Anti‐apoptotic protein BCL‐XL as a therapeutic vulnerability in gastric cancer

Abstract

Background

New therapeutic targets are needed to improve the outcomes for gastric cancer (GC) patients with advanced disease. Evasion of programmed cell death (apoptosis) is a hallmark of cancer cells and direct induction of apoptosis by targeting the pro‐survival BCL2 family proteins represents a promising therapeutic strategy for cancer treatment. Therefore, understanding the molecular mechanisms underpinning cancer cell survival could provide a molecular basis for potential therapeutic interventions.

Method

Here we explored the role of BCL2L1 and the encoded anti‐apoptotic BCL‐XL in GC. Using Droplet Digital PCR (ddPCR) technology to investigate the DNA amplification of BCL2L1 in GC samples and GC cell lines, the sensitivity of GC cell lines to selective BCL‐XL inhibitors A1155463 and A1331852, pan‐inhibitor ABT‐263, and VHL‐based PROTAC‐BCL‐XL was analyzed using (CellTiter‐Glo) CTG assay in vitro. Western Blot (WB) was used to detect the protein expression of BCL2 family members in GC cell lines and the manner in which PROTAC‐BCL‐XL kills GC cells. Co‐immunoprecipitation (Co‐IP) was used to investigate the mechanism of A1331852 and ABT‐263 kills GC cell lines. DDPCR, WB, and real‐time PCR (RTPCR) were used to investigate the correlation between DNA, RNA, protein levels, and drug activity.

Results

The functional assay showed that a subset of GC cell lines relies on BCL‐XL for survival. In gastric cancer cell lines, BCL‐XL inhibitors A1155463 and A1331852 are more sensitive than the pan BCL2 family inhibitor ABT‐263, indicating that ABT‐263 is not an optimal inhibitor of BCL‐XL. VHL‐based PROTAC‐BCL‐XL DT2216 appears to be active in GC cells. DT2216 induces apoptosis of gastric cancer cells in a time‐ and dose‐dependent manner through the proteasome pathway. Statistical analysis showed that the BCL‐XL protein level predicts the response of GC cells to BCL‐XL targeting therapy and BCL2L1 gene CNVs do not reliably predict BCL‐XL expression.

Conclusion

We identified BCL‐XL as a promising therapeutic target in a subset of GC cases with high levels of BCL‐XL protein expression. Functionally, we demonstrated that both selective BCL‐XL inhibitors and VHL‐based PROTAC BCL‐XL can potently kill GC cells that are reliant on BCL‐XL for survival. However, we found that BCL2L1 copy number variations (CNVs) cannot reliably predict BCL‐XL expression, but the BCL‐XL protein level serves as a useful biomarker for predicting the sensitivity of GC cells to BCL‐XL‐targeting compounds. Taken together, our study pinpointed BCL‐XL as potential druggable target for specific subsets of GC.

We demonstrated that both selective BCL‐XL inhibitors and VHL‐based PROTAC BCL‐XL can potently kill GC cells that reliant on BCL‐XL for survival. We also confirmed that the effect caused by DT2216 or selective BCL‐XL inhibitors is predominantly BAX/BAK dependent.

1. INTRODUCTION

The current standard‐of‐care therapies with surgery and adjuvant chemo/radiotherapy for patients with GC have markedly improved their outcomes, especially for those in the early stages of the disease. However, the overall clinical outcome for patients with advanced GC remains poor, with a median survival time of only 12–15 months and a 5‐year OS of approximately 5–25%. ¹ The introduction of targeted therapy and immunotherapy that include the anti‐HER2 agents (e.g. trastuzumab, trastuzumab deruxtecan), anti‐VEGFR2 ramucirumab and anti‐PD‐1 pembrolizumab have had some success in patients with advanced GC. ² , ³ However, these therapies are only effective in subgroups of GC patients who possess certain biomarkers, such as HER2 overexpression/amplification, PD‐L1/MSI status etc. Notably, the high rate and rapid emergence of resistance further limit the use of the treatments. ² , ³ Therefore, there is an unmet need to identify new therapeutic vulnerabilities or targets for GC interventions.

