Imaging with labeled monoclonal antibodies may be useful in detecting, staging, and monitoring tumors. Despite their high affinity and specificity, a critical limitation of antibody imaging is the high background signal due to prolonged clearance from the blood, which reduces the tumor-to-background ratio. To address this problem, we developed a molecular imaging probe consisting of multiple self-quenching fluorophores [Cy5.5 or Alexa Fluor 680 (Alexa680)] conjugated to a monoclonal antibody (trastuzumab) to synthesize Tra-Cy5.5(SQ) or Tra-Alexa680(SQ), respectively. This agent only becomes fluorescently “active” after cellular internalization but is quenched in the unbound state leading to high tumor-to-background ratios. The in vitro quenching capacity for both conjugates was ∼9-fold. In vivo imaging experiments were done in mice bearing both 3T3/HER-2+ and BALB/3T3/ZsGreen/HER-2− xenografts. Tra-Alexa680(SQ) produced specific enhancement in the 3T3/HER-2+ tumors but not in the HER-2− control tumors. However, Tra-Cy5.5(SQ) produced nonspecific enhancement in both 3T3/HER-2+ and control tumors. In conclusion, whereas Cy5.5-conjugates produced nonspecific results as well as rapid liver accumulation, conjugating multiple Alexa680 molecules to a single monoclonal antibody resulted in a near-infrared optical agent that activated within specific target tumors with high tumor-to-background ratio with considerable potential for clinical translation. [Mol Cancer Ther 2009;8(1):232–9]
Molecular imaging with antibodies has the potential not only to improve the detection of tumors but also to characterize them by their cell surface expression profiles (1, 2). However, antibody delivery to a tumor relies on the high binding affinity and the low off-rate of antibodies to their cell surface antigens as well as their abundant blood supply (3, 4) with leaky tumor vasculature leading to enhanced permeability and retention (EPR; refs. 5, 6), thus increasing antibody accumulation. Because the EPR effect depends only on the physical characteristics of the macromolecules injected and not on their binding characteristics, it often leads to nonspecific tumor uptake. To achieve specific antibody imaging, sufficient time for clearance of the unbound antibody is needed to reduce background signal resulting in favorable tumor-to-background ratios. Meanwhile, the long clearance times of antibodies make delayed imaging a necessity, raising practical issues with regard to patient and physician acceptance. Therefore, in vivo antibody-based target-specific molecular imaging is limited by the EPR effect and prolonged clearance times leading to reduced tumor-to-background ratio, which lowers both sensitivity and specificity.
Humanized antibodies, which are antigen-specific complementary determination region-grafted human IgG molecules, have been used for clinical cancer therapy because they produce antigen-dependent cellular cytotoxicity with minimal toxicity due to low immunogenicity. Therefore, the humanized antibody is a realistic choice as a targeting moiety for molecular imaging probes. However, imaging with humanized antibodies has achieved limited success. Despite their highly specific accumulation in target tumors, a critical limitation of humanized antibody imaging is the high background signal due to prolonged blood clearance, which reduces the tumor-to-background ratio. Of the clinically available imaging techniques for labeling antibodies, only positron emission tomography and single-photon emission computed tomography have been widely used in vivo and then only with long-lived isotopes. However, because positron emission tomography or single-photon emission computed tomography probes constantly emit signal (decreasing as a function of half-life of the radioisotope), EPR-related signal and background signal are quite high, especially when humanized antibodies are used. Therefore, to optimize the pharmacokinetics and clearance, genetic or enzymatic modifications of antibodies have been investigated; however, these alterations may reduce the therapeutic value of the antibody (2). Optically labeled antibodies, in theory, suffer from the same limitations as radioisotopes; however, optical probes differ because they can be activated or switched on only at the target cancer cells in response to specific intracellular environmental stimuli. By activating the fluorescence signal only within the target cells, nonspecific accumulation due to EPR and in the blood pool is minimized.
