Humanized anti-DEspR IgG4S228P antibody increases overall survival in a pancreatic cancer stem cell-xenograft peritoneal carcinomatosis ratnu/nu model

Pancreatic peritoneal carcinomatosis (PPC), with the worst median overall-survival (mOS), epitomizes the incurability of metastatic cancer. Cancer stem cells (CSCs) underpin this incurability. However, inhibitors of CSC-stemness fail to increase mOS in cancer patients despite preclinical tumor-reduction. This shortfall reinforces that preclinical efficacy should be defined by increased mOS in the presence of cancer comorbidities, CSC-heterogeneity and plasticity. The primary objectives of this study are: to test the dual endothelin-1/signal peptide receptor, DEspR, as a nodal therapeutic target in PPC, given DEspR induction in anoikis-resistant pancreatic CSCs, and to validate humanized anti-DEspR antibody, hu-6g8, as a potential therapeutic for PPC. We used heterogeneous pools of CSCs selected for anoikis resistance from reprogrammed Panc1 and MiaPaCa2 tumor cells (TCs), and adherent TCs reprogrammed from CSCs (cscTCs). We used multiple anti-DEspR blocking antibodies (mAbs) with different epitopes, and a humanized anti-DEspR recombinant mAb cross-reactive in rodents and humans, to test DEspR inhibition effects. We measured DEspR-inhibition efficacy on multiple prometastatic CSC-functions in vitro, and on tumorigenesis and overall survival in a CSC-derived xenograft (CDX) nude rat model of PPC with comorbidities. Here we show that DEspR, a stress-survival receptor, is present on subsets of PDAC Panc1-TCs, TC-derived CSCs, and CSC-differentiated TCs (cscTCs), and that DESpR-inhibition decreases apoptosis-resistance and pro-metastatic mesenchymal functions of CSCs and cscTCs in vitro. We resolve the DNA-sequence/protein-function discordance by confirming ADAR1-RNA editing-dependent DEspR-protein expression in Panc1 and MiaPaCa2 TCs. To advance DEspR-inhibition as a nodal therapeutic approach for PPC, we developed and show improved functionality of a recombinant, humanized anti-DEspR IgG4S228P antibody, hu-6g8, over murine precursor anti-DEspR mabs. Hu-6g8 internalizes and translocates to the nucleus colocalized with cyto-nuclear shuttling galectins-1/3, and induces apoptotic cell changes. DEspR-inhibition blocks transperitoneal dissemination and progression to peritoneal carcinomatosis of heterogeneous DEspR±/CD133 ± Panc1-derived CSCs in xenografted nude rats, improving mOS without chemotherapy-like adverse effects. Lastly, we show DEspR expression in Stage II-IV primary and invasive TCs in the stroma in PDAC-patient tumor arrays. Collectively, the data support humanized anti-DEspR hu-6g8 as a potential targeted antibody-therapeutic with promising efficacy, safety and prevalence profiles for PPC patients.


Background
Cancer metastasis causes 90% of all cancer-related deaths and remains a high unmet need despite decades of research [1]. Pancreatic ductal adenocarcinoma (PDAC) epitomizes the unmet need with a 44:55 death-to-case ratio per year [2]. While second to PDAC liver metastasis in prevalence, peritoneal metastasis exhibits the worst median overall-survival (mOS) of all PDAC metastases [3] with rapid feed-forward dissemination-progression to pancreatic peritoneal carcinomatosis (PPC) [4]. Development of PPC after curativeintent surgery, despite post-resection adjuvant therapy, especially in patients with tumor cell-positive peritoneal fluid cytology 6g [5], indicates inherent PPC therapy-resistance that is not simply due to delayed diagnosis. PPC does not benefit from surgical debulking, unlike colorectal/ovarian peritoneal metastases [6], thus reiterating the therapeutic challenges and high unmet need in PPC.
Cumulative research implicates cancer stem cells (CSCs) [7] in post-resection cancer recurrence and metastasis [8,9], especially in PDAC [8]. However, the failure to meet primary endpoints in clinical trials of "CSC-only" inhibitors indicates the need for inhibition of both CSCs and TCs [9] in order to address therapy-resistance arising from bi-directional CSC/ TC-reprogramming or plasticity, and CSC/TC heterogeneity [9,10]. The dual endothelin-1/signal peptide VEGF receptor (DEspR) is a cell-surface accessible target induced on subsets of multi-potential, highly tumorigenic PDAC (Panc1-derived) CSCs and TCs [11]. DEspR-inhibition decreases CSC anoikis resistance, stress-survival, vasculo-angiogenesis, and RNA levels of pro-survival Mcl1 and cIAP2 proteins [11]. PDAC TCs express both DEspR ligands, endothelin-1 (ET1) [12] and signal peptide of VEGF (SP VEGF ) cleaved from the VEGF-propeptide [13]. Expression levels of both ET1 and VEGF (and by extension SP VEGF ) are associated with aggressive PDAC and poor outcomes [12,13]. However, while both ET1 and VEGF-axes are implicated in PDAC progression, inhibitors of ET1-A and/or ET1-B receptor [14] or the VEGF/ receptor-axis [15] are not FDA-approved targeted therapies for PDAC. By hypotheses elimination, DEspR, activated by ET1 and SP VEGF , remains the missing receptor pathway to inhibit when ET1 and VEGF and perforce, SP VEGF , are elevated.
DEspR is indeed a 'missing puzzle piece' as it is annotated as a non-coding gene, due to a stop codon [T-G-A] in the NCBI DNA database [16], in lieu of tryptophan-codon [T-G-G] at amino acid position-#14 (tryp#14). Given multiple experimental evidence detecting DEspR protein, functionality, and RNA-sequences with [T-G-A/G]-tryp#14 in placenta RNA-seq dataset [11,17,18], analysis for potential ADAR1 RNA-editing would clarify this DNA-protein discordance.
Here we show that DEspR, a stress-survival receptor, induced on CSCs and expressed in CSC-derived tumor cells, is a nodal therapeutic target in PPC. To advance a potential therapy, we developed, tested and validate the humanized anti-DEspR IgG4 S228P antibody as a potential targeted therapy for patients with PPC with a promising preclinical efficacy and safety profile, and a clinically relevant expression profile in primary and metastatic pancreatic cancer.

