eliminate the signal from hemoglobin and thereby highlight the signal from the ICG-α v β 6 antibody 18 .
Several limitations of our study need to be addressed. Although un-mixing the signal of ICG-α v β 6 antibody by subtraction of 710 nm and 900 nm images may restrain blood signals, signals from other components, such as melanin and lipids, may not be totally restrained. We will attempt more sophisticated algorithmic analysis for the extraction of signal of the probe from tissues. The random conjugation of ICG to antibody may lead to a decrease of binding affinity of the ICG-α v β 6 antibody for integrin α v β 6 . Site-specific conjugation will be considered in future studies. An ultraviolet-induced early cSCC model in mice that has biologic character more similar to human cSCC will also be used. The efficacy and correlations between antibody lots should be assessed particularly before clinical application.
Our results suggest that ICG-α v β 6 antibody is a promising probe targeting the integrin α v β 6 for the detection of cSCC tumors by PAI and fluorescence imaging. It may find clinical application in diagnosis of cSCC as well as in intraoperative navigation.

Methods
Production of anti-α v β 6 antibody. The gene segments of the heavy-chain (HC) and light-chain (LC) encoding a derivative motif 10 targeting integrin α v β 6 were identified after bio-panning for phages in the phage display system 19 . The HC sequence was then inserted into the vector pFUSEss_CHIg_hG1 (Invivogen, San Diego, CA) between EcoRI and NheI restriction sites to form plasmid pFUSE_H. The LC sequence was inserted into the multiple cloning site of the vector pFUSE2ss-CLIg-hl2 (Invivogen) between EcoRI and AvrII restriction sites to form plasmid pFUSE_L.
The plasmids were used to transform E. coli competent cells. Zeocin (for HC) or blasticidin (for LC) resistant clones were cultured and plasmids isolated using a maxi kit (Sigma Aldrich, St Louis, MO). DNA sequencing was performed to confirm the correct sequence.
A scrambled gene sequence was inserted into the variable domain of HC for production of a scrambled IgG (ScrIgG) without specificity for α v β 6 .
The antibody was biosynthesized using FreeStyle ™ 293 Expression System (Thermo Fisher Scientific, Pittsburgh, PA). FreeStyle 293F cells were grown in DMEM supplemented with 10% FBS and penicillin/streptomycin (Thermo Fisher Scientific) to 90% confluence in plates. Cells were washed and then digested with 2 ml Higher maximum radiance in tumors was observed than in the non-targeted ICG-ScrIgG control group. No significant difference between the two groups was observed among other organs. Higher maximum radiance was observed in liver, followed by spleen, intestine and kidney.
Scientific RepoRts | 7:42442 | DOI: 10.1038/srep42442 trypsin. The suspended cells were centrifuged (1500 r.p.m, 5 min) to pellets and then resuspended with 10 ml of serum free FreeStyle ™ 293 expression medium. Cell concentrations were calculated with a Countess cell counter (Thermo Fisher Scientific). Cell viability should be greater than 90%. Cells were then separated into 125-ml square flasks with final viable cell concentration of 1 × 10 6 /ml and volume of 28 ml in each flask for transfection.
Sixty microliters of lipofectamine (Thermo Fisher Scientific) was mixed gently with Opti-MEM ® I to a total volume of 1 ml and incubated for 5 min at room temperature. Twelve micrograms of pFUSE_H and 18 μ g of pFUSE_L were mixed gently with Opti-MEM ® I to a total volume of 1 ml in a 2-ml Eppendorf tube. Then, the diluted lipofectamine was mixed into the plasmids gently and incubated for 30 min at room temperature. After the incubation, the 2 ml of lipofectamine-plasmid complex was added to the prepared flask of cells. Each flask should contain a total volume of 30 ml, with a final cell density of approximately 1 × 10 6 viable cells/ml. The cells were then incubated in a 37 °C incubator with a humidified atmosphere of 8% CO 2 in air. At 24 hours after transfection, valproic acid sodium salt (Sigma Aldrich, St. Louis, MO) was added into the cultures with final concentration of 2 mM. Seven days after transient transfection, the culture medium was harvested and centrifuged at 1500 r.p.m. for 5 min. The supernatant was purified with a NAb Protein G Spin Kit (Thermo Fisher Scientific). Sufficient antibody for the experiment was harvested at one time to maintain the same efficacy of the antibody. This means approximately 30 flasks of 293F cells (with 30 ml expression medium in each flask) were transfected simultaneously.
The quality of the preparations was analyzed on reducing and non-reducing SDS-PAGE. Molecular masses were confirmed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS; Foster City, CA). The eluents were condensed and the solvent was changed to 0.1 M NaHCO 3 (pH = 7.3) using Vivaspin 500 50 kDa MWCO spin membrane (Sartorius Biotech, Goettingen, Germany) and stored at 4 °C or − 20 °C for short-term (< 1 month) or long-term storage. Concentration of the purified antibody was detected with a Nanodrop 2000 UV-Vis spectrophotometer (Thermo Fisher Scientific).
