Concurrent targeting of GSK3 and MEK as a therapeutic strategy to treat pancreatic ductal adenocarcinoma

Abstract

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal malignancies worldwide. However, drug discovery for PDAC treatment has proven complicated, leading to stagnant therapeutic outcomes. Here, we identify Glycogen synthase kinase 3 (GSK3) as a therapeutic target through a whole‐body genetic screening utilizing a ‘4‐hit’ Drosophila model mimicking the PDAC genotype. Reducing the gene dosage of GSK3 in a whole‐body manner or knocking down GSK3 specifically in transformed cells suppressed 4‐hit fly lethality, similar to Mitogen‐activated protein kinase kinase (MEK), the therapeutic target in PDAC we have recently reported. Consistently, a combination of the GSK3 inhibitor CHIR99021 and the MEK inhibitor trametinib suppressed the phosphorylation of Polo‐like kinase 1 (PLK1) as well as the growth of orthotopic human PDAC xenografts in mice. Additionally, reducing PLK1 genetically in 4‐hit flies rescued their lethality. Our results reveal a therapeutic vulnerability in PDAC that offers a treatment opportunity for patients by inhibiting multiple targets.

In this study, we found that GSK3 and MEK are therapeutic targets in PDAC. Co‐targeting of these two kinases provides a therapeutic opportunity for PDAC patients.

Abbreviations

AsPC‐1‐Luc

AsPC‐1 cells expressing luciferase

ATF2

Activating transcription factor 2

AURKB

Aurora kinase B

C9

CHIR99021

CT

a combination of CHIR99021 and trametinib

dRet ^M955T

an active M955T isoform of Drosophila Ret

DRSC

Drosophila RNAi Screening Center

FOLFIRINOX

a combination of leucovorin, fluorouracil, irinotecan, and oxaliplatin

GEMMs

genetically engineered mouse models

GSK3

Glycogen synthase kinase 3

IACUC

Institutional Animal Care and Use Committee

L2

LY2090314

LCK

Lymphocyte‐specific protein tyrosine kinase

MTC

edullary thyroid cancer

MTD

maximum tolerated dose

PDAC

Pancreatic ductal adenocarcinoma

PLK1

Polo‐like kinase 1

ptc

patched

PTX

paclitaxel

ROCK

Rho‐associated protein kinase

RTK

receptor tyrosine kinase

Ser

Serrate

TMA

Tissue microarray

Tram

trametinib

UAS

Upstream activation sequence

1. INTRODUCTION

Pancreatic ductal adenocarcinoma (PDAC) is one of the most aggressive and lethal malignancies worldwide. ¹ With the number of PDAC patients on the rise, PDAC will become the second leading cause of cancer‐related death in the USA by 2025. ² Currently, the standard treatment for PDAC includes surgical resection and chemotherapy, such as FOLFIRINOX (a combination of leucovorin, fluorouracil, irinotecan, and oxaliplatin) or gemcitabine combined with nab‐paclitaxel (PTX). Although these regimens provide greater anti‐tumor effects than conventional chemotherapy, they have serious adverse effects that limit the survival benefit to patients. ³ , ⁴ Even targeted therapies, including immune checkpoint inhibitors, have also been disappointing in their efficacy against the majority of PDAC. ⁵ Collectively, there is an unmet clinical need for PDAC therapy.

Thus far, conventional experimental models, such as cultured human cells and GEMMs, have contributed significantly to PDAC research. ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ In fact, these models have provided significant insights into the mechanisms of PDAC development, including the critical role of Kras ^G12D in inducing precursor lesions in the pancreas ⁶ as well as the progression mechanisms of PDAC upon Kras and Trp53 alterations. ⁷ However, these mouse models are costly and labor intensive for high‐throughput genetic and drug screening. Hence, one of the major problems in PDAC research is the lack of an efficient and comprehensive whole‐body animal model to screen out therapeutic targets and drug candidates. To solve this issue effectively, we have been utilizing the fruit fly Drosophila in combination with the above conventional models, leveraging the advantages of flies. First, flies are well characterized for their genome, with orthologs of >70% of genes associated with human diseases, including cancer. ¹¹ , ¹² , ¹³ Second, flies offer a useful genetic toolkit for whole‐body analysis, ¹⁴ , ¹⁵ , ¹⁶ as flies with complex genotypes can be easily produced by genetic modifications or are readily available from resource banks. Third, flies mature rapidly and are highly reproductive. In addition, they do not require massive laboratory equipment, resulting in low breeding costs (<0.1% of mice per animal) and fast and cheap phenotypic screening.

