Optical imaging probes with targeting ligands directed at cell surface markers overexpressed on cancers have the potential to detect and characterize tumors in vivo with high sensitivity and specificity.1, 2, 3, 4 Such probes could help direct surgery5 or guide nonconventional treatments, such as photodynamic therapy.6 However, malignant tumors are notorious for their heterogenous and diverse expression of cell surface markers;7 the degree and variety of target molecule expression may vary between and within tumor foci. Moreover, there is inevitably some low-level expression in normal tissue. However, by utilizing a strategy that employs more than one type of targeting ligand in a single imaging session, it should be possible to detect more tumor volume with great specificity than can be accomplished with a probe directed at a single target. Thus, if the simultaneous diagnosis of several targets on a tumor cell were possible in vivo, it could improve both tumor detection and characterization, potentially leading to more effective targeted therapy.
Herein, we describe the simultaneous use of two optical probes, each targeting a different cell surface marker and each conjugated to a different optical fluorophore. We employed two human ovarian cancer cell lines, SHIN38 and SKOV3,9 which produce peritoneal dissemination when introduced into mice via intraperitoneal injection. Although both cell lines express D-galactose receptor (D-galR),10 only SKOV3 also overexpresses HER2/neu.9 We chose these receptor systems to demonstrate simultaneous multitarget molecular imaging because both of them are feasible for clinical application. D-galR is expressed by various ovarian cancer cells,10 therefore, it is a good target for developing fluorescence-enhanced surgical guidance for disseminated peritoneal ovarian cancer. HER2/neu is a well-established cancer cell surface marker expressed on various cancers, including breast, colon, and ovarian. A humanized monoclonal antibody, trastuzumab, has been approved for a molecular-specific target cancer reagent by the FDA for patients with breast cancers that overexpress HER2/neu and is used commonly in clinical practice. We employed galactosyl serum albumin (GSA) conjugated Rhoda mine Green (RhodG) to bind D-galR-expressing cells and trastuzumab conjugated Alexa680 to bind HER2/neu-expressing cells.2, 3 Given that SKOV3 expresses both targets, we predicted these cells would bind both GSA-RhodG and trastuzumab-Alexa680 while SHIN3 tumors, expressing only D-galR would only bind GSA-RhodG. Using this strategy we demonstrate multiple targeting to a single tumor type and characterize the distribution of the surface marker expression in vivo.
Materials and Methods
Cell Lines and Culture
Two established human ovarian cancer cell lines (SKOV3 and SHIN3) were used in this study. SKOV3 overexpresses HER2/neu,9 but both express D-galR.10 SHIN3 cells were transfected with a plasmid expressing red fluorescent protein (RFP) to create a red fluorescent phenotype (SHIN3/RFP) that permitted optical identification of SHIN3-derived tumor implants, as previously described.11 All cell lines were grown in RPMI 1640 medium (Invitrogen Corporation, Carlsbad, CA) containing 10% fetal bovine serum [(FBS), Invitrogen Corporation], 0.03% L-glutamine at , penicillin, and streptomycin in 5% .
Synthesis of Rhodamine Green-Conjugated GSA
GSA was purchased from Sigma Chemical (St. Louis, MO), amino-reactive Rhodamine Green (RhodG-NHS) was purchased from Invitrogen Corporation. At room temperature, of GSA in was incubated with of RhodG-NHS, at pH 8.5 for . The mixture was purified with a gel filtration column (Sephadex G50 column, PD-10; GE Healthcare, Piscataway, NJ) and the GSA binding fraction was eluted by PBS at pH 7.4 and collected in a test tube. GSA-RhodG was kept at in the refrigerator as stock solutions. The protein concentrations of GSA-RhodG samples were determined with the Coomassie Plus protein assay kit (Pierce Biotechnology, Rockford, IL) by measuring the absorption at with a ultraviolet-visible (UV-vis) system (8453 Value UV-Visible Value System, Agilent Technologies, Santa Clara, CA) using standard solutions of known concentrations of GSA . The concentration of RhodG was then measured by absorption at with the UV-vis system to confirm the number of fluorophore molecules conjugated to each GSA molecule. The number of fluorophore molecules per GSA was adjusted to approximately 3–3.5.
