Article Text
PDF
Abstract
Background PARP inhibitors (PARPi) have been licensed for the maintenance therapy of patients with metastatic pancreatic cancer carrying pathogenic germline BRCA1/2 mutations. However, mutations in BRCA1/2 are notably rare in pancreatic cancer.
Objective There is a significant unmet clinical need to broaden the utility of PARPi.
Design RNA sequencing was performed to screen potential targets for PARPi sensitivity. The synthetic lethal effects were verified in patient-derived xenograft (PDX), xenograft and patient-derived organoid models. Mechanisms were explored via LC‒MS/MS, coimmunoprecipitation, laser microirradiation, immunofluorescence, the homologous recombination (HR) or non-homologous end joining (NHEJ) reporter system, in situ proximity ligation assay and live-cell time-lapse imaging analyses.
Results Targeting protein for Xenopus kinesin-like protein 2 (TPX2) is an exploitable vulnerability. TPX2 was downregulated in PDX models sensitive to PARPi, and TPX2 inhibition conferred synthetic lethality to PARPi both in vitro and in vivo. Mechanistically, TPX2 functions in a cell cycle-dependent manner. In the S/G2 phase, ATM-mediated TPX2 S634 phosphorylation promotes BRCA1 recruitment to double-strand breaks (DSBs) for HR repair, whereas non-phosphorylated TPX2 interacts with 53BP1 to recruit it for NHEJ. The balance between phosphorylated and non-phosphorylated TPX2 determines the DSB repair pathway choice. During mitosis, TPX2 phosphorylation enhances Aurora A activity, promoting mitotic progression and chromosomal stability. Targeting TPX2 S634 phosphorylation with a cell-penetrating peptide causes genomic instability and mitotic catastrophe and enhances PARPi sensitivity. Additionally, the inhibition of TPX2 or S634 phosphorylation combined with gemcitabine further sensitised pancreatic cancer to PARPi.
Conclusions Our findings revealed the dual-functional significance of TPX2 in controlling DNA DSB repair pathway choice and mitotic progression, suggesting a potential therapeutic strategy involving PARPi for patients with pancreatic cancer.
- PANCREATIC CANCER
- CHEMOTHERAPY
- DRUG RESISTANCE
- CELL BIOLOGY
- CELL SIGNALLING
Data availability statement
Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. The transcriptome data generated by the study were deposited in the National Omics Data Encyclopedia (OER399218) (https://www.biosino.org/node/). The mass spectrometry proteomics data were deposited to the ProteomeXchange Consortium via the iProX partner repository with the dataset identifier PXD043145 (https://www.iprox.cn/page/home.html). All the relevant data that support the findings of this study are available from the corresponding author on request.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
PARP inhibitors (PARPi) have been licensed for the maintenance therapy of patients with metastatic pancreatic cancer carrying pathogenic germline BRCA1/2 mutations. However, mutations in BRCA1/2 are notably rare in pancreatic cancer (less than 10%), which limits the broader applicability of this promising therapeutic strategy.
PARPi, by inhibiting single-strand DNA break repair, are synthetically lethal in homologous recombination (HR)-deficient cancer cells, highlighting the importance of identifying strategies to exploit genome instability vulnerabilities in pancreatic cancer.
HR and non-homologous end joining (NHEJ) repair pathway compete to repair DNA double-strand breaks (DSBs) throughout the cell cycle and play crucial roles in determining cell fate.
Mitotic progression is also a determinant of cell fate.
WHAT THIS STUDY ADDS
The balance between phosphorylated and non-phosphorylated targeting protein for Xenopus kinesin-like protein 2 (TPX2) determines the DSB repair pathway choice.
ATM-mediated TPX2 S634 phosphorylation is required for M-phase progression and chromosomal stability.
Low expression of TPX2 is synergistically lethal with PARPi in pancreatic cancer.
Targeting TPX2 S634 phosphorylation with a cell-penetrating peptide sensitises pancreatic cancer cells to PARPi.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
TPX2 is a predictor of PARPi sensitivity and a potential target to increase the susceptibility of patients with pancreatic cancer to PARPi.
Combining PARPi treatment with gemcitabine has a synergistic effect and is well-efficacious in pancreatic cancer patients with low TPX2 expression.
Introduction
Pancreatic cancer is one of the most aggressive cancers, with approximately 13% of patients surviving 5 years after diagnosis.1 In addition to surgery, the majority of conventional pancreatic cancer therapies involve DNA-damaging agents that cause the formation of either covalent DNA adducts and crosslinks or single-strand or double-strand breaks (DSBs), inducing genomic instability and leading to cell death. Despite their dismal survival rates, certain subsets of patients, such as those with homologous recombination (HR)-deficient pancreatic cancer, are indeed more vulnerable to DNA-damaging treatments.2–4
Recently, PARP inhibitors (PARPi) have shown promising synthetic lethality in HR-deficient cancers with BRCA1/2 mutations.5–7 Although PARPi have been approved for the maintenance treatment of patients with metastatic pancreatic cancer carrying pathogenic germline BRCA1/2 mutations, mutations in BRCA1/2 are notably rare.2 8–10 Aside from the inactivation of mutations in recognised HR mediators such as BRCA1/2, other mechanisms, such as the silencing of BRCA1 activity and disturbances in the HR pathway in other known and undiscovered mediators of this route, may result in HR deficiency. This has sparked much interest in ways to induce ‘BRCAness’ phenotypes. The identification and discovery of the full extent of the ‘BRCAness’ phenotype in pancreatic cancer highlights the need for new strategies to improve the efficacy and applicability of PARPi.
DNA repair is crucial to genomic stability. Knowledge of how DNA repair works and how to target it can be used to induce ‘BRCAness’. HR and non-homologous end joining (NHEJ) compete to repair DNA DSBs throughout the cell cycle. HR activity is inhibited in G1, although NHEJ and HR are active in G2 and S. The reason that NHEJ does not regularly outperform HR in S and G2 is unknown. ATM coordinates the recruitment of 53BP1 or BRCA1 to damaged DNA in response to DSBs, determining whether the DSB repair route is chosen.11–13 Tumour cells choose between DSB repair pathways that compete for DSBs to determine their fate. NHEJ directly binds to DNA breaks in the error-prone repair pathway, which can cause mutations and cell death. A homologous template makes HR a precise and error-free DNA repair mechanism.14 53BP1 and BRCA1 oppose each other, switching the DSB repair route to NHEJ or HR.12 15–17 If HR is absent, error-prone DNA repair pathways such as NHEJ are used, causing genomic instability.
