Anti Tumor Immunity Can Be Uncoupled From AutoimmunityFollowing Hsp70-Mediated Inflammatory Killing Of NormalPancreas1

Timothy Kottke1,*, Jose Pulido1,2,*, Jill Thompson1, Luis Sanchez-Perez1,3,#, HeungChong4, Stuart K. Calderwood5, Peter Selby7, Kevin Harrington6, Alan Melcher7, andRichard Vile1,3,71Department of Molecular Medicine, Mayo Clinic, Rochester, MN 559052Department of Ophthalmology and Ocular Oncology, Mayo Clinic, Rochester, MN 559053Department of Immunology, Mayo Clinic, Rochester, MN 559054St George's Hospital Medical School, Tooting, London, UK5Harvard Medical School, Beth Israel Deaconess Medical Center, Boston, MA 022156Institute of Cancer Research, 237 Fulham Road, London, SW37Cancer Research UK Clinical Centre, St. James' University Hospital, Leeds, UK

AbstractWe have a long term interest in the connectivity between autoimmunity and tumor rejection.However, outside of the melanocyte/melanoma paradigm, little is known about whetherautoimmune responses to normal tissue can induce rejection of tumors of the same histologicaltype. Here, we induced direct, pathogen-like cytotoxicity to the normal pancreas in associationwith the immune adjuvant hsp70. In sharp contrast to our studies with a similar approach for thetreatment of prostate cancer, inflammatory killing of the normal pancreas induced a Th-1-like,anti-self response to pancreatic antigens which was rapidly suppressed by a concomitantsuppressive regulatory T cell (Treg) response. Interestingly, even when Treg cells were depleted,the Th-1-like response was insufficient to induce significant ongoing autoimmunity. However, theTh-1-like response to antigens expressed in the pancreas at the time of damage was sufficient toinduce rejection of tumors expressing either a foreign (ova) antigen, or fully syngeneic tumorantigens (on PancO2 tumor cells), provided that Treg were depleted prior to inflammatory killingof the normal pancreas. Taken together, these data indicate that profound differences existbetween the immunoprotective mechanisms in place between different tissues (pancreas andprostate) in their response to pathogen-like damage. Moreover, they also show that, althoughmultiple layers of immunological safeguards are in place to prevent the development of severeautoimmune consequences in the pancreas (in contrast to the prostate), tumor rejection responsescan still be de-coupled from pathological autoimmune responses in vivo, which may provide novelinsights into the immunotherapeutic treatment of pancreatic cancer.

1This work was supported by the Mayo Foundation, by NIH Grants RO1CA085931 and 1RO1CA132734, by the Richard M. SchulzeFamily Foundation, by Cancer Research UK and by an unrestricted grant from Research to Prevent Blindness Inc., NYAddress correspondence to Dr. Richard G. Vile / Dept Molecular Medicine / Guggenheim 1836 / Mayo Clinic / 200 1st Street SW,Rochester, Minnesota, 55902, USA; Tel: 507 284 9941; Fax: 507 266 2122; vile.richard@mayo.edu.*These authors contributed equally to the study.#Current address: The Robert Preston Tisch Brain Tumor Center at Duke, Division of Neurosurgery, Department of Surgery, 491Research Drive, Sands Bldg Room 210, Durham, NC 27710

NIH Public AccessAuthor ManuscriptCancer Res. Author manuscript; available in PMC 2011 March 1.

Published in final edited form as:Cancer Res. 2009 October 1; 69(19): 7767–7774. doi:10.1158/0008-5472.CAN-09-1597.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Keywordshsp70; Fusogenic membrane glycoproteins; immunotherapy; tumor antigens; autoimmunity

IntroductionWe have hypothesized that the connectivity between autoimmune reactivity and anti tumorimmunity might be exploited as an immunotherapeutic approach for treatment of tumors(1-6). This stems from multiple reports of a close association between successful inductionof tumor immunity with emergence of concomitant autoimmunity in melanoma, wheremany melanoma antigens are unaltered self proteins of melanocytes(7,8).

As a result, we developed a novel approach to the immunotherapy of cancer by targetingnormal tissues in situ as a source of immunogen(5). We hypothesized that by inducing‘stressful death’ of normal cells (melanocytes) it would be possible to generate autoimmuneresponses effective against tumor cells (melanoma) which share antigens with the normaltissue(7). In this respect, we, and others, have shown that hsp70 converts tolerogenic antigenpresentation into immunostimulatory presentation to break tolerance by acting as a keymediator of immunogenic cell death (9-12). We went on to show that killing normalmelanocytes, in the presence of hsp70 (inflammatory killing) generates T cell reactivityagainst established melanomas. Thus, intradermal plasmid DNA injections of atranscriptionally targeted cytotoxic gene, along with a plasmid expressing hsp70 (9),induced direct in vivo, inflammatory killing of normal melanocytes, broke tolerance to selfantigens and led to rejection of established melanomas (1,3,4). Progressive autoimmunedisease was inhibited by regulatory T cells (Treg) (1,4). However, co-expression of CD40Lincreased anti tumor efficacy and now induced aggressive autoimmunity (3).

