Toward the Clinical Development and Validation of a Thy1-Targeted Ultrasound Contrast Agent for the Early Detection of Pancreatic Ductal Adenocarcinoma

Abstract

Early detection of pancreatic ductal adenocarcinoma (PDAC) represents the most significant step toward the treatment of this aggressive lethal disease. Previously, we engineered a preclinical Thy1-targeted microbubble (MBThy1) contrast agent that specifically recognizes Thy1 antigen overexpressed in the vasculature of murine PDAC tissues by ultrasound (US) imaging. In this study, we adopted a single-chain variable fragment (scFv) site-specific bioconjugation approach to construct clinically translatable MBThy1-scFv and test for its efficacy in vivo in murine PDAC imaging, and functionally evaluated the binding specificity of scFv ligand to human Thy1 in patient PDAC tissues ex vivo.

Materials and methods: We recombinantly expressed the Thy1-scFv with a carboxy-terminus cysteine residue to facilitate its thioether conjugation to the PEGylated MBs presenting with maleimide functional groups. After the scFv-MB conjugations, we tested binding activity of the MBThy1-scFv to MS1 cells overexpressing human Thy1 (MS1Thy1) under liquid shear stress conditions in vitro using a flow chamber setup at 0.6 mL/min flow rate, corresponding to a wall shear stress rate of 100 seconds, similar to that in tumor capillaries. For in vivo Thy1 US molecular imaging, MBThy1-scFv was tested in the transgenic mouse model (C57BL/6J - Pdx1-Cre; KRas; Ink4a/Arf) of PDAC and in control mice (C57BL/6J) with L-arginine-induced pancreatitis or normal pancreas. To facilitate its clinical feasibility, we further produced Thy1-scFv without the bacterial fusion tags and confirmed its recognition of human Thy1 in cell lines by flow cytometry and in patient PDAC frozen tissue sections of different clinical grades by immunofluorescence staining.

Results: Under shear stress flow conditions in vitro, MBThy1-scFv bound to MS1Thy1 cells at significantly higher numbers (3.0 ± 0.8 MB/cell; P < 0.01) compared with MBNontargeted (0.5 ± 0.5 MB/cell). In vivo, MBThy1-scFv (5.3 ± 1.9 arbitrary units [a.u.]) but not the MBNontargeted (1.2 ± 1.0 a.u.) produced high US molecular imaging signal (4.4-fold vs MBNontargeted; n = 8; P < 0.01) in the transgenic mice with spontaneous PDAC tumors (2-6 mm). Imaging signal from mice with L-arginine-induced pancreatitis (n = 8) or normal pancreas (n = 3) were not significantly different between the two MB constructs and were significantly lower than PDAC Thy1 molecular signal. Clinical-grade scFv conjugated to Alexa Fluor 647 dye recognized MS1Thy1 cells but not the parental wild-type cells as evaluated by flow cytometry. More importantly, scFv showed highly specific binding to VEGFR2-positive vasculature and fibroblast-like stromal components surrounding the ducts of human PDAC tissues as evaluated by confocal microscopy.

Conclusions: Our findings summarize the development and validation of a clinically relevant Thy1-targeted US contrast agent for the early detection of human PDAC by US molecular imaging.

Figure 1.. Characterization of the Thy1-specific scFv-(Gly)₅-Cys.

(A) Flow-chart showing study design and pre-clinical validation work-up of Thy1-targeted microbubble (MB) contrast agent (MB_Thy1-scFv). Dotted lines represent validation experiments with only the scFv protein. (B) SDS-PAGE comparison of recombinantly produced scFv (with TrxA fusion protein) before and after the addition of C-terminal Cysteine (Cys) linked by a penta-Glycine linker (Gly₅). A major band for the monomer form (~44 kDa) and a minor band for the dimer form (~89 kDa) are observed for scFv-(Gly)₅-Cys in the non-reducing SDS-PAGE. (C) Mass Spectroscopy of the recombinantly expressed scFv-(Gly)₅-Cys (with N-terminal TrxA and His₆ fusion proteins as shown in the expression plasmid map) show a prominent singly charged species of scFv monomer (m/z = 44,328) and dimer (m/z = 89,105). Full sequence of the Thy1-scFv is provided with a heavy chain (blue) and a light chain (brown) separated by (G₄S)₃ linker sequence (black). (D) Flow cytometry based binding detection of the Dynabeads-immobilized 100 nM scFv-(Gly)₅-Cys to the 66 pmol soluble human (h) or murine (m) Thy1 recombinant protein as detected by anti-Thy1-APC antibody. Naked beads without scFv was used as a negative control sample.

Figure 2.. Characterization of MBs.

(A) A general schematic representation of the MBs with its functional components. A phospholipid shell composed of DPPC, DPPE and DSPE lipids surrounds a gas core consisting of octafluoropropane (OFP). A PEG chain (PEG5000) covalently attached to phospholipids enhances stability of the MBs while it also serves as a linker to bear an active Maleimide (MA) functional group for targeting by bioconjugation with proteins or neutral group (MPEG) in control MBs. (B) Schematic representation of thiol-MA coupling strategy to form the targeted MBs (MB_Thy1-scFv). C-terminal cysteine residue of the Thy1-specific scFv-(Gly)₅-Cys is reduced for stable covalent bonding to the MA group in the DSPE-PEG(5000)-MA components of the targeting MB shell. (C) Bright-field and fluorescence images of MB_Thy1-scFv confirming the surface conjugation of scFv-(Gly)₅-Cys onto the MB shell. A composite image shows biotinylated scFv-(Gly)₅-Cys (red color; detected by streptavidin-Alexa Fluor 647) signal overlapping the DiO-labeled shell of targeting MBs (green color).

