Collagenase-Expressing Salmonella Targets Major Collagens in Pancreatic Cancer and Improves Immune Checkpoint Blockade

: Therapeutic resistance in pancreatic ductal adenocarcinoma (PDAC) can be attributed, in part, to a dense extracellular matrix containing excessive collagen deposition. Here, we describe a novel Salmonella typhimurium (ST) vector expressing the bacterial collagenase Streptomyces omiyaensis trypsin (SOT), a serine protease known to hydrolyze collagens I and IV, which are predominantly found in PDAC. Utilizing aggressive models of PDAC, we show that ST-SOT selectively degrades intratumoral collagen leading to enhancement of immune checkpoint blockade (ICB) therapy in tumor-bearing mice. Ultimately, we found that ST-SOT treatment significantly modifies the intratumoral immune landscape to generate a microenvironment more conducive to ICB.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) represents >90% of all pancreatic cancers and currently has a dismal five-year survival rate of 10% [1,2]. Diagnostic efforts to detect early disease are hindered due to lack of specific symptoms, advanced local tumor growth, and early metastasis, the latter of which frequently excludes surgical resection as a viable treatment [3,4]. Treatment options beyond resection are limited due to the chemoresistance of PDAC, attributed to its characteristic desmoplastic reaction that results in significant increases in collagen content compared to healthy pancreas [5,6]. Of the twenty-eight types of collagen, collagen I (fibrillar) constitutes the majority of PDAC stroma and has been suggested to promote PDAC cell only) and induced (increasing L-arabinose) conditions, followed by western blot (WB) of pellet lysates. WB revealed robust expression of His-tagged SOT at the correct molecular weight (~31 kD) at all percentages of L-arabinose as well as tight regulation of protein expression under uninduced and induced conditions ( Figure   1B). To determine the subcellular location of SOT expressed under induced conditions, we performed immunofluorescent staining to detect the His-tag fused to the C-terminus of SOT protein ( Figure 1C).
Immunofluorescent staining of His-tagged SOT under induced conditions revealed clear localization outside of the bacterial cytoplasm and on the surface, defined by DAPI staining of genomic DNA. Altogether, these data ST-SOT under uninduced and induced conditions were stained at indicated time points with acridine orange (live, green) and ethidium bromide (dead, orange) and imaged by fluorescence microscopy at 100X magnification. Scale bar confirm that expression of the SOT transgene is tightly regulated using an inducible pBAD system and that the recombinant SOT protein is auto-displayed on the bacterial surface of ST.
Whereas tight regulation of transgene expression is important for minimizing toxicity during bacterial growth stages, sustained viability following induction will also be critical to maximize SOT activity. Thus, we determined growth kinetics of ST-SOT over 24 hours in non-inducing and inducing (2% L-arabinose) conditions at various starting optical densities (O.D.). Under uninduced conditions, ST-SOT reached a maximum O.D. within 6 hours that was maintained through 24 hours ( Figure 1D). Following induction, there was no observable increase or decrease from initial O.D., suggesting that SOT expression attenuates replication without causing immediate ST loss. To further investigate bacterial viability after induction, we performed live/dead staining using a mixture of acridine orange (AO) and ethidium bromide (EB), respectively, at 4 and 24 hours under uninduced and induced conditions [32,33]. As shown in Figure 1E, the percentage of live bacteria 4 hours after induction was not significantly different through 24 hours but was significantly different from percent viability observed for uninduced ST-SOT. Overall, these results confirm robust expression of SOT by ST that is associated with fitness cost to ST, thus emphasizing the importance of using an inducible system to allow for initial expansion of ST-SOT and robust SOT expression for subsequent in vitro and in vivo functional studies.