Therefore, self-reactive T cells which have escaped thymic deletion can, if correctlyactivated, kill both melanoma cells and normal melanocytes (7,8). Recently, we investigatedwhether a similar connectivity exists between autoimmune responses to the prostate andrejection of prostate tumors(2). Previously, we had shown that (tumor) cell killing using aviral fusogenic membrane glycoprotein (FMG) can be potently immunogenic through fusionof cells into syncytia(13,14), and we used an adenoviral vector expressing VSV-G, the FMGfrom Vesicular Stomatitis Virus (VSV)(14), to kill normal prostate cells(2). Inflammatorykilling of normal prostate tissue in situ, using vectors expressing VSV-G and hsp70, inducedIL-6, which triggered a CD4/CD8-dependent autoimmunity, associated with IL-17. Thisautoimmune response induced rejection of established prostate tumors, but not otherhistological types of tumors(2). Therefore, an intimate connectivity between autoimmune,and tumor rejection, responses extends beyond the classical melanoma paradigm.

The loss of melanocytes, and normal prostate tissue, can be tolerated considerably betterthan metastatic melanoma or prostate cancer (8). We were interested, therefore, in whetherthe immunological safeguards that control autoimmunity following pathological damage to atissue/organ might be more rigorous when damage to the tissue has more acute physiologicalconsequences (15). Here, we investigated the link between inflammatory killing of normalpancreas, the development of autoimmunity and the link with immunity against pancreatictumors. We show for the first time that prostate and pancreas tissues respond verydifferently to exactly the same pathogen like damage; that this has profound consequencesfor the development of autoimmune disease associated with these different tissues and thatthe principle of inflammatory killing of normal cells to treat neoplastic disease can beextended to pancreatic cancer as well as melanoma and prostate cancer.

Kottke et al. Page 2

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Materials and MethodsCell lines, plasmids and viruses

Mouse ductal pancreatic adenocarcinoma Panc02 cells (16) were a kind gift from Dr PinkuMukherjee (Mayo Clinic, Scottsdale). Murine melanoma B16.F1, cells have beendescribed(14). B16ova cells (H2-Kb) were derived from B16 by transduction with thechicken ovalbumin gene (14). Cell lines were grown in Dulbecco's modified Eagle'sminimal essential medium (Life Technologies, Maryland) supplemented with 10% (v/v)fetal calf serum (Life Technologies) and L-glutamine (Life Technologies). All cell lineswere monitored routinely and free of Mycoplasma.

Replication defective adenoviral vectors were all E1 deleted, serotype 5 vectors with thecytomegalovirus (CMV) immediate-early gene promoter-enhancer driving the transgene.Ad-VSV-G expresses the G glycoprotein of Vesicular Stomatitis Virus (VSV-G) (13,14);Ad-hsp70 contains the inducible murine heat shock protein 70 gene (17); and Ad-GFPcontains the Green Fluorescent protein gene(18).

In the CMV-ova plasmid, the ovalbumin gene (14) is driven by the CMV promoter inpCR3.1 (Invitrogen).

Histopathology of tumor sectionsOrgans were fixed in 10% Formalin in PBS, paraffin embedded and sectioned. H&E-stainedsections were prepared and examined by two independent pathologists blinded toexperimental design.

Reverse Transcriptase Polymerase Chain ReactionOrgans were snap frozen in liquid nitrogen. RNA was prepared with QIAGEN RNAextraction kits. 1μg total cellular RNA was reverse transcribed in 20μl using oligo-(dT) asprimer. A cDNA equivalent of 1ng RNA was amplified by PCR as described (14,19) (detailsof primers upon request).

Hsp70 treatment of explanted tissuesProstates or pancreas from three mice were recovered, weighed and two blocks of 30mg permouse were dissociated separately and allowed to settle overnight. The following day, oneof the two cultures from each animal was incubated with recombinant murine hsp70 (Sigma,St Louis) (10μg/ml) for 24 hours. The second duplicate culture was treated with BSA. 24 hrslater, supernatants were assayed for IL-6 by ELISA.

Treg-mediated inhibition of IFN-γ secretion from activated T cellsT cells of OT-I transgenic mice express the Vα2 chain of the OT-I T cell receptor whichrecognizes the SIINFEKL peptide from the chicken ovalbumin protein in the context ofH-2Kb as expressed by B16ova cells(20). Spleen and lymph nodes from OT-I-transgenicmice were crushed through a 100-μm filter to prepare a single-cell suspension. RBC wereremoved by a 2-min incubation in ACK buffer (sterile dH2O containing 0.15 M NH4Cl, 1.0mM KHCO3, and 0.1 mM EDTA, pH 7.2–7.4). To activate OT-I, single cell suspensionsfrom spleen and lymph nodes were adjusted to 1.0×106 cells/ml in IMDM plus 5% FCS,10−5 M 2-ME, 100 U/ml penicillin, and 100 μg/ml streptomycin and stimulated with 1μg/mlSIINFEKL peptide and 50 IU/ml human IL-2 (Mayo Clinic Pharmacy).

To assay for T cell suppressive (Treg) activity, 250,000 freshly harvested splenocytes fromtreatment groups were plated in triplicate with 105 naive OT-I CD8+ve T cells with eitherno added peptide, an irrelevant non-activating peptide (TRP-2 180-188 SVYDFFVWL(21)) or

Kottke et al. Page 3

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

with synthetic, H-2Kb-restricted ova peptide SIINFEKL(20). Supernatants were assayed forIFN-γ by ELISA. The degree of suppressive activity in splenocyte cultures is reflected bytheir ability to inhibit IFN-γ responses of naïve OT-I T cells when presented with activatingSIINFEKL. Dependence of T cell suppression TGF-β(22) was assayed using recombinanthuman TGF-β sRII/Fc chimera (R&D Systems), a 159 amino acid extracellular domain ofhuman TGF-β Receptor Type II fused to the Fc region of human IgG1.