Several activatable optical probes have recently been reported (7–11). These are largely based on self-quenching mechanisms whereby enzymatic cleavage of fluorophores held in close steric alignment results in fluorescent activation as the fluorophores move away from each other. Among the various choices for imaging fluorophores, near-infrared (NIR) probes have the advantage of better depth penetration within tissue and are amenable to self-quenching (12). For instance, when two or more Cy5.5 dyes are conjugated to generation-6 polyamidoamine dendrimers, which are similar in hydrodynamic diameter to antibody molecules, self-quenching occurs and the degree of self-quenching increases as the number of Cy5.5 dyes increases (13). However, conjugation of multiple fluorophores to the same macromolecule risks altering the pharmacokinetics of the conjugate. Relatively few reports focus on activatable optical probes conjugated to antibodies.
In this study, we synthesized and tested a self-quenching activatable probe conjugated to a monoclonal antibody using cyanine-based NIR fluorophores, Alexa Fluor 680 (Alexa680) and Cy5.5, as described in Supplementary Fig. S1.1
Supplementary material for this article is available at Molecular Cancer Therapeutics Online (http://mct.aacrjournals.org/).
Materials and Methods
Trastuzumab, a Food and Drug Administration-approved humanized anti-HER-2 antibody, which has a complementary determination region against HER-2 grafted on a human IgG1 framework, was purchased from Genentech. Cy5.5-NHS ester was purchased from GE Healthcare. Alexa680-NHS ester was purchased from Invitrogen. ZsGreen plasmid was purchased from Clontech Laboratories. All other chemicals used were of reagent grade.
Structural Analysis of Cy5.5 and Alexa680
The chemical structure of Cy5.5 has been widely published (15). The structure of Alexa680 has been determined with mass spectroscopic analysis and nuclear magnetic resonance as shown in the Supplementary Figs. S3 and S4.1
Briefly, mass spectroscopic and tandem mass spectroscopic analyses were done with LCMS-IT-TOF systems (Shimadzu America). H- and C-nuclear magnetic resonance analyses were done with Gemini or a Mercury 300 MHz spectrometer (Varian). By summarizing both results with the reference from a patent document (16), the structure of Alexa680 was suggested as shown in Supplementary Fig. S1.1
Synthesis of Cy5.5- or Alexa680-Conjugated Antibodies
To synthesize the “always-on” control conjugate, trastuzumab (1 mg, 6.8 nmol) was incubated with Cy5.5-NHS (11 μg, 10 nmol, 5 mmol/L in DMSO) or Alexa680-NHS (16 μg, 14 nmol, 5 mmol/L in DMSO) in 0.1 mol/L Na2HPO4 (pH 8.5) at room temperature for 30 min. The mixture was purified with a Sephadex G50 column (PD-10; GE Healthcare). The protein concentration was determined with Coomassie Plus protein assay kit (Pierce Biotechnology) by measuring the absorption at 595 nm with a UV-Vis system (8453 Value UV-Visible Value System; Agilent Technologies). The concentration of Cy5.5 or Alexa680 was measured by absorption with the UV-Vis system to confirm the number of fluorophore molecules conjugated to each trastuzumab molecule. The number of Cy5.5 or Alexa680 per antibody was ∼1 in these control probes. The resulting compounds Tra-Cy5.5(ON) and Tra-Alexa680(ON) (the designation “ON” indicates that these probes are always active) were kept at 4°C in the refrigerator as a stock solution.
For the self-quenched conjugate, trastuzumab (1 mg, 6.8 nmol) was incubated with Cy5.5-NHS (77 μg, 68 nmol, 5 mmol/L in DMSO) or Alexa680-NHS (158 μg, 137 nmol, 5 mmol/L in DMSO) in 0.1 mol/L Na2HPO4 (pH 8.5) at room temperature for 30 min. Then, the mixture was purified and protein and Cy5.5 and Alexa680 concentrations were determined as described above. The number of Cy5.5 or Alexa680 per antibody was ∼7. The resulting conjugates Tra-Cy5.5(SQ) and Tra-Alexa680(SQ) (the designation “SQ” indicates that these compounds are self-quenched) were kept at 4°C in the refrigerator as a stock solution.