Indeed, we successfully utilized a transgenic fly model carrying dRet ^M955T mimicking the RET ^M918T mutation found in MTC patients ¹⁷ , ¹⁸ to develop a new method ‘Rational Polypharmacology’ to widen the therapeutic window of the approved kinase inhibitor drug sorafenib. ¹⁹ , ²⁰ , ²¹ The key to this method is to conduct chemical genetic screening for the whole kinome in dRet ^M955T flies to reveal ‘anti‐target’ kinases of sorafenib such as MKNK1 whose inhibition causes sorafenib toxicity. Subsequent in silico modeling enabled the identification of sorafenib analogs with significantly reduced binding to the anti‐target, thereby improving the efficacy of sorafenib.

In addition, we recently established a whole‐animal platform also for PDAC research by combining Drosophila with mammalian models in a complementary manner. ²² PDAC is often characterized by somatic mutations in four specific genes, namely activation of the oncogene KRAS and inactivation of the tumor suppressor genes CDKN2A, TP53, and/or SMAD4. ²³ In fact, patients with all four mutations have the worst prognosis among all the PDAC patients, ²⁴ , ²⁵ highlighting the importance of developing a corresponding animal model carrying this ‘4‐hit’ genotype to cause endogenous tumors. Therefore, we used Drosophila to generate the first effective 4‐hit animal model, resulting in the identification of novel therapeutic targets and candidates for PDAC in a whole‐body context. Namely, we conducted comprehensive genetic screening of the entire kinome in 4‐hit flies and identified kinases, including Mitogen‐activated protein kinase kinase (MEK) and Aurora kinase B (AURKB), as therapeutic targets for PDAC. Through chemical testing, we discovered that a combination of the MEK inhibitor drug trametinib and the AURKB inhibitor BI‐831266 consistently suppressed the growth of human PDAC xenografts in mice. ²²

To obtain more therapeutic options for PDAC, we further explored other candidates of therapeutic targets that we discovered during the screening. Our study resulted in the identification of GSK3 as an additional therapeutic target, which holds promise in accelerating the development of new PDAC treatments.

2. MATERIALS AND METHODS

2.1. Drosophila genetic testing

Fly stocks carrying a kinase gene mutation or small interfering RNA (siRNA) were obtained from BDSC, KYOTO Stock Center (Kyoto, Japan), and the National Institute of Genetics (Mishima, Japan). Stocks balanced with balancer chromosomes without the Tubby (Tb) marker were rebalanced with the balancer FM7c‐Tb‐RFP, CyO‐Tb‐RFP, or TM6B carrying Tubby (Tb) as a visible marker. ²² To knockdown kinases specifically in the transformed cells, Ser>4‐hit or ptc>4‐hit flies were crossed with UAS‐siRNA flies. Resulting progenies were cultured until adulthood for 11 days at 27°C, and the number of adults was divided by that of total pupae to determine percent viability.

2.2. Tissue microarray analysis

In total, 86 specimens were used to develop TMA (Table S1) ²² and were analyzed for expression of GSK3 using anti‐GSK3β antibody (RRID, AB_2636978; 1:400; Cell Signaling Technology, Berkeley, CA) (Table S2). Specimens were scored for the ratio of the number of GSK3β‐positive cancer cells to that of cancer cells in the whole core by two researchers (JF and KCH) who were blinded to the clinical information of the patients. The patients were classified into GSK3β‐positive or GSK3β‐negative groups (cutoff: >0%), and overall survival and other characteristics were compared between the two groups using appropriate statistical methods.

2.3. Drosophila chemical testing

All chemicals were dissolved in dimethyl sulfoxide (DMSO; Sigma Aldrich, St. Louis, MO) to prepare stock solutions (Table S3). Before screening, the MTD of each chemical in 0.1% final DMSO concentration was determined by evaluating the viability of non‐transgenic control flies fed with the chemical. Virgin UAS‐4‐hit females were crossed with Ser‐gal4 males to obtain Ser‐gal4;UAS‐4‐hit (Ser>4‐hit) offspring, and their viability was determined as in genetic testing at 22°C (Figure S1A). For wing area analysis, Ser>4‐hit progenies were raised until adulthood on fly food, with or without chemicals, for 21 days at 16°C. Then, the adult wings were dissected, and their size was quantified using ImageJ software.

2.4. Mouse xenograft assay

In total, 1 × 10⁶ AsPC‐1 cells (ATCC, Manassas, VA) transduced with luciferase cDNA were injected into the exposed pancreatic tails in anesthetized mice. The mice were assigned randomly to one of the four arms (n = 7/group) based on similar average tumor size, as determined by bioluminescence measured using the IVIS Spectrum Imaging System (Caliper Life Science, Mountain View, CA). The mice were then dosed five times a week for a period of 3 weeks with vehicle (5% DMSO in saline p.o. plus 25% Cremophor EL and 25% ethanol in saline i.p.), CHIR99021 (C9;10 mg/kg i.p.), trametinib (tram;1.0 mg/kg p.o.), or a combination of C9 (10 mg/kg i.p.) and tram (1.0 mg/kg p.o.). Bioluminescence signals were analyzed on a weekly basis using the IVIS system and quantified using Living Image Ver.4.2 (Caliper, Hopkinton, MA).

