The right microbial stimulus can direct innate immune effector cells to specific organ sites to clear pathology

Recent developments in understanding how the functional phenotype of the innate immune system is programmed has led to paradigm-shifting views on immunomodulation. These advances have overturned two long-held dogmas: only adaptive immunity confers immunological memory and innate immunity lacks specificity. This work describes the novel observation that innate immune effector cells can be recruited to specific tissues of the body where pathology is present by using a microbial-based immune stimulus that consists of an inactivated pathogen that typically resides or causes infection in that target tissue site. We demonstrate this principle using experimental models of cancer and infection for which different subcutaneously delivered microbial-based treatments were shown to induce the recruitment of immune effector cells to specific diseased organs. Amelioration of disease in a given organ niche was dependent on matching the correct microbial stimulus for the affected organ site but was independent of the nature of the pathology. This observation intriguingly suggests that the immune system, upon pathogen recognition, tends to direct its resources to the compartment in which the pathogen has previously been encountered and would be the most likely source of infection. Importantly, this phenomenon provides a novel means to therapeutically target innate immune effector cells to sites of specific disease localization to potentially treat a wide spectrum of pathologies, including cancer, infection, and chronic inflammatory disorders. AUTHOR SUMMARY Vaccines that target adaptive immune memory have revolutionized medicine. This study describes a novel strategy that works as a modified innate immune “vaccine” that exploits the trained response of innate immune effector cells to clear pathology in a specific tissue site. Unlike memory of the adaptive immune system, which functions like a lock and key, innate immune memory is more akin to a reflex response – like experienced muscle or neural cells that are changed by a stimulus to respond more efficiently upon re-exposure. This change in behavior through experience is the definition of learning. Our study suggests that this innate immune learning occurs at different levels. Emergency hematopoiesis trains new innate immune cells in the bone marrow to respond quickly and effectively to a non-specific threat; whereas, pathogen-specific training occurs at sites where cells making up the immunologic niche have had interactions with a particular pathogen and have been trained to respond more robustly to it upon re-presentation in the context of a danger signal. The speed with which new immune cells are trained in the bone marrow in response to an imminent microbial threat and their subsequent recruitment to the target organ site where that microbe typically resides suggests there are ways the immune system communicates to coordinate this rapid response that are yet to be fully delineated. These findings provide a novel highly proficient way to harness the potent effector functions of the innate immune system to address a wide range of immune-based diseases.


