Abstract

   Metabolic reprogramming is a key contributor to cancer therapeutic
   resistance. Peroxisomes are highly metabolic organelles essential for
   lipid metabolism and reactive oxygen species (ROS) turnover. Recent
   studies pointed out that targeting peroxisomal genes could be a
   promising strategy for treating therapy-resistant cells. However, the
   role of peroxisomes in CRC chemoresistance remains largely unexplored.
   This study aimed to investigate the function of peroxisomes in CRC
   chemoresistance and uncover the underlying mechanisms. Our results
   showed that the protein level of peroxisome marker PMP70 was strongly
   correlated with oxaliplatin (LOHP)-treated tumor recurrence in CRC.
   LOHP was confirmed to induce pexophagy in CRC cells, whereas
   LOHP-resistant cells maintained stable peroxisome levels and resisted
   this selective autophagy. Moreover, depletion of PMP70 significantly
   reduced the viability of resistant CRC cells in response to LOHP, both
   in vitro and in vivo. Mechanistically, PMP70 acted as a potential
   protector against excessive lipid peroxidation (LPO) in PMP70^High and
   LOHP-resistant CRC cells. Additionally, PMP70-depleted cells exhibited
   an altered metabolic profile, characterized by reduced neutral lipids,
   fewer lipid droplets (LDs), and cell cycle arrest, indicating that
   PMP70 downregulation resulted in metabolic impairment. In conclusion,
   our study unveiled the pivotal role of PMP70-mediated peroxisomal
   functions in conferring chemoresistance, particularly in the context of
   LOHP resistance in CRC.

   Subject terms: Gastrointestinal cancer, Chemotherapy, Peroxisomes

Introduction

   Metabolic reprogramming and deregulation are hallmark features of
   cancer. Alterations in lipid metabolism have gradually been proven to
   be important adaptations that support tumor progression, metastasis,
   and therapy resistance, rather than merely passive processes
   [[38]1–[39]3]. Several studies have validated that enhanced
   lipogenesis, lipid catabolism, and LD formation and turnover have
   emerged as key features of therapy-resistant tumor cells [[40]4–[41]8].
   Targeting aberrant lipid metabolism could serve as a promising
   therapeutic strategy for overcoming therapy resistance.

   Peroxisomes are highly metabolic organelles present in most eukaryotic
   cells, well-known for their roles in lipid metabolism [[42]9]. The very
   long chain fatty acids that have 22 or more carbons undergo β-oxidation
   in peroxisomes before being oxidized in the mitochondria
   [[43]9–[44]11]. The β-oxidation process in peroxisomes generates
   reactive oxygen species (ROS), which is mainly eliminated by catalase
   (CAT). As highly dynamic organelles, peroxisomes can modify their
   shape, size, location, and abundance in response to nutritional or
   environmental signals. For example, under hypoxic conditions (1%
   oxygen), peroxisomal biogenesis genes and plasmalogen synthesis genes
   are selectively essential for cell survival [[45]12]. Under nutrient
   stress, the physical interaction between peroxisomes and LDs increases
   significantly, leading to the release of fatty acids through lipolysis
   in white adipose tissue [[46]13]. Peroxisomal plasticity thus becomes a
   powerful weapon for cells in resisting external stresses.

   Peroxisomal dysfunction has also been implicated in tumor development.
   Notably, a reduction in peroxisome abundance has been reported in
   several neoplastic tissues, including colon carcinoma [[47]14]. Recent
   studies have shown that genes involved in peroxisomal β-oxidation, such
   as NUDT7 and ACOX1, can suppress CRC progression [[48]15, [49]16]. And
   our previous work has reported that CRC patients with more active
   peroxisome pathways had favorable overall survival [[50]17]. All these
   suggest that peroxisome potentially may act as a tumor-suppressive
   organelle in CRC. In addition, recent studies have highlighted the role
   of peroxisomal activity in influencing the response to systemic
   therapies in breast cancer and lymphoma, indicating that peroxisomes
   adapt to therapeutic stress, thereby protecting cells from
   therapy-induced cell death [[51]18–[52]20]. Therefore, these findings
   underscore the need for further investigation into the distinct
   functions of peroxisomes in different contexts, particularly in CRC.

   PMP70 is a 70 kDa protein locates on the peroxisomal membrane encoded
   by the ATP Binding Cassette Subfamily D Member 3 (ABCD3) gene. It is
   the most enriched peroxisomal membrane protein and the first observed
   peroxisomal ATP Binding Cassette(ABC) transporter. It participates in
   the peroxisomal import of dicarboxylic acids, branched-chain fatty
   acids, and C27 bile acid intermediates [[53]21]. PMP70 is commonly
   recognized as a specific marker of peroxisomes. Clinical evidence
   showed that ABCD3 mRNA and protein levels are reduced in CRC tumor
   tissues, and low ABCD3 mRNA levels are correlated with poor prognosis
   in CRC patients [[54]22], suggesting a potential tumor-suppressive role
   for PMP70 in CRC. However, ABCD3 amplification has been observed in 19
   drug-resistant cancer cell lines [[55]23]. These findings highlight the
   potential dual role of PMP70, which may act as a tumor suppressor in
   CRC while also being involved in adaptive mechanisms under therapeutic
   pressure. This duality underscores the need to investigate its specific
   functions in CRC, particularly in the context of chemoresistance.

