Fibroblast Activation Protein-Targeting Minibody-IRDye700DX for Ablation of the Cancer-Associated Fibroblast with Photodynamic Therapy

Fibroblast activation protein (FAP), expressed on cancer-associated fibroblasts, is a target for diagnosis and therapy in multiple tumour types. Strategies to systemically deplete FAP-expressing cells show efficacy; however, these induce toxicities, as FAP-expressing cells are found in normal tissues. FAP-targeted photodynamic therapy offers a solution, as it acts only locally and upon activation. Here, a FAP-binding minibody was conjugated to the chelator diethylenetriaminepentaacetic acid (DTPA) and the photosensitizer IRDye700DX (DTPA-700DX-MB). DTPA-700DX-MB showed efficient binding to FAP-overexpressing 3T3 murine fibroblasts (3T3-FAP) and induced the protein’s dose-dependent cytotoxicity upon light exposure. Biodistribution of DTPA-700DX-MB in mice carrying either subcutaneous or orthotopic tumours of murine pancreatic ductal adenocarcinoma cells (PDAC299) showed maximal tumour uptake of 111In-labelled DTPA-700DX-MB at 24 h post injection. Co-injection with an excess DTPA-700DX-MB reduced uptake, and autoradiography correlated with FAP expression in the stromal tumour region. Finally, in vivo therapeutic efficacy was determined in two simultaneous subcutaneous PDAC299 tumours; only one was treated with 690 nm light. Upregulation of an apoptosis marker was only observed in the treated tumours. In conclusion, DTPA-700DX-MB binds to FAP-expressing cells and targets PDAC299 tumours in mice with good signal-to-background ratios. Furthermore, the induced apoptosis indicates the feasibility of targeted depletion of FAP-expressing cells with photodynamic therapy.


Introduction
Fibroblast activation protein (FAP) is a type II integral serine protease that is expressed by activated fibroblasts. It facilitates remodelling of the extracellular matrix and therefore is involved in functions such as cell adhesion, invasion, and motility. In over 90% of epithelial tumours, FAP expression is found on the cancer-associated fibroblasts (CAFs). These cells have a critical role in promoting tumour growth, invasion, metastasis, and immunosuppression and ultimately in the clinical outcome of patients [1][2][3][4][5]. The development and application of tracers for non-invasive imaging FAP-expressing cells have therefore gained widespread interest over the past years [6,7].
Besides imaging, targeted depletion of FAP-expressing cells is an appealing treatment option. Depletion of FAP-expressing cells was investigated in transgenic mice with FAP-promotor driven diphtheria toxin receptor (DTR) [5] or using pharmacological FAP inhibitors [8], FAP-targeting vaccines and T-cell therapies [9][10][11][12], and FAP-targeting molecules coupled to cytotoxic compounds [13][14][15][16][17]. These compounds inhibit tumour growth and induce anti-tumour immune responses in various preclinical models; however, some studies also describe severe systemic toxicity, including cachexia and anaemia [18,19]. This indicates the importance of FAP-expressing cells in maintaining normal muscle mass and haematopoiesis. Furthermore, it warrants caution, as FAP expression is also found in wound healing and normal tissues such as the placenta, uterine stroma, embryonic tissue, and multipotent bone marrow stromal cells [5,20,21].
Targeted photodynamic therapy offers a potential solution to these limitations. Here, a targeting compound conjugated to a photosensitizer is activated by light of a specific wavelength, inducing production of reactive oxygen species and resulting in cell death. As opposed to non-targeted photodynamic therapy using the photosensitizer alone, in this approach, off-target toxicity is avoided. Furthermore, as the photosensitizer conjugate is only activated in the region of interest, the effects of eliminating target-expressing cells in other tissues or organs are avoided. The silica-phthalocyanine derivative IRDye700DX is highly suitable and often used for targeted photodynamic therapy, as the water-soluble axial ligands increase hydrophilicity, and the NHS ester enables conjugation to targeting proteins such as antibodies. Furthermore, it is activated with near-infrared light (690 nm), is very photostable, and = has a high singlet oxygen quantum yield [22][23][24].
We and others have proven the feasibility of FAP-targeted photodynamic therapy with a monoclonal antibody [25][26][27][28][29][30][31] or with nanoparticles [32,33]. Though high binding affinity and long circulation time often lead to high uptake of monoclonal antibodies at the target site, smaller molecules could lead to better tissue penetration. In addition, faster blood clearance could lead to a shorter time interval between injection and exposure to light. Other studies have shown similar or increased efficiency of photodynamic therapy with smaller molecules such as nanobodies (~15 kDa), diabodies (~45 kDa), or minibodies (~75 kDa) when compared to a monoclonal antibody [34,35]. Here, we optimized FAP-targeted photodynamic therapy with a minibody. These are comprised of antigen binding single-chain fragment variable regions fused to the C H 3 region of the human Fc domain. Minibodies maintain the specificity and avidity of full-length antibodies while being biologically inert and lacking immune effector functions. The use of minibodies in non-invasive imaging has shown successful in preclinical [36,37] and initial clinical results [38,39].
Here, we generated a minibody conjugated to the phthalocyanine-based photosensitizer IRDye700DX for application in photodynamic therapy and the chelator DTPA for 111 In-labeling and quantification or visualization of the tracer in biodistribution and imaging studies. We aim to provide proof of concept for the feasibility of FAP-targeted photodynamic therapy in FAP-expressing cell lines in vitro and in a syngeneic murine model carrying subcutaneous and orthotopic tumours of a murine pancreatic ductal adenocarcinoma (PDAC) cell line, as PDAC is a tumour type in which FAP-expressing CAFs play a large role.