Materials and methods for fly assay, cell culture, mouse experiments, immunohistochemistry, phospho‐antibody array, and statistical analysis are described in Data S1.

3. RESULTS

3.1. Identifying GSK3 as a therapeutic target of PDAC through genetic testing in 4‐hit flies modeling PDAC genotype

To mimic the ‘4‐hit’ genotype observed in PDAC patients, we generated a quadruple‐hit animal model through the application of Drosophila transgenic technology ²² (Figure 1A). Namely, we targeted cDNA sequences of Drosophila Ras ^G12D (dRas ^G12D) and Cyclin E (dCycE) and shRNA sequences for a TP53 ortholog p53 and a SMAD4 ortholog Med to developing wing discs in larvae. Several reports, including our own, have substantiated the utility of the disc as a valuable tissue for studying epithelial transformation characterized by increased cell proliferation and migration, as well as for testing drug candidates targeting RTK‐Ras and various kinase signaling pathways. ²⁰ , ²² , ²⁶ , ²⁷ , ²⁸ , ²⁹ In fact, we observed high lethality in Serrate (Ser)‐gal4,UAS‐GFP,UAS‐4‐hit flies (Ser>4‐hit flies; hereafter referred to as 4‐hit flies) replicating the four‐gene alterations under the control of the Ser‐gal4 driver active in the disc, indicating that this strain is a useful model for determining the therapeutic vulnerability of PDAC in a whole‐body manner. We identified MEK and AURKB as novel therapeutic targets in PDAC through a comprehensive dominant modifier screening that introduced a heterozygous mutation in each of 220 kinase genes into 4‐hit flies to determine the impact of their reduced function in a whole‐body context ²² (Figure 1B). In addition to these kinases, we were particularly intrigued by the investigation of another crucial target, sgg [fly ortholog of human lycogen synthase kinase 3α/β (GSK3α/β)], due to a significant improvement in 4‐hit fly viability upon its reduction reaching 73%, which is comparable with the effect observed with Dsor1 (MEK) heterozygosity (Figure 1C, Table S4). These findings suggest that GSK3 also plays an important role in tumorigenesis dependent on the 4‐hit alterations.

FIGURE 1.

Identifying GSK3 and MEK as candidate therapeutic targets in PDAC using 4‐hit flies that model the PDAC genotype. (A) Utilizing a ‘4‐hit’ Drosophila model mimicking PDAC genotype. G12D, G to D missense activation mutation. LOF, loss of function. shRNA, short hairpin RNA. HA, hemagglutinin tag. dCycE, Drosophila Cyclin E. (B) Genetic screening using Ser>4‐hit flies to determine the therapeutic target kinases in PDAC. A heterozygous mutation of each kinase gene was transferred to 4‐hit flies by crossing. The viability of the resulting offspring was compared with that of progenies obtained from a cross between 4‐hit flies and kinase‐proficient control flies. (C) Heterozygosity of sgg or Dsor1 rescued the lethality of 4‐hit flies. Parentheses, human ortholog of fly gene. Error bars, standard deviation (SD) in technical triplicate. (D) Heterozygosity of sgg or Dsor1 suppressed transformation in the developing wing disc of third instar patched (ptc)>4‐hit larvae. Transformed cells are marked by GFP. Arrowheads, representative expanded areas consisting of proliferating cells. Blue, DAPI staining outlining the disc margin. Boxed areas are magnified in the insets. Scale bars, 50 μm. (E) Expanded relative ptc area in 4‐hit wing discs compared with kinase‐proficient control. Each dot represents one disc. (F) Knockdown of sgg in transformed cells rescued the lethality of 4‐hit flies. siRNA was induced under the control of the Ser enhancer. Legend is the same as in (C). (G) Knockdown of sgg or Dsor1 suppressed transformation in the wing disc of ptc>4‐hit larvae. Legend is the same as in (D). (H) Relative ptc area to the whole wing disc. Legend is the same as in (E). *p < 0.001 in Dunnett's test compared with control ptc>4‐hit flies.

To determine the role of sgg in the transformation in the disc, we introduced sgg heterozygosity into ptc‐gal4,UAS‐GFP,UAS‐4‐hit (ptc>4‐hit) flies, which display aggressive cell proliferation and migration. ²² Consistent with the observed rescue of fly viability, reducing sgg or Dsor1 dosage suppressed the transformation (Figure 1D,E). To examine whether sgg and Dsor1 operate specifically in the transformed cells, we utilized UAS‐kinase ^siRNA flies for kinase knockdown in a GAL4‐dependent manner. We found that knockdown of sgg or Dsor1 rescued 4‐hit fly lethality and significantly suppressed the transformation, strongly suggesting that these kinases transduce pro‐tumorigenic signals in a cell‐autonomous manner (Figure 1F–H, Table S4). Based on these findings, we hypothesized that inhibition of GSK3 had anti‐tumorigenic effects on 4‐hit PDAC.

