Rapid imaging of lung cancer using a red fluorescent probe to detect dipeptidyl peptidase 4 and puromycin-sensitive aminopeptidase activities

Rapid identification of lung-cancer micro-lesions is becoming increasingly important to improve the outcome of surgery by accurately defining the tumor/normal tissue margins and detecting tiny tumors, especially for patients with low lung function and early-stage cancer. The purpose of this study is to select and validate the best red fluorescent probe for rapid diagnosis of lung cancer by screening a library of 400 red fluorescent probes based on 2-methyl silicon rhodamine (2MeSiR) as the fluorescent scaffold, as well as to identify the target enzymes that activate the selected probe, and to confirm their expression in cancer cells. The selected probe, glutamine-alanine-2-methyl silicon rhodamine (QA-2MeSiR), showed 96.3% sensitivity and 85.2% specificity for visualization of lung cancer in surgically resected specimens within 10 min. In order to further reduce the background fluorescence while retaining the same side-chain structure, we modified QA-2MeSiR to obtain glutamine-alanine-2-methoxy silicon rhodamine (QA-2OMeSiR). This probe rapidly visualized even borderline lesions. Dipeptidyl peptidase 4 and puromycin-sensitive aminopeptidase were identified as enzymes mediating the cleavage and consequent fluorescence activation of QA-2OMeSiR, and it was confirmed that both enzymes are expressed in lung cancer. QA-2OMeSiR is a promising candidate for clinical application.

Among malignant neoplasms, primary lung cancer is the leading cause of death worldwide 1 . Aggressive surgical resection improves the prognosis, and is increasingly used for elderly patients, patients with low lung function associated with emphysema, and patients with incidentally detected tiny lung cancers. In these cases, the risk-benefit ratio has led to an increasing number of patients undergoing sublobar resections, such as wedge resection and segmentectomy, instead of classic lobectomy. However, sublobar resection may shorten the distance between the tumor and the resection margin, which increases the risk of local recurrence. Indeed, the local recurrence rate is reported to be about 6% after reduction resection of lung cancer 2 . It is difficult to detect microcarcinomas near the surgical margins with the naked eye, and it is not practical to submit all margins for rapid pathological diagnosis. Therefore, in order to reduce local recurrence and improve the prognosis, there is a need for auxiliary diagnostic methods that can rapidly visualize microcarcinomas.
Fluorescent probes have many medical applications, including rapid diagnosis of cancer cells and evaluation of pathological conditions such as infection and inflammation. For example, focusing on the fact that the activity of ɤ-glutamyl transpeptidase is increased in certain tumors, we developed ɤ-glutamyl hydroxymethyl rhodamine green (gGlu-HMRG) as a fluorescent probe that is activated by this enzyme 3 (Supplemental Fig. 1A). This probe www.nature.com/scientificreports/ is expected to be applicable for rapid diagnosis and imaging-guided surgery of cancers such as breast cancer 4 and liver cancer 5 . Furthermore, gGlu-HMRG showed 43.8% sensitivity and 84.9% specificity for detecting lung cancer in surgical specimens 6 , and its utility for non-small cell lung cancer was confirmed in a prognostic study 7 . However, the sensitivity and specificity of gGlu-HMRG are not sufficient for clinical application to detect lung cancer. We considered that the probe performance might be improved in two ways. First, a probe emitting red fluorescence is expected to provide increased sensitivity, because autofluorescence of lung tissue is minimal in this wavelength range. Secondly, a probe targeting plural target enzymes whose activity is increased in lung cancer is expected to show increased specificity. Our laboratory has developed a series of red fluorescent probes (Supplemental Fig. 1B) based on the 2-methyl silicon rhodamine (2MeSiR) scaffold, which has an absorption maximum at 593 nm and a fluorescence maximum at 613 nm 8,9 . Notably, its absorption wavelength shifts to the shorter wavelength side when a peptide chain is introduced on the amino group of one of the xanthene rings via an amide bond. Thus, the resulting compound does not emit fluorescence upon excitation at around 590 nm, but after cleavage, the released fluorophore is efficiently excited at this wavelength and emits strong fluorescence. We also modified 2MeSiR to obtain 2-methoxy silicon rhodamine (2OMeSiR). In this case, the fluorescence of the peptide derivative is quenched by photoinduced electron transfer, and activated after cleavage of the peptide bond, affording a red fluorescent probe with a higher fluorescence activation ratio 10 (Supplemental Fig. 1C).
Building on that work, the purpose of the present study is to select and validate the best red fluorescent probe for rapid diagnosis of lung cancer from a 400-probe library of 2MeSiR-based probes, as well as to identify the target enzymes that activate the probe, and to confirm their expression in cancer cells. We first screened the probes against lysates from lung cancer surgical specimens, and then tested the most effective candidates on fresh tissue specimens. Next, we modified the selected probe, QA-2MeSiR, to obtain glutamine-alanine-2-methoxy silicon rhodamine (QA-2OMeSiR), which has the same side chain structure as QA-2MeSiR but shows lower baseline fluorescence. The enzymes that activate QA-2OMeSiR fluorescence were identified and validated by several methods (Fig. 1).

