ASXL1 promotes adrenocortical carcinoma and is associated with chemoresistance to EDP regimen

Introduction

Adrenocortical carcinoma (ACC) originated from adrenal cortex cells, as a rare neoplasm with annual incidence of 0.7~2 per million [1], is highly aggressive and is prone to metastasize. Lack of heralding symptoms often render patients with ACC diagnosed at a late stage. Mitotane, as the only FDA approved specific drug for ACC, together with surgery, chemo- and radio-therapy forms a multi-modality treatment scheme for ACC at present. However, for stage IV patients catalogued by traditional European Network for the Study of Adrenal Tumors (ENSAT) classified method, their 5-years survival could merely reach 6% - 13% [2]. In general, current treatment for stage III-IV ACC is disappointing.

Current clinical practice guidelines on the management of adrenocortical carcinoma was issued by European Society of Endocrinology (ESE) in 2018. In terms of drug treatment, based on the results of the FIRM-ACT trial, the guidelines recommend the utilization of mitotane +EDP (etoposide, Adriamycin, cisplatin combination chemotherapy) as the most effective remedy for advanced ACC patients who are not amenable to local treatment [3].

Recent high throughput sequencing studies revealed some genomic mutations remarks that be identified as “drive events” of ACC tumorigenesis including mutations in IGF-2, TP53, CTNNB1, MEN1 and PRKAR1A. Somatic mutations and fusions of these genes mostly relate to continuous activation of WNT/β-catenin pathway which cause the overexpression of corresponding growth factors and inactivation of tumor suppressor TP53. In the meantime, there is a dearth of studies on the impacts of copy number variation (CNV) in ACC. Assie, Fernandez, Lerario et al. reported partial or complete deletion of chromosome 1p, 17p, 22p, 22q, 2q and 11q by comparative genomics via the microsatellite method loss of heterozygosity and allele imbalance in ACC [4–6]. However, insightful mechanistic analyses are still lacking.

The Cancer Genome Atlas (TCGA) project of ACC demonstrated unprecedented genomic and genetic panorama of the rare cancer which substantially promoted researches on ACC from bench to bedside [7]. However, many copy number variance (CNV) in ACC revealed in TCGA still lacked insightful functional investigation. In the current study, we set off to evaluate role of 20q gain in ACC and identified Additional sex combs like transcriptional regulator 1 (ASXL1) as target gene that could mediate chemoresistance with the aim of deepening the understanding of biology of ACC.

Materials and Methods

Results

CNV of ASXL1 was first queried across cancer types and ACC was amongst top cancers that harbored ASXL1 gain (Figure 1A). We then focused on TCGA ACC cohort and found gain-of-function was the predominant alteration in all 3 ASXLs (Figure 1A). Of note, gain of ASXL1 occurred in ~57% of ACC cases (Figure 1B). We further plotted mRNA expression against copy number in all 3 ACXLs and found only ASXL1 showed increased mRNA expression along with increased copy number, indicating functional output of the gene (Figure 1B). Overexpression of AXSL1 conferred both worsened overall and progression-free survival in ACC patients (Figure 1C). Enrichment analysis showed CTNNB1 mutation, one of the established driver events in ACC, was significantly enriched in ASXL1-gained cases (Figure 1D, 1E). Gene expression enrichment analysis also showed significantly enriched drug metabolism pathway (Figure 1F) and WNT signaling (Figure 1G).

We then pursued drug metabolism function of ASXL1 gain. By querying cytoband that encompassed ASXL1 that was amplified in GDSC dataset, we showed that ASXL1 gain was associated with significant resistance to a variety of compounds in both GDSC1 and GDSC2 pools (Figure 2A). Amongst the compounds were three chemo-agents (EDP) that consisted of the first line chemotherapy of ACC (Figure 2B). ASXL1 functioned in chromatin remodeling and regulates the expression of many genes. To identify target gene(s) of ASXL1 we performed analysis of the ChIP-Atlas and found FSCN1 ranking top 2 in binding affinity to ASXL1 (Figure 2C). Of note, FSCN1 was reported to play important pro-tumorigenic role in ACC [10]. Reproduction of TCGA ACC cohort showed significant linear positive correlation between expressions of ASXL1 and FSCN1 (Figure 2D). Interestingly, this correlation did not exist in normal adrenal tissue (Figure 2E). Analysis of clinicopathological parameters in TCGA ACC cohort showed ASXL1 expression was significantly higher in tumors with incomplete removal, with marginal significance in tumors of higher stage and progressed disease upon adjuvant treatment (Supplementary Table 1). In our IHC cohort, we validated linear correlation of expression of ASXL1 and FSCN1 using IHC scores (r = 0.823, P <0.001) (Figure 2F) and showed AXSL1 expression being significantly higher in tumors with advanced staging and in nodal positive cases with marginal significance (Table 1). We then used an alternative strategy to validate ASXL1/FSCN1 axis by plotting drug sensitivity profile against expressions of ASXL1 and of FSCN1 in the CTRP platform that incorporated multiple datasets. We showed austocystin D being the top hit in both queries (Figure 2G).

