PET Imaging of TIGIT Expression on Tumor-Infiltrating Lymphocytes

This article is featured in Highlights of This Issue, p. 1825

Immune checkpoint receptors expressed on activated T and natural killer (NK) cells play a key role in maintaining physiologic self-tolerance (1). However, these same regulating pathways are exploited by tumor cells to evade immune surveillance (2). This tumor survival strategy has been targeted with checkpoint blockade therapies such as anti-PD-1 and anti-CTLA-4 agents. While these therapies have shown remarkable responses in a subset of patients, primary and adaptive resistance often occurs (3). This is attributed to a variety of mechanisms including downregulating MHC to prevent antigen presentation and upregulation of other inhibitory receptors such as TIM-3, LAG-3, and VISTA (4).

To reduce the occurrence of tumor escape, other inhibitory targets that act through alternative pathways are being extensively explored. One such checkpoint inhibitor is the T-cell Ig and ITIM domain receptor (TIGIT). TIGIT is an inhibitory receptor expressed on CD8⁺ cytotoxic T cells, CD4⁺ Th cells, FOXP3⁺ regulatory T cells, and NK cells (5). Both TIGIT and PD-1 are expressed in >70% of human melanoma tumor microenvironments, with PD-1 mainly expressed on CD8⁺ T cells and TIGIT expressed on CD8⁺ T cells, regulatory T cells, and NK cells (6). The ratio of PD-1 to TIGIT expressed by tumor-infiltrating lymphocytes (TIL) is variable, ranging from 0.75 to 4.0 in colorectal cancer(5). TIGIT can inhibit immune responses of effector T and NK cells by acting through inhibitory signals (7–9) and affecting effector functions of T cells and antigen-presenting cells (10, 11). TIGIT and PD-1 blockade additively increased proliferation, cytokine production, and degranulation on TILs (12), making TIGIT a potential key addition to current immunotherapies. For this reason, a number of clinical trials in solid tumors are combining TIGIT therapies with current checkpoint blockade therapies (13–15).

The expression of TIGIT on TILs has been reported via IHC, Western blotting, and PCR on hepatocellular carcinoma (16), follicular lymphoma (17), squamous cell cancers (5), Hodgkin lymphoma (18), renal cell carcinoma (19), small cell lung cancer (20), and melanoma (21). TIGIT has been validated as a therapeutic target and anti-TIGIT therapies are currently being evaluated in multiple clinical trials (Supplementary Table S1). An antibody specific for the TIGIT receptor (AB154) is being evaluated in a variety of solid cancers (NCT03628677). This phase I trial is evaluating the safety, tolerability, pharmacokinetics, pharmacodynamics, and clinical activity of AB154 as either a monotherapy or in combination with an anti-PD-1 antibody (AB122). In another phase I trial, the anti-TIGIT agent MK-7684 and pembrolizumab was administered in 34 patients with advanced solid tumors for whom standard treatment options had failed (NCT02720068). The response rate was reported as 19% and the disease control rate was 47% (22). While these results are promising for a subset of patients, patient stratification for these therapies would benefit from a quantitative, whole-body measure of TIGIT expression.

Generally high but variable TIGIT expression is seen in solid tumors, making this therapy likely applicable to many cancer types. However, to determine TIGIT expression, invasive biopsy sampling followed by IHC is currently required. While this procedure is the current gold standard, it does not allow accurate quantitation in heterogeneous samples or on distant metastases. A noninvasive imaging agent could detect and quantify TIGIT within the tumor microenvironment and aid in determining what patients would benefit from TIGIT immunotherapies (23). This approach has been tested with PET agents for PD-1, PD-L1, and Ox40, all being developed as clinical diagnostics (24–26). These PET tracers allow noninvasive, whole-body, quantitative assessment of receptor expression to aid in patient stratification. Here we develop TIGIT-specific antibody PET tracers and evaluate their application in quantifying TIGIT expression in vivo (Fig. 1).