   In this work, we found that peroxisome abundance and PMP70 level were
   associated with LOHP treatment recurrence in CRC patients. Knockdown of
   PMP70 significantly overcame LOHP resistance in CRC cells. Our findings
   unveil that peroxisomes and PMP70 play a significant role in conferring
   chemoresistance in CRC.

Materials and methods

Human CRC specimens and RNA-seq

   Human colorectal cancer specimens were obtained from the Department of
   Colorectal and Anal Surgery at Zhongnan Hospital of Wuhan University.
   The tissue sections were stained with H&E and diagnosed through
   pathological examination. Our study was approved by the Ethics
   Committee of Zhongnan Hospital of Wuhan University (No.
   2020106,2020-05). All patients signed written informed consent forms.
   Clinical characteristics of our CRC RNA-seq cohort are shown in
   Supplementary Table [56]1. RNA samples from clinical CRC tissues and
   cells (n = 3) were assessed for integrity using RNA Nano 6000 (Agilent
   Technologies, CA, USA) and then subjected to library construction. The
   libraries were sequenced on Illumina Novaseq 6000 using RNA sequencing
   technology provided by Novogene (Beijing, China).

Cell culture

   Human HCT116, HCT8, HT29, DLD1, SW620, SW480, and RKO cells were
   obtained from the American Type Culture Collection (ATCC) and we
   cultured them in ATCC-recommended medium supplemented with 10% fetal
   bovine serum (Gibco, NY, USA) and 1% penicillin-streptomycin (Gibco).
   The LOHP-resistant cell lines, HCT116L and HCT8L, were developed as
   previously described [[57]24]. These cell lines were generated by
   exposing the cells to increasing concentrations of LOHP (ranging from
   0.1 μmol/L to 2 μmol/L), which allowed for the selection of
   LOHP-resistant cells capable of being cultured in 2 μmol/L LOHP. LOHP
   (HY-17371), Nec1s (Necrostatin 2 racemate, HY-14622A), and NAC
   (Acetylcysteine, HY-B0215) were obtained from MedChemExpress (MCE).
   Lip-1 (Liproxstatin-1, S7699) and Z-VAD-fmk (S7023) were purchased from
   Selleck.

Lipid ROS assay and cell cycle profiling

   Cells were seeded on 12-well plates and incubated overnight. The
   following day, cells were treated with compounds for the specified
   duration, harvested by trypsinization, and suspended in 200 μL of PBS
   containing 5 μM C11-BODIPY 581/591 (Invitrogen). After a 30-min
   incubation at 37 °C in a water bath, cells were washed three times with
   PBS. LPO was evaluated using the Beckman CytoFLEX S Flow Cytometer
   equipped with a 488 nm laser on an FL1 detector. At least 10,000 single
   cells were analyzed per well. For cell cycle profiling, cells were
   seeded in 6-well plates to achieve 70–80% confluency at the time of
   harvest. Cells were fixed in 70% ethanol and stored at –20 °C for up to
   two weeks. The cell cycle was analyzed using FxCycle™ PI/RNase Staining
   Solution (Thermo, [58]F10797) and quantified by Flow Cytometer.

Mitochondrial ROS and stress profiling

   MitoSox Red (Thermo Fisher Scientific, [59]M36008) was used to measure
   mitochondrial superoxide levels. Cells were incubated with 5 µM MitoSox
   Red in PBS at 37 °C for 30 min in the dark. After washing with PBS,
   fluorescence was measured using a flow cytometer (Ex/Em: 510/580 nm).
   Mitochondrial function was assessed using the Seahorse XF Mito Stress
   Test Kit (Agilent). Cells were seeded in XF96 plates, equilibrated with
   XF Assay Medium, and treated sequentially with oligomycin (1 µM), FCCP
   (2 µM), and rotenone/antimycin A (0.5 µM). Oxygen consumption rate
   (OCR) was measured, and parameters such as basal and maximal
   respiration were calculated using Seahorse Wave software.

Untargeted lipidomic analysis

   Lipids were extracted from the organic phase using a modified Bligh and
   Dyer method [[60]25]. The lipids were then separated and analyzed using
   a Qexactive mass spectrometer (Thermo Fisher Scientific, USA) with a
   CSH C18 column (1.7 mm × 2.1 × 100 mm, Waters, USA). Elution was
   performed using the following gradient: 0–2 min at 40–43% liquid B
   (10 mM ammonia formate, 0.1% formic acid, 90% isopropyl alcohol, and
   10% acetonitrile); 2–2.1 min at 43–50% liquid B; 2.1–7 min at 50–54%
   liquid B; followed by an increase to 70%-99% from 7 .1 to 13 min. Flow
   rate was set to 0.35 mL/min. For positive/negative ionization modes,
   the spray voltage was set to 3.8/-3.2 kV respectively while maintaining
   the aux gas heater temperature at 350 °C and capillary temperature
   constant at 320 °C. The full scan range of m/z values ranged from 200
   to 2000 with a resolution of 70,000 and AGC target for MS acquisitions
   set to 3,000,000 with maximum ion injection time of 100 ms. For
   subsequent MS fragmentation, the top three precursors were selected
   with a maximum ion injection time of 50 ms, resolution of 17,500, and
   AGC fixed at 100,000. Stepped normalized collision energy was set to
   15, 30, and 45 eV. Lipid identification and quantitation were performed
   using LipidSearch 4.1 SP2 (Thermo Fisher Scientific Inc., USA),
   followed by data scaling and normalization using metaX.

