Development of a Universal Anti–Polyethylene Glycol Reporter Gene for Noninvasive Imaging of PEGylated Probes

Cells and Animals

NIH 3T3 mouse fibroblasts and human HeLa cervical cancer cells were purchased from American Type Culture Collection. EJ human bladder carcinoma cells and CNL Chang liver cells were obtained from Academia Sinica. The cells were cultured in Dulbecco minimal essential medium (Sigma–Aldrich) supplemented with 10% heat-inactivated bovine calf serum, penicillin (100 units/mL), and streptomycin (100 μg/mL) at 37°C in a humidified atmosphere of 5% CO₂. Female BALB/c nude mice (6–8 wk old) and female BALB/c mice (6–8 wk old) were obtained from the National Laboratory Animal Center, Taipei, Taiwan. All mice receiving radioiodinated probes were pretreated with Lugol solution (Sigma) in their drinking water for 2 d to block thyroidal uptake of free radioiodine (18). All animal experiments were performed in accordance with institutional guidelines and approved by the Animal Care and Use Committee of the Kaohsiung Medical University.

Generation of Anti-PEG Reporter–Expressing Cells

The V_L-Cκ and V_H-C_H1 domains of the anti-PEG reporter were cloned from complementary DNA prepared from the AGP3 hybridoma (19) following a previously described method (10). Primers used in the cloning of V_L-Cκ and V_H-C_H1 are as follows: V_L-Cκ sense: 5′-tgctggggcccagccggccgatattgtgttgacgcaggct-3′; V_L-Cκ antisense: 5′-ccgctcgagacactcattcctgttgaagct-3′; V_H-C_H1 sense: 5′-gaagatctgaagtgcagctggtggagtct-3′; and V_H-C_H1 antisense: 5′-caggtcgacagctggaatgggcacatgcag-3′. The V_L-Cκ and V_H-C_H1 genes were joined by a composite furin-2A protease cleavage site (20) and fused to the complementary DNA sequence encoding the C-like extracellular-transmembrane-cytosolic domains of the mouse B7-1 antigen (eB7) in the pLNCX-eB7 retroviral vector (10) to generate pLNCX-anti-PEG-eB7. A plasmid (pLNCX-anti-DNS-eB7) that encoded a membrane Fab with specificity for 5-(dimethylamino)naphthalene-1-sulfonyl chloride (DNS) (21) was constructed in an analogous fashion. EJ/anti-PEG or EJ/anti-DNS cells that stably expressed approximately equivalent levels of anti-PEG or anti-DNS reporters were generated by retroviral transduction as previously described (22). Characterization of the EJ/anti-PEG or EJ/anti-DNS cells by Western blotting and by flow cytometry is described in the Supplemental Methods (supplemental materials are available online only at http://jnm.snmjournals.org).

Synthesis of PEG-NIR797 and 4-Arm ¹²⁴I-PEG-SHPP, ¹²⁴I-FIAU Imaging Probes

The protocols of probe syntheses are provided in the Supplemental Methods. The 4-arm ¹²⁴I-PEG-N-succinimidyl-3-(4-hydroxyphenyl)propionate (SHPP) and ¹²⁴I-2′-fluoro-2′-deoxy-beta-d-arabinofuranosyl-5-iodouracil (FIAU) were more than 95% pure, with specific activities calculated to be 27.8 and 1.35 GBq/μmol, respectively. The half-life of 4-arm ¹²⁴I-PEG-SHPP was about 8.2 min.

Optical Imaging of Anti-PEG Reporter In Vitro and In Vivo

EJ/anti-PEG and EJ/anti-DNS cells (5 × 10⁵) were stained with 0.02, 0.1, and 0.5 μM PEG-NIR797 in phosphate-buffered saline (PBS) containing 0.05% bovine serum albumin on ice for 45 min. The cells were washed with cold PBS, and the surface immunofluorescence was measured on an IVIS50 optical imaging system (excitation, 760 nm; emission, 835 nm) (Caliper Life Sciences). For in vivo imaging, BALB/c nude mice bearing established EJ/anti-PEG and EJ/anti-DNS tumors (200 mm³) in their right and left hind leg regions, respectively, were intravenously injected with PEG-NIR797 fluorescent probes (2 mg/kg of body weight), and the optical images were acquired with an IVIS50 at 4, 24, and 48 h after injection. The regions of interest of tumor areas were drawn and analyzed with Living Image software, version 2.50 (Caliper Life Sciences).