ST-SOT Hydrolyzes Major Collagen Substrates
To evaluate the collagenase activity of SOT expressed by ST, we employed the use of gelatin-agar plates and fluorescently-labeled substrate assays [34,35]. Gelatin is the product of partially-hydrolyzed collagen and when further hydrolyzed, in agar plates, forms a visible white precipitate [36]. We proceeded to culture ST-SOT under uninduced or induced (3% L-arabinose) conditions for 3 hours and then spotted 5µL of culture (1x10 8 colony forming units (CFUs)) onto gelatin-agar plates overnight. As anticipated, an area of hydrolysis was observed for induced ST-SOT that was limited to the size of the colony, suggesting that SOT is anchored to the bacterial membrane and not secreted to cause diffused hydrolysis outside the perimeter of the colony (Figure   2A). The ability of ST-SOT to hydrolyze gelatin was further confirmed using quenched FITC-labeled gelatin, which upon hydrolysis, is converted into fluorescent peptides. As shown in Figure 2B, induced ST-SOT (+Ara) caused significant increases in fluorescence intensity, compared to uninduced control (-Ara), within 4 hours and continued to increase through 24 hours. These results are the first to suggest that SOT expressed by ST has sufficient functional activity to hydrolyze the less-complex gelatin. We next determined whether ST-SOT could hydrolyze the major collagen types found in PDAC, namely I and IV. Indeed, induced ST-SOT also caused significant increases in fluorescence intensity when co-incubated with FITC-labeled collagen I or IV ( Figure   2C, D), further confirming that SOT expressed by ST exhibits collagenolytic function. Additionally, we observed no significant change in fluorescence intensity when FITC-labeled substrates were co-incubated with uninduced ST-SOT, emphasizing tight regulation by the inducible pBAD promoter system.

In Vivo Depletion of Collagens by ST-SOT is Restricted to PDAC Tumor Tissue and Augments ST Diffusion
We next determined the ability of ST-SOT to colonize and deplete collagen in orthotopic (o.t.) Kras G12D p53 R172H Cre Pdx1 (KPC) 4662.5 and subcutaneous (s.c.) Pan02 tumors when delivered intravenously into wildtype C57BL/6 mice [37,38]. We first verified that the YS1646 vector used in the construction of ST-SOT was capable of colonizing o.t. and s.c. tumors by using a constitutive bacterial reporter construct encoding the bioluminescent LUX operon [39]. When 5x10 6 CFU recombinant YS1646 encoding LUX (ST-LUX) was injected intravenously (i.v.) into C57BL/6 mice bearing o.t. KPC4662.5 or s.c. Pan02 tumors (>150 mm 3 ), we observed bioluminescence localized to the area of the tumor, which was typically detected by day ~2 and    Figure 3C), which was also associated with significantly greater ST diffusion throughout tumor tissue ( Figure 3D, Figure S1B). ST colonization and reduction in collagen content were not observed in healthy tissue such as the skin ( Figure 3E) and joints ( Figure S1C) under inducing conditions. These data suggest that ST-SOT effectively colonizes PDAC tissue, whether located o.t. or s.c., and degrades tumor-associated collagens that allow for greater influx of large molecular weight objects, such as ST, from the bloodstream.

ST-SOT Treatment Reduces Frequency of Suppressive Intratumoral Immune Subsets
Significantly high collagen content is known to increase intratumoral frequencies of suppressive immune subsets that blunt anti-tumor responses [19,40]. To determine the effects of reducing collagen content in PDAC tissue, we evaluated intratumoral immune subsets following ST-SOT treatment by flow cytometry in mice bearing Pan02 tumors. From our initial gating of total CD45 + cells, we immediately observed dramatic decreases in CD3 + T cells following induction ( Figure 4A, p<0.05). Within this CD3 + population, CD4 + T cells were significantly decreased in induced mice (p<0.05), compared to uninduced, specifically those co-expressing PD-L1 + (Figure 4B, p<0.0001), which are known to induce apoptosis or anergy in effector T-cells [41]. Moreover, induced ST-SOT treatment was shown to decrease the intratumoral frequency of PD-1-expressing CD8 + T cells, macrophages (F4/80 + CD11b + ), and dendritic cells (CD11b + CD11c + ) ( Figure   4C, p<0.001). Modulation of these specific immune subsets by collagen is consistent with previous studies [19,42]. Overall, these results suggest that ST-SOT treatment can modulate the intratumoral landscape to produce a microenvironment more conducive to immunotherapy. Additional analyses of CD3 + immune subsets were performed to quantify frequency of (B) CD4 + and PD-L1 + CD4 + cells.