Purified anti-mouse TIGIT mAb (TIGITmAb, clone 1G9) raised in mouse was purchased from Bio X cell (catalog no. BE0274) or BioLegend (catalog no. 142102). NHS ester-DOTA (NHS-DOTA, catalog no. B-280) and p-SCN-Deferoxamine (DFO-NCS, catalog no. B705) were purchased from Macrocyclics. Copper-64 (⁶⁴Cu; t_1/2 = 12.7 hours, 17.76–162.8 GBq/μmol) was obtained from the University of Wisconsin (Madison, WI). Zirconium-89 (⁸⁹Zr) in 1.0 mol/L oxalic acid (t_1/2 = 3.3 days, purity >99%; 2–3 GBq/mL) was obtained from the University of Alabama at Birmingham (Birmingham, AL). All other general reagents were obtained from Sigma-Aldrich unless otherwise stated.

B16-F10 (murine melanoma) and HeLa (human cervical) cells were purchased from ATCC (CRL-6475). A HeLa cell line expressing the TIGIT receptor was generated using a plasmid from VectorBuilder consisting of the mouse TIGIT sequence driven by the EF1α promoter, eGFP driven by the cytomegalovirus promoter, and a puromycin sequence for selection (Supplementary Fig. S1). Transfection was performed using Lipofectamine 3000 (Thermo Fisher Scientific). Cells were sorted three times on a FACSAria II sorter with the top 1% of TIGIT staining collected. The transfected HeLa cells (TIGIT-HeLa) were kept under selection using DMEM supplemented with 10% FBS, 1% Antibiotic-Antimycotic, and 40 μmol/mL of puromycin. All cell lines were screened to exclude infection with Mycoplasma and preserved for a period of 20 passages in culture.

Qualitative flow cytometry was performed on TIGIT-HeLa cells and nontransfected HeLa cells by staining with PE-mTIGIT antibody (2 μL/10⁶ cells) and DAPI (1 μL of 10 mg/mL/10⁶ cells). Quantitative flow cytometry was performed on an LSR II instrument to determine the average number of TIGIT receptors per cell using a QIFIKIT (Agilent).

Flow cytometry was completed on isolated, activated C57B/l6J (Jackson Laboratory) mouse splenocytes to assess antibody affinity for mouse TIGIT. Splenocytes were activated via phorbol 12-myristate 13-acetate (PMA) and calcium ionomycin (50:500 ng/mL for 4 hours) at 37°C (27). Cell staining was performed in 96-well microtiter plates (U-bottom). A total of 100 μL of the antibody was mixed with 100 μL cell suspension containing 2 × 10⁵ splenocytes and incubated for 1 hour at 4°C. Cells were washed three times with FACS buffer, followed by a secondary stain with 100 μL of anti-mouse IgG-APC (BioLegend, catalog no. 409305) prediluted to 1:100. The plate was incubated for 30 minutes on ice in the dark and washed three times. Propidium iodide was used as a live/dead stain. Analysis was carried out on a flow cytometer (FACS Aria III, BD Biosciences) and data were analyzed by FlowJo FACS analysis software (Tree Star).

DOTA was conjugated to mouse TIGITmAb at pH 7.3, 1 mg/mL Ab concentration, and a 20:1 DOTA:TIGITmAb ratio at 4°C for 14 hours. The DOTA-TIGITmAb conjugate was purified using 50 kDa spin filters (Millipore) with 0.1 mol/L ammonium acetate (pH 5.4) as the buffer. DOTA-TIGITmAb was washed three times and stored at 4°C at a concentration of 1–2 mg/mL. DFO-TIGITmAb was conjugated to mouse TIGITmAb at pH 8.5 in HEPES buffer, 6 mg/mL Ab concentration, and a 5:1 DFO:TIGITmAb ratio at 37°C for 14 hours. The DFO-TIGITmAb conjugate was purified using a SEC 2000 high-performance liquid chromatography (HPLC) with 0.1 mol/L ammonium acetate buffer (pH 7) as the mobile phase eluted at 1 mL/minute. The purified DFO-TIGITmAb was concentrated to 5 mg/mL using a Vivaspin 30 kDa cut-off centrifugal filter and stored in 100 μL aliquots in 0.1 mol/L ammonium acetate buffer (pH 7.0) at −20°C.