T2-Weighted MRI of Anti-PEG Reporter In Vitro and In Vivo

EJ/anti-PEG and EJ/anti-DNS cells (3 × 10⁶) were incubated with 48, 24, 12, and 6 μM PEG₂₀₀₀ conjugated superparamagnetic iron oxide (PEG-SPIO, kindly provided by Yun-Ming Wang of the National Chiao Tung University, Taiwan) on ice for 30 min. After being washed with cold PBS, the PEG-SPIO labeled cells were collected by centrifugation (1,500 rpm) and imaged with a clinical 3.0-T MRI scanner (Sigma; GE Healthcare) equipped with a high-resolution head coil. All samples were measured by a T2-weighted spin-echo sequence (repetition time, 2,500 ms; echo time, 90 ms).

For in vivo imaging, BALB/c nude mice bearing established EJ/anti-PEG and EJ/anti-DNS tumors (300 mm³) in their right and left hind leg regions, respectively, were intravenously injected with PEG-SPIO (10 mg of iron/kg of body weight). Pentobarbital-anesthetized mice were sequentially imaged at 1, 2, and 24 h after injection with a 7-T horizontal-bore magnet (Bruker Instruments). A T2-weighted spin-echo sequence (repetition time, 2,500 ms; echo time, 90 ms) was used for MRI.

In Vivo Imaging of Anti-PEG and HSV-tk Reporters by Small-Animal PET

BALB/c nude mice bearing established EJ/anti-PEG and EJ/anti-DNS or EJ/TK (an EJ cell line expressing functional HSV-tk) tumors (200 mm³) in their right and left hind leg regions, respectively, were anesthetized with halothane vapor and then intravenously injected with 3,700 kBq (in 100 μL) of 4-arm ¹²⁴I-PEG-SHPP or ¹²⁴I-FIAU. The mice were sequentially imaged at 4, 24, and 48 h after injection with a micro-PET R4 scanner (Concorde Microsystems), following an imaging protocol previously described (18).

Immune Response to Anti-PEG and HSV-tk Reporter Proteins

The immunogenicity of the anti-PEG and the HSV-tk reporters was compared using a previously described protocol (22). Briefly, BALB/c mice were intravenously injected with 20 μg of pLNCX (vector control, n = 6), pLNCX-anti-PEG-eB7 (n = 8), or pLNCX-HSV-tk (n = 8). Preimmune and immune (day 10) sera (diluted 1:200 in PBS) were added to the wells of microtiter plates precoated with BALB 3T3 cells transfected with pLNCX, pLNCX-anti-PEG-eB7, or pLNCX-HSV-tk-eB7 (which codes for a membrane form of HSV-tk). Binding of antibodies on the cells was detected by enzyme-linked immunosorbent assay as previously described (22). All readings were background-adjusted by subtracting the absorbance of the wells incubated with preimmune sera.

Construction of Humanized Anti-PEG Reporter

The anti-PEG reporter was humanized using a previously reported protocol with minor modifications (23). Briefly, to select human framework sequences for complementarity-determining-region grafting, we compared the V_H and V_L sequences of the anti-PEG Fab with the National Center for Biotechnology Information database of human immunoglobulin germline sequences in the variable and joining regions using the IgBLAST program (http://www.ncbi.nlm.nih.gov/igblast/). DP35-JH4 and A2c-JK2 were found to be the most homologous to the V_H and V_L of anti-PEG reporter, respectively. For construction of the humanized V_L, the complementarity-determining regions of anti-PEG V_L, determined by the rule of Kabat et al. (24), was grafted onto human A2c-JK2. For construction of the humanized V_H, the complementarity-determining regions and 1 FR3 residue (Ser73) of anti-PEG V_H were similarly grafted onto DP25-JH4. The humanized anti-PEG V_L and V_H segments were synthesized by assembly polymerase chain reaction (23). The pLNCX-human anti-PEG-eB7 plasmid and tumor cells expressing humanized anti-PEG reporters (EJ/human anti-PEG and Hela/human anti-PEG) were constructed as described above.