In addition to traditional DNA damage response (DDR) factors, mitotic proteins influence DSB repair pathways. MAD2L2 prevents DNA end resection, enhancing NHEJ.18–20 In BRCA-deficient cells, MAD2L2-shieldin complex inhibition impairs NHEJ and causes PARPi resistance.21 ATM phosphorylation of Bub3 increases its association with the Ku70/Ku80/DNA-PKcs complex and promotes NHEJ.22 The mitotic checkpoint serine/threonine-protein kinase BuB1 must associate with 53BP1 for NHEJ.23 Targeting protein for Xenopus kinesin-like protein 2 (TPX2) is a microtubule-associated protein that controls the mitotic spindle and activates the cell cycle kinase Aurora A.24 25 The role of TPX2 in different stages of the cell cycle is unclear. Researchers have reported that TPX2 plays a role in the DDR.26 However, the detailed mechanism is not clear. A high-throughput screen suggested that TPX2 may be a substrate of ATM/ATR kinase in HEK293 cells.27 Does the phosphorylation of TPX2 occur in pancreatic cancer? If so, which kinase can phosphorylate TPX2? The functional significance of TPX2 phosphorylation also remains unknown.
Herein, we demonstrated that TPX2 operates in a cell cycle-dependent manner in pancreatic cancer, with ATM-mediated S634 phosphorylation determining DSB repair pathway choice and mitotic progression. Inhibiting TPX2 or S634 phosphorylation enhances PARPi sensitivity. We further proposed that patients with low TPX2 or S634 phosphorylation may benefit from combined PARPi and gemcitabine treatment. We also designed a cell-penetrating peptide (CPP) to target TPX2 S634 phosphorylation, offering mechanistic insights and a potential therapeutic strategy for pancreatic cancer.
Materials and methods
Additional materials and methods are included in the online supplemental material.
Results
TPX2 inhibition sensitises pancreatic cancer to olaparib
Six patient-derived xenograft (PDX) models with varying olaparib responses were created in this investigation (online supplemental figure S1A–F). We used RNA sequencing to examine the gene expression profiles of olaparib-resistant and olaparib-sensitive PDX models to identify targets that confer synergistic effects with olaparib (figure 1A). We identified TMEM117, CCDC86, MYBBP1A, TPX2 and CHERP as the top five different expression genes (figure 1B). Suppressing TPX2, CCDC86, MYBBP1A or CHERP in combination with olaparib significantly inhibited PANC-1 cell proliferation (online supplemental figure S1G), with TPX2 showing the most synthetic lethal effects, suggesting that TPX2 may be a synergistically lethal target. We also validated TPX2 expression in these six PDXs and found that TPX2 was downregulated in PDX models sensitive to olaparib (online supplemental figure S1H,I).
Supplemental material
TPX2 inhibition sensitises pancreatic cancer to olaparib. (A) Schematic model for comparing gene expression between olaparib-resistant and olaparib-sensitive PDX models. (B) Different gene expression profiles of olaparib-resistant and olaparib-sensitive PDX models. (C–E) TPX2-high PDX and TPX2-low PDX xenografts were treated daily with intraperitoneal olaparib or PBS (6 mice per cohort). The mice were sacrificed after 21 days of therapy. (C) Photograph of the tumours harvested at the end of the experiment with the PDX models. (D) Quantification of the weights of the tumours in the different groups of PDX models. The data are presented as the means±SDs (two-way ANOVA) and (E) growth curves of the PDX models treated with vehicle or olaparib. The data are presented as the means±SDs (one-way ANOVA). (F) TPX2-high PDX and TPX2-low PDX xenografts were treated daily with intraperitoneal olaparib or PBS (6 mice per cohort) for 21 days. Mice were presumed to be dead when the tumour volume reached 1500 mm3. K-M survival curves for the indicated PDX groups. The p value was calculated by the log rank test. (G–K) PDX tumours were subjected to immunological staining to detect the apoptosis marker cleaved caspase 3, the cell proliferation marker Ki67, and the DSBs markers γ-H2AX and TPX2 in serial sections. (G) Representative IHC micrographs, and (H–K) the histological score (H score) or the relative fluorescence intensity of the indicated markers were quantified from three separate fields from six tumours from six individual mice in each of the indicated treatment groups (two-way ANOVA). Scale bar, as indicated. (L,M) Representative image and quantification of metaphase chromosome spreads in the indicated PDXs (two-way ANOVA). Scale bar, 10 µm. (N–O) Representative images (N) and quantification (O) of tumour viability in the indicated groups after treatment with vehicle, olaparib (1 µM) for 72 hours. Scale bars, 100 µm. The data are presented as the means±SDs from three independent experiments (Student’s t test). (P) Representative micrographs showing TPX2 expression correlated with fluorescence intensity of γ-H2AX in 104 pancreatic cancer tissues from Fudan University Shanghai Cancer Center. Scale bar, 100 µm. IHC, immunohistochemical; PDX, patient-derived xenograft.
We then constructed stable TPX2-knockdown PANC-1 and MIAPaCa-2 cell lines to evaluate the impact of TPX2 on the sensitivity of pancreatic cancer cells to olaparib (figure S1J). Interestingly, although TPX2 is an essential gene, its knockdown did not significantly affect cell proliferation or apoptosis in these cell lines (online supplemental figure S1K–N). However, reduced TPX2 expression significantly increased the sensitivity of these cells to olaparib both in vitro and in vivo (online supplemental figure S1K–R). Additionally, we developed an olaparib-resistant PANC-1 cell line (online supplemental figure S2A). Knocking down TPX2 in this resistant cell line (online supplemental figure S2B) had no effect on cell proliferation or apoptosis but markedly increased cell sensitivity to olaparib both in vitro and in vivo (online supplemental figure S2C–G).
