Recurrent XPO1 mutations alter pathogenesis of chronic lymphocytic leukemia

Chronic lymphocytic leukemia (CLL) is the most prevalent adult leukemia in Western societies, characterized by clonal expansion of mature CD5⁺/CD19⁺ expressing B lymphocytes with significant genetic and clinical heterogeneity [1]. The majority of CLL patients are diagnosed at an early stage with low tumor volume and moderate disease activity [2]. While a small percentage of CLL cases will never progress from this indolent disease phase, most patients will maintain a watchful waiting period until worsening cytopenias or disease symptoms emerge, upon which treatment begins. Recent advances in treatment of patients with CLL have dramatically improved patient outcomes, prompted by innovations in drugs targeting B-cell receptor (BCR) signaling pathways including Bruton’s tyrosine kinase (BTK) [3‐8] or phosphoinositide 3-kinases (PI3K) [9‐13], or disruption of the B-cell lymphoma 2 (BCL2) [14‐17] inhibitor of apoptosis, among others [18]. Each class of drug yields extended remissions with variable time periods of treatment ranging from intermittent to continuous. Sequential treatment with these agents is also effective. However, a subset of patients develop resistance to inhibitors of BCR pathways and BCL2 and have a relatively poor prognosis, spotlighting the need for further investigation of novel therapies [19]. Preventing this from occurring in high risk patients is also important for which new therapies are needed. Moreover, approximately 2–10% of CLL cases will progress to develop a highly aggressive lymphoma morphologically resembling a diffuse large B cell lymphoma (DLBCL) termed Richter’s Transformation (RT). Patients who develop RT have a poor prognosis with short survival due to the lack of effective therapies [20, 21].

Investigating mechanisms behind the genetic and clinical heterogeneity of CLL, genomic studies have identified a number of recurrent somatic mutations affecting several critical regulatory proteins, including, among others, those affecting the nuclear export protein Exportin-1 (XPO1; also known as Chromosome Region Maintenance 1, CRM1) [22‐24]. Those affecting XPO1 are primarily observed as single point mutations to a residue within the NES-binding groove (most frequently E571K or E571G), altering the charge and structural basis of the amino acid in this critical location [25]. These recurrent hot-spot point mutations have been reported in ~ 5% of CLL cases and between ~ 25 and 35% of CLL cases progressing to RT, suggesting either a pathogenic role for this genetic aberration or association with other high risk features [24, 26‐28].

Nuclear vs cytoplasmic compartmentalization is a defining trait of eukaryotic cells, and remains critical for proper cell maintenance and function. Selective permeability of the nuclear envelope and the nuclear pore complex (NPC) allows for passive diffusion of small ions and metabolites while facilitating active transport of large macromolecules (> 40 kDa) with energy-dependent nuclear transporters. The karyopherin-β superfamily of importins and exportins mediates the transport of large macromolecules containing either nuclear localization signals (NLSs) or nuclear export signals (NESs), respectively [29]. The major nuclear export receptor, XPO1, exports hundreds of functionally diverse proteins, binding to highly diverse 10–15 amino acid NES motifs in cargos that contain 4–5 hydrophobic residues that interact with the hydrophobic XPO1 NES-binding groove (Fig. 1a) [30‐33]. XPO1 is expressed in all human cells and has been cited as a core essential gene, where loss of XPO1 is a lethal event due to the ablation of proper nuclear and cytosolic organization [34, 35].

While maintaining proper homeostatic balance in eukaryotic cells, XPO1 sustains relevance to cancer through several proposed mechanisms; most widely cited is its ability to shuttle tumor-suppressor proteins out of the nucleus (e.g. p53, IkB, BRCA1) thereby preventing their anti-apoptotic function [36]. In addition, XPO1 is a transporter for > 1000 putative protein cargos [37‐39], influences the localization and metabolism of oncoprotein-encoding mRNAs and various RNA species (miRNA, rRNA, snRNA, tRNA) [40], and plays an essential role in regulating mitosis and chromosome structure [41], suggesting many contributing oncogenic mechanisms may involve XPO1. Moreover, alterations to NESs of recurrently mutated genes in cancer have been shown to drive mis-localization of the NES-containing protein and enhance downstream tumorigenic effects [42]. In this regard, XPO1 is frequently overexpressed and/or present with recurrent mutations in many cancer types. Surprisingly, these abnormalities have remarkable specificity for hematologic malignancies including Hodgkin’s lymphoma (HL), DLBCL, primary mediastinal B cell lymphoma (PMBCL), and CLL [24, 43]. Notably, the XPO1 gene is located on chromosome 2p15, suggesting XPO1 overexpression may be a critical leukemogenic mechanism in the subset of CLL patients with gain of the short arm of chromosome 2 (+ 2p) [44].

