Treating cancer through inhibition of nuclear export is one of the best examples of basic research translation into clinical application. Nuclear export factor chromosomal region maintenance 1 (CRM1; Xpo1 and exportin-1) controls cellular localization and function of numerous proteins that are critical for the development of many cancer hallmarks. The diverse actions of CRM1 are likely to explain the broad ranging anti-cancer potency of CRM1 inhibitors observed in pre-clinical studies and/or clinical trials (phase I–III) on both advanced-stage solid and hematological tumors. In this review, we compare and contrast the mechanisms of action of different CRM1 inhibitors, and discuss the potential benefit of unexplored non-covalent CRM1 inhibitors. This emerging field has uncovered that nuclear export inhibition is well poised as an attractive target towards low-toxicity broad-spectrum potent anti-cancer therapy.
Nuclear export, mainly mediated by the nuclear export factor exportin-1 (better known as chromosomal region maintenance 1, CRM1), is an essential function in all eukaryote that transport nuclear export signal (NES) containing cargoes from the nucleus to the cytoplasm.1 Upregulation of this process is a common characteristic for a broad spectrum of cancers; inhibition of nuclear export kills cancer cells effectively, although its anti-cancer mechanism is not conclusive thus far.2,3 In addition, CRM1 has been shown to mediate drug resistance.4,5 Among dozens of CRM1 inhibitors discovered, a few were clinically tested or are undergoing clinical trials, including the first generation of CRM1 inhibitor, leptomycin B (LMB), and the second-generation CRM1 inhibitor SINE (specific inhibitor of nuclear export).6 In this review, we first present the background of nuclear–cytoplasmic transport, the nuclear export factor CRM1 and the cancer hallmark pathways affected by CRM1 inhibition. We then discuss the details of LMB and SINE, with both being covalent CRM1 inhibitors. Finally, we propose non-covalent CRM1 inhibitors as the next generation of anti-cancer drugs, and discuss their advantage over covalent inhibitors.
Eukaryotes are characterized by the presence of the cell nucleus, which is enclosed by a nuclear envelope and separated from the rest of the cell. The nuclear pore complex (NPC) is the sole gateway on the nuclear envelope that governs protein and nucleic acid exchange between the nucleus and cytosol.7 Although small molecules are freely permeable across the NPC, permeability is increasingly restricted as the molecular size approaches 30 kDa.8 Movement of bigger molecules or more efficient passage of smaller molecules in and out of the nucleus is mediated by active transport of soluble transport factors called karyopherin proteins.9,10 The human genome encodes ~20 different karyopherin proteins, functioning as importin (for nuclear import), exportin (for nuclear export) or transportin (both import and export), each being responsible for transporting a set of cargoes (protein or RNA) containing specific sequences/motifs known as nuclear localization signal (NLS) or NES or both.11,
For nuclear import, a cargo with accessible NLS and an importin form a complex, which is imported into the nucleus together through the NPC (Figure 1).22,23 The small GTPase RanGTP in the nucleus then dissociates the cargo from the importin through direct or indirect competition.24,25 The RanGTP–importin complex is then recycled to the cytoplasm. After GTP hydrolysis by RanGAP and concomitant RanGDP dissociation, importin is ready for another cycle of nuclear import.26,27 For a cargo to exit the nucleus, it must display an NES, which cooperatively forms a tight trimeric complex with an exportin and RanGTP.28,
Nuclear export factor CRM1
Of the known exportins, CRM1 is an essential and most often used exportin in cells, which exports numerous cargoes including both proteins and RNAs.1,34,
CRM1 exports a long list of tumor suppressors or oncogenes, such as p53, FOXOs, p27, nucleophosmin, BCR–ABL, eIF4E and survivin, and these proteins are mislocalized to the cytoplasm in many cancer cell types (Figure 2).6,68,
Downregulating cancer hallmarks through CRM1 inhibition
During the multistep development of cancer, cells acquire unique biological properties that enable them to become neoplastic and eventually malignant.74 These properties include genomic instability, sustained proliferation, resistance to cell death, reprogramming of cellular energetics and so on, which are summarized as the hallmarks of cancer by Hanahan and Weinberg.75 Intriguingly, many CRM1 cargoes are found to be critical for at least nine hallmark features of cancer (Figure 2). Next, we will briefly discuss how the altered cellular distribution of CRM1 cargoes contributes to a particular cancer hallmark, and how CRM1 inhibition may reverse these processes, hopefully bringing some insights into CRM1 inhibitors’ broad-spectrum anti-cancer activity.