We established two new PDX models with varying levels of TPX2 expression (online supplemental figure S2H,I) to evaluate tumour size and survival. Tumours with lower TPX2 expression showed increased sensitivity to olaparib (figure 1C–E). While olaparib slightly prolonged survival in mice with high TPX2 tumours, it significantly prolonged survival in mice with low-TPX2 tumours (figure 1F). Histological analysis of γ-H2AX levels confirmed the extent of the DSBs, revealing a negative correlation between TPX2 expression and γ-H2AX levels after olaparib treatment (figure 1G–I), indicating that lower TPX2 expression leads to increased DSB accumulation. Compared with the TPX2-high groups, the TPX2-low PDX presented higher cleaved caspase 3 expression and lower Ki67 levels after olaparib treatment (figure 1G, J and K). Additionally, the number of olaparib-induced chromosomal aberrations was significantly greater in TPX2-low PDXs (figure 1L, M). We also used patient-derived organoid (PDO) models to assess olaparib sensitivity. TPX2-knockdown PDOs displayed increased sensitivity to olaparib (figure 1N, O, online supplemental figure S2J–M). These results support the idea that TPX2 inhibition synergistically enhances the efficacy of olaparib in pancreatic cancer.
Since PARPi and other chemotherapy agents for patients with pancreatic cancer cause DNA damage, we examined TPX2’s role in DSB accumulation in a clinical setting. Using γ-H2AX as a marker to measure DSB levels in 104 patients with pancreatic cancer, we found that high TPX2 expression correlates with fewer DSBs (figure 1P, online supplemental figure S2N).
ATM mediates TPX2 S634 phosphorylation in response to DNA damage in S and G2 phases
Since the accumulation of DSBs is generally associated with a defect in the DNA-repair capacity of the cell, we first investigated the impact of TPX2 on DNA repair in pancreatic cancer (online supplemental figures S3,S4). A high-throughput screen suggested that ATM/ATR kinase may phosphorylate TPX2 at S634 in HEK293 cells,27 we validate ATM-induced TPX2 S634 phosphorylation in response to DNA damage (online supplemental figure S5). We generated an antibody targeting TPX2 S634 phosphorylation and confirmed that this modification is DNA damage-induced and ATM-dependent (figure 2A,B, online supplemental S6A,B). To investigate TPX2 S634 phosphorylation in DNA repair, we exposed cells to ionising radiation (IR) and analysed TPX2 S634 phosphorylation at different time points. TPX2 S634 phosphorylation was rapidly induced, indicating its role in DDR (figure 2C, online supplemental S7A). Given that TPX2 is a cell cycle-dependent protein, PANC-1 cells were synchronised in late G1 using a double-thymidine block and then released to progress through the S phase to the G2/M phase to determine the specific cell cycle stage at which IR-induced TPX2 S634 phosphorylation occurs (figure 2D, online supplemental S7B). We found that IR-induced TPX2 S634 phosphorylation is also cell cycle dependent and increases from the S to G2 (figure 2E). In addition, UV laser microirradiation showed TPX2 recruitment primarily in S and G2 phases (figure 2F).
ATM-mediated TPX2 S634 phosphorylation is required for DDR in the S and G2 phases. (A) Dot blot showing the specificity of the anti-phospho-S634 TPX2 antiserum for the WG010939 (TPX2 S634 peptide) and control peptides at the indicated concentrations. (B) PANC-1 cells were pretreated with DMSO or an ATM inhibitor (Ku55933, 10 µM) for 1 hour and then further treated with or without IR (6 Gy) for 2 hour. Cell lysates were immunoblotted with the indicated antibodies. (C) Western blot analysis with the indicated antibodies at the indicated time points. (D) Schematic of the double-thymidine synchronisation experiments. (E) Immunoblotting analysis of TPX2 and TPX2 S634 expression in synchronised and released PANC-1 cells. (F) The indicated PANC-1 cells were subjected to laser-microirradiation. TPX2 and γ-H2AX were analysed by immunofluorescence. (G) Schematic of the DR-GFP/EJ5-GFP reporter system. (H–I) HR (H) and NHEJ (I) repair efficiency in the indicated DR-U2OS or EJ5-U2OS cells. The data are presented as the means±SDs (Student’s t test). (J) Representative immunofluorescence micrographs and quantification of BRCA1, RAD51, 53BP1 and RIF1 in the indicated PANC-1 cell lines at 6 hour after IR. Scale bar, 5 µm. At least 50 cells were analysed per cell line/experiment in IF experiments. The data are presented as the means±SDs of three independent experiments (Student’s t test). (K) Western blot analysis of total or chromatin-enriched extracts from the indicated PANC-1 cells. IR (6 Gy). (L) Western blot analysis of total or chromatin-enriched extracts from the indicated PANC-1 cells. IR (6 Gy), and Ku55933 (10 µM). (M) Representative immunofluorescence micrographs and quantification of BRCA1 and 53BP1 in the indicated PANC-1 cell lines by double-thymidine synchronisation experiments (Student’s t test). Scale bar, 5 µm. DMSO, dimethyl sulfoxide; DDR, DNA damage response; HR, homologous recombination; NHEJ, non-homologous end joining.
To assess the role of TPX2 S634 phosphorylation in DSB repair, we used a DR-GFP/EJ5-GFP reporter system (figure 2G). The S634A mutation reduced HR repair efficiency and increased NHEJ efficiency (figure 2H–I). We expressed WT or S634A TPX2 in TPX2-knockdown PANC-1 and MIAPaCa-2 cells (online supplemental figure S7C) to study the function of ATM-mediated S634 phosphorylation. S634A cells showed decreased BRCA1 and RAD51 IR-induced foci (IRIF) but increased 53BP1 and RIF1 IRIF (figure 2J, online supplemental figure S7D–H). Chromatin-enriched fractions revealed reduced BRCA1 and RAD51 and increased 53BP1 and RIF1 accumulation in S634A cells (figure 2K). No difference in chromatin protein accumulation was observed between WT and S634A cells treated with the ATM inhibitor Ku55933, indicating TPX2 S634 phosphorylation is ATM-dependent (figure 2L). In addition, TPX2-knockdown cells showed significantly reduced BRCA1, RAD51, 53BP1 and RIF1 chromatin accumulation (online supplemental figure S7I), indicating that TPX2 is upstream of these proteins. Moreover, S634A cells showed decreased BRCA1 IRIF and increased 53BP1 IRIF in S and G2 phases (figure 2M). These results indicated that ATM-mediated-TPX2 S634 phosphorylation promotes BRCA1 and RAD51 recruitment to chromatin and inhibits 53BP1 and RIF1 accumulation, specifically during S and G2 phases.