3.2. Association of nuclear GSK3β expression with poor prognosis in PDAC patients

GSK3, a class of serine/threonine kinases comprising two prominent isoforms, GSK3α and GSK3β, was initially identified as a pivotal enzyme implicated in glycogen metabolism. ³⁰ The differentiation between the roles of GSK3α and GSK3β in PDAC development remains elusive, notwithstanding the observed elevation in the expression levels of both isoforms relative to those detected in normal pancreatic tissue (Figure S2). In particular, GSK3β is considered a tumor promoter that correlates with poor prognosis in diverse solid malignancies in the colorectum, lung, breast, brain, ovary, kidney, and bladder. ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ Although a report has demonstrated nuclear accumulation of GSK3β in poorly differentiated PDAC, ⁴¹ the relationship between its expression and the prognosis of patients with PDAC remains largely unexplored. Therefore, we performed an immunohistochemical analysis of GSK3β in PDAC tissues. Of the 86 specimens, 29 (33.7%) were positive for GSK3β staining in the nucleus (Figure 2A, Table S1). Importantly, these patients displayed significantly reduced overall survival compared with those without detectable GSK3β staining (Figure 2B). These results indicate an association between nuclear GSK3β expression and an unfavorable prognosis in patients with PDAC.

FIGURE 2.

Correlation between GSK3β expression and unfavorable prognosis of PDAC patients. (A) Representative immunohistochemical staining of GSK3β in PDAC specimens. Arrowheads, cancer cells positive for GSK3β in the nucleus (brown staining). The boxed area is magnified in the inset. Scale bars, 50 μm. (B) Nuclear expression of GSK3β is associated with poor overall survival of PDAC patients. Kaplan–Meier analyses were conducted by classifying the patients into two groups based on the presence or absence of GSK3β‐positive cells. p‐values were calculated by log‐rank test. Patient characteristics are provided in Table S1.

3.3. Suppression of proliferation of 4‐hit human PDAC cells by GS K 3 knockdown

To elucidate the distinct contributions of GSK3α and GSK3β to PDAC cell proliferation, we used RNA interference in a soft agar colony formation assay. The knockdown of GSK3α and GSK3β in human AsPC‐1 cells carrying the 4‐hit alterations resulted in a significant reduction in their mRNA levels (Figure 3A,B). AsPC‐1 cells that were knocked down for GSK3α or GSK3β exhibited significantly reduced colony numbers compared with the non‐silencing control group. Markedly, the knockdown of both GSK3α and GSK3β further reduced the number of colonies in AsPC‐1 cells compared with single gene knockdown (Figure 3C,D). These results indicated the indispensability of GSK3α and GSK3β activities in the proliferation of human PDAC cells, thereby suggesting their crucial roles in PDAC progression. Furthermore, we speculated that the inhibition of either GSK3α or GSK3β provides a therapeutic effect on PDAC, with the potential for increased potency through their combined inhibition.

FIGURE 3.

Knockdown of GSK3α or GSK3β suppressed the proliferation of human PDAC cells. (A, B) Knockdown of GSK3α (A) or GSK3β (B) reduced their mRNA levels in AsPC‐1 cells. AsPC‐1 cells were transfected with either of two independent siRNA sets, each consisting of distinct siRNAs against GSK3α (siGSK3α #1 and #2) or GSK3β (siGSK3β #1 and #2), and the level of RNA interference was determined by reverse transcription‐qPCR (RT‐qPCR) 48 h post transfection. *p < 0.05 and **p < 0.01 in Dunnett's test. Ns, non‐silencing siRNA. (C) Colony formation assay in soft agar with RNA interference targeting GSK3α or GSK3β in AsPC‐1. Boxed areas are magnified in the insets. Scale bars, 200 μm. (D) Suppression of AsPC‐1 colony formation in soft agar upon knockdown of GSK3α or GSK3β. Colony numbers were scored after seeding for 7 days. *p < 0.001, ^# p < 0.001, and ^§ p < 0.001 in Dunnett's test. Ns. Error bars, SD in technical triplicate. HPF, High‐power field.

