Increased mitochondria are responsible for the acquisition of gemcitabine resistance in pancreatic cancer cell lines

Abstract

Pancreatic ductal adenocarcinoma has a particularly poor prognosis as it is often detected at an advanced stage and acquires resistance to chemotherapy early during its course. Stress adaptations by mitochondria, such as metabolic plasticity and regulation of apoptosis, promote cancer cell survival; however, the relationship between mitochondrial dynamics and chemoresistance in pancreatic ductal adenocarcinoma remains unclear. We here established human pancreatic cancer cell lines resistant to gemcitabine from MIA PaCa‐2 and Panc1 cells. We compared the cells before and after the acquisition of gemcitabine resistance to investigate the mitochondrial dynamics and protein expression that contribute to this resistance. The mitochondrial number increased in gemcitabine‐resistant cells after resistance acquisition, accompanied by a decrease in mitochondrial fission 1 protein, which induces peripheral mitosis, leading to mitophagy. An increase in the number of mitochondria promoted oxidative phosphorylation and increased anti‐apoptotic protein expression. Additionally, enhanced oxidative phosphorylation decreased the AMP/ATP ratio and suppressed AMPK activity, resulting in the activation of the HSF1–heat shock protein pathway, which is required for environmental stress tolerance. Synergistic effects observed with BCL2 family or HSF1 inhibition in combination with gemcitabine suggested that the upregulated expression of apoptosis‐related proteins caused by the mitochondrial increase may contribute to gemcitabine resistance. The combination of gemcitabine with BCL2 or HSF1 inhibitors may represent a new therapeutic strategy for the treatment of acquired gemcitabine resistance in pancreatic ductal adenocarcinoma.

Mitochondria were increased in acquired GEM‐resistant pancreatic cancer cell lines, which induced oxidative phosphorylation (OXPHOS) and upregulation of BCL2 and BCL‐xL anti‐apoptotic proteins. As a result of OXPHOS, the AMP/ATP ratio and AMP‐activated protein kinase (AMPK) activity were decreased, which activated the heat shock factor 1 (HSF1)–heat shock protein (HSP) pathway to promote environmental stress tolerance.

Abbreviations

AMPK

AMP‐activated protein kinase

FIS1

mitochondrial fission 1 protein

HSF1

heat shock factor1

HSPs

heat shock proteins

MFF

mitochondrial fission factor

OXPHOS

oxidative phosphorylation

PDAC

pancreatic ductal adenocarcinoma

TCA

tricarboxylic acid

TFAM

mitochondrial transcription factor

1. INTRODUCTION

Among all malignancies, PDAC has a particularly poor prognosis. In 2022, the American Cancer Society reported a 5‐year survival rate of approximately 10%. ¹ Surgical resection is currently the only curative PDAC treatment. However, despite diagnostic advances, ² it is often detected at an advanced stage; thus, chemotherapy is used to improve the prognosis. Gemcitabine (GEM) has been an important chemotherapeutic drug for pancreatic cancer since first reported by Burris et al. in 1997. ³ GEM plus nab‐paclitaxel ⁴ and GEM plus S‐1 ⁵ therapies are also used. Once the efficacy of such GEM‐based regimens was demonstrated, their use quickly spread worldwide. However, despite showing some effectiveness, the results were unsatisfactory owing to the development of drug resistance. The 5‐year survival rate for distant metastases of PDAC is below 5%. Overcoming GEM resistance is expected to improve this poor outcome.

Various GEM resistance causes have been reported, including inhibition of drug absorption through the development of dense stromal structures unique to pancreatic cancer, ⁶ , ⁷ adaptation to the environment through cell metabolic changes, ⁸ , ⁹ resistance‐conferring genetic mutations, such as those that result in an increased anti‐apoptotic protein expression, ¹⁰ , ¹¹ , ¹² and proliferation of cancer stem cells, a cell subgroup that differs from most cancer cells in its chemotherapy resistance. ¹³ , ¹⁴

Mitochondria, organelles involved in cellular energy production and cell death regulation, through processes such as autophagy and apoptosis, ¹⁵ are involved in chemoresistance of various carcinomas, including melanoma, ovarian cancer, head and neck squamous cell carcinoma, and diffuse histiocytic lymphoma. ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ Mitochondria constantly change in number and morphology through repeated fusion and division. These mitochondrial dynamics ²⁰ are related to various diseases. ²¹