Programmed cell death pathways, including apoptosis, serve as natural barriers to cancer pathogenesis. ⁴ The essential executioner proteins in the intrinsic (mitochondrial or BCL2‐regulated) pathway to apoptosis are BAX and BAK; once these become activated, they drive mitochondrial outer membrane permeabilization (MOMP), committing the cell irreversibly to apoptosis. In the absence of apoptosis‐inducing signals, BAX/BAK are kept in check by the pro‐survival proteins including BCL2, BCL‐XL, MCL1, BCLw and BFL‐1/BCL2A1. ⁵ In many cancers, the propensity to undergo apoptosis is impaired because of sustained pro‐survival signaling. ⁶ , ⁷ Accordingly, pharmacologic inhibitors targeting pro‐survival BCL2 proteins have been developed to induce apoptosis in cancer cells. ⁸ , ⁹

Various malignancies show recurrent genetic amplifications affecting pro‐survival members of this family, particularly MCL1 and BCL2L1, ¹⁰ but there are also studies indicating that BCL2L1 copy number variations (CNVs) are not associated with corresponding expression levels ¹¹ and that BCL2L1 gain/amplification may not exert the same biological function as overexpression. ¹² Amplification of BCL2L1 is reported in 11 (10.7%) of the 103 GC cases analyzed by aCGH, while the putative amplification rate of BCL2L1 using the GISTIC algorithm with the TCGA dataset is 2.7% (6/220) (www.cbioportal.org). ¹³ Using siRNA targeting BCL‐XL or ABT737, a BH3 mimetic compound inhibiting BCL2, BCL‐XL and BCLw, showed that BCL2L1‐amplified GC cell lines are more susceptible to this inhibitor than the BCL2L1‐nonamplified cells. However, it was later reported that the BCL2L1‐amplified GC cell lines MNK‐28 and MKN‐74 used in this study were cross‐contaminated. Given the elusive role of BCL2L1 in GC, we were therefore determined to characterize the role of BCL2L1 and the encoded anti‐apoptotic BCL‐XL in GC cell survival and to look for biomarkers predicting the response of GC cells to BCL‐XL‐targeting therapy.

Here, we provide evidence that BCL‐XL is a promising therapeutic target in a subset of GC cases characterized by increased BCL‐XL protein expression. Functionally, we demonstrated that both selective BCL‐XL inhibitors and VHL‐based PROTAC BCL‐XL are active in killing GC cells that rely on BCL‐XL for survival. Furthermore, we demonstrated that BCL2L1 CNVs cannot reliably predict BCL‐XL expression and that the BCL‐XL protein level, but not BCL2L1 CNVs, serves as a useful biomarker for predicting sensitivity of GC cells to BCL‐XL targeting compounds. Taken together, our study identifies BCL‐XL as a potential druggable target for specific subsets of GC.

2. METHODS

2.1. Clinical samples

Fresh tissue samples of GC and paired normal gastric tissue were obtained from newly diagnosed GC patients at surgery with informed consent. Peripheral blood samples were also collected from five of these GC patients and platelets were isolated with the Peripheral Blood Platelet Isolate Kit (Solarbio, P6390) according to the manufacturer's instructions. The study was performed with the approval of National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College (Approval number: 17‐156/1412), and the Institute of Laboratory Animal Sciences, Chinese Academy of Medical Sciences (Approval number: GJN22002).

2.2. Cell lines

Human GC cell lines, including 23132/87, SNU216, NCI‐N87, MKN1, AGS, HGC27, and SNU719, were purchased from Cobioer (Nanjing). Human multiple myeloma cell lines, including MM1S, KMS12PE and KMS12BM, were purchased from ATCC and DSMZ. All cancer cell lines were cultured in RPMI‐1640 medium (Gibco, CA, USA) supplement with 10% fetal bovine serum (FBS) (Gibco, CA, USA) at 37°C with 5% CO₂. The human embryonic kidney cell line HEK293T was cultured in DMEM supplemented with 10% FBS at 37°C with 10% CO₂. Cells were tested regularly using the Myco‐Lumi Luminescent Mycoplasma Detection Kit (Beyotime, C0297M)and were consistently negative for mycoplasma.

2.3. Plasmids

The Constitutive Cas9 vector FUCas9Cherry (Addgene #70182), the inducible guide RNA vector FgH1tUTG (Addgene #70183), the constitutive guide RNA vector pKLV‐U6 gRNA(BbsI)‐PGKpuro2ABFP (Addgene #50945) and the 3rd generation lentiviral packaging plasmids pMDLg/pRRE (Addgene #12251), pRSV‐Rev (Addgene#12253) and pCMV VSV‐G (Addgene#8454) were used in this study.

2.4. Reagents

ABT‐263, A1155463, A1331852, DT2216, MG132 and Doxycycline were purchased from MedChemExpress (MCE). The pan‐caspase inhibitor Q‐VD‐Oph was purchased from Selleck.