Results
I-1. Screening using lung adenocarcinoma lysates. The peptide sequences of the probes and amino acid abbreviations are shown in Supplemental Tables 1 and 2. Lysates from 11 lung adenocarcinoma cases were used for initial screening. Among the 400 probes, 62 probes met the criteria of "elevated fluorescence of tumor lysate/normal lysate (TN ratio) ≥ 2.5" and "top 30% of elevated fluorescence of tumor lysate" in four or more cases; these are listed in Supplemental Table 3. An example of the change in fluorescence intensity over time is shown in Supplemental Fig. 2. I-2. Screening using fresh lung cancer specimens. A total of 23 fresh lung cancer specimens (16 adenocarcinoma, 5 squamous cell carcinoma, one atypical carcinoid, and one small cell carcinoma) were used to screen the 62 probes identified by lysate screening, together with 2MeSiR without any side chain. An exam-   Fig. 3, together with a dot chart showing the increase in fluorescence in tumor and normal tissues, and sequential pseudo-color images. Overall, the highest AUC (0.962) was observed at 10 min. The sensitivity was 96.3% and the specificity was 85.2%. The results are summarized in Supplemental Table 5. At 10 min, the average T/N ratio www.nature.com/scientificreports/ was 2.93, and T/N was ≥ 2 in 21/27 cases (78%). There were no significant differences by histological type, gender, or pathological stage within the measured range.

I-4. In-vitro comparison of QA-2MeSiR and QA-2OMeSiR.
In both fluorescence images measured by Maestro ® (CRi Inc., Woburn, MA) at 640 nm (Supplemental Fig. 3A) and digital camera images (iPhone6 ® Apple, Cupertino, CA) (Supplemental Fig. 3C), which are considered to be similar to naked-eye observation, QA-2OMeSiR showed weaker fluorescence than QA-2MeSiR ( Fig. 3C). On the other hand, there was no marked difference between 2OMeSiR and 2MeSiR in either case. The fluorescence intensity of 2OMeSiR was 52.8 times higher than that of QA-2OMeSiR, while that of 2MeSiR was 10.2 times higher than that of QA-2MeSiR (Supplemental Fig. 3D).

I-5. Application of QA-2OMeSiR to marginal lung cancer lesions. Examples of QA-2OMeSiR
imaging of borderline lesions of adenocarcinoma and squamous cell carcinoma are shown in Fig. 4A-E and in Fig. 4F-J, respectively. HE staining confirmed the presence of tumor cells in the highly fluorescent areas and normal tissue in the weakly fluorescent areas. The image closest to that seen with the naked eye through the filter was the image taken by the digital camera, and this also showed a clear difference between tumor and normal tissues. Based on these results, the following studies were performed using QA-2OMeSiR.