In Vitro Binding and Internalization of DTPA-700DX-MB
3T3, 3T3-FAP, and PDAC299 cells were grown to 80% confluency in 6-well plates and incubated with 1600 Bq 111 In-labelled DTPA-700DX-MB in binding buffer (DMEM with 0.5% BSA) at 37 • C for 1, 2.5, or 24 h. After washing twice with PBS, cells were incubated with ice-cold 0.1 M acetic acid and 154 mM NaCl (pH 2.6) for 10 min at 4 • C to collect the membrane-bound fraction. After washing, cells were lysed using 0.1 M NaOH, the lysate was collected, and activity in both fractions was counted in a γ-counter (2480 Wizard 3 , LKB/Wallace, Perkin-Elmer, Boston, MA, USA).

Singlet Oxygen Production
250 nM DTPA-700DX-MB was incubated with 50 µM p-nitrosodimethylaniline (RNO) and 400 µM imidazole as an acceptor of singlet oxygen in PBS pH 7.4, in clear, flat-bottom 96well plates (Costar) and exposed to 690 nm light at 200 mW/cm 2 with a light emitting diode (LEDfactory, Leeuwarden, The Netherlands) [41]. Absorbance at 440 nm was measured every minute up to 13 min (156 J/cm 2 total light dose) with the Tecan Infinite ® 200 Pro (Tecan, Männedorf, Switzerland) to determine singlet oxygen-induced bleaching of RNO.

In Vitro Targeted Photodynamic Therapy with DTPA-700DX-MB
3T3, 3T3-FAP, and PDAC299 cells were grown to 80% confluency in 96-well plates and incubated with DTPA-700DX-MB in binding buffer (DMEM with 0.5% w/v BSA, BB) or with BB alone at 37 • C in a humidified atmosphere with 5% CO 2 for 2.5 h. After washing with PBS, regular culture medium was added. The cells were then exposed to 0 or 60 J/cm 2 690 nm light at a fluency rate of 200 mW/cm 2 . Cell viability was measured 24 h after tPDT using the CellTiter-Glo assay (Promega, Madison, WI, USA). Furthermore, 3T3-FAP cells were labelled with Vybrant DiO dye according to manufacturer's protocol (Thermo Fisher Scientific, Waltham, MA, USA) and cocultured with PDAC299 tumour cells in Costar 96-well flat-bottom, clear plates in DMEM with 10% FCS. Cells were incubated with 3 nM DTPA-700DX-MB for 2 h at 37 • C, and upon washing, the plate was irradiated with 60 J/cm 2 690 nm light. After 24 h, the cells were incubated with 1 µg/mL propidium iodide (PI) in PBS. After washing twice with PBS, cells were imaged with the EVOS FL cell imaging system with the suitable LED cubes (PI = RFP; DiO = GFP).