Due to the over-expression of XPO1 in many cancer cells and its role in exporting tumor suppressor genes such as TP53, development of selective inhibitors of nuclear export (SINEs) that inhibit the nuclear export activity of XPO1 in malignant cells has garnered increased attention [43, 45]. The SINE molecules selinexor (KPT-330/XPOVIO™) [46] or Eltanexor (KPT-8602) [47] have each demonstrated anti-leukemia activity in pre-clinical studies and have advanced to clinical trials in patients with CLL and other types of cancer. Moreover, Selinexor has received accelerated approval from the US Food and Drug Administration (FDA) for treatment of patients with relapsed and/or refractory (R/R) multiple myeloma or R/R DLBCL [48, 49].

A previous report described studies indicating that germline insertion of the E571K XPO1 mutation could accelerate development of lymphoma in mice made to express high-level c-Myc and BCL-2, suggesting such XPO1 mutations could contribute to oncogenic transformation [24]. Additionally, we previously showed that the E571K mutation affects the binding of a small subset of NESs that have highly charged β-strands located close enough to make strong interactions with residue 571 [50], suggesting a select group of cargos are affected by this mutation. One of these cargos is the RNA regulator eIF4E-Transporter (4E-T), which forms a weaker association with E571K XPO1, resulting in inappropriate 4E-T localization in cells carrying the mutation. However, these previous studies employed mouse models used to study lymphomagenesis or models primarily for general in vitro tumor cell investigation. Although the E571K mutation in XPO1 has been observed in CLL, the role they play driving leukemogenesis remains unclear.

Herein, we present a retrospective analysis of CLL cases in which we have identified a subset of patients harboring an E571 XPO1 mutation. This analysis reveals the E571K/E571G XPO1 mutation associated with high-risk genetic and epigenetic co-factors and at the time of diagnosis. To elucidate whether such mutations in XPO1 could play a causal role in CLL leukemogenesis, we developed Eµ-XPO1 transgenic mice harboring human XPO1 under control of µ heavy chain enhancer (Eµ) elements driving constitutive expression of XPO1. We found that overexpression of XPO1 (wt, E571K or E571G) was sufficient to initiate development of a CLL-like disease in a subset of aged mice, but lacked the ability to drive a spontaneous leukemic expansion in young mice. However, when crossed with the established Eµ-TCL1 CLL-mouse model (Eµ-XPO1/TCL1), concurrent presence of E571K or E571G XPO1 mutations with overexpression of the TCL1 oncogene sufficiently enhanced the rate in which these mice developed their CLL-like phenotype. Lastly, crystal structures of XPO1 (wt or E571K) bound to several SINE molecules (KPT-185, KPT-330/Selinexor, and KPT-8602/Eltanexor) show the compounds binding similarly in each case, which was further confirmed by a similar cytotoxicity profile from this class of compounds in patients’ cells with or without an XPO1 mutation. This suggests XPO1 inhibitors are likely to remain effective pharmaceutic candidates for cancer patients with E571 XPO1 mutations.

