Abstract

Natural killer (NK) cells are innate lymphocytes with remarkable cytotoxic abilities that control cancer and viral infections independent of antigen specificity. Indeed, NK cells are the first induced pluripotent stem cell (iPSCs)-derived hematopoietic cells to be tested in clinical trials against hematological tumors. However, limited persistence in vivo and complexity of differentiation protocols, pose significant obstacles to widespread NK-based therapeutics. We hypothesize that direct cell reprogramming mediated by cell type-specific transcription factors (TFs) can be employed to generate NK lymphocytes from somatic cells. To define combinations of TF that induce NK cell identity we tested a list of 19 candidate TFs for their ability to activate a NK-specific reporter in mouse embryonic fibroblasts (MEFs). We identified Ets1, Nfil3, T-bet, Eomes (TENE) that activated a NCR1-driven reporter and induce NK progenitor and mature NK global gene expression programs as assessed by single-cell mRNA seq. To evaluate species conservation, we transduced human fibroblasts with 4 TFs and assessed activation of NK-specific markers by flow cytometry. We show that CD34 and CD56 expression is induced by enforced expression of TENE, suggesting that this combination of TFs to induce NK cell identity are conserved in human. CD34 expression increased with time in culture and was enhanced by the addition of cytokines important for lymphocyte development, suggesting cell expansion. Finally, we employed a barcoded TF approach coupled with single-cell mRNA-seq to inform the specification of progenitor versus mature programs. We screened 48 TFs and first confirmed the involvement of Ets1 in immature NK development, T-bet and Eomes in mature NK, and Nfil3 in both stages. We also suggest Runx family TFs and Ikzf1 as novel
regulators of NK development.

Taken together, our results contribute to the understanding of the transcriptional network driving NK lineage commitment and pave the way for the generation of patient- specific NK cells by direct cell reprogramming approaches.

Abstract

Dendritic cells (DCs) are professional antigen-presenting cells able to induce potent and long-lasting adaptive immune responses. Within the DC family, type 1 conventional dendritic cells (cDC1s) excel on the ability to cross-present exogenous antigens to cytotoxic T cells, a critical step for inducing antitumor immunity. However, cDC1s are rare in peripheral blood and in vitro differentiation of monocytes, CD34+ progenitors or induced pluripotent stem cells results in the generation of heterogenous DC populations with poor antigen presentation capacity and limited ability to migrate to lymph nodes. Our group has recently identified the transcription factors PU.1, IRF8 and BATF3 (PIB) as sufficient to reprogram mouse and human fibroblasts into functional cDC1-like cells. However, fibroblasts are not readily available in
sufficient numbers, in contrast to monocytes that are easily accessible in peripheral blood at high numbers.

Here, we investigated cDC1 reprogramming in THP-1 monocytic cell line and explored multiple viral and non-viral delivery systems to achieve efficient gene delivery to primary monocytes. We showed that overexpression of PIB mediated by lentiviral vectors allowed cDC1 reprogramming of THP-1 cells at high efficiency. Reprogrammed THP-1 cells expressed the cDC1 surface markers CLEC9A and CD141, and the co-stimulatory molecules CD40 and CD80. Interestingly, high levels of CLEC9A expression were detected as early as 3 days after transduction, suggesting that reprogramming progresses with fast kinetics in monocytic cells. We observed that toll-like receptor stimulation in reprogrammed THP-1 cells induced upregulation of co-stimulatory molecules and increased secretion of inflammatory cytokines. Interestingly, cDC1 reprogramming was associated with reduced tumorigenicity of THP-1 cells. Lastly, we tested different viral and non-viral delivery systems to further optimize the transduction of primary monocytes and peripheral blood mononuclear cells and observed that AAV vectors and mRNA allowed higher transgene delivery to primary monocytes when compared to lentiviral vectors.

This work demonstrates that overexpression of PIB in human monocytic cells allows efficient and fast reprogramming to cDC1-like cells and gives insights into alternative viral and non-viral systems allowing transgene delivery to primary monocytes and other hematopoietic cells. Ultimately, this study will open the opportunity to develop efficient cDC1-based vaccines for cancer immunotherapy.