3.4. Synergistic rescue effects by a combination of CHIR99021 and trametinib on 4‐hit fly viability

We next investigated the potential of inhibiting GSK3 as a therapeutic strategy for PDAC patients by assessing the ability of their inhibitors to rescue the transformation phenotypes of 4‐hit flies. To this end, we conducted a series of chemical tests utilizing two highly selective GSK3α/GSK3β inhibitors, CHIR99021 (C9) and LY2090314 (L2) ⁴² , ⁴³ (Table S3). Because we identified modest efficacy of the MEK inhibitor drug trametinib, ²² we also tested their combinations with trametinib. As monotherapy, both C9 and L2 failed to exhibit therapeutic efficacy. Nevertheless, C9 combined with trametinib rescued 4‐hit fly lethality more efficiently than L2 combined with trametinib (Figure 4A,B). Because the Bliss synergy score verified that C9 and trametinib acted more synergistically than L2 and trametinib, we validated C9 in this study.

FIGURE 4.

Combination of CHIR99021 and trametinib as a therapeutic candidate as revealed in 4‐hit flies. (A) Rescue of 4‐hit fly lethality by combinations of GSK3α/β inhibitor CHIR99021 (C9) and the MEK inhibitor drug trametinib (tram). A combination of C9 and tram rescued 4‐hit fly lethality (left). The synergy between C9 and tram was statistically significant (right). MicroM, final concentration in fly food. Bliss synergy score (right) was used to generate a heatmap to indicate the effect as synergistic (red), additive (white), or antagonistic (green). (B) Effects of a GSK3α/β inhibitor LY2090314 (L2) and tram on 4‐hit fly viability. L2 and tram also rescued fly lethality in 4‐hit flies (left). The synergy between L2 and tram was statistically significant (right). (C) C9 combined with tram suppressed wing defects in Ser>4‐hit flies. Compared with the wings of wild‐type adults, those of 4‐hit adults were wrinkled and had a smaller wing area. Scale bars, 50 μm. (D) Wing area was increased by the combination treatment. *p < 0.001 in one‐way ANOVA with Tukey's honestly significant difference (HSD) post hoc test. **p < 0.001, ^# p < 0.05, and ^## p < 0.001 in Dunnett's test. Error bars, SD in technical triplicate.

Therefore, we also examined in 4‐hit flies the effect of C9 and trametinib on adult wing malformation, an established indicator of Ras signaling activation. ²² , ⁴⁴ When cultured at 16°C, 4‐hit flies were able to survive because of reduced GAL4 activity, and untreated 4‐hit flies manifested significantly enhanced wing venation, leading to wing malformation. The combination of C9 trametinib consistently suppressed the abnormality, and the effect was more prominent than that of single‐agent treatment (Figure 4C,D). Based on these findings, we speculated that the simultaneous targeting of GSK3α, GSK3β, and MEK presents a novel therapeutic strategy for PDAC patients.

3.5. Suppression of orthotopic 4‐hit human PDAC xenograft in mice by a combination of CHIR99021 and trametinib

To validate the therapeutic efficacy of CT in mammals, we utilized a PDAC xenograft mouse model. Specifically, we established the model by orthotopically inoculating AsPC‐1‐Luc, harboring the four‐gene alterations, in immunodeficient mice. Following the determination of the optimal dose through maximum tolerated dose (MTD) (Figure S3A), and pharmacokinetics (Figure S3B) assays in mice, we administered C9 intraperitoneally and/or trametinib orally to mice bearing AsPC‐1 xenografts for a period of 3 weeks. Tumor progression was evaluated by measuring the bioluminescence signal intensities using in vivo imaging in the presence of luciferin.

Notably, CT treatment caused significantly superior suppression of xenograft growth compared with C9 or trametinib monotherapy (Figure 5A,B). While no mice displayed a reduction in tumor volume upon monotherapy, two out of seven mice demonstrated a partial response to CT treatment (Figure 5C). Importantly, no significant reduction in body weight was observed in mice receiving CT when compared with the control group during the course of treatment, indicating the safety of CT in vivo (Figure S3C). Collectively, our findings provide compelling evidence that the efficacy of CT identified in 4‐hit flies can be translated into mammals.

FIGURE 5.

Anti‐tumor effect of the CHIR99021 and trametinib combination on orthotopic PDAC xenografts in mice. (A) Growth of orthotopic AsPC‐1 xenografts in nude mice treated with vehicle, CHIR99021 (C9; 10 mg/kg i.p.), trametinib (tram; 1.0 mg/kg p.o.), or C9 combined with tram (CT). Luciferase‐expressing human AsPC‐1 xenografts were monitored using in vivo bioluminescent imaging. Bioluminescence signal intensities were shown as photons/sec/cm²/steradian (p/s/cm²/sr). (B) CT suppressed the growth of AsPC‐1 xenografts. *p < 0.05 and **p < 0.01 in the Mann–Whitney U‐test on day 21. Error bars, SD in seven mice. Data from a single experiment. (C) CT inhibited xenograft growth more efficiently than a single treatment. A waterfall plot displaying percent changes in tumor volume on day 21 relative to pretreatment baselines in each group. Each bar represents a single mouse. Red arrowheads indicate partial response according to the RECIST criteria (at least a 30% tumor size reduction from baseline). ⁴⁵