INTRODUCTION
In 1899, cancer researcher D'Arcy Power noted: "Where malaria is common, cancer is rare." [1] 64 For centuries, physicians have observed that acute infection can be associated with spontaneous 65 cancer regression, and it is now well appreciated that the potent immune response to the threat of 66 acute infection can overcome malignancy [2,3]. One of the best known early clinical applications 67 of this observation was developed in the 1890's by Dr. William Coley who used an inactivated 68 bacterial cocktail, which came to be known as Coley's Toxin, to treat various types of cancers and 69 achieved success rates similar to modern treatments in many cases [2]. At least 15 epidemiological 70 studies have examined the possible link between infectious disease and cancer, with all but one 71 supporting an association between infection and reduced incidence of neoplasms [2]. Yet, intra-72 vesical administration of Bacillus Galmette-Guerin (BCG) for the treatment of high-risk non-73 muscle invasive bladder cancer is the only microbe-based therapy that is currently part of the 74 approved standard of care [4]. The failure to realize the full potential of this immunotherapeutic 75 approach can be attributed to a number of factors, including lack of sufficient characterization of 76 the immunological mechanism(s) driving the therapeutic outcome and difficulties with obtaining 77 consistent results. However, recent pivotal developments in understanding the basis of what is now 78 coined trained innate immunity, which encompasses changes in the programming of the innate 79 immune response through experience, have far-reaching potential to improve our understanding 80 of the anti-cancer immunological response induced by acute infection [5,6]. This new deeper 81 appreciation of innate immune system adaptation and reprogramming is projected to meaningfully 82 advance the use of microbial-based treatments for not only cancer, but also for many of the 83 increasingly prevalent immune-based diseases we now contend with, including inflammatory 84 bowel disease, allergies and autoimmune disorders [5][6][7]. 5 85 We recently characterized the key immune cellular and molecular pathways by which mimicking 86 an acute infection using a microbe-based stimulus markedly reduces the tumor burden in two 87 different lung cancer models [8]. The treatment used was formulated from an inactivated lung 88 pathogen, a derivative of a clinical isolate of Klebsiella, which is subcutaneously administered 89 every second day [8]. Efficacy in reducing lung tumor burden was shown to be 1) dependent on 90 the animals having had previous exposure to Klebsiella and 2) independent of adaptive immunity 91 [8]. In the work presented here, we describe a newly discovered phenomenon of organ-specific 92 trained innate immune cell recruitment and disease amelioration. We show organ specificity is 93 determined by the organ niche of the bacterial species from which the microbial stimulant is 94 derived. These findings suggest that there are levels of greater complexity of innate immune 95 training that are yet to be fully elucidated. By applying this immunological targeting 96 therapeutically, these findings provide a means to direct trained innate immune effector cells to 97 specific tissue sites of pathology. Klebsiella, called QBKPN, was previously tested in a Lewis Lung Carcinoma model [8]. It was 104 observed that mice without prior lung exposure to Klebsiella did not experience a reduction in 105 tumor burden with QBKPN treatment, whereas mice with prior exposure to Klebsiella did [8]. 106 Here we show that when we administer a similarly formulated immunotherapy based on 6 107 Escherichia coli, called QBECO, there is little relative efficacy in reducing lung cancer burden in 108 the Lewis Lung Carcinoma model, despite all animals likely having exposure to E. coli (Fig. 1A). 109 We replicated the previous findings in a different lung cancer model using B16F10 melanoma cells 110 (Supplemental Fig. 1A), which suggests QBKPN's anti-cancer efficacy is not cancer cell type 111 specific, rather it appears to be applicable to different cancer cell types growing in the lungs. We 112 hypothesized that QBECO may have a therapeutic anti-tumor effect in a compartment likely to be 113 exposed to E. coli. We first used an intraperitoneal cancer model since E. coli infection is the most anti-cancer efficacy in the lungs. Collectively, these findings suggest there are organ-specific 122 immune effects of these microbial treatments and that these subcutaneously administered products 123 are sampled by immune cells in the targeted organ niche. To exclude the possibility that the 124 efficacy of these microbial-based treatments was a result solely of reduced seeding of injected 125 cancer cells, we tested both QBKPN and QBECO in a chemically-induced colon cancer model in 126 which cancer develops with exposure to DSS/AOM over a 70 day period [11,12]. Because both 127 Klebsiella species and E. coli are enteric bacteria in mice, both QBKPN and QBECO treatment 128 succeeded in reducing tumor burden in the colon compared to vehicle-treated mice in this model 129 (Fig. 1C). A biodistribution study using in vivo imaging of fluorescently labelled QBKPN (Cy5.5-7 130 QBKPN) confirmed that the product is systemically distributed (Supplemental Fig. 2A); peak 131 levels in blood were detected at 2 hours after administration and dropped subsequently 132 (Supplemental Fig. 2B).

133
Characterizing the lung-specific immune response induced by QBKPN compared to QBECO 134 in mice with lung cancer 135 Immuno-phenotyping was used to characterize the lung specific immune changes induced by 136 QBKPN compared to QBECO treatment using the murine B16F10 lung cancer model. QBKPN 137 administration increased the percentage of NK cells and interstitial macrophages in the lungs when 138 assessed at both days 5 and 17 after tumor inoculation (corresponding to 15 and 27 days of QBKPN 139 treatment, respectively). In contrast, QBECO treatment did not change the proportion of either NK 140 cells or interstitial macrophages in the lungs relative to vehicle treated control mice ( Fig. 2A-B).