Antigen priming assays - Splenocyte preparation and antigen presentationSplenocytes enriched in lymphocytes were prepared by standard techniques(23). 500,000freshly purified splenocytes were incubated either with tissue lysates, tumor cells (at ratiosof 100:1; 10;1 or 1:1), or were pulsed with 1 μg/ml of target peptide (SIINFEKL forresponses to ova(20) or TRP-2 180-188 SVYDFFVWL(21) as irrelevant antigen control).48-72 hours later supernatants were tested for IFN-γ by ELISA (Pharmingen). TRP-2 180-188SVYDFFVWL(21) and Ova SIINFEKL(20) were synthesized at the Mayo Foundation Corefacility. To test for specificity against normal pancreatic antigens, splenocytes wereincubated every 48 hrs for 6 days with 100μl of freeze/thaw lysates of normal pancreas, orprostate (specificity control), previously dissociated in vitro and supernatants tested for IFN-γ.

ELISA analysisSupernatants were tested by specific ELISA for IFN-γ or IL-6 (BD Biosciences) accordingto the manufacturers' instructions.

In Vivo StudiesAll procedures were approved by the Mayo Foundation Institutional Animal Care and UseCommittee. C57Bl/6 mice were purchased from Jackson Laboratories at 6-8 weeks of age.To establish subcutaneous tumors, 2×105 B16 or B16ova cells, or 5×105 Panc02 cells, in100μl PBS were injected into the flank of mice. Intra- pancreatic injections (50μl), wereperformed under anesthetic. For survival, tumor diameter in two dimensions, was measuredthree times weekly using calipers and mice were killed when tumor size was approximately1.0×1.0 cm in two perpendicular directions.

For Treg depletion, 0.5mg of PC-61 antibody (Monoclonal Antibody Core Facility, MayoClinic) was given intra-peritoneally 2 and 1 days before the first viral injection. FACSanalysis of spleens and lymph nodes confirmed subset specific depletions.

StatisticsSurvival data was analyzed using the logrank test (24), and the two-sample unequal variancestudent's t test analysis was applied for in vitro assays. Statistical significance wasdetermined at the level of p<0.05.

ResultsInflammatory killing of normal pancreas does not induce autoimmunity

When delivered intra-prostatically, the combination of Ad-VSV-G and Ad-hsp70 causedsevere infiltration, necrosis and tissue destruction (Figs.1A-C) (2). This inflammation waspersistent (>3 weeks) and was associated with ongoing autoimmune destruction of theprostate, as reflected by a progressive decrease of the wet weight of prostates recoveredfrom treated animals (not shown) (2). In contrast, the effects of injection of Ad-VSV-G+Ad-hsp70 into the pancreas were markedly less severe. Significantly, the only abnormal areasreported histologically up to 3 weeks following injection with Ad-VSV-G+Ad-hsp70 were

Kottke et al. Page 4

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

restricted to the site of injection (Fig.1D); typically, these abnormal areas encompassedatrophic excocrine glands, with scattered necrotic cells, and low levels of lymphocyticinfiltration, along with abnormally small glands in these areas (Fig. 1D). However, even indirectly adjoining areas, the organ was apparently unaffected with normal exocrine glandsand islets (Fig. S1A). Animals injected with any of the Ad-VSV-G, Ad-hsp70 (not shown)or combination (Ad-VSV-G+Ad-hsp70) vectors did not show distress, symptoms of damageto the pancreas, or abnormal blood glucose levels (Fig. S1B/1C), indicating that no chronicautoimmune effects were induced. To discount the possibility that the differences betweenthe effects of Ad-infection of prostate and pancreas could be due to different susceptibilitiesto adenoviral infection between the tissues, we confirmed that transgene expression frominjected organs was comparable using FACs for GFP (not shown) and rtPCR for expressionof the hsp70 gene (Fig. S1D).

Inflammatory killing of pancreas induces a potent Treg responseWe have shown previously that IL-6 is induced both in prostates in situ, and in draining LN,from mice injected with Ad-hsp70 (whether or not Ad-VSV-G or Ad-GFP were alsoinjected) (2). In addition, IL-6 was also produced from explanted/dissociated prostate tissueex vivo when treated with recombinant hsp70 (Fig.2A) (2). In contrast, IL-6 protein was notinduced from three dissociated pancreas treated with recombinant hsp70 at concentrationswhich induced high levels of IL-6 from explanted prostates (Fig.2A). In addition, mRNA forIL-6 was neither detectable in the pancreas in situ (not shown), nor in the LN draining thepancreas (Fig.2B), following direct intra-pancreatic injection with Ad-hsp70 alone, Ad-VSV-G alone, or with Ad-VSV-G+Ad-hsp70. Significantly, however, TGF-β was expressedin the majority of the LN following intra-pancreatic injection with Ad-hsp70, Ad-VSV-G, orthe combination (Fig. 2B).