2.5. Lentivirus production and infection

The lentivirus packaging plasmids pMDLg/pRRE, pRSV‐Rev and pCMV VSV‐G were transiently transfected into HEK293T cells with the constructs of interest using Lipofectamine™ 3000 Transfection Reagent (Thermo, L3000015). Supernatant containing infectious virus particles was harvested 48 h later. A second viral harvest was made following a further 24 h incubation with fresh medium. Virus‐containing supernatant was filtered through a 0.45 μm filter and stored at 4°C or −80°C until used.

Typically, GC cells were seeded into 6‐well plates at 5000,00 cells/well. An equivalent volume of virus‐containing culture medium was added along with polybrene (Sigma) to a final concentration of 5 μg/mL. Cells were spin infected (1800 g, 25°C, 1 h) and then incubated at 37°C for 20 h. Cells were then washed and resuspended in fresh culture media.

2.6. CRISPR/Cas9 gene editing

BAX ^−/− BAK ^−/− cells were generated using CRISPR/Cas9. 23 132/87 and SNU216 were infected with lentiviruses expressing Cas9 (mCherry) and the pKLV‐gRNAs targeting BAX and BAK (BFP). MCherry⁺BFP⁺ cells were sorted into 96‐well plates at one cell per well using a BD FACSAriaTMII flow cytometer. Mutation of the targeted DNA was then confirmed by targeted PCR followed by Sanger DNA sequencing. Single cell clones with frameshift mutations in both BAX and BAK were used for this study. The inducible guide RNA vector FgH1tUTG was used to delete BCL‐XL. The sgRNA sequences and primer sequences are described in Supplementary Table S1.

2.7. Cell viability assays

To test the response of GC cell lines to BH3‐mimetics, cells were seeded in 96‐well plates at 3 × 10³ cells/well and treated with titrated concentrations (0–10 μM) of indicated drugs for 24 h. Similar experiments were performed with DT2216 with cell viability tested at 24–96 h. Cell viability was determined using CellTiter‐Glo assay (Promega) according to the manufacturer's instructions. Percentage cell viability was calculated by normalizing to the viability of cells treated with DMSO (vehicle control). GraphPad Prism software was used to calculate the concentration at which cell viability was reduced to 50% (IC50).

2.8. Droplet digital PCR (ddPCR)

Genomic DNA from both GC cell lines and primary samples was extracted with the AllPrep DNA/RNA Micro Kit (Qiagen, #80284) according to the manufacturer's guidelines. A 1 μg sample of genomic DNA was restricted with MseI (New England Biolabs) enzyme for 1 h at 37°C. The PCR mixture was assembled in 20 μL solution containing 2 × ddPCR supermix (Bio‐Rad), 20 × primer and probes, and 20 ng of the restricted DNA. The reaction mixture and droplet generation oil (Bio‐Rad) were loaded into the eight‐channel disposable droplet generator (QX‐200; Bio‐Rad). The droplets were transferred to a 96‐well PCR plate and PCR reaction was performed as follows: enzyme activation for 10 min at 95°C, then 40 cycles at 94°C for 30 s, then 60°C for 1 min, followed by enzyme deactivation by holding for 10 min at 98°C and 4°C (performed with a ramp rate of 2°C/s in all steps). The PCR plate was placed in a droplet reader (Bio‐Rad). Analysis of the ddPCR data was performed with QuantaSoft analysis software (Bio‐Rad), which accompanied the droplet reader. The amplification‐threshold value was set at 3.0 for both primary GC samples and cell lines. The BCL2L1 and RPP30 assay sequences are described in Supplementary Table S1.

2.9. RT‐PCR

Total RNA was extracted using Trizol reagent (Thermo Fisher Scientific). The complementary DNA was synthesized using the PrimeScript™ RT reagent Kit (Perfect Real Time) (TaKaRa) according to the manufacturer's instructions. TB Green Premix Ex Taq II(Tli RNaseH Plus) (TaKaRa) was used in the real‐time PCR reaction. Messenger RNA (mRNA) levels were normalized to expression of the housekeeping gene GAPDH. The −ΔCT method was used in the analysis of PCR data. Primers used in RT‐PCR are described in Supplementary Table S1.

