Zusammenfassung Neueste Daten deuten daraufhin, dass maligne Tumoren der immunologischen Überwachung über eine Herunterregulierung der T-Zell-Aktivierung mittels eines PD-1 (programmed death 1)/ PDL-1/PDL-2 Signals ausweichen. Dies führt offensichtlich zu einer herabgesetzten Immunantwort und kann somit das Tumorwachstum fördern. Das Oberflächenmolekül PD-1, welches auf T- und B-Zellen, myeloischen Zellen und auf vielen menschlichen Karzinomen exprimiert wird, gehört zu der CD28 Familie, und PDL-1 (B7H1) und PDL-2 (B7DC) wurden als Liganden zu PD-1 beschrieben. Der CD28/B7 Signalweg gehört zu der Gruppe kostimulatorischer Signale, die beim zustande kommen einer T-Zell-gerichteten Immunantwort als zweites kostimulatorisches Signal notwendig sind. In dieser Arbeit wurde die molekulare Expression von PD-1, PDL-1 und PDL-2, von Zytokinen und T-Zell-Subpopulationen im Tumorgewebe von 81 Patienten analysiert, die sich einer kurativen oder palliativen Operation (gemäß UICC- Stadien I-IV) eines primären kolorektalen Karzinoms unterzogen hatten. Es zeigte sich auf Protein- als auch auf molekularer Ebene erstmals, dass PDL-1 und PDL-2 im Tumorgewebe fortgeschrittener Tumorstadien (UICC III/IV) signifikant überexprimiert wurden, PD-1 dagegen erniedrigt exprimiert war. PD-1 wurde dagegen deutlich auf infiltrierenden CD4+ Zellen bei Patienten fortgeschrittener Tumorstadien (UICC III/IV) detektiert, wohingegen PDL-1 auf CD4+ Zellen in frühen Stadien gefunden wurde. Im Vergleich zu den frühen Stadien (UICC I/II) wurde eine grössere Anzahl an T-Zellen mit regulatorischem Charakter in den Tumorstadien III/IV beobachtet. Letzteres würde dem Tumor im Bezug auf sein fortschreitendes Wachstum von Vorteil sein, da regulatorische T-Zellen T-Effektorzellen inhibieren können. Im Tumorgewebe fand sich zudem eine verminderte Expression an IFN-gamma, welches unter anderem T-Effektorzellen aktiviert und damit eine Immunantwort verstärkt. Das Zytokin IL-10, welches mit regulatorischen T-Zellen, aber auch mit T-Helfer (Th)2-Zellen und Tumorgewebe assoziiert ist und antiinflammatorische Eigenschaften besitzt, wurde in höheren Stadien verstärkt nachgewiesen. Dadurch verschafft es dem Tumor einen Überlebensvorteil. Die Beobachtungen in dieser Arbeit zeigen, dass PDL-1 und PDL-2 eine Schlüsselrolle während der Tumorprogression zukommen. Regulatorische T-Zellen sind offensichtlich in den Prozeß der Immunantwort gegen den Tumor eingebunden. Die deutliche PDL-1- und PDL-2-Expression auf T-Zellen insbesondere in frühen Tumorstadien lässt darauf schließen, dass die PD-1/PDL-1- bzw. PD-1/PDL-2-Interaktion inhibitorische Signale zwischen Tumorzellen und den T-Zellen vermittelt. In fortgeschrittenen Stadien (UICC III/IV) waren diese kostimulatorischen Signale auf den T-Zellen nur vermindert vorzufinden. Dies hat zur Folge, dass der Tumor sein Wachstum ungehindert fortsetzen kann, da die anti-Tumor-T-Zell-Antwort, die den Tumor normalerweise in seiner Expansion beeinträchtigt, gestört ist. Es wird somit festgestellt, dass eine Blockade der untersuchten Oberflächenmoleküle PDL-1 oder PDL-2 auf Tumorzellen eine wertvolle Option in der Immuntherapie des humanen kolorektalen Karzinoms darstellen könnte. Angesichts der diskutierten Tatsachen im Hinblick auf das Verhalten der regulatorischen T-Zellen in höheren Tumorstadien könnte sich eventuell eine