TPX2 S634 phosphorylation facilitates the recruitment of BRCA1, whereas the unphosphorylated form of TPX2 interacts with 53BP1
To investigate the role of TPX2 S634 phosphorylation in the recruitment of HR or NHEJ factors, coimmunoprecipitation (co-IP) following mass spectrometry (MS) was used to identify HR and NHEJ proteins that interact with TPX2 (online supplemental figure S8A,B). We found that the interaction of TPX2 with ATM, 53BP1 and BRCA1 was increased after IR (figure 3A, online supplemental figure S8C,D) and confirmed their interactions in S and G2 phases (figure 3B). Truncation studies revealed that TPX2 residues 561–747 are essential for binding ATM and BRCA1, whereas residues 149–393 are critical for the 53BP1 interaction (figure 3C–D). Computer simulations indicated that TPX2 S634 does not interact with 53BP1 but can bind to the C-terminus of BRCA1 (online supplemental figure S8E,F).
ATM-mediated TPX2 S634 phosphorylation promotes BRCA1 recruitment to DSBs for HR repair, whereas unphosphorylated TPX2 interacts with 53BP1 to recruit it for NHEJ. (A) Flag-tagged TPX2 in PANC-1 cells was immunoprecipitated with a Flag antibody, and Western blotting was performed using the indicated antibodies. (B) Co-IP analysis of the associations of TPX2 with the indicated proteins in the indicated cells by double-thymidine synchronisation experiments. (C) Summary of the binding domains between TPX2 and ATM, BRCA1 or 53BP1. (D) PANC-1 cells expressing Flag-tagged WT TPX2 or the indicated Flag-tagged TPX2 fragments were immunoprecipitated with an anti-Flag antibody and immunoblotted with the indicated antibodies. (E) PANC-1 cells expressing with Flag-tagged TPX2 WT or TPX2 S634A were treated with or without IR (6 Gy) for 2 hour. Cell lysates were immunoprecipitated with IgG or a Flag antibody and immunoblotted with the indicated antibodies. (F) PANC-1 cells expressing with the Flag-tagged vector or Flag-tagged TPX2 WT were pretreated with shATM for 24 hours and then further treated with or without IR (6 Gy) for 2 hour. Cell lysates were immunoprecipitated with a Flag antibody and immunoblotted with the indicated antibodies. (G) PANC-1 cells expressing with the Flag-tagged vector or Flag-tagged TPX2 WT were pretreated with DMSO or an ATM inhibitor (Ku55933, 10 µM) for 1 hour and then further treated with or without IR (6 Gy) for 2 hour. Cell lysates were immunoprecipitated with Flag antibody and immunoblotted with the indicated antibodies. (H) Representative images and quantification of Duo-linked proteins (Flag and BRCA1) in situ in Flag-TPX2 WT or S634A PANC-1 cells with or without IR. At least 50 cells were analysed. The data are presented as the means±SDs from three independent experiments (Student’s t test). (I) Representative images and quantification of Duo-linked proteins in situ in (Flag and 53BP1) PANC-1 cells with or without IR. At least 50 cells were analysed. The data are presented as the means±SDs from three independent experiments (Student’s t test). (J) Cell lysates were immunoprecipitated with IgG or BRCA1 antibody and immunoblotted with the indicated antibodies. (K) Cell lysates were immunoprecipitated with IgG or 53BP1 antibody and immunoblotted with the indicated antibodies. (L) Representative immunofluorescence micrographs and quantification of TPX2 pS634 and BRCA1 foci formation in PANC-1 cells. (M) Representative immunofluorescence micrographs and quantification of TPX2 pS634 and 53BP1 foci formation in PANC-1 cells. (N) Western blot analysis of total or chromatin-enriched extracts from the indicated PANC-1 cells without IR or at 6 hour after IR with 6 Gy using the indicated antibodies. DMSO, dimethyl sulfoxide; DSBs, double-strand breaks; HR, homologous recombination; NHEJ, non-homologous end joining.
We then examined the effect of TPX2 S634 phosphorylation on its interactions with BRCA1 and 53BP1. IR enhanced both TPX2-BRCA1 and TPX2-53BP1 interactions, but S634A reduced TPX2-BRCA1 binding and increased TPX2-53BP1 binding, specifically in G2 and S phases (figure 3E, online supplemental figure S8G). ATM inhibition also decreased TPX2-BRCA1 and increased TPX2-53BP1 interactions (figure 3F–G, online supplemental figure S8H). We also employed an in situ proximity ligation assay to determine the direct interaction. S634A significantly reduced TPX2-BRCA1 foci and increased TPX2-53BP1 foci after IR compared with WT (figure 3H–I).
Additionally, BRCA1 interacted with phosphorylated TPX2, while 53BP1 did not (figure 3J–K). Phosphorylated TPX2 co-localised with BRCA1 IRIF (figure 3L) but not with 53BP1 foci (figure 3M). Given that TPX2 residues 149–393 are critical for the interaction with 53BP1 (figure 3D), the retention of residues 394–747 interferes with 53BP1 binding, resulting in the inhibition of its chromatin localisation (figure 3N). Together, these results indicated that phosphorylated TPX2 interacts with BRCA1, facilitating BRCA1 and RAD51 recruitment to chromatin, while unphosphorylated TPX2 interacts with 53BP1, promoting its chromatin localisation. The phosphorylated and non-phosphorylated forms of TPX2 control whether it recruits to BRCA1 or to 53BP1 in chromatin.
The ratio of phosphorylated- to non-phosphorylated TPX2 determines the DSB repair pathway choice
We synthesised five short peptides fused with CPPs targeting TPX2 phosphorylated at S634 (figure 4A) to investigate the role of phosphosphorylated and non-phosphorylated TPX2 in the choice of the DSB repair pathway. S634 peptide 3# had the most potent inhibitory effect on TPX2 phosphorylation (figure 4B). We also synthesised the S634A peptide, a scrambled peptide in which an A residue was used to replace the S634 residue, based on S634 peptide 3#. The phosphorylation of TPX2 at S634 was significantly reduced by S634-peptide 3#, but not by S634A-peptide (figure 4C–E).