Mouse experiments

   We purchased male athymic nude mice (3 weeks old) from GemPharmatech
   (Jiangsu, China). The Zhongnan Hospital of Wuhan University
   Institutional Animal Care Animal Welfare Committee approved the
   experiments. We implanted sgNC and sgPMP70#2 HCT116L cells (3 × 10^6
   cells/0.1 ml) subcutaneously in the right side of the dorsal midline of
   the mice (8 weeks old). Once tumor volume reached 80–180 mm^3, we
   randomly assigned mice to different treatment groups and treated them
   with vehicle or LOHP (5 mg/kg, i.p., twice a week) for 20 days. After
   drug administration, we measured tumors every three days and sacrificed
   the mice at day 42 (after injection of cells).

Statistical analysis

   Peroxisome pathway score was calculated as previously described
   [[61]17]. To determine the PMP70 protein score, we categorized CRC
   patients into two groups based on WB results: PMP70-High and PMP70-Low.
   Random Forest (RF) was used, a commonly used ensemble machine learning
   algorithm [[62]26], to establish a prediction model using RNA-seq data
   from these patients. GraphPad Prism 8 was used to derive P values,
   which are indicated as *P < 0.05, **P < 0.01 and ***P < 0.001. We
   employed t test, one or two-way ANOVA, and Dunnett’s or Bonferroni’s
   post hoc test to obtain these values.

Additional materials and methods

   Further details on the methods can be found in the Supplementary
   Methods. The antibodies and siRNAs are provided in Supplementary Tables
   [63]S2–[64]S3. Original data for western blot analyses are included as
   original data 1.

Results

Peroxisome abundance correlates with LOHP-based chemotherapy in CRC

   To explore the potential role of peroxisome in CRC, we measured the
   expression of peroxisomal marker PMP70 in CRC tumor tissues and
   investigated the correlation between clinical characteristics and PMP70
   level. Western blotting analysis revealed elevated levels of PMP70 in
   CRC tumors went through LOHP-based chemotherapy compared to the
   untreated tumors (Figs. [65]1A, [66]B). To further investigate the
   correlation between peroxisomes and LOHP therapy in CRC, a systemic
   treatment database (GEO Number: [67]GSE106584) was incorporated into
   our study. Since neither ABCD3 mRNA nor genes in the peroxisome pathway
   reliably reflect peroxisome abundance, the random forest algorithm was
   employed to generate a PMP70 score, which could be used to predict the
   likelihood of peroxisome abundance in the tissues. The prediction was
   based on data from our CRC cohort, which included RNA-seq data along
   with the matched PMP70 protein levels (Fig. [68]S1A). Analyzing the
   dataset by this method, we found that within the FOLFOX treatment group
   (LOHP, 5-Fu, and Leucovorin), the PMP70 scores of the tissues from the
   recurrent CRC subgroup were remarkably higher. This trend was not
   evident in the 5-Fu/Leucovorin, radiotherapy, or untreated groups.
   These findings strongly implied that higher peroxisome abundance was
   associated with the resistance to LOHP treatment in CRC (Fig. [69]1C).

Fig. 1. Peroxisome abundance and PMP70 protein levels correlate with CRC
progression and LOHP treatment.

   [70]Fig. 1
   [71]Open in a new tab

   A, B Western blot analysis and quantification of PMP70 protein levels
   in 38 primary tumor tissues and 9 LOHP-treated CRC tumor tissues.
   Protein levels were quantified using ImageJ. C PMP70 scores in CRC
   primary tumor tissues with and without recurrence, subgrouped by
   treatment types in the [72]GSE106584 dataset. FOLFOX:
   LOHP + 5FU+Leucovorin; RT: Radiotherapy. D Western blot analysis of
   PMP70 protein levels and RNA-seq analysis of ABCD3 mRNA expression in
   38 CRC primary tumor tissues at different T stages. Protein levels were
   quantified with ImageJ. E Representative immunofluorescence images
   showing colocalization of PMP70 and CD133. F Western blot analysis of
   PMP70 and CD133 protein levels in 13 CRC primary tumor tissues. G
   Correlation analysis of PMP70 and CD133 protein levels in 38 CRC tumor
   tissues based on ImageJ quantification of Western blots. H CD133
   protein levels in CRC tumor tissues at early versus advanced T stages.
   *P < 0.05, **P < 0.01, ***P < 0.001.

   Intriguingly, there is a significant increase in PMP70 protein levels
   in tumor tissues of CRC patients with higher T stages, while the mRNA
   levels of ABCD3 do not show a similar elevation (Fig. [73]1D). Cancer
   stem cells (CSCs) were involved in tumor initiation, progression,
   recurrence and therapy resistance [[74]27]. A previous work has
   reported that peroxisome abundance was elevated during colon stem cell
   differentiation [[75]28]. Therefore, expression of CSCs marker CD133
   was examined in CRC tumor tissues. Immunofluorescence staining showed
   CD133 was expressed and partially colocalized with PMP70 in the CRC
   tumor (Fig. [76]1E). The protein levels of PMP70 were positively
   correlated with CD133 (Fig. [77]1F, G), both of which were upregulated
   in CRC tissues at advanced T stages (Fig. [78]1H).