II-1. Selection of target enzyme candidates using diced electrophoresis gel assay. Diced elec-
trophoresis gel (DEG) assay was performed to search for target enzymes, using QA-2OMeSiR and adenocarcinoma lysates #10 and #12. In the first assay, three peaks were observed (#10) (Supplemental Fig. 4A). These peaks were subjected to peptide mass fingerprinting (PMF) analysis as described below. DEG assay was also performed on lung adenocarcinoma sample lysate #12, and similar peaks were observed (Supplemental Fig. 4B). Next, DEG assay was performed with QA-2OMeSiR and squamous cell carcinoma lysate #21 (Supplemental Fig. 4C). In the squamous cell carcinoma lysate, there was no increase in fluorescence at the location of the major peak observed in the adenocarcinoma lysate, but there was a peak at the location corresponding to the weak peak observed in the adenocarcinoma lysate (③ in Supplemental Fig. 4A,B). The PMF results afforded a list of candidate target enzymes (Supplemental Tables 6-8). The analysis of part ① in Supplemental Fig. 4 identified acylamino-acid-releasing enzyme (AARE) and dipeptidyl peptidase 4 (DPP4) as candidate target enzymes. The analysis of part ② yielded bleomycin hydrolase, calpain 1, and DPP4 as candidates, while part ③ afforded puromycin-sensitive aminopeptidase (PSA). Next, the putative involvement of these five enzymes was examined using inhibitors. www.nature.com/scientificreports/
These results indicate that the target enzymes of QA-2OMeSiR are DPP4 and PSA. Since neither DPP4 nor PSA inhibitor alone fully blocked the fluorescence increase of QA-2OMeSiR, we performed an additional experiment using both sitagliptin and puromycin simultaneously (Fig. 5A). The combined inhibitors showed potent inhibition at concentrations above the IC50 (the results are shown in light blue in the figure for 10 nM sitagliptin, 1 µM puromycin and higher concentrations). Therefore, both DPP4 and PSA were considered to be target enzymes.  www.nature.com/scientificreports/  www.nature.com/scientificreports/ cell line and the results of the t-test for differences between the siRNAs used. Knockdown of either DPP4 or PSA suppressed the response of QA-2OMeSiR. The fact that the suppression was only partial in each case supports the involvement of both enzymes in the reaction.  www.nature.com/scientificreports/ II-5. Immunohistochemical staining. As shown in Fig. 7, both adenocarcinoma and squamous cell carcinoma showed positive immunostaining for DPP4. On the other hand, in normal lung tissue, some macrophages and alveolar epithelium were DPP4-positive, while others were negative. PSA staining was also positive in both adenocarcinoma and squamous cell carcinoma cells, while normal lung tissue was negative.

II-6. Expression of DPP4 and PSA in lung cancer and normal tissue.
Whole RNA was extracted from lung cancer and normal lung tissues, and the expression levels of DPP4 and PSA were evaluated by means of quantitative polymerase chain reaction (qPCR) (Fig. 8). In adenocarcinoma, both DPP4 and PSA were more highly expressed in the tumor than in normal lung tissue (n = 3). In squamous cell carcinoma cases, DPP4 was weakly expressed in both normal and tumor tissues, whereas PSA was highly expressed in tumor tissues, but not in normal tissues (n = 3).