In cancer cells, the energy and macromolecular compounds required for rapid expansion are supplied by glycolysis, the mitochondrial tricarboxylic acid cycle, and OXPHOS. ⁸ , ²² In addition to metabolic plasticity, these processes are essential for cancer cell survival because they contribute to the maintenance of the cellular redox balance, ion homeostasis, including Ca²⁺ signaling, and apoptosis regulation. ²³ , ²⁴

Mitochondrial dynamics associated with anticancer drug exposure and acquisition of resistance have been reported to increase the number of mitochondria in cisplatin‐resistant ovarian cancer. ²⁵ In pancreatic cancer, increased mitochondria in cancer stem cells ²⁶ , ²⁷ and decreased TFAM expression in GEM‐resistant cells ²⁸ have been reported; however, few reports have evaluated mitochondrial dynamics before and after GEM‐resistance acquisition by GEM‐sensitive pancreatic cancer cell lines.

Reports showing increased mitochondria in cancer cells have also observed concurrently increased OXPHOS, suggesting that metabolic changes had occurred. Because AMPK, which has various roles in the cell, is regulated by metabolic changes, ²⁹ , ³⁰ evaluation of AMPK changes may help elucidate the AMPK‐related resistance mechanisms.

The involvement of anti‐apoptotic proteins, which are expressed and function in the mitochondria, ³¹ in chemoresistance has also been reported. The manner in which mitochondrial dynamics change after the chemoresistance acquisition and the way these proteins are affected by this change are noteworthy.

In this study, we aimed to clarify the relationship between mitochondrial dynamics and GEM resistance in human pancreatic cancer cell lines (MIA PaCa‐2 and Panc1) by comparing mitochondrial dynamics, metabolic changes, and the expression of proteins involved in chemoresistance before and after GEM‐resistance acquisition, to propose a new therapeutic strategy.

2. MATERIALS AND METHODS

2.1. Cells and cell culture

Authenticated human pancreatic cancer cell lines were purchased: MIA PaCa‐2 (lot no. 01272017; JCRB) and Panc1 (ID TKG 0606; RIKEN). None of the hosts had GEM‐resistant pancreatic cancers. Additionally, PK59 (ID.TKG 0492; RIKEN) from a GEM‐resistant pancreatic cancer patient was used. MIA PaCa‐2 cells were routinely cultured in DMEM (low‐glucose; WAKO) containing 4.5 g/L glucose. Panc1 and PK59 were routinely cultured in RPMI medium (WAKO) containing 4.5 g/L glucose. Both media were supplemented with 10% fetal bovine serum and 1% penicillin–streptomycin. GEM‐resistant MIA PaCa‐2 and Panc1 cells were obtained by treating parental MIA PaCa‐2 and Panc1 cells with increasing concentrations of GEM (WAKO), ranging from 12.5 nM to a final concentration of 50 nM (MIA PaCa‐2) and 100 nM (Panc1). After establishing GEM resistance, the cells were passaged at least 10 times with the final concentration of GEM (Figure 1A). GEM‐resistance acquisition was confirmed through viability assays. The parental cells were concurrently cultured.

FIGURE 1.

Establishment of gemcitabine (GEM)‐resistant cells. (A) MIA PaCa‐2 and Panc1 cells were treated with increasing doses of GEM (12.5 nM to 50 or 100 nM, respectively). (B) Cell survival curves for each cell line after 72‐h treatment with the indicated doses of GEM (n = 5, mean ± SD).

2.2. Viability assay and IC50 calculation

For cytotoxicity assays, 1500–3000 cells in 100 μL culture medium were seeded into each well of a 96‐well plate and 2.5–10,000 nM GEM or placebo was added to the culture medium, 24 h after seeding. Next, 10 μL CCK8 solution (Dojindo) was added to each well and plates were incubated for 72 h. After 2–4 h of incubation, the absorbance at 450 nm was measured using a Gen5 Microplate Reader (BioTek). Cell viability was calculated as relative viability using absorbance values referenced to the untreated control. Inhibition experiments were performed using the BCL2 inhibitor ABT263 (Cayman Chemical), the HSF1 inhibitor DTHIB (Selleck Chemicals), and the AMPK inhibitor BAY3827 (Selleck Chemicals).

2.3. Metabolic assays

Glycolysis and OXPHOS are the most important metabolic pathways in all cell types. Therefore, we examined related metabolic changes before and after GEM‐resistance acquisition. The oxygen consumption rate (OCR) of each cell line and the extracellular acidification rate (ECAR) of glycolytic metabolism were determined using a Seahorse XFp bioenergetic analyzer with the XF Cell Mito Stress Test Kit (Agilent Technologies), as recommended. To standardize the cell number at the measurement time‐point, 40,000 cells/well were plated in XFp culture plates using CELL‐TAK (Corning). The cells were cultured in a CO₂‐free incubator for 1 h in the measurement medium before analysis. Analyses were performed using Wave Desktop Software version 2.6.1.53 (Agilent Technologies).