Wide scale genomic analysis across a range of cancer types have discovered the recurrent E571 XPO1 mutations to be particularly enriched in hematologic malignancies. In CLL, presence of an E571 XPO1 mutation also suggests association with other high-risk CLL prognostic indicators, but to date, a complete understanding of their impact on the genesis and progression of these hematologic malignancies remains unresolved. We have herein provided an extensive evaluation of the leukemogenic potential of E571 XPO1 mutations in both human and murine CLL. Thanks to collaboration between a diverse set of clinical CLL research groups, we retrospectively analyzed a total of 1286 CLL cases, identifying 72 of which having presented with an E571 XPO1 mutation at their time of diagnosis. The overall distribution of XPO1-mutated patients in this cohort and their association with other high-risk CLL co-factors falls in line with previous reports for this mutation, providing further evidence for the E571 XPO1 mutation as a cancer-associated genetic abnormality. The high VAF seen among all patients suggests XPO1 is likely part of the early initiating event of CLL and is most likely not acquired as a later sub-clone with treatment. Collectively, this suggests a dominant role for XPO1 in CLL leukemia initiation. Despite these associations, the results of our study also confirm that the E571 XPO1 mutation does not impact survival outcomes in these patients, measured either in surrogate by TTFT or from time from diagnosis to death.

With regards to cytogenetic evaluation, poor survival outcomes paired with a significant association with del(13q14) and an abrupt dissociation with trisomy 12 (+ 12) in XPO1-mutated CLL patients offer potential pathways to explain the consequences of these mutations. Deletion of the q14 region of chromosome 13 in CLL cases is often met with a favorable prognosis, citing the loss of miR-15/16, DLEU7, or RB1 as major contributors to disease pathogenesis [81‐83]. Roughly 70% of evaluable XPO1-mutated patients presented del(13q14) as a co-occurring genetic abnormality, a significant increase from the ~ 50% reported in the general CLL population and the ~ 42–45% reported in CLL cells with wt XPO1. Despite a historically fair prognosis, the survival outcomes of these patients were among the shortest reported in this study, suggesting some combination of dysfunctional molecular signaling stemming from these genetic aberrations are acutely advantageous for proliferation of the CLL cells. Moreover, the alternative activation of survival pathways in XPO1-mutated CLL cells suggests the XPO1 mutation may be skewing the molecular dependence of specific survival pathways for continued proliferation of neoplastic B cells and encourages the emergence of additional factors to contribute to leukemogenesis. Enrichment in factors such as CD2, CD28, GZMB, and INFG also suggests the disrupted signaling in XPO1-E571K CLL cells may influence surrounding immune cells, contributing to the immune dysfunction attributed to CLL patients. The role for complications stemming from E571 XPO1 mutations in modulating the immune system of CLL patients, or increasing susceptibility to opportunistic infections [84], remain unresolved. Further genetic evaluation will be necessary to completely describe these mechanisms.

Of the most striking results from cytogenetic evaluation in these patients was the exclusivity of the E571 XPO1 mutation and the gain in copy number for chromosome 12. This suggests a divergent biology of trisomy 12 and XPO1-initiated CLL. Copy number gains in entire chromosomes remain particularly challenging events to study in cancer pathogenesis due to the immensity of information stored in each chromosome. Regarding chromosome 12, however, a recent report suggests hyperactive NFAT signaling in + 12 CLL cases may act as a driving mechanism in these neoplastic cells [85]. Interestingly, RNA-sequencing of CLL cells with E571K XPO1 mutations in our study revealed an upregulation of genes converging on NFAT and NFAT-related signaling pathways. This relationship may reveal a scenario in which convergence of the disrupted molecular pathways from both genetic aberrations is a lethal event in CLL cells. Thus, development of an E571 XPO1 mutation would push towards an aggressive CLL phenotype with worse overall CLL prognosis, and conversely, development of a trisomy 12 leading to an intermediate CLL course.

It is important to note that the CLL patient data described in our study were collected from tertiary care centers and are likely selecting for advanced disease cases that are referred to these institutions. This remains a limitation to most large-scale investigations determining trends in cancer data, highlighting the need for widespread access to affordable, high-throughput evaluation of tumor samples in primary care centers.