The balance between phosphorylated and non-phosphorylated TPX2 determines the choice of DSB repair pathway. (A) Schematic illustration of the designed peptides. (B–E) PANC-1 cells were immunoprecipitated with an anti-Flag antibody, and Western blotting was performed using the indicated antibodies. (F–K) Representative immunofluorescence micrographs and quantification of BRCA1, RAD51, 53BP1 and RIF1 in the indicated PANC-1 cell lines at 6 hour after IR. Scale bar, 5 µm. At least 50 cells were analysed per cell line/experiment in IF experiments. The data are presented as the means±SDs of three independent experiments (one-way ANOVA). (L) Western blot analysis of total or chromatin-enriched extracts from the indicated PANC-1 cells at 6 hour after IR with 6 Gy using the indicated antibodies. (M,N) HR and NHEJ repair efficiency in the indicated DR-U2OS or EJ5-U2OS cells. The data are presented as the means±SDs (one-way ANOVA). (O) Cell lysates were immunoprecipitated with an anti-Flag antibody and immunoblotted with the indicated antibodies. (P) Proposed model for the role of TPX2 in DSB repair pathway choice. DSBs, double-strand breaks; HR, homologous recombination; NHEJ, non-homologous end joining.
In response to DNA damage, S634-peptide #3 dose-dependently reduced BRCA1 and RAD51 foci while increasing 53BP1 and RIF1 foci (figure 4F–K, online supplemental figure S9A–F). The chromatin-enriched fractions also presented decreased BRCA1 and RAD51 recruitment and increased 53BP1 and RIF1 accumulation at chromatin after S634-peptide 3# treatment (figure 4L). Additionally, S634-peptide 3# reduced HR repair efficiency and increased NHEJ efficiency (figure 4M–N) and enhanced TPX2-53BP1 interaction while decreasing TPX2-BRCA1 interaction in a dose-dependent manner (figure 4O). These data indicate that ATM phosphorylates TPX2 to enhance its binding to BRCA1, promoting HR by facilitating BRCA1 recruitment to DSBs. Unphosphorylated TPX2 interacts with 53BP1, aiding its binding to damaged chromatin for NHEJ (figure 4P). The balance between phosphorylated and non-phosphorylated TPX2 determines the choice of DSB repair pathway.
TPX2 S634 phosphorylation is required for mitosis and chromosomal stability
To explore if ATM-mediated TPX2 S634 phosphorylation affects mitosis, we found that nocodazole-induced mitotic arrest increases TPX2 S634 phosphorylation and enhances its interaction with ATM in PANC-1 cells (figure 5A–C). The S634A mutation in TPX2 reduced its interaction with Aurora A (figure 5D) and decreased Aurora A phosphorylation (Thr 288) (figure 5E), indicating that TPX2 S634 phosphorylation impacts mitotic function by regulation of Aurora A activity. To exclude the potential effect of nocodazole to distinguish pure mitotic events, mitotic shake-off cells were collected from exponentially growing PANC-1 and MIAPaCa-2 cells. We discovered that TPX2 can be phosphorylated at S634 during mitosis and that this phosphorylation enhances the binding of TPX2 to Aurora A and promotes Aurora A activity (figure 5F–H, online supplemental figure S10A–C).
TPX2 S634 phosphorylation is required for mitosis and chromosomal stability. (A) Schematic of the synchronisation experiments. PANC-1 cells were synchronised with nocodazole (Noco) treatment (100 nM). (B) Nocodazole was used to arrest cells during mitosis. PANC-1 cells were treated with nocodazole (100 nM) for 17 hours, and the cell lysates were subjected to immunoblotting with the indicated antibodies. (C) PANC-1 cells treated with DMSO or nocodazole (100 nM) for 17 hours were immunoprecipitated with an anti-Flag antibody and immunoblotted with the indicated antibodies. (D) Flag-tagged TPX2 WT or S634A PANC-1 cells treated with nocodazole (100 nM) for 17 hours were immunoprecipitated with an anti-Flag antibody and immunoblotted with the indicated antibodies. (E) The indicated PANC-1 cells treated with nocodazole (100 nM) for 17 hours were immunoblotted with the indicated antibodies. (F) Mitotic shake-off was used to arrest PANC-1 cells during mitosis. The indicated mitotic shake-off cells were immunoblotted with the indicated antibodies. (G) The indicated mitotic shake-off PANC-1 cells were immunoprecipitated with an anti-Flag antibody and immunoblotted with the indicated antibodies. (H) The indicated mitotic shake-off PANC-1 cells were immunoblotted with the indicated antibodies. (I) The indicated PANC-1 cells transfected with H2B-GFP plasmids were imaged at the onset of mitosis to monitor chromosomal dynamics. Representative fluorescence images are shown. Scale bar, 10 µm. (J) The average time from nuclear envelope breakdown (NEB) to anaphase onset in the indicated PANC-1 cells was measured by time-lapse microscopy. At least 50 cells were analysed per cell line/experiment. The data are presented as the means±SDs (Student’s t test). (K–L) The indicated PANC-1 cells were treated with nocodazole (100 nM) for 17 hours and stained with a flow cytometry-based anti-phospho-histone H3-Ser10 antibody to determine the mitotic index. (K) Representative images are shown and (L) quantification of mitotic cells. The data are presented as the means±SDs from three independent experiments (Student’s t test). (M–N) The indicated PANC-1 cells were stained with α-Tubulin and DAPI to visualise microtubules (green) and chromosomes (blue). (M) Representative images are shown. Scale bar, 10 mm, and (N) quantification of the mitotic defeat cells. At least 50 cells were analysed per cell line/experiment. The data are presented as the means±SDs (Student’s t test). (O,P) (O) Representative images of the chromosome spread assay. Scale bar, 10 µm and (P) quantification of aneuploidy in the indicated cells. At least 50 cells were analysed. The data are presented as the means±SDs (Student’s t test). (Q) Proposed model for the role of TPX2 in mitosis. DMSO, dimethyl sulfoxide.