2.4. Quantitative real‐time RT‐PCR

Quantitative RT‐PCR (qRT‐PCR) was performed to evaluate the expression of apoptotic genes associated with drug resistance. MIA PaCa‐2 and Panc1 cells were cultured and harvested at ~80% confluency. Total RNA was prepared using a NucleoSpin RNA Plus Kit (Macherey‐Nagel), and cDNA was prepared using PrimeScript RT Master Mix (Takara Bio Inc.). qRT‐PCR was performed using a Mastercycler ep Realplex (Eppendorf). The gene‐specific primers used for TB Green (Takara Bio Inc.) are detailed in Table S1.

2.5. mtDNA content

Total DNA was isolated using the QIAmp DNA kit (Qiagen) according to the manufacturer's instructions. Mitochondrial DNA copy number was estimated vi qRT‐PCR using the Human Mitochondrial DNA (mtDNA) Monitoring Primer Set (Takara Bio Inc.) consisting of NADH dehydrogenase subunit 1 (ND1), ND5, solute carrier organic anion transporter family member 2B1 (SLCO2B1), and serpin family A member 1 (SERPINA1).

2.6. Mitochondria isolation

Mitochondrial fractions were isolated using the Mammalian Mitochondria Isolation Kit for Tissue and Cultured Cells (BioVision) according to the manufacturer's instructions.

2.7. Measurement of AMP and ATP

AMP and ATP levels were determined using AMP colorimetric (BioVision) and ATP (Dojindo) assay kits, respectively, and were normalized to cell number.

2.8. Statistical analysis

Data were obtained from three or more independent experiments. The collected data were statistically evaluated using JMP software version 16 (SAS Institute) and Microsoft 365 Excel (Microsoft Corp.). Data were expressed as the mean ± SD. Statistical analysis was performed using Student's t test. Statistical significance was set at p < 0.05.

For transmission electron microscopy analysis, western blotting, and xenograft experiments, please refer to Appendix S1.

3. RESULTS

3.1. Establishment of GEM‐resistant human pancreatic cancer cells

Both the MIA PaCa‐2 and Panc1 human pancreatic cancer cell lines acquired resistance to repeated long‐term GEM treatment (Figure 1A). A comparison of IC₅₀ values after 72 h of incubation showed that GEM resistance was ~27‐fold greater in MIA PaCa‐2‐R than in MIA PaCa‐2 cells. The viability of Panc1‐R cells did not reach the IC₅₀, even at GEM concentrations 50‐fold higher than that of Panc1. IC₅₀ values were 24, 648, 236, and >10,000 nM GEM for MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R, respectively.

The viability of these GEM‐resistant cells was compared with that of PK59 cells established from patients with GEM‐resistant pancreatic cancer (Figure 1B), which indicated that our resistant strain had the same or slightly lower resistance than the cell lines established from GEM‐resistant hosts.

3.2. Metabolic assay showed high OXPHOS capacity in GEM‐resistant cells

Glycolytic reserve did not change in either MIA PaCa‐2 or Panc‐1 cells from before to after GEM‐resistance acquisition (Figure 2A). In OXPHOS evaluations, spare capacity and maximal respiration, as calculated using OCR, were both increased after resistance acquisition (p < 0.05; Figure 2B). This indicated that, under ATP‐deficient conditions, the resistant cells were more capable than the parent cells of producing ATP by OXPHOS.

FIGURE 2.

Gemcitabine (GEM)‐resistant pancreatic cancer cells show increased OXPHOS capacity compared to the parental cells. (A) Extracellular acidification rate (ECAR) of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R. n = 5, each. Mean ± SD, *p < 0.05. (B) oxygen consumption rate (OCR) of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R. n = 5, each. Mean ± SD, *p < 0.05.

3.3. Total amount of mitochondria is increased in GEM‐resistant cells

We examined whether enhanced OXPHOS was owing to the increased metabolic capacity of individual mitochondria. The mitochondrial fraction was isolated and examined for respiratory chain complex expression, using the same number of mitochondria, but no difference was observed (Figure 3A).

FIGURE 3.