Since, in the presence of TGF-β, IL-6 acts a critical mediator between the differentiation ofCD4 cells into either Th-17 (IL-6 present) or Treg (IL-6 absent) cells(25-27), we tested forthe generation of Treg responses. Splenocytes from mice injected intra-pancreatically withAd-GFP suppressed IFN-γ secretion from activated T cells in the presence of their cognateantigen only moderately (Fig. 2C, lanes 1 and 3, p>0.01) or not at all (Fig.2D, Lanes 1&2).However, splenocytes from mice injected intra-pancreatically with Ad-VSV-G+Ad-hsp70almost completely suppressed IFN-γ secretion from activated T cells in vitro (Fig. 2C, lane 2compared to lanes 1 or 3, p<0.001). This suppression was dependent in large part uponTGF-β(22), since addition of 341-BR TGF-β sRII/Fc, which neutralizes secreted TGF-β, tothe splenocyte/T cell co-cultures restored the ability of the T cells to secrete IFN-γ to levelsclose to those seen on co-culture of T cells with splenocytes from un-injected mice (Fig. 2D,lanes 4 and 5, p=0.001). These results contrast sharply with those using splenocytes frommice injected intra-prostatically with Ad-VSV-G+Ad-hsp70, which were unable to exert anysuppression of activated T cells in this assay even when spleens were harvested at differenttimes after prostatic injection (not shown) (2). We also confirmed that IL-17 was notdetectable in pancreas injected with Ad-VSV-G+Ad-hsp70 by rtPCR or ELISA (not shown)and that only background levels of IL-17 protein were observed in the draining LN (notshown). Taken together, these data indicate that inflammatory killing of normal pancreasactively induces a potent TGF-β-dependent Treg response; this contrasts with the sametreatment of normal prostate, following which Treg were absent and an autoimmune Th17response was induced associated with induction of IL-6 (2).

Autoimmune consequences of inflammatory killing of the pancreasSplenocytes recovered from mice treated with Ad-VSV-G+Ad-hsp70 secreted IFN-γ at onlymoderate, but significantly (p=0.05), higher levels than splenocytes from control treatedmice when co-cultured with lysates of normal pancreatic tissue (Fig.3A), although this

Kottke et al. Page 5

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

significance was not reproducible over separate experiments. In contrast, splenocytes frommice treated with PC-61 prior to Ad-VSV-G+Ad-hsp70, which effectively depletes theresultant Treg response in these animals (Fig.2), displayed significantly increased reactivityagainst normal pancreas compared to non-Treg-depleted mice (P<0.01) (Fig.3A), consistentwith the unmasking of a T cell response against normal tissue antigens induced byinflammatory killing of the normal pancreas. These splenocytes did not secrete significantlevels of IFN-γ upon similar co-culture with normal prostate tissue (not shown), indicatingthat inflammatory killing raises organ specific autoimmune T cell responses.

Consistent with the low level of active self reactive T cell response (Fig.3A), and asdescribed in Fig.1, inflammatory killing of the normal pancreas did not invoke histologicalsigns of widespread damage to the organ outside of the site of injection (Figs.1D and S1A).Moreover, these mice did not develop diabetes (Fig.1) nor any other overt signs of toxicityassociated with pancreatitis. Surprisingly, however, despite the presence of circulatingsplenocytes with high specificity for pancreatic antigens (Fig.3A), mice treated with PC-61prior to Ad-VSV-G+Ad-hsp70, also did not develop diabetes (Fig.3B-D). Moreover, thesetreated mice did not have elevated serum levels of amylase over several months ofobservation, indicating that little or no additional immune damage was being done to thepancreas (not shown). In addition, these mice had only very mild histological abnormalities(Fig. S2B-S2C) compared to control treated mice (Fig.S2A) or Ad-VSV-G+Ad-hsp70-treated mice without Treg depletion (not shown). These abnormalities were associated withnon-suppurative, mild focal inflammation and low levels of increased necrosis, which werestill highly localized rather than being pervasive across the whole organ.

These data indicate that hsp70-mediated inflammatory killing of normal pancreas results in aTreg response, which suppresses an IFN-γ/Th-1-like reactivity against normal pancreas-associated antigens, but depletion of which alone is insufficient to unmask significantautoimmunity.

Immunizing potential of inflammatory killing of normal pancreas – foreign antigensWe investigated the ability of inflammatory killing of normal pancreas to support thepriming of T cell responses against a foreign antigen expressed within the pancreaticmicroenvironment using the chicken ovalbumin protein (ova), against which no tolerance isin place in C57Bl/6 mice. When a plasmid expressing OVA was co-injected into thepancreas, along with adenoviral vectors, anti-ova responses were only detected upon Ad-VSV-G+Ad-hsp70 treatment with prior depletion of Treg (Fig.4A; p=0.001 lane 4 comparedto all other groups). Although no tolerance exists in C57Bl/6 mice against OVA, it is stillpossible to grow tumors in which ova is stably expressed. Therefore, we repeated theexperiments of Fig.4A by co-injecting pCMV-ova along with adenoviral vectors into thepancreas of C57Bl/6 mice bearing subcutaneous B16ova tumors. Although these tumors aremelanomas, rather than from pancreatic origin, they stably express the ova antigen which, inthis system, is only expressed in the pancreas at the time of inflammatory killing. Consistentwith the results of Fig.4A, intra-pancreatic injection of pCMV-ova+(Ad-VSV-G+Ad-hsp70)was completely ineffective at inducing rejection of a tumor expressing ova (Fig. 4B).However, if the mice were previously depleted of Treg cells, established ova expressingtumors (B16ova), but not identical tumors lacking expression of the pancreas specificantigen (B16), were now rejected (Fig.4B). These data confirm that inflammatory killing ofpancreas allows priming of rejection responses against foreign antigens present in thepancreatic micro-environment. However, concomitantly primed Treg responses are highlyprotective against effector activity – which includes potential anti-tumor rejection responses- resulting from this priming.