To investigate the process of mitosis, we stably transfected a GFP-tagged histone plasmid into PANC-1 cells with stable knockdown of TPX2 and reconstituted the cells with WT or S634A, after which chromosomal dynamics and the timing of mitosis were monitored via time-lapse microscopy (figure 5I–J, online supplemental movies 1–4). We found that TPX2 knockdown resulted in chromosome misalignment and prolonged the duration of mitosis. TPX2 WT rescued this process, whereas TPX2 S634A did not. We also detected a substantial increase in mitotic arrest in TPX2 knockdown cells, as evaluated by flow cytometry analysis of histone H3-serine 10 phosphorylation in response to nocodazole treatment (figure 5K–L). TPX2 WT rescued mitotic arrest, whereas S634A TPX2 did not. These findings indicated that TPX2 phosphorylation is required for mitosis.
Supplementary video
Supplementary video
Supplementary video
Supplementary video
We further assessed chromosomal stability in these cells. We discovered that TPX2 knockdown or TPX2 S634A mutation resulted in a considerably greater proportion of cells with multipolar spindles, suggesting that TPX2 phosphorylation is required for mitotic cell division (figure 5M–N). In addition, the number of aneuploid TPX2 knockdown or TPX2 S634A mutant cells was much greater than that of control or WT cells (figure 5O–P). Taken together, these data suggested that during mitosis, ATM-mediated TPX2 S634 phosphorylation enhances Aurora A activity, hence facilitating mitotic progression and maintaining chromosomal stability (figure 5Q).
TPX2 inhibition or TPX2 S634A mutation synergises with olaparib, causing genomic instability and mitotic catastrophe
To investigate the effect of olaparib, we tested whether it causes TPX2 S634 phosphorylation. Olaparib increased TPX2 S634 phosphorylation in a dose-dependent manner (figure 6A, online supplemental figure 11A) and enhanced the TPX2-BRCA1 and TPX2-53BP1 interactions. The S634A mutation promoted the TPX2-53BP1 interaction and reduced TPX2-BRCA1 binding (figure 6B). Following olaparib treatment, ATM knockdown or Ku55933 treatment decreased the TPX2-BRCA1 interaction and increased TPX2-53BP1 interaction (online supplemental figure 11B,C). Phosphorylated TPX2 did not interact with 53BP1 but did interact with BRCA1 (figure 6C,D). TPX2 knockdown or S634A mutation exacerbated olaparib-induced DNA damage, as evidenced by increased tail moments and γ-H2AX foci (figure 6E,F, online supplemental figure S11D,E). Similar to IR, TPX2 S634A increased 53BP1 and RIF1 in chromatin and decreased BRCA1 and RAD51 in cells treated with olaparib, with no difference in chromatin protein levels when ATM was suppressed (figure 6G,H).
TPX2 knockdown or the TPX2 S634A mutation synergises with olaparib, causing genomic instability and mitotic catastrophe. (A) PANC-1 cells were treated with olaparib (0 µM, 0.25 µM, 0.5 µM, 0.75 µM, 1 µM or 2 µM) for 24 hours, and the cell lysates were immunoblotted with the indicated antibodies. (B–D) Co-IP of cellular extracts from the indicated PANC-1 cells treated with olaparib (1 µM). (E) Quantification of the neutral comet assay in the indicated PANC-1 cells at 24 hours after olaparib (1 µM) treatment. Scale bar, 10 µm. The tail moment was analysed using CometScore software. At least 50 cells were analysed. The data are presented as the means±SDs (Student’s t test). (F) Quantification of γ-H2AX foci in the indicated PANC-1 cells at 24 hours after olaparib (1 µM) treatment. Scale bar, 5 µm. At least 50 cells were analysed. The data are presented as the means±SDs from five independent experiments (Student’s t test). (G,H) Western blot analysis of total or chromatin-enriched extracts from the indicated PANC-1 cells treated with Olaparib (1 µM) and the indicated antibodies. (I–L) The indicated PANC-1 cells transfected with H2B-GFP plasmids and treated with DMSO or olaparib (1 µM) were imaged at the onset of mitosis to monitor chromosomal dynamics. Scale bar, 10 µm. (M) Experimental procedure for analysing the cell cycle distribution of the indicated PANC-1 cells after olaparib (1 µM) treatment. DNA>4 n was detected by fluorescence-activated cell sorting for DNA content with propidium iodide. The data are presented as the means±SDs of three independent experiments (Student’s t test). (N) Immunoblotting for TPX2-pS634 in the indicated PDAC cell lines. β-actin was used as a loading control. (O) An MTS assay was used to evaluate the proliferation of the indicated PDAC cell lines after exposure to olaparib for 72 hours. The data are presented as the means±SDs from three independent experiments (two-way ANOVA). DMSO, dimethyl sulfoxide; PDAC, pancreatic ductal adenocarcinoma.
We used live-cell imaging to examine whether olaparib induced chromosomal aberrations in TPX2-knockdown or TPX2 S634A cells. Olaparib caused mitotic catastrophe in TPX2-knockdown and TPX2 S634A cells, but not in control or WT cells (figure 6I–L, online supplemental movies 5–12). TPX2-knockdown and TPX2 S634A cells also showed increased DNA with more than 4 N after olaparib treatment (figure 6M). Figure 6N,O revealed that cells with lower levels of TPX2 S634 phosphorylation were more sensitive to olaparib. These results indicated that the TPX2 S634A mutation enhances genomic instability and mitotic catastrophe in combination with olaparib.
Supplementary video
Supplementary video
Supplementary video
Supplementary video
Supplementary video
Supplementary video
Supplementary video
Supplementary video
Coupling of gemcitabine with TPX2 inhibition or TPX2 S634A mutation obviously improved pancreatic cancer sensitivity to olaparib in vitro and in vivo
In patients with defective homologous DNA repair, combining chemotherapy with PARPi treatment can be beneficial. Given that gemcitabine is a standard chemotherapy for treating pancreatic ductal adenocarcinoma (PDAC), we hypothesised that combining gemcitabine with TPX2 inhibition would increase sensitivity to olaparib. TPX2 knockdown or the S634A mutation, either alone or in combination with gemcitabine, significantly reduced the IC50 of olaparib in PANC-1 cells (figure 7A,B, online supplemental figure 12A). Furthermore, both TPX2 knockdown and the S634A mutation enhanced cell sensitivity to olaparib and gemcitabine. The combination of gemcitabine with TPX2 inhibition or the S634A mutation markedly increased pancreatic cancer sensitivity to olaparib (figure 7C,D, online supplemental figure S12B,C). These treatments also increased olaparib-induced apoptosis (figure 7E, online supplemental figure S12D). We also examined the in vivo combination efficacy in xenografts. Tumours with TPX2 knockdown and gemcitabine treatment were more sensitive to olaparib (figure 7F,G, online supplemental figure 12E).