Materials and methods
Please see Supplementary Methods and Materials (Additional file 1) for specifics.

Study design
The purpose of this study was to investigate anti-DEspR therapy as a translatable pathway for curative-intent therapy for PPC. DEspR-inhibition was attained using different anti-DEspR antibodies binding to their respective epitopes on human-DEspR. In vitro experiments were designed to assess the impact of DEspR-inhibition on TC and CSC pro-tumor stress survival and functionalities. Two PDAC cell lines were used representing different KRAS mutations: Panc1 with G12D mutant KRAS detected in 70-95% of PDAC cases, and MiaPaCa2 PaCa2 with G12C mtKRAS detected in 1-3% of PDAC cases. Tumor cells and CSCs comprised different subsets of permutations of DEspR±/CD133 ± in order to represent heterogeneity. Studies were done on all key components of the CSC-TC spectrum: DEspR± CSCs, TCs, and cscTCs, in order to demonstrate efficacy of DEspRinhibition regardless of CSC/TC plasticity. Independent biological replicates were performed on different days with different experimenters, and technical replicates were performed in triplicate to demonstrate methodological rigor, unless otherwise stated. Different informative, functional endpoints were selected to affirm reproducibility of DEspR-inhibition efficacy in vitro.
In vivo experiments in CSC-derived xenograft (CDX)subcutaneous and PPC nude rat models evaluated the impact of DEspR-inhibition on tumorigenicity and progression, and overall survival in female and male PPC rats. PK/PD experiments were performed to better characterize anti-DEspR therapeutics. Outbred Rowett nude rat models can attain larger allowable tumors (20% of 250 g BW), hence longer timecourse with more complex tumors and cancer comorbidities, compared to tumor-to-bodyweight ratio limits in inbred nude mice (20% of 25 g BW). Contemporary age-matched controls were used. Treatment-group assignments were based on pre-study defined distribution scheme that ascertained that rats in treated and control groups were littermatched to the best possible, of identical ages, equivalent weights, and received treatment/mock-treatment injections under identical conditions. All animals were monitored by blinded Lab Animal Science (LASC) technicians, with prior IACUC approved study-endpoints without modification. Sample size was calculated based on pilot studies in order to allow significant statistical power to assess primary endpoints. DEspR expression in human PDAC tumor arrays was quantified in blinded manner. Tumor cores, in duplicates, represented different PDAC stages of disease. All available data points were included in analysis.

Collagen-1-α1 ELISA
Detection of secreted COL1A1 present in Col1/3 was determined using COL1A1-ELISA kit (CusaBio Technology, CSB-E13445h) following manufacturer's specifications. Col1A1-secretion was measured from supernatants of 7c5-treated and control Panc1 CSCs. Optical density from the assay and standards were collected at 450 nm with corrections subtracted from 540 nm, using a SpectraMax M3 microplate reader (Molecular Devices, San Jose, CA).

7c5-AF568 immunofluorescence staining
Immunofluorescence staining was performed as previously described [11,18]. Binding was performed at 4°C with 10 μg/ml 7c5-AF568. Internalization was induced with prewarmed 37°C cell media. Cells were then fixed after 15min, 30-min, 1-h, and 2-h as noted above. Images were performed on a Zeiss Axioskop fluorescence microscope and Leica SP5 confocal microscope. Additional information is listed in Supplementary Methods (see Additional file 1).

