SPECT/CT Imaging, Biodistribution and Radiation Dosimetry of a 177Lu-DOTA-Integrin αvβ6 Cystine Knot Peptide in a Pancreatic Cancer Xenograft Model

Introduction Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive malignant neoplasms, as many cases go undetected until they reach an advanced stage. Integrin αvβ6 is a cell surface receptor overexpressed in PDAC. Consequently, it may serve as a target for the development of probes for imaging diagnosis and radioligand therapy. Engineered cystine knottin peptides specific for integrin αvβ6 have recently been developed showing high affinity and stability. This study aimed to evaluate an integrin αvβ6-specific knottin molecular probe containing the therapeutic radionuclide 177Lu for targeting of PDAC. Methods The expression of integrin αvβ6 in PDAC cell lines BxPC-3 and Capan-2 was analyzed using RT-qPCR and immunofluorescence. In vitro competition and saturation radioligand binding assays were performed to calculate the binding affinity of the DOTA-coupled tracer loaded with and without lutetium to BxPC-3 and Capan-2 cell lines as well as the maximum number of binding sites in these cell lines. To evaluate tracer accumulation in the tumor and organs, SPECT/CT, biodistribution and dosimetry projections were carried out using a Capan-2 xenograft tumor mouse model. Results RT-qPCR and immunofluorescence results showed high expression of integrin αvβ6 in BxPC-3 and Capan-2 cells. A competition binding assay revealed high affinity of the tracer with IC50 values of 1.69 nM and 9.46 nM for BxPC-3 and Capan-2, respectively. SPECT/CT and biodistribution analysis of the conjugate 177Lu-DOTA-integrin αvβ6 knottin demonstrated accumulation in Capan-2 xenograft tumors (3.13 ± 0.63%IA/g at day 1 post injection) with kidney uptake at 19.2 ± 2.5 %IA/g, declining much more rapidly than in tumors. Conclusion 177Lu-DOTA-integrin αvβ6 knottin was found to be a high-affinity tracer for PDAC tumors with considerable tumor accumulation and moderate, rapidly declining kidney uptake. These promising results warrant a preclinical treatment study to establish therapeutic efficacy.

Introduction: Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive malignant neoplasms, as many cases go undetected until they reach an advanced stage. Integrin avb6 is a cell surface receptor overexpressed in PDAC. Consequently, it may serve as a target for the development of probes for imaging diagnosis and radioligand therapy. Engineered cystine knottin peptides specific for integrin avb6 have recently been developed showing high affinity and stability. This study aimed to evaluate an integrin avb6-specific knottin molecular probe containing the therapeutic radionuclide