LOHP-resistant CRC cells are resistant to pexophagy

   For generating an in vitro model of LOHP-resistant CRC, we treated
   HCT116 and HCT8 CRC cell lines with low-dose LOHP for several rounds.
   The selected cells termed as HCT116L and HCT8L, which showed high
   resistance to LOHP treatment (Fig. [79]S1B). Peroxisome-resident
   proteins, including PEX3, PEX16, CAT, and PMP70, were all upregulated
   in HCT116L comparing to HCT116 (Fig. [80]S1C). The increase of
   peroxisome, indicated by PMP70, was also proved through imaging as
   HCT116L and HCT8L had significantly higher levels of PMP70 puncta than
   parental cells. Driving for peroxisomal changes in responding to LOHP
   treatment in LOHP-resistant and parental CRC cells, we further treated
   the cells with LOHP. Unexpectedly, we observed a significant decrease
   of PMP70 puncta in HCT116 and HCT8 but an opposite increase of it in
   LOHP-resistant cells upon LOHP treatment (Fig. [81]2A, B), suggesting
   peroxisomal abundance increase might be critical to the CRC cells in
   responding to the LOHP-caused stress.

Fig. 2. LOHP-resistant CRC cells exhibit reduced LOHP-induced pexophagy.

   [82]Fig. 2
   [83]Open in a new tab

   A Representative confocal images of PMP70 puncta in parental and
   LOHP-tolerant CRC cells treated with LOHP. B Quantification of PMP70
   puncta in panel (A). C Confocal images of HCT116 cells expressing
   dKeima-SKL with or without LOHP treatment (60 μM). D Flow cytometry
   analysis and quantification of HCT116 cells expressing dKeima-SKL
   treated with LOHP (40 μM). Data represent 4–5 images and >15 cells per
   condition. E, F Confocal images and quantification of PMP70 puncta in
   HCT116 cells treated with vehicle or 10 μM CQ in the presence or
   absence of LOHP. Data represent 5–6 images and >25 cells per condition.
   G, H Colocalization of PMP70 puncta and LC3 puncta in HCT116 and
   HCT116L cells, both treated by vehicle and 60 μM LOHP. Panel (H) is the
   quantification of panel G. 5 ~ 6 images, total of 25+ cells counted per
   condition. I Ubiquitinated-PMP70 increased upon LOHP treatment in
   HCT116 than in HCT116L cells. HCT116 and HCT116L cells were transfected
   with HA-Ub for 24 h and then treated with LOHP (60 μM) for 24 h.
   Co-immunoprecipitation of PMP70 with HA, and corresponding inputs.
   *P < 0.05, **P < 0.01, ***P < 0.001.

   Peroxisome homeostasis is mainly regulated by pexophagy, a type of
   selective autophagy process that degrades peroxisomes [[84]29]. To
   characterize whether the peroxisome abundance decrease in HCT116 caused
   by LOHP treatment is mediated by pexophagy or not, we generated the
   Keima-SKL expression HCT116 cell line, in which the SKL sequence leads
   a pH-sensitive reporter Keima locating on the peroxisome. LOHP
   treatment increased the Ex.561/Ex.488 (acidic environment/neutral
   environment) ratio of Keima-SKL protein, suggesting that LOHP treatment
   induced pexophagy in HCT116 (Fig. [85]2C, [86]D). Besides, the loss of
   peroxisomes caused by LOHP treatment was completely rescued by
   chloroquine (CQ), an autophagy inhibitor blocking
   autophagosome-lysosome fusion (Fig. [87]2E, [88]F). These data
   uncovered that LOHP induced pexophagy and the subsequent loss of
   peroxisomes in HCT116.

   Since the peroxisome abundance of HCT116L would not decrease upon LOHP
   treatment, we hypothesized that HCT116L is resistant to LOHP-induced
   pexophagy. LOHP treatment significantly increased the colocalization of
   PMP70 puncta and LC3 puncta in HCT116, but not in HCT116L (Fig. [89]2G,
   [90]H). The ubiquitination of peroxisomal proteins, such as PMP70,
   which can be recognized by autophagy receptors, is commonly served as a
   signal for pexophagy. We found that PMP70 is ubiquitinated upon LOHP
   treatment in HCT116, but not in HCT116L (Fig. [91]2I). Together, these
   results demonstrated that HCT116L was resistant to LOHP-induced
   pexophagy.

PMP70 depletion overcome LOHP resistance in CRC cells

   To investigate PMP70 functions in LOHP-resistant CRC cells,
   PMP70-depleted HCT116L cell lines were generated using CRISPR-Case9
   interference. PMP70 depletion significantly reduced the proliferation
   and sphere formation in HCT116L cells (Fig. [92]3A, [93]B).
   PMP70-depletion also led to a decrease of CD133 at the protein level
   (Fig. [94]3C). Notably, knocking out PMP70 enhanced the efficacy of
   LOHP in HCT116L and HCT8L cells (Fig. [95]3D).

Fig. 3. Silencing PMP70 overcomes LOHP resistance.

   [96]Fig. 3
   [97]Open in a new tab

   Representative images (A) and quantification (B) of colony formation
   and sphere formation assays of PMP70-depleted and control HCT116L
   cells. C Western blot analysis of PMP70, CD133, CD44, and Nanog
   expression in sgPMP70 and sgNC HCT116L cells. D Effects of LOHP on cell
   survival measured by CCK8 assays in sgNC and sgPMP70 HCT116L cells
   (left panel). Survival curves for HCT8L cells transfected with siRNA
   targeting PMP70 or control siRNA in response to LOHP (right panel). The
   IC50 of LOHP for each condition is indicated in the plots. E
   Representative images of xenograft tumors derived from sgPMP70 or sgNC
   HCT116L cells treated with LOHP or vehicle. F Tumor growth curves and
   tumor weight quantification in different treatment groups. G IHC
   staining of PMP70 and CD133 in tumor tissues from control or
   LOHP-treated mice bearing sgNC or sgPMP70 HCT116L xenografts. H
   Quantification of panel (G). *P < 0.05, **P< 0.01, ***P < 0.001.