CSC tumorsphere assay
Tumorsphere assay was modified based on previously described protocols [11]. Panc1-WT and ADAR1-KO TCs, as well as MiaPaCa2-WT and ADAR1-KO TCs (p5) were plated on 96-well cell culture ultralow adherence plates, in tumorsphere media, at 250, 500, and 1000 cells/well. Cells were imaged using Celigo Imaging Cytometer (Nexcelom, Lawrence, MA), with tumorsphere colonies counted as ≥50 μm in diameter, using the Celigo tumorsphere counting protocol. A graphical representation of the experimental design is provided in Figure S2b.
3D hu-6g8 modeling 3-D modeling of hu-6g8 was done using the ABody-Builder tool in Therapeutic Antibody Profiler (UK).
Heavy and light chains were numbered using the IMGT, Chothia, Kabat, North/Aho, and Contact numbering scheme via "Antigen receptor Numbering and Receptor Classification" ANARCI tool. Then "ABodyBuilder" was used to create a homology model of antibody sequence using SAbDab to find framework templates, FREAD to homology model loops, MODELLER/SPHINX if FREAD fails, and PEARS to model side chains [http://opig.stats. ox.ac.uk/webapps/newsabdab/sabpred/tap].

Binding saturation
Binding saturation was performed as previously described [11]. Direct antibody binding to live Panc1 TCs and CSCs was evaluated by flow cytometry using hu-6g8-AF568 or 6g8-AF568 in sequential, serial concentrations (0.3-30 μg/ml), under conditions identical to those described above. Each data point was performed in duplicate. Summary of data is provided in Table S2.

Angiogenesis assay
HUVEC assays were performed as previously described [11], comparing hu-6g8 or 6g8 in sequential, serial concentrations (0.5-30 μg/ml), under identical conditions. Assay conditions were performed in quadruplicate. After 16-h of treatment, tube formations were digitally photographed and analyzed using ImageJ.

Animals
Outbred Rowett nude nu/nu (Charles River Labs) were used for all in vivo experiments. Rats were 4-5-week-old (female) or 3-4-week-old (male) at time of cell injection. All studies were performed in accordance with IACUC approved protocol. See Additional file-1 Supplemental Methods for additional information.
Heterotopic subcutaneous Panc1 PPC model Two-million CSCs were pretreated with 200 μg/ml or vehicle control for 1 h at 4°C in M2 media (Sigma, M7167) prior to injection into female rats. Study ended when vehicle-control reached maximum allowable tumor size or reached 100-days (5g12). Tumor volumes were assessed at study endpoints via caliper measurements. Tumor volumes were calculated using the formula (4/3πr 1 2 × r 2 ) where r 1 is the larger, and r 2 the smaller radius [11].

Orthotopic Panc1 PPC model
For the Panc1-CDX PPC model, nude nu/nu rats received an intraperitoneal injection of 200 mg/kg cyclophosphamide (Sigma, cat# C7397), 3 days prior to intraperitoneal injection of 2-million Panc1 CSCs in M2 media. For CSC pre-treatment studies, Panc1 CSCs were pretreated with 200 μg/ml 5g12 or 6g8, or vehicle control for 1 h at 4°C. For the PPC eight-dose treatment studies, PPCnude-rats received twice-weekly intraperitoneal injections of 1 mg/kg 6g8 or 7c5, 26 mg/kg gemcitabine (Sigma, G6423), or saline for 4-weeks, starting 7-days post-CSC injection. For the female-PPC single-dose study, PPC-nude-rats received a single-iv injection of 3 mg/kg or 15 mg/kg hu-, or a single-iv injection of 100 mg/kg intraperitoneal gemcitabine or saline.
21-days after cell injection. For the male-PPC singledose study, PPC-nude-rats study received a single-iv injection of 15 mg/kg hu-6g8 or a single-iv injection of saline 21-days after engraftment. For neutrophil, platelet, and neutrophil-lymphocyte ratio (NLR) determinations, blood was collected at days 28, 35, and 42 post-injection in 1% EDTA. Blood was analyzed using HEMAVET 950 FS Auto Blood Analyzer (Drew Scientific, Miami Lakes, FL) with rat-species settings.
Pharmacokinetic study PPC rats received a single-iv bolus of 3 mg/kg or 15 mg/kg hu-6g8 4-weeks after tumor engraftment. Blood was drawn at 5-min, 15-min, 30-min, 1-h, 8-h, 24-h, 72-h, 1wk, and 4-wks in 1% EDTA tubes, with plasma isolated by centrifugation. Protein levels were assessed using Nano-Drop™ as described above, and antibody levels were measured by Western blot. Equal (10 μg) protein was loaded into 12% Mini-Protean Gels (Bio-Rad 4,561,045). Protein transfer was performed using Immuno-Blot PVDF membrane. Anti-GAPDH was used as a protein loading control under conditions identical to those above. Hu-6g8 was detected using anti-human IgG HRP (Sigma AP112P, 1: 10000) for 14 h at 4°C. Blots were developed and imaged as described above. Analysis was performed using ImageJ with densometry analysis, with concentrations determined from loaded protein standards.
Target-engagement study PPC models were identical to those described above. Once tumors were palpable in the greater omentum of orthotopic PPC male and female rats, they received a single-iv injection of 3 mg/kg hu-6g8 or IgG4 isotype control. After 24 h, rats were anesthetized, receiving PBS aortic perfusion prior to tissue collection under isoflurane anesthesia. PPC tumor samples and adjacent abdominal organ tissues were collected and fixed in PBS-buffered, pH 7.4, 4% paraformaldehyde. Paraffin-embedded slides were prepared (AML Labs, St. Augustine, FL), and immunofluorescence staining was performed as previously described [11]. Slides were treated overnight with 10 μg/ml AF568-anti-human IgG in a humidified chamber. Imaging was performed with a Zeiss Axioskop fluorescence microscope, as previously described [11]. To determine bioeffects, immunohistochemistry was performed using antibody to activated Caspase-3 and Ki67, with DAB secondary detection system (Mass Histology Services, Worcester, MA).