2.10. Western blotting and co‐IP

Total protein was extracted using RIPA buffer containing complete protease inhibitors (Roche, Cat. #4693132001). Protein content was quantified using the Enhanced BCA Protein Assay Kit (Beyotime, P0010S). Lysates were diluted with 5 × SDS‐PAGE sample loading buffer at a 4:1 ratio and denatured by boiling at 95°C for 10 min. Protein samples (20–40 μg) were separated by SDS‐PAGE using 10% SDS‐polyacrylamide gels and transferred onto nitrocellulose membranes (Invitrogen). The membranes were blocked with 5% skimmed milk in TBS and 0.1% Tween‐20 (blocking buffer) before incubation with antibodies. Monoclonal antibodies to BCL2 (clone ERP17509, abcam, ab182859, 1:1000), BCL‐XL (clone E18, abcam, ab32370, 1:1000), MCL1 (clone 19C4‐15, abcam, ab243136, 1:1000), BAX (clone D2E11, CST, #5023S, 1:1000), BAK (clone D4E4, CST, #12105 s, 1:1000), BIM (clone C34C5, CST, #2933S, 1:1000), NOXA (clone D8L7U, CST, #14766S, 1:750), VHL (CST, #68547, 1:1000), MDM2 (clone D1V2Z, CST, #86934,1:1000), β‐actin (clone OTI1, ZSGB, TA‐09, 1:1000) and GAPDH (clone OTI2D9, ZSGB, TA‐08, 1:1000) were used in this study. All antibodies were diluted in blocking buffer.

23132/87 cells were pre‐incubated with 20 μM Q‐VD‐Oph for 30 min and treated with indicated concentrations of A1331852 and ABT‐263 for 6 h before collection. Co‐immunoprecipitation was performed using the Immunoprecipitation Kit with Protein A + G Agarose Gel (Beyotime) according to the manufacturer's instructions. Briefly, immunoprecipitation was performed with anti‐BCL2 (clone100/D5, abcam, ab692) and anti‐BCL‐XL antibody (Proteintech, #66020‐1‐Ig) and blotted with anti‐BIM antibody (clone C34C5, CST, #2933S, 1:1000). The immunoprecipitants and inputs were analyzed by immunoblotting as described above.

3. RESULTS

3.1. A subset of GC cell lines relies on BCL‐XL for survival

To study the role of BCL‐XL in GC cell survival, we collected 7 different human GC cell lines and treated them with the two selective BCL‐XL inhibitors A1155463 ¹⁴ and A1331852, ¹⁵ as well as with ABT‐263, ¹⁶ which targets BCL2, BCL‐XL and BCLw, and is in clinical trials. Interestingly, 3 out of 7 GC cell lines could be readily killed by the BCL‐XL‐selective inhibitors A1155463 and A1331852 (Figure 1A). Of note, both selective BCL‐XL inhibitors showed higher rates of killing these GC cell lines than ABT‐263 (Figure 1A). Importantly, we confirmed that the effect exerted by these inhibitors is via apoptosis as genetic removal of the downstream pro‐apoptotic executioners BAX and BAK abolished the killing of these GC cell lines by these BH3‐mimetic drugs (Figure 1B). The role of BCL‐XL in maintaining GC cell survival was further confirmed by genetic deletion of BCL‐XL using the inducible CRISPR/Cas9 system in 23 132/87 cells (Figure 1C). ¹⁷

FIGURE 1.

A subset of GC cell lines are susceptible to selective BCL‐XL inhibitors. (A) Sensitivity of human GC cell lines to the small molecular inhibitors of BCL‐XL. The sensitivity of the 7 human GC lines to A1155463, A1331852 and ABT‐263 was determined using the CellTiter‐Glo assay after culturing in 0–10 μM of the indicated compounds for 24 h. A discrete heat map representation of the mean ± SD of IC50s from 3 independent experiments is shown in this panel. Red represents potent killing (IC50 < 0.1 μM), whereas blue indicates resistance (IC50 > 10 μM). (B) BAX/BAK‐dependent killing of GC cells by these selective BCL‐XL inhibitors. Top panel: loss of BAX and BAK expression in the engineered knock‐out clones was confirmed by Sanger DNA sequencing and western blotting. Bottom panel: the viability of WT 23132/87 cells or the BAX−/−BAK−/− subclones after treatment with 0–10 μM of the indicated compounds was determined using the CellTiter‐Glo assay. (C) Genetic deleting BCL‐XL rapidly induces cell death in 23 132/87 cells. The viability of 23 132/87 cells 0–120 h after addition of doxocycline (DOX) to induce expression of guide RNAs (sgRNAs) that target BCL‐XL was determined with the CellTiter‐Glo assay. Two sgRNAs targeting BCL‐XL were tested. (D) ABT‐263 is less effective at releasing BIM from BCL‐XL than A1331852. Equivalent lysates prepared from 23 132/87 cells treated with the indicated concentrations of ABT‐263 or A1331852 in the presence of 20 μM QVD were immunoprecipitated with the BCL‐XL or BCL2 antibody. The amount of BIM associated with BCL‐XL upon indicated treatments was determined. Total amount of BIM and β‐actin in the whole cell lysates served as the loading controls. Data in (A–C) represent the means ± SD of ≥3 independent experiments; data shown in (D) are from a representative of two experiments.