⁶⁴Cu was added to DOTA-TIGITmAb (1 mg/mL) in 0.1 mol/L ammonium acetate (pH 5.3) at a ratio of 0.37 MBq/μg Ab and incubated at 37°C for 30–45 minutes. A total of 10 μL of 50 mmol/L ethylenediaminetetraacetic acid (EDTA) was added for 15 minutes to scavenge free ⁶⁴Cu. ⁶⁴Cu-TIGITmAb was purified using 50 kDa spin filters with saline as the buffer. ⁸⁹Zr in 1.0 mol/L oxalic acid was first brought to a pH of 7.0–7.4 with 1.0 mol/L sodium carbonate. ⁸⁹Zr (74–75 MBq; 200 μL) was then added to DFO-TIGITmAb (400 μg; 5 mg/mL) in 1× PBS (pH 7.3) and incubated at 37°C for 1 hour. After incubation, 10 μL of 50 mmol/L EDTA was added for 15 minutes to scavenge free ⁸⁹Zr. Purification of ⁸⁹Zr-TIGITmAb was achieved via size exclusion (SE) HPLC in PBS buffer [0.1 mol/L NaCl, 0.05 mol/L sodium phosphate (pH 7.4)] at a flow rate of 1.0 mL/minute. The ⁸⁹Zr-TIGITmAb peak corresponding to the antibody was collected and concentrated using a Vivaspin 30 kDa centrifugal filter and centrifuged at 3,000 × g for 15 minutes. The final product was filtered through a 0.2-micron filter into a sterile vial. Serum stability for ⁶⁴Cu-TIGITmAb and ⁸⁹Zr-TIGITmAb was conducted by addition of 3.7 MBq radiotracer to 100 μL of mouse or human serum with time points taken every 24 hours.

The immunoreactivity of ⁶⁴Cu-TIGITmAb and ⁸⁹Zr-TIGITmAb was assessed via Lindmo assay (28). ⁶⁴Cu-TIGITmAb and ⁸⁹Zr-TIGITmAb were prepared at a concentration of 0.037 MBq in 2 mL of PBS with 1% BSA. A total of 50 μL of the radiolabeled antibody was added to a serial dilution of TIGIT receptor–positive HeLa cells ranging from 5×10⁴ to 5×10⁶ cells per vial (500 μL in PBS with 1% BSA added). After an incubation at room temperature for 30 minutes, the solutions were centrifuged at 300 × g, supernatant removed, and pellets counted using a gamma counter. Nontransfected HeLa cells were also tested with the radiolabeled antibodies, as described above.

⁸⁹Zr-TIGITmAb binding was also assessed on isolated splenocytes. Two sets of experiments were performed (blocking and nonblocking; n = 3). Blocking samples received a 25-fold excess of cold TIGITmAb prior to the addition of ⁸⁹Zr-TIGITmAb (2.3 μCi/mL; 200 μL/tube). The solutions were incubated for 1 hour at 25°C, washed, and activity bound to the cell pellets were counted on a gamma counter.

Animal studies were performed in compliance with and approval from the Administrative Panel on Laboratory Animal Care at Stanford University (Stanford, CA). PET/CT imaging was performed on a Siemens Inveon small animal multimodality PET/CT system (Preclinical Solutions; Siemens Healthcare Molecular Imaging). CT imaging was performed at 80 kVp at 500 μA, second bed position, half scan 220 degrees of rotation, and 120 projections per bed position with a cone beam micro-X-ray source (50 μmol/L focal spot size) and a 4064 × 4064–pixel X-ray detector. The data were reconstructed using Shepp-Logan filtering and cone-beam filtered backprojection.