141
To account for the greater recruitment of NK cells and macrophages to the lungs by QBKPN vs.

142
QBECO treatment, lung chemokine concentrations were quantified. Relative to both vehicle 143 control and QBECO, treatment with QBKPN led to higher levels of chemokines important for the 144 recruitment and activation of NK cells and macrophages, including CCL2, CCL5, and IFNγ-145 inducible chemokines, CXCL9 and CXCL10, at both 5 and 17 days following tumor inoculation 146 (Fig. 2C). IFNγ gene expression was increased with QBKPN treatment compared to QBECO at 147 both timepoints measured (Fig. 2D); however, protein levels of IFNγ were measurably greater at 148 only the day 5 timepoint (Supplemental Fig. 3). QBKPN's anti-lung cancer efficacy was 149 previously found to be dependent on the activation of the NKG2D pathway, which is fundamental The contribution of the presence of a lung pathology to the type and extent of the immune response 159 triggered by QBKPN vs. QBECO was investigated using cancer-free mice. Among the most 160 notable differences in the immune response in healthy animals was the proportion of NK cells 161 recruited to the lungs with QBKPN treatment, which was markedly attenuated in healthy mice 162 compared to lung cancer bearing mice (Fig. 4A vs. Fig. 2A). In contrast, the proportion of 163 interstitial macrophages increased in both healthy mice and mice with cancer ( Fig. 4B, Fig. 2B).

164
Another difference in healthy vs. lung cancer bearing mice was the degree to which certain immune 165 responses were sustained. In particular, the expression of IFNγ and Rae1, which decreased or 166 remained the same over time in healthy mice (Fig. 4), increased over time in cancer-bearing mice 167 (Fig. 3). However, similar to cancer-bearing mice, there was greater lung immune stimulation with 168 QBKPN than with QBECO treatment, but this difference was generally less pronounced in healthy 169 mice. The exception to this observation was in the gene expression, but not protein levels, of 170 granzyme B, granzyme A, and perforin, which showed equal increases with 27 days of either 171 QBKPN or QBECO treatment in disease-free animals (Fig. 4G). Collectively, these findings 172 suggest that the presence of pathology dictates the extent, duration and type of lung-specific

Demonstration of the broader application of immune modulation by organ-specific
223 microbial-based therapies to treat immune-related diseases 224 The ability to target trained innate immune cells to specific organ sites of pathology by the strategic 225 selection of the correctly matched microbial-based therapy has potential to treat a broad range of 226 immune-related diseases. Adding to the evidence provided for this novel approach for cancer using 227 specific gram-negative pathogens targeting the lungs, peritoneal cavity and GI tract, we next tested microbial components as potential new treatment modalities for the many serious immune-related 265 diseases growing in incidence and prevalence [5,24,25]. In this work, we describe what we believe 266 to be a seemingly unknown phenomenon of the existence of immunological memory contained 267 within specific regional niches for microbes that normally inhabit or infect it. We were able to  Two long-held dogmas have recently been brought into question with paradigm-shifting advances 282 in our understanding of innate immunity: 1) immunological memory is the domain of the adaptive 283 arm of the immune system, and 2) innate immunity lacks specific memory [5,6,26,27]. The 284 treatment strategy we employed is rooted in both the trainability of the innate immune system and 285 its more nuanced organ specificity, which is intrinsically different from that of the adaptive by which this microbial regulation is instituted was not defined.

317
It makes intuitive sense that there exists a means by which the innate immune system directs its 318 resources to the most likely source of infection upon detecting microbial components that are 319 presented as an acute threat. We propose this organ-specific microbial recognition reflects a more 320 sophisticated level of immunological education than the non-specific training that initially occurs 321 in the bone marrow. It is reminiscent of a similar phenomenon observed in invertebrates, which 322 lack adaptive immunity, yet show a stronger immune response to bacterial strains previously 323 encountered compared to strains to which they are naïve [26]. It was previously demonstrated that 324 a Klebsiella-based immunotherapy was effective in reducing lung cancer burden in Rag2 deficient 325 mice that are largely devoid of classical adaptive immune function [8]. The contribution of 326 adaptive immunity to the observed microbial mediated organ-specificity cannot be ruled out; 327 however, we believe it is unlikely the primary driver of this organ specific memory. Cells not 328 conventionally considered to be part of the immune system, such as epithelial and endothelial cells 329 that function as the interface between the host and its microbiota, are also capable of forming long-