3.2. ABT‐263 is not an optimal inhibitor of BCL‐XL

BCL‐XL plays an important role in cancer pathogenesis and in mediating drug resistance to standard chemotherapy and targeted therapies, which underpins the ongoing clinical trials of combination therapy testing ABT‐263 in solid tumors and lymphoid malignancies. ⁹ Intriguingly, we only observed subtle activity of ABT‐263 in the BCL‐XL‐dependent GC cell lines (Figure 1A). To understand the low activity of ABT‐263 in inhibiting BCL‐XL even though it exhibited comparable binding affinity to BCL2, BCL‐XL and BCLw in vitro, ¹⁶ we compared the ability of A1331852 and ABT‐263 to release the pro‐apoptotic initiator of apoptosis, BIM, from its restraint by BCL‐XL. Consistent with the functional data, ABT‐263 was much less effective at displacing BIM from BCL‐XL than A1331852 (Figure 1D). Moreover, we found that ABT‐263 was much more potent at displacing BIM from BCL2 than from BCL‐XL, confirming that ABT‐263 is a more potent inhibitor of BCL2 than BCL‐XL (Figure 1D).

3.3. VHL‐based PROTAC‐BCL‐XL DT2216 appears to be active in GC cells

Despite the superior ability of small molecular inhibitors of BCL‐XL to induce killing in BCL‐XL‐reliant GC cells, the clinical utility of direct BCL‐XL inhibitors is largely limited by their on‐target and dose‐limiting platelet toxicity. ¹⁸ , ¹⁹ The recent development of proteolysis targeting chimeras (PROTACs) to induce degradation of targeted proteins has gained momentum, ²⁰ which has changed the landscape of drug development. Identification of the most promising protein targets to be degraded and a ligase that is highly expressed in tumors compare with normal tissues ensures augmented efficacy and reduced toxicity. Accordingly, several BCL‐XL specific PROTACs have been reported with reduced toxicity in platelets but potent antitumor activity in hemotopoeitic malignancies ²⁰ , ²¹ , ²² and small‐cell lung cancer. ²³

Interestingly, a recent study using in silico analyses showed that the E3 ligase MDM2 is highly expressed in GC cells and concluded that BCL‐XL PROTACs coupled with the MDM2 ligase represents a potential therapeutic approach in GC. ²⁴ , ²⁵ We therefore firstly determined the expression of MDM2 in our GC cell lines together alongside three blood cancer cell lines. Unexpectedly, heterogenous expression of MDM2 was observed in GC cell lines, and the abundance of MDM2 was even lower in three BCL‐XL‐dependent GC cell lines: 23132/87, NCI‐N87 and SNU216 (Figure 2A). Instead, we detected ubiquitously high expression of the E3 ligase VHL in all the GC cell lines, which is comparable to that in the blood cancer cell lines (Figure 2A). More importantly, only low levels of both MDM2 and VHL were detected in platelets (Figure 2B). The abundant expression of VHL in GC cells but not in platelets indicated that BCL‐XL‐PROTACs coupled with VHL are likely to achieve anti‐tumor effects with only low toxicity in platelets. To test this, we detected the activity of a VHL‐based BCL‐XL‐PROTAC DT2216 in our panel of GC cell lines. ²⁰ Notably, DT2216 induced both dose‐ and time‐dependent killing of all three BCL‐XL‐dependent GC cell lines (23 132/87, SNU216 and NCI‐N87) (Figure 2C). Importantly, we confirmed that the effect caused by DT2216 is predominantly BAX/BAK dependent, although minor BAX/BAK‐independent killing was observed at a very high concentration (10 μM) (Figure 2D, Supplementary Figure S1A).

FIGURE 2.

VHL‐based PROTAC‐BCL‐XL DT2216 is also active in killing BCL‐XL‐dependent GC cells. (A) Immunoblotting analysis of E3 ligase MDM2 and VHL expression in both GC and blood cancer cell lines. (B) Minimal expression of VHL and MDM2 in human platelets. (C) Sensitivity of human GC cell lines to the VHL‐based PROTAC‐BCL‐XL DT2216. The viability of 23 132/87, SNU216 and NCI‐N87 to DT2216 after treatment with 0–10 μM of DT2216 for 24–96 h was determined using the CellTiter‐Glo assay. (D) BAX/BAK‐dependent killing of GC cells by DT2216. The viability of WT 23132/87 cells or the BAX−/−BAK−/− subclones after treatment with 0–10 μM of DT2216 was determined at 72 h using the CellTiter‐Glo assay. (E) Protein levels of BCL‐XL, MCL1 and BCL2 upon DT2216 treatment. BAX−/−BAK−/− SNU216 cells were treated with 0–10 μM of DT2216 for 96 h and the protein levels of BCL‐XL, BCL2 and MCL1 were determined. (F) DT2216 induces degradation of BCL‐XL in a time‐dependent manner. (G) Proteasome inhibition blocks the BCL‐XL degradation by DT2216. Immunoblotting analyses of BCL‐XL protein levels in BAX−/−BAK−/− SNU216 cells after they were either left untreated or pretreated with the proteasome inhibitor MG132 (5 μM) for 1 h, and then treated with or without DT2216 for 24 h. The immunoblots in (A,B,E–G) are representatives of ≥2 independent experiments; data in (C) and (D) represent the means ± SD of ≥3 independent experiments.