Animals
The Radboud University animal ethics committee approved study protocols (CCD number AVD1030020209645). All procedures were performed according to the Institute of Laboratory Animal Research guide for Laboratory Animals. All 6-8-week-old female C57BL/6 mice (6JRj, Janvier) were acclimatized for 1 week upon arrival at our institution. They were fed standard chow ad libitum and housed on a 12 h day-night cycle.
Subcutaneous model-Mice were injected subcutaneously at one or both shoulders with 5 × 10 5 PDAC299 tumour cells resuspended in 100 µL of DMEM. Mice were weighed, and tumours were measured with a calliper twice a week. Tumour size was calculated as follows: tumour size = (4/3) * π * (L/2) * (W/2) * (D/2), with L being tumour length, W being tumour width, and D being tumour depth. When (one of the) tumours reached a size of 200 mm 3 , usually within 10-14 days, animals were included in the experiments.
Orthotopic model-The pancreas was exposed by making a surgical incision just below the spleen. Either 5 × 10 3 (n = 5) or 2.5 × 10 4 (n = 4) PDAC tumour cells resuspended in 30 µL of DMEM and Matrigel (1:1) were injected into the pancreas with a 27G needle. The wound was closed with a surgical suture, and mice were weighed and imaged with ultrasound (Vevo 2100 preclinical imaging system, Visual Sonics, Amsterdam, The Netherlands, MS-550D transducer) twice per week. At 32 days post injection, when the tumours reached a size of 300-500 mm 3 , mice were included in the experiment. , and 48 h (n = 5) post injection, mice were sacrificed through CO 2 /O 2 asphyxiation. Three additional mice were co-injected with a 7.4-fold excess nonconjugated and unlabelled minibody and sacrificed at 24 h to determine specificity of uptake. Tumour, blood, and organs were collected, weighed, and counted in a γ-counter (WIZARD, 2480 Automatic Gamma Counter, Perkin-Elmer, Boston, MA, USA). Uptake in the tissues was calculated as the percentage of the injected activity per gram of tissue (%IA/g). Tumours of the mice injected with 10 MBq were processed for autoradiography analyses as described below.
Orthotopic model-0.3 nmol 10 MBq (for autoradiography analyses and ex vivo biodistribution) or 1 MBq (for ex vivo biodistribution only) 111 In-labelled DTPA-700DX-MB in 200 µL PBS was injected intravenously in mice carrying orthotopic PDAC299 tumours (5 × 10 3 cells n = 3; 2.5 × 10 4 cells n = 4). Two additional mice in the group injected with 5 × 10 3 tumour cells were co-injected with a 7-fold excess non-conjugated and unlabelled minibody. At 24 h post injection, mice were sacrificed through CO 2 /O 2 asphyxiation. Tumour, blood, and organs were collected, weighed, and counted in a γ-counter (WIZARD, 2480 Automatic Gamma Counter, Perkin-Elmer, Boston, MA, USA). Uptake in the tissues was calculated as the percentage of the injected activity per gram of tissue (%IA/g). Tumours of the mice injected with 10 MBq were processed for autoradiography analyses as described below.

MicroSPECT/CT
A SPECT/CT scan was acquired for 30 min 24 h post injection of one mouse carrying an orthotopic PDAC299 tumour with 10 MBq 111 In-labelled DTPA-700DX-MB using a U-SPECT/CT-6 (MILabs, Utrecht, The Netherlands). Images were acquired using a 1 mm diameter pinhole ultra-high-sensitivity mouse collimator. The SPECT scan was reconstructed with three iterations and a voxel size of 0.4 mm (MILabs reconstruction software version 12.00), and the image was made with VivoQuant software (Version 2020R2 build9).