Previous reports have noted that E571 XPO1 mutations are exclusively found as heterozygous missense mutations, never appearing in the homozygous or hemizygous state [24, 86]. This finding reinforces the notion that XPO1 contributes critical homeostatic functions to the eukaryotic cell. As CLL is most often defined as a clonal expansion of malignant B lymphocytes, identifying a median E571 VAF in XPO1-mutated CLL patients to be 0.47 suggests the E571 XPO1 mutation is in the dominant clone of the total disease burden. This finding proposes the E571 XPO1 mutation as an early transformative event in progression of CLL, from which these clones progress to populate the majority of the leukemic tumor burden. This same explanation, however, implies the E571 XPO1 mutation in B lymphocytes alone is not sufficient to spur development of an overt leukemia, as very few CLL patients presented with a low VAF at the time of diagnosis. This idea is reaffirmed by data generated from our novel Eμ-XPO1 transgenic mouse model, where E571K-XPO1 and E571G-XPO1 overexpressing mice displayed the same rates of disease development as mice overexpressing the WT-XPO1 transgene. However, splenic populations from Eμ-XPO1^E571K and Eμ-XPO1^E571G mice, but not Eμ-XPO1^WT mice, were largely comprised of clonally related B cells, suggesting these mutations are advantageous for proliferation and survival, and would be primed for leukemic transformation with the addition of another genetic abnormality.

When paired with another CLL-driving oncogene, like has been recently shown by Taylor and colleagues in mouse models combining the E571K XPO1 mutation with c-Myc or Bcl-2 alterations [24], recurrent mutations to this site stimulate a more aggressive disease phenotype. In our study, crossing Eμ-XPO1 mice with the Eμ-TCL1 mouse, which is the most common murine model of CLL, Eμ-XPO1xTCL1 mice expressing E571K- or E571G-XPO1 displayed a more rapid onset of their CLL-like disease. This accelerated disease phenotype was not produced in Eμ-XPO1xTCL1 mice expressing WT-XPO1. Interestingly, however, no significant impact on survival was observed between each Eμ-XPO1xTCL1 transgenic cross and Eμ-TCL1 mice. Moreover, the histologic disease phenotype in all Eμ-XPO1xTCL1 transgenic crosses retained similarity to that observed in the Eμ-TCL1 model, suggesting the overall phenotype resulting from this double transgenic cross is dominated by the TCL1 oncogene. Observations with this Eμ-XPO1xTCL1 mouse fall in line with survival data collected from human CLL patients, suggesting that once leukemogenic transformation is complete, other high risk factors may provide a more significant contribution to the overall mortality of this disease.

As the presence of an E571 XPO1 mutation in CLL almost exclusively coincides with a negative overall prognosis, development of an effective therapeutic regimen is of critical importance. Recently, small molecule inhibitors of XPO1 termed SINEs have gained increasing attention as anti-cancer molecules, with specific emphasis toward hematologic malignancies. The covalent binding site of all SINE compounds, C528, resides in the XPO1 NES-binding groove within proximity of the E571 locus, the latter of which potentially contributing to the biochemical environment uniquely conducive for SINE binding. However, by solving crystal structures of SINE-bound XPO1, we determined the E571K mutation is unlikely to impact the inhibitory potential of this class of drugs. These data support the use of SINE molecules in cancer patients possessing an E571 XPO1 mutation without suspicion of alternative drug-target interactions altering the therapeutic potential. Our data using a human tumor cell model and CLL patient cells with heterozygous XPO1-E571K expression support this claim, as we observed no change in proliferation potential between wt and E571K mutant cells upon treatment with these molecules. In both mutant models, interestingly, a modest increase in sensitivity to SINE inhibition was observed at select doses, suggesting manipulation of molecular signaling pathways stemming from the E571 XPO1 mutation may make these cells more susceptible to this form of anti-cancer treatment. Further characterization of these interactions will be necessary to fully describe the altered molecular pathways stemming from E571 XPO1 mutations that may influence the sensitivity and cytotoxic efficacy of tumor cells to inhibition by SINE molecules.

The similar disease course in human and murine CLL observed in our study suggests the mislocalization of intracellular machinery stemming from E571-XPO1 mutations may be priming pre-neoplastic B cells for complete leukemogenic transformation with the addition of oncogenic molecular defects converging on similar critical regulatory or signaling pathways. However, a complete understanding of the leukemogenic consequences stemming from the E571 XPO1 mutation are not yet fully understood and will require further investigation to provide a complete description of this unique genetic abnormality.