To validate the mechanism of action of DT2216 against GC cells, we used BAX/BAK deficient SNU216 cells as a model system, as loss of BAX/BAK excludes the nondirect degradation of BCL2 family proteins due to the activation of downstream caspase cascade. ²⁶ We found that DT2216 caused a dose‐ and time‐dependent decrease of BCL‐XL protein (Figure 2E,F, Supplementary Figure S1B). In addition, we found that pre‐incubation of SNU216 cells with the proteasome inhibitor MG‐132 prevented the degradation of BCL‐XL induced by DT2216 in the GC cell lines (Figure 2G). Consistent with the previous studies, ²⁰ , ²² no BCL2 degradation was detected, but we also observed a reduction of MCL1 protein level at higher concentrations(10 μM) (Figure 2E). Although previous studies suggested that the activation of caspase‐3 upon DT2216 treatment contributes to the reduction of MCL1 protein, ²⁰ , ²² we excluded this mechanism in our studies as no cleaved caspase‐3 was detected at all doses tested using BAX/BAK deficient cells (data not shown). This therefore needs to be further addressed by future studies. Collectively, these data confirm that DT2216 can potently induce apoptosis in a BAX/BAK‐dependent manner in BCL‐XL‐dependent GC cells via proteasome mediated BCL‐XL degradation.

3.4. BCL‐XL protein level predicts the response of GC cells to BCL‐XL targeting therapy

It was previously reported that GC cell lines with BCL2L1 gene amplification were more susceptible to ABT737 treatment, but the study was only conducted in two BCL2L1 gene‐amplified GC cell lines, MNK‐28 and MKN‐74, and a cross‐contamination between these two cell lines was later reported. Given the limited number of cell lines used in the previous study, we examined whether BCL2L1 CNVs predict the susceptibility of GC cells to the BCL‐XL‐targeting drugs using our panel of GC cell lines. To address this, we determined BCL2L1 CNVs as well as the mRNA and protein levels of the anti‐apoptotic protein BCL‐XL in our 7 GC cell lines (Figure 3A). Surprisingly, although there was a significant correlation between BCL‐XL mRNA and BCL‐XL protein expression, we did not detect a correlation between BCL2L1 CNVs and BCL‐XL mRNA or BCL‐XL protein expression (Figure 3B). Of note, we found that GC cell lines expressing high levels of BCL‐XL protein were more susceptible to BCL‐XL inhibition, with no significant association between BCL2L1 CNVs and response to the BCL‐XL inhibitors detected (Figure 3C).

FIGURE 3.

Biomarkers predicting the dependence of GC cells on BCL‐XL. (A) BCL2L1 CNVs, BCL‐XL mRNA and BCL‐XL protein levels in the GC cell lines used in this study. Top: error bars indicate the Poisson 95% confidence intervals for each determination. The dashed line indicates the ddPCR threshold cut‐off of 3.0 copies for calling a sample BCL2L1‐amplified. Middle: relative BCL‐XL mRNA expression normalized to GADPH. Bottom: BCL‐XL protein expression was quantified by western blotting and densitometric analysis using ImageJ. Relative BCL‐XL protein levels normalized to SNU216 were used for subsequent correlation analysis. (B) Simple linear regression of BCL2L1 CNVs, BCL‐XL mRNA or BCL‐XL protein levels. (C) Simple linear regression between IC50s to A1331852 and BCL2L1 CNV, BCL‐XL mRNA or BCL‐XL protein levels. (D) Protein levels of other BCL2 family members in the GC cell lines used in this study. (E) Simple linear regression between IC50s to A1331852 and BCL2, MCL1, BIM or NOXA protein levels.