3.6. Suppression of MAPK pathway and PLK1 phosphorylation in PDAC cells upon CT treatment

Although the dosing experiment on mice demonstrated anti‐PDAC effects in a whole‐body context, it remained unclear whether CT targeted specifically cancer cells. To answer this question, we performed a soft agar colony formation assay using cultured human PDAC cells in the presence of the chemicals. In accordance with the findings obtained in the fly and xenograft assays, C9 and trametinib exhibited a synergistic suppression of AsPC‐1 proliferation (Figure 6A,B, Tables S5–S7, Figure S5A). Of note, these treatments also inhibited colony formation of PANC‐1 cells harboring 3‐hit alterations (KRAS, TP53, and CDKN2A; Tables S5–S7, Figure S4A,B), suggesting that our therapeutic seeds can be applied to multiple patient populations of PDAC carrying distinct mutation profiles.

FIGURE 6.

The CHIR99021 and trametinib combination inhibits PDAC cell proliferation. (A) AsPC‐1 colony formation in soft agar. AsPC‐1 cells were cultured in the presence of CHIR99021 (C9) and/or trametinib (tram) for 7 days. *p < 0.05, **p < 0.01 in Dunnett's test. ^# p < 0.05, ^## p < 0.01 in Dunnett's test. Error bars, SD. HPF, High‐power field. (B) Suppression of AsPC‐1 colony formation in soft agar upon C9 and tram. Scale bars, 200 μm. (C) Immunohistochemical analyses of AsPC‐1 orthotopic xenografts. Representative staining images from the vehicle, C9, tram, or combinations (CT) are displayed. Scale bars, 50 μm. (D) Tram and CT decreased the proportion of cells positive for Ki‐67. *p < 0.05 and **p < 0.01 in Dunnett's test. (E) Tram and CT decreased the population of cells expressing phosphorylated ERK (pERK). Legend the same as in (D). (F) Phospho‐antibody array analysis of AsPC‐1 xenografts. Fold changes in phosphorylation signal ratio (vehicle vs. CT) were visualized. The phosphorylation signal ratio was computed as the ratio of phosphoprotein signal to non‐phosphoprotein signal. Each dot represents one phosphoprotein. Green dots, significantly down‐regulated phosphoproteins in the CT group. (G) Heterozygosity of polo rescued the lethality of 4‐hit flies. *p < 0.05 in Student's t‐test. Legend the same as in Figure 1B. (H) A schematic representation of the mechanism of action of CT.

Subsequently, we conducted immunohistochemical analyses of the xenografts to determine the mode of action of the CT combination further. The combination caused a significant reduction in the level of Ki‐67, a proliferation marker, compared with vehicle control or monotherapy, which was consistent with the observations in the xenograft assay (Figure 6C,D). Furthermore, we confirmed a reduced level of phosphorylated ERK (pERK) as a marker of MAPK pathway activity upon trametinib and CT treatments (Figure 6C,E).

Conversely, GSK3 plays a pivotal role in suppressing the Wnt/β‐catenin pathway by phosphorylating β‐catenin. Even though loss of antigen‐presenting cell (APC) function drives colorectal tumorigenesis, the precise functions of GSK3 remain unclear in PDAC. To determine the activity of the Wnt/β‐catenin pathway in 4‐hit PDAC, we conducted immunohistochemistry for β‐catenin on the orthotopic xenografts and its neighboring normal pancreatic tissue. We found that β‐catenin was present primarily on the plasma membrane of normal ductal cells, whereas β‐catenin accumulated in the cytoplasm and/or nucleus in AsPC‐1 cells (Figure S5A,B). We also examined C9‐treated AsPC‐1 xenografts and observed no changes in the subcellular localization of β‐catenin upon C9 treatment (Figure S5C). These results suggest that the Wnt/β‐catenin pathway is active in 4‐hit PDAC and that GSK3 inhibition has no effects on the pathway activity in this type of cancer.