Discussion
First, in connection with the probe screening, it should be noted that the results in the initial screen using frozen tissue lysates were not fully consistent with the results of the subsequent screen using fresh tissue samples, possibly due to loss of enzyme activity in the frozen samples. Therefore, it is important to confirm screening results in fresh specimens. We confirmed the absence of interference from tissue autofluorescence by using 2MeSiR (Supplemental Table 4). In this study, QA-2MeSiR imaging of surgically resected fresh specimens showed a sensitivity of 96.3% and specificity of 85.2% at 10 min, representing a significant improvement over the results previously obtained with gGlu-HMRG 6 . Another probe, glutamic acid-proline-HMRG (EP-HMRG), showed a sensitivity of 96.9% and specificity of 85.7% for esophageal cancer 11 , and has also been applied to detect early-stage esophagogastric junction adenocarcinoma 12 as well as head and neck cancer 13 . EP-HMRG is currently under clinical trial. In the present study, we did not find a correlation between fluorescence elevation and malignancy (histological type and progression) (Supplemental Table 5), but this may be due to the small number of cases and the lack of fresh specimens of early-stage cancer. Further studies will be needed to address this issue.
Two target enzymes, DPP4 and PSA, were identified according to the method of Komatsu et al. 14 . The finding that the fluorescence increase was almost completely suppressed by inhibitors of these two enzymes in the Figure 8. Quantitative PCR analysis of DPP4 and PSA expression in lung cancer. In adenocarcinoma, both DPP4 and PSA were more highly expressed in the tumor than in normal lung tissue (n = 3, #11, #14, #17). In squamous cell carcinoma, DPP4 was rarely observed in either normal or tumor tissues, but PSA was frequently observed in tumor tissues (n = 3, #22, #23, #24). www.nature.com/scientificreports/ experiment with lung cancer tissue lysates also supports their involvement. In contrast, inhibitors of the other candidate enzymes had no effect in similar experiments, ruling out a contribution of these enzymes. This was confirmed by the use of siRNA-transfected cell lines, purified enzymes, immunostaining and qPCR. We found that the Km values of QA-2OMeSiR for DPP4 and PSA were 24.4 µM and 24.6 µM, respectively, and these values are similar to the reported Km value of 17.7 µM for EP-HMRG, which also targets DPP4 11 . Interestingly, not all of the samples expressed both DPP4 and PSA. For example, the qPCR results in Fig. 8 show that ad 1,2 and sq 3 express both DPP4 and PSA, but ad 3 expresses mainly DPP4, while Sq 2,3 expresses mainly PSA. This suggested that the lung cancers tissues examined here were heterogeneous. DPP4 is a membrane-bound soluble protein that is widely distributed in the body 15 and specifically cleaves the C-terminal peptide bond of the second alanine or proline from the N-terminal 16 . Physiologically, it inactivates incretin, an intestinal hormone that promotes insulin secretion, and is involved in immune regulation, signal transduction, and apoptosis 17 . It also plays a role in cancer progression and suppression, and is highly expressed in lung cancer cells 18 . PSA is a cytosolic alanyl aminopeptidase 19 , which may have a role in nervous system function 20,21 . Recent studies indicate that PSA inhibitors may be useful in the treatment of hematological malignant diseases 22 . Our siRNA, immunostaining and qPCR results indicated that not only DPP4 but also PSA is highly expressed in lung cancer cells. Thus, although neither DPP4 nor PSA is expressed specifically in cancer cells, the combination may be more effective than either enzyme alone to detect lung cancer cells. DPP4 is localized on the cell surface and PSA in the cytosol, and the reaction of QA-2OMeSiR with the enzymes is expected to occur at these locations after application of the probe (see Supplemental Fig. 11). It would be difficult to rationally design probes targeting both enzymes, and we consider that screening is an effective approach to find such multi-target probes.
We found that some normal lung tissues showed slight fluorescence elevation, which appeared to be associated with interstitial pneumonia or emphysema. Macrophages were immunostained for DPP4 (Fig. 7), and there are many reports showing that DPP4 has a role in the inflammatory process 23 . In addition, it has been reported that DPP4 is expressed in the alveolar epithelium during the development of fibrosis in interstitial pneumonia 24 . Thus, it is necessary to take into account the possibility that increased fluorescence may be observed in normal tissues in patients with such conditions.
In conclusion, our results indicate that the red fluorescent probe QA-2OMeSiR, which targets both DPP4 and PSA, is effective for the rapid detection of even borderline cancer lesions. This probe is a promising candidate for early clinical introduction. It may also have potential for application to detect metastatic lymph nodes and disseminated lesions. Furthermore, the screening strategy described here should be applicable to find probes for visualizing other types of malignant lesions that overexpress different enzymes. Application of probes to fresh specimens. Fresh specimens were divided into 1 mm pieces of tumor and normal tissue, which were placed on µ-Slide 8-Well high Bionert (ibidi GmbH, Gräfelfing, Germany). Whitelight (measurement wavelength: 600 nm, exposure time: 10 ms) and fluorescence (measurement wavelength: 640 nm, exposure time: 100 ms) images were acquired using the Maestro ® . 200 µL of probe solution, adjusted to 50 µM, was dropped into each dish. In the case of 2MeSiR, a concentration of 5 µM was used. Fluorescence images were acquired after 1, 3, 5, 10, 20, and 30 min, and the average signal of the ROI (region of interest) was calculated in the 640 nm image. The difference between the values at 3, 5, 10, 20, and 30 min and the value at 1 min was taken as the fluorescence increase, and the values for normal and tumor tissues were obtained. Filter settings were 570/40 nm for excitation and 610 nm for emission. Wavelengths from 600 to 850 nm were measured at intervals of 10 nm. All subsequent measurements were performed with the same settings. www.nature.com/scientificreports/ Comparison of QA-2MeSiR and QA-2OMeSiR. QA-2OMeSiR was obtained in powder form as the trifluoroacetate salt from Goryokayaku Co., Tokyo, Japan. 2OMeSiR was obtained from our university's stock in powder form 7 . Each probe was dissolved in DMSO and stocked as a 10 mM solution, which was stored in a freezer at −30 °C and used as required. QA-2MeSiR and QA-2OMeSiR were used at the concentration of 50 µM, while 2MeSiR and 2OMeSiR were used at the concentration of 5 µM. Images were obtained with Maestro ® and with a digital camera (iPhone 6 ® ).