ST-SOT Treatment Augments Anti-Tumor Efficacy of Immune Checkpoint Blockade (ICB) Therapy
To determine the therapeutic effects of ST-SOT treatment on tumor cell growth, we performed a series of in vitro and in vivo studies. Tumor growth control by ST alone is dependent on the presence of innate immune subsets, such as neutrophils and macrophages, that exert both bacteriocidal and cytolytic activity [43,44].
However, engineered strains of attentuated ST may also exert anti-tumor activity when cultured directly with tumor cells that is independent of media depletion or pH change, which can be observed in the first 24 hours of co-incubation [45]. Whereas co-culturing with ST-SOT did not affect the overall viablity of Pan02 tumor cells under inducing conditions (Figure 5A), a significant decrease in growth kinetics was observed over 24 hours compared to tumor cells cultured under uninduced conditions or only L-arabinose (no ST treatment) ( Figure   5B). These results suggest that SOT expressed by ST may delay tumor cell growth in culture, possibly through a mechanism involving collagen degradation [46].  the I+IgG group but not U+IgG and U+ICB groups (Figure 5E, p<0.05). Of note, larger tumors, such as those from the U+IgG group may generally show increased expression of cleaved caspase-3 in tumor centers due to hypoxia and lack of blood flow, compared to smaller tumors. Thus, the increased cleaved caspase 3 in the I+ICB group compared to similarly sized (small) tumors (I+IgG) is likely due to direct action of the treatment.
Taken together, these data indicate that tumors from the I+ICB group demonstrate significantly decreased cell proliferation and/or increased apoptosis compared to tumors from all other groups.

Combination ST-SOT with ICB therapy prevents increases in collagen deposition and checkpoint expression concurrent with ICB therapy alone
In

Discussion
Therapeutic resistance continues to be a major factor contributing to poor survival rates in pancreatic cancer. While many anticancer drugs prove potent in vitro, many have failed in the clinic because they are unable to penetrate tumor tissue in sufficient amounts to be therapeutic while also averting adverse effects.
Eliminating tumor ECM components, such as collagen, is hypothesized to improve anticancer drug delivery by decreasing solid stresses and normalizing tissue vasculature [50]. Various preclinical studies utilizing collagen-targeted therapies have demonstrated promising outcomes but remain controversial due to the abundance of collagens throughout the body and lack of tumor-specificity associated with these approaches [20,51,52]. In this study, we engineered tumor-targeting ST to express SOT under an inducible promoter in order to provide optimal tumor-targeted degradation of collagen. Under inducing conditions, the SOT enzyme was found to anchor to the bacterial surface, minimizing potential for secretion into systemic circulation. We found that induction of tumor-colonizing ST-SOT resulted in significant reduction of collagen content and greater bacterial diffusion. Interestingly, significant reduction of suppressive intratumoral subsets was observed Previous studies have shown that a high-density collagen matrix within tumors is associated with decreased cytotoxic T cell abundance and increased regulatory T cell infiltration [53][54][55]. In line with this, collagen molecules are known to act as a ligand for the inhibitory leukocyte-associated immunoglobulin-like receptor 1 (LAIR-1), which is encoded by NK, T and B cells [56,57]. Signaling through this receptor induces T cell exhaustion and increases resistance to ICB [19,58]. ST-SOT pre-treatment likely interferes with this ligand-receptor interaction, sensitizing tumors to ICB treatment. Interestingly, in this study we observed a 2-fold increase in collagen content by 48 hours after ICB treatment, which has been observed previously in a pre-clinical colon cancer model [49]. The combination of ICB with induction of ST-SOT in our model prevented this increase and synergized to decrease collagen in the stroma greater than induction with IgG, possibly implying that IgG treatment increases collagen somewhat similarly to ICB treatment. Regardless, increased collagen after an initial treatment with ICB may decrease the diffusion of further ICB treatments to the tumor, consistent with a more recently described mechanism of acquired resistance to ICB [19].
Fifty percent of PDAC diagnoses occur at late-stage, and nearly all histopathologic analyses of primary tumor and metastases show extensive desmoplasia [59]. However, the abundance of each ECM component contributing to desmoplasia is not always uniform. Collagen imparts mechanical stresses that act in concert with other major ECM components, such as hyaluronan, to limit tumor perfusion [60,61]. Whereas an overabundance of hyaluronan contributes to increased interstitial fluidic pressure leading to vessel compression, collagen fibers impart rigidity to maintain tissue-level compression that prevents stromal collapse and, in turn, vessel decompression. Thus, strategies employing only individual depletion of hyaluronan or collagen, even if done effectively, may not result in maximum tumor permeability. A combinatorial strategy to target both hyaluronan and collagen simultaneously not only presents an approach to maximize drug delivery in PDAC with broader applicability, but also presents greater risk of adverse events. To overcome this possibility, we have also engineered recombinant ST expressing bacterial hyaluronidase (bHs-ST) [25]. A mixture containing both ST-SOT and bHs-ST is predicted to simultaneously deplete tumor-associated collagen and hyaluronan with minimal off-tumor toxicity. This is in contrast to previously tested small-molecule inhibitors, which inadvertently increase tumor aggressiveness and metastasis as a result of targeting signaling pathways involved in collagen and hyaluronan synthesis [62,63]. Our ST-based platform allows for safe targeting of both components for the first time and will require additional studies to determine the added benefits of dual versus single depletions as well as effects on metastatic potential, which remains a major concern with ECM remodeling. In addition to PDAC, our findings may have broader application to other desmoplastic tumor types such as those originating in the breast and lung [64]. containing 10% FBS, 2mM L-glutamine and pen/strep. KPC4662.5 cells, prior to orthotopic implantation (survival surgery) into C57BL/6 mice, were passaged ≤5 times and maintained at ≤80% confluency in DMEM containing 10% FBS, 2mM L-glutamine and pen/strep.