Apart from BCL2L1 gene amplification, other BCL2 family members (BIM ²⁷ , ²⁸ and NOXA ²⁹ ), were also reported to affect the reliance of cancer cells on BCL‐XL. To determine whether other BCL2 family proteins might impact the dependency of GC cells on BCL‐XL, the expression levels of BCL2, MCL1, NOXA and BIM proteins were correlated with A1331852 sensitivity, but no significant correlation was detected (Figure 3D,E).

3.5. BCL2L1 gene CNVs do not reliably predict BCL‐XL expression

Our results in GC cell lines suggested that BCL‐XL expression is controlled at multiple levels, which cannot be predicted solely by its gene amplification. We therefore further addressed this question using primary samples derived from GC patients. Paired normal and tumor tissues from 18 GC patients were collected and targeted ddPCR was performed to determine BCL2L1 CNVs. Consistent with the previous studies, 10.5% (2/18) of the GC cases studied showed BCL2L1 gene amplification (≥ three copies). Although both samples with BCL2L1 gene amplifications (GC#1 and #2) showed relatively high levels of BCL‐XL protein expression, similarly high levels of BCL‐XL were also detected in another 5 tumor samples that had no BCL2L1 gene amplification (GC#3–#7) and no correlation between BCL2L1 CNVs and BCL‐XL protein levels was detected in these samples (Figure 4A). Surprisingly, 6 of the 18 GC cases with no BCL2L1 gene amplification (GC#8–#13) showed decreased BCL‐XL protein expression. In addition, 5 of the 18 GC cases with no BCL2L1 gene amplification (GC#8–#13) showed comparable BCL‐XL protein levels in paired normal and tumor samples. Nevertheless, no correlation between BCL2L1 CNVs and BCL‐XL protein expression was detected in all samples studied (right panels in Figure 4A–C, Supplementary Figure S2). These data together with the results from GC cell lines confirmed that BCL2L1 gene CNVs do not reliably predict BCL‐XL protein expression and that high levels of BCL‐XL expression are much more prevalent than BCL2L1 gene amplification in GC samples.

FIGURE 4.

BCL2L1 CNV cannot reliably predict BCL‐XL expression. (A) BCL2L1 CNVs and BCL‐XL protein levels in primary GC samples with elevated BCL‐XL protein expression in paired tumor tissues. BCL2L1 CNVs were estimated by ddPCR. BCL‐XL protein expression was quantified by western blotting and densitometric analysis using ImageJ. Left: summary of BCL2L1 CNVs and BCL‐XL protein levels in paired normal and tumor tissues. The dashed line indicates the ddPCR threshold cut‐off of 3 copies for calling a sample BCL2L1‐amplified. Fold change (FC) of BCL‐XL protein levels normalized to the paired normal tissues was used. Middle: immunoblotting analysis to determine BCL‐XL protein level in primary GC samples. Right: correlation analysis between BCL2L1 CNVs and BCL‐XL protein levels in these samples. Relative BCL‐XL protein levels normalized to the common sample loaded in different blots were used (see Supplementary Figure S2 for the original full blots). (B) Similar experiments to (A) with decreased BCL‐XL protein expression in paired tumor tissues. (C) Similar experiments to (A) with comparable BCL‐XL protein expression in paired normal and tumor tissues.

4. DISCUSSION

The main focus of our study was to elucidate the role of BCL2L1 and the encoded anti‐apoptotic BCL‐XL in GC cell survival and to look for biomarkers predicting the response of GC cells to BCL‐XL‐targeting therapy. Using selective BCL‐XL inhibitors (Figure 1) as well as a proteolysis targeting chimera to induce BCL‐XL degradation (Figure 2) and a genetic approach (Figure 1), we showed that a subset of GC cell lines (3/7) requires BCL‐XL for survival and that BCL‐XL protein levels, but not BCL2L1 gene CNVs, serve as a useful biomarker predicting sensitivity of GC cells to BCL‐XL‐targeting drugs (Figure 3).

Amplification of the BCL2L1 gene has been implicated in promoting aberrant survival of malignant cells in various cancers, including GC. ¹⁰ , ¹³ However, unlike in the case of CNVs of the MCL1 gene, ¹¹ the relationship of BCL2L1 CNVs and BCL‐XL protein expression has been elusive. Early studies in non‐small‐cell lung cancer and ovarian cancer showed that BCL2L1 CNVs are not associated with high levels of BCL‐XL protein and that BCL2L1 gene gain/amplification may not exert the same biological function as overexpression. ¹¹ , ¹² Here, using GC cell lines and primary GC samples, we showed that BCL2L1 CNVs do not reliably predict high levels of BCL‐XL protein expression (Figures 3 and 4), suggesting the regulation of BCL‐XL expression at multiple levels in these cancer cells. Accordingly, the BCL‐XL protein level, but not BCL2L1 CNVs, predicts sensitivity of GC cells to BCL‐XL targeting compounds.