Autoradiography
Tumours were harvested and formalin-fixed and paraffin-embedded (FFPE). They were cut into 4 µm sections and mounted on superfrost glass slides. A phosphor screen (Fuji Film BAS-IP SR 2025, Raytest, Straubenhardt, Germany) was exposed to the slides for 72 h in a Fujifilm BAS cassette 2025. Then, the images were acquired with a Typhoon FLA 7000 laser scanner (GE healthcare Life Sciences, Chicago, IL, USA) at a pixel size of 25 × 25 µm. Images were analysed with Aida Image analysis software (Version 4.21, Elysia Raytest, Germany).

In Vivo FAP-tPDT with DTPA-700DX-MB in the Subcutaneous PDAC299 Model
Mice carrying subcutaneous tumours of PDAC299 on both shoulders were injected with 0.6 nmol DTPA-700DX-MB in PBS (n = 6) or PBS alone (n = 5). At 24 h post injection, mice were anaesthetized with isoflurane (1.5% in 1 L/min oxygen). Uptake in both tumours was visualized with fluorescence imaging (Excitation filter 640 nm, Emission filter Cy5.5 (IVIS Lumina II, PerkinElmer, Waltham, MA, USA)), and then, one of the tumours was exposed to 50-200 J/cm 2 690 nm light at 230-300 mW/cm 2 (DTPA-700DX-MB group: one dose of 55 J/cm 2 at 280 mW/cm 2 (n = 1), three doses of 55 J/cm 2 at 280 mW/cm 2 (n = 1), two doses of 100 J/cm 2 at 230 mW/cm 2 (n = 1), and one dose of 100 J/cm 2 at 230 mW/cm 2 (n = 3); and the PBS group: one dose of 100 J/cm 2 at 230 mW/cm 2 (n = 5)). We irradiated the left tumour unless indicated otherwise. The rest of the body was shielded from light using paper towels and aluminium foil. After irradiation, the treated tumour was imaged for fluorescence again. Mice were sacrificed at 24 h post light irradiation with CO 2 /O 2 asphyxiation. Tumours, livers, kidneys, and spleens were obtained and formalin-fixed and paraffin-embedded.
For the FAP staining, antigen retrieval was performed by heating in EDTA for 10 min at 96 • C in a PT Module (Thermo Fisher Scientific, Waltham, MA, USA). The endogenous peroxidase activity was blocked by incubating with 3% H 2 O 2 in PBS for 10 min at room temperature, and endogenous biotin was blocked with a biotin/avidin blocking kit (VEC-TASTAIN, Thermo fisher) for 15 min at room temperature each. Non-specific binding was blocked through preincubation with 20% normal goat serum for 30 min at room temperature. The slides were subsequently incubated with the primary antibody for 1 h at room temperature and with secondary biotinylated goat-anti-rabbit (VECTASTAIN, 1/200 in PBS 1%/BSA) for 30 min, followed by labelling with the avidin-biotin complex (VECTASTAIN, Thermo Fisher, 1/100). The antibody complex was visualized using 8 min incubation with diaminobenzene (bright DAB, Immunologic, VWR, Dublin, Ireland). All slides were counterstained with haematoxylin (Klinipath, Olen, Belgium) for 5 s and mounted with a cover slip (Permount, Thermo-Fisher, Waltham, MA, USA).
For the cleaved caspase-3 staining, antigen retrieval was performed by heating in 10 mM citrate, pH 6.0, for 10 min at 96 • C in a PT Module (Thermo Fisher Scientific, Waltham, MA, USA). The peroxidase activity was blocked by incubating with 3% H 2 O 2 in PBS. Non-specific binding was blocked through preincubation with 20% normal goat serum for 30 min at room temperature. The slides were subsequently incubated overnight at 4 • C with the primary antibody and with secondary biotinylated goat-anti-rabbit (VEC-TASTAIN, Thermo Fisher, 1/200 in PBS 1%/BSA) for 30 min, followed by labelling with the avidin-biotin complex (VECTASTAIN, Thermo Fisher, 1/100). The antibody complex was visualized using 8 min incubation with bright DAB. All slides were counterstained with haematoxylin for 5 s and mounted with a Permount cover slip.