In Vivo Imaging of Human Anti-PEG Reporter

The human anti-PEG reporter was examined in both subcutaneous and metastatic tumor models in vivo. For imaging the subcutaneous tumor model, BALB/c nude mice (n = 3) bearing established EJ/human anti-PEG and EJ/anti-DNS tumors (200 mm³) in their right and left hind leg regions, respectively, were imaged by small-animal PET as described above.

For imaging the metastatic tumor model, BALB/c nude mice (n = 3) were intravenously injected with Hela or Hela/human anti-PEG cells (5 × 10⁶ cells). After 3 wk, the mice were intravenously injected with PEG-NIR797 fluorescent probes (2 mg/kg of body weight). Whole-body images and lung-tissue images of these mice were acquired by the IVIS50 optical imaging system (excitation, 760 nm; emission, 835 nm) at 48 h after injection.

Statistical Analysis

Differences in tumor uptake of PEG-NIR797 were compared by the signed rank test. Data were considered significant at a P value of 0.05 or less. The cytotoxicity of iodinated 4-arm PEG-SHPP and FIAU and the immunogenicity of anti-PEG and HSV-tk reporters were compared by 1-way ANOVA using InStat software (version 3.0; GraphPad Software) followed by the Student t test. Bonferroni adjustment was used to adjust the significance level for multiple (3) comparisons. A P value of no more than the Bonferroni-corrected significance level (0.05/3 = 0.0167) was considered significant.

Characterization of Surface Anti-PEG Reporter

We constructed a membrane-anchored Fab with specificity to PEG as a reporter gene, or anti-PEG reporter. The light chain (V_L-C_K) is joined—through a furin cleavage site and 2A peptide (20)—to the heavy chain (V_H-C_H1), which is itself fused to the C-like extracellular-transmembrane-cytosolic domains of the mouse B7-1 antigen (Fig. 1A). A membrane-anchored Fab with specificity for DNS was constructed in an analogous fashion. The DNS Fab does not bind PEG and was used as a negative control reporter.

To analyze the anti-PEG and the control anti-DNS reporters, total cellular, membrane, and cytosolic proteins were separated by SDS-PAGE under reducing conditions and analyzed by Western blotting using antibodies to the hemagglutinin epitope present in the light chain (Supplemental Fig. 1A), or to the c-myc epitope present in the heavy chain (Supplemental Fig. 1B). Two protein bands with apparent molecular weights of approximately 26 and 50 kDa, corresponding to the light chain and the heavy chain of the anti-PEG reporter, respectively, were detected. Most of the membrane-anchored anti-PEG reporters were present as fully cleaved Fab fragments. A minor band of approximately 84 kDa, representing uncleaved V_L-C_K-furin-2A-V_H-C_H1-B7, constituted only a small proportion of the fusion protein expressed on the cell membrane. These results indicate that correctly processed anti-PEG reporter was expressed on the plasma membrane. Similar expression of the anti-DNS reporter on EJ cells was also noted.

EJ/anti-PEG and EJ/anti-DNS cells were stained with a commercial PEG–quantum dot to assess the PEG binding activity of the anti-PEG reporter. As shown in Figure 1B, EJ/anti-PEG cells but not EJ/anti-DNS cells selectively bound PEG–quantum dot. For testing of whether functional anti-PEG reporters can be expressed in vivo, cell suspensions prepared from freshly resected EJ/anti-PEG or EJ/anti-DNS tumors were stained with a hemagglutinin antibody or PEG–quantum dot. Figure 1C shows that EJ/anti-PEG and EJ/anti-DNS cells recovered from tumors retained expression of the respective reporters, but only EJ/anti-PEG cells displayed PEG–quantum dot binding activity. These results show that functional anti-PEG reporters were expressed on cell membranes in vitro and in vivo.