INTRODUCTION
Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive malignant neoplasms and accounts for 80-90% of all pancreatic cancer cases. Although the incidence of PDAC is low, in cancer-related deaths, it ranks seventh globally (1). Owing to its poor prognosis, the 5-year survival rate is 9% with merely 24% of patients surviving for a year (2). This is mainly because PDAC patients rarely exhibit symptoms before an advanced stage of the disease has been reached, and due to the lack of appropriate diagnostic and therapeutic options. Surgical resection of the tumor, part of the pancreas, and other nearby digestive tract organs remains to be the only curative treatment for early-stage PDAC patients. Gemcitabine has been used for several years as a baseline chemotherapeutic treatment. Lately, combination therapy of gemcitabine with folfirinox and nab-paclitaxel demonstrated improved results in comparison to the use of gemcitabine alone (3,4). However, high resistance of PDAC to chemotherapy dilutes its efficacy (5,6).
Integrins are heterodimeric transmembrane cell surface proteins that mediate cell-to-cell and cell-to-extracellular matrix (ECM) adhesion (7,8). Many integrins, including integrin avb6, were reported to be upregulated in various cancers such as breast cancer, gastric cancer, colorectal cancer, lung cancer, ovarian cancer and PDAC (9)(10)(11)(12)(13)(14). In welldifferentiated PDACs, integrin avb6 overexpression was identified in 100% of the samples (13,14). Moreover, integrin avb6 overexpression has been recognized as a prognostic marker for reduced survival in non-small cell lung cancer (15), gastric carcinoma (16), colorectal cancer (17) cervical squamous cell carcinoma (18) and PDAC (19). Remarkably, integrin avb6 expression was found to be higher in PDAC than in chronic pancreatitis (20). These findings support the utilization of integrin avb6 as a target for the development of new diagnostic and therapeutic tools.
Cystine knot peptides (knottins) represent small peptides of approximately 4 kDa with three threaded disulfide bonds that form a rigid topological knot constraining the peptide's conformational flexibility. Such a structural motif is known as a cystine knot (21,22). One of the advantages of knottins is the high variability of backbone residues that may be used to modulate tumor and kidney uptake (23). Knottins are well suited for in vivo tumor-targeting applications as their disulfide-bonded core confers outstanding thermal stability and resistance against proteolytic degradation; moreover, they have been shown to be nonimmunogenic. Furthermore, knottins may be chemically modified to tailor their in vivo pharmacokinetic properties for a variety of clinical applications. Previously, we have developed the optimized knottin R 0 1-MG that shows low single-digit nanomolar binding affinity for integrin avb6 (23,24). Recently, the first clinical study with R 0 1-MG demonstrated clinical potential for targeting PDAC in humans (24).
In the current imaging, biodistribution and dosimetry study, a lutetium-177 DOTA conjugate of this knottin was evaluated as a candidate for therapeutic purposes in a PDAC xenograft model. Binding affinity of the 177 Lu tracer was found to be in the low single-digit nanomolar range. SPECT/CT imaging and biodistribution of the tracer 177 Lu-DOTA-integrin avb6 knottin revealed substantial accumulation of the tracer in the tumor as well as faster renal clearance. Our studies demonstrate the potential of the 177 Lu-DOTA integrin avb6 knottin as a therapeutics candidate in PDAC.

Immunofluorescence
For immunofluorescent staining of tumor tissue, cells grown on glass coverslips were fixed with 1:1 methanol/acetone for two minutes and air-dried. After washing with PBS (Biochrom, Berlin, Germany) and blocking with 5% goat serum in PBS for 30 minutes, coverslips were incubated with a rabbit IgG against human integrin b6 (#HPA023626, Atlas Antibodies, Bromma, Sweden) diluted in 0.1% BSA in PBS in a wet chamber for one hour at room temperature. After washing, coverslips were incubated with the secondary antibody goat-anti-rabbit-Cy3 (Jackson ImmunoResearch, West Grove, USA; 2.5 μg/mL diluted in 0.1% BSA in PBS) for 30 minutes. After washing with PBS, nuclei were stained with 1 μg/mL TOTO-3 (Invitrogen) in PBS for 5 minutes. Finally, the cells were fixed with 96% ethanol for two minutes, embedded with Immu-Mount (Thermo Fisher Scientific, Waltham, USA) and analyzed using a confocal laser-scanning microscope (LSM510, Carl Zeiss, Jena, Germany).

Non-Radioactive Metalation With nat Lu
Non-radioactive complexes of DOTA-integrin avb6 knottin with natural lutetium were generated by incubation of 0.9 nmol (final concentration of 30 μM) peptide conjugate dissolved in 5 μl buffer (sodium acetate/acetic acid buffer, 0.5 M, pH 5.4) with 27 nmol (a 30-fold molar excess) of the Lu 3+ ion ( nat Lu-DOTA-knottin) or no metal ion (control DOTA-knottin). The reaction volume was made up to 30 μl with water. The reaction was carried out for 10 minutes at 80°C. HPLC showed >98% purity of the nat Lu-labeled peptide.