Metabolic function of isolated mitochondria were unchanged and total number of mitochondria is increased in GEM‐resistant cells. (A) Western blot (respiratory chain complexes and COX4) of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R. (B) mitochondrial mtDNA copy number normalized by nuclear DNA content in MIA PaCa‐2 and MIA PaCa‐2‐R. n = 5, each. Mean + SD, *p < 0.05. (C) mitochondrial mtDNA copy number normalized by nuclear DNA content in Panc1 and Panc1‐R. n = 3, each. Mean + SD, *p < 0.05. (D) Representative electron microscopic images of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R. Scale bars: 2.0 μm. Arrows point to mitochondria. (E) The graph shows the percentage of mitochondria per cytosol of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R. n = 3, each. Mean + SD, *p < 0.05.

As mitochondrial OXPHOS function was comparable before and after GEM‐resistance acquisition, we investigated whether the increase in OXPHOS was because of an increase in total mitochondrial mass. The amount of mtDNA normalized to invariant nuclear DNA revealed that GEM‐resistant MIA PaCa‐2‐R and Panc‐1‐R cells had significantly higher mtDNA copy numbers (p < 0.05; Figure 3B,C). Additionally, the mitochondria were observed using electron microscopy and were clearly visible (Figure 3D). The mitochondrial percentage in the cytoplasm was measured and was found to be higher in resistant cells (p < 0.05; Figure 3E). These results indicated that the mitochondrial number was increased in GEM‐resistant cells.

3.4. Evaluation of mitosis indicated decreased mitophagy in resistant cells

To investigate the mechanism through which the mitochondrial number increased in resistant cells, we focused on mitochondrial mitosis, mitophagy, and mitochondrial autophagy. The mitochondrial division site reportedly changes the subsequent course of the mitochondria. The central mitochondrial division has been reported to increase the mitochondrial number, whereas peripheral mitosis is associated with mitophagy. ³² Expression of the MFF promotes central mitochondrial division, whereas FIS1 regulates peripheral mitochondrial division. In our study, parental cells exhibited higher FIS1 expression than that of the resistant cells, suggesting increased peripheral division in the parental cells. MFF expression was similar between the resistant and parental strains. DRP1 expression, which reflects the total number of mitochondrial divisions, was similar between MIA PaCa‐2 and MIA PaCa‐2‐R cells, whereas its expression in Panc1‐R was upregulated compared with that in Panc‐1 cells (Figure 4A). These results indicated that, at least in Panc1 cells, mitochondrial mitosis increased after resistance acquisition. Interestingly, aberrant mitochondrial mitosis was reduced after GEM‐resistance acquisition.

FIGURE 4.

Increased number of mitochondria in resistant cells may be associated with decreased mitophagy. (A) Western blot (FIS1, MFF, DRP and β‐Actin) of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R. Graphs show FIS1, MFF and DRP1 protein expression normalized to β‐actin of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. n = 5, each. Mean + SD, *p < 0.05. (B) Western blot (LC3 and β‐Actin) of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. Graphs show LC3IILC3I ratio of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. n = 5, each. Mean + SD, *p < 0.05. (C) Western blot (PINK1, Parkin, SQSTM1/p62, Optineurin, NDP52 and β‐Actin) of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. Graphs show PINK1, Parkin, SQSTM1/p62, Optineurin and NDP52 protein expression normalized to β‐actin of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. n = 5, each. Mean + SD, *p < 0.05. (D) Western blot (BNIP3, BNIP3L/Nix and β‐actin) of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. Graphs show BNIP3 and BNIP3L/Nix protein expression normalized to β‐actin of MIA PaCa‐2 and MIA PaCa‐2‐R cells. n = 5, each, and Panc1 and Panc1‐R cells. n = 3, each. Mean + SD, *p < 0.05.

Expression of LC3, an autophagy marker, was also examined. Although LC3‐II expression tended to increase in the resistant strains, differences were not significant (Figure 4B). In terms of mitophagy‐related factors, PINK1 was clearly elevated in GEM‐resistant strains. However, parkin was not identified. In MIA PaCa‐2, optineurin was clearly decreased in the resistant strain. NDP52, and phosphorylated ubiquitin levels were similar between the parental and resistant strains. SQSTM1/p62 levels were elevated in the MIA PaCa‐2 but decreased in the Panc1 parental strain (Figure 4C).

In contrast, BNIP3 expression was similar in MIA PaCa‐2 and MIA PaCa‐2‐R cells, whereas BNIP3L/NIX was upregulated in the parental strain. In Panc1 cells, BNIP3 was markedly upregulated in the parental strain, whereas BNIP3L/NIX was not significantly different between Panc1 and Panc1‐R cells; however, the resistant strain was more upregulated (Figure 4D).