The most remarkable trait that cancer cells acquire is their ability to perpetually divide, resulting in uncontrolled proliferation.76 Many tumor-specific mechanisms are involved in this particular cancer cell trait. For instance, the proto oncogene BCR–ABL is formed by a fusion of the ABL1 (Abelson murine leukemia viral oncogene homolog 1) and the BCR (breakpoint cluster region) genes, resulting in a BCR–ABL chimeric protein, which constantly stimulates proliferation of myeloid cells.77,78 BCR–ABL is exported to the cytoplasm of cancer cells where it activates the PI3K/Akt pathway.77,79 CRM1 inhibition traps BCR–ABL in the nucleus, re-sensitizes leukemia cells to the BCR–ABL inhibitor imatinib, resulting in strong reduction of tumor cell proliferative potential with limited toxicity to normal myeloid precursors.80,81 In addition, the expression level of several master growth regulators, such as c-Myc, c-Met and epidermal growth factor receptor, is reduced by CRM1 inhibition through different mechanisms, which might be crucial for the reduced rate of tumor proliferation observed.53,59,60,82,83
Evading growth suppressors
Tumors evade powerful negative regulation of cell proliferation imposed by different growth suppressors such as retinoblastoma protein, p21 and p27.75 These cell cycle inhibitors function in the nucleus in normal cells, but are mislocalized to the cytoplasm by CRM1 in various cancers.84,
Genome instability and mutation
In cancer cells, the DNA maintenance machinery is often mutated or mislocalized, thereby facilitating alterations of the genome and the acquisition of multiple hallmarks subsequently.75 p53 is a well-known genome guardian, which has pivotal roles in sensing and repairing DNA damage.92 Besides p53 mutations, cancer cells can evade p53 survillience through CRM1-mediated p53 nuclear export.93,94 Treatment with CRM1 inhibitors results in increased nuclear p53 level, triggering p53-mediated transcription and apoptosis.58,95 Similarly, several other proteins critical for genome stability are exported to the cytoplasm in different types of cancer cells, including HSP90, nucleophosmin and PTEN.96,
Resisting cell death
In addition to the sustained proliferation ability, cancer cells must bypass programmed cell death by apoptosis.75 Survivin, a member of the inhibitor of apoptosis family, is localized in both the nucleus and the cytoplasm of tumor cells, and it is the cytosolic fraction that exerts the cancer-promoting activity.99,100 Inhibition of nuclear export by survivin NES antibodies promotes the nuclear accumulation and degradation of survivin, which abolishes its cytoprotective function.101,102 In another study, nuclear accumulation of pro-apoptotic protein Bok (Bcl-2-related ovarian killer) by mutation of its NES or CRM1 inhibition causes apoptosis in breast cancer cells.103 Another example pertains to FOXO family proteins, which are important transcription factors controlling the expression of apoptosis-related genes.104 Through phosphorylation events, their NESs are exposed, leading to FOXO’s cytoplasmic localization and loss of pro-apoptotic activity in cancer cells.18
Enabling replicative immortality
Maintenance of telomeres by telomerase is important for chromosome stabilization and cell immortalization.105 As such, telomerase is activated in germ cells and most cancers.106 Telomerase RNA subunit TLC1 must be exported into the cytoplasm to recruit the protein subunits for complete assembly of the enzyme, which is then imported into the nucleus to extend telomeres.107 Nuclear export of TLC1 requires both CRM1 and the messenger RNA export machinery.108 It is reported that nuclear export of TLC1 is an essential step for the formation of the functional RNA containing enzyme, and blocking TLC1 export prevents its cytoplasmic maturation and leads to telomere shortening.108
