In Vivo Imaging of Methionine Aminopeptidase II for Prostate Cancer Risk Stratification

Prostate cancer is the second most commonly diagnosed malignancy among men and a leading cause of cancer-related deaths worldwide (1). Most cases of prostate cancer are asymptomatic and are detected within the prostate as a result of screening. However, screening has also resulted in overdetection of indolent prostate cancers resulting in overtreatment and unnecessary morbidity, necessitating the development of prognostic models in early disease (2). Current risk stratification uses serum PSA measurements, Gleason grade on biopsy (microscopic morphology of the cancer cells), and clinical stage for prognostication, although these provide only modest prediction of clinical outcomes (3). Identification of new biomarkers suitable for development of noninvasive prognostic technology is highly desirable.

PET imaging hybridized with CT or MRI is sensitive, noninvasive, three-dimensional (3D), and has demonstrated utility in improving the diagnosis of prostate cancer, especially for recurrent and advanced disease (4). Current prostate cancer imaging tracers approved by the FDA or in clinical trial typically rely on binding of a radiolabeled ligand, inhibitor, or antibody to a membrane receptor (5). However, not all prostate cancers express the membrane receptors targeted by these imaging agents. For example, neuroendocrine prostate cancer (NEPC), a highly aggressive form of advanced prostate cancer, commonly lacks expression of prostate-specific membrane antigen (PSMA), the target of a recently approved PET tracer, gallium-68 PSMA-11 (6, 7). PET tracers that can traverse cell membrane and report on activity of molecules essential for intrinsic biochemical processes, which undergo alterations during prostate cancer initiation and progression independent of the androgen receptor (AR) signaling axis, could offer attractive imaging targets.

Methionine aminopeptidase II (MetAP2) is a cytosolic metalloprotease that catalyzes the cotranslational removal of the N-terminus initiator methionine residue from nascent proteins (8). MetAP2 overexpression stimulates cancer cell proliferation (9) and is well known as a druggable target of angiogenesis, a pathologic determinant of progression in solid tumors (10). Considerable effort over many years has been devoted to developing MetAP2 inhibitors as anticancer drugs (11–13), because its expression has been correlated with proliferation and angiogenesis in advanced prostate cancer (14, 15). High MetAP2 expression has been observed in prostatic carcinoma (9); we now report that high levels of MetAP2 are predictive for shorter time to biochemical recurrence in prostate cancer treated surgically. Analysis of MetAP2 levels in a cohort of more than 400 patient prostate cancer samples from radical prostatectomies confirmed the overexpression of MetAP2 in more than half of prostate cancer samples regardless of grade, and MetAP2 overexpression was associated with a 2-fold increased risk of biochemical recurrence within 5 years. MetAP2, therefore, represents a promising target for noninvasive functional imaging of clinically significant prostate cancer.

We have devised a strategy termed target-enabled in situ ligand aggregation (TESLA) to develop molecular probes for imaging the activity of proteolytic enzymes in vivo that can be used with a variety of imaging modalities, including fluorescence imaging, photoacoustic imaging, MRI, and PET (16–19). This strategy was inspired by the rapid condensation reaction between 6-hydroxy-2-cyanobenzothiazole (CBT) and d-cysteine in the synthesis of firefly d-luciferin under physiologic conditions (20). The cysteine can be masked at its amino and thiol groups, for example, by a peptide substrate, and linked to the CBT unit in designing a TESLA probe. Such probes can undergo self-condensation and assemble into nanoaggregates after the target enzyme activation (17, 21–23). Recently, we have further explored this condensation reaction from CBT to other aromatic nitriles, such as 2-cyanopyrimidine and aminothiols (24). In this study, we designed novel MetAP2-sensitive nanoaggregation tracers (M-SNAT; Supplementary Figs. S1–S4) and validated them for imaging the activity of MetAP2 in a series of in vitro and in vivo models. The first demonstration of MetAP2-activatable small molecule for PET imaging in preclinical models highlights its potential for noninvasive diagnosis and risk stratification of early prostate cancer.

PC3, DU145, 22Rv1, LNCaP, and NCI-H660 carcinoma lines were obtained from the ATCC. LNCaP-Trop2 are engineered LNCaP cells overexpressing Trop2, as reported previously (25). Culture was maintained at 37°C with 5% CO₂. Cells were split using 0.25% Trypsin-EDTA. PC3, DU145, 22Rv1, LNCaP, and LNCaP-Trop2 were cultured in RPMI1640 medium supplemented with 10% FBS, 100 U/mL penicillin, 100 μg/mL streptomycin, and 1% l-glutamine. H660 cells were cultured in RPMI1640 medium supplemented with 5% FBS, 0.005 mg/mL insulin, 0.01 mg/mL transferrin, 30 nmol/L sodium selenite, 10 nmol/L hydrocortisone, 10 nmol/L beta-estradiol, and 4 mmol/L l-glutamine. Human prostate epithelial cell line, PrEC, was cultured in prostate epithelial cell basal medium (PCS-440-030, ATCC) supplemented with 100 U/mL penicillin, 100 μg/mL streptomycin, 1% l-glutamine, and a Prostate Epithelial Cell Growth Kit (ATCC, PCS-440-040). PC3, DU145, 22Rv1, LNCaP, NCI-H660, and LNCaP-Trop2 were maintained below 10 passages. PrEC was maintained below three passages. All cell lines were routinely tested for Mycoplasma contamination (MycoAlert Mycoplasma Detection Kit purchased from Lonza), with authentication performed at Stanford Functional Genomics Facility (Stanford, CA) for short tandem repeat profiling.