Automated Quantification Cleaved Caspase-3 IHC
Slide digitization was performed using a 3DHistech P1000 scanner digital slide scanner (3DHistech, Budapest, Hungary) with a 20× objective at a resolution of 0.24 µm/pixel. To quantify the amount of cleaved caspase-3 after FAP-tPDT, the total tumour area was manually annotated on tissue sections from two different depths in the tumour (approximately 150 µm apart). Tissue folds and staining artefacts were excluded from the annotation. A previously developed automated colour deconvolution algorithm [42] was used to extract the DAB staining from the background haematoxylin staining. This algorithm automatically determines the optimal stain matrix per slide, which is used for unmixing the slide for an optimal colour unmixing. To achieve this, an adapted version of the deconvolution of Ruifrok et al. was used [43]. We extended this algorithm by computing the ratio (positive pixels for a staining per region of interest) using automated Otsu thresholding on a resolution of 2 µm/pixel, which resulted in a thresholded image. The ratio was calculated on this thresholded image by dividing the stain of interest by the total amount of stained pixels.

Distribution Visualisation FAP IHC
To compare the autoradiography signal with the FAP IHC staining, we used the same algorithm to analyse the FAP staining. Instead of using the ratio as output, the thresholded image was used. To compare the FAP IHC staining with autoradiography, which are different in terms of resolution, we generated FAP IHC density maps, which mimic the resolution of the autoradiography (25 µm/pixel). A density map was calculated based on the thresholded binary image of the FAP staining. For this purpose, we defined a circular kernel of 0.49 mm 2 at a resolution of 32 µm/pixel and applied it to the slide in a sliding window fashion, computing the ratio of positive pixels over the circle area for every position in the slide. To limit the analysis to valid pixels in the tumour area within the slide, only positions within the tumour annotations were considered while sliding the circular kernel.

Statistics
All quantitative data are expressed as mean ± SD. The statistical analyses were performed using GraphPad Prism (Version 5.0, GraphPad Software Inc., San Diego, CA, USA). Student's t-test was used to determine significance.

DTPA-700DX-MB Binds to FAP-Expressing Cells and Causes Light-Induced Toxicity
The FAP-binding minibody was conjugated to DTPA-ITC and IRDye700DX-NHS using standardized protocols, reaching a ratio of 1.25 IRDye700DX molecules per minibody ( Figure 1A). Though FPLC showed an increase of 5.93% to 14.88% of aggregates upon conjugation (Supplementary Figure S1A,B), spectral properties of the IRDye700DX were retained upon conjugation ( Figure 1B). The construct was labelled with 111 In in 0.5 M MES buffer (pH 5.4) for 20 min, and successful labelling was determined with ITLC and FPLC (Supplementary Figure S1C). Membrane binding of DTPA-700DX-MB to 3T3-FAP cells was confirmed (3.4 ± 0.07, 6.7 ± 0.9, and 11.1 ± 0.13 percentage of added activity at 1, 2.5, and 24 h incubation, respectively), and the construct was efficiently internalized (6.8 ± 0.14, 11.8 ± 0.41, and 16.2 ± 0.98 percentage of added activity at 1, 2.5, and 24 h incubation, respectively) ( Figure 1C). Lack of binding or internalization in native 3T3 cells as well as in 3T3-FAP cells pre-incubated with an excess unlabelled minibody confirm FAP specificity. The half-maximal inhibitory concentration of DTPA-700DX-MB was 44 nM ( Figure 1D). The conjugate produced singlet oxygen upon irradiation with 690 nm light with similar efficiency as the non-conjugated NHS-IRDye700DX ( Figure 1E). Upon incubation with varying doses of DTPA-700DX-MB and subsequent irradiation with 60 J/cm 2 200 mW/cm 2 690 nm light, efficient cell death was induced ( Figure 1F). This effect was not observed when incubating native 3T3 cells or without irradiation, indicating receptor specificity and lack of dark toxicity of DTPA-700DX-MB ( Figure 1F). Since on some tumour cell lines expression of FAP is observed, we investigated the binding of DTPA-700DX-MB to PDAC299 cells, but no binding or light-induced cytotoxicity was observed (Supplementary Figure S2A,B). This indicates that these cells do not express FAP, which was also verified in co-cultures of PDAC299 and 3T3-FAP, in which only 3T3-FAP cells were killed upon treatment with DTPA-700DX-MB and light (Supplementary Figure S2C).