Option damit eröffnen, die regulatorischen T-Zellen mittels eines spezifischen Antikörpers gegen PDL-1 und/oder PDL-2 zu beeinflussen, um somit die hemmenden Auswirkungen gegenüber der anti-Tumor-Immunantwort zu verhindern oder zu revidieren. / Summary Recent data have suggested that malignant tumors are able to escape from the immunological surveillance on a downregulation of the T cells via a PD-1 (programmed death 1) / PDL-1/PDL-2 signal. This obviously leads to a reduced immune response and thus may promote tumor growth. The surface molecule PD-1, which is expressed on T and B cells, myeloid cells and on many human carcinoma, belongs to the CD28 family, and PDL-1 (B7H1) and PDL-2 (B7DC) are ligands to PD-1. The CD28/B7 signal belongs to the group of costimulatory pathways, which is needed for a T-cell-directed immune response as a second costimulatory signal. In this work, the molecular expression of PD-1, PDL-1 and PDL-2, cytokines and T-cell-subpopulations was analyzed in tumor tissue of 81 patients, which have undergone a curative or palliative surgery (according UICC- stages I-IV) of a primary colorectal cancer. For the first time it was shown on the protein and molecular level, that PDL-1 and PDL-2 were significantly overexpressed in tumor tissue of advanced tumor stages (UICC III/IV), the expression of PD-1, on the other hand was decreased. PD-1 was much more detected on infiltrating CD4+ cells in patients of advanced tumor stages (UICC III/IV), while PDL-1 was found on CD4+ cells in the early stages. In comparison to the early stages (UICC I/II), a larger number of T cells with regulatory character in the tumor stage III/IV was observed. The latter would be an advantage for the tumor in relation to its progressive growth, because regulatory T cells are able to inhibit T-effector-cells. In tumor tissue the expression of IFN-gamma was also reduced, which among others activate T-effector-cells and thus reinforced an immune response. The cytokine IL-10, which is associated with regulatory T cells, but also with T-helper(Th)2-cells and tumor tissue, has antiinflammatory properties and was demonstrated enhanced at higher stages. Thus it gives the tumor a survival advantage. The observations in this study show that PDL-1 and PDL-2 plays a key role during tumor progression. Regulatory T cells are obviously involved in the process of immune response against the tumor. The significant PDL-1- and PDL-2-Expression on T-cells, particularly in the early stages of tumor suggests that the PD-1/PDL-1- or PD-1/PDL-2-interaction mediated inhibitory signals between tumor cells and the T-cell. In advanced stages (UICC III/IV) the costimulatory signals on T-cells were decreased. This means that the tumor's growth can continue unhindered, because the anti-tumor-T-cell-response which the tumor usually affected is disrupted in its expansion. It is thus found that a blockade of the investigated surface molecules PDL-1 and PDL-2 in tumor cells could be a valuable option in the immunotherapy of human colorectal cancer. Given the facts discussed in relation to the behavior of regulatory T cells in advanced tumor stages could possibly be an option to influence the regulatory T cells with a specific antibody against PDL-1 and / or PDL-2, thus to prevent or revise the inhibitory effects against the anti-tumor-immunresponse.