Abstract

Hematopoietic stem cell transplantation has been used as the primary curative treatment for a variety of hematologic malignancies. Generation of patient-tailored HSCsin vitro by direct cellular reprogramming has the potential to overcome major limitations of current treatments. Using this method, somatic cells can be converted into different lineages through enforced expression of transcription factors (TFs). The expression of GATA2, FOS and GFI1B (GaFoGi) TFs reprograms human dermal fibroblasts (HDFs) into HSC-like cells. The mechanisms and regulators underlying this dynamic process remain elusive. Genome-wide engineering screening approaches provide an opportunity to map reprogramming regulators and to improve the efficiency of the process.

This project aimed to optimize a CRISPR/Cas9 screening toolbox for defining regulators of hemogenic reprogramming. We first compared knockout efficiency of conditional and constitutive Cas9 enzymes and showed that the constitutive Cas9 yields higher knockout efficiency when compared to the inducible Cas9 system. We have also demonstrated that a multiplicity of infection of 1 is both sufficient and optimal to achieve effective knockout. In parallel, we have optimized the transduction of a sgRNA library targeting 104 genes required for HSC self-renewal and proliferation. Finally, we have improved the delivery of the hemogenic TFs to fibroblasts using a single lentiviral vector. By comparing lentiviral vectors expressing GaFoGi under the control of multiple promoters, we showed that expression of GaFoGi under the spleen focus forming virus promoter generates CD34+CD9+ACE+CD49f+ hemogenic cells at highest efficiency. This strategy has allowed us to define reprogrammed and partial reprogrammed populations for cell sorting and screening of regulators.

The findings of this study provide a foundation for CRISPR/Cas9 screening to define transcriptional and epigenetic regulators of hematopoietic reprogramming. Ultimately, the identification of molecular facilitators and barriers of reprogramming will allow generation of human HSC-like cells at high efficiency and fidelity for autologous or allogenic transplantation.

Abstract

Immunotherapy utilizes the patient’s own immune system for the treatment of cancer. Due to exceptional antigen-presentation capacity Dendritic Cells (DCs) have great potential for cancer immunotherapy. Using direct cellular reprogramming our group has identified a combination of transcription factors, PU.1, IRF8, and BATF3 that reprograms mouse and human fibroblasts and a range of cancer cell lines into antigen-presenting cells resembling type 1 conventional DCs (cDC1) in terms of morphology, transcriptional, epigenetic profiles, and functional features. To support clinical translation of this approach based on induced antigen presentation of cancer antigens, we addressed cDC1 reprogramming in primary cancer tissues obtained from patients with head and neck, urothelial, lung carcinoma, and melanoma. we showed that all primary samples tested were permissive to cDC1 reprogramming with varying efficiencies according to the cancer cell type of origin. Phenotypic analysis demonstrated that reprogrammed primary cancer cells upregulated expression CD45 and HLA-DR which represent well the reprogramming trajectory. Furthermore, reprogrammed cells expressed cDC1-specific marker CD226 as well as co-stimulatory molecules CD40 and CD80, suggesting that reprogrammed primary cells became competent for antigen presentation. In addition, reprogrammed tumor cells secreted pro- inflammatory cytokines- TNF-α and IL-12p70 which may further enhance anti-tumor immunity.

To model cDC1 reprogramming within the tumor microenvironment (TME), we generated cancer derived organoids in the presence or absence of fibroblasts. We showed that reprogramming was feasible in organoid 3D models, despite an overall decrease in transduction efficiency, that could be improved by combining transduction protocols with dissociation methods. Interestingly, the efficiency of cDC1 reprogramming was not hampered in cancer organoids generated with fibroblasts, suggesting that the immunosuppressive TME does not negatively impact the reprogramming process. This study brings valuable information for clinical translation of cDC1 cancer cell reprogramming and contributes to the development of novel immunotherapies based on direct reprogramming.