DTPA-700DX-MB Targets Subcutaneous PDAC299 Tumours In Vivo
PDAC299 is a tumour cell line that is derived from the pancreatic tumour of a Ptf1a WT/Cre ;Kras WT/LSL-G12D P53 LSL-R172H/fl mouse [40]. Upon subcutaneous injection into C57BL/6 mice, these tumours show a moderately well-differentiated morphology organized in glandular structures that are typical for adenocarcinomas (Figure 2A, HE). Furthermore, FAP-expressing fibroblasts were present (Figure 2A, FAP IHC), and excessive collagen formation was found in these tumours (Figure 2A, Sirius red). In a time-optimization biodistribution in mice carrying subcutaneous tumours of PDAC299, a maximum relative tumour uptake of 7.3 ± 1.02 %IA/g and a tumour-to-blood ratio of 8.3 ± 0.82 were observed at 24 h post injection of 0.3 nmol 111 In-labelled DTPA-700DX-MB ( Figure 2B,C and Supplementary  Table S1). Tumour uptake was significantly lower upon co-injection of a 7.4x excess of unlabelled and unconjugated minibody (4.4 ± 1.45 %IA/g, p = 0.0059), indicating at least a partly FAP-specific uptake ( Figure 2B and visualization of IRDye700DX-derived fluorescence in Figure 2D). The main clearance route of DTPA-700DX-MB is the liver, and some uptake was observed in the kidney and spleen. Furthermore, FAP-dependent uptake was observed in the tibia (9.85 ± 1.09 vs. 3.64 ± 0.80 %IA/g upon blocking, p < 0.0001), which is most likely due to presence of FAP-expressing cells in the bone marrow. Autoradiography of PDAC299 tumour sections showed heterogenous distribution of DTPA-700DX-MB in the tumour ( Figure 2E). To compare the autoradiography signal with the FAP IHC staining, which are different in terms of resolution, we generated FAP IHC density maps, which mimic the resolution of the autoradiography. In general, regions with a high signal on autoradiography also showed a higher signal on FAP IHC density maps visually, while in regions with a low signal on autoradiography, little FAP expression was observed on immunohistochemistry ( Figures 2E and S3).

DTPA-700DX-MB Induces Cell Death in Subcutaneous PDAC299 Tumours In Vivo
To determine in vivo efficacy, mice carrying two PDAC299 tumours were injected with 0.6 nmol DTPA-700DX-MB (n = 6) or PBS as a control (n = 5). One of the tumours was exposed to 690 nm light. Bleaching of IRDye700DX-derived fluorescence indicated efficient activation of the photosensitizer. A light dose of 55 J/cm 2 at a dose rate of 280 mW/cm 2 was not sufficient to induce apoptosis (n = 1, Figure S4). Irradiation with three separate doses of 55 J/cm 2 at 280 mW/cm 2 and two separate doses of 100 J/cm 2 at 230 mW/cm 2 light did induce apoptosis; however, in these conditions, we saw immediate bleeding of the skin and induction of crust formation, most probably due to excessive heating (both n = 1, Figure S4). A single dose of 100 J/cm 2 at 230 mW/cm 2 did not result in bleeding or crust formation but did induce apoptosis, as illustrated by the increase in caspase-3-positive fraction when compared to the contralateral non-irradiated tumours (n = 3, 0.41 ± 0.17 vs. 0.01 ± 0.01) (Figures 3A and S5). In the group injected with PBS, some upregulation of apoptosis was observed as well when comparing the caspase-3-positive fraction of the irradiated tumour with the non-irradiated tumour (n = 5, 0.23 ± 0.15 vs. 0.029 ± 0.012); however, this effect was less apparent than for the minibody-treated group (p = 0.0432 when comparing the MB and PBS light-exposed groups, Figures 3B and S6).