In Vitro and In Vivo Imaging of Anti-PEG Reporter

We tested 3 major mainstream imaging technologies for detecting the anti-PEG reporter: optical imaging, MRI, and PET. EJ/anti-PEG and EJ/anti-DNS cells were stained with PEG-NIR797, PEG-SPIO, or 4-arm ¹²⁴I-PEG-SHPP. Probe binding on cells was quantified by optical imaging, T2-weighted MRI, or γ-counting, respectively. Figure 2A shows quantitative binding of PEG-NIR797 to EJ/anti-PEG or EJ/anti-DNS cells in vitro. On average, signal intensity (defined as photons/s/cm²) was 6.5-, 7.5-, and 9.8-fold stronger in the EJ/anti-PEG cells than in the EJ/anti-DNS cells after the addition of 0.5, 0.1, and 0.02 μM PEG-NIR797, respectively. Similarly, PEG-SPIO was selectively trapped by the EJ/anti-PEG cells at all concentrations examined (darker images in Fig. 2C). The T2 signal intensity was reduced by more than 23% (range, ∼23.1%−94.1%) in EJ/anti-PEG cells, compared with EJ/anti-DNS cells. In line with the optical imaging and the MRI results, more than 55-fold (range, ∼55- to 159-fold) of 4-arm ¹²⁴I-PEG-SHPP was retained on EJ/anti-PEG cells than on EJ/anti-DNS cells (Fig. 2E). These data indicate that the anti-PEG reporter can trap various PEGylated probes for optical imaging, MRI, and PET.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

Imaging of anti-PEG reporter by multiple imaging systems. (A) In vitro optical imaging of EJ/anti-PEG cells or EJ/anti-DNS cells stained with graded concentrations of PEG-NIR797, with SEM of triplicate determinations indicated. (B) In vivo optical imaging of mice bearing EJ/anti-PEG tumors (right hind legs) and EJ/anti-DNS tumors (left hind legs) at 4, 24, and 48 h after injection of PEG-NIR797. Representative images of 3 independent experiments (n = 6) are shown. (C) In vitro MRI of EJ/anti-PEG cells (bottom) and EJ/anti-DNS cells (top) stained with graded concentrations of PEG-SPIO. (D) In vivo MRI of mice bearing EJ/anti-PEG tumors (solid line) and EJ/anti-DNS tumors (dashed line) at 1, 2, and 24 h after injection of PEG-SPIO (10 mg of iron/kg of body weight). Note that images of EJ/anti-PEG tumors are darker. Representative images of 3 independent experiments (n = 3) are shown. (E) In vitro binding of 4-arm ¹²⁴I-PEG-SHPP to EJ/anti-PEG cells or EJ/anti-DNS cells after staining with graded concentrations of 4-arm ¹²⁴I-PEG-SHPP. (F) In vivo small-animal PET of mice bearing EJ/anti-PEG tumors (right hind legs) and EJ/anti-DNS tumors (left hind legs) at 4, 24, and 48 h after injection of 4-arm ¹²⁴I-PEG-SHPP. Small-animal PET images were acquired in coronal and transverse planes. Representative images of 3 independent experiments (n = 3) are shown. αPEG = anti-PEG; αDNS = anti-DNS.