The small animal PET scanning (static scans with a coincidence timing window of 3.4 ns and energy window of 350–650 keV) was performed at the following time points after the tracer injection: 4, 24, and 48 hours for 10 minute, and 72 hours for 15-minute scans. The acquired images were reconstructed with two-dimensional ordered-subset expectation maximization (OSEM 2D) algorithm. PET/CT images and three-dimensional regions of interest (ROI) of organs of interest were computed using Inveon Research Workplace software. The mean pixel values within the ROI volume was converted to radioactivity concentration in counts per milliliter per minute (cpm) by using a predetermined conversion factor. The percentage-injected dose per gram of tissue (%ID/g) was determined by dividing each tissue's cpm obtained from the ROI by the injected dose. At each time point before imaging, animals were anesthetized and scanned at the same conditions as detailed above. A Hidex gamma counter was used for ex vivo radioanalysis.

Six- to 10-week-old nu/nu female mice (average weight = 23.0 ± 2.0 g) were injected in the flank with 7.5×10⁵ cells in 100 μL matrigel. TIGIT-positive HeLa cells were injected into the left flank and TIGIT-negative HeLa cells were injected into the right flank. Xenografts were allowed to develop for 10 days. ⁶⁴Cu-TIGITmAb (or ⁶⁴Cu-mouse IgG, n = 4/probe) was injected via tail vein at 2.2–3.7 MBq (10–15 μg) in 100 μL of saline and imaged at 24 and 48 hours followed by a terminal ex vivo biodistribution study. ⁸⁹Zr-TIGITmAb (or ⁸⁹Zr-mouse IgG, n = 4 per probe) were injected via tail vein at 1.85–3.7 MBq (15–20 μg) in 100 μL of saline and imaged at 24, 48, and 72 hours followed by a terminal ex vivo biodistribution study. Gamma counter values were background subtracted and decay corrected to the time of injection and the %ID/g for each tissue sample was calculated by normalization to the total activity injected. Animal experiments were conducted twice.

Six- to 8-week-old female B6 mice were implanted with 1 × 10⁶ B16F10 melanoma cells in the right shoulder and developed allografts over 10–12 days. Two groups of mice (blocking and nonblocking; n = 5) received ⁸⁹Zr-TIGITmAb (200 μL, 1.7–2.1 MBq, 15–16 μg of DFO-TIGITmAb) via tail vein injection. Blocking was completed by injection of cold TIGIT (200 μg in 100 μL) 2 hours prior to ⁸⁹Zr-TIGITmAb administration. The mice were imaged at 4, 24, 48, and 72 hours after injection and immediately followed by a terminal biodistribution study.

B6 mice (n = 5) were implanted with B16F10 melanoma allografts as described above. After 10 days of tumor growth, mice were euthanized and tumors were removed. Liberase TL (300 μg/mL) and DNase I (100μg/mL) in 5 mL Hank's Balanced Salt Solution (HBSS) were added to dissociated tumors for 1 hour at 37°C with gentle mixing every 15–20 minutes followed by washing in HBSS. Subsequently, ACK buffer was added to samples for 10 minutes to lyse red blood cells followed by two washes in cell stain buffer. Single-cell suspensions were prepared and stained for mTIGIT (2 μL Ab/10⁶ cells) and DAPI (10 μg/mL final concentration) as a viability stain as described above. Samples were run in triplicate, and at least 2×10⁵ live cells per sample were collected. Analysis was completed on an LSR II flow cytometer. Ex vivo flow cytometry was completed twice.

Unpaired Student t test (two-tailed, unequal variance) was used for data comparisons. P values of less than 0.05 were considered statistically significant.

Following three rounds of cell sorting under puromycin selection, >98% of HeLa cells expressed the mouse TIGIT receptor (Fig. 2A; Supplementary Fig. S2) with 95,200 ± 15,900 TIGIT receptors per cell (Fig. 2B). Splenocytes activated ex vivo were positive for TIGIT and DFO-conjugated TIGITmAb showed similar affinity as the unconjugated TIGITmAb for activated splenocytes, demonstrating that DFO conjugation did not adversely affect binding (Fig. 2C). These results validate the expression of TIGIT in transfected HeLa cells, as well as the specificity of TIGITmAb for the TIGIT receptor on primary splenocytes.