Prostates were collected at Stanford Hospital (Stanford, CA) under institutional review board approval and written informed consent. All patient samples were deidentified prior to inclusion in this study. Subsequently, tissues were processed at the Stanford Department of Urology tissue bank and consisted of benign prostatic hyperplasia (n = 62), benign tissue adjacent to cancer (n = 63), low-grade cancer (Gleason 3 + 3 to 3 + 4; n = 249), or high-grade cancer (Gleason 4 + 3 to 5 + 5; n = 56). Samples were sliced to display clinically relevant disease, fixed overnight, then sectioned at 4 μm, and mounted on lysine-coated glass slides. Samples were stained for MetAP2 by IHC and scored by intensity by a pathologist from 0 to 3 (0, negative; 1, uninterpretable; 2, low intensity positive staining; and 3, high intensity positive staining).

Tissue microarrays (TMA) for clinical evaluation of MetAP2 were constructed at Stanford University (Stanford, CA) using an internal cohort of patients with a median follow-up of 10 years, as described previously (26). This cohort consisted of recurrent (n = 58) and nonrecurrent patients (n = 176) and included patients with Gleason score 3 + 3 to 4 + 4. Three cores per patient were included on the TMA. MetAP2 staining was performed as described next. Samples were scored for MetAP2 intensity by a genitourinary pathologist (C.A. Kunder) who was blinded to sample information.

Tumor slices or TMAs were heated to 65°C for 1 hour prior to deparaffinization in Clearify and rehydration. Antigen retrieval was performed for 20 minutes in 10 mmol/L sodium citrate buffer pH 6 in a steam bath (95°C) and washed, followed by 3% H₂O₂ for 5 minutes. Slides were blocked for 1 hour in 2.5% goat serum at room temperature. Mouse primary antibody (OriGene, concentration 1:75 and Santa Cruz Biotechnology, 1:150) was added overnight at 4°C.

Prostate cancer cells (5 × 10⁵) were implanted with 100% Matrigel into the rear flank of 6- to 8-week-old NOD/SCID/IL2Rγ-null (NSG) mice. Tumors were grown to approximately 500 mm³ and harvested. Tumors were divided for either protein (flash-frozen in liquid nitrogen) or histologic analysis (fixed overnight at 4°C in 10% buffered formalin, paraffin embedded, and sliced at 4 μm thickness). IHC for xenografts was performed as described above.

Cells were trypsinized, pelleted, and washed three times with ice-cold PBS and lysed in RIPA buffer. Tumor tissues were cut into small pieces and ground in 1.5 mL Eppendorf tubes with microtube pestles for lysis in RIPA buffer. Protein concentrations of centrifuged whole lysates were determined with BCA assay. As described previously (27), in brief, either 40 or 50 μg of cell or tumor lysates were loaded in each lane in NuPAGE 4% to 12% bis-tris protein gels for electrophoresis at 200 V for 90 minutes. Wet transfer was performed using a Bio-Rad Transfer Kit at 300 mA for 90 minutes. The transferred nitrocellulose membrane was blocked in PBS containing 5% BSA and 0.1% Tween-20 for 1 hour. Primary antibody incubation was performed in the blocking buffer overnight at 4°C at the recommended concentration. The membrane was then washed with PBS containing 0.1% Tween-20 for 5 minutes, four times. Secondary antibody incubation (LI-COR donkey anti-mouse IgG IRDye 680 or anti-rabbit IgG IRDye 800CW, 1:10,000) was performed in the blocking buffer for 2 hours at room temperature. After washing four times with PBS containing 0.1% Tween-20, membranes were analyzed in a LI-COR Odyssey Imaging System.

M-SNAT (10 μmol/L, 10 μL of 1 mmol/L stock in DMSO) was diluted in reaction buffer containing 50 mmol/L HEPES, 100 mmol/L NaCl, and 0.1 mmol/L MgCl₂ at pH 7.5 with or without 1 mmol/L tris(2-carboxyethyl)phosphine (TCEP; 2 μL of 500 mmol/L stock in water) and 7.5 μg recombinant human MetAP2 in 1 mL final concentration. Reactions were performed at 37°C for 24 hours and monitored by high-performance liquid chromatography (HPLC).

For microscopic imaging, PC3, DU145, 22Rv1, PC3 MetAP2-knockdown (KD), and 22Rv1 MetAP2-KD cells were seeded at about 50% confluence on cover glasses in 6-well plates a day before incubation with 20 μmol/L M-SNAT (10 μL of 2 mmol/L stock in DMSO) in medium containing 1% FBS and TNP-470 (2 μmol/L, 2 μL of 1 mmol/L stock in DMSO). After incubation at 37°C for 24 hours, cells were washed three times with PBS, and then fixed with 10% formalin for 30 minutes. Cells were permeabilized in PBS-Triton X100 (0.1% volume/volume) at room temperature for 15 minutes, and then washed three times with PBS. Post-click assay solution was freshly prepared with 100 mmol/L ascorbic acid (100 μL of 1 mol/L stock in water, freshly prepared), 1 mmol/L CuSO₄ (10 μL of 100 mmol/L stock in water), 15 μmol/L (BimC₄A)₃ (0.5 μL of 30 mmol/L stock in water), and 5 μmol/L Cy5-azide (1 μL of 5 mmol/L in DMF) in 1 mL PBS. Cells were incubated with assay solution for 4 hours at 37°C, and then washed three times with PBS. Cells were then stained with 300 nmol/L DAPI in PBS for 10 minutes at room temperature, washed with PBS three times, and mounted on glass slides with antifade mounting medium.