Coupling gemcitabine with TPX2 inhibition or the TPX2 S634A mutation improved pancreatic cancer sensitivity to olaparib in vitro and in vivo (A,B) Olaparib IC50 (A) and dose-response curves (B) of the indicated cells treated with or without gemcitabine (100 nM) for 72 hours. (C,D) Colony formation assays and relative survival of the indicated cell lines after exposure to olaparib (1 µM) and/or gemcitabine (100 nM) (Student’s t test). (E) Quantification of flow cytometry results for propidium iodide and annexin V staining in the indicated cell lines after exposure to olaparib (1 µM) and/or gemcitabine (100 nM) for 48 hours. The data are presented as the means±SDs of three independent experiments (Student’s t test). (F,G) Xenografts and tumour weights obtained from mice in different groups treated with vehicle, olaparib (50 mg/kg per day), gemcitabine (5 mg/kg per 3 days) or their combination (Student’s t test). Scale bars, 1.0 cm. (H) Representative images of tumour viability in the indicated groups after treatment with vehicle, olaparib (1 µM), gemcitabine (100 nM) or combined for 72 hours. Scale bars, 100 µm. (I) The combined effect of olaparib and gemcitabine in the indicated patient-derived tumouroid. PDAC, pancreatic ductal adenocarcinoma.
Furthermore, we employed PDO models to assess sensitivity to olaparib (online supplemental figure S12F). The TPX2 knockdown and TPX2 S634A PDO models showed greater sensitivity to olaparib as well as to gemcitabine (figure 7H, online supplemental figure S12G). Moreover, olaparib more effectively synergised with gemcitabine in the TPX2 knockdown and TPX2 S634A PDO models (figure 7I). Taken together, these findings suggest the therapeutic potential of gemcitabine in combination with olaparib in PDAC patients with lower TPX2 expression or TPX2 S634 mutation.
Targeting TPX2 S634 phosphorylation with a peptide inhibitor enhances olaparib sensitivity
Finally, we investigated the effect of S634-peptide 3# on olaparib sensitivity. TPX2 S634 phosphorylation was significantly reduced by S634-peptide 3#, but not by S634A-peptide in response to olaparib (figure 8 A, B). In addition, S634-peptide 3# significantly decreased the formation of BRCA1 and RAD51 foci and enhanced the formation of 53BP1 and RIF1 foci after olaparib treatment (figure 8C, online supplemental figure A–D). S634-peptide 3# significantly inhibited mitosis progression, and in cells treated with S634-peptide 3#, olaparib resulted in mitotic catastrophe (figure 8D, online supplemental figure S13E, online supplemental movies 13–18).
Supplementary video
Supplementary video
Supplementary video
Supplementary video
Supplementary video
Supplementary video
S634-peptide 3# is synergistically lethal with PARPis in pancreatic cancer. (A, B) PANC-1 cells were immunoprecipitated with an anti-Flag antibody, and western blotting was performed using the indicated antibodies. (C) Representative immunofluorescence micrographs of BRCA1 and 53BP1 foci in the indicated PANC-1 cell lines treated with olaparib. Scale bar, 5 μm. (D) CPP- and/or olaparib-treated PANC-1 cells were imaged at the onset of mitosis to monitor chromosomal dynamics. Scale bar, 10 μm. (E,F) An MTS assay was used to evaluate the proliferation of the indicated cell lines after exposure to olaparib or CPPs for 72 h. The data are presented as the means ± SDs (Student’s t test). (G,H) PDX tumour images and tumour weights are shown (Student’s t test). (I) Kaplan-Meier survival curves for the indicated PDX groups. (J) Immunological staining was used to detect cleaved caspase 3, Ki67, γ-H2AX and TPX2. Scale bar, as indicated. (K) Proposed model for dual functional significance of TPX2 in regulating olaparib sensitivity via the modulation of DSB repair pathway choice and mitotic progression. CPP, cell-penetrating peptide; DSBs, double-strand breaks; PDX, patient-derived xenograft.
S634-peptide 3# significantly increased pancreatic cancer cell sensitivity to olaparib, as shown in figure 8E, F, online supplemental figure S13F,G. In the TPX2-high pancreatic cancer PDX model, the combination of S634-peptide 3# and olaparib had a stronger synergistic lethal effect than the other treatments did (figure 8G, H, online supplemental figure S13H). While olaparib marginally prolonged survival in the S634A-peptide group, it significantly prolonged survival in the S634-peptide 3# group (figure 8I). TPX2 expression remained unchanged by either peptide, but olaparib with S634-peptide 3# increased γ-H2AX fluorescence (figure 8J, online supplemental figure 13I,J). Additionally, cleaved caspase 3 levels decreased, while Ki67 increased with the combination treatment (figure 8J, online supplemental figure S13K–L), highlighting the synergistic lethality of S634-peptide 3# with PARPi in pancreatic cancer.
Discussion
PARP inhibition is a new potential treatment option for PDAC, a refractory disease for which little therapeutic improvement has been achieved. While PARPi are FDA-approved for metastatic PDAC patients with germline BRCA1/2 mutations, there remains a significant unmet need to enhance their effectiveness. Beyond BRCA1/2 mutations, no other predictors of PARPi response in PDAC have been identified. Here, we demonstrate that TPX2 is a strong predictor of PARPi sensitivity in pancreatic cancer and a potential target for improving patient response to PARPi. These findings were confirmed in vivo using multiple PDX models. We propose a molecular mechanism underlying the role of TPX2 in DSB repair pathway choice and mitosis, which bolsters the utility of PARPi therapy for pancreatic cancer. We also demonstrated that combining PARPi with gemcitabine shows synergistic, well efficacious in pancreatic cancer cell lines and PDOs with low TPX2 expression.