Statistical analysis
Statistical analyses were performed using GraphPad PRISM 8.3 and Sigma Stat software. Paired Student's ttest was used to compare means between two groups. Chi-square tests-of-independence were used to compare categorical data. Analysis of variance and nonparametric Kruskal-Wallis one-way analysis of variance were performed when appropriate for ≥3 study groups. Correlation analysis was performed using the Pearson correlation coefficient (R) between continuous variables. Colocalization across imaging was performed using Manders coefficient. Differences in OS were calculated with the Kaplan-Meier survival curve, Mantel-Cox log rank statistic, and Holm-Sidak multiple comparison test. P values were corrected using the Bonferroni multiple comparison testing. Statistically significant values were indicated as follows: *p ≤ 0.05,**p ≤ 0.01,***p ≤ 0.001, and ****p ≤ 0.0001 unless otherwise stated.

Results
To study DEspR as a potential therapeutic target for PPC, we used multiple blocking anti-DEspR monoclonal antibodies (mAbs) raised against two extracellular domain epitopes of human DEspR (Fig. 1a). Epitope-1 anti-DEspR mAbs recognize a human-specific domain (7c5, 5g12), whereas the epitope-2 mAb, 6g8, binds to a conserved epitope in humans, rats, and monkeys, and spans the contested tryptophan [W] #14 currently annotated as a stop codon in the NCBI DNA-seq database (Fig. 1a) [18]. Both are upstream to consensus sequences for experimentally proven N-glycosylation and internalization recognition signal (IRS) consensus sequences (Fig. 1a).
We assessed the efficacy of DEspR-inhibition in PPC using a stepwise, experimental system that modeled CSC-heterogeneity and plasticity, two factors which underpin cancer therapy resistance. We modeled CSCheterogeneity by using DEspR±/CD133 ± CSCs and DEspR± TCs and cscTCs in all tests. We modeled CSC-plasticity by testing in the presence of both TC-to-CSC and CSC-to-cscTC reprogramming in in vitro and in vivo experiments. To advance a clinically-translatable therapeutic paradigm, we prioritized the study of parameters assessing anti-cancer efficacy of DEspR-inhibition at the cellular level in vitro, and subject overall survival in vivo.

DEspR-inhibition decreases Panc1-TC and CSC stresssurvival in vitro
To gain insight into efficacy, we first studied the impact of DEspR-inhibition on TC stress-survival by studying apoptotic cell morphology changes induced by anti-DEspR-mAb receptor binding and internalization. This would unify previous data reporting that DEspR cell signaling supports key pro-survival pathways [11], DEspR-inhibition decreases pro-survival gene RNA expression levels in stress-resistant Panc1-CSCs [11], and 7c5-antibody/ DEspR complexes internalize and translocate to the nucleus in Panc1-TCs [18], into a putative therapeutic paradigm. We therefore characterized DEspR-bound 7c5internalization by treating Panc1-TCs with AF568-labeled 7c5 (7c5-AF568) and tracking 7c5/receptor internalization using serial fixed-immunofluorescence microscopy. We analyzed intracellular quantity and localization of 7c5-AF568, and determined concomitant apoptotic cell morphology changes in relation to internalized 7c5-AF568.
After 1 h, we detected fluorescently-labeled AF568-7c5/DEspR internalization into the cytoplasm with some nuclear localization, in contrast to minimal non-specific internalization of the AF568-IgG2b isotype-control (Fig.  1b). By 2-h, we observed further nuclear localization and a significant increase in various apoptotic cell morphology changes in AF568-7c5-treated Panc1 TCs in contrast to the lack of nuclear localization in isotype-treated controls (Fig. 1b). A few cells in 7c5-AF568 treated TCs exhibited rounded cell swelling rather than apoptosis cell shrinkage, consistent with necroptosis phenotype; none detected in isotype controls (Fig. 1b). These data are concordant with previous data showing that overnight DEspR-inhibition using 7c5 in Panc1-CSCs decreases anti-apoptotic Mcl-1 and anti-apoptosis/necroptosis CIAP2 or BIRC3 RNA levels [11], both key pro-survival proteins in PDAC TCs [19,20].