The epifluorescence microscope images were acquired by 1 × 81 Inverted Microscope (Olympus) equipped with pE-4000 Illumination Systems (CoolLED) and ORCA-Flash4.0 Digital CMOS Camera (HAMAMATSU), with excitation at 405 nm for DAPI and 650 nm for Cy5. Digital images were reconstructed by MetaMorph software (v. 7.8.11.0) and analyzed using the ImageJ (NIH) software package.

The superresolution images were acquired by OMX BLAZE 3D-structured illumination, super-resolution microscope (Applied Precision) with simultaneous excitation at 405 nm for DAPI (emission filters 435/31 nm) and 642 nm for Cy5 (emission filters 683/40 nm), according to the standard procedure in the manufacturer’s instructions, and reconstructed with OMX softWoRx image processing software. Images were further processed using Volocity (v. 6.3.0, PerkinElmer) to generate the video.

For flow cytometry analysis, PC3 cells were trypsinized and resuspended after incubation in 20 μmol/L M-SNAT and medium containing 1% FBS with or without 2 μmol/L TNP-470 for 24 hours. Cells were centrifuged at 300 × g for 10 minutes and washed with PBS once, and then resuspended and fixed in 10% formalin at room temperature for 30 minutes. Cells were washed with PBS for three times by centrifugation, and then resuspended in post-click assay buffer at 37°C for 4 hours. Cells were washed with PBS for three times, and then analyzed with a 4-laser, 12-color DxP12 Cytek Upgrade (Becton Dickinson, Cytek Biosciences) and evaluated according to their size (forward scatter, FSC) on a linear scale, excited by 640 nm later, filtered by 655 nm long pass filter and 616/661 nm band pass filter. Flow cytometry data analysis was done using FlowJo V10 software. The mean fluorescence intensity was collected and plotted against the natural autofluorescence of cells.

Cell viability assay was performed with a CellTiter 96 AQueous One Solution Cell Proliferation Assay (MTS) Kit (Promega) according to the manufacturer’s protocol. PC3 cells were incubated with M-SNAT (100 pmol/L–100 μmol/L), cisplatin (500 μmol/L), and solvent (DMSO, 1%) in a 96-well plate at 37°C for 24 hours. The second day, medium was changed, and cells were incubated with MTS reagent for additional 3 hours. The absorbance at 490 nm was measured and plotted.

pLKO.1 control scramble short hairpin RNA (shRNA) vector (Addgene, plasmid #1864) or pLKO.1-MetAP2 shRNAs (Millipore Sigma, TRCN0000294143: GGATCATATACAGCG CAATTT; TRCN0000286742: GCAGAAGCACATCGACAAGTT; and TRCN0000306917: GACAAAGGCAAACCGTCTAAT) were cotransfected with pMDLg/pRRE, pRSV-Rev, and pMD2.G into 293T cells based on standard calcium phosphate transfection. The media with viral particles were collected at 24 and 48 hours, and 0.45 μm filtered for usage. A total of 2 × 10⁵ PC3 or 22Rv1 cells were seeded in 6-well plates for infection with 2 mL viral media containing polybrene (10 μg/mL). Seventy-two hours after transfection, the infected cells underwent selection using puromycin (1 μg/mL). The media with puromycin were changed every 3 days and the stable lines were generated after 9 days for further validation utilizing Western blot analysis to determine the KD efficiency of MetAP2.

M-SNAT (100 μmol/L, 1 μL of 2 mmol/L stock in DMSO) was diluted in 20 μL mouse serum and incubated at 37°C for 0 minute, 15 minutes, 30 minutes, 1 hour, and 2 hours. Serum proteins were denatured by mixing cold methanol (980 μL) and precipitated by centrifugation at 8,000 × g for 10 minutes. Supernatants were analyzed by HPLC and LC/MS. Percentage of tracer (relative area) was calculated as (peak area of tracer/total peak area on the HPLC chromatogram) × 100.

All experimental procedures using mice were performed in agreement with protocols approved by the Institutional Animal Care and Use Committee (IACUC) of Stanford University, as well as the USAMRMC Animal Care and Use Review in accordance with the laws of the United States and regulations of the Department of Agriculture.

For the MetAP2 inhibition model, 6- to 8-week-old male nu/nu nude mice were purchased from Charles River Laboratories. Mice were housed in the Research Animal Facility at the Clark Center of the Stanford University. A total of 5 × 10⁵ PC3 cells were injected subcutaneously into the right upper flank region of nu/nu nude mice in 50% Matrigel. When the tumors reached approximately 100 mm³, TNP-470 (stock in DMSO) was sequentially dissolved in ethanol and PBS, and then injected subcutaneously at a dose of 30 mg/kg at a position remote from the tumor (lower left back) every other day for 4 days and then injected every day for 5 additional days, with the last injection 4 hours prior to injection of PET tracer. Tumor dimensions were measured every other day by external caliper and tumor volume was calculated using the modified ellipsoid formula (length × width²)/2.