ScrIgG was also manufactured at the same time in the same way. The ICG-α v β 6 antibody was also mixed with venous blood of mice in Eppendorf tubes and imaged by a Nexus 128 photoacoustic instrument (Endra, Boston, MA) at laser light 710 nm and 900 nm using continuous rotation (24 s total scan time, 240 views, 1 pulse/view). The data were analyzed using Osirix software (Apple, Cupertino, CA).

Synthesis and characterization of ICG
For the photobleaching study, 50 μ L of ICG-α v β 6 antibody (1 μ M) was analyzed with the photoacoustic instrument (Endra) and scanned at 710 nm and 900 nm 30 times, each scan lasting for 1 min. Photobleaching was determined by the change in photoacoustic intensity over time.
ScrIgG was also labeled in the same way to form ICG-ScrIgG as a non-targeted control.
Binding affinity of ICG-α v β 6 antibody. Competitive binding enzyme-linked immunosorbent assay (ELISA) was performed to measure the binding affinity of ICG-α v β 6 antibody as described elsewhere 20 . Briefly, each well of a Costar 96-well plate (Corning, Tewksbury, MA) was coated with 0.5 μ g of P2W7 antibody (anti-α v; Abcam, Cambridge, MA) at 4 °C overnight. Wells were washed with DPBS, blocked with blocking buffer (1% bovine serum albumin in DPBS) for 3 h and then washed. Fifty microliters per well of integrin α v β 6 (or integrin α v β 3 or integrin α v β 5 , 3 μ g/mL; R&D Systems, Minneapolis, MN) was added for 1 h and washed with wash buffer (TBS, 0.2% Tween 20; Sigma Aldrich). Serial dilutions (1 μ M-0.1 nM) of the anti-α v β 6 antibody, ICG-α v β 6 antibody or ICG-ScrIgG and biotinylated fibronectin (Sigma Aldrich) were incubated for 1 h and washed. Fifty microliters per well of extravidin HRP conjugate (Sigma Aldrich) was added to each well for 1 h and washed. 3,3′ ,5,5′ -Tetramethylbenzidine (TMB; Sigma Aldrich) was then added to each well and allowed to incubate for 15 min. Absorbance at 450 nm was measured in a plate reader (Synergy Mx; BioTek, Winooski, VT). The human squamous carcinoma cells A431 (integrin α v β 6 -positive) 10 were plated at 1 × 10 4 cells/cm 2 in dishes and grown to ~85% confluence. ICG-α v β 6 antibody or ICG-ScrIgG (0.4 μ M) was added and incubated at 37 °C for 4 h. After washing with DPBS three times, fluorescence images were collected on an Eclipse Ti-E inverted microscope (Nikon, Melville, NY). Human embryonic kidney 293 T cells (integrin α v β 6 -negative 10 ) were also incubated with ICG-α v β 6 antibody in the same way. Additionally, for the blocking study, a 10-fold molar excess of unlabeled antibody was added to final concentration of 4 μ M, 10 minutes before addition of ICG-α v β 6 antibody into the plates of A431 cells.
Sensitivity Study. Human squamous carcinoma cells A431 were plated at 1 × 10 4 cells/cm 2 in dishes and grown to ~85% confluence. The cells were incubated with 0.4 μ M ICG-α v β 6 antibody for 24 h at 37 °C. The cells were washed then detached with trypsin (Thermo Fisher Scientific) and centrifuged. The harvested cell pellets were re-suspended and serially diluted to 1.8, 0.9, 0.45, 0.23, 0.12, 0.06, 0.03 million in 100 μ L of DPBS for PAI and fluorescence imaging. Small Animal Imaging and Tissue Biodistribution. All animal experiments were performed in accordance with the National Guides for the Care and Use of Laboratory Animals approved by the Institutional Animal Care and Use Committee (IACUC).
Eight-week-old female nude mice (nu/nu; Charles River Laboratories, Boston, MA) were used in this study. Two million A431 (in 100 μ L DPBS) cells were injected subcutaneously in the left dorsum of the mice (n = 9). Mice bearing 0.5-0.8 cm xenografted tumors were scanned using the PAI system (Endra) at 710 nm and 900 nm using continuous rotation (24 s total scan time, 240 views, 1 pulse/view). The data were obtained before and 1 min, 1 h, 2 h, 4 h, 24 h, 48 h after administration of 200 μ L of probe (2.5 μ M) via tail-vein injection. The data were analyzed using Osirix software (Apple). The mice were also imaged using an optical imaging system (Bruker, Billerica, MA) for fluorescence imaging.
For in vivo blocking studies, a 30-fold molar excess of unlabeled anti-α v β 6 antibody was intravenously injected into the mice 10 min before administration of ICG-α v β 6 antibody.
Mice were sacrificed 48 h post injection of probe. The excised organs were imaged using an IVIS200 optical imaging system (Caliper Life Sciences). The data were analyzed using Living Image 4.0 software. The tumor tissue was frozen and cryosectioned at − 20 °C into 5-μ m slices, and fluorescence images were collected on an inverted microscope.
Statistical Analysis. All statistical analyses were performed with SPSS 20 (IBM, Chicago, IL). The data are presented as the mean ± standard deviation. Means were compared using independent-samples t test. P values of 0.05 or less indicated significance. Nonlinear regression was performed with GraphPad Prism 5 (GraphPad, La Jolla, CA).