Figure 3.. MB_Thy1-scFv specifically binds to Thy1 protein in vitro.

(A) Binding comparison of targeted MBs [MB_Thy1-scFv; anti-Thy1-antibody (Ab) coated MBs (MB_Thy1-Ab)] and control MBs [scrambled scFv coated MBs (MB_{scFv-scrambled}); Isotype-antibody coated MBs (MB_Non-targeted)] to 20 nM biotinylated human Thy1 (hThy1) protein by flow cytometry. DiOC₁₈ fluorophore incorporated in MB shell was used to gate the micron-sized particles and detect associated streptavidin-AlexaFluor-647 dye signal from the MB bound biotinylated-Thy1 protein. Only MB_Thy1-scFv and the positive control MB_Thy1-Ab showed binding signal to the soluble Thy1 protein. (B) Upper panel: A simplistic schema of the flow chamber cell attachment setup shows controlled flow (0.6 mL/ minute) of MBs. MS1 cells overexpressing Thy1 (MS1_Thy1) grown on a glass slide was inverted (curved arrow) with cells directly contacting the liquid containing MBs flowing at the set rate. Lower panel: Bar graph quantifications of MB_Thy1-scFv show significantly higher (P<0.01) number of MBs attached per MS1_Thy1 cell compared to the MB_Non-targeted under liquid shear stress conditions in flow chamber assay. Error bars represent standard deviation.

Figure 4.. MB_Thy1-scFv enhances Thy1-specific ultrasound (US) molecular signal in murine PDAC tumors.

(A) Left: A transgenic mouse model (PDAC mouse model; n=8) injected with MBs via the tail vein for US imaging. Right: Representative transverse-section contrast mode ultrasound images and bar graph quantification show significantly (4.4 fold; P<0.01) higher imaging signal in PDAC tumors (green outline) with MB_Thy1-scFv (5.3 ± 1.9 a.u.) compared to imaging with MB_Non-targeted (1.2 ± 1.0 a.u.). Thy1 differential targeted-enhancement (dTE) signal is represented as a colored overlay over the contrast-mode images. (B) Schematic treatment plan for L-arginine-induced pancreatitis in C57BL/6 mice (n=8). No significant (ns) differences in Thy1 imaging signal was observed in pancreatitis tissues with both MB constructs as shown in the dTE images (upper panel) and bar graph quantification (lower panel). B-mode images were used as reference to draw a region of interest (green outline) around the entire pancreas. (C). Bar graph show no significant (ns) difference in imaging signal in the normal pancreas (n=3) with both MB constructs. Error bars represent standard deviation.

Figure 5.. Validation of scFv-(Gly)₅-Cys binding activity in cells and tissues expressing human Thy1.

The process of scFv (without TrxA fusion protein; 29 kDa) scale-up production was optimized in a commercial facility and binding specificity validated in vitro. (A) Flow cytometry histograms confirming binding specificity of biotinylated scFv-(Gly)₅-Cys (100 nM) to MS1_Thy1 cells by streptavidin-AlexaFluor-647 detection. 150 nM of a commercially available biotinylated—anti-Thy1-antibody (Ab) is used as a positive control for the confirmation of cell-surface Thy1 expression. Detection signal intensity of scFv and antibody binding could be affected due to differences in their biotinylation levels. (B) SigmaPlot of dose-dependent (0 – 1000 nM) normalized scFv-(Gly)₅-Cys (biotinylated) flow cytometry binding signal (streptavidin-AlexaFluor-647) to MS1_Thy1 cells relative to background MS1_WT binding signal. (C) Immunofluorescence images of MS1 cells grown on slides confirm Thy1-binding specificity of scFv-(Gly)₅-Cys-AlexaFluor-647 conjugate (red) to MS1_Thy1 cells. Blue color indicates DAPI nuclear stain. (D) RNA-seq-based gene expression correlation analysis (Pearson’s R= 0.77; Spearman’s R=0.88) between Thy1 and endothelial/ angiogenesis gene signature (16 genes) in the human pancreatic cancer (red outline; n=179) and normal pancreatic (blue outline; n=171) tissue samples. Data available from The Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) databases. (E) Multi-panel immunofluorescence staining and corresponding H&E images of patient PDAC frozen tissue sections. The top and middle rows show representative staining with anti-VEGFR2 antibody (Alexa Fluor 488; green) and the scFv-(Gly)₅-Cys-AlexaFluor-647 conjugate (red). VEGFR2 staining is used as a reference maker for vasculature and tumor cells. Note the incidences where co-immunostaining occurs (yellow color). Thy1-scFv stained mostly vasculature (red arrows) and cancer associated fibroblasts around the ducts in the tumor environment. Left to right: R160052 (poorly differentiated adenocarcinoma); R160055 (invasive ductal adenocarcinoma); R170125 (invasive poorly differentiated adenocarcinoma); R160131 (invasive well-differentiated ductal adenocarcinoma); R160197 (poorly differentiated invasive ductal adenocarcinoma); R160226 (invasive ductal adenocarcinoma).