Tumor growth requires new blood vessels formation to supply nutrients for increasing mass of tumor cells.109 The well-known prototype of angiogenesis inducer is vascular endothelial growth factor.110 CRM1 inhibition causes nuclear retention of the NES-containing cargo Fbw7, a subunit of a ubiquitin ligase that promotes the degradation of nuclear Notch-1 and further leads to decreased vascular endothelial growth factor level.66 Copper metabolism MURR1 domain 1 (COMMD1) protein, an inhibitor of HIF-1, is actively exported to the cytoplasm by CRM1 under low oxygen concentrations.111 Disruption of the NESs or CRM1 inhibition results in nuclear accumulation of COMMD1, enhancing the repression of transcriptional activity of HIF-1 by COMMD1.111
Activating invasion and metastasis
The transcription factor Snail has important roles in epithelial–mesenchymal transition, tumor invasion and metastasis.112 CRM1 inhibition leads to nuclear accumulation of FBXL5 (F-Box and leucine-rich repeat protein 5), which is a negative regulator of Snail.113 Silencing CRM1 or Snail results in nuclear accumulation of FBXL5 and inhibition of epithelial–mesenchymal transition.113 Similarly, APC (adenomatous polyposis coli) protein, a negative regulator of nuclear β-catenin, is mislocalized to the cytoplasm by CRM1 in cancer cells, resulting in uncontrolled β-catenin transactivation of metastasis-related proteins.40,114 Further, cytoplasmic promyelocytic leukemia (cPML) promotes a mesenchymal phenotype and increases the invasiveness of prostate cancer cells through transforming growth factor-β signaling.115 cPML nuclear export is mediated by CRM1, co-expression of which with cPML correlates with reduced disease-specific survival in patients.115
Deregulating cellular energetics
Cancer cells usually display upregulated energetic metabolism to adapt to their high rate of proliferation.75 The ribosome is an effective cancer drug target because ribosome inhibition limits cellular energetics by affecting global protein synthesis.116,117 CRM1-mediated nuclear export is essential for nuclear export of pre-mature ribosome subunits and inhibition of CRM1 causes immature 40S and 60S ribosome production.118,
The importance of inflammation in tumor development has been increasingly recognized.124 Cox-2 and NF-κB are the key cellular mediators of inflammation that are often upregulated in cancer cells.125,126 It is shown that CRM1 inhibitor downregulates Cox-2 level by limiting its messenger RNA export.72 Treatment of ovarian cancer cells with a CRM1 inhibitor revealed a reduction in COX-2 expression and concomitant reduction of cell proliferation and increased apoptosis.47 NF-κB inhibitor IκBα is also a cargo of CRM1.127 IκBα is rapidly locked in the nucleus by CRM1 inhibition and forms a transcriptional inactive complex with NF-κB.128,129
Although it is impossible to summarize all proteins involved in nuclear export and cancer, the above examples clearly illustrate the strong link between CRM1 inhibition and reversion of cancer hallmarks. Many mechanism-based cancer drugs only target one particular aspect of cancer. For instance, epidermal growth factor receptor inhibitors reduce cancer proliferation, CDK inhibitors stop cell cycle and BH3 mimetics promote cell death.130,
Different classes of nuclear export inhibitors
Over the past two decades, plenty of nuclear export inhibitors (NEIs) have been discovered or developed, tested against diseases such as cancer, virus infection and neuronal degeneration.2,41,134 By their origins, these inhibitors can be classified into four groups as follows: bacterial products, herbal ingredients, fungal or animal NEIs, and synthetic NEIs (Figure 3). (1) Bacterial NEIs include leptomycin A/B, ratjadone A/C and anguinomycin A/B/C/D, which all have a long polyketide chain with a lactone ring.136,
It should be noted that all these inhibitors form a covalent bond with Cys528 of human CRM1 protein, through a Michael addition reaction. Cys528 lies in the NES-binding groove of CRM1 (Figures 4a and b). Thus, these inhibitors directly inhibit NES binding to CRM1 (Figures 4c and d). Mutation of Cys528 disables the anti-tumor action of these compounds.48,62,70,113,151,152,155 Next, we will focus on two classes of the most characterized NEIs: LMB and SINEs.