ST strains and generation of ST-SOT
YS1646 (ST) was obtained from ATCC® (202165™) and cultured in modified LB media containing MgSO4 and CaCl2 (LB-0) in place of NaCl. The SOT amino acid sequence (GenBank Accession no. KP313606) fused to Fla (N-terminus) and 6xHis-tag (C-terminus) was used to synthesize a codon-optimized cDNA (Biomatik) inserted in-frame into a pBAD bacterial expression vector (kind gift from Michael Davidson, Addgene #54762) to generate pBAD-SOT. ST-LUX was generated using the pAKlux2 plasmid (pAKlux2 was a gift from Attila Karsi, Addgene #14080). Plasmids were electroporated into YS1646 using a BTX electroporator (1mm gap cuvettes, settings: 1.8kV, 186 ohms), spread onto LB-0 plates containing 100 µg/mL ampicillin and incubated overnight at 37C. Glycerol stocks were generated for pBAD-SOT positive clones identified by colony PCR and restriction digest of plasmid preparations.

Bacterial growth, viability, and analysis of SOT expression
ST clones electroporated with pBAD-SOT were cultured in media with or without 2% (w/v) L-arabinose at  incisions were closed using absorbable sutures and staples, respectively. Analgesics were administered pre-and post-surgery. Pan02 cells (2x10 5 ) were implanted subcutaneously above the right flank.

Quantification of collagen degradation
Paraffin-embedded tumors were sectioned and stained using Masson's Trichrome. Whole tumors were imaged using the Zeiss Axio Observer II microscope (Carl Zeiss Inc.; White Plains, NY) and trichrome images were color deconvoluted using the "Masson's Trichrome" setting in ImageJ (U. S. National Institutes of Health; Bethesda, Maryland). Thresholds were set for individual channels just under background and quantified using the "analyze particles" option under "measure." Total collagen area (blue channel) was divided by the addition of collagen area plus total cytoplasm area (red channel) to obtain the percentage of collagen. Technologies). DAB staining out of total nuclei per field was done using ImageJ (NIH). DAB and hematoxylin channels were separated using color deconvolution (H-DAB preset), and thresholds were set to cover positive staining area for DAB (brown) and positive staining nuclei for hematoxylin (blue) just under background.

Immunohistochemistry
DAB-positive nuclei were quantified by dividing DAB threshold area by DAB threshold area plus hematoxylin threshold area.

Flow cytometry
One million live cells were counted using trypan blue and first stained with a fixable viability dye (eBiosciences

Tumor Growth Measurements and Treatment
For the Pan02 subcutaneous tumor model, tumors were allowed to grow for two weeks until reaching an Tumor volumes were measured thrice weekly using digital calipers. Tumor growth is represented as fold change in growth compared to volume at initial ST-SOT treatment.

Statistics
All statistical analyses were done using the Prism software by GraphPad (V8). Unless otherwise noted in figure legends, statistics were obtained by performing a two-way ANOVA followed by Tukey's multiple comparisons test.

Conclusions
Whereas cancer treatments continue to improve at a rapid pace, survival rates for PDAC patients have, unfortunately, not followed suit. In the case of PDAC, it has become increasingly clear that combinatorial strategies to eliminate tumor fibrosis may be requisite for therapeutic success. Increasing evidence also suggests that reducing levels of extracellular matrix (ECM) components not only relieves interstitial pressures to enhance drug delivery, but may also modulate tumor-associated immune subsets to generate a tumor microenvironment more conducive to immunotherapy. Indeed, we and others have observed dramatic improvements in immunotherapy following direct depletion of hyaluronan and collagen. One major hurdle in targeting ECM components has been major adverse events associated with systemic, off-tumor toxicity.
Incorporation of degradative enzymes into ST vectors represents a novel strategy to restrict ECM degradation to tumor tissue.

Supplementary Materials:
The following are available online at www.mdpi.com/xxx/s1, Figure S1: ST colonization and diffusion.