For PET imaging of MetAP2 stable KD models, NOD-SCID-IL2Rγ (NSG; Jackson Laboratory) male mice were bred at the Canary Center (Stanford, CA) in accordance with IACUC guidelines. A total of 2.5 × 10⁵ wild-type (WT), sh-ctrl, or sh-MetAP2 PC3 cells were injected subcutaneously into bilateral flanks of 6-week-old male NSG mice in 100% Matrigel. Tumors were measured every third day by external caliper, and tumor volume was calculated by use of the formula (length × width × height)/2. Tumors were grown to around 300–1,000 mm³ volume on average at the time of imaging.

For biodistribution studies, 1 × 10⁶ sh-ctrl or sh-MetAP2 PC3 cells were subjected to bilateral subcutaneous injection into the flanks of 6-week-old male NSG mice in 100% Matrigel. Tumors were measured every second or third day by external caliper, and tumor volume was calculated by use of the formula (length × width × height)/2. Tumors were grown to around 300 to 600 mm³ volume on average at the time of studies.

In experiments using the MetAP2 inhibitor, PET/CT imaging was performed using a Siemens Inveon micro PET/CT scanner (CT attenuation corrected). Totally, 8 mice were imaged in two batches, 4 mice each. Approximately 145 μCi of [¹⁸F]-M-SNAT (radiochemical purity of 99% as determined by HPLC; first batch, ∼14 pmol and second batch, ∼26 pmol) was injected intravenously into tumor-bearing mice. For the MetAP2 stable KD mouse model, PET imaging was performed using a Siemens Inveon D-PET micro-PET scanner (transmission attenuation corrected). CT was performed using a Sofie G-Next PET-CT scanner. Totally, 8 mice were imaged in two batches, 4 mice each. Approximately 79 μCi of [¹⁸F]-M-SNAT (radiochemical purity of 99% as determined by HPLC; first batch, ∼46 pmol and second batch, ∼80 pmol) was injected intravenously into tumor-bearing mice. Dynamic scans were acquired in list mode format over 90 minutes, and sorted into 22 times of frames, 0.5-mm sinogram bins for image reconstruction (4 × 15 seconds, 4 × 60 seconds, 11 × 300 seconds, and 3 × 600 seconds). Mice were anesthetized with 2% isoflurane near the center of the FOV to ensure the highest image resolution and sensitivity. Iterative reconstruction was performed using 3D-ordered subsets expectation maximization (OSEM), followed by fast maximum posteriori (MAP) with the following parameters: MAP OSEM iterations, 2; MAP subsets, 16; and MAP iterations, 18. PET images were registered with CT images. Siemens Inveon Research Workplace software v.4.2 was used for visualization of radiotracer uptake in the tumor, and to define the 3D volumes of interest (VOI). The relative tumor or organ radioactivity concentrations were from mean pixel values within the multiple VOIs and converted to counts per milliliter per minute and then divided by the injected dose (ID) to obtain an imaging VOI-derived percentage of the injected radioactive dose per cubic centimeter of tissue (%ID/cc). We assumed the average tissue density is close to density of water and converted %ID/cc to percentage of the injected radioactive dose per gram tissue (%ID/g).

Ex vivo biodistribution studies were performed to measure tumor- and tissue-associated radioactivity. Totally, 18 mice were analyzed in two batches, 9 mice in the first 40 μCi (∼38.28 pmol) injection batch (sh-ctrl, n = 4 and sh-MetAP2, n = 5) and 9 mice in the second 300 μCi (∼520.90 pmol) injection batch (sh-ctrl, n = 4 and sh-MetAP2, n = 5). [¹⁸F]-M-SNAT (radiochemical purity of 99% as determined by HPLC) was injected intravenously into tumor-bearing mice. Mice were sacrificed 60 minutes after injection to collect tumor, heart, lung, spleen, muscle, liver, kidney, prostate, and bone. The radioactivity in each tissue was measured using a gamma counter (Hidex automatic gamma counter) and normalized to tissue weight and the amount of radioactivity administered to each mouse, which was decay corrected to the time of radiotracer injection. Data were expressed as %ID/g values.

GraphPad Prism 7 was utilized for plotting and statistical analysis. The significant difference was determined by performing one-way or two-way ANOVA followed by Bonferroni multiple comparisons test and t tests (Mann–Whitney test, nonparametric) to determine the statistical significance with 95% confidence intervals with *, P < 0.0332; **, P < 0.0021; ***, P < 0.0002; ****, P < 0.0001; ns, not significant.

All data associated with this study are present in the article or Supplementary Materials and Methods. Materials are available and will be provided under the material transfer policies of Stanford University. These requests should be directed to the corresponding authors.

We first measured the protein levels of MetAP2 in a panel of prostate cell lines, including a benign human prostate epithelial cell line, PrEC (28), and prostate cancer cell lines PC3, DU145, and 22Rv1, and their corresponding tumor xenografts. PrEC cells expressed significantly lower levels of MetAP2 when compared with PC3, DU145, and 22Rv1 prostate cancer cells and xenografts (Fig. 1A). MetAP2 was also expressed in LNCaP prostate cancer xenografts and xenografts from two different models of NEPC, NCI-H660, and NEPC driven by the Trop2 oncogene (TD-NEPC; Fig. 1B and C; ref. 25). In contrast to MetAP2, PSMA was only detected in LNCaP xenografts, suggesting that utilizing MetAP2 as a target for new imaging modalities may detect additional subsets of prostate cancer missed by PSMA tracers. Together, these findings demonstrate that MetAP2 levels are elevated in models of adenocarcinoma and NEPC.