Kottke et al. Page 6

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Immunizing potential of inflammatory killing of normal pancreas – tumor antigensWe are particularly interested in whether the autoimmune response induced by inflammatorykilling of normal cells can be exploited to treat tumors of the same histological type sharingcommon antigens(1-6). Since ova is a foreign antigen in the context of C57Bl/6 mice, wealso investigated whether inflammatory killing of normal pancreas could be exploited as ananti tumor immunotherapy against pancreatic tumors where the tumor cells express only selfantigens to which tolerance is in place in fully syngeneic mice. Direct intra-pancreaticinjection of control adenoviruses (including Ad-hsp70 alone, Ad-VSV-G alone or Ad-GFPalone) was unable to impact on the growth of subcutaneous Panc02 tumors, which aremurine pancreatic cancer cells syngeneic to C57Bl/6 mice, either with (Fig.5A), or without(not shown), Treg depletion compared to mock injected mice. Interestingly, intra-pancreaticinjection of Ad-VSV-G+Ad-hsp70 actually induced subcutaneous Panc02 tumors to growmarginally faster than tumors in mice injected with control treatments (p=0.05) (Fig.5B).We hypothesize that this small, but significant and reproducible, increased rate of tumorgrowth, is attributable the induction of high levels of Treg by the inflammatory killing of thepancreas, which are protective against endogenous immune control of these Panc02 tumors.In contrast, the combination of prior depletion of Treg with intra-pancreatic Ad-VSV-G+Ad-hsp70 induced a potent tumor rejection response in which all Panc02 tumors underwentcomplete regression (Fig. 5C). This rejection response was highly pancreas specific sincemice treated in the same way with PC-61+Ad-VSV-G+Ad-hsp70 were unable to reject s.c.B16 melanoma tumors, or cause any slowing of their growth compared to control treatedmice (not shown). Splenocytes harvested from mice injected with control treatments, or withAd-VSV-G+Ad-hsp70, did not secrete IFN-γ in response to normal, or tumor-associated,pancreatic antigens (Figs.5D and S3A). However, consistent with their ability to mediaterejection of pancreatic, but not non pancreas-derived tumors, splenocytes from mice treatedwith PC-61+Ad-VSV-G+Ad-hsp70 secreted IFN-γ in response to both pancreas antigensexpressed on Panc02 cells, and, to a lesser extent, normal pancreatic tissue (Fig.S3B), butnot in response to antigens expressed on B16 cells or normal prostate tissue (Fig.S3B). In asingle experiment, a tumorigenic dose of Panc02 cells was completely rejected by 4 of 4mice which had previously been cured of pre-established Panc02 tumors by treatment withPC-61/Ad-VSV-G+Ad-hsp70 as in the protocol of Fig.5. In contrast, an additional 3 micecured of Panc02 tumors by the same treatment were unable to reject a re-challenge with B16melanoma tumor cells.

Taken together, these data show that the inflammatory killing of normal pancreas can beexploited as an anti-tumor immunotherapy against pancreatic tumors expressing fullysyngeneic tumor antigens; moreover, tumor rejection responses can be engineered (in thiscase by inflammatory killing combined with Treg depletion) which are not associated withdevelopment of pathological autoimmune consequences.

DiscussionWe show here that an IFN-γ/Th-1-like, anti-self response to normal pancreatic antigens isinduced by inflammatory killing of normal pancreatic tissue (Fig.3A). Unlike the responseto the same pathogen-like insult in the prostate(2), hsp70 expression does not generate IL-6in either the pancreas, or draining LN (Fig.2A,B). The absence of IL-6, in the presence ofTGF-β in these LN (Fig.2B), led instead to potent Treg responses (Fig.2C,D), which restrainthe activity of pancreas-specific splenocytes activated in response to the inflammatorykilling (Fig.3A). We believe that this Treg response is one component of an intrinsicimmunological safeguard, which protects the normal organ from autoimmune damagefollowing pathological insult. Interestingly, however, despite high levels of reactivity againstnormal pancreatic antigens (Fig.3A), even mice depleted of Treg did not developpathological autoimmunity against the pancreas (Fig.3). This would be consistent with a

Kottke et al. Page 7

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

hypothesis that multiple checkpoint controls protect against unrestrained activity ofendogenous T cells activated by pathological damage to an organ as physiologicallyimportant as the pancreas.

We are currently testing the hypothesis that the response to hsp70 in the pancreas ismediated through cross-talk between cells, and/or receptors, which negatively regulate theability of potentially responsive cells to secrete IL-6 within the pancreatic micro-environment and the draining LN(11). In this respect, co-stimulatory molecules onpancreatic cell types may restrict the activity of self reactive T cells which become activatedin vivo. For example, the negative co-stimulator PD-L1(28), plays a pivotal role incontrolling induction of diabetes in prediabetic NOD mice (29)

The major impetus for these studies was our interest in the connectivity between thegeneration of autoimmune responses and anti-tumor immunity(1-6,8). We have previouslyshown that normal prostate expresses antigens which serve as targets for Th17-mediatedtumor rejection following inflammatory killing of normal prostate(2). Here, generation of astrongly suppressed Th-1-like response to normal pancreatic antigens did not, per se, induceautoimmune disease (Fig.3). However, it was sufficient to reject tumors expressing either aforeign (ova) antigen, or fully syngeneic tumor antigens (on Panc02 tumors), provided thatTreg were depleted (Figs.4&5). PC-61/Ad-VSV-G+Ad-hsp70 treatment cured mice ofPanc02 tumors seeded between 1 (Fig.5) to 4 days prior to the first injection of PC-61 (4-7days prior to Ad-VSV-G+Ad-hsp70 intra-pancreatic injections). If treatment was initiatedlater, although significant tumor inhibition could be achieved, fewer than 50% of mice werecured. Tumor rejection was highly tissue type specific. In our previous studies, intra-prostatic injection of Ad-hsp70 alone gave significant therapy compared to Ad-GFP (2). Inthe Panc02 model, we never observed any therapeutic difference between Ad-hsp70 alone,Ad-VSV-G alone or the same dose of Ad-GFP, indicating that both killing of normalpancreas and co-expression of hsp70 are required for anti tumor effects. Moreover, longterm surviving mice which controlled Panc02 tumors, developed no overt signs ofpancreatitis or diabetes. These data suggest, therefore, that it is possible to uncouple antitumor immunity from highly pathological autoimmunity in the context of immunologicalhomeostasis of the pancreas.