   We further tested the response of PMP70-depleted HCT116L cells to LOHP
   in vivo. HCT116L transfected with non-targeting sgRNA (sgNC) and
   sgPMP70 were subcutaneously injected in the right flank of nude mice,
   respectively. PMP70-depleted tumors showed minor regression compared to
   control tumors, and PMP70 depletion enhanced the tumor-suppression
   efficacy of LOHP (Fig. [98]3E, [99]F). Immunohistochemical (IHC)
   staining confirmed that PMP70-depleted tumors exhibited a significantly
   lower number of PMP70-positive cells compared to control tumors. A
   decreased level of CD133 positive cells was also observed in
   PMP70-depleted tumors (Fig. [100]3G, [101]H), which was consistent with
   the findings in vitro. These data suggest that PMP70 depletion overcame
   LOHP resistance in HCT116L, both in vitro and in vivo.

PMP70 protect CRC cells from LPO

   Since PMP70 is a peroxisomal membrane protein essential for lipid
   transportation and metabolism, we compared the lipidome of
   PMP70-depleted and control HCT116L cells, and found that the abundance
   of several polyunsaturated phosphatidylethanolamines (PUFA-PE) was
   significantly higher in PMP70-depleted HCT116L cells (Fig. [102]4A).
   Since PUFA-PE was known as the main substrate of LPO, we hypothesized
   that PMP70 depletion would change the LPO level in cells and used
   BODIPY-C11 probe to verify it. PMP70-depleted HCT8L and HCT116L cells
   exhibited higher LPO levels compared to their parental counterparts
   (Fig. [103]4B). Metabolic interplay between peroxisomes and
   mitochondria is crucial for cellular metabolic homeostasis [[104]9,
   [105]30–[106]32]. Therefore, we further tested mitochondrial functions.
   Results showed that mitochondrial ROS was comparable between
   PMP70-depleted cells and control cells (Fig. [107]4C, [108]D), and
   mitochondria functional assay showed no obvious change in
   PMP70-depletion cells (Fig. [109]4E). These results suggest that PMP70
   deletion induced excessive LPO without influencing mitochondrial stress
   in LOHP-resistant CRC cells.

Fig. 4. PMP70 protects CRC cells from LPO.

   [110]Fig. 4
   [111]Open in a new tab

   A Volcano plot showing lipid species significantly altered between
   sgPMP70#2 and control HCT116L cells. PE: phosphatidylethanolamine; ePE:
   ether-PE; PUFA: polyunsaturated fatty acid. B Lipid ROS levels measured
   using BODIPY 581/591-C11 in sgPMP70 and control HCT116L cells, as well
   as in HCT8L cells transfected with siRNA targeting PMP70 or control
   siRNA. C Fluorescence histograms of BODIPY 581/591-C11 and MitoSOX Red
   staining in sgPMP70 and control HCT116L cells, illustrating LPO and
   mitochondrial ROS levels, respectively. D Quantification of mean
   fluorescence intensity of MitoSOX Red staining in sgPMP70 and control
   HCT116L cells. E Mitochondrial stress profile of sgPMP70 and control
   HCT116L cells, assessed using Seahorse extracellular flux analysis. F
   Western blot analysis of PMP70 and GPX4 in various CRC cell lines. G
   Correlation between PMP70 and GPX4 protein levels in the CPTAC-COAD
   dataset (n = 97). H LPO in HT29 and DLD1 cells transfected with siRNA
   targeting PMP70 (siPMP70#1, #2), GPX4 (siGPX4#1, #2), or control siRNA.
   *P < 0.05, **P < 0.01, ***P < 0.001.

   GPX4 is a critical enzyme for the primary reduction of lipid peroxides.
   The protein levels of PMP70 and GPX4 were examined in several CRC cell
   lines. Of note, CRC cells with higher expression of PMP70 (PMP70^High;
   DLD1, SW480, RKO) exhibited relatively lower level of GPX4 and vice
   versa (Fig. [112]4F). Proteomic data of 97 colon tumor tissues from
   TCGA-CPTAC dataset validated the negative correlation of PMP70 and GPX4
   protein level (Fig. [113]4G, r = –0.3376, P = 0.0007). We then
   hypothesized that PMP70 and GPX4 might be complementary factors in CRC
   cells for preventing LPO. PMP70 and GPX4 were knockdown in HT29 and
   DLD1 cells, respectively. GPX4 knockdown induced LPO in HT29 cells
   (PMP70^Low) but not in DLD1 cells (PMP70^High); PMP70 knockdown
   increased LPO in DLD1 but not in HT29 (Fig. [114]4H). Treatment with
   the GPX4 covalent inhibitor 1S,3R-RSL3 (RSL3) in control and
   PMP70-depleted (or -knockdown) PMP70^High cells (DLD1 and HCT116L) did
   not significantly increase the sensitivity of the cells to RSL3-induced
   LPO and cell death (Fig. [115]S2). Together, these results suggest that
   PMP70 might play an important role in preventing LPO in PMP70^High CRC
   cells.