Abstract

Conventional type 1 dendritic cells (cDC1s) are professional antigen-presenting cells with key roles in initiating and regulating potent and long-lasting anti-tumor immune responses. As such, they are critical for the response to checkpoint blockade and adoptive T-cell transfer but their rarity in peripheral blood has limited the clinical exploitation of cDC1s for cancer immunotherapy. We have recently identified PU.1, IRF8, and BATF3 transcription factors that convert mouse and human fibroblasts into functional cDC1s.

However, low reprogramming efficiency was a major roadblock for clinical translation. Here, we developed a microscopy-based high-content screening platform using the dendritic cell specific reporter system Clec9a-tdTomato, combined with flow cytometry analysis for cDC1 surface markers to identify and validate small molecules (SMs) that enhance reprogramming efficiency. We identified 358 SMs that increased Clec9a-tdTomato reporter activation, providing proof of principle that SMs can augment cDC1 reprogramming. We then validated these SMs by flow cytometry and showed that microscopy-based high-content screening results correlate with flow cytometry. We identified retinoid acid receptor (RAR) agonists, in particular RARβ and RARγ, promoting early reprogramming events as indicated by Clec9a-tdTomato reporter activation. Moreover, inhibition of WNK2 and B-RAF facilitated late events of cDC1 reprogramming as shown by increased CD45 and MHC-II marker expression. Also, WNK2 inhibition increased CD40 and XCR1 marker expression, hence enhancing cDC1 maturation and restricting a cDC1 cell fate.

These findings suggest that a stepwise activation of retinoic acid signaling followed by promoting MEK/MAPK pathway may result in an optimized cDC1 reprogramming protocol. Our work provides new insights into molecular mechanisms underlying cDC1 specification and reprogramming, paving the way to efficiently generate cDC1s for cancer immunotherapy.

Abstract

Conventional dendritic cells type 1 (cDC1s) are a rare subset of circulating immune cells that
excel in cross-presentation ability and induction of antigen-specific, cytotoxic CD8+ T cells.
The transcription factor core that controls the specification of human cDC1 is still largely unexplored. PU.1, IRF8 and BATF3 were previously identified as sufficient in inducing cDC1 fate in murine and human fibroblasts, but how these reprogramming factors engage chromatin and rewire the transcriptional profile remains to be elucidated.

Here, we have dissected the reprogramming mechanisms by performing ChIP-seq analysis of the chromatin targeting by the three reprogramming factors when expressed individually or in combination in human fibroblasts. We show that PU.1 displays dominant and independent chromatin targeting capacity at early stages of reprogramming and recruits IRF8 and BATF3 to a cohort of common genomic targets, including cDC1 and antigen cross-presentation genes. This cooperative binding is reflected by the engagement of open enhancers and promoters, disrupting transcriptional initiation of fibroblast-specific genes. We show the three reprogramming factors physically interact and reprogramming is abolished by a point mutation in IRF8 that affects interaction with PU.1. While IRF8 only depends on PU.1 for chromatin engagement, BATF3 requires both PU.1 and IRF8 expression for targeting at early stages of reprogramming. Indeed, we provide evidence for physical interaction between BATF3 and PU.1, indicating a mediating role of BATF3 in the formation of the “cDC1 reprogramming complex” to fine-tune chromatin engagement.

These results shed light on the molecular mechanisms underlying specification and reprogramming of human cDC1s and provide a platform for the generation of patient-tailored cDC1s, a long-sought DC subset for vaccination strategies in cancer immunotherapy.

Abstract

Direct cell reprogramming through enforced expression of transcription factors (TFs) converts somatic cells into functional, differentiated cells of other lineages without transiting through pluripotency. Recently, PU.1, IRF8, and BATF3 were identified as the TF network capable of reprogramming murine and human fibroblasts into conventional dendritic cells type 1 (cDC1s). Dendritic cells (DCs) are a remarkable heterogenous lineage bridging innate and adaptive immunity. In addition to antigen-presenting conventional DCs, plasmacytoid DCs (pDCs) represent a specific subset specialized at producing type I interferons (IFN-I) during the antiviral response. pDCs’ ontogeny and the TF networks required to specify pDC-fate and function remain unclear.

This thesis aims to define the TF codes required to induce functional pDCs from unrelated cell-types, utilizing cell fate reprogramming approaches. We have employed an IRF8- based screening approach, given its key role and high expression in pDCs. 36 candidate TFs were screened with IRF8 using mouse embryonic fibroblasts carrying a Clec9a- based reporter system. Overexpression of IRF8 and SPIB was sufficient to induce reporter activation. By performing a secondary screen, we identified IKZF2, when combined with IRF8 and SPIB, that in addition to maintaining Clec9a activation, increases hCD2-based reporter expression and kick-start expression of hematopoietic and pDC- specific surface markers. Thus these three TFs were identified as the optimal network to induce pDC-fate. Induced pDCs express the surface markers B220, CCR9, Ly6C, and antigen-presentation MHC-II molecules. Functionally, induced cells acquire the ability to secrete IFN-I and inflammatory cytokines including TNF-a, IL-6, CCL5, and CXCL10, in
response to TLR9 triggering.

This study identified a combination of three TFs to induce IFN-I-producing pDCs from a non-related cell type by direct cell reprogramming. This finding provides insight into the unique developmental program and functional properties of pDCs. In the future, these results open the possibility to explore induced pDCs to induce immune responses for immunotherapy.

Abstract

The potential of dendritic cells (DCs) for cancer immunotherapy has been highlighted for decades. Type 1 conventional DCs (cDC1s) gained recent attention due to their superior ability to perform antigen cross- presentation, a critical step for inducing antitumor cytotoxic T cell responses. However, the clinical testing of cDC1s has been hindered by their shortage in peripheral blood. We have previously identified a combination of three transcription factors PU.1, IRF8, and BATF3 able to reprogram mouse and human fibroblast to cDC1-like cells. Here, we show that the same combination of transcription factors can reprogram human monocytes isolated from peripheral blood into cDC1-like cells. Reprogramed monocytes express cDC1-specific surface markers including CLEC9A, CD141, and XCR1, show increased migratory capacity towards chemokine gradients and produce IL-12. Importantly, we show that reprogrammed monocytes acquire antigen cross presentation capacity to efficiently activate cytotoxic CD8+ T cells. This study sets the foundation to generate clinically applicable cDC1-like cells from easily accessible blood cells for DC vaccination.

Abstract

Despite tremendous efforts and achieved advances, cancer is still the leading cause of death worldwide, which underlines the need to develop novel curative approaches. In the past decade, we witnessed the rapid development of immunotherapy – novel treatment strategies based on harnessing the patient´s very own immune system to fight cancer. Although immunotherapy revolutionized the treatment in multiple malignancies, including advanced melanoma, in the vast majority of patients the response is very limited.

Recently, the Pereira lab at Lund University identified a combination of three proteins, PU.1, IRF8, and BATF3 (PIB) which were sufficient to generate immunogenic type 1 dendritic cells from mouse fibroblasts via direct cellular reprogramming. These new cells had a transcriptional program and cell morphology resembling conventional DC type 1 (DC1), which are specialized in antigen cross-presentation and initiating cytotoxic T cell responses. Furthermore, preliminary data has shown that the reprogramming towards dendritic cell fate can also be applied in cancer cells directly. Therefore, inducing antigen presentation directly in tumor cells may help to bypass current limitations of other immunotherapies, such as tumor cell heterogenicity, immune evasion, and neoantigen identification.

Now, to support the translational efforts of these findings, the application of dendritic cell reprogramming was tested in patient samples. I investigated the reprogramming efficiency mediated by PIB in cancer samples across several malignancies including breast, lung, bladder, pancreatic, head, and neck carcinomas and melanoma as well as cancer associated fibroblasts. I have further evaluated the global gene expression reprogramming at the single cell level with a focus on DC1- and antigen presentation-specific genes. Overall, I observed that all patient samples underwent significant changes during reprogramming with more than 50% of the cells expressing at least one of the reprogramming markers CD45 or HLA-DR. Particularly, lung carcinoma showed highest reprogramming efficiency where more than 60% of the cells were partially reprogrammed and 15% were fully reprogrammed. Interestingly the reprogramming efficiencies observed from patient samples were higher than cancer cell lines, showing a convergent global switch in the transcriptional program to the DC1 fate at the single cell level. Moreover, I also evaluated the reprogramming efficiency in cancer cells tumor organoids constructed by the forced-floating method. Tumor organoids were reprogrammed in similar patterns as their parent cells cultured through conventional 2D methods. However, in the 3D models, it was observed low transduction efficiency, hinting at the potential need to improve the delivery system of PIB into tissues.

These findings support the application of DC reprogramming in patients across multiple malignancies, thus paving the way for the development of a novel cancer gene therapy approach based on dendritic cell reprogramming.

Abstract

Cell fate reprogramming towards pluripotency or alternative somatic cell-types has highlighted the plasticity of adult somatic cells, providing new technologies to generate desired cell types for tissue repair or for disease modeling. There is momentum to bring these concepts to immunology by specifying unique immune cellular identities that set in motion immune responses. Dendritic cells (DCs) are professional antigen presenting cells specialized in the recognition, processing and presentation of antigens to T cells, inducing adaptive immune responses. Within the DC compartment, conventional type 1 DCs (cDC1s) excel in antigen cross- presentation, a critical step to initiate cytotoxic T cell responses. Recent studies have highlighted complementary local and systemic roles of cDC1s in inducing anti-tumor immunity. Nevertheless, their rarity in peripheral blood and the lack of methodologies enabling the generation of a pure population of mature cDC1-like cells limits their study and therapeutic utility. Here, I explore direct cellular
reprogramming from non-hematopoietic cell-types as an alternative approach to generate cDC1.

In this thesis, I screened 34 transcription factors and identified PU.1, IRF8, and BATF3 as the minimal combination required to reprogram mouse and human fibroblasts into cDC1-like cells. Induced DCs (iDCs) generated by cell
reprogramming acquire a DC-like morphology and express cDC1 surface markers and surface molecules required for antigen presentation to T cells. I used single- cell mRNA sequencing to explore the gradual and asynchronous nature of DC reprogramming and demonstrated the downregulation of fibroblast and cell cycle progression genes coupled with the upregulation of cDC1 and antigen presentation genes. Importantly, this approach generated exclusively cDC1-like cells, and not DCs from other subsets. Reconstruction of successful human DC reprogramming trajectories highlighted gene modules associated with successful reprogramming and identified inflammatory cytokine signalling and the DC-inducing transcription factor network as key drivers of the process. Motivated by these observations, I
combined IFN-γ, IFN-β and TNF-α with constitutive expression of PU.1, IRF8 and BATF3 to increase human DC reprogramming efficiency by 190-fold. Functionally, iDCs respond to inflammatory stimuli, engulf dead cell material, secrete inflammatory cytokines and cross-present antigens to CD8+ T cells. Interestingly, I observed that intra-tumoral vaccination in syngeneic mouse models increased infiltration of antigen-specific CD8+ T cells, promoted a T cell cytotoxic profile in draining lymph nodes and controlled tumor growth. Mechanistically, I
show that PU.1 displays dominant and independent targeting capacity by engaging enhancer and promoter regions in open chromatin and recruiting IRF8 and BATF3 to the same binding sites. This cooperative binding allows the downregulation of fibroblast genes and activation of cDC1 genes required to achieve DC reprogramming. Finally, I adapted the DC reprogramming protocol to allow high- content screening of small molecules and identified several small molecules that increase reprogramming efficiency and potentially replace the action of PU.1, IRF8 and BATF3 in DC reprogramming.

These findings should open avenues to better understand cDC1 specification and
functional specialization. Ultimately, DC reprogramming might represent a platform for the future generation of cDC1s for therapy.