A comparison between inflammatory killing in both prostate and pancreas providesimportant insights into the aetiology, and treatment, of infectious, autoimmune, andmalignant disease. The data show that different tissues are protected by different arrays ofimmune checkpoints which must be subverted if an initiating infection is to lead toautoimmunity. Thus, the same pathogenic insult (in this case Ad-VSV-G+Ad-hsp70) canhave very severe long term autoimmune consequences in one organ (prostate) but minimaleffects in another (pancreas). These data also have important implications for understandinghow the site of infection (or vaccination) can determine whether protective, or suppressive,immunity is raised against the immunogen(30). By characterizing cytokine profiles indifferent organs undergoing progressive autoimmune attack, novel therapies (such asblockade of cytokines) to treat such diseases will become apparent.

Finally, a further implication of these data is that fewer safeguards exist which suppress antitumor immune responses (CTL responses which need to be unmasked only by depletion ofTreg) (Fig.5), than are in place to ensure that aggressive autoimmunity does not occurfollowing pathogenic-like insults to the pancreas. Further experiments in transgenic modelsof spontaneous pancreatic cancer (31) will confirm, or refute, this hypothesis. Theseconsiderations suggest that it may be possible to devise a protocol in which a balancebetween anti-tumor activity, and autoimmune severity, can be found using intra-pancreaticinjection and Treg depletion. However, the physiological consequences of allowing

Kottke et al. Page 8

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

aggressive anti-pancreatic immune responses to go unchecked could potentially be verytoxic(15) and further long term toxicology studies are required before these studies are readyfor clinical translation.

In summary, hsp70-mediated inflammatory killing of the pancreas induces an IFN-γ/Th-1-like immune reactivity against the normal tissue which is concomitantly suppressed by apotent Treg response. Activation of Treg by inflammatory killing of the normal pancreasrepresents only one (of probably several) protective responses that ensures that majorautoimmune reactivity does not result from pathological-like damage to this critical organ.In addition, we show that tumor rejection responses can be de-coupled from pathologicalautoimmune responses in vivo, which may provide novel insights into theimmunotherapeutic treatment of pancreatic cancer.

AcknowledgmentsThe authors thank Toni L. Higgins for expert secretarial assistance.

References1. Daniels G, Sanchez-Perez L, Kottke T, et al. A simple method to cure established tumors by

inflammatory killing of normal cells. Nature Biotechnol 2004;22:1125–32. [PubMed: 15300260]2. Kottke T, Sanchez-Perez L, Diaz RM, et al. Induction of hsp70-mediated, Th17 autoimmunity can

be exploited as immunotherapy for metastatic prostate cancer. Cancer Research 2007;67:11970–9.[PubMed: 18089828]

3. Sanchez-Perez L, Kottke T, Daniels G, et al. Killing of normal melanocytes, combined with hsp70and CD40L expression, cures large established melanomas. J Immunol 2006;177:4168–77.[PubMed: 16951382]

4. Sanchez-Perez L, Kottke T, Diaz RM, et al. Potent selection of antigen loss variants of B16melanoma following inflammatory killing of melanocytes in vivo. Can Res 2005;65:2009–17.

5. Ferrone S. Immunotherapy dispenses with tumor antigens. Nat Biotechnol 2004;22:1096–8.[PubMed: 15340476]

6. Calderwood SK. Chaperones and slow death--a recipe for tumor immunotherapy. Trends Biotechnol2005;23:57–9. [PubMed: 15661338]

7. Uchi H, Stan R, Turk MJ, et al. Unraveling the complex relationship between cancer immunity andautoimmunity: lessons from melanoma and vitiligo. Adv Immunol 2006;90:215–41. [PubMed:16730265]

8. Gogas H, Ioannovich J, Dafni U, et al. Prognostic significance of autoimmunity during treatment ofmelanoma with interferon. N Engl J Med 2006;354:758–60. [PubMed: 16481646]

9. Melcher AA, Todryk S, Hardwick N, Ford M, Jacobson M, Vile RG. Tumor immunogenicity isdetermined by the mechanism of cell death via induction of heat shock protein expression. Nat Med1998;4:581–7. [PubMed: 9585232]

10. Millar DG, Garza KM, Odermatt B, et al. Hsp70 promotes antigen-presenting cell function andconverts T-cell tolerance to autoimmunity in vivo. Nat Med 2003;9:1469–76. [PubMed:14625545]

11. Murshid A, Gong J, Calderwood SK. Heat shock proteins in cancer vaccines: agents of antigencross presentation. Expert Rev Vacc 2008;7:1019–30.

12. Binder RJ. Heat shock protein vaccines: from bench to bedside. Int Rev Immunol 2006;25:353–75.[PubMed: 17169780]

13. Bateman A, Harrington K, Kottke T, et al. Viral fusogenic membrane glycoproteins kill solidtumor cells by non-apoptotic mechanisms which promote cross presentation of tumor antigens bydendritic cells. Can Res 2002;62:5466–6578.

14. Linardakis E, Bateman A, Phan V, et al. Enhancing the efficacy of a weak allogeneic melanomavaccine by viral fusogenic membrane glycoprotein-mediated tumor cell-tumor cell fusion. Can Res2002;62:5495–504.