Determination of Quenching Ability In vitro
The quenching abilities of each conjugate were investigated by denaturalizing with 5% SDS and 2-mercaptoethanol. Briefly, the conjugates were incubated with 5% SDS and 1% 2-mercaptoethanol in PBS at 95°C for 2 min. As a control, the samples were incubated in PBS. The fluorescence signal intensity of each sample was measured with a fluorescence spectrometer (Perkin-Elmer LS55).
Measurement of the Lipophilicity of Cy5.5 and Alexa680
To evaluate the lipophilicity of Cy5.5 and Alexa680, the partition coefficient (logD) was determined. Cy5.5-NHS or Alexa680-NHS (50 nmol) was mixed with 1 mL each of 1-octanol and 0.1 mol/L phosphate buffer (pH 5.2, 7.3, and 8.5) in a test tube. Three tubes were used for each condition. The tubes were shaken intensely (3 × 1 min) and incubated for 20 min at room temperature, and the whole process was repeated twice to ensure that the reaction had reached equilibrium. Then, 1 mL aliquots of each phase were collected, and the concentration of Cy5.5-NHS or Alexa680-NHS was measured by absorption with the UV-Vis system. The distribution ratios were determined as the logarithm value of the octanol-to-buffer ratio (logD).
For HER-2 targeting studies, the HER-2 gene-transfected NIH3T3 (3T3/HER-2+) cell line was used. As a negative control, green fluorescence protein transfected BALB/3T3 cell line (BALB/3T3/ZsGreen/HER-2−) was employed. The BALB/3T3/ZsGreen cell line does not express HER-2 receptor. The cell lines were grown in RPMI 1640 (Life Technologies) containing 10% fetal bovine serum (Life Technologies), 0.03% l-glutamine, 100 units/mL penicillin, and 100 μg/mL streptomycin in 5% CO2 at 37°C.
Fluorescence Microscopy Studies
3T3/HER-2+ cells (1 × 104) were plated on a cover glass-bottomed culture well and incubated for 16 h. Then, Tra-Cy5.5 or Tra-Alexa680 was added to the medium (30 μg/mL), and the cells were incubated for either 1 or 8 h. Cells were washed once with PBS, and fluorescence microscopy was done using an Olympus BX51 microscope (Olympus America) equipped with the following filters: excitation wavelength 590 to 650 nm and emission wavelength 662.5 to 747.5 nm. Transmitted light differential interference contrast images were also acquired. To investigate the receptor specificity, the conjugates were also incubated with BALB/3T3 (HER-2−) cells.
All procedures were carried out in compliance with the Guide for the Care and Use of Laboratory Animal Resources (1996), National Research Council, and approved by the local Animal Care and Use Committee. Both receptor-positive and receptor-negative tumor cell lines were implanted in the mice. For HER-2 targeting studies, 3T3/HER-2+ cells (2 × 106 in PBS) were injected subcutaneously in the left dorsum of the mice, and BALB/3T3/ZsGreen cells (2 × 106 in PBS) were injected subcutaneously into the right dorsum. For orthotopic tumor models, 3T3/HER-2+ cells (1 × 106 in PBS) and BALB/3T3/ZsGreen cells (1 × 106 in PBS) were injected into the right and left mouse mammary pads, respectively. The experiments were done at 5 to 8 days after cell injection.