DEspR-positive/negative subsets among TCs, cscTCs and CSCs
To assess DEspR-accessibility as a therapeutic target in PPC, we assessed its relative cell surface expression by flow cytometry. We compared Panc1-TCs and Panc1-CSCs as components of CSC-TC plasticity and heterogeneity [10], and tested whether DEspR cell-surface expression increases in anoikis-resistant Panc1-CSCs. To this end, we studied functionally selected anoikis-resistant CSCs from reprogrammed Panc-1-TCs regardless of marker cell-surface expression. These CSCs exhibited high tumorigenicity previously validated in vivo [11].

DEspR-inhibition decreases CSC anoikis-resistance and spheroid formation
To determine the impact of DEspR inhibition on CSC anoikis resistance and tumorsphere formation in the presence of different CSC-subsets, we treated DEspR± Panc1-CSCs with fluorescently-tagged AF568-7c5, then cell-sorted fluorescent DEspR+ from DEspR-CSCs by MoFlo Cell sorting. Testing survival in low adherence anoikis-culture conditions, we observed that Mo-Flo sorted DEspR[−] CSCs grew by 5 days, forming sheetlike clusters, rather than tumorspheres (Fig. 1e-third  panel). However, after the 7c5-bound/sorted DEspR+ CSCs were non-viable (Fig. 1e-fourth panel). Interestingly, the subsequent passage of the DEspR[−] CSC-pool in low adherence culture conditions re-established a DEspR+ pool in 30-40% of CSCs as detected by flow cytometry. These observations confirm DEspR roles in CSC stress (anoikis)-survival which impacts stemnessassociated spheroid formation.

Detection of ACTA2 (αSMA) and Col1A1 expression on CSCs and cscTCs
To determine the effects of DEspR inhibition on Panc1-CSC and cscTC mesenchymal functions relevant to peritoneal dissemination, we assessed alpha-smooth muscle actin (αSMA) expression, and concomitant downstream expression and release of collagen-1 (Col1A1) given their prometastatic functionality and potential contritubution to desmoplasia in the tumor microenvironment. Immunofluorescence detected αSMA expression in TNF-α stimulated cscTCs (Fig. 1f), which prompted evaluation of collagen1 (Col1A1) expression, as αSMA expression in activated fibroblasts is associated with upregulation of Col1A1 expression and secretion, but TNF-α decreases Col1A1 in dermal fibroblasts. Surprisingly, αSMA+ cscTCs co-expressed Col1A1 in pure cscTC cultures (Fig. 1f) without fibroblast co-cultures.
To examine the impact of DEspR-inhibition on αSMA/Col1A1 expression, we tested anti-DEspR 6g8and 7c5-mediated DEspR inhibition in Panc1-cscTCs and CSCs. Anti-DEspR 6g8 treatment suppressed both the expression of αSMA in cscTCs (Fig. 1g) and the expression/secretion of Col1A1 from CSCs in vitro (Fig.  1h). Similarly, anti-DEspR 7c5 suppressed Col1A1 expression/secretion from CSCs (Fig. 1h). Whether this suppression is direct or via induction of apoptosisassociated function-shutdown remains to be determined.
To study DEspR-Col1A1 co-expression in vivo, we examined tumor sections from Panc1-CSC derived xenograft (CDX)-PPC nude rats by immunofluorescence (IF) using human-specific DEspR (7c5) and human/rat reactive Col1A1 fluorescently labeled antibodies. Multiplex-IF shows that PPC-TCs in tumor cell islands co-express human-specific DEspR and Col1A1, but that DEspR −/Col1A1+ TCs are also present in the same tumor cell islands (Fig. 1i). Human-specific DEspR+/Col1A1 Panc1-CSC-derived TCs in PPC tumor sections are distinguished from rat host stromal fibroblasts by cell morphology, location, and negativity for human-DEspR+ expression (Fig. 1i). Concordantly, representative Masson Trichrome and H&E-stained section (Fig. 1j) shows pericellular collagen deposition. These data demonstrate that Panc1 cscTCs and CSCs also contribute to pericellular collagen deposition in PPC tumors.

