Prdm16 belongs to the PR domain-containing protein family of transcription factors, which control a plethora of essential cellular processes, including specification of cell lineage during development (Chi and Cohen, 2016). Prdm16 was first identified in leukemia, where truncation mutants lacking functional domains behaved as oncogenic (Zhou et al., 2016), providing the first indication that Prdm16 might function as a tumor suppressor. In addition its involvement in leukemia, several studies have subsequently shown that Prdm16 controls brown fat cell differentiation as well as dedifferentiation of white fat to beige fat (Harms et al., 2015; Hiraike et al., 2017; Seale et al., 2008; Seale et al., 2007). Moreover, Prdm16 is required for stemness in multiple tissues, including hematopoietic and nervous systems (Chi and Cohen, 2016). Germline deletion of Prdm16 in mice impairs the maintenance of neural and hematopoietic stem cells during fetal development, resulting in neonatal death (Shimada et al., 2017). As such, this lethal phenotype hampered any further investigation to delineate a possible role of Prdm16 in cell fate determination in other organ systems, such as pancreas, where the same progenitor cells give raise to all pancreas lineages, e.g., ductal, acinar, and islet (Gu et al., 2003). In this study, we found that conditional deletion of Prdm16 in early pancreatic progenitor cells had no discernible impact on animal health or pancreas physiology, indicating that Prdm16 is dispensable for pancreas development and function. Because mutational inactivation of PRDM16 has been shown to be associated with leukemia (Zhou et al., 2016), we went on to explore whether Prdm16 could contribute to the pathogenesis and/or progression of PDAC, in which acquisition of oncogenic KRAS endows acinar cells with stemness traits that facilitate their differentiation toward a ductal-like lineage, thereby culminating in acinar-to-ductal metaplasia and attendant emergence of premalignant lesions (Bardeesy et al., 2006a; Gu et al., 2003; Park et al., 2008; Tuveson et al., 2004). Progression of premalignant lesions either follows the PanIN-to-PDAC sequence, MCN-to-PDAC, or IPMN-to-PDAC sequence, depending on the nature of the secondary genetic events (Bardeesy et al., 2006a; Bardeesy et al., 2006b; Gu et al., 2003; Tuveson et al., 2004). Yet, among the most studied secondary genetic alterations in PDAC, only Smad4 inactivation stood out as the main mechanism that enables progression through the IPMN-to-PDAC sequence (Bardeesy et al., 2006b; Whittle et al., 2015). To the best of our knowledge, how Smad4 inactivation facilitates this IPMN-to-PDAC transition phenotype has never been addressed experimentally. Using the KrasG12D-based mouse model of PDAC, we confirmed that KSC mice develop mostly IPMN lesions as described initially (Bardeesy et al., 2006b) rather than MCN lesions described in a subsequent study (Izeradjene et al., 2007). Most importantly, we found that concomitant ablation of Prdm16 and Smad4 (KSPrC) resulted in highly aggressive tumors, which develop with very short latencies to the full-blown PDAC and frequently metastasize to the lung, a site associated with the human disease (Connor and Gallinger, 2021; Hidalgo, 2010). Comprehensive histopathological analyses revealed that these tumors follow the PanIN-to-PDAC progression route rather than the IPMN-to-PDAC progression route that proceeds with ablation of Smad4 alone. Because inactivating Smad4 led to the increased expression of Prdm16, we proposed a model in which Prdm16 functions as a molecular switch to dictate whether the malignant transformation process follows the IPMN-to-PDAC route or the PanIN-to-PDAC route (Fig. 8). This model also posits that Prdm16 might function to suppress PDAC pathogenesis at very early stages of the malignancy. In further support of this notion, we found that ablating PRDM16 in the human PDAC cancer cell lines BxPC-3 and PANC-1 did not influence their proliferative or metastatic behaviors, as evidenced using a variety of in vivo and in vitro tumor growth and invasion assays. In light of these findings, a more comprehensive investigation using genetic and histological approaches are needed to firmly establish whether Prdm16 indeed elicits its tumor suppressor activity at early stages, and if so, whether this occurs through direct effects on cancer cell growth or tumor microenvironment reprogramming. As such, our findings open up unique frameworks that would ultimately leverage general efforts to unravel mechanistic paradigms of PDAC, for which very limited therapeutic interventions are currently available.