Saturation Binding Assay
To determine the dissociation constant (K d ) and the maximum number of binding sites (B max ) of the radioligand, saturation binding assay was performed on BxPC-3 and Capan-2 cell lines. For this, approximately 50,000 cells per well were seeded in a 96 well plate and incubated overnight at 37°C. Cells were then incubated with binding buffer and varying concentration (0, 0.01, 0.05, 0.1, 0.2, 0.5, 1, 2, 5, 10, 15, 20 nM) of 125 I-DOTAintegrin avb6 knottin either with (non-specific binding) or without 1 μM of DOTA-integrin avb6 knottin (total binding). Non-specific binding was subtracted from total binding to obtain specific binding. All three datasets (non-specific binding, specific binding and total binding) were plotted with GraphPad Prism and fitted using nonlinear regression (one site -total and nonspecific, one site -specific binding). The software provides Bmax in the same value as the respective y-axis, in this case cpm. The following calculations were performed to obtain B max in receptor sites per cell. In the first step, the specific activity of 125 I (2175 Ci/ mmol) was transformed into dpm/mmol by multiplication with 2.22*10 12 . This was multiplied with the counter efficiency of 50% to get cpm/mmol and subsequently converted to cpm/fmol: In a second step, the B max value, calculated by the software in cpm, was divided by cpm/fmol to derive the amount of substance in fmol. This was multiplied with the Avogadro constant (6.02*10 8 /fmol) to obtain the number of molecules before division by seeded cells (50,000 in this case): sites=cell = ½Bmax(cpm) * 6:02 * 10 8 fmol −1 =cpm=fmol * 50, 000

Cell Irradiation
Cells were irradiated using a GSR D1 gamma irradiator (Gamma-Service Medical GmbH, Leipzig, Germany) with a cesium-137 source. The cells with the respective culture medium were placed inside the radiation chamber and required dose was achieved by adjusting the tray level and duration of the exposure. After irradiation, the cell lines were incubated at 37°C without medium change for the appropriate duration.

Cell Count (DAPI Staining)
Cells were seeded at a density of 5,000 cells per well in a 96 well plate and incubated overnight at 37°C. After irradiating with 0, 0.2, 2, 4, 6, 8, 10, 15 or 30 Gy, cells were incubated at 37°C for 96 hours. Cells were then washed with PBS and fixed with 4% PFA for 10 min at room temperature. Thereafter, cells were stained with DAPI (1:5,000 in PBS/0.1% Tx-100) and incubated at room temperature for 10 min. Before image acquisition, wells were filled with 80 μL of PBS and images were acquired using an automated microscope (IN Cell Analyzer 1000; GE, Reading, UK) with a 4x objective. Image stacks were analyzed and nuclei were counted by the IN Cell software.

Clonogenic Assay
Cells were seeded at a density of 5,000 cells per well in a 12 well plate and incubated overnight at 37°C. After irradiation with 0, 2, 4, 6 or 8 Gy, cells were incubated at 37°C without medium change for 7 days. Cells were then washed with PBS and fixed with 70% ethanol for 10 min, stained with 0.2% crystal violet solution for another 10 min and carefully rinsed with tap water. Plates were dried overnight and digitized using an Odyssey infrared scanner (700 nm channel, intensity 3, 84 mm resolution and medium quality). For quantification, images were analyzed using the Colony Area plugin for ImageJ.

Xenografts
For in vivo experiments, at least 8-week-old female athymic NMRI-Foxn1 nu /Foxn1 nu mice (Janvier Labs, Saint-Berthevin, France) were used. Animal care followed institutional guidelines and all experiments were approved by local animal research authorities. For the generation of tumor xenografts, 5 x 10 6 cells of Capan-2 cells were inoculated subcutaneously into the left and right shoulder (1:1 phosphate-buffered saline [PBS]/ Matrigel Basement Membrane Matrix High Concentration, Corning, Corning, USA). Tumors were allowed to grow for two to four weeks (tumor volume > 100 mm 3