Synthesis of Rhodamine Green and Alexa680-Conjugated Trastuzumab
Trastuzumab (Herceptin®), an FDA-approved humanized anti-HER2 antibody, which has a complimentary determination region against HER2 grafted on a human IgG1 framework, was purchased from Genentech Inc. (San Francisco, CA). RhodG-NHS and Alexa680-NHS were purchased from Invitrogen Corporation. At room temperature, of trastuzumab in was incubated with of RhodG-NHS or Alexa680-NHS, at pH 8.5 for . The mixture was purified with a gel filtration column (Sephadex G50 column, PD-10; GE Healthcare, Piscataway, NJ) and the antibody binding fraction was eluted by PBS at pH 7.4 and collected in a test tube. Trastuzumab-RhodG and trastuzumab-A680 were kept at as stock solutions. The protein concentrations of samples were determined with the Coomassie Plus protein assay kit (Pierce Biotechnology, Rockford, IL) by measuring the absorption at with a UV-vis system (8453 Value UV-Visible Value System, Agilent Technologies) using standard solutions of known concentrations of trastuzumab . The concentration of RhodG and Alexa680 was then measured by absorption at 503 and , respectively, with the UV-vis system to confirm the number of fluorophore molecules conjugated to each trastuzumab molecule. The number of fluorophore molecules per trastuzumab was adjusted to approximately 3–3.5.
In Vitro Fluorescence Microscopy
Fluorescence microscopy was performed with an Olympus BX61 microscope (Olympus America, Inc., Melville, NY) equipped with the following filters: for the blue light filter, a bandpass filter from and a bandpass filter from were used for excitation and emission light, respectively; for the green light filter, the values were from and from , respectively, and for the red light filter, the values were from and from , respectively. Transmitted light differential interference contrast (DIC) images were also acquired.
Either SHIN3 cells or SKOV3 cells were plated on a cover glass–bottomed culture well and incubated at in an atmosphere of 5% for . GSA-RhodG , trastuzumab-Alexa680 , or both were added to the culture media. The cells were incubated and removed at the following time points: 1, 8, . Following removal, the cells were washed once with PBS, followed by fluorescence microscopy with the blue and red filter sets. To better localize the optical probes intracellularly, we also performed a colocalization study, adding of the lysosomal marker (LysoTracker Red DND-99, Molecular Probe Inc., Eugene, OR) prior to imaging SKOV3 cells treated with either GSA-RhodG or trastuzumab-RhodG. LysoTracker selectively accumulates in the acidic compartments of the cell (i.e., the lysosome) and emits red fluorescence.12 Fluorescence microscopy was performed with the blue, green, and red light filters at and time points.
To demonstrate dual targeting in vitro, a coculture of SKOV3 and SHIN3/RFP cells was created. Both SHIN3-RFP cells and SKOV3 cells were plated on a cover glass–bottomed culture well and incubated at in a 5% atmosphere for . Both GSA-RhodG and trastuzumab-Alexa680 were added to the culture media. The cells were incubated and removed at 1 and . Following removal, cells were washed once with PBS and fluorescence microscopy was immediately performed with the blue, green, and red light filters.
All procedures were approved by the National Cancer Institute Animal Care and Use Committee. The tumor implants were established by intraperitoneal injection of cells suspended in of PBS in female nude mice (National Cancer Institute Animal Production Facility, Frederick, MD). Experiments with tumor-bearing mice were performed at for the SHIN3 mouse model and for the SKOV3 mouse model. Also, to establish a coincident tumor model, another group of mice received the intraperitoneal injection of SKOV3 cells ( cells) followed seven days later by an intraperitoneal injection of SHIN3/RFP cells ( cells). Experiments with SKOV3 and SHIN3/RFP tumor-bearing mice were performed at after injection of the SKOV3 cells.
In Vivo Spectral Fluorescence Imaging
After establishing the intraperitoneal dissemination model, groups of mice (four per group) with SHIN3, SKOV3, or the coincident model of SHIN3/RFP and SKOV3 were injected i.p. with of trastuzumab-Alexa680 diluted with PBS, prior to imaging, followed by of GSA-RhodG diluted with PBS, prior to imaging. Mice were sacrificed with carbon dioxide, and then the abdominal cavities were exposed and the peritoneal membranes were spread on nonfluorescent black plates. Spectral fluorescence images were obtained using the Maestro In Vivo imaging system (CRi, Inc., Woburn, MA). Magnified multicolor spectral fluorescence images of the peritoneal membranes were obtained with a multiexcitation acquisition13 using the blue-filter setting (excitation filter: , emission filter: long pass) and the red-filter setting ( , long pass) for the single-injected model, and the blue-filter, green-filter ( , long pass), and red-filter settings for the coincident tumor (SKOV3 and SHIN3/RFP) model. The tunable filter was automatically stepped in increments from for the blue-filter settings and from for red-filter settings, while the camera sequentially captured images at each wavelength interval. The spectral fluorescence images consisting of autofluorescence spectra, RhodG, RFP, and Alexa680 were unmixed, using commercial software (Maestro software, CRi).