We further evaluated the levels of MetAP2 in human prostate cancer by IHC staining of radical prostatectomy sections and TMAs containing samples from 422 patients with prostate cancer, of which, 212 patients had 10 years of follow-up (29). As shown in Fig. 1D, benign prostate tissues had low levels of MetAP2, while low-grade prostate cancers (Gleason score 3 + 3) expressed moderate to high levels of MetAP2, and high-grade cancers (Gleason score 4 + 4 and 5 + 5) expressed high MetAP2 levels throughout the cancerous tissues (Fig. 1D and E; Supplementary Fig. S5). Benign prostatic hyperplasia (BPH) and benign adjacent to cancer tissues had significantly lower levels of MetAP2 than both low- and high-grade prostate cancers, while there was no significant difference between BPH and benign adjacent to prostate cancer (Fig. 1F). Among the 212 patients with long-term follow-up (recurrent, n = 52; nonrecurrent, n = 160), high levels of MetAP2 at the time of surgery were associated with biochemical recurrence (Fig. 1G; Supplementary Fig. S6).

Figure 2A illustrates the design of M-SNAT: the probe carries a methionine (P1, green) followed by a cysteine (P2) on the N-terminus that is connected to a 2-cyanopyrimidine (blue) through a linker. In eukaryotes, MetAP2 substrates prefer penultimate residues (P2) with a small nonpolar side chain (e.g., Gly, Ala, Ser, Thr, Val, and Cys; ref. 30). After intravenous administration, M-SNAT extravasates into tissues. In normal prostatic or BPH tissues with minimal MetAP2 activity, M-SNAT will mostly remain intact or be partially reduced and will, therefore, diffuse away freely. In prostate cancer with high MetAP2 activity, cleavage and release of the P1 methionine residue by MetAP2, as well as the reduction of the disulfide bond by intracellular glutathione (GSH), trigger the intramolecular condensation reaction. Unlike the unprocessed probes, cyclized macromolecules become rigid and tend to interact with each other to form molecular nanoaggregates (Fig. 2B). By tagging the probe with a radioisotope like fluorine-18, the location and amounts of these nanoaggregates can be detected in vivo through PET/CT imaging (Fig. 2C, top compound). For in vitro studies, a post-click reaction at the propargylglycine with an azide-conjugated far-red fluorophore, Cy5, was utilized to pinpoint the aggregated and precipitated M-SNAT in cells (Fig. 2C, middle compound). A control compound (M-SNAT-ctrl), with a methylated thiol that prevents intramolecular cyclization after methionine cleavage by MetAP2, was synthesized to examine the role of cyclization and aggregation in the activity of M-SNAT (Fig. 2C, bottom compound).

Macrocyclization of M-SNAT to produce M-SNAT-cyclized was monitored in solution using HPLC and mass spectrometry. After 24-hour incubation with recombinant human MetAP2 in the presence of TCEP to mimic the intracellular reducing environment, M-SNAT (10 μmol/L; HPLC retention time, T_R = 15.5 minutes) was converted completely into M-SNAT-cyclized (T_R = 11.8 minutes; Fig. 3A; Supplementary Fig. S7). In contrast, the M-SNAT-ctrl cannot cyclize due to the nonreducible methylated thiol group (Supplementary Fig. S8). Dynamic light scattering (DLS) analysis showed the formed nanoparticles with an average diameter of 861 nm (531–1281 nm; Fig. 3B), which was confirmed by transmission electron microscopy (TEM; Fig. 3C; Supplementary Fig. S9). MetAP2 catalyzed removal of the methionine residue and activated M-SNAT to undergo intramolecular macrocyclization and subsequent nanoassembly.

The uptake and MetAP2-specific activation of M-SNAT were evaluated in PC3 and DU145 prostate cancer cells. Post-click fluorescence labeling with Cy5-azide was performed to pinpoint the nanoaggregation in cells and to avoid possible interference with probe uptake incurred by conjugating the probe with a bulky Cy5 fluorophore prior to treating the cells (Fig. 4A). A well-characterized MetAP2 inhibitor (TNP-470), which inhibits the aminopeptidase activity by alkylating the His²³¹ residue in the active center, was used to evaluate the probe specificity (10). Under microscope, red fluorescence was observed in M-SNAT–treated PC3 and DU145 cells in a concentration-dependent manner after 24-hour incubation, while there was minimal fluorescence signal in cells with the control probe M-SNAT-ctrl (Fig. 4B; Supplementary Figs. S10–S12). When TNP-470 (2 μmol/L) was added, the fluorescence in both PC3 and DU145 cells was significantly reduced (Fig. 4B; Supplementary Fig. S11). Flow cytometry analysis of PC3 cells incubated with M-SNAT and postlabeled with Cy5-azide revealed 22.7% inhibition by TNP-470 (Fig. 4C; Supplementary Fig. S13). These data indicate the requirement of MetAP2 hydrolysis in conversion of M-SNAT to cyclized product and the macrocyclization-dependent probe nanoaggregation and retention in cells. We evaluated the toxicity of M-SNAT up to 100 μmol/L in PC3 cells and found little effect on the cell viability (Supplementary Fig. S14).