To investigate whether the anti-PEG reporter could be noninvasively imaged in vivo, we intravenously injected PEG-NIR797, PEG-SPIO, or 4-arm ¹²⁴I-PEG-SHPP into BALB/c nude mice bearing subcutaneous EJ/anti-PEG and EJ/anti-DNS tumors. Whole-body images of the tumor-bearing mice were acquired by optical imaging, T2-weighted MRI, and small-animal PET, respectively. Figure 2B shows that PEG-NIR797 probe was selectively retained at EJ/anti-PEG tumors but not at EJ/anti-DNS tumors. The near-infrared signal lasted as long as 48 h after injection and clearly indicated the locations of the EJ/anti-PEG tumors. Fluorescent intensities in the region of interest were 9.0-, 10.9-, and 16.9-fold greater in the EJ/anti-PEG tumors than in the EJ/anti-DNS tumors at 4, 24, and 48 h, respectively (P < 0.028 for all time points). Similarly, PEG-SPIO was selectively retained in EJ/anti-PEG tumors, leading to markedly darker signals (a positive T2-weighted signal) in the EJ/anti-PEG tumors but not in the control EJ/anti-DNS tumors (Fig. 2D). Finally, consistent with the in vitro tests, the 4-arm ¹²⁴I-PEG-SHPP preferentially accumulated in EJ/anti-PEG tumors (Fig. 2F). On average, EJ/anti-PEG tumors accumulated 4.4-, 6.0-, and 10.4-fold more radioactivity than did EJ/anti-DNS tumors at 4, 24, and 48 h, respectively (Supplemental Fig. 2). Though slight signals could initially be detected in the lung, liver, and kidney, these nonspecific signals faded over time whereas specific signals remained conspicuous in the EJ/anti-PEG tumors. Collectively, these results indicate that the anti-PEG reporter, when paired with PEGylated probes, is useful for noninvasive imaging in vivo by optical imaging, MRI, and PET.

Comparisons Between Anti-PEG and HSV-tk Reporters

HSV-tk is currently regarded as a gold standard reporter gene. We therefore sought to compare the imaging specificity, immunogenicity, and probe toxicity of the anti-PEG and HSV-tk reporters. EJ/TK was generated by retroviral transduction. Western blotting indicated that EJ/anti-PEG and EJ/TK cells expressed similar levels of anti-PEG or HSV-tk reporter proteins (Supplemental Fig. 3). BALB/c nude mice bearing EJ/anti-PEG and EJ/TK tumors in their right and left hind legs were intravenously injected with 4-arm ¹²⁴I-PEG-SHPP or ¹²⁴I-FIAU. Specific targeting of radioiodinated probes was analyzed by small-animal PET. Figure 3 shows that 4-arm ¹²⁴I-PEG-SHPP selectively accumulated in EJ/anti-PEG tumors but not in EJ/TK tumors whereas ¹²⁴I-FIAU displayed the opposite specificity. Importantly, imaging specificity attained by the anti-PEG reporter was comparable to HSV-tk.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

Imaging comparison of anti-PEG and HSV-tk reporters. Mice bearing established EJ/anti-PEG (right hind legs) and EJ/TK (left hind legs) tumors were intravenously injected with 3,700 kBq of 4-arm ¹²⁴I-PEG-SHPP (top) or ¹²⁴I-FIAU (bottom). Small-animal PET images were acquired in coronal planes and transverse planes at 4, 24, and 48 h after injection of probes. αPEG = anti-PEG.

The immunogenicity of the anti-PEG and HSV-tk reporters was examined after hydrodynamic-based gene transfer of anti-PEG- or HSV-tk–encoding plasmids to BALB/c mice. We did not detect noticeable antibody titer in mice injected with anti-PEG–encoding plasmids (Fig. 4). In line with previous reports (3), however, antibodies against HSV-tk were readily detected in the mice injected with HSV-tk–encoding plasmids, indicating that HSV-tk was indeed immunogenic.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4. 

Immunogenicity of anti-PEG and HSV-tk reporter systems. Serum samples collected from BALB/c mice 10 d after hydrodynamic-based gene transduction were assayed by enzyme-linked immunosorbent assay for presence of antibodies against anti-PEG or HSV-tk proteins expressed on surface of 3T3 fibroblasts. Representative data from 3 independent experiments are shown. Horizontal bars indicate mean antibody titer in each group. αPEG = anti-PEG; mTK = membrane-anchored HSV-tk.