Radiochemical Labeling With Lutetium-177 ( 177 Lu)
Radiolabeling of the DOTA-integrin avb6 knottin was carried out manually using [ 177 Lu]LuCl 3 and a reagent kit from ITG Isotope Technologies Garching GmbH (Garching, Germany). A total of 40 μL (stock solution of peptide at 1 μg/μL in 10% DMSO in water) was added to 500 μL of ascorbate buffer solution pH 5.0, prepared from the kit (composition not disclosed by vendor) and the resulting mixture was then added to 35 μL of [ 177 Lu]LuCl 3 (2 GBq in aqueous 0.04 M HCl) and the reaction mixture was heated at 80°C for 35 minutes followed by cooling for 10 minutes at room temperature. The product was then diluted with 0.5 mL of saline and the pH was adjusted to 7 using a 1 M aqueous solution of NaHCO 3 prior to injection in mice. A reverse-phase HPLC system (Knauer GmbH, Germany) with a column (Eurospher II, C18, 250 × 4 mm) was used to determine the radiochemical purity of the radiotracer. The HPLC system was equipped with an Azura P.6.1L pump coupled with ultraviolet (Azura UVD 2.1L) and radiometric (g-Raytest-Isotopenmessgeraete GmbH, Germany) detectors. The gradient elution system used mobile phase A (100% acetonitrile) and mobile phase B (deionized H 2 O containing 0.1% trifluoroacetic acid) and a flow rate of 1.0 mL/min. Starting with 0% A and 100% B, the gradient was increased to 100% A over 25 min and finally returned to initial gradient conditions within 5 min. The retention time of the tracer was found to be 10.37 min and the peak of free Lu-177 was observed at 2.64 min. Radiochemical purity determined by HPLC was found to be higher than 97% (Supplementary Figure S1) in all cases and the product had a molar activity of approximately 44 GBq/μmol.

SPECT/CT Imaging
SPECT and CT imaging were performed using the nanoSPECT/ CTplus scanner (Mediso, Hungary/Bioscan, France). Mice were anesthetized using 1-2% isoflurane with oxygen at a flow rate of approximately 0.5 l/min. After a low-dose CT scan for positioning the tumors and kidneys in the scan range the SPECT acquisition was started directly before intravenous injection of approximately 50 MBq of 177 Lu-DOTAintegrin avb6 knottin (0.1-0.15 ml).
Nine consecutive multi-pinhole SPECT images of 10 min duration each (5 angular steps a 60 sec, 2 bed positions) were acquired. Additional individual scans of 35-60 min duration (5 angular steps á60-180 sec, 2 bed positions) were performed up to 8 days to assess biodistribution kinetics.

Biodistribution Studies
Tumor-bearing mice (n=4-5 per time point) were injected with approximately 1 MBq (22 pmol) of 177 Lu-DOTAintegrin avb6 knottin to the tail vein via a catheter. Mice (n=4-5 per time point) were sacrificed and dissected one, two, three and eight days post injection. Tumor, blood, stomach, pancreas, small intestine, colon, liver, spleen, kidney, heart, lung, muscle and femur samples were weighed and uptake of radioactivity was measured by a gamma counter.

Dosimetry of Tissue and Tumor
Assuming that the organ-to-whole-body activity concentration ratio in mice would equal that in humans, the injected activity concentration (%IA/g) acquired from the mouse biodistribution study was transformed to human whole-organ percentage of injected activity per gram of tissue (%IA/g) human. The mouse uptake data were extrapolated to humans by relative scaling of mass and time using the following two equations, (1) t human = t animal ½M human =M animal 0:25 (2) where %IA/g is percentage of injected activity per gram of tissue, %IA/organ is percentage of injected activity per organ; m is organ mass of mouse or human, M is body mass of mouse or human; t is time. The absorbed doses per activity were calculated by using the extrapolated human source organ residence times as input in the OLINDA/EXM 1.

Target Expression and Tracer Affinity
As integrin b6 forms heterodimers only with integrin av, detection of the b6 subunit mRNA or protein will be informative about the dimer, too. To identify the optimal animal model for imaging, integrin b6 mRNA was measured by RT-qPCR in human PDAC cell lines, corresponding xenografts, A549 lung adenocarcinoma and HT29 colorectal adenocarcinoma cells as well as mouse kidney and liver. Capan-2 and BxPC-3 cells and xenografts showed the highest abundance of integrin b6 mRNA ( Figure 1A). Mouse kidney and liver showed around 1,000-fold less integrin b6 mRNA, indicating a significantly lower expression of the corresponding mouse receptor in these organs. Immunofluorescence staining detected integrin b6 on the plasma membrane of Capan-2 cells ( Figure 1B).
To determine the binding affinity of the DOTAintegrin avb6 knottin to its target, saturation and competition binding assays on BxPC-3 and Capan-2 cells were performed ( Figures 1C, D and Table 1). Saturation binding assays were performed to determine dissociation constant (K d ) and maximum number of binding sites per cell (B max ). K d values for BxPC-3 and Capan-2 cells were found to be 0.30 nM and 0.75 nM, respectively. Correspondingly, the B max values for BxPC-3 and Capan-2 were shown to be approximately 11,800 and 11,500 binding sites/cell. 125 I-labeled DOTAintegrin avb6 knottin showed binding to both cell lines, BxPC-3 and Capan-2, displaced by the unlabeled peptide in a concentration-dependent manner. The inhibitory constant (K i ) values of DOTA-integrin avb6 knottin for BxPC-3 and Capan-2 were calculated to be 1.69 nM and 9.46 nM respectively. To determine the effect of lutetium chelation on the probe's affinity for integrin avb6, binding of nat Lu-DOTA-knottin and control-DOTA-knottin (without nat Lu) on BxPC-3 and Capan-2 was examined. The complexation of the nat Lu ion did not compromise the high affinity of the tracer ( Table 1).