In vivo Spectral Imaging Studies
Tra-Cy5.5(ON), Tra-Cy5.5(SQ), Tra-Alexa680(ON), or Tra-Alexa680(SQ) (50 μg/100 μL PBS) were injected via the tail vein into tumor-bearing (3T3/HER-2+ and BALB/3T3/ZsGreen/HER-2−) mice. The mice were anesthetized with intraperitoneally administered 10% sodium pentobarbital with 0.1% scopolamine butylbromide, and spectral fluorescence images were obtained using the Maestro In vivo Imaging System (CRi) using two filter sets 2 days after the conjugate injection. The deep red filter sets were used to image Cy5.5 or Alexa680 fluorescence and the blue filter sets were used for ZsGreen fluorescence. The deep red filter sets use a band-pass filter from 642 to 680 nm (excitation) and a long-pass filter over 702 nm (emission); the blue filter sets use a band-pass filter from 437 to 476 nm (excitation) and a long-pass filter over 493 nm (emission). The tunable emission filter was automatically stepped in 10 nm increments from 650 to 950 nm for the deep red filter sets and from 500 to 800 nm for the blue filter sets while the camera captured images at each wavelength interval with constant exposure. The spectral fluorescence images consist of autofluorescence spectra and the spectra from Cy5.5, Alexa680, and ZsGreen, which were then unmixed based on their spectral patterns using commercial software (Maestro software; CRi). Mice were sacrificed with carbon dioxide immediately after completion of imaging. Then, the tumors were resected and ex vivo imaging was done using the same Maestro settings.
Chemical Structures of Cy5.5 and Alexa680 Indicates Differences in Nonspecific Protein Binding
The chemical structures of Alexa680 and Cy5.5 based on previously published results (16) and our own analysis (see supportive data and method for nuclear magnetic resonance and mass spectroscopy as shown in Supplementary Figs. S2-S4)1 are shown in Supplementary Fig. S1.1 In summary, Alexa680 is smaller with fewer aromatic rings than Cy5.5. Additionally, although Cy5.5 is negatively charged, Alexa680 has the neutral charge. From these chemical structures, it is predicted that Alexa680 is less lipophilic and charged with lower nonspecific protein binding than Cy5.5.
Lipophilicity of Cy5.5 Is Higher Than That of Alexa680
The partition coefficients of Cy5.5 and Alexa680 were obtained to determine their lipophilicity, which might alter the pharmacokinetics of the conjugated antibody. The resulting logD values (higher values indicating higher lipophilicity) are shown in Table 1. As predicted from the chemical structures, the lipophilicity of Cy5.5 was higher than Alexa680 for pH range of approximately 5 to 9 and was pH-independent, whereas the lipophilicity of Alexa680 was pH-dependent. Therefore, Cy5.5 would be predicted to more strongly alter pharmacokinetics by nonspecific protein binding.
|.||pH 5.2 .||pH 7.3 .||pH 8.5 .|
|Cy5.5||−1.07 ± 0.06||−0.98 ± 0.14||−1.06 ± 0.02|
|Alexa680||−1.91 ± 0.03||−1.72 ± 0.04||−1.42 ± 0.14|
|.||pH 5.2 .||pH 7.3 .||pH 8.5 .|
|Cy5.5||−1.07 ± 0.06||−0.98 ± 0.14||−1.06 ± 0.02|
|Alexa680||−1.91 ± 0.03||−1.72 ± 0.04||−1.42 ± 0.14|
NOTE: Mean ± SD of three experiments.
Quenching Capacity of Both Antibody-Fluorophore Conjugates Is High
The quenching capacity was measured by adding 5% SDS and 2-mercaptoethanol. The molecular interaction can be dissociated with SDS and the heavy and light chains of IgG can be separated from each other with 2-mercaptoethanol treatment. The quenching capacities were 9-, 2-, 8-, and 2-fold for Tra-Cy5.5(SQ), Tra-Cy5.5(ON), Tra-Alexa680(SQ), and Tra-Alexa680(ON), respectively.