ADAR1 regulation of DEspR protein in Panc1 and MiaPaCa2 nonCSC-TCs
To address the discordant NCBI annotation of the DEspR locus as a transcribed non-coding gene, FBXW7antisense RNA (AS-1), with our data showing DEspR protein expression and functionality [18], we tested whether ADAR1 RNA-editing would reconcile this discrepancy via A/I(G) RNA editing. Since site-specific RNA editing can occur at levels as low as 0.1%, with the average being only 20% RNA-edited species [21], a low, but physiologically relevant, percentage of RNA-edited transcripts could be vulnerable to algorithm-based exclusion as 'noise/error' in high-throughput RNA-seq databases. To affirm that DEspR meets the structural requirements for ADAR1 binding, DNA database analysis shows that the DEspR RNA spanning the contested tryp-#14 exists as double-strand-RNA (dsRNA) with the FBXW7-RNA on the antisense strand, thus meeting the dsRNA-requirement for ADAR1-binding for RNAediting [21]. Furthermore, the DEspR-RNA sequence spanning the contested RNA-edited A/I(G) site contains a putative hairpin loop with three smaller loops that is concordant with secondary structure requirements sufficient to guide ADAR1 binding for nucleotide-specific editing [22] (Fig. S1a).
To obtain experimental evidence for ADAR1dependent DEspR expression, we first performed double-immunostaining of two PDAC-TC lines, Panc1 [KRAS G12D ] and MiaPaCa2[KRAS G13D ] to determine if every ADAR1-expressing TC also expresses DEspR. Data show that ADAR1 and DEspR co-expression overlap in both Panc1-TCs and MiaPaCa2-TCs, such that all DEspR+ TCs expressed ADAR1 (Fig. S1b-c). Next, we performed ADAR1-knockout using a CRISPR/Cas9 knockout system (Fig. S2a). To determine the effects of ADAR1-knockout, we performed a timed-series flow cytometry analysis. This confirmed co-expression of DEspR and ADAR1 proteins in DEspR+ cells in Panc1 (Fig. 2a) and MiaPaCa2 (Fig. 2c) TCs at baseline, loss of ADAR1 expression by 3rd passage after transfection, and subsequent loss of DEspR-expression in both Panc1 (Fig.  2a) and MiaPaCa2 (Fig. 2c) by the 4th passage (Table  S1). Mock-control knockout of Panc1 and MiaPaCa2 cells using the mouse-specific ADAR1 construct did not reduce human-ADAR1 nor human-DEspR expression, but confirmed transfection via GFP-reporter gene expression (Fig. 2e), thus affirming the specificity of human ADAR1 knockout experiments (Fig. 2a, c), and ruling out confounders from transfection process and/or from puromycin-based selection.