Radiosensitivity of BxPC-3 and Capan-2 Cells
To evaluate the suitability of BxPC-3 and Capan-2 cells as model cell lines in terms of their radiosensitivity, two approaches were taken. Both cell lines were irradiated with different doses (0, 0.05, 0.2, 2, 4, 6, 8, 10, 15 and 30 Gy) of gamma radiation from a 137 Cs source. After 96 hours of incubation, nuclei were stained and counted. The IC 50 values for the resulting growth inhibition/cell death in BxPC-3 and Capan-2 cells were found to be 4.3 Gy and 5.5 Gy, respectively (Figures 2A, B). To evaluate radiation effects on colony formation, both cell lines were treated with different doses (0, 2, 4, 6 and 8 Gy) of radiation. After 7-8 days of incubation, colonies were fixed, stained and counted ( Figures 2C, D). BxPC-3 cells (IC 50 1.3 Gy) appeared to be slightly more radiation-sensitive compared to Capan-2 (IC 50 2.2 Gy) in this assay.

SPECT Imaging in a Capan-2 Xenograft Mouse Model of Pancreatic Cancer
While the integrin avb6 knottin and its DOTA conjugate previously had been used for tumor imaging studies employing 18 F, 64 Cu or 99m Tc, no data regarding the biodistribution of the 177 Lu-DOTA-integrin avb6 knottin were available. To obtain such data, 177 Lu-DOTA-integrin avb6 knottin was synthesized, yielding a product with 98% radiochemical purity and a molar activity of approximately 44 GBq/μmol. The tracer was injected intravenously in mice bearing Capan-2 xenografts on both shoulders. SPECT/CT images were taken at different times post injection. Figure 3A shows SPECT/CT images taken at 22 hours p.i. The maximum intensity projection (MIP) reveals an accumulation of 177 Lu-DOTA-integrin avb6 knottin in both the tumors and the kidneys. Uptake by other organs was low or moderate. Short-term and long-term tracer kinetics for kidney and tumor were quantified from SPECT images, up to 3 and 187 hours post-injection, respectively. Indeed, kidneys showed a higher initial accumulation of 177 Lu-DOTA-integrin avb6  knottin ( Figure 3B), yet clearance was also faster than from the tumor ( Figure 3C).

Biodistribution Analysis of Organ Uptake
Ex vivo biodistribution analysis of mice bearing Capan-2 xenografts demonstrated a tumor uptake of the 177 Lu-DOTA-integrin avb6 knottin of 3.1 ± 0.6, 2.5 ± 0.4, 3.5 ± 0.9 and 1.2 ± 0.2%IA/g (mean ± S.E.M.) on day 1, 2, 3 and 8, respectively ( Figure 4A). Nevertheless, tracer uptake by the kidney of 19.2 ± 2.5, 12.5 ± 0.6, 14.7 ± 4.5 and 2.3 ± 0.4%IA/g was detected on days 1, 2, 3 and 8 p.i., respectively. On the other hand, low or moderate activity was discovered in organs like stomach, colon and lung at later time points ( Table 2). The time-activity curve from ex vivo biodistribution data ( Figure 4B) confirms the analysis of SPECT kinetics ( Figure 3C) with activity from kidneys being washed out more rapidly than from tumors. Based on ex vivo biodistribution inputs, tumor-to-organ ratios were calculated (Table 2 and Figure 5). Except for kidney, these ratios were favorable in pancreas, liver, blood, lung and muscle. Ex-vivo analysis of H&E-stained xenografts did not reveal significant differences in tissue structure between samples obtained on day 1 or day 8 after injection of the tracer (Supplementary Figure S2).