All spectral fluorescence images of tumor-bearing mice with a single tumor type underwent a region-of-interest (ROI)–based analysis. First, all nodules, whose long axis was , were identified on either unmixed RhodG or Alexa680 spectral images. Then, single maximum-sized ROIs were placed over those nodules using Image J software (http://rsbweb.nih.gov/ij/), and the average fluorescence intensity of each ROI was calculated. A “positive” nodule was defined as having an average fluorescence intensity of arbitrary units on unmixed images, whereas a “negative” nodule was defined as having an average fluorescence intensity of arbitrary units. The number of positive nodules for both RhodG and Alexa680 and the number of positive nodules for either RhodG or Alexa680 were recorded.
In Situ Fluorescence Microscopy
After the surgical procedure described in the previous section, the abdominal contents were isolated and spread on a glass slide, and then subjected to in situ fluorescence microscopy using an Olympus BX61 microscope (Olympus America, Inc., Melville, NY). DIC imaging was used to guide spectral fluorescence imaging, which was performed either with a two-filter set for the single-tumor model or a three-filter set for the coincident tumor model. Only nodules sized from on DIC images were included in this analysis.
In Vitro Fluorescent Microscopic Study
Serial observations of SHIN3 and SKOV3 cells incubated with GSA-RhodG, trastuzumab-Alexa680, or the both were done using fluorescence microscopy (Fig. 1, 2, 3, 4 ). Fluorescence microscopic images of SHIN3 cells demonstrated binding of GSA-RhodG to SHIN3 cells denoted by small, peripheral fluorescent dots within cells at postincubation [Fig. 1a]. After , the bright fluorescent dots had coalesced and remained present in the cytoplasm at the time point. Additionally, GSA-RhodG demonstrated a similar phenomenon when incubated with SKOV3 cells. Incubation of the two cell types individually with trastuzumab-Alexa680 failed to reveal binding of this probe to SHIN3 cell at any of the time points [Fig. 1a], whereas SKOV3 cells demonstrated cell-surface binding of trastuzumab-Alexa680 at [Fig. 1b]. The specific binding of each reagent was demonstrated with experiments with the similar setting but with blocking by addition of 100-fold excess nonlabeled reagent as shown in Fig. 5 . At the probe was internalized, as evidenced by the presence of numerous intracellular fluorescent foci homogenously distributed in the cytoplasm, and remained present when imaged at the time point. These studies revealed that binding of GSA-RhodG to SKOV3 cells and SHIN3 cells was comparable while trastuzumab-Alexa680 bound only to SKOV3 cells. Additionally, even in the coculture model of SKOV3 and SHIN3/RFP cells, SKOV3 cells were depicted by both GSA-RhodG and trastuzumab-Alexa680, while SHIN3/RFP cells were depicted only by GSA-RhodG [Fig. 1c].
Fluorescence microscopy of SKOV3 cells exposed to either GSA-RhodG or trastuzumab-RhodG and incubated with Lysotracker was performed to localize the optical probes within the cell (Fig. 6 ). At postincubation, trastuzumab-RhodG bound only to the cell surface, and did not colocalize with the Lysotracker [Fig. 6a]. However, by postincubation, trastuzumab-RhodG had been internalized and could be identified as small dots, which colocalized to the lysosome. These results indicate that trastuzumab-RhodG binds to HER2/neu receptor expressed on the cell surface and then the complex is internalized forming an endosome, later fusing with the lysosome to form an endolysosome structure. A similar phenomenon was seen with incubation with GSA-RhodG at postincubation, as small fluorescent dots were identified and corresponded to lysosomes labeled by Lysotracker [Fig. 6b].
In Vivo Multicolor Spectral Fluorescence Imaging with Single-Tumor Model
Seventy-one tumor nodules could be identified on unmixed spectral images in four mice with SHIN3 tumors, while 69 tumor nodules could be identified on unmixed spectral images in four mice with SKOV3 tumors. The mean size of the nodules was in the SHIN3 group, and in SKOV3 group. In the SHIN3 group, all 71 tumor nodules had emission spectra corresponding only to GSA-RhodG ( : 100%) and no tumor nodules had an emission spectra corresponding to trastuzumab-Alexa680 [Fig. 7 and 8a ]. In the SKOV3 group, 64 nodules were successfully targeted with both targeting probes, visualized by the presence of a distinct spectra pattern comprised of both GSA-RhodG and Tratuzumab-Alexa680 spectra ( : 94.2%) [Fig. 8b]. However, five nodules failed to demonstrate the distinct emission spectra; three nodules were positive only for GSA-RhodG, while two nodules were positive only for tratuzumab-Alexa680.