Silencing PMP70 enhances LOHP-induced LPO in CRC cells

   Since peroxisome protein PMP70 can reduce LPO level, and LOHP treatment
   would reduce peroxisome abundance (Figs. [116]4B, [117]2B), we
   investigated whether LOHP treatment induces LPO in CRC. Liproxstatin-1
   (Lip-1) was used to trap lipophilic radicals in HCT116 and HCT8 cells.
   As shown, Lip-1 partially rescued LOHP-induced cell death (Fig.
   [118]5A), suggesting that LPO may be one of the mechanisms underlying
   LOHP-induced cell death. Flow cytometry analysis also revealed that the
   LOHP treatment-induced increase in LPO levels was significantly lower
   in LOHP-resistant cells compared to LOHP-sensitive cells (Fig.
   [119]5B).

Fig. 5. Silencing PMP70 enhances LOHP-induced LPO in CRC cells.

   [120]Fig. 5
   [121]Open in a new tab

   A Cell viability of HCT116 and HCT8 cells treated with liproxstatin-1
   (Lip-1, 2 μM) under different doses of LOHP (n = 6). B Percentage of
   oxidized lipids measured by BODIPY 581/591-C11 in HCT116L, HCT8L, and
   their parental cells, with or without LOHP treatment (20 μM, 48 h). C
   LPO levels in HCT116L cells transduced with sgPMP70 or sgNC and in
   HCT8L cells transfected with siPMP70 or siNC after LOHP treatment
   (10 μM). D Cell viability of siPMP70#1 and control cells treated with
   LOHP (HCT8L:20 μM, HCT116L:40 μM) and LOHP combined with different cell
   death inhibitors (2 μM Lip-1, 10 μM Z-VAD-fmk, 1 μM Nec1s, and 10 μM
   NAC). E 4-HNE IHC staining of control or LOHP-treated mice bearing
   HCT116L tumors transduced with sgNC or sgPMP70 (n = 5). Right chart:
   quantification of the staining intensity using ImageJ. F Survival curve
   and LPO levels in siPMP70 or control DLD1 cells after LOHP treatment
   (20 μM). *P < 0.05, **P < 0.01, ***P < 0.001.

   We then investigated the responses of PMP70-interfered LOHP-resistant
   cells to LOHP-induced LPO and cell death, both in vitro and in vivo.
   Knockdown of PMP70 significantly enhanced LOHP-induced LPO in HCT116L
   and HCT8L cells (Fig. [122]5C). Furthermore, we tried to investigate
   the type of cell death induced by PMP70 knockdown under LOHP treatment
   in HCT116L and HCT8L. However, the sensitivity of resistant cells to
   LOHP-induced cell death cannot be fully rescued by any of the cell
   death inhibitors tested, even though apoptosis inhibitors can slightly
   restore cell viability. This phenomenon suggests that the increased
   sensitivity to drugs induced by PMP70 knockdown do not follow a known
   form of programmed cell death, indicating a more complex underlying
   mechanism (Fig. [123]5D). In xenograft tumors, LPO was assessed by IHC
   staining of 4-HNE, a major end-stage product of LPO. PMP70-depleted
   tumors exhibited a higher number of 4-HNE-positive cells compared to
   control tumors (Fig. [124]5E), further supporting the role of PMP70 in
   modulating LPO. Additionally, downregulation of PMP70 in PMP70^High
   CRCs (DLD1) significantly enhanced the cell-killing efficacy of LOHP
   and increased LOHP-induced LPO (Fig. [125]5F). Together, all these
   results suggest that silencing PMP70 could enhance LOHP
   treatment-caused LPO and its killing efficacy in LOHP-resistant and
   PMP70^High CRC cells.

PMP70 stabilizes neutral lipids and LDs in LOHP-resistant CRC cells

   We further conducted a lipidomic analysis to investigate the altered
   lipidome of PMP70-depleted and control cells upon LOHP treatment. LOHP
   treatment upregulated 4 out of 6 cholesteryl esters (ChEs) and part of
   lysophosphatidylinositols (LPIs) in HCT116L cells (Fig. [126]6A). In
   PMP70-depleted HCT116L cells, most lysophosphatidylcholines (LPCs) and
   LPIs increased, and only 2 out of 6 ChEs significantly increased upon
   LOHP treatment. Compared to the control HCT116L cells, most neutral
   lipids, including ChEs and triacylglycerols (TGs), were significantly
   decreased in PMP70-depleted HCT116L cells with or without LOHP
   treatment (Fig. [127]6B).

Fig. 6. PMP70 stabilizes neutral lipids and LDs in LOHP-resistant CRC cells.

   [128]Fig. 6
   [129]Open in a new tab

   A Volcano plots of lipid species significantly altered by LOHP compared
   to vehicle treatment in sgPMP70 #2 (right panel) and sgNC (left panel)
   HCT116L cells. LPI: lysophosphatidylinositols; ChE: cholesteryl ester;
   LPC: lysophosphatidylcholine; PS: phosphatidylserine; PC:
   phosphatidylcholine; PE: phosphatidylethanolamine. B Volcano plots of
   lipid species significantly altered between sgPMP70 #2 and sgNC HCT116L
   cells under LOHP (right panel) or vehicle treatment (left panel). TG:
   triacylglycerol. C–E HCT116L cells transduced with sgPMP70 (#2, #3) or
   sgNC were transfected with mCherry-SKL for 48 h. Scale bar = 10 μm. LDs
   were stained with Lipi-Blue (100 nM, Dojindo) for 30 min and analyzed
   by confocal microscopy (panel C). Quantification of LDs per cell is
   shown in panel (D). Quantification of peroxisome-LD contacts per cell
   is shown in panel (E). F Immunoblots of SOAT1, DGAT1, and GAPDH in
   sgPMP70 and sgNC HCT116L cells. *P < 0.05, **P < 0.01, ***P < 0.001.