3.5. Increased number of mitochondria leads to GEM resistance by increasing anti‐apoptotic proteins

The possibility that apoptosis‐related proteins contributed to GEM resistance was also examined. The mRNA expression of apoptosis‐related genes was evaluated using real‐time qRT‐PCR before and after GEM‐resistance acquisition. Expression of the anti‐apoptotic gene BCL2 was enhanced in MIA PaCa‐2‐R cells compared with that in MIA PaCa‐2 cells. BCL2 and BCL2L1 (BCL‐xL) mRNA expression levels were higher in Panc1‐R than in Panc1 cells. mRNA levels of the apoptotic genes BAX, BAD, and BIM were similar before and after resistance acquisition (p < 0.05; Figure 5A).

FIGURE 5.

Evaluation of apoptosis‐related proteins. The expression of anti‐apoptotic proteins was enhanced in resistant cells. (A) Quantitative reverse transcription polymerase chain reaction (qRT‐PCR) of apoptosis‐related protein genes from MIA PaCa‐2 and MIA PaCa‐2‐R. n = 5, each, and Panc1 and Panc1‐R. n = 3, each. Mean ± SD, *p < 0.05. (B) Western blotting (BCL2, BCL‐xL, BAX and β‐actin) of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. Graphs show anti‐apoptotic protein (BCL2, BCL‐xL) expression normalized to β‐Actin in MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. n = 5, each. Mean ± SD, *p < 0.05. (C) Western blotting (BCL2, BCL‐xL and COXIV) of mitochondrial proteins isolated from MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. Graphs show anti‐apoptotic protein (BCL2, BCL‐xL) expression normalized to COXIV in MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. n = 5, each, and Panc1 and Panc1‐R. n = 3, each. Mean ± SD, *p < 0.05.

Western blotting showed increased protein expression of BCL2 in MIA PaCa‐2‐R cells and of BCL2 and BCL‐xL in Panc1‐R cells but no changes in BAX expression in either cell line (p < 0.05; Figure 5B). Thus, as shown in previous reports, we observed an increased expression of anti‐apoptotic proteins in GEM‐resistant cells.

We then examined whether the expression of anti‐apoptotic proteins was more prevalent in the mitochondria of resistant cells; however, no change was observed in anti‐apoptotic protein expression (Figure 5C). These results indicated that the amount of anti‐apoptotic proteins present in the mitochondria did not change before or after resistance acquisition.

We then tested the effects of BCL inhibitors on tumor growth inhibition. No difference in cell viability was observed between MIA PaCa‐2 and MIA PaCa‐2‐R cells when the BCL inhibitor ABT263 was used alone; however, it was more effective at inhibiting the proliferation of Panc1‐R cells than that of Panc1 cells (Figure 6A,B). PK59 cells expressed more anti‐apoptotic proteins than did the other cell lines. ABT263 markedly inhibited the proliferation of this cell line (Figure S1, Figure 6C). Chou–Talalay synergy analysis showed several drug combinations with a combination index (CI) ³³ <1, indicating synergy. For MIA PaCa‐2‐R cells, the CI was <1 for 500 and 1000 nM GEM when combined with 5.0 μM Direct Targeted HSF1 Inhibitor, SISU‐102 (DTHIB). The most efficient combination was 1000 nM GEM/5.0 μM DTHIB (CI = 0.722). None of the combinations showed synergistic effects on Panc1‐R cells (Figure 6D,E). The strongest synergistic effect in combination with GEM was seen in PK59 cells (Figure 6F). These results suggested that the expression of anti‐apoptotic proteins was increased in resistant cells, which may have contributed to the resistance in some cell lines.

FIGURE 6.

Effect of BCL inhibitor (ABT263) on gemcitabine (GEM)‐resistant cells. Synergistic effect of the GEM combination was observed in MIA PaCa‐2‐R and PK59. (A–C) Graph shows the cell viability of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, Panc1‐R, and PK59 at the indicated doses of ABT263. n = 3, each. Mean ± SD, *p < 0.05; ns, p > 0.05. (D–F) Graph shows the combination index (CI) of MIA PaCa‐2‐R, Panc1‐R, and PK59 according to drug combination effect (Fa).

3.6. Increased ATP production due to enhanced metabolic capacity activates the AMPK–HSF1 pathway, resulting in GEM resistance

Given that the resistant cells produced more energy than their parental cells, due to increased OXPHOS associated with the increased number of mitochondria, we measured AMP/ATP levels and confirmed that energy production was indeed increased in GEM‐resistant cells (Figure 7A).