To further validate the specificity of M-SNAT activation by MetAP2, we generated MetAP2-KD PC3 and 22Rv1 cells. Three KD lines (M1–M3) of each cell line were generated via transduction of lentivirus carrying different shRNA sequences. A scramble shRNA was used to generate KD controls. The cell lines with the highest KD (PC3-M1 and 22Rv1-M2) were selected for cell-based studies (Fig. 4D and E). Previous work has shown that cells deficient in p53 show less inhibition of proliferation by TNP-470 due to disruption of p53-dependent induction of p21^Waf1/CIP1 to induce cell-cycle arrest (31, 32). As expected, MetAP2 KD slowed the growth of 22Rv1-M2 cells expressing WT p53, but had minimal impact on the growth of PC3-M1 (p53 null) cells (Fig. 4F and G; ref. 33). High doses of TNP-470, however, were reported to slow the growth of PC3 cells 30 days after tumor implantation (15). After incubation with M-SNAT for 24 hours, both PC3- and 22Rv1-KD cells showed much weaker red fluorescence than their WT counterparts (Fig. 4H; Supplementary Figs. S15 and S16). Taken together, these cell studies demonstrate specific intracellular activation and retention of M-SNAT due to MetAP2 activity.

Aggregated and Cy5 postlabeled M-SNAT assemblies observed in Fig. 4B were further visualized at a single-cell level using superresolution structured illumination microscope (SR-SIM), which provided both lateral (XY) and axial (Z) resolutions at twice the diffraction limit of conventional high-resolution light microscope systems, such as confocal (Fig. 5; Supplementary Movie S1). The size of particles formed in situ varied from a few hundred nanometers to more than 1 μm, consistent with the values measured by DLS and TEM (Fig. 3B and C). The postlabeled red fluorescence aggregates occurred throughout the cytoplasm with a relatively higher density surrounding the nucleus.

The ability of [¹⁸F]-M-SNAT to report MetAP2 activity in vivo was assessed in two different models of MetAP2 inhibition. To test the specificity of [¹⁸F]-M-SNAT for MetAP2 activity independent of tumor growth, we used a PC3 prostate cancer model in which cell growth was not affected by MetAP2 knockdown (Fig. 4F). PC3 cells (5 × 10⁵) were implanted subcutaneously in male nude mice (15) and measured every other day for 18 days until reaching approximately 100 mm³. Starting at day 18, PBS with or without TNP-470 (30 mg/kg) was injected subcutaneously every other day for 4 days, and then daily for 5 additional days, with the last injection 4 hours prior to PET imaging. Consistent with the in vitro knockdown experiments, there were no significant differences in tumor volumes (∼400 mm³) between the inhibitor- or PBS-treated groups over the short period of treatment (Fig. 6A). [¹⁸F]-M-SNAT was prepared in a rapid two-step chemical synthesis using the labeling chemistry as described previously (Fig. 2C; Supplementary Fig. S3; ref. 20). Mice received an intravenous injection of approximately 145 μCi [¹⁸F]-M-SNAT and underwent a programed 90-minute micro-PET/CT whole-body scan. As shown in Fig. 6B and C (corresponding 3D projection in Supplementary Fig. S17), coregistered PET (sky color) and CT (gray and white) images demonstrated that PC3 xenografts in PBS-treated mice retained higher radioactivity (up to 1.6 %ID/g) than tumors in mice treated with TNP-470 (around 1 %ID/g). Tumors were resected 20 hours after tracer injection for ex vivo analysis. As antibodies used for IHC recognize epitopes upstream and distant from the active binding pocket (β sheet 1–8, from Asn223; ref. 34), changes in expression of MetAP2 by IHC were not seen after inhibitor treatment (Supplementary Fig. S18). CD31 (PECAM1) staining revealed slightly decreased neovascularization in TNP-470–treated tumors, reflecting its known antiangiogenic effect (Supplementary Fig. S18).

To further evaluate the specificity of [¹⁸F]-M-SNAT for imaging MetAP2 activity, we developed MetAP2-KD PC3 xenograft models. A total of 2.5 × 10⁵ WT, MetAP2-KD (sh-MetAP2-M1), or scramble shRNA–transfected (sh-ctrl) PC3 cells (Fig. 4D) were suspended in 100% Matrigel and subcutaneously injected bilaterally into rear flanks of male NSG mice (n = 5, each group). Tumors were measured every 3 days after implantation. As observed in vitro (Fig. 4F), growth of PC3 tumors in NSG mice was not affected by MetAP2 KD (Fig. 6D). At day 30, mice carrying WT (n = 2), sh-MetAP2-M1 (n = 3), and sh-ctrl (n = 3) tumors were injected with [¹⁸F]-M-SNAT (∼79 μCi) and imaged by a micro-PET/CT for 90 minutes. Tumors larger than 900 mm³ or smaller than 300 mm³ were excluded from analysis to minimize the impact of tumor volumes on tracer uptake. The PET signal in sh-MetAP2-M1 tumors was significantly lower than WT and sh-ctrl tumors, with less than 1% uptake in sh-MetAP2-M1 tumors compared with 1.5% uptake in both WT and sh-ctrl tumors (Fig. 6E and F; corresponding 3D projection in Supplementary Fig. S19). Ex vivo analysis confirmed the lower expression of MetAP2 in imaged sh-MetAP2-M1 tumors than WT and sh-ctrl tumors (Fig. 6G). Whole-body distribution of [¹⁸F]-M-SNAT in both models at 65 minutes postinjection was analyzed by defining the 3D VOIs and showed that most radioactivity was in the kidney and bladder, with additional uptake in the liver (Fig. 6H). We compared the uptake in liver between PBS- and inhibitor-treated groups, but found no statistical differences, suggesting that uptake in the liver was not due to MetAP2 activity (Supplementary Fig. S20). Bone, muscle, and the pelvic region, in contrast, had very little uptake of the tracer, making it ideal for imaging primary and potentially metastatic prostate cancer lesions at these sites (Fig. 6H). NSG mice have been shown to have a faster clearance of antibodies and antibody–drug conjugates compared with nude mice due to an abnormal Fc interaction (35). We compared the biodistribution of [¹⁸F]-M-SNAT in NSG and nude mice, particularly in the spleen, and found they were similar overall (Fig. 6H). The extreme immunodeficiency in NSG mice did not appear to interfere with the biodistribution of [¹⁸F]-M-SNAT.