DTPA-700DX-MB Targets PDAC299 Orthotopic Tumours In Vivo
Maximum relative tumour uptakes of 8.97 ± 2.01 %IA/g and 9.28 ± 1.44 %IA/g were observed at 24 h post injection of 0.3 nmol 111 In-labelled DTPA-700DX-MB in mice injected with 5 × 10 3 and 2.5 × 10 4 tumour cells, respectively ( Figure 4A and Table S2). Tumour uptake was 7.89 ± 2.03 %IA/g upon co-injection of a 7-fold excess of unlabelled and unconjugated minibody (not significant). As in the mice carrying subcutaneous tumours, we observed uptake in the liver, kidney, and spleen. Furthermore, FAP-dependent uptake was observed in the tibia (6.61 ± 0.55 vs. 4.15 ± 0.12 %IA/g upon blocking, p = 0.0096). The signal in the orthotopic tumour could be visualized with SPECT/CT imaging ( Figure 4B, tumour area is encircled). Autoradiography of PDAC299 tumour sections showed heterogenous distribution of DTPA-700DX-MB. Similar to the subcutaneous tumours, comparable patterns were observed for the autoradiography and FAP-IHC density maps in the orthotopic tumours (Figures 4C and S7). were observed for the autoradiography and FAP-IHC density maps in the orthotopic tumours ( Figures 4C and S7).

Discussion
FAP-expressing CAFs are shown to have pro-tumourigenic effects in multiple solid tumour types. In this work, we characterize DTPA-700DX-MB, which can be used for light-induced ablation of FAP-expressing cells.
The anti-FAP minibody was efficiently functionalized with the photosensitizer IRDye700DX, which is one of the most suitable photosensitizers for targeted photody-