First-generation NEI: LMB
The first and most well-known inhibitor of CRM1 is LMB.156 LMB is a natural product produced by bacteria Streptomyces.157 It was first identified as an anti-fungal agent.158 Later on, it is discovered that LMB potently kills cancer cells.159 Scientists had proved that CRM1 is the cellular target of LMB as early as 1998; however, the LMB–CRM1 complex structure was not solved until recently.160,161
LMB is a long polyketide molecule with an α,β-unsaturated lactone warhead (Figure 3). In complex with CRM1, its long polyketide chain aligns with the hydrophobic NES-binding groove, forming extensive hydrophobic interactions (Figure 4c). The α,β-unsaturated lactone warhead links to a Cys528 equivalent residue of yeast CRM1 at α-position (Figure 4e). Unexpectedly, the lactone ring of LMB is hydrolyzed, forming an opened up structure of hydroxy acid (Figure 4e). The same hydrolysis is also observed in two other CRM1 inhibitors (anguinomycin A and ratjadone A) from bacteria.161 After hydrolysis, LMB forms extra interactions with CRM1, gaining one salt bridge and one hydrogen bond with CRM1.161 The hydrolysis is caused by three basic residues (in CRM1) adjacent to the LMB lactone ring, mutation of which results in CRM1 mutant that does not hydrolyze LMB’s lactone ring.161
LMB lactone hydrolysis leads to an unexpected finding that the covalent bond between LMB and CRM1 is reversible under certain condition. Using dialysis and pull-down assay, it is shown that LMB with lactone ring de-conjugates from CRM1 (slowly), whereas the hydrolyzed LMB does not.161 Thus, hydrolysis prevents de-conjugation of inhibitor and increases inhibitor potency. A permanent CRM1 inhibitor could provide Streptomyces bacteria more space to grow by killing fungi more efficiently, as CRM1-mediated nuclear export pathway is essential for fungi.1 More importantly, this finding could also explain the clinically strong cytotoxicity of LMB. LMB entered clinical trial for the treatment of cancers in 1996.136 However, strong dosage-limiting toxicities produced barely any clinical benefits and the mechanism of toxicity was unknown at the time.136 Indeed, recent studies also show that LMB-treated cells have permanently blocked nuclear export, which is lethal not only for cancer cells, but also for normal cells.1,48,155
Second-generation NEI: SINE
Although the early clinical failure of LMB spelled doom for CRM1-targeted drug development, it did not stop the endeavor of academic and pharmaceutical researchers, who discovered NEIs with improved pharmacological properties through library screens, of which PKF050-638 was found as an anti-viral hit.150 Its N-azolylacrylate scaffold is later on adopted by SINEs.2
Like LMB, SINEs also form covalent bond with Cys528 of CRM1. However, SINEs differ from LMB in several aspects (Table 1).162 First, SINEs are much more compact when compared with LMB. Whereas LMB occupies almost the entire NES groove, smaller SINEs occupies <50% of the space (Figure 4d). Second, in contrast to LMB, KPT-185 (one of the SINEs) is not hydrolyzed after binding to CRM1 and does not form salt bridge with CRM1 (Figure 4f).56,163,164 As such, its covalent bond to CRM1 is found to be slowly reversible.161 Third, SINEs but not LMB treatment induces CRM1 degradation, followed by CRM1 re-synthesis after drug removal.59,165 Therefore, through bond reversibility and re-synthesis, nuclear export inhibition by SINE treatment is not permanent, but rather transient. It is found that nuclear export in KPT-treated mouse at 10 mg kg−1 reversed to ~20% after 1 day and 50% after 3 days.155 These observations altogether could explain the fact that SINE is significantly less toxic than LMB observed in clinical trials.