FIGURE 7.

Enhanced oxidative phosphorylation in resistant cells causes a decrease in the AMP/ATP ratio and increases HSF1 by suppressing AMPK activation. (A) AMP/ATP ratio in resistant cells normalized to that in parental cells. n = 3, each. Mean ± SD, *p < 0.05. (B) Western blotting (AMPKα, pAMPKα, Thr172, HSF1 and β‐actin) of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. Graphs show the pAMPKα/AMPKα ratio in GEM‐resistant cells normalized to that in parental cells. n = 5, each. Mean ± SD, *p < 0.05. and HSF1 protein expression normalized to β‐actin of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells. n = 5, each. Mean ± SD, *p < 0.05. (C) The graph shows the cell viability of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1 and Panc1‐R at the indicated doses of GEM plus BAY3827. n = 5, each. Mean + SD, *p < 0.05; ns, p > 0.05.

As a decrease in the AMP/ATP ratio suppresses AMPK activity, we examined changes in the AMPK–HSF1‐HSP pathway, which leads to cell survival in the presence of environmental stress in GEM‐resistant cells(Figure S2). We found a decreased AMP/ATP ratio in GEM‐resistant cells, which suppressed AMPK phosphorylation and the AMPK–HSF1–HSP pathway, leading to cell survival in the presence of environmental stresses in GEM‐resistant cells. In both resistant cell lines, we observed that AMPK phosphorylation at Thr172, reflecting its activity, was suppressed by the decreased AMP/ATP ratio, compared with the parental cells, whereas HSF1 expression was increased (Figure 7B).

Furthermore, HSP90 and HSP70 expression, regulated by HSF1, was higher in resistant than in parental cells (Figure S3). Therefore, we evaluated cell viability using an HSF1 inhibitor (DTHIB). The single‐agent effect in resistant cells was significantly higher than that in parental cells (Figure 8A,B) and showed a synergistic effect when combined with GEM in GEM‐resistant cells. For MIA PaCa‐2‐R cells, the CI was <1 for 12.5, 25, and 50 nM GEM combined with 2.5 μM DTHIB. The most effective combination was 12.5 nM GEM/2.5 μM DTHIB (CI = 0.689 (Figure 8C). In Panc1‐R cells, the CI was <1 for 500 or 1000 nM GEM combined with 2.5 μM DTHIB. The most efficient combination was 500 nM GEM/2.5 μM DTHIB (CI = 0.752) (Figure 8D). These results suggested that the AMPK–HSF1–HSP pathway is involved in GEM‐resistance acquisition. To evaluate whether inhibition of the AMPK pathway in the parental strain is beneficial for cell survival, the AMPK inhibitor BAY3827 (Figure S2) was used in combination with GEM, which resulted in improved cell viability (Figure 7C).

FIGURE 8.

Synergistic effect of combination treatment with gemcitabine (GEM) is observed in MIA PaCa‐2‐R and Panc1‐R cells. (A, B) Graph shows the cell viability of MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells at the indicated doses of DTHIB. n = 5, each. Mean ± SD, *p < 0.05. (C, D) Graph shows combination index (CI) of MIA PaCa‐2‐R and Panc1‐R according to drug combination effect (Fa). (E) Tumor xenograft images from mice treated with gemcitabine (5 mg/kg, IP, q3d) or DTHIB (5 mg/kg, IP, every day) and/or gemcitabine plus DTHIB. (F) The graph shows the relative tumor volume (RTV) and tumor weight for each subgroup in MIA PaCa‐2 and MIA PaCa‐2‐R cells. RTV was calculated using the formula: V/V0 where V is the tumor volume on a sacrifice and V0 is the volume on 1 week after tumor injection. n = 5, each. Mean ± SD, *p < 0.05.

We generated murine xenografts from MIA PaCa‐2, MIA PaCa‐2‐R, Panc1, and Panc1‐R cells, and investigated the efficacy of DTHIB in vivo. Similar to the in vitro results, the combination of GEM (5 mg/kg, once every 3 days) and DTHIB (5 mg/kg/day) showed a significant growth inhibition effect as compared to GEM alone (Figure 8E,F; Figure S4). Tumor weight also tended to decrease with the GEM and DTHIB combination as compared to that with GEM alone in resistant cell xenografts but without significant differences (Figure 8G).