To validate the aggregation-dependent tracer retention in vivo, we performed a titration study with two different doses of [¹⁸F]-M-SNAT in our NSG/MetAP2-KD prostate cancer xenograft model: low (40 μCi) and high (300 μCi). Because the cyclic products from MetAP2 are more hydrophobic than the tracer itself and tend to interact with each other, a higher local concentration should promote nanoaggregation of the cyclic products, the mechanism for contrast enhancement. As shown in Fig. 6I (left), a high (300 μCi, 520.90 pmol) dose of [¹⁸F]-M-SNAT showed higher tracer radioactivity (%ID/g) at 1 hour postinjection than a low dose (40 μCi, 38.28 pmol; P = 0.0019) in NSG mice bearing PC3 xenografts (sh-ctrl). This result is consistent with the proposed nanoaggregation mechanism, and also provided direct evidence that the MetAP2-activated nanoaggregation could occur at the concentration as low as 40 μCi (P = 0.0005). Consistent to the imaging data, the distribution results showed that most radioactivity was in the liver and kidney. The prostate, bone, and muscle, in contrast, had very little uptake of the tracer (Fig. 6I, right). Finally, M-SNAT seems to be relatively stable in the serum, with a gradual decline from 100% to approximately 25% in 2-hour incubation at 37°C mainly because of hydrolysis of the methionine residue (Supplementary Fig. S21).

Using the TESLA strategy, we developed a novel molecular imaging technology for noninvasive imaging of MetAP2 activity based on conversion of a small, diffusible molecule into cyclized, nondiffusible form that undergoes nanoaggregation in the cytoplasm of cancer cells. The probe chemistry is flexible and allows easy modification for generating fluorescence read out at a single-cell level, as well as a radiative output signal for detection in vivo. Because the tracer retention directly depends on MetAP2 expression and activity to produce contrast (Figs. 4 and 6), the images correlate with the catalytic activity of the enzyme in situ. This MetAP2 imaging strategy has potential for clinical translation because there are ongoing needs to develop imaging tools that identify clinically aggressive prostate cancers, and MetAP2 tends to be expressed in some high-risk localized cancer and in model systems of castrate-resistant prostate cancer and NEPC.

Several PET tracers have been studied for imaging prostate cancer in localized disease to identify lesions in the prostate, when planning localized therapy to identify lymph node metastases, at time of biochemical recurrence after definitive local therapies, such as surgery or radiotherapy, and when metastasized to bone. The 2-[¹⁸F]fluoro-2-deoxy-D-glucose ([¹⁸F]FDG) PET is used in many cancers, but has limited sensitivity for imaging prostate cancer. Choline agents, [¹¹C]-acetate, and [¹⁸F]-fluciclovine have better sensitivity than conventional imaging technologies under particular clinical settings (36). The advent of PSMA PET imaging with either gallium-68 or fluorine-18 conjugated small molecules is revolutionizing the diagnosis and management of prostate cancer (37). However, PSMA is expressed in benign prostatic epithelium, resulting in false positive findings within the prostate (38), and is incrementally upregulated in primary cancer, lymph node, and distant metastases (39). PSMA expression does not correlate with tumor aggressiveness in localized prostate cancer or with risk of recurrence after surgery (40–42). Discovery of reliable prognostic biomarkers, especially as imaging agents capable of early-stage disease stratification, remains a clinical demand that could be met, in part, through MetAP2 imaging.

MetAP2 catalyzes critical N-terminus initiator methionine hydrolysis and has been studied as a target of antiangiogenic compounds in a variety of solid tumors, including prostate cancer (43, 44). Although MetAP2 has been heavily investigated as a therapeutic target, its use as a diagnostic imaging target had yet to be explored. In localized prostate cancer, high expression of MetAP2 in 54% of low-grade and 59% of high-grade cancers, and low or absent expression in noncancerous prostatic tissues (BPH and normal adjacent tissues) make it an attractive imaging target. Furthermore, the finding that prostate cancer cases with elevated MetAP2 have twice the risk of biochemical recurrence within 5 years following surgery suggest that MetAP2 imaging could be particularly useful for identifying clinically significant localized disease, a major unmet clinical need (2). In addition, the finding that MetAP2 highly expressed in model systems of castrate resistant and NEPC merits additional study since this imaging approach identifies MetAP2 enzymatic activity, thereby identifying tumors that could be treated with MetAP2 targeted therapies.