DOTA and DFO were successfully conjugated to the TIGITmAb with 1.59 ± 0.30 DOTA molecules and 0.90 ± 0.17 DFO molecules per TIGITmAb (Supplementary Table S2), as determined by MALDI-MS (Supplementary Fig. S3). The radiochemical yield of ⁶⁴Cu-TIGITmAb was 60%–75% and the specific activity of 0.222–0.370 MBq/μg, while ⁸⁹Zr-TIGITmAb had a radiochemical yield of 60%–80% and a specific activity of 0.074–0.185 MBq/μg. The final purity of both radiotracers was >98% (Supplementary Fig. S4). The serum stability of ⁶⁴Cu-TIGITmAb was >95% out to 48 hours (Fig. 3A) and that of ⁸⁹Zr-TIGITmAb was >95% out to 96 hours (Fig. 3D). Finally, ⁶⁴Cu-TIGITmAb was further evaluated using SE HPLC and showed a radiochemical purity of 98.3 ± 1.5% (mean ± SD; Supplementary Fig. S4C).

⁶⁴Cu-TIGITmAb and ⁸⁹Zr-TIGITmAb showed at least 40-fold higher binding to TIGIT-HeLa cells, compared with the radiolabeled isotype control and low nonspecific binding to TIGIT-negative HeLa cells (P < 0.05; Fig. 3B and E). ⁶⁴Cu-TIGITmAb and ⁸⁹Zr-TIGITmAb had immunoreactivities of 72.7 ± 1.7% and 80.5 ± 2.2%, respectively (Fig. 3C and F). ⁸⁹Zr-TIGITmAb demonstrated high specificity on activated splenocytes (Fig. 3G).

Both PET probes showed specific uptake in HeLa-TIGIT–expressing xenografts (n = 4), compared with TIGIT-negative xenografts or isotype controls (P < 0.05). ⁶⁴Cu-TIGITmAb showed specific uptake in HeLa-TIGIT at 24 hours, with increased uptake at 48 hours (Fig. 4A and B). The 48 hours ex vivo biodistribution (Fig. 4C) of ⁶⁴Cu-TIGITmAb and ⁶⁴Cu-mIgG showed ⁶⁴Cu-TIGITmAb uptake in HeLa-TIGIT xenografts of 21.9 ± 3.5%ID/g, compared with ⁶⁴Cu-mIgG uptake of 9.8 ± 2.2%ID/g (P < 0.05). However, blood activity levels at 48 hours were 13.4 ± 0.8%ID/g due to the pharmacokinetics of the antibody.

At 72 hours postinjection, ⁸⁹Zr-TIGITmAb demonstrated higher tumor-to-background signal ratios than ⁶⁴Cu-TIGITmAb due to higher tumor uptake and lower blood activity (Fig. 5A). This was confirmed by ex vivo biodistribution results at 72 hours (Fig. 5B), where ⁸⁹Zr-TIGIT uptake in HeLa-TIGIT xenografts was 29.3 ± 4.5%ID/g, compared with ⁸⁹Zr-mIgG uptake of 8.8 ± 0.8%ID/g (P < 0.05). Blood activity levels at 72 hours were 9.7 ± 1.0%ID/g, significantly lower than ⁶⁴Cu-TIGITmAb at 48 hours (P < 0.05). ⁸⁹Zr-TIGITmAb was chosen for further studies due to the combination of higher tumor uptake and lower blood activity levels compared with ⁶⁴Cu-TIGITmAb at later time points.

⁸⁹Zr-TIGITmAb PET signals at the tumor site indicated the presence of TIGIT on TILs. Representative axis and coronal immunoPET images of mice bearing melanoma at various time points (4, 24, and 72 hours after injection of tracer) are shown in Fig. 6A. PET ROI quantification (Fig. 6B) correlated with biodistribution results at 72 hours after injection (Fig. 6C). ⁸⁹Zr-TIGITmAb uptake by TIGIT-bound TILs in nonblocking and blocking mice at 72 hours was (mean %ID/g ± SD) 7.4 ± 0.9 and 3.81 ± 0.75, respectively (P < 0.05). Tumor-to-muscle uptake ratios for ⁸⁹Zr-TIGITmAb in nonblocking mice were 4.1 ± 0.2 and 7.4 ± 0.7 at 4 and 72 hours (black bar), respectively (P < 0.05; Fig. 6D).