En raison du manque d'efficacité et de la toxicité des thérapies conventionnelles contre le cancer, la recherche s'est concentrée sur le développement de nouvelles stratégies. Ces efforts ont été à l'origine de l'essor de l'immunothérapie, dont les acteurs les plus récents sont les anticorps monoclonaux ciblant les points de contrôles immunitaires (PCI). Parmi ces inhibiteurs des PCI, on retrouve les anticorps ciblant la protéine de surface « Programmed Cell Death 1 », les anti-PD1, et ceux ciblant son ligand, « Programmed Cell Death Ligand 1 », les anti-PDL-1. Ces anticorps ont démontré une efficacité spectaculaire dans plusieurs types de cancers, et sont aujourd'hui couramment utilisés en clinique comme thérapies dans le mélanome, le cancer du poumon, de la vessie et du rein. Cependant, ces traitements ne profitent pas à tous les patients atteints de cancer, avec en moyenne 60% de résistance innée, et 25% de résistance acquise après une réponse primaire aux anticorps, variable selon le type de tumeur. Les phénomènes impliqués dans la résistance sont à l'heure actuelle peu connus. Ainsi, l'objectif de mon projet de recherche consistait à établir des modèles in vivo de résistance acquise aux anti-PD1 et anti-PDL 1. Pour ce faire, nous avons utilisé des tumeurs syngéniques de rein (RENCA), de vessie (MB49 et MBT-2) et de colon (MC38) et des souris immunocompétentes, que nous avons rendues résistantes aux traitements en les soumettant à des séries de réimplantation de tumeurs et de traitements, induisant une pression de sélection jusqu'à l'obtention d'un phénotype résistant. Le succès du blocage de l'axe PD1/ PDL-1 étant fortement lié à l'état du microenvironnement tumoral, nous avons mis en place un protocole d'immunophénotypage. Nous avons ainsi pu observer les cellules au profil « anti-tumoral », telles que les cellules T, les Natural Killer, et les macrophages M1, mais également les cellules ayant une fonction immunosuppressive, telles que les macrophages M2, les MDSC, les Treg. Enfin, certaines études ayant identifié une sur-régulation des PCI inhibiteurs alternatifs dans les cas de résistance acquise à l'anti-PD1, nous avons également observé l'expression de LAG3, TIM3 et TIGIT en plus de l'expression de PD1 et PDL-1. Nous avons ainsi pu déterminer que la résistance semble très fortement dépendante du modèle tumoral, même si nous avons pu identifier une diminution des macrophages M1 anti-tumoraux dans l'ensemble des modèles résistants à l'anti- PD1, et une augmentation des Treg dans les modèles résistants à l'anti-PDL-1, suggérant un mécanisme commun de résistance propre respectivement à l'anti-PD1 et à l'anti-PDL-1. Suite à l'identification par Zaretsky et al. de gènes impliqués dans la voie interféron dans des cas de résistance acquise de mélanome traité à l'anti- PD1, nous avons également décidé d'étudier le profil moléculaire des tumeurs résistantes. Cela nous a permis d'identifier 5 gènes communs entre les modèles anti- PD1 et anti-PDL-1 résistants, dont SERPINF1 et FCNA qui semblent prometteurs comme cibles à valider. Enfin, en parallèle de l'établissement et de la caractérisation des modèles de résistance acquise, nous avons testé de nouvelles approches thérapeutiques de potentialisation des anticorps anti-PD1 et anti-PDL-1 en combinaison avec des chimiothérapies de référence pour le cancer étudié. Nous avons ainsi démontré une potentialisation dans les modèles sauvages de cancer du côlon MC38 et de la vessie MB49, aucun effet de la combinaison dans le modèle de cancer du sein métastatique 4T1, et une inhibition de l'effet de l'anti-PDL-1 avec la combinaison dans le modèle de vessie MBT-2. L'immunophénotypage nous a permis de constater ici aussi des différences très importantes entre les modèles tumoraux, au niveau basal et après traitement [etc...] / Because of the limited efficacy and the toxicity of conventional therapies to fight cancer, researchers focused on the new trategies. These efforts lead to the emergence of immunotherapies, whose msot recent actors are the monoclonal antibodies targeting immune checkpoint (ICP). Among those ICP inhibitors, we found antibodies targeting the surface protein « Programmed Cell Death 1 », called anti- PD1, and those targeting its ligand, « Programmed Cell Death Ligand 1 », called anti- PDL-1. Those antibodies shown a great efficacy in a wide diveristy of cancers, and are currently used for clinical practice in the case of melanoma, lung cancer, bladder cancer and renal cell carcinoma. However, those treatments don’t benefit to all tumor bearing patients, with a mean of 60% of innate resistance, and 25% of acquired resistance following a primary response, variable according to tumor type. Phenomena involved in resistance are currently poorly described. In this context, the aim of my project was to establish in vivo preclinical models of acquired resistance to anti-PD1 and anti-PDL-1. To do that, we used syngeneic renal cancer (RENCA), bladder cancer (MB49 and MBT-2), and colorectal cancer (MC38), and immunocompetent mice, that we have made