Reported SINEs include KPT-185, KPT-249, KPT-251, KPT-276, KPT-330 and KPT-335. Their half-maximal effective concentration for cancer cells lies between 10 nm to 1 μm, and 5–20 μm for non-neoplastic cells.133 KPT-185 is the most studied compound, with limited bioavailability in murine and monkey pharmacokinetic studies.95 KPT-276 has been shown to block inflammation and nerve cell damage in mouse models of inflammatory demyelination.164 KPT-335 has received a Minor Use/Minor Species designation from the Center for Veterinary Medicine of the Food and Drug Administration for the treatment of lymphomas in canines.
Of all the SINEs, KPT-330 (Selinexor) is the most promising compound and is undergoing numerous human hematologic and solid tumor clinical trials. Before KPT-330’s clinical trial, many expected that inhibiting nuclear export will generate profound side effects, as CRM1 is essential for viability and maintaining the proper localizations of its target cargoes required for normal functions of the cell.1,166,167 However, side effects of SINE observed are much milder and controllable, including nausea, vomiting, anorexia, diarrhea, fatigue, weight loss and hepatotoxicity, by no means comparable to chemotherapy treatment.133,168,169 The efficacy and toxicity of SINE compounds were reviewed elsewhere recently.6,63,69,170
Future perspective: non-covalent NEIs
To our best knowledge, all known NEIs are covalent CRM1 inhibitors and rely on the conserved Cys528 residue on CRM1 to exert its therapeutic effects. Mutation of Cys528 renders current CRM1 inhibitors inactive.62,113,155,171,172 Interestingly, Cys528 mutation has been found in a huge number of fungal CRM1, which provides resistance to the anti-fungal agent LMB.17 Here we would like to propose a different class of NEI: non-covalent NEI. Non-covalent NEIs can be used in various cancers like their covalent counterparts by exerting transient inhibition, yet being insensitive to Cys528 mutation. In addition, they may possess intrinsically lower toxicity and higher efficacy.
The majority of drugs in clinical use are non-covalent in nature.173 Generally, covalent inhibitors are more prone to non-specific conjugation and causing undesirable off-target effects.174,175 All the current NEIs have a Michael acceptor, which is mild reactive and could bind to various targets.148,161,176 Side effects of SINE are mainly gastrointestinal and liver related.168,177,178 As SINE compounds are taken orally, its higher gastrointestinal concentration may result in more local off-target covalent binding, possibly accounting for the observed side effects.
In addition, SINE resistance can occur in tumor cells expressing very high level of CRM1, and in cells that tumor suppressor proteins are relocalized to cytoplasm after SINE treatment.155,178,179 With reduced side effects, non-covalent NEI could be used at higher dosage to achieve tighter transient inhibition, to increase its anti-tumor activity and to improve response rate in patients.
Cancer is not a single disease, as many different pathways are activated in different cancer cells. Nuclear export factor CRM1 exports/regulates many tumor suppressors and oncoproteins. Notably, CRM1 inhibition can attenuate many cancer hallmarks simultaneously, likely explaining the broad-spectrum anti-cancer potencies observed. The first-generation NEI LMB failed in phase I clinical trial due to high cytotoxicity. Second-generation inhibitors display much reduced cytotoxicity, owing to its reversible inhibition of nuclear export. It should be emphasized that ‘broad-spectrum’ anti-cancer does not imply ‘all-spectrum’, in other words—effective against all cancers. Although KPT-330 clinical trials have been very encouraging, non-covalent NEIs remain an interesting alternative for its possibly low-toxicity and broad-spectrum action as a CRM1-targeted anti-cancer therapy.
We thank Drs Yuh Min Chook (UTSW) and Chong Chen (SKLB) for comments. QS is supported by National Natural Science Foundation of China (NSFC), fund number 80502629. EB is supported by NIH R01-DK073639 and CPRIT grants. DJ is a ‘Junior One Thousand Talents’ program scholar. We apologize for those relavent works overlooked by this review.