DEspR-inhibition of Panc1-CSC tumorigenicity and transperitoneal dissemination
Having demonstrated efficacy of DEspR-inhibition of CSC/TC stress-survival and functionalities in vitro, we next tested whether DEspR-inhibition of mixed DEspR± Panc1-CSCs would suffice to reduce tumorigenicity in vivo, given targeted-sparing of all DEspR[−] CSCs. We used a heterotopic-subcutaneous xenograft model in nude rats (Fig. 4a) to attain 10x larger and hence, more complex and heterogeneous tumors than would be attainable in mice, and to facilitate tracking tumor volume and invasiveness [11]. We observed that one-hour treatment of DEspR± Panc1-CSCs with human-specific epitope-1 5g12-mAb prior to subcutaneous injection decreased tumorigenicity as measured by tumor volume, a net decrease that was sustained over time compared to non-treated Panc1-CSCs (Fig. 4b). Similarly, 1-h pretreatment of mixed DEspR± Panc1-CSCs prior to intraperitoneal injection (Fig. 4a), using epitope-1 (5g12) and epitope-2 (6g8) mAbs (Fig. 1a), decreased Panc1-CSC peritoneal dissemination, resulting in significantly increased survival of PPC-rats compared to non-treated CSC-controls (Fig. 4c). Since 7c5 and 5g12 are human-DEspR-specific, and CSC mixed-pools comprise both DEspR+ and DEspR-CSCs, net-decreased tumorigenicity reflects the importance of DEspR+ CSC subset in tumor establishment, dissemination, and progression. The more robust impact on tumorigenicity in subcutaneous vs peritoneal xenografting is concordant with greater metastatic-permissiveness in the peritoneal microenvironment, as observed in PDAC patients [25].
To test the impact of anti-DEspR therapy on overall survival (OS) as a translatable and clinically relevant endpoint, we used a Panc1-CDX-PPC nude rat model (Fig. 4d). This Panc1-CDX-PPC model recapitulates key clinical PPC features and predilection for the greater omentum [26], along with key clinical comorbidities including: ascites, jaundice, gut dysfunction, and high mortality [27]. We excluded primary PDAC to attain a timed onset of PPC to enhance reproducibility of survival studies which a priori requires quantitatively equivalent onset and tumor burden for each study subject. Additionally, a PPC-only model can determine the impact of PPC per se on overall survival and comorbidities, as well as test CSC capability in peritoneal dissemination-progression without need for premetastatic niche formation [28]. Treatment started 7-days after Panc1-CSC intraperitoneal (ip) injection, a time point when PPC rats exhibit multiple (> 20) visible (1-3 mm) peritoneal tumors, which in patients would contraindicate curative-intent surgery. Murine mAbs 7c5-and -treatments significantly increased OS compared to saline mock-treated controls (Fig. 4e, Table S4). In contrast, standard-of-care gemcitabine (26 mg/kg/dose ip), equivalent to mouse dose 198 mg/kg (see Additional file 1: Supplementary Methods), or human dose 1000 mg/m 2 did not significantly extend OS compared to saline-control rats (Fig.  4e, Table S4).
Next, we tested the humanized clinical candidate hu-6g8 in PPC nude rats at a more advanced stage, approximately 3-4 weeks post-Panc1-CSC intraperitoneal xenografting (Fig. 4d), recording both efficacy and adverse events. We compared hu-, at 3-and 15-mg/kg/dose given intravenously (iv), to gemcitabine at 100-mg/kg/dose intraperitoneally (ip), equivalent to~3x-human 1000-mg/ m 2 dose, or 760 mg/kg mouse dose (see Additional file 1: Supplementary Methods). At this stage, PPC rats typically exhibit multiple palpable peritoneal tumors (> 20 tumors: ≥ 5-mm diameter), at times in matted confluence, and with dissemination to the other abdominal organs and retroperitoneal space, but prior to comorbid ascites, jaundice, or intestinal dysfunction. We used single-dose treatment regimens to eliminate confounders from rat-host immunogenic response to foreign/human IgG4-protein, and used single high-dose 100 mg/kg ip gemcitabine with dose-limiting toxicities if given 2 doses.
Notably, single-dose hu-6g8 extended median survival significantly in a dose-dependent manner, compared to mock-treated saline control (Fig. 4f). Survival outcomes were equivalent between 3-mg/kg hu-6g8 and high dose gemcitabine (Fig. 4f); whereas, 15-mg/kg hu-6g8 single-dose treatments showed significant improvement in survival compared to dose-limiting gemcitabine therapy.
To assess for potential sex-dependent efficacy, we tested Panc1-CDX-PCC male rats using the identical single dose of 15-mg/kg hu-6g8, the more efficacious dose in PPC female rats. We observed a similar improvement in survival between female (Fig. 4f) and male rats (Fig. 4g) with advanced-PPC, consistent with sexindependent efficacy. We also initiated a survival study in MiaPaCa2 CDX-PPC nude rat model but aborted this survival-study due to < 100% tumor penetrance in nontreated controls despite identical conditions with Panc1-CDX-PPC model. Additionally, intraperitoneal administration of hu-6g8 mAb gave equivocal results suggesting vulnerability of hu-6g8 mAb to proteases present in advanced PPC-peritoneal/ascites fluid.
In addition to survival benefits, hu-6g8 treated CDX-PPC rats had decreased tumor burden and comorbidities compared to untreated PPC-rats. Comparing the tumor burden of saline mock-treated PPC rats at time of death to a contemporary age-matched 15-mg/kg hu-6g8treated PPC rat that was euthanized to match tumor duration, we observed greater omental tumor burden, as well as distended gut, ascites, jaundice, and biliary obstruction from tumor invasion at the porta hepatis in the saline-control PPC rat, but not in the hu-6g8 treated rat (Fig. 5a).
To further assess pharmacological parameters in support of in vivo efficacy, we tested for hu-6g8 target engagement and bioeffects in CDX-PPC female and male rats with established peritoneal metastatic tumors (≥ 3 weeks after CSC ip-injections) 24-h after iv-infusion of a single 3-mg/kg dose. In contrast to human IgG4-isotype infused tumors, with minimal immunofluorescent-positive target-engagement, tumors from hu-6g8 treated PPC rats exhibited human-specific IgG immunofluorescence in the majority of TCs, indicating target-engagement, (Fig. 5b,  Fig. S4b). Furthermore, we detected good tumor penetration of hu-6g8 and high tumor specificity when contrasted to normal pancreas exhibiting zero-immunofluorescence (Fig. 5b, Fig. S4b). These data support clinical feasibility as potential PPC therapy.
To test for predicted target-bioeffects 24-h after infusion, we analyzed adjacent serial sections for apoptosis measuring activated caspase-3 by immunohistochemistry (IHC). Isotype mock-treated tumors exhibited minimal activated caspase-3 in contrast to increased activated caspase-3 immuno-staining in hu-6g8-treated PPCtumor cells (Fig. 5c). Importantly, adjacent tissues from normal stomach, duodenum, liver, pancreas, lung, spleen, vasculature, and adipose (Fig. 5d) did not exhibit induced-apoptosis in hu-6g8-treated PPC-CDXs, indicating tumor-targeted specificity and sparing of normal DEspR[−] tissues. To assess impact on tumor cell proliferation, we observed that hu-6g8-treated tumors demonstrated decreased cellular proliferation measured by number of Ki67+ tumor cells on immunohistochemistry (IHC) (Fig. S4c). We also observed decreased number of tumor microvessels in tumor areas with decreased Ki67-IHC and increased tumor cell loss (Fig. S4c).