To assess the safety of anti-PEG or HSV-tk imaging probes, we incubated iodinated 4-arm PEG-SHPP or FIAU with parental EJ, EJ/anti-PEG, EJ/TK, and CNL cells. Consistent with previous results indicating that PEG is a nontoxic polymer (14), iodinated 4-arm PEG-SHPP (≤125 μM) did not affect the survival of CNL, EJ, or EJ/anti-PEG cells (Supplemental Fig. 4A). In contrast, FIAU was considerably cytotoxic to CNL cells (inhibitory concentration of 50%, 65.9 μM). Moreover, FIAU induced more pronounced cytotoxicity to the cells that expressed HSV-tk (inhibitory concentration of 50%, 0.18 μM, P < 0.00001, comparing EJ/TK to EJ; Supplemental Fig. 4B). These results and previous reports (15,16) firmly demonstrate that PEGylated probes display little toxicity to mammalian cells. On the contrary, the dose of HSV-tk probes should be carefully adjusted to prevent toxicity.

In Vivo Imaging of Humanized Anti-PEG Reporter

Before the anti-PEG reporter can be used in humans, the mouse-derived sequences must be humanized to minimize human antimouse immune responses. The complementarity-determining regions of the anti-PEG V_H and V_L were grafted onto the human germline DP-35 + JH4 and A2c + JK2 segments to create humanized anti-PEG V_H and V_L (Supplemental Fig. 5). The humanized anti-PEG V_H and V_L genes were fused to human IgM C_H1 and Cκ genes to construct the heavy chain and light chain of the humanized anti-PEG reporter (human anti-PEG). The human anti-PEG reporter was functional as shown by binding of PEG–quantum dot to EJ/human anti-PEG cells (Fig. 5A). As a test of its in vivo utility as a reporter gene, BALB/c nude mice bearing established EJ/human anti-PEG and EJ/anti-DNS tumors in their right and left hind legs, respectively, were intravenously injected with 4-arm ¹²⁴I-PEG-SHPP and then imaged with a small-animal PET scanner. Figure 5B shows that specific signals selectively accumulated at EJ/human anti-PEG tumors at 4 h after injection. The radiointensity of the region of interest was 4.5-, 5.0-, and 10.4-fold greater in the EJ/human anti-PEG tumors than in the EJ/anti-DNS tumors at 4, 24, and 48 h, respectively (similar to EJ/anti-PEG tumors vs. EJ/anti-DNS tumors). In addition, a metastatic lung model of Hela/human anti-PEG cells was also monitored by optical imaging. Figure 5C shows that specific fluorescent signals were detected in the lung foci after intravenous injection of Hela/human anti-PEG cells but not control Hela cells. Collectively, these results indicate that humanized anti-PEG may be an optimal reporter gene for noninvasive imaging of gene expression and cell delivery in humans.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5. 

Analysis and imaging of humanized anti-PEG reporter. (A) Flow cytometric analysis of surface expression (dashed line, stained by hemagglutinin antibody) and function (solid line, stained by PEG–quantum dot) of human anti-PEG reporter in EJ cell line. Shaded area of graph shows mock staining with PBS. (B) In vivo small-animal PET of mice bearing EJ/human anti-PEG tumors (right hind legs) and EJ/anti-DNS tumors (left hind legs) at 4, 24, and 48 h after injection of 4-arm ¹²⁴I-PEG-SHPP. Small-animal PET images were acquired in coronal and transverse planes. Representative images of 3 independent experiments (n = 3) are shown. (C) In vivo optical imaging of naive mice (left panel) or mice bearing Hela (middle panel) and Hela/human anti-PEG (right panel) lung metastatic tumors at 48 h after injection of PEG-NIR797. Lung tissue was also harvested to confirm tumor foci and specific fluorescent signals. Representative images of 3 independent experiments (n = 3) are shown. h-αPEG = human anti-PEG; αDNS = anti-DNS.