Kottke et al. Page 9

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

15. Finkelberg DL, Sahani D, Deshpande V, Brugge WR. Autoimmune pancreatitis. New Engl J Med2006;355:2670–6. [PubMed: 17182992]

16. Corbett TH, Roberts BJ, Leopold WR, et al. Induction and chemotherapeutic response of twotransplantable ductal adenocarcinomas of the pancreas in C57Bl/6 mice. Cancer Res 1984;44:717–26. [PubMed: 6692374]

17. Melcher A, Murphy S, Vile RG. Heat shock protein expression in target cells infected by poorlydetectable levels of replication competent virus contributes to the immunogenicity of adenoviralvectors. Human Gene Ther 1999;10:1431–42. [PubMed: 10395369]

18. Ahmed A, Thompson J, Emiliusen L, et al. A conditionally replicating adenovirus targeted totumor cells through activated RAS/MAPK-selective mRNA stabilisation. Nat Biotechnol2003;21:771–7. [PubMed: 12794639]

19. Vile RG, Castleden SC, Marshall J, Camplejohn R, Upton C, Chong H. Generation of an anti-tumour immune response in a non-immunogenic tumour: HSVtk-killing in vivo stimulates amononuclear cell infiltrate and a Th1-like profile of intratumoural cytokine expression. Int JCancer 1997;71:267–74. [PubMed: 9139853]

20. Hogquist KA, Jameson SC, Health WR, Howard JL, Bevan MJ, Carbone FR. T cell receptorantagonistic peptides induce positive selection. Cell 1994;76:17. [PubMed: 8287475]

21. Dyall R, Bowne WB, Weber LW, et al. Heteroclitic immunization induces tumor immunity. J ExpMed 1998;188:1553–61. [PubMed: 9802967]

22. Thomas DA, Massague J. TGF-beta directly targets cytotoxic T cell functions during tumorevasion of immune surveillance. Cancer Cell 2005;8:369–80. [PubMed: 16286245]

23. Coligan, JE.; Kruisbeek, AM.; Margulies, DH.; Shevach, EM.; Strober, W. Current Protocols inImmunology. Wiley and Sons, Inc; 1998.

24. Altman, DG. Practical Statistics for Medical Research. Chapman Hall; London: 1991. Analysis ofsurvival times; p. 365-95.

25. Veldhoen M, Hocking R, Atkins C, Locksley R, Stockinger B. TGFbeta in the context of aninflammatory cytokine milieu supports de novo differentiaton of IL-17-producing T cells.Immunity 2006;24:179–89. [PubMed: 16473830]

26. Mangan P, Harrington L, O'Quinn D, et al. Transforming growth factor-beta induces developmentof the TH17 lineage. Nature 2006;441:231–4. [PubMed: 16648837]

27. Bettelli E, Carrier Y, Gao W, et al. Reciprocal developmental pathways for the generation ofpathogenic effector TH17 and regulatory T cells. Nature 2006;441:235–8. [PubMed: 16648838]

28. Sharpe AH, Wherry EJ, Ahmed R, Freeman GJ. The function of programmed cell death 1 and itsligand in regulating autoimmunity and infection. Nature Immunology 2007;8:239–45. [PubMed:17304234]

29. Ansari MJ, Salama AD, Chitnis T, et al. The programmed cell death-1 (PD-1) pathway regulatesautoimmune diabetes in nonobese diabetic (NOD) mice. JExpMed 2003;198:63–9.

30. Bonnotte B, Gough M, Phan V, et al. Intradermal injection, as opposed to subcutaneous injection,enhances immunogenicity and suppresses tumorigenicity of tumor cells. Cancer Research2003;63:2145–9. [PubMed: 12727832]

31. Mukherjee P, Ginardi AR, Madsen CS, et al. Mice with spontaneous pancreatic cancer naturallydevelop MUC-1-specific CTLs that eradicate tumors when adoptively transferred. J Immunol2000;165:3451–60. [PubMed: 10975866]

32. Bateman A, Bullough F, Murphy S, et al. Fusogenic membrane glycoproteins as a novel class ofgenes for the local and immune-mediated control of tumor growth. Cancer Res 2000;60:1492–7.[PubMed: 10749110]

Kottke et al. Page 10

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 1. Ad-VSV-G+Adhsp70 Treatment of the Prostate, But Not Pancreas, Is Associated withOngoing Autoimmune DestructionA-C. C57Bl/6 mice were injected intraprostatically with (5×108p.f.u. Ad-VSV-G+5×108p.f.u. Ad-hsp70) (13,14,32) (A&B at low and high powers) or with 109p.f.u. Ad-GFP (C). Mice were euthanized 3 weeks later. Prostates were analyzed histologically usingH&E sections. Ad-hsp70 induced moderate levels of inflammatory infiltrate (not shown) (2)but extensive immune infiltration/necrosis was recorded with Ad-VSV-G+Adhsp70.Essentially normal prostate architecture was recorded with Ad-GFP (C). D/S1A. Mice wereinjected intra-pancreatically with (5×108p.f.u. Ad-VSV-G+5×108p.f.u. Ad-hsp70) andeuthanized 3 weeks later. A pancreas injected with Ad-VSV-G+Ad-hsp is shown at the siteof injection (D) where local inflammation and tissue damage was evident; an adjoiningsection of tissue is shown with a normal islet (S1A) indicating that very little inflammationwas initiated elsewhere throughout the pancreas. S1B/S1C. Blood samples were drawn fromtail veins following intra-pancreatic injection of PBS or Ad-VSV-G+Ad-hsp70 and glucoselevels were measured. S1D. cDNA from 2 prostates and 2 pancreas injected as above witheither 109p.f.u. of Ad-GFP, or 109p.f.u. of Ad-hsp70, and normalized for equal weights oftissue, was screened for expression of hsp70 using primers specific for vector expressedhsp70. Similar levels of transgene expression were detected in both types of organs.