Insight into clinical translational relevance
Analysis of hu-6g8 immunofluorescence in 133-patient PDAC-tumor arrays detected no DEspR expression in normal pancreas (Fig. 6a-b), in contrast to DEspR+ expression in TCs and microvessels in all PDAC stages (I-IV), as well as in invasive TCs in tumor-stroma ( Fig.  6a-b). Isotype-IgG4 immunostaining verified specificity of hu-6g8 immunostaining (Fig. 6a). Additionally, we observed DEspR+ expression in the majority of hepatic, omental, and peritoneal metastatic TCs (Fig. 6b). Quantitation by blinded scoring revealed that 82% of PDAC tumor cores with > 50% of tumor proportion scores are DEspR+ (Fig. 6c), and that 90% of 79 tumors with tumor cells invading the stroma are DEspR+ (Fig. 6d).

Discussion
DEspR as a nodal therapeutic target in PPC Cumulative data demonstrate DEspR as a nodal therapeutic target across the PDAC CSC-to-TC spectrum. DEspR is coexpressed in CD133 + ALDH1+ CSCs, and in αSMA+ cscTCs. Blocking anti-DEspR murine precursor and humanized antibodies bind cell-surface DEspR, internalize, and cotranslocate with gal1/gal3 nuclear shuttling proteins to the nucleus, subsequently inducing apoptosis in vitro and in vivo. In the PPC-only nude rat model, DEspR-inhibition decreases CSC transperitoneal seeding and subsequent tumor progression and re-dissemination, associated PPCcomorbidities, thereby increasing mOS in both male and female xenografted nude rats.

Insights into CSCs in PPC
Specific to the CSC-paradigm in cancer metastasis, the development of PPC after intraperitoneal injection of Panc1-CSCs indicates CSC self-sufficiency in tumor-seeding, and in orchestrating the feed-forward dissemination-progression to pancreatic peritoneal carcinomatosis without requiring prior priming of metastatic beds observed in hematogenous metastases [37]. Specific to CSC-heterogeneity and CSC-subset hierarchy, inhibition of DEspR+ CSCs/TCs, in the presence of DEspR[−] CSCs/TCs, support nodal roles of DEspR+ CSCs/TCs in PPC tumorigenicity and transperitoneal dissemination-progression. Furthermore, conversion of DEspR[−] to DEspR+ CSCs in low adherence stress-culture conditions supports DEspR-roles in anoikis resistance. This also confirms molecular interconversion reported between CSCs subsets [9]. The demonstration of CSC-expression and secretion of Col1A1 independent of fibroblasts or pancreatic stellate cells, and DEspR+ CSC-derived tumor microvessels in vivo [11], suggest CSC self-sufficiency in orchestrating tumorigenesis and progression in the peritoneal space. These data delineate DEspR+ CSC-subset as a nodal driver in transperitoneal dissemination-progression.

Resolution of DNA-database discordance
In addition to validating the efficacy of DEspR as a therapeutic target in PPC, we show that ADAR1-dependent DEspR expression, via CRISPR/cas9-knockout studies. This reconciles the stop-codon detected in NCBI-DNA databases to cumulative data showing DEspR protein functionality and detection of tryptophan-codon#14-TGG by amplification-refractory mutation system (ARMS)-PCR and in placenta RNA seq-database [18]. These molecular data are further supported by detection of epitope-2, which spans the questioned tryptophan-#14 at epitopemidpoint, by mAbs 6g8 and hu-6g8. We confirm protein function by detection of gal1 and gal3 colocalization, which were previously detected on glycosylated DEspR pull-down experiments [18]. Altogether, data confirm that the DEspR protein is a functional protein upon ADAR1 RNA-editing. Importantly, concordance of pro-cancer roles of both DEspR presented here and prior [11,18,38] and ADAR1-editase, and ADAR1-dependency of PDAC cell lines [39] strengthen observations in this study.

Limitations of the work
We acknowledge limitations to our studies. We cannot comment on putative differences of DEspR-inhibition between KRAS-mutant vs. wild-type PDAC tumor cells, however we note that > 80-90% of PDAC patients have KRAS mutations which are associated with worse prognosis. Similarly, we cannot comment on preclinical efficacy of DEspR-inhibition in PDX or in immunecompetent PPC/PDAC models, however, insights gained in current survival studies using the PPC model will guide future preclinical efficacy studies in different PDAC and metastatic-PDAC models. Furthermore, while we have elucidated part of the downstream targets of DEspR that are decreased upon DEspR-inhibition such as pro-survival protein Mcl-1 in CSCs and TCs, as well as αSMA and Col1A1 in cscTCs and CSCs (presented here), and multipotential CSC-vasculogenesis [11], we recognize that further work remains to be done to dissect each paradigm. Nevertheless, the work presented here comprehensively support DEspR as an important and clinically relevant therapeutic target in PPC, with DEspR-inhibition providing a promising therapeutic approach for patients with pancreatic peritoneal carcinomatosis.
Boston University (BU) policy; and available to industry partners in compliance with BU or option-licensee NControl Therapeutics, Inc.

Declarations
Ethics approval All animal model studies were done in compliance with Boston University IACUC approved protocol AN15160 to V. Herrera and N. Ruiz-Opazo, and in compliance with ARRIVE 2.0 (Additional file-1: Supplemental Methods and Materials).