Methods
Observation of borderline lesions using QA-2OMeSiR. In the case of resected specimens that contained relatively large tumors, the transition zones between normal and tumor tissues were sectioned in the pathology department. The fresh samples were placed on a µ-Dish 35 mm, low (ibidi GmbH, Gräfelfing, Germany) and 500 µL of 50 µM QA-2OMeSiR was dropped onto the entire sample.
Identification of target enzyme candidates by diced electrophoresis gel (DEG) assay and peptide mass fingerprinting (PMF). DEG assay is a technique developed by Komatsu et al., in which enzyme extracts such as cell lysates are fractionated by nondenaturing two-dimensional electrophoresis. The gels are diced and the pieces are dispensed into multiwell plates, in which enzyme assays using fluorescent probes are performed 15 (Supplemental Fig. 12). Fluorescence intensity changes were measured by EnVision ® immediately and after 12 h. Proteins in gel pieces showing a large fluorescence increase were analyzed by the PMF method at Intégrale Co., Ltd., Tokyo, Japan. water, deparaffinized, and microwaved at 98 °C for 20 min to activate the antigen. The sections were then treated with 3% hydrogen peroxide to block endogenous peroxidase activity, followed by endogenous biotin inactivation using an avidin/biotin blocking kit. The primary antibody CD26 (H-270) (sc-9153 Santa Cruz Biotechnology) was used at a dilution of 1/500 and PSA (sc-390184: mouse monoclonal) was used at a dilution of 1/50. The reaction was carried out for 2 h at room temperature, and counter-stained with hematoxylin.

Reaction of lung cancer lysates with
Expression of DPP4 and PSA in lung tissue. Frozen specimens of lung tumors and normal tissues were shredded with scissors and RNA was extracted as previously described.Primer sequences for hDPP4 and hPSA, as well as hGAPDH and hHPRT1 (internal standards), are shown in Supplemental Table 11. Other details are shown in Supplemental www.nature.com/scientificreports/ using a LightCycler ® 480 System II (F. Hoffmann-La Roche Ltd, Basel, Switzerland). The mRNA expression levels of DPP4 and PSA were expressed relative to internal standards and compared between tumor and normal tissues.