Dosimetric Calculation
Dosimetric calculations for the human male adult for 177 Lu ware generated ( Table 3). The expected absorbed doses per injected activity in humans were calculated using the mouse biodistribution data. The two interspecies scaling (mass and time) model (25,26) was implemented for extrapolating animal to human data. The calculated expected effective dose was 0.04 mSv/MBq. Additionally, the calculated absorbed doses for the common organs are indicated. As expected from the mouse biodistribution data, kidney showed the highest absorbed dose of 0.02 mSv/MBq followed by lung and stomach of 0.01 and 0.005 mSv/MBq respectively ( Table 3).

DISCUSSION
Due to its high and selective tumor overexpression, integrin avb6 is emerging as a target in cancer for nuclear imaging.  Data are presented as mean ± SD %IA/g of tissue (4 ≤ n ≤ 5).
Several tumor-targeting strategies based on integrin avb6 have been developed for either diagnostic or therapeutic purposes (15,(27)(28)(29)(30)(31). To that end, ligands had been conjugated either with a radionuclide ( 18 F, 64 Cu, 111 In or 177 Lu) or an anti-cancer drug (e.g. tesirine) (31). The majority of approaches involved the use of linear peptides derived from foot-and-mouth disease virus, which show a high affinity to the receptor. However, due to reduced stability and specificity for the receptor, in vivo results were ambiguous. In a recent study, a cyclic radiotracer specific for integrin avb6 ( 68 Ga-cycratide) was used in a pancreatic mouse model for PET imaging (32). The DOTA-integrin avb6 knottin applied here has recently been used in a first-in-human clinical study and has demonstrated a high potential for PDAC targeting (24). In this study, the same DOTA-conjugated engineered cystine knot peptide (knottin) specific for integrin avb6 was chosen. The knottin peptide's stability and affinity to the receptor were not compromised in vivo and it had previously exhibited favorable tumor uptake (21,23). Previously, 64 Cu, 18 F and 99m Tc have been investigated in conjugation with this knottin. However, no data were available on the pharmacology of the tracer coupled to a therapeutic radionuclide such as 177 Lu. This study was designed to examine the in vitro pharmacology, biodistribution and dosimetry of a 177 Lu-labeled DOTA-knottin in a xenograft mouse model, to pave the way for a potential clinical translation of this compound. Likewise, fluorescencelabeled knottins specific for integrin avb6 had shown promising preclinical results, which could be further translated for the early detection of PDAC in patients (33,34). Such an optical imaging agent could also play a role in fluorescenceguided surgery for PDAC patients. Indeed, the specificity of the agent would assist in discerning PDAC from pancreatitis and normal pancreatic tissue.
In the current study, integrin avb6 mRNA expression in BxPC-3 and Capan-2 was found to be highest among all tested PDAC and other cell lines. Additionally, sustained expression of integrin avb6 in mouse xenografts of these cell lines was confirmed. More importantly, mRNA expression of integrin avb6 in mouse liver and kidney was found to be very low compared to BxPC-3 and Capan-2 cell lines. This, however, does not rule out expression in specific anatomical substructures of these two organs of excretion. Immunofluorescence  experiments established the expression of integrin avb6 on the surface of Capan-2 cells. This is of relevance as RT-qPCR data will only be informative about mRNA levels, not protein. In addition, many cell membrane receptors occur in equilibrium of distribution between plasma membrane and intracellular compartments, e.g. the trans-Golgi network or endosomes. Proof of a high degree of surface expression may therefore be a meaningful predictor of in vitro and in vivo tracer binding. Along with mRNA and protein expression data, the presence of a substantial number of functional receptors on the cell surface is an essential benchmark for nuclear imaging. In agreement with previous findings, the affinity of the DOTA-integrin avb6 knottin for BxPC-3 and Capan-2 cells was found to be in the low nanomolar range (1.69 and 9.46 nM) as revealed by radioligand binding assay. It is interesting to note that the affinity of the knottin did not change upon either the addition of a DOTA chelator moiety or incorporation of natural non-radioactive lutetium ( nat Lu) into the chelator.