   Neutral lipids are mainly stored in LDs in the cells. Therefore, we
   used Lipi-blue staining to evaluate LD abundance of PMP70-depleted and
   control HCT116L cells. Consistent with the lipidomic results, PMP70
   depletion caused significant reduction of LDs in LOHP-resistant cells
   (Fig. [130]6C, [131]D). Critical enzymes for neutral lipid synthesis
   and LD formation, including SOAT1 and DGAT1, were reduced in
   PMP70-depleted HCT116L cells (Fig. [132]6F). Unexpectedly, a decrease
   in peroxisome-LD connections was also observed in PMP70-depleted cells
   (Fig. [133]6C, [134]E). These data indicate that PMP70 contributes to
   the LD homeostasis regulated by peroxisomes in LOHP-resistant CRC
   cells.

PMP70 deletion leads to cell cycle arrest

   To investigate the potential pathways impacted by PMP70 in
   LOHP-resistant CRC cells, we performed RNA-seq analyses on
   PMP70-depleted and control HCT116L cells. Pathway enrichment analysis
   revealed that the cell cycle pathway was significantly downregulated in
   PMP70-depleted cells (Fig. [135]7A). Among the most differentially
   expressed genes, apart from ABCD3, key regulators of DNA replication
   including MCM4, MCM5, MCM7, and CDT1 were downregulated in
   PMP70-depleted cells (Fig. [136]7B). Of note, E2F1, a crucial
   transcription factor for cell cycle progression and proliferation, was
   also down-regulated due to PMP70 depletion. Changes of the expression
   levels of MCM4, MCM7, CDT1, and E2F1 were validated by qPCR (Fig.
   [137]7C). Consistently, flow cytometry analysis demonstrated that
   PMP70-depleted cells exhibited a prolonged S phase compared to the
   control cells (Fig. [138]7D). Together, these data suggest that PMP70
   depletion in LOHP-resistant CRCs would cause the cell cycle arrest.

Fig. 7. PMP70 deletion led to cell cycle arrest in LOHP-resistance CRC cells.

   [139]Fig. 7
   [140]Open in a new tab

   A Gene Set Enrichment Analysis (GSEA) showing the downregulated cell
   cycle pathway (KEGG HSA04110) in sgPMP70#2 vs. sgNC cells. B Volcano
   plots showing significantly differentially expressed genes between
   sgPMP70#2 or sgPMP70#3 and sgNC cells. ABCD3 and cell cycle-related
   genes (e.g., E2F1, CDT1, and MCM family members) are labeled. C
   Quantitative real-time PCR analysis of cell cycle-related gene
   expression in sgPMP70 or sgNC HCT116L. Data represent relative mRNA
   expression levels normalized to control. D Cell cycle distribution of
   sgPMP70#2 cells compared to sgNC, assessed by flow cytometry. The
   percentage of cells in G0/G1, S, and G2/M phases is shown. E Schematic
   representation of the effects and potential mechanisms underlying PMP70
   downregulation in LOHP-resistant CRC cells. **p < 0.01, ***p < 0.001.
   Data are presented as mean ± SD.

Discussion

   Recent studies have highlighted the role of peroxisomes in cancer
   biology, particularly in lipid metabolism and oxidative stress
   regulation. Peroxisomal regulators involved in β-oxidation,
   α-oxidation, and ether lipid synthesis have been shown to promote tumor
   progression in breast cancer, hepatocellular carcinoma, and prostate
   cancer [[141]16, [142]33–[143]35]. Catalase, a key peroxisomal ROS
   controller, is implicated in various cancers, as cancer cells often
   reduce ROS levels to avoid apoptosis [[144]19, [145]36, [146]37].
   Peroxisomes have also been linked to drug resistance. In lymphoma,
   inhibiting peroxisomal genes involved in biogenesis or pexophagy has
   been shown to enhance apoptosis in drug-resistant cells [[147]19,
   [148]20]. These findings suggest that targeting peroxisomal function
   could be a potential strategy to overcome drug resistance and enhance
   therapy-induced cell death.

   In CRC, peroxisomes are found to be reduced in tumor tissues, and
   certain proteins involved in their biosynthesis and β-oxidation act as
   tumor suppressors [[149]38], suggesting a role in anti-tumorigenesis.
   Peroxisomes are versatile organelles with context-dependent functions,
   adapting to various cellular stresses. In the context of
   chemoresistance, our data suggested that peroxisomes may promote the
   survival of resistant CRC cells by shielding them from drug-induced
   killing. Specifically, we demonstrated that peroxisomes protect CRC
   cells from LOHP treatment, and silencing PMP70 can restore their
   sensitivity to LOHP.