4. DISCUSSION

In this study, GEM‐resistant human pancreatic cancer cell lines were generated, and mitochondrial dynamics were investigated before and after GEM‐resistance acquisition, which revealed that the mitochondrial number had increased. Anti‐apoptotic BCL2 family protein levels as well as OXPHOS were increased in the mitochondria of the resistant strain and more intracellular ATP was produced. These results suggest that the AMPK–HSF1–HSP system is activated and that cell death can be suppressed in GEM‐resistant cells.

In pancreatic cancer, the involvement of mitochondria in GEM resistance has been studied and mechanisms related to metabolic plasticity, autophagy, apoptosis, and mtDNA damage have been reported ¹² , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ ; however, few reports have discussed the role of mitochondrial dynamics in GEM resistance.

Kleele et al. reported an interesting mechanism through which mitochondrial fate depends on the fragmentation site. Peripheral mitochondrial division leads to mitophagy, whereas mitochondrial fission normally occurs when mitochondria are divided centrally. The DRP1 and MFF complexes have been reported to be responsible for central division, whereas FIS1 regulates peripheral division through lysosome mobilization. ³² In this study, FIS1 expression was clearly reduced in the GEM‐resistant strains (Figure 4A), suggesting that mitophagy was reduced. PINK1 is normally transported to the mitochondrial inner membrane, fragmented by PARL, localized to the inner membrane, and rapidly degraded. ³⁹ In depolarized mitochondria, PINK1 is transported to the inner mitochondrial membrane, is degraded by PARL, and then accumulates in the outer mitochondrial membrane. ⁴⁰ The phosphorylation of Ser65 of ubiquitin by parkin triggers mitophagy. ⁴¹ This signals ubiquitin‐binding cargo receptors, such as SQSTM1/p62, optineurin, and NDP52, to induce mitophagy. ⁴² , ⁴³ , ⁴⁴ However, because PINK1 levels are increased in resistant strains, we believe that mitochondria are depolarized but parkin is deficient; thus, mitophagy remains stagnant. Furthermore, BNIP3 and BNIP3L/NIX expression was suppressed, further indicating mitophagy suppression. Parkin expression has been reported to decrease in breast cancer, ⁴⁵ and receptor‐dependent mitophagy via BNIP3 and BNIP3L/NIX may play a central role in cancer cells. Future studies should examine the contribution of mitochondrial metabolic turnover in GEM‐resistance acquisition in pancreatic cancer.

Overexpression of anti‐apoptotic proteins has been reported to contribute to the acquisition of resistance to GEM as well as to various chemotherapies. ¹⁰ , ¹¹ , ¹² In this study, the levels of anti‐apoptotic proteins BCL2 and/or BCL‐XL were increased in resistant cells (Figure 5A,B). Interestingly, instead of increasing the levels of BCL2 family proteins in the mitochondria, the total mitochondrial BCL2 family protein amount increased with increasing mitochondrial mass (Figure 5B,C). ABT‐263 (Navitoclax), a nonselective BCL2 inhibitor used in this study, inhibits both BCL2 and BCL‐xL. ⁴⁶ The synergistic effect of ABT‐263 with GEM in GEM‐resistant cell lines not only proved one of the putative mechanisms of GEM‐resistance acquisition but may also lead to the development of a new treatment for pancreatic cancer in combination therapy with GEM.

In this study, an increase in the total mitochondrial mass also increased OXPHOS (Figures 2B and 3B–E). In pancreatic cancer stem cells, the enhancement of OXPHOS with increased mitochondria and high sensitivity to OXPHOS inhibition, and the antitumor effect of metformin, an OXPHOS inhibitor, have also been reported. ²⁶ , ²⁷ , ³⁴ However, Fujiwara et al. reported that, although the basal respiration of a GEM‐resistant cell line was high, GEM treatment reduced TFAM expression and OXPHOS, resulting in reduced ROS production and contributing to GEM resistance. ²⁸

mtDNA‐encoded genes are essential for mitochondrial OXPHOS. ⁴⁷ , ⁴⁸ In this study, no change was noted in respiratory chain complex expression in the resistant strains (Figure 3A), suggesting that no pathogenic mtDNA mutations had occurred. Therefore, we concluded that the mtDNA copy number increase upon GEM‐resistance acquisition indicated a mitochondrial number increase. This mitochondrial number increase in the GEM‐resistant cell lines likely enhances OXPHOS. Furthermore, OXPHOS capacity was clearly improved, not only in basal respiration but also in terms of reserve capacity and maximal respiration (Figure 2B), which contrasted with Fujiwara et al.'s results. ²⁸ The reason for the discrepancy is unknown; however, it is possible that the experimental conditions may play a role, because of the large discrepancies in the OCR baselines. Most likely, differences in the methods used to establish resistant cell lines and in the concentrations and types of GEM used in the experiments had an impact. In other cancers, Zampieri et al. reported increased mitochondria and OXPHOS in human ovarian cancer cell lines evaluated before and after the acquisition of cisplatin resistance, ²⁵ similar to our results. Furthermore, in this study, an increase in ATP levels was observed in GEM‐resistant cell lines. This confirmed an increase in OXPHOS levels (Figure 7A).