Accumulating evidence suggests that Prdm16 functions as a potent inhibitor of TGF-β/Smad signaling under various physiological contexts (Chuikov et al., 2010; Stine et al., 2019; Takahata et al., 2009). TGF-β/Smad signaling is well known to play a dual role during cancer progression, functioning at early stages as a tumor suppressor to restrict the malignant transformation, and at late stages as a tumor promoter to facilitate cell invasion and metastasis (Feng and Derynck, 2005). To date, the most appealing speculations as to TGF-β dual function during PDAC progression have been that loss of the TGF-β cytostatic function enables cells to escape growth-inhibitory regulation, which would ultimately culminate in malignant transformation (David et al., 2016; Feng and Derynck, 2005; Massagué, 2008). Once the tumor has developed, other TGF-β responses unrelated to its cytostatic function then supposedly prevail presumably in a manner that facilitates PDAC invasion and metastasis (Bardeesy et al., 2006b; Feng and Derynck, 2005; Ijichi et al., 2006; Massagué, 2008). Interestingly, high levels of TGF-β expression in human PDAC strongly correlates with poor prognosis (Friess et al., 1993; Parajuli et al., 2019), which raises a conundrum as to whether activation of TGF-β signaling could contribute directly to malignant transformation in addition to driving cell invasion and metastasis. However, subsequent studies have shown that Smad4 inactivation in the context of KrasG12D (KSC) led to the acceleration of PDAC (Bardeesy et al., 2006b; Izeradjene et al., 2007), unequivocally confirming the tumor suppressor role of TGF-β signaling in PDAC. Nevertheless, the tumors deficient for Smad4 retained epithelial differentiation and manifested an attenuated metastatic potential (Bardeesy et al., 2006b; Whittle et al., 2015), which is also in favor of a tumor promoter role of TGF-β signaling. So far, definitive experimental evidence on whether inactivation of canonical TGF-β/Smad signaling per se is sufficient to suppress PDAC invasion and metastasis in an irreversible manner is still lacking. Here, we found that ablating Prdm16 in a Smad4 null-background was sufficient to render the PDAC tumors again highly invasive and metastatic. Intriguingly, concomitant ablation of Prdm16 in KSC mice also resulted in a shift from IPMN to PanIN, which could conceivably contribute to metastasis in KSPrC mice, as the vast majority of PDAC GEMMs that develop PanINs also develop highly metastatic PDAC, including KSPC mice (Smad4 deletion and p53.R172H expression), which behave similarly to our KSPrC mice (Bardeesy et al., 2006a; Bardeesy et al., 2006b; Tuveson et al., 2004; Whittle et al., 2015). These findings, together with the observation that Prdm16 expression is lost during late stages of PDAC, highlight Prdm16 as a key player in PDAC progression and metastasis when Smad4 is inactivated. Because TGF-β signaling activation leads to the accumulation of Prdm16 through the suppression of Smad4 inhibitory effects, one would speculate that Smad4 and Prdm16 might function in the same signaling network that integrates the TGF-β tumor promoter effects during PDAC progression. However, it is also conceivable that Prdm16 might function to suppress metastasis induced by other TGF-β superfamily members, such as Activins and BMPs, which are known to signal through Smad4, and can enhance malignancy and promote cancer metastasis in a variety of human malignancies (Attisano and Wrana, 2000; Feng and Derynck, 2005; Pickup et al., 2017). As such, a comprehensive investigation of the mechanisms by which Smad4 and Prdm16 interact to influence PDAC progression may uncover the existence of additional key players and/or pathways that are amenable to therapeutic interventions.

Perhaps the most intriguing finding in this study was the persistent increase in Prdm16 expression during the progression from IPMN to PDAC in KSC mice, which at first glance seems to support a hypothesis in which Smad4 might function as a repressor of Prdm16 during PDAC progression, and hence conceivably that canonical TGF-β/Smad signaling might also repress Prdm16 expression. Quite unexpectedly, we found that activation of TGF-β signaling did not repress Prdm16 expression, but rather resulted in a strong accumulation of both Prdm16 mRNA and protein both in KSC mice and human PANC-1 cells. Noteworthy, we also detected relatively high expression in the stromal compartment, which likely occurs because of the increased TGF-β signaling, which is known to take place during PDAC progression and contribute to the desmoplastic stroma of this malignancy (Friess et al., 1993). In efforts to probe the underlying mechanisms, we found that inactivating Smad4 was sufficient to recapitulate the effects of TGF-β signaling, inducing Prdm16 expression to an extent similar to that elicited by TGF-β1. Based on these observations, we reasoned that activation of TGF-β signaling might relieve the transcriptional repression imposed by Smad4 on the Prdm16 promoter. However, although we found that Smad4 associated strongly with the PRDM16 promoter at steady state, this binding was not affected by the activation of TGF-β1 signaling, indicating that other factors are involved in TGF-β-mediated Prdm16 expression. Probing this possibility, we detected a strong binding of Prdm16 to its own promoter at steady state, which was almost completely abolished by TGF-β stimulation, suggesting that activation of TGF-β signaling might dislodge Prdm16 from its own promoter. Of note, Prdm16 failed to bind to its promoter in cells deficient for SMAD4, suggesting that Smad4 might associate with and recruit Prdm16 to the PRDM16 promoter. Because Prdm16 has been shown to function as a potent transcriptional repressor in various contexts (Pinheiro et al., 2012; Seale et al., 2008; Seale et al., 2007; Stine et al., 2019; Takahata et al., 2009), we proposed a model in which Prdm16 mediates its own repression once it has been recruited to its promoter by Smad4. While these data demonstrate for the first time that Prdm16 can repress its own expression, we cannot exclude the possibility that other mechanisms might also contribute to this phenomenon. Despite this limitation, our study sheds light on a previously uncharacterized interplay between Smad4 and Prdm16, which appears to dictate the progression trajectory of PDAC. Going forward, we anticipated that our discovery will guide forthcoming studies seeking to understand mechanistic paradigms of PDAC, which could ultimately pave the way for innovative therapeutic breakthroughs to curb this deadly disease.