   We have explored peroxisome homeostasis in CRC chemoresistance. As data
   shown, LOHP triggered pexophagy resulting in the degradation of
   peroxisomes, whereas LOHP-resistant CRC cells show resistance to
   pexophagy induced by LOHP. Dahabieh et al. have revealed the role of
   pexophagy in therapy resistance [[150]20]. They found that histone
   deacetylase inhibitor (HDACi) upregulated peroxisome abundance in
   lymphoma U937 cells, different from our findings in HCT116 cells.
   HDACi-resistant U937 cells possessed enriched peroxisomes as well as
   activated pexophagy, which is similar to the pattern observed in
   HCT116L cells, where we found resistant cells maintained stable and
   enriched peroxisomes and had higher colocalization of PMP70 puncta and
   LC3 puncta. They also reported that disrupting peroxisomal matrix
   protein transport through PEX26 silencing can induce pexophagy and
   contribute to delayed therapy resistance in lymphoma. The results we
   presented and previous reports suggest that drug-resistant cancer cells
   tightly regulate peroxisome homeostasis through enhanced pexophagy,
   despite differing reactions of peroxisomes in lymphoma and CRC cells
   when attacked by drugs. Since LOHP-resistant CRC cells tend to
   protecting and maintaining their abundant peroxisomes under LOHP
   treatment, disrupting peroxisome homeostasis may represent a promising
   strategy to eliminate these resilient cells.

   Intriguingly, several PUFA-PEs, which are substrates of LPO,
   significantly increased in PMP70-depleted HCT116L cells that had
   elevated levels of LPO. Previously, Tabassum et al reported that LOHP
   can induce excess LPO [[151]39]. Our study found that LPO is one of the
   mechanisms underlying LOHP-induced cell death in CRC. Silencing PMP70
   in LOHP-resistant cells can induce overwhelming LPO in response to
   LOHP. This discovery suggested a new function for PMP70—protecting
   cells from LPO. GPX4 is a pivotal enzyme for preventing excess LPO in
   cells. Our data suggested that PMP70 plays a crucial role in limiting
   LPO in CRC cells with low GPX4 levels, where PMP70 is dominant. Besides
   GPX4, two other enzymes that play a crucial role in defending against
   LPO are ferroptosis suppressor protein 1 (FSP1) and dihydroorotate
   dehydrogenase (DHODH). Mao et al identified DHODH operates in parallel
   to mitochondrial GPX4 to inhibit LPO by reducing ubiquinone to
   ubiquinol and repressing DHODH induces extensive mitochondrial LPO in
   Low-GPX4 cancer cells [[152]40]. Although our data suggests that PMP70
   works in parallel with GPX4, further research are needed to understand
   the specific mechanisms by which PMP70 or peroxisomes defend against
   LPO.

   To gain insight into the mechanism of PMP70 in preventing LPO, we
   conducted untargeted lipidomic analysis and RNA-seq analyses in
   PMP70-depleted HCT116L and control cells.

   Firstly, we found that when PMP70 was silenced in HCT116L, numbers of
   LDs and most of the neutral lipids significantly decreased.
   Lippincott-Schwartz and coworkers validated that the transferring of
   fatty acids from LDs to peroxisomes requiring M1 Spastin tethering
   complex found in LDs and ABCD1 in peroxisomes [[153]41]. The
   LD-peroxisome trafficking mediated by M1 Spastin and ABCD1 can relieve
   LPO in LDs. ABC transporters of subfamily D include four proteins:
   ABCD1, ABCD2, ABCD3 and ABCD4. Except for ABCD4, the other three are
   located on peroxisome membrane responsible for lipid transport. Our
   data also showed that depletion of PMP70 (ABCD3) in HCT116L cells
   caused a significant decrease of contacts between LDs and peroxisomes.
   These results indicates that peroxisomal lipid transporters, such as
   ABCD1 and ABCD3, may be important for communication between peroxisomes
   and LDs. Moreover, we identified that knockout of PMP70 decreased LD
   numbers and enzymes critical for LD biosynthesis in LOHP-resistant
   cells. In cancer cells, inhibition of LD synthesis can channel PUFA
   away from TAG into phospholipids and finally sensitize the cells to LPO
   [[154]42, [155]43]. This may be a possible explanation of the enhanced
   sensitivity of sgPMP70 cells to LPO, since PUFA were not sequestered in
   LD, but the mechanisms of how PMP70 regulate LD abundance need further
   investigation.

   Secondly, RNA-seq and flow cytometry analyses revealed that PMP70
   depletion in HCT116L cells induced cell cycle arrest, particularly with
   a prolonged S phase. A more recent study demonstrated that arrested
   cells are more susceptible to LPO due to alterations in lipid
   composition of plasma membrane, which are regulated by MBOAT1 and EMP2
   [[156]44]. This raises the possibility that PMP70 depletion inhibited
   cell cycle progression, leading to decreased expression of MBOAT1 and
   EMP2, and consequently enhanced LPO. However, it would be of greater
   interest in the future to explore how PMP70 directly affects lipid
   composition, given its role in lipid metabolism. Furthermore,
   investigating the mechanisms by which PMP70 influences cell cycle
   progression is also warranted.

   In summary, our study highlights the pivotal role of peroxisomes and
   PMP70 in maintaining chemoresistance in CRC through the regulation of
   LPO balance. An aberrant metabolic profile of PMP70-depleted cells was
   also observed, characterized by excessive LPO, loss of LDs and cell
   cycle arrest (Fig. [157]7E). Targeting PMP70 presents a promising
   therapeutic strategy to overcome LOHP resistance and improve treatment
   outcomes in CRC.

Supplementary information

   [158]Supplementary Materials^ (493.9KB, pdf)
   [159]Original data 1^ (1.8MB, pdf)

Acknowledgements