Our findings indicate that TPX2 expression may serve as an independent marker for PARPi sensitivity in pancreatic cancer, regardless of BRCA mutation status. Approximately 24% of patients in our centre exhibited low TPX2 expression, suggesting that this subset could benefit from PARPi treatment. Moreover, our research demonstrates that TPX2 not only regulates BRCA1 but also influences the recruitment of 53BP1. Based on this, we hypothesise that TPX2 expression could further refine the stratification of PARPi response in patients with BRCA1/2 mutations. Moving forward, we plan to initiate a prospective clinical trial to evaluate the efficacy of PARPi in patients with low TPX2 expression.
Although HR suppression during G1 is widely understood, it is unknown why profuse NHEJ does not occur in S and G2, indicating that it is an active HR-promoting factor during the S/G2 phase. In our investigation, we discovered that ATM kinase, but not ATR kinase, phosphorylates TPX2 at S634 in response to DNA damage in the S and G2 phases. In the S/G2 phase, ATM-mediated TPX2 S634 phosphorylation increases BRCA1 recruitment to DSBs for HR repair through its association with BRCA1. Moreover, the interaction of non-phosphorylated TPX2 with 53BP1 promotes the recruitment of 53BP1 to damaged chromatin to execute NHEJ. The ratio of phosphorylated to non-phosphorylated TPX2 determines the DSB repair pathway choice. As a result, cells with high levels of TPX2 S634 phosphorylation undergo error-free HR to repair DSBs, whereas those with low levels of TPX2 S634 phosphorylation must undergo error-prone NHEJ to repair DSBs.
A crucial step in deciding between error-free HR and error-prone NHEJ is DNA-end processing to generate 3' tails, which is controlled by the interplay between 53BP1 and BRCA1.28 Although we demonstrated that TPX2 regulates the mutual exclusivity of 53BP1 and BRCA1, the direct involvement of TPX2 in DNA-end resection has yet to be established. Intriguingly, a study revealed that the complex of TPX2 and Aurora A seems to control 53BP1 during DSB repair, blocking BRCA1 antagonism and consequently permitting DNA-end resection and HR.26 However, the function of TPX2 phosphorylation in DSB repair has not been determined. The results of our studies and those in the literature are identical and complementary. A few studies have implicated TPX2 in HR,29–31 but none has demonstrated that TPX2 directly affects NHEJ. In the present study, we provide evidence that TPX2 is responsible for regulating both HR and NHEJ.
TPX2 is known to regulate spindle formation by increasing microtubule nucleation from chromatin and stabilising spindle microtubules in a Ran-dependent manner.32 In our investigation, we demonstrated a novel mechanism by which TPX2 is engaged in bipolar spindle formation and maintenance during mitosis in an ATM-dependent manner. ATM promotes TPX2 S634 phosphorylation and increases Aurora A activity during mitosis. The TPX2 S634A mutation significantly increases the frequency of aneuploid cells, demonstrating that TPX2 S634 phosphorylation is necessary for mitosis.
In this study, we developed a CPP that effectively blocks TPX2 S634 phosphorylation and demonstrates synergistic lethality with PARPi in pancreatic cancer. While peptides often encounter delivery challenges, small-molecule inhibitors could serve as viable alternatives with optimised selectivity and minimal off-target effects. Although TPX2 knockdown did not significantly increase cell death, it did lead to chromosome misalignment, suggesting that targeting TPX2 may pose toxicity and patient stratification challenges. However, advanced drug delivery methods, partial inhibition, schedule optimisation, formulation adjustments, and personalised dosing strategies could help mitigate these issues. Small-molecule inhibitors targeting essential genes like the CDK family, which are directly or indirectly involved in DNA repair, have demonstrated effectiveness in combination therapies without exhibiting toxicity in animal studies.33 34 Therefore, the development strategies for CDK inhibitors may offer valuable insights for the development of TPX2 inhibitors.
Overall, our findings reveal the dual-functional significance of TPX2 in controlling DNA DSB repair pathway choice and mitotic progression, suggesting that TPX2 serves as a novel target for expanding the utility of PARPi in pancreatic cancer through conferring synthetic lethality (figure 8K).
Supplemental material
Data availability statement
Data are available upon reasonable request. All data relevant to the study are included in the article or uploaded as supplementary information. The transcriptome data generated by the study were deposited in the National Omics Data Encyclopedia (OER399218) (https://www.biosino.org/node/). The mass spectrometry proteomics data were deposited to the ProteomeXchange Consortium via the iProX partner repository with the dataset identifier PXD043145 (https://www.iprox.cn/page/home.html). All the relevant data that support the findings of this study are available from the corresponding author on request.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants was approved by the Institutional Research Ethics Committee of FUSCC, and written informed consent was obtained from all patients (2401290-13). Participants gave informed consent to participate in the study before taking part.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1
- Data supplement 2
- Data supplement 3
- Data supplement 4
- Data supplement 5
- Data supplement 6
- Data supplement 7
- Data supplement 8
- Data supplement 9
- Data supplement 10
- Data supplement 11
- Data supplement 12
- Data supplement 13
- Data supplement 14
- Data supplement 15
- Data supplement 16
- Data supplement 17
- Data supplement 18
- Data supplement 19
- Data supplement 20
Footnotes
MX, RT and HP are joint first authors.
XY and SS are joint senior authors.
MX, RT and HP contributed equally.
Contributors MX, RT and HP contributed equally to this paper. MX, RT and HP performed the experiments and statistical analysis. JY, XT, HX, YG, YlL, DW, YbL, YH, ZM, WW and JX acquired data. RT contributed to the bioinformatics analysis. MX, SS and XY wrote the manuscript. SS and XY conceived, designed and supervised the study and provided a critical review of the manuscript. MX and SS are responsible for the overall content as the guarantor.
Funding This study was jointly supported by the National Natural Science Foundation of China (92374102 to SS), the National Natural Science Foundation of China (U21A20374 to Xianjun Yu), Natural Science Foundation of Shanghai (23ZR1479300 to Si Shi), Shanghai Municipal Science and Technology Major Project (21JC1401500 to Xianjun Yu) and Scientific Innovation Project of Shanghai Education Committee (2019-01-07-00-07-E00057 to Xianjun Yu)
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer-reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.