Discussion
FAP-expressing CAFs are shown to have pro-tumourigenic effects in multiple solid tumour types. In this work, we characterize DTPA-700DX-MB, which can be used for light-induced ablation of FAP-expressing cells.
The anti-FAP minibody was efficiently functionalized with the photosensitizer IRDye700DX, which is one of the most suitable photosensitizers for targeted photodynamic therapy due to its high hydrophilicity, photostability, and singlet oxygen quantum yield [22]. We demonstrated efficient and specific binding to murine FAP-transfected 3T3 fibroblasts and toxicity upon exposure to 690 nm light. In vivo, we showed targeting to both subcutaneous and orthotopic PDAC299 tumours. The uptake is at least partly specific, as shown by inhibition of accumulation after administration of an excess unlabelled minibody and colocalization of the minibody with FAP expression, as determined immunohistochemically. The limited molar excess of minibody (7-fold as compared to up to 100-fold used in other blocking studies) could lead to an underestimation of FAP-specific uptake. In the in vivo therapy experiment, we observed induction of apoptosis in the PDAC299 subcutaneous tumours, as illustrated by cleaved caspase-3 expression, in both the minibody and PBS-treated groups following light exposure. The upregulation in the minibody-treated group was, however, more pronounced, and part of the upregulation of caspase-3 expression may have been induced by excessive heating [44]. Surprisingly, we also observed a low auto-fluorescent signal in the tumours of PBS-injected mice, which was bleached upon irradiation. Though the origin of this fluorescent signal is unknown and has not been described in other studies, this may also be another contributor to the increase of caspase-3 staining that we observed in these tumours.
In these studies, we used a syngeneic mouse model PDAC299 using a murine cell line with PDAC typical KRAS and Trp53 mutations. We have shown that the subcutaneous PDAC299 tumours have an abundant stroma with collagen bundles containing FAP-expressing fibroblasts. As this model shows physiological overexpression of FAP on the stromal component, as is seen in human tumours as well, we believe that it better recapitulates the physiological situation as opposed to models that have been used in other studies, such as mixtures of cancer cells and fibroblasts and human tumour cell lines with either physiological or artificial FAP overexpression. Though the subcutaneous model is adequate to provide insight into pharmacokinetics and targeting properties of the DTPA-700DX-MB, we do believe that further studies into the biological effects of the FAP-targeted PDT should be investigated in orthotopic models because those better reflect the local environment in which the tumours develop.
Other groups have employed FAP-targeted PDT with a nanoparticle and an antibody [26,[30][31][32][33]. Zhen et al. showed efficient killing of fibroblasts in the murine breast cancer model 4T1, which resulted in increased CD8+ T-cell infiltration. Katsube et al. reported efficient killing of CAFs in a murine model that was co-inoculated with human squamous carcinoma cells (TE4) and activated human foetal oesophageal fibroblasts (FEF3), which improved the sensitivity to conventional chemotherapy. These results indicate the potential therapeutic efficacy of FAP-targeted photodynamic therapy in multiple tumour types. As mentioned above, further studies in an orthotopic PDAC299 model with varying light and protein doses are needed to determine the effect on immunomodulation and efficacy of other systemic therapies such as chemo-and immunotherapies.
Besides FAP-targeted PDT, radionuclide therapy with FAP-targeting small molecules has gained attention over the past years. FAP-targeting small molecules labelled with actinium-226 or lutetium-177 showed modest growth inhibition in preclinical studies, and initial clinical studies are emerging [45][46][47]. Though systemic toxicities are a potential problem of radionuclide therapies, an advantage is that the therapeutic effect occurs in all lesions in the body, and it is thus suitable for application in metastasized disease. PDT can only be used for tumours that are accessible to light that has a limited penetration depth (up to 1 cm for NIR light). We therefore envision clinical application of FAP-tPDT to be applied intraoperatively or through multiple laser fibres for larger and more deeper-seated lesions of localized disease. Furthermore, we expect this therapy to be applied in an adjuvant manner, sensitizing the tumours for other systemic therapies such as chemotherapy and immunotherapy.

Conclusions
In conclusion, we have shown that the DTPA-700DX-MB conjugate efficiently binds to FAP-expressing cells and induces cell death upon light exposure in vitro. Furthermore, it targets FAP-expressing fibroblasts in PDAC299 subcutaneous and orthotopic tumours and induces apoptosis in the subcutaneous tumours upon light exposure. This study provides proof-of-concept of FAP-targeted PDT in PDAC, and future studies should demonstrate the feasibility in other tumour types and the effects on tumour growth, immunomodulation, and efficacy of other systemic therapies.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/cells12101420/s1, Figure S1: FPLC data of the native minibody, DTPA-MB-700DX and 111 In-labeled DTPA-MB-700DX; Figure S2: Lack of association of DTPA-700DX-MB with PDAC299 cells in vitro, Figure S3: Autoradiography of subcutaneous PDAC299 tumour sections, Figure S4: In vivo efficacy of targeted photodynamic therapy with DTPA-700DX-MB, Figure S5: In vivo efficacy of targeted photodynamic therapy with DTPA-700DX-MB, Figure S6: PBS control group for FAP-targeted photodynamic therapy in vivo, Figure S7: Autoradiography of orthotopic PDAC299 tumour sections, Table S1: Uptake in various tissues in the subcutaneous model, Table S2: Uptake in various tissues in the orthotopic model.  Institutional Review Board Statement: The animal study protocol was approved by the Institutional Review Board (or Ethics Committee) of The Radboud University animal ethics committee (CCD number AVD103002015209 and AVD1030020209645).

Data Availability Statement:
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.