The synthesis and radiolabeling of [¹⁸F]-M-SNAT are straightforward and ideal for large-scale preparation. We characterized [¹⁸F]-M-SNAT in both inhibitor (TNP-470) treated and stable KD PC3 xenograft models. Because TNP-470 (50–200 mg/kg) treatment in nude mice has been reported with toxicity, and even death (15), we lowered the injection dose to 30 mg/kg and performed multiple injections till the tumors reached approximately 400 mm³ on average for PET imaging. Overall, a good tolerance was observed and no obvious variances in tumor size were noticed with this short treatment plan. There are several opportunities for improving the relatively moderate uptake of [¹⁸F]-M-SNAT by tumors (up to ∼2 %ID/g). For example, increasing stability of the tracer by incorporating a more bulky disulfide on the cysteine residue of [¹⁸F]-M-SNAT might impede nonspecific hydrolysis in the serum, but likely will not impact MetAP2 hydrolysis kinetics in cancer cells where it is reduced by GSH. A previous study by Zhou and colleagues reported a tripeptide analogue containing a mercaptoacetyl residue (P2) and alanyl (or phenylalanyl) residue (P3) showed excellent efficiency for human MetAP2 [k_cat/K_M = 2.8 × 10⁵ L/(mol*s); ref. 45]. Replacing the current glycyl residue by alanine may improve the hydrolytic kinetics by MetAP2, and, thus, increase signal retention.

MetAP2 PET imaging has promise to advance current imaging strategies in several ways. First, MetAP2 imaging is more likely to be active in more aggressive cancer, based on patterns of expression and, therefore, could enrich for identification of clinically significant prostate cancers. Second, loss of expression of PSMA in NEPC is often acquired after androgen deprivation therapies (7). As we observed, MetAP2 is highly expressed in model systems of castrate-resistant adenocarcinoma and NEPC, suggesting a unique role of MetAP2 PET imaging for tracking these histologic variants (Fig. 1B and C). Third, noninvasive PET imaging of MetAP2 activity could be invaluable as a companion diagnostic test to predict and monitor responses to MetAP2 inhibitors in the near future. Collectively, these data indicate that MetAP2 may be a valuable addition to the oncologist armamentarium for use as a prognostic indicator of biochemical recurrence and diagnostic imaging agent in prostate cancer.

J. Xie reports a patent for US20200085980A1 pending. Z. Chen reports a patent for caspase-3–triggered molecular self-assembling pet probes and uses thereof pending. J. Rao reports grants from NIH and U.S. Army Medical Research Acquisition Activity through the Congressionally Directed Medical Research Program during the conduct of the study, as well a patent for the imaging technology described in this article and filed by Leland Junior Stanford University pending. No disclosures were reported by the other authors.

Opinions, interpretation, conclusions, and recommendations are those of the authors and not necessarily endorsed by the U.S. Army and the funding agencies.

J. Xie: Conceptualization, software, formal analysis, funding acquisition, investigation, writing–original draft, writing–review and editing. M.A. Rice: Software, validation, investigation, methodology, writing–original draft, writing–review and editing. Z. Chen: Conceptualization, investigation. Y. Cheng: Conceptualization, investigation. E.-C. Hsu: Software, investigation, methodology. M. Chen: Software, investigation, methodology. G. Song: Investigation. L. Cui: Investigation. K. Zhou: Investigation. J.B. Castillo: Software, investigation. C.A. Zhang: Resources, software, validation, investigation. B. Shen: Resources, supervision, validation, investigation. F.T. Chin: Resources, formal analysis, supervision, validation, investigation. C.A. Kunder: Conceptualization, resources, formal analysis, supervision, validation, investigation, writing–review and editing. J.D. Brooks: Conceptualization, resources, supervision, funding acquisition, validation, writing–review and editing. T. Stoyanova: Conceptualization, resources, supervision, validation, investigation, writing–review and editing. J. Rao: Conceptualization, resources, supervision, funding acquisition, validation, writing–review and editing.

This work was supported by NIH grant U54CA199075 (Center for Cancer Nanotechnology Excellence for Translational Diagnostics). J. Xie thanks the Molecular Imaging Program at Stanford for the Molecular Imaging Young Investigator prize. M.A. Rice was supported by the U.S. Army Medical Research Acquisition Activity through the Congressionally Directed Medical Research Program (CDMRP) under award number W81XWH1810141. T. Stoyanova acknowledges the support from NIH/NCI (R37CA240822 and R03CA230819) and CDMRP (award number W81XWH1810323). M. Chen acknowledges the support from the Stanford Cancer Translational Nanotechnology Training Program funded by NCI award T32CA196585. L. Cui was supported by a CDMRP Breast Cancer Research Program Breakthrough postdoctoral fellowship award (award number W81XWH1810591). J.D. Brooks was supported by the NIH/NCI (U01CA196387). The SR-SIM was funded by an NIH High End Instrumentation grant (award number 1S10OD01227601) from the National Center for Research Resources. The authors thank J. Mulholland at Stanford Cell Sciences Imaging Facility for help on SR-SIM, Stanford Shared FACS Facility for instrumentation and assistance with flow cytometry, Functional Genomics Facility at Stanford for access to LI-COR Odyssey for Western blots, and Stanford Center for Innovation in In Vivo Imaging (SCI³) for micro-PET/CT imaging, and continued support of the Canary Foundation.

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.