Flow cytometry on single-cell suspensions of B16 tumors from five mice showed TIGIT expression on infiltrating TILs. A total of 10.91 ± 2.22% of live cells in the tumor microenvironment were positive for TIGIT (Fig. 6E), showing TIGIT expression in this tumor microenvironment as shown previously (29). Furthermore, from a separate group (n = 4), melanoma allograft tissue was harvested and prepared several sections from intermelanoma and intramelanoma. These FFPE-tissue slides were stained to determine the expression pattern of both TILs and TIGIT by using CD45 and TIGITmAb (Supplementary Fig. S5). Low TIGIT expression correlated with low CD45 staining (Supplementary Fig. S5D) with the majority of TIGIT expression associated with the CD45⁺ TILs (Supplementary Fig. S5C).

This work develops and validates TIGIT-specific PET agents for quantification of TIGIT expression in the tumor microenvironment. We demonstrate that ⁶⁴Cu-TIGITmAb and ⁸⁹Zr-TIGITmAb are highly specific molecular imaging probes for assessing TIGIT expression in vivo. While both probes detect the TIGIT receptor in TIGIT-HeLa xenografts, ⁸⁹Zr-TIGITmAb showed superior contrast due to better matching of full-length antibody pharmacokinetics, which are slow to clear from circulation. ⁸⁹Zr-TIGITmAb was therefore evaluated in immunocompetent B6 mice with B16 melanoma allografts and demonstrated the ability to quantitatively image TIGIT expression from 4 to 72 hours after injection. Blocking studies with excess cold TIGITmAb decreased the PET signals in the tumor, demonstrating specificity of ⁸⁹Zr-TIGITmAb. Flow cytometry and ex vivo immunofluorescence confirmed TIGIT expression on TILs in the tumor microenvironment. These results indicate ⁸⁹Zr-TIGITmAb as a potential companion diagnostic for current and upcoming TIGIT therapies. The ability to quantify TIGIT expression in solid tumors could aid in treatment strategies prior to starting a particular immunotherapy regimen.

While this work has demonstrated the ability of quantitative TIGIT imaging in solid tumors, it must be noted that the murine TIGIT receptor was targeted. This was necessary to test the probe in TIGIT-expressing immune cells in immunocompetent mice; however, developing a human TIGIT imaging probe is necessary for clinical translation. A companion diagnostic for TIGIT therapy, much like ⁶⁴Cu-pembrolizumab for human PD-1 could aid in therapeutic decisions prior and throughout the regimen. These results warrant the future development of a human TIGIT PET imaging agent for quantifying TIGIT expression in the tumor microenvironment for stratification of patients for anti-TIGIT therapy.

In conclusion, we developed and radiolabeled novel anti-TIGIT probes for immunoPET imaging of TIGIT expression in xenograft and melanoma allograft murine models. Furthermore, specific in vivo binding of anti-TIGIT probes were confirmed via quantitative flow cytometry and ex vivo biodistribution studies. This study demonstrates the potential for clinical translation of these radiotracers.

No disclosures were reported.

T. Shaffer: Conceptualization, formal analysis, validation, writing–original draft. A. Natarajan: Conceptualization, data curation, software, formal analysis, validation, investigation, visualization, methodology, writing–original draft, writing–review and editing. S.S. Gambhir: Conceptualization, resources, project administration, writing–review and editing.

This work was supported by a Cancer-Translational Nanotechnology Training grant (T32 CA196585, to T. Shaffer) and a Stanford Bio-X seed grant. Small animal PET/CT imaging and gamma counter measurements were performed in the SCi³ Stanford Center for Innovation In Vivo Imaging. Data were collected on an instrument in the Shared FACS Facility obtained using NIH S10 Shared Instrument Grant S10RR027431-01.

The co-authors would like to dedicate this work to S.S. Gambhir (1962–2020) for his excellent mentoring and tremendous support through the years.

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.