Given the central role of BCL‐XL in maintaining cancer cell survival and mediating drug resistance in solid tumors, ³⁰ ABT‐263 (Navitoclax) in combination with chemotherapy or targeted anti‐cancer therapies is currently being evaluated in multiple clinical trials. ³¹ , ³² However, although ABT‐263 showed high binding affinity to both BCL2 and BCL‐XL in vitro, ¹⁶ we only observed subtle activity of ABT‐263 in the BCL‐XL‐dependent GC cell lines. Our mechanistic study showed that ABT‐263 is a more potent inhibitor of BCL2 than BCL‐XL (Figure 1). This is in consistent with the previous study in human non‐Hodgkin lymphomas, which showed that high expression of BCL2 but not BCL‐XL predicted sensitivity to ABT‐263. ³³ These data suggested that effective targeting of BCL‐XL using highly specific inhibitors, rather than BH3 mimetic drugs targeting not only BCL‐XL but also BCL‐2 (and BCL‐W), may be required to exert optimal anti‐tumor effects in clinical studies.

Apart from the sub‐optimal activity of ABT‐263 in inhibiting BCL‐XL, the on‐target and dose‐limiting toxicity in platelets of direct targeting of BCL‐XL further limited their clinical utility thus far. ¹⁸ , ¹⁹ Strategies that restrict the action of BCL‐XL inhibitors to tumor cells can potentially reduce the on‐target toxicity in platelets and thereby yield an enhanced therapeutic index. ²⁰ , ³⁴ , ³⁵ , ³⁶ Here, we found that E3 ligase VHL, but not MDM2, is highly expressed in GC cells with minimal expression in platelets, suggesting that BCL‐XL‐PROTAC coupled with VHL may achieve potent anti‐tumor efficacy with reduced toxicity. Indeed, our in vitro studies with the VHL‐based PROTAC BCL‐XL DT2216 showed high activity in inducing apoptosis in BCL‐XL‐dependent GC cells via proteasome mediated degradation of BCL‐XL, achieving an effect similar to the selective BCL‐XL inhibitor A1331852. However, the anti‐tumor effect of DT2216 with spared activity on platelets needs to be further investigated using in vivo GC models.

In summary, our study highlighted the importance of understanding the molecular mechanisms underpinning cancer cell survival and identified BCL‐XL as a promising therapeutic target in GC cases with high levels of BCL‐XL protein expression. Development of strategies such as PROTAC‐based and antibody‐drug conjugates that restrict the action of BCL‐XL targeting BH3 mimetics to tumor cells holds promise for the treatment of GC patients with advanced diseases.

AUTHOR CONTRIBUTIONS

YMW, LPZ, JNG and RG designed the study; DBZ provided the primary GC samples and reviewed the paper; YMW, ZLP, CW, GMX, XJY and MYL performed experiments; MJL analyzed the data; ZFL and SYR collected the primary GC samples used in this study. This study was supervised by DBZ, JNG and RG.

CONFLICT OF INTEREST STATEMENT

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Ran Gao and Jia‐Nan Gong are Editorial Board member of AMEM and a co‐author of this article. To minimize bias, they were excluded from all editorial decision‐making related to the acceptance of this article for publication.

ETHICS STATEMENT

All clinical samples were collected from the Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College from 2021 to 2023. Written informed consent was obtained from all participants enrolled in this study, and ethical approval was obtained from the following institutional review boards in accordance with the Declaration of Helsinki: National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital, Chinese Academy of Medical Sciences and Peking Union Medical College. Approval number: 17‐156/1412. The Institute of Laboratory Animal Sciences, Chinese Academy of Medical Sciences. Approval number: GJN22002.

Supporting information

Table S1. The sgRNAs and primer sequences used in this study.

FIGURE S1 BAX/BAK‐dependent killing of GC cells by DT2216. (A) The viability of BAX ^−/− BAK ^−/− 23 132/87 subclones after treatment with 0–10 μM of DT2216 for 24–96 h was determined using the CellTiter‐Glo assays. (B) Loss of BAX and BAK expression in the engineered BAX/BAK deficient SNU216 subclones was confirmed by Western Blotting.

FIGURE S2 The full Western blots used in Figure 4 were shown.

Wei Y, Zhang L, Wang C, et al. Anti‐apoptotic protein BCL‐XL as a therapeutic vulnerability in gastric cancer. Anim Models Exp Med. 2023;6:245‐254. doi: 10.1002/ame2.12330