Compared to previous biodistribution studies ( 18 F, 99m Tc) using the same knottin (35,36), a higher uptake of the 177 Lu tracer in tumors was observed here. Some organs like lung, stomach, colon and kidney showed tracer uptake at early time points yet this was washed away at later time points. As integrin avb6 expression in mouse kidney had been found several orders of magnitude lower than in xenograft tumors (Fig. 1A), this uptake in the renal medulla may be attributed to unspecific uptake by transporters, e.g. OATPs. A high amount of tracer was accumulated in the kidney at early time points (day 1: 19.2 ± 2.5%IA/g); there was a significant reduction at later time points (day 8: 2.3 ± 0.40%IA/g). However, in the tumor the activity was cleared at a much lower rate than in the kidney. Higher tracer activity in the kidney resulted in a lower tumor-to-kidney ratio at day-1, however, Capan-2 tumors were clearly recognized. Similarly, a higher tumor-to-organ ratio at all-time points gives an advantage for contrast and favorable tumor imaging. These findings again underline that the tracer binds in the tumor specifically and with greater affinity. However, tumor activity is still limiting the full impact of this targeting approach. In addition to characterizing the principal tracer pharmacology with the therapeutic radionuclide 177 Lu, this study was also designed to see whether tumor uptake would be higher with this nuclide than with the previously tested ones. As 177 Lu has a comparatively long half-life and potentially allows for a smaller peptide/radiometal ratio, the resulting higher molar activity was hoped to lead to higher tumor accumulation. Indeed, a molar activity of 44 GB/μmol was achieved, which was 2.4-fold higher than the one reported for the 64 Cu conjugate before (19). Unfortunately, tumor uptake was not found to be higher than in this study. A further improved initial tracer uptake in the tumors should be the goal for the continued path towards translation into the clinic. With affinities already in a very favorable range, overall design of the conjugate could be modulated, e.g. by the introduction of a different spacer. In addition, increasing the molar activity of the tracer formulation may result in increased tumor uptake.
For dosimetry, biodistribution data were extrapolated to human adult male using OLINDA/Exm software for the therapeutic radionuclide 177 Lu. This dosimetric projection provided a preliminary estimate for assessing the therapyassociated risk of radiation damage. The absorbed doses in the kidney propose it as the dose-limiting organ. As this method involves two scaling methods to extrapolate human data, the resulting prediction is of preliminary nature and needs further confirmation, e.g. by collecting corresponding data from human subjects.
In the framework of this study, only a single dose of tracer was injected into the mice, and the animals were monitored for no longer than eight days. Consequently, no significant impact was observed in H&E-stained tissue sections of day 1 and day 8 post injections. A therapy study, potentially involving multiple/ repeated administrations of tracer is required to establish therapeutic efficacy. Still, the uptake of tracer by xenograft tumors at different time points and faster renal clearance confirms the suitability of the DOTA-integrin avb6 knottin as a tracer and of integrin avb6 as a valid target for diagnosis and potential targeted radionuclide therapy in PDAC.

CONCLUSION
In this study, binding of the peptide tracer 177 Lu-DOTAintegrin avb6 knottin to its target both in vitro and in vivo was investigated. 177 Lu-DOTA-integrin avb6 knottin exhibited high affinity and specific binding to target-positive cells and tumors. The study demonstrated the translational potential of this tracer for imaging and therapy of integrin avb6overexpressing tumors like PDAC.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material.

ETHICS STATEMENT
The animal study was reviewed and approved by Landesamt für Gesundheit und Soziales, Land Berlin.

AUTHOR CONTRIBUTIONS
Concept and experimental design: BW, SG, RK, SR, and CG. Development of methodology: RK and NB. Acquisition of data: SS, TH, SE, SP, and NB. Analysis and interpretation of data: SS and CG. Writing, review, and/or revision of the manuscript: SS, CG and all other authors. Study supervision: BW, SG, and CG. All authors contributed to the article and approved the submitted version.

FUNDING
This study was supported by a grant from the Will Foundation.