resistant by serial reimplantations of tumors pieces and serial treatments, inducing a selection pressure until we obtained a resistant phenotype. The efficiency of PD1/PDL-1 axis blocking is strongly linked to the microenvironment composition, as a result we realized an immunophenotyping protocol. We observed anti-tumor cells as T cells, Natural Killer cells, and M1 macrophages, but also cells harboring immunosuppressive functions, as M2 macrophages, MDSC, and Treg. Moreover, some studies have identified an upregulation of alternatives ICP in the context of acquired resistance to anti-PD1, so we also observed the expression of LAG3, TIM3 and TIGIT besides PD1 and PDL-1 expression. We shown that resistance is strongly dependant to the tumor model, even if we identified a decrease of anti-tumor M1 macrophages is models resistant to anti-PD1, and an increase of Treg in models resistant to anti-PDL-1, suggesting a common mechanism of resistance specific to respectively anti PD1 and anti-PDL-1. Following Zaretsky and al. identification of genes involved in interferon pathway in the case of acquired resistance to anti-PD1 in melanoma, we decided to study the molecular profile of resistant tumors. We identified 5 common genes differently modulated between anti-PD1 and anti-PDL-1 resistant models, including SERPINF1 and FCNA which seems to be promising as targets to validate. Lastly, in parallel to establishment and characterization of preclinical models of acquired resistance, we tested new therapeutical approches of anti-PD1 and anti- PDL-1 potentiation in combination with reference chemotherapies. We shown a synergy in wild-type colorectal and bladder cancers (MC38 and MB49), no effect of the combination in metastatic breast cancer 4T1, and an inhibition of anti-PDL 1 effect in bladder cancer MBT-2. Immunphenotyping of tumors allowed us to observe here also high differences between tumor models, both at baseline and after treatments initiation. To conclude, even if our results need a validation with patients samples, we demonstrated that different cellular and molecular modifications could be involved in resistance to anti-PD1 and anti-PDL-1, and that resistance could be bypass with chemotherapy combination, according to tumor type

Developing personalized cancer therapies based on cancer genomics methodologies forms the basis for future cancer therapeutics. A genomics platform was developed based on canine cancer to produce a proof-of-concept for personalized genomics led therapeutic choices but also developing personalized therapeutics for canine cancer patients themselves. The platform identified the genetic state of a canine cancer patient within two drugable pathways; p53 and HSP90/IRF1. The former gene was wild-type p53 thus directing the use of p53 activating molecules. The latter mutations in both HSP90 and IRF1 suggested an investigation into HSP90 and interferon signalling molecules as drug leads. Drugs that target both of these pathways were subsequently used to measure drug effects in cell line models but also to identify novel biomarkers of drug responses. My study focused on the effect of the HSP90-inhibitor Ganetespib had on its client proteins, particularly IRF-1. Briefly my results indicated the following:(i) Ganetespib downregulated IRF-1 protein levels in A375 cell lines and this attenuation was not mediated by either MDM2 or CHIP (E3 ligase). IFNγ- induced IRF-1 was also observed to be downregulated when Ganetespib was used in combination therapy.(ii) Insitu proximity ligation assay showed induced HSC70 upregulation upon HSP90 inhibition by Ganetespib and HSC70/MDM2 complexes were seen to be stabilized compared to the usage of MDM2/p53 inhibitor-nutlin. I hypothesize that MDM2/HSC70 complex might chaperon IRF-1 into lysosome for degradation via chaperon mediated autophagy pathway. (iii) My results also indicate that Ganetespib can downregulate IFN γ- induced PDL-1 expression in melanoma cell lines. Pre-sensitizing the cells with Ganetespib prior to the addition of IFNγ could attenuate PDL-1 to basal levels. (iv) My results also showed that the downregulation of PDL-1 by Ganetespib is an IRF-1 dependent mechanism. Therefore, my results suggest that HSP90 represents an important emerging target for cancer therapy because its inactivation results in the simultaneous blockade of multiple signalling pathways and can also sensitize tumor cells to other anticancer agents. Targeting HSP90 could also help to disrupt PD1/PDL- 1 interaction and activate immune system to recognise tumor cells. I conclude that HSP90 and IRF-1 play a critical role in types II interferon pathways and these findings establish a novel basis for the design of future Ganetespib-based combinatorial approaches to improve patient outcomes in this disease. These approaches finally demonstrate that cancer genomics can stratify choice of cancer drugs used on patients but also provide evidence that cancer patient samples can be used for the specific stratification of cancer drug choice based on cancer genomics data.