Fluorescence Microscopy Shows Fluorescent Activation in Target Cells
Fluorescence microscopy studies were carried out to visualize the cellular binding location and behavior of fluorescently labeled antibody in vitro. All the investigated trastuzumab conjugates showed fluorescent signal on the surface of 3T3/HER-2+ cells 1 h after incubation (Fig. 2A). The fluorescence was observed inside the cell after 8 h incubation for all conjugates, and the fluorescent dots were brighter for the self-quenched conjugates than always-on conjugates, which were not self-quenching (Fig. 2A). The microscopic images were also obtained using HER-2− BALB/3T3 cells (Fig. 2B). Tra-Alexa680(ON), Tra-Alexa680(SQ), and Tra-Cy5.5(ON) showed no fluorescent signal, thus showing their specificity for HER-2-expressing tumors. However, Tra-Cy5.5(SQ) showed nonspecific fluorescence inside HER-2− BALB/3T3 cells after 8 h incubation probably because of nonspecific protein binding. These findings are well correlated and supported by the fluorescence-activated cell sorting analysis shown in the Supplementary Fig. S5.1
Self-Quenched Antibody-Alexa680 Conjugate Showed High Tumor-to-Background Ratio during In vivo Fluorescence Imaging, Whereas Cy5.5 Conjugates Did Not
The results of imaging with trastuzumab conjugates in both 3T3/HER-2+ (HER-2+) and BALB/3T3/ZsGreen/HER-2− tumor-bearing mice are summarized in Fig. 3. In the whole-body image (Fig. 3A), the targeted tumor (3T3/HER-2+) was greatly enhanced with Tra-Alexa680(SQ), whereas the control tumor (BALB/3T3/ZsGreen) was minimally enhanced. With Tra-Alexa680(ON), the NIR signal could be detected not only in the targeted tumor but also in the control tumor, although the signal was higher in the targeted tumor. The NIR signal was also shown around the control tumor with Tra-Alexa680(ON). On the other hand, the targeted tumor could not be clearly imaged with Tra-Cy5.5(SQ), and the NIR signal was higher for the control tumor, including the peripheral regions, indicating nonspecific binding. High NIR signal was observed in both the target and the control tumors with Tra-Cy5.5(ON). The resected tumors also showed the advantage of Tra-Alexa680(SQ), that is, the NIR from the labeled antibody in the HER-2+ tumors did not overlap with the ZsGreen fluorescence signal arising from the HER-2− tumors (Fig. 3B). With Cy5.5-conjugated trastuzumab, high NIR signal could be seen in nontarget tumor. In addition, the whole-body abdominal images showed reduced background for Tra-Alexa680(SQ) but high liver uptake with Tra-Cy5.5(ON) likely related to the latter's lipophilicity (Fig. 4). Similar results were obtained with the orthotropic mammary pad tumor model (Fig. 5).
The fundamental barriers to optical imaging in tissue are high light scattering, autofluorescence, high absorption by hemoglobin and other macromolecules, and reduced depth penetration with decreasing wavelength. One solution for these problems is to use a very bright fluorescence source such as quantum dots (17–19) or high-efficiency fluorescence proteins (20). In reality, even single cells labeled with fluorescence proteins could reportedly be tracked in vivo (21, 22). However, the physical sizes of such high-efficiency fluorescence sources strongly influence the biodistribution of the agent. As a result, it is difficult to use such agents as injectable fluorescent molecular imaging probes especially in concert with antibodies, which are themselves large molecules. Therefore, smaller organic fluorophores remain the preferred choice as labels for intravenous optical molecular imaging probes. Among the many organic fluorophores available, NIR dyes are often selected because of the advantages of reduced autofluorescence and improved tissue penetration compared with the visible dyes (12).
In this study, we investigated two kinds of NIR fluorophores attached to the monoclonal antibody trastuzumab: Cy5.5 and Alexa680. Trastuzumab is a complementary determination region-grafted humanized monoclonal antibody (IgG1 subclass), which binds to HER-2 expressed on the cell surface of some tumors, and the antibody has been highly successful in treating patients with breast cancers that have amplified expression of HER-2. Because the humanized antibody is made by using the human IgG1 framework and grafting 12 complementary determination region sequences, the advantage of using the same subclass of humanized antibody is that >98% of protein sequences have homology, even when the target antigen of the antibody is changed from HER-2 to other cancer-specific target molecules. Therefore, one would expect the optical features of the antibody conjugate to be the same regardless of the antigen used. This was shown when the self-quenching efficiency of the conjugate was identical, when we switched the antibody from trastuzumab to daclizumab, which is also a complementary determination region-grafted humanized monoclonal antibody (IgG1 subclass) targeting CD25 (interleukin-2 receptor α subunit). Indeed, the identical chemistry was used underscoring the flexibility of the concept. The quenching capacities were 12-, 2-, 10-, and 1-fold for Dac-Cy5.5(SQ), Dac-Cy5.5(ON), Dac-Alexa680(SQ), and Dac-Alexa680(ON), respectively.
The pharmacokinetic characteristics of humanized antibodies, with clearance times typically measured in weeks, are well matched to the self-quenching system described here (23). Prolonged circulating humanized antibody is advantageous from a therapeutic perspective because of the continuous delivery of drug to the target tumor; however, this becomes a disadvantage when trying to image with the same antibody labeled with radioisotopes or “always-on” fluorophores. By employing a self-quenching activatable strategy, the unbound circulating agent yields minimal signal, whereas the bound agent activates after internalization leading to high signal from target tumors.
Quenching was achieved by conjugating a large number (∼7) of Cy5.5 and Alexa680 to each antibody. In vitro microscopy revealed that fluorescence markedly increased after 8 h incubation due to internalization (Fig. 2). The quenching effect and the fluorescence increase after the internalization was also shown for Tra-Alexa680(SQ) with fluorescence-activated cell sorting flow cytometry studies (Supplementary Fig. S5).1 Among these conjugates, the most successful for in vivo imaging was Tra-Alexa680(SQ) (Fig. 3). The background fluorescence from the blood pool with this agent was low compared with Tra-Alexa680(ON) (Fig. 4). In contrast, Tra-Alexa680(ON) resulted in signal from nontarget tumor as well as target tumor. Although the rich blood supply and EPR effect might cause this nonspecific fluorescence with an agent, which is always on (always fluoresces), the self-quenched conjugates were able to activate only within tumor cells, thus taking advantage of the EPR effect without suffering the consequences of possible nonspecific uptake. The identical findings were observed when the Dac-CD25 targeting tumor system was employed (Supplementary Fig. S2).1
Tra-Cy5.5(SQ) was less successful as a tumor-specific imaging agent. The multiplicity of Cy5.5 molecules altered the biodistribution and pharmacokinetics of the antibody. For instance, as shown in Fig. 4, fluorescence was detected in liver immediately after intravenous injection of Tra-Cy5.5(SQ). In comparison, the fluorescence in the liver was substantially lower for Tra-Alexa680(SQ) and even Tra-Cy5.5(ON). The rapid uptake by the liver could sequester Tra-Cy5.5(SQ) and prevent it from binding to the HER-2 receptor in the tumor. The lipophilicity and charge of Cy5.5 likely results in its uptake and metabolism in the liver. Because the lipophilicity was higher for Cy5.5 than Alexa680 (Table 1), the pharmacokinetics could be largely affected by the conjugated fluorophore itself because of the large number of fluorophore molecules on a single antibody. The minimal liver accumulation with Tra-Cy5.5(ON) supports this hypothesis.
In addition, it is known that nonspecific binding to plasma protein increases with lipophilicity (24). In microscopic studies, Tra-Cy5.5(SQ) showed nonspecific uptake for HER-2− BALB/3T3 cells. The nonspecific protein binding of Cy5.5 may be related to its lipophilicity, although other factors (e.g., charge and molecular weight) might also affect binding.
In conclusion, we have successfully shown an antibody-targeted, activatable NIR molecular imaging probe based on self-quenching. The high quenching ability of Tra-Alexa680(SQ) and low nonspecific binding derived from its more hydrophilic character leads to a high tumor-to-background ratio in both the xenograft and the orthotopic models of cancer. Humanized antibodies conjugated with self-quenched Alexa680 can image HER-2 and CD25 in vivo without the disadvantages typical of antibody imaging. This technology could be universally applied to humanized antibodies of a similar class (IgG1).
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant support: Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
We thank Drs. Masatoshi Takahashi and Masayuki Nishimura (Shimadzu Scientific Instruments) for great assistance for the various mass spectroscopic analyses of fluorescence dyes.