Kottke et al. Page 11

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 2. Inflammatory killing of pancreas induces a TGF-β-dependent Treg responseA. Supernatants from pancreas and prostates from C57Bl/6 mice (3/grp), incubated withrhsp70 or BSA, were assayed for IL-6. Error bars for each pancreas sample represent thestandard deviation from three wells per sample in the ELISA assay; error bars from theprostate samples represent the standard deviation of mean IL-6 readings per prostate (n=3).Representative of two separate experiments. B. C57Bl/6 mice were injected intra-pancreatically with 109p.f.u Ad-GFP, Ad-VSV-G, or Ad-hsp70 or with (5×108 p.f.u Ad-VSV-G+5×108 p.f.u Ad-hsp70). 8d later, draining LN were analyzed by rtPCR for IL-6,TGF-β or GAPDH. Representative of multiple experiments. C. C57Bl/6 mice were injectedintra-pancreatically with 109p.f.u of Ad-GFP (lane 1) or with (5×108p.f.u Ad-VSV-G+5×108p.f.u Ad-hsp70) (lanes 2&4). 14 days later, 250,000 splenocytes were plated with105 naïve OT-I CD8+T cells with either ova peptide SIINFEKL (lanes 1-3) or with the non-activating TRP-2 peptide (lane 4), in triplicate. 48hrs later supernatants were assayed forIFN-γ. Lane 3: OT-1 with no added splenocytes. D. The experiment of C. was repeated withOT-I cells co-cultured with SIINFEKL and splenocytes from C57Bl/6 mice either notinjected (lane 1), injected intra-pancreatically with 109p.f.u Ad-GFP (lanes 2,3), or with(5×108p.f.u Ad-VSV-G+5×108p.f.u Ad-hsp70) (lanes 4,5). Splenocytes in lanes 3&5 werealso cultured with 50ng/ml of 341-BR TGF-β sRII/Fc to neutralize TGF-β.

Kottke et al. Page 12

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 3. Inflammatory killing of normal pancreas induces pancreas specific responsesA. C57Bl/6 mice (3/grp) were injected intra-pancreatically with 109p.f.u of Ad-GFP, or with5×108p.f.u. Ad-VSV-G+5×108p.f.u. Ad-hsp70. Mice were injected i.p with the Tregdepleting PC61 antibody, or control Ig, (days -2 and -1 prior to intra-pancreatic treatmentwith adenoviruses). 10 days following intra-pancreatic injections, 500,000 splenocytes wereincubated with freeze/thaw lysates of dissociated pancreatic tissue every 48 hrs for 6 daysafter which supernatants were assayed for IFN-γ. B-D. The experiment of A. above wasrepeated (n=5/grp). Blood samples from tail veins were assayed for glucose levels. S2A-S2C. 20 days following intrapancreatic injection with PC-61+ Ad-GFP (S2A) or withPC-61+Ad-VSV-G+Ad-hsp70 (S2B 10X, S2C 100X), as described in A. above, pancreas ofmice were inspected histologically.

Kottke et al. Page 13

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 4. Inflammatory killing of normal pancreas primes antigen-specific responses which canreject associated tumorsA. C57Bl/6 mice were injected intra-pancreatically with 0.1mg pCMV-ova along with109p.f.u. of Ad-VSV-G or Ad-hsp70 or with (5×108p.f.u. Ad-VSV-G+5×108p.f.u. Ad-hsp70). All groups received either a control Ig or Treg depleting PC61 2 and 1 days beforethe intra-pancreatic injection. 1 week later, 500,000 splenocytes per group were co-culturedwith the ova peptide SIINFEKL in triplicate and 48hrs later supernatants were assayed forIFN-γ. Representative of two experiments. B. 2×105 B16, or B16ova, cells were seeded s.c.in C57BL/6 mice. On day 6 mice were injected intra-pancreatically with 0.1mg of pCMV-ova along with (5×108p.f.u. Ad-VSV-G+5×108p.f.u. Ad-hsp70). Groups also received eithera control Ig or PC61 antibody 2 and 1 days before the intra-pancreatic injection. Survival(tumor = 1.0cm) following seeding of tumors is shown.

Kottke et al. Page 14

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 5. Inflammatory killing of normal pancreas requires depletion of Treg to unmask anti-tumor immune responsesA-C. C57Bl/6 mice were injected subcutaneously with 5×105 Panc02 cells. 1 and 2 dayslater mice were administered 0.5mg PC61 antibody or control Ig. The following day (d,4),mice were injected intra-pancreatically with 109p.f.u. of Ad-GFP (A), or with (5×108p.f.uAd-VSV-G+5×108p.f.u Ad-hsp70) (B,C). Tumor size with time is shown (5/6 per group).D/S3A-S3B. 21 days following intra-pancreatic injections, at the time of completeregression of Panc02 tumors in mice treated with PC-61+Ad-VSV-G+Ad-hsp70, 500,000splenocytes per mouse were incubated with either 5×105 Panc02 or B16 tumor cells for 72hrs, or with freeze/thaw lysates of dissociated prostate or pancreas every 48 hrs for 6 daysafter which supernatants were analyzed for IFN-γ.

Kottke et al. Page 15

Cancer Res. Author manuscript; available in PMC 2011 March 1.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript