It Takes a Village to Overcome KRAS Dependence in Pancreatic Cancer

Cancer is considered a multistep process. Pancreatic ductal adenocarcinoma (PDAC), in particular, has a well-established sequence of recurring mutations in four driver genes including KRAS, CDKN2A, SMAD4, and TP53. KRAS variants are the most abundant driver mutations found in upward of 95% of cases and have been linked to several features of the PDAC development and the malignant phenotype, making it an ideal therapeutic target (1). Unfortunately, to date, these driver genes have largely remained undruggable and despite advances in sequencing technologies, very few additional significantly recurrent PDAC drivers have been identified (2). Given its prevalence, KRAS remains the most ideal target for treatment, although identification of tolerable, effective targeted therapy has remained elusive. A seeming breakthrough was announced last year with the development of a small-molecule inhibitor (AMG-510) that specifically targets KRAS^G12C (3). Although this holds promise in the management of subsets of lung or colorectal adenocarcinoma, the most common KRAS mutation in PDAC is KRAS^G12D. Nonetheless, this discovery makes targeting oncogenic KRAS signaling caused by other mutations a possibility in the near future.

However, given that cancer is an evolutionary process, treatment resistance is nearly inevitable as a consequence of not only genetic heterogeneity within the cancerous cells themselves, but also the physiologic heterogeneity of the tumor ecosystem composed of a number of “normal” cell populations that interact in dynamic ways with the tumor (4). Appreciating the PDAC tumor as a complex system like this, consisting of a stromal compartment, vascular endothelium, and immune infiltrates, not only expands the number of potential therapeutic targets, but also can inform acquired resistance to therapy in this precision medicine era. The current study (5) utilized an inducible Kras^G12D Trp53^−/− (iKPC) mouse to effectively model oncogenic KRAS signaling extinction so as to identify effective bypass mechanisms. Using a human cDNA library of nearly 300 epigenetic regulators, the authors infected iKPC cancer cells after KRAS extinction and subsequently orthotopically transplanted the cells back into nude mice. Monitoring was then conducted for tumors that “escape” dependency of oncogenic KRAS signaling. qPCR of escape tumors identified histone deacetylase 5 (HDAC5) as a mediator of tumorigenesis in all cases in which it was expressed. This finding was specific for HDAC5 and dependent on an intact deacetylase domain.

Overexpression of HDAC5 in spheroids derived from iKPC cell lines derived after withdrawal of doxycycline (turning off oncogenic KRAS) failed to promote organoid growth suggests the need for tumor microenvironment (TME) to support escape tumor growth. Gene set enrichment analysis of HDAC5-driven escape tumors revealed increased expression of genes related to inflammatory pathways. Taken together, this raised the hypothesis that such HDAC5-driven tumors may affect the immune microenvironment to provide trophic support. Intersecting transcriptomic profiles of HDAC5-driven tumor cytokines with expression of receptors from such in vivo tumors identified marked upregulation of TGFβ1 and its cognate receptor TGFβ receptor 3 (TGFβR3), respectively. Furthermore, supplementation of TGFβ to the HDAC5-driven spheroids was sufficient to support their growth through canonical SMAD3/4-dependent signaling.

To connect these two observations, HDAC5-driven tumor escape and TGFβ1-induced trophic signaling, the investigators noted that TGFβ1 growth support was not dependent on intact HDAC5, potentially further supporting its cell-extrinsic source. Thus, the focus shifted toward examining differences between the TME of oncogenic KRAS–driven tumors and that of escape tumors. The TME consists of several cell populations including vascular endothelium, fibroblasts with an extensive desmoplastic reaction, and an immune cell infiltrate. Given the extent of desmoplasia, much attention has been focused on the role of the stromal compartment in PDAC pathobiology and response to therapy (6). However, this study identified a critical role of the inflammatory infiltrate.

To interrogate differences in the TME, combinations of mass cytometry (CyTOF), flow cytometry, and IHC of the different tumor types were used. Interestingly, the authors noted a clear switch from a neutrophil-predominant (CD45⁺ CD11b⁺ Ly6G^hi Ly6C^lo) infiltrate to a macrophage-rich (CD45⁺ CD11b⁺ F4/80⁺ Ly6C⁻) one in the escape tumors relative to KRAS-driven tumors. These macrophages prominently expressed TGFβ1. Interestingly, not only did this macrophage switch occur in both orthotopic and subcutaneous murine models, but phenotypic characterization revealed that a majority originated from the bone marrow as opposed to the local tissue-resident cells. Together, this suggests these tumor-associated macrophages (TAM) were actively recruited from the peripheral circulation by HDAC5-driven tumors. Furthermore, depletion of this cell population in vivo using clodronate impaired KRAS-independent tumor growth significantly.

TAM represents a critical myeloid subpopulation in the PDAC TME. It has previously been demonstrated that TAM secretion of IL6 promotes PDAC progression through STAT3 signaling (7). Thus, the alternative trophic signaling axis mediated by TGFβ1 suggests redundant mechanisms by which the PDAC cancer cells can co-opt the immune infiltrate. TAMs have also been implicated in treatment resistance, making them potentially promising targets for therapy (8). The observed remodeling of the immune TME was also stimulated using conditioned media derived from HDAC5-driven escape tumors. On the basis of transcriptomic analysis, macrophage chemokine ligands CCL2 and CCL7 were both found to be relatively highly expressed. Overexpression of CLL2 and neutralization of CLL2 and cognate chemokine receptor type 2 (CCR2) promoted and blocked KRAS-independent tumor growth, respectively.

To make a mechanistic connection between HDAC5-driven tumor escape and TME remodeling, the authors cross-referenced a number of datasets. This included chromatin immunoprecipitation sequencing, probing for HDAC5 occupancy at gene promoters, and integrating these findings with transcriptomic analyses of differentially expressed genes both in HDAC5 escape tumors and after HDAC5 knockdown. The candidate genes identified included the negative regulator of JAK/STAT signaling, Socs3. Markers of active gene expression including histone 3 lysine 9 acetylation (H3K9ac) and histone 3 lysine 27 acetylation (H3K27ac) at the Socs3 promoter decreased with overexpression and increased with knockdown of HDAC5. Not only was acetylation of the Socs3 promoter inversely correlated with CCL2 expression, but knockdown of Socs3 also resulted in increased CCL2 levels.

Through the utilization of a sophisticated inducible oncogenic KRAS–driven mouse model, the authors comprehensively and eloquently elucidated a novel mechanism by which PDAC can remodel its local TME to overcome dependency of KRAS signaling for growth. With extinction of oncogenic KRAS, HDAC5 promotes recurrent tumor growth by forming a corepressor complex to modulate several inflammatory signaling-related genes including Socs3. Through deacetylation at the Socs3 genetic locus, expression of this JAK/STAT-negative regulator is inhibited, which ultimately results in derepressed expression of CCL2. This macrophage chemokine is secreted to recruit M2 macrophages, resulting in TME immune remodeling with a neutrophil-to-macrophage switch. Recruited macrophages then secrete TGFβ1, which stimulates SMAD4-dependent signaling (Fig. 1). Detailed mechanistic studies were validated in a host of models including not only in vitro murine-derived cell lines and human cell lines, but also orthotopic and subcutaneous allografts in addition to immune-competent mouse models with different genetic backgrounds.

This clearly has significant translational implications for PDAC treatment. Although standard of care for PDAC remains chemotherapy, given the prevalence of KRAS-driven disease, therapeutically targeting this signaling axis is a primary goal. However, preclinical investigations targeting KRAS and/or downstream proxies reveal adaptive resistance mechanisms to this treatment strategy. Specifically, they tested the use of KRAS inhibitor ARS-1620, PI3K inhibitor and MEK inhibitor trametinib. In oncogenic KRAS–driven tumors, treatment with ARS-1620 and trametinib resulted in increased HDAC5 expression and TAM infiltration into the TME. In fact, supplementation of TGFβ1 to human and murine KRAS-driven spheroids resulted in attenuated sensitivity to ARS-1620 treatment, whereas knockdown of SMAD4 or neutralization of TGFβR resensitized cells to KRAS abrogation. Although HDAC5 inhibitors were not effective alone in KRAS-driven tumors, addition of an HDAC inhibitor to combined MEK and PI3K blockade was superior to the dual combination alone.

Finally, the detailed mechanism presented here allows for a variety of therapeutic interventions including HDAC4/5 inhibition, small-molecule inhibition of TGFβR, CCR2 blockade, or even neutralization of CCL2. All of these modalities were tested in a preclinical study using the iKPC mouse. As expected, these agents had minimal effect in the presence of oncogenic KRAS signaling. However, after extinguishing this signaling axis, mice were treated with one of these various strategies. Although all treatments attenuated tumor growth and increased mouse survival of extinction alone, the most pronounced effects were seen when targeting HDAC5 or CCR2. In fact, CCR2 inhibition has already been demonstrated to be safe and tolerable in a phase Ib trial in combination with standard-of-care chemotherapy (9).

TAM-secreted TGFβ1 supports HDAC5-driven PDAC tumor growth in a SMAD4-dependent manner. This microenvironment-dependent mechanism of treatment resistance could serve as rationale for a novel therapeutic adjunct for a subset of cases, although it should be noted that approximately 55% of cases harbor mutant SMAD4. This important study, among others (10), lends further credence to the need not only to understand the pathobiology of PDAC itself, but also to more comprehensively understand PDAC pathobiology including TME dynamics to inform more effective, personalized treatments at the bedside.

No potential conflicts of interest were disclosed.