Furthermore, we conducted a phospho‐antibody array analysis comparing vehicle‐ and CT‐treated xenograft tumors, and identified Polo‐like kinase 1 (PLK1), LCK, ATF2, and p38α as proteins exhibiting reduced phosphorylation levels (Figure 6F, Table S8). Previous publications have suggested a crucial role for PLK1 in PDAC pathogenesis. Namely, PLK1 is overexpressed in PDAC compared with normal pancreas and correlates with poor prognosis in patients. ⁴⁶ Additionally, PLK1 inhibition suppressed PDAC cell proliferation by interfering with mitosis and inducing apoptosis. ⁴⁷ In fact, introducing a heterozygous mutation in polo (fly ortholog of human PLK1) into 4‐hit flies rescued their lethality, implying a pivotal role for PLK1 in 4‐hit‐dependent transformation (Figure 6G, Table S4). AURKB, the therapeutic target we recently identified in PDAC, ²² and PLK1 are both pivotal mitotic regulators. Therefore, it is intriguing to validate the effects of combining a MEK inhibitor with a well‐established mitotic inhibitor on PDAC. We thus tested the MEK inhibitor trametinib combined with PTX, and observed a pronounced inhibitory effect on the growth of AsPC‐1 spheroids (Figure S6B,C). Collectively, these results suggest that CT exerts its suppressive effects on PDAC cell proliferation by suppressing PLK1 activity in GSK3‐ and KRAS‐dependent PDAC (Figure 6H).

4. DISCUSSION

To understand the mechanisms underlying PDAC pathogenesis and to develop novel therapeutic strategies for this devastating disease, we recently generated the first animal model to generate endogenous tumors, 4‐hit Drosophila. Notably, this model enables comprehensive screening of the entire kinome to identify potential therapeutic targets and rapid testing of therapeutic candidates within a whole‐body context. ²² In this study, we made a significant discovery that GSK3 is also a novel therapeutic target for PDAC, demonstrating that the CT combination exhibited marked anti‐tumor effects across multiple models. These results further confirmed the usefulness of the 4‐hit model, in conjunction with mammalian models, as a useful platform for advancing PDAC research.

Currently, the standard treatment for PDAC includes chemotherapy regimens such as FOLFIRINOX or gemcitabine in combination with nab‐PTX, which underscores the importance of combination therapies. In a previous clinical trial targeting PDAC, however, the combination of trametinib with standard gemcitabine therapy yielded only marginal benefits compared with gemcitabine monotherapy. ⁴⁸ One of the pivotal advantages of our approach lies in its ability to identify novel therapeutic candidates, including combination therapies, through high‐throughput screening in a whole‐body context. ²² The novel seed C9 that we identified is a highly selective GSK3 inhibitor initially developed as an anti‐diabetic compound. ⁴⁹ Curiously, recent reports have indicated its effectiveness as an anticancer agent in xenograft models of non–small‐cell lung cancer ⁵⁰ and epithelioid sarcoma. ⁵¹ In our assays, C9 monotherapy showed only marginal effects on AsPC‐1 xenograft growth, whereas trametinib significantly potentiated its efficacy, suggesting that simultaneous inhibition of GSK3 and MEK is critical for suppressing 4‐hit PDAC. Of note, in our recent study, the combination of the AURKB inhibitor BI‐831266 and trametinib failed to suppress the AsPC‐1 xenograft. ²² This is possibly because AsPC‐1 cells carry the R465C missense mutation in FBXW7, an F‐box family member involved in AURKB ubiquitination for degradation. ⁵² Given that CT has shown efficacy against AsPC‐1 xenografts, we speculate that this treatment is useful to overcome the resistance of PDAC against the simultaneous targeting of AURKB and MEK. Furthermore, our immunohistochemical analysis of PDAC tissues demonstrated that more than 30% of patients were positive for GSK3β (Figure 2A,B). This frequency is much higher than the proportion with microsatellite instability and high tumor mutation burden eligible for immune checkpoint inhibitors approved recently as a second‐line treatment. ⁵³ , ⁵⁴ Collectively, within the context of clinical implementation, we anticipate that the utilization of CT treatment will enhance potential avenues for PDAC therapy.

Over 90% of PDAC contain activation mutations in the KRAS gene, which promote the proliferation and survival of PDAC cells. ⁵⁵ , ⁵⁶ Over the past decades, mutant KRAS has been considered an elusive target for drug intervention because of its high affinity for GTP and its surface lacking suitable drug‐binding sites. Recent efforts have generated the irreversible KRAS^G12C‐specific inhibitors sotorasib (AMG510) and adagrasib (MRTX849) for the treatment of non–small‐cell lung cancer ⁵⁷ , ⁵⁸ as well as the KRAS^G12D‐specific inhibitor MRTX1133 as a potential anti‐PDAC drug. ⁵⁹ , ⁶⁰ However, resistance to these treatments has emerged as an issue that limits their therapeutic benefits. ⁶¹ One solution to overcome this limitation is to leverage synthetic lethal interactions within the cancer. In fact, a previous study successfully identified the hypersensitivity of KRAS‐active cancer cells to PLK1 inhibition. ⁶² Additionally, a combination of a PLK1 inhibitor and a ROCK inhibitor suppressed the growth of patient‐derived xenografts of KRAS‐active lung cancer, resulting in significantly prolonged mouse survival. ⁶³ These results are consistent with our hypothesis that PLK1 promotes oncogenic signaling in KRAS‐dependent PDAC (Figure 6H). Considering also our recent seed inhibiting one of the key mitotic regulators AURKB, ²² it is intriguing to speculate that targeting the cell cycle machinery offers a useful strategy to treat PDAC. As expected, dual treatment with trametinib and the mitotic inhibitor PTX effectively suppressed the PDAC cell proliferation (Figure S5B,C). A previous report demonstrated that combining trametinib with nab‐PTX suppresses the growth of AsPC‐1 xenograft by inducing apoptosis in cancer cells. ⁶⁴ Therefore, we propose therapeutic advantages associated with concurrent targeting of MEK and the mitotic kinases AURKB and PLK1. Indeed, the utilization of PLK1 inhibitors in the treatment of PDAC has garnered considerable attention. Despite the ineffectiveness observed in combining the PLK1 inhibitor rigosertib with gemcitabine in patients with metastatic PDAC, ⁶⁵ , ⁶⁶ it is noteworthy that alternative PLK1 inhibitors have been developed, with the imminent prospect of forthcoming clinical trials.