GEM is metabolized into active nucleosides inside the cell and induces apoptosis by inhibiting DNA strand elongation and synthesis. ⁴⁹ , ⁵⁰ Additionally, GEM has a metabolic property, called self‐potentiation, whereby inhibition of enzymes related to deoxynucleotide metabolism by GEM triphosphate (dFdCTP) and GEM diphosphate (dFdCDP) leads to the intracellular/intratumoral accumulation of GEM and its phosphorylated forms. ⁵¹ , ⁵² Wang et al. reported that HSP70 increased BCL2 expression and exhibited anti‐apoptotic effects in pancreatic cancer and that the RNA binding and regulatory functions of HSP70 are associated with drug resistance. ⁵³ Qin et al. ⁵⁴ reported that increased HSF1 expression caused GEM resistance in pancreatic cancer cell lines. Therefore, we here focused on the ATPK–HSF1–HSP pathway.

In the present study, increased ATP levels resulted in decreased AMPK levels. Consequently, HSF1 was de‐repressed, and HSP70 and HSP90 levels increased (Figure S3). These results suggested that the AMPK–HSF1–HSP pathway may contribute to GEM‐resistance acquisition by protecting cells from environmental stress. Furthermore, HSF1 inhibitors were more effective in GEM‐resistant cell lines than in the parental cell lines and showed synergistic effects with GEM in GEM‐resistant cell lines (Figure 8). However, in a study of oxaliplatin tolerance in HCC, HSF1 was reported to promote OXPHOS by suppressing AMPK. ⁵⁴ Thus, AMPK inactivates HSF1 by phosphorylation, whereas HSF1 reciprocally suppresses AMPK by physically forcing it into an inactive conformation.

This study had several limitations. First, because we used cell lines generated by administering GEM to pancreatic cells, they may have different characteristics from cell lines established from the tumors of patients who received GEM. However, because we intended to investigate the detailed changes that occur by comparing pre‐ and post‐resistance acquisition, we established GEM‐resistant cells from GEM‐sensitive cell lines and performed experiments. This was not possible in PK59 cells cultured from patient tissues. Whenever possible, we conducted experiments to determine whether similar events could be confirmed in PK59 cells. Further research is needed to elucidate the mechanism of drug resistance acquisition in pancreatic cancer using resected tissues from patients.

In human pancreatic cancer cell lines, mitochondrial dynamics are altered before and after GEM‐resistance acquisition, with increased mitochondria appearing in resistant cells. An increased mitochondrial number results in increased OXPHOS and BCL2 expression. Furthermore, increased HSP expression through the AMPK–HSF1–HSP pathway contributed to tolerance to environmental stress. The combination of GEM with BCL2 or HSF1 inhibitors may be a novel therapeutic strategy for treating GEM‐resistant pancreatic cancer.

AUTHOR CONTRIBUTIONS

Hitoshi Masuo, Satoru Miyazaki, Takahiro Yoshizawa, Kiyotaka Hosoda, Hiroki Sakai, and Hikaru Hayashi contributed to the acquisition and analysis of the experimental data and drafting of the manuscript. Koji Kubota, Akira Shimizu, Tsuyoshi Notake, and Yuji Soejima contributed to the conception and design of this study and made critical revisions related to the important intellectual content of the manuscript. All authors read and approved the final version of the manuscript for submission.

FUNDING INFORMATION

This study received no specific grants from any funding agency in the public, commercial, or not‐for‐profit sectors.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

Approval of the research protocol by the institutional review board: N/A.

Informed Consent: N/A.

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

Animal Studies: N/A.

Supporting information

Appendix S1.

Appendix S2.

Masuo H, Kubota K, Shimizu A, et al. Increased mitochondria are responsible for the acquisition of gemcitabine resistance in pancreatic cancer cell lines. Cancer Sci. 2023;114:4388‐4400. doi: 10.1111/cas.15962