In Situ Fluorescence Microscopy Single-Tumor Model
Fifty-nine tumor nodules could be identified on DIC images in four mice with SHIN3 tumors, while 36 tumor nodules could be identified on DIC images in four mice with SKOV3 tumors. Nodule size averaged in the SHIN3 group, and in the SKOV3 group. In the SHIN3 group, all 59 tumor nodules were depicted with GSA-RhodG and no signal corresponding to trastuzumab-Alexa680 was observed ( ; 100%) [Fig. 9a ]. In the SKOV3 group, all 36 nodules could be depicted with both GSA-RhodG and trastuzumab-Alexa680 ( ; 100%) [Fig. 9b].
In Vivo Multicolor Spectral Fluorescence Imaging and In Situ Fluorescence Microscopy in a Combined SKOV3 and SHIN3/RFP Tumor Model
After establishing a reproducible mixed-tumor model composed of SKOV3 and SHIN3/RFP, we injected both GSA-RhodG and trastuzumab-Alexa680 into mixed-tumor–bearing mice to determine the capability of multiple targeting to distinguish SKOV3 containing tumors from SHIN3 containing tumors in vivo and in fusion tumors composed of both cell lines in the same animal. In vivo multicolor spectral fluorescence imaging showed that tumor nodules from SKOV3 lesions were easily identified by detection of Alexa680 and RhodG, but not the RFP spectrum that was the intrinsic marker for SHIN3 [Fig. 10a ]. SHIN3/RFP lesions demonstrated RhodG and RFP spectra, but no Alexa680 spectrum. Furthermore, in situ fluorescence microscopy confirmed these results [Fig. 10b].
In this study, we demonstrated the ability of two distinct optical probes, GSA-RhodG and trastuzumab-Alexa680, targeting different cell-surface markers, D-galR and HER2/neu, respectively, to distinguish two different cell lines of human ovarian cancer in vivo. Thus, SKOV3 cells, which express both D-galR and HER2/neu, can be distinguished from SHIN3 cells, which express only D-galR, even in tumor nodules that are composed of both cell types. Although such distinctions have been made in vitro, only recently has multitargeted in vivo imaging become feasible. This is made possible by multicolor spectral fluorescence imaging and fluorescence microscopy, which permit the spectral separation of RhodG and Alexa680 by using different excitation and emission filter sets as well as by using multiplexed targeted optical probes.
Muticolor fluorescence imaging has been performed mostly in the visible light range with fluorescent proteins.14 With injectable fluorescent probes, a wider variety of wavelengths can be used for imaging. However, the only reason why Alexa680, a near-infrared fluorophore, was used was that it provided adequate spectral separation from RFP, which was the marker of SHIN3 cells. For surface applications, surgery assistance, or endoscopy, visible fluorescence is advantageous compared to near-infrared fluorescence. Because we used RFP as a validation tool, RhodG was a natural choice as an exogenous fluorophore because there was sufficient spectral separation between these two fluorophores. TAMRA, another desirable fluorophore,15 has an emission spectra that overlaps with RFP. However, in clinical applications, where RFP is not present, the combination of TAMRA and RhodG might be ideal for dual targeting.
Although single target probes have demonstrated remarkably high specificity and sensitivity, false positives persist.3 One advantage to using multiplexed targeted optical probes for tumor detection is that this approach reduces the number of false positives by increasing the specificity for tumor detection compared to a single-target-probe approach. Another advantage of this strategy is that it enables the detection and characterization of different types of tumor lesions, as shown in Fig. 10, in which all tumor nodules were detected by GSA-RhodG (a broadly sensitive probe), but only HER2/neu-expressing cells were detected with trastuzumab-Alexa680 (a highly specific probe). This strategy is akin to “in vivo immunohistochemistry” commonly used in modern pathology labs. Moreover, by changing the combination of targeted optical probes, it might be possible to distinguish a tumor lesion from surrounding inflammatory changes that are frequently found in clinical specimens (e.g., lung cancer associated with pneumonia), if probes specific for inflammatory markers were developed for optical imaging. Therefore, if combined with new technical advances in real-time fluorescence cameras,5, 16 a multiplex targeting approach could enable more detailed identification of tumors in real time and be useful during surgery or endoscopy.
In this study, there were five false negative nodules that were not detected with in vivo multicolor spectral fluorescence imaging. Although these tumor nodules yielded a signal, it was too weak and below the assigned cutoff value such that these tumors were not diagnosed correctly with the in vivo camera. It is likely that the delivery of the imaging agent and/or excitation light to these tumors was insufficient due to the particular anatomic location or shape of these specific tumors. This problem would be less likely to occur during fluorescence microscopy because of the restricted and flat field of view and homogeneously strong lighting achievable with microscopy. With this in mind, it would be more likely to detect a lesion with fluorescence microscopy than with in vivo imaging, leading to false-negative results.
In conclusion, a method for in vivo simultaneous multitargeted optical imaging of different receptors was successfully established. This method may have implications for improving the diagnostic accuracy and therapeutic efficacy in cancer diagnosis and treatment.
This research was supported by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research.