We recognize that one of the limitations of this study is the lack of stromal components in flies, such as acquired immunity, which can affect the therapeutic outcomes of chemical treatments. However, it is encouraging that our seed CT identified in the fly screening successfully suppressed the growth of human PDAC cells in mice. Therefore, it would be intriguing to test the efficacy of CT in combination with stromal modulators, including immune checkpoint inhibitor drugs developed in mammalian models. Consistent with this idea, inhibition of PLK1 was reported to sensitize PDAC xenografts to immune checkpoint therapy through the activation of anti‐tumor immune responses. ⁶⁷ Deciphering the modes of PLK1 regulation by GSK3 and/or MEK, as well as the roles of the other three effectors identified in the antibody array, will provide further understanding of the mechanisms of PDAC development, which should accelerate drug discovery.

In summary, we identified concurrent targeting of GSK3 and MEK as a novel therapeutic strategy to treat PDAC, which underscores the versatility and utility of 4‐hit Drosophila. Rational targeting of multiple targets according to the results obtained in whole‐body screening can open a new horizon in the development of PDAC therapeutics.

AUTHOR CONTRIBUTIONS

Junki Fukuda: Conceptualization; data curation; formal analysis; investigation; methodology; writing – original draft; writing – review and editing. Shinya Kosuge: Formal analysis; investigation; writing – review and editing. Yusuke Satoh: Investigation; writing – review and editing. Sho Sekiya: Investigation; writing – review and editing. Ryodai Yamamura: Formal analysis; investigation; writing – review and editing. Takako Ooshio: Investigation; writing – review and editing. Taiga Hirata: Investigation; writing – review and editing. Reo Sato: Investigation; writing – review and editing. Kanako C. Hatanaka: Investigation; writing – review and editing. Tomoko Mitsuhashi: Investigation; writing – review and editing. Toru Nakamura: Investigation; writing – review and editing. Yoshihiro Matsuno: Investigation; writing – review and editing. Yutaka Hatanaka: Investigation; writing – review and editing. Satoshi Hirano: Investigation; supervision; writing – review and editing. Masahiro Sonoshita: Conceptualization; data curation; funding acquisition; investigation; project administration; supervision; validation; writing – original draft; writing – review and editing.

FUNDING INFORMATION

Japan Society for the Promotion of Science grants 19H05412 and 20H03524 (MS). Japan Agency for Medical Research and Development grant JP20cm0106273 (MS). Princess Takamatsu Cancer Research Fund grant (MS). Takeda Science Foundation grant (MS).

CONFLICT OF INTEREST STATEMENT

Masahiro Sonoshita is a shareholder of FlyWorks.

ETHICS STATEMENT

Approval of the research protocol by an Institutional Reviewer Board: Patient specimens for tissue microarray were used in accordance with the protocols approved by the Hokkaido University Ethical Review Board for Life Science and Medical Research (approval number: 019–0154).

Informed Consent: Written informed consent to provide surgical samples for tissue microarray was obtained from all the patients before surgery.

Registry and the Registration No. of the study/trial: N/A.

Animal Studies: All animal studies were conducted using protocols approved by the Hokkaido University Safety Committee on Genetic Recombination Experiments (approval numbers: 2019‐007 and 2022‐029) and the Hokkaido University Animal Research Committee (approval numbers: 19‐0121, 2022‐0117, and 2022‐0118).

Supporting information

Data S1.

Table S1.

Figure S1.

Fukuda J, Kosuge S, Satoh Y, et al. Concurrent targeting of GSK3 and MEK as a therapeutic strategy to treat pancreatic ductal adenocarcinoma. Cancer Sci. 2024;115:1333‐1345. doi: 10.1111/cas.16100