We then evaluated effect of ASXL1 in vitro in 2 ACC cell lines. Silencing of ASXL1 significantly inhibited proliferation in both cell lines (Figure 3A). ASXL1-KD also significantly inhibited colony formation (Figure 3B). Silencing of ASXL1 significantly lowered population in G1phase and increased population in S phase in both cell lines (Figure 3C). ASXL1-KD also led to increased early and late apoptosis in NCI-H295R cell but showed no effect on apoptosis in SW-13 cells (Figure 3D). For cell motility, we showed that ASXL1-KD significantly decreased cell invasion (Figure 3E) and wound healing (Figure 3F) in both cell lines. In the rescue assays, we first found that transcriptional activity being significantly increased in both cell lines with ASXL1-OE (Figure 4A). FSCN1 silencing restored promotion of proliferation and cell migration by ASXL1-OE in both ACC cell lines (Figure 4B, 4C). However, FSCN1 silencing did not alter cell cycle profile and apoptosis interfered by ASXL1-OE in either cell line (Figure 4D, 4E). We then examined drug sensitivity profile of EDP. We found both ASXL1 and FSCN1 silencing decreased IC50 of doxorubicin in both ACC cell lines (Figure 4F). Nonetheless, only ASXL1-KD but not FSCN1-KD could cold lower IC50 of cisplatin (Figure 4G). Similarly, FSCN1-KD showed minimal effect on sensitivity of etoposide whereas ASXL1-KD increased sensitivity (Figure 4H).

We next evaluated role of ASXL1 in vivo. Both silencing of ASXL1 and EDP-M regimen significantly inhibited tumor growth of NCI-H295R cells, whereas addition of ASXL-KD further inhibited tumor growth (Figure 5A). Survival analysis showed significantly prolonged survival in group with ASXL1-KD and EDP-M (Figure 5B). Interestingly, whereas tumors in SW-13 xenograft models grew at a similar rate to NCI-H295R, SW-13 tumors showed trend towards acquired insensitivity upon prolonged treatment starting on day 39 (Figure 5C). Combination of EDP-M and ASXL-KD also showed similar trend although the combination group showed most potent inhibition (Figure 5C, 5D). As ASXL3 was an important paralog of ASXL1 [11] and ASXL3 was constitutively overexpressed in SW-13 cells, we evaluated ASXL3 expression in both cells and found EDP could only induced increased ASXL3 expression in SW-13 cells (Figure 5E). IHC in harvested tumors also showed decreased ASXL1 expression in ASXL1-KD group and strong ASXL3 expression in SW-13 cells (Figure 5F).

Discussion

Stage IV ACC confers a 5-year overall survival of 0% and novel treatment modality is at urgent need [12]. Our study highlights both the protumorigenic role and chemoresistance of ASXL1 gain in ACC. Normal ASXL1 plays a role in embryonic development. ASXL1 mutations are observed primarily in myelodysplastic syndromes, but they are also observed in colorectal and endometrial cancer [13, 14]. ASXL1 is altered in 4.42% of all cancers with colon adenocarcinoma, lung adenocarcinoma, breast invasive ductal carcinoma, acute myeloid leukemia, and myelodysplastic syndromes. The most common alterations in ASXL1 is mutation (2.62%) followed by amplification (0.67%). As a driver event in several cancers, ASXL1 status has been recognized as an inclusion criteria in cancer trials targeting hematological malignancies such as acute myeloid leukemia, myelodysplastic syndromes, acute myeloid leukemia arising from previous myelodysplastic syndrome, chronic myelomonocytic leukemia, and secondary acute myeloid leukemia [15].

Our study has been the first to identify the role of ASXL1 in ACC. Though ASXL1 is solely amplified in 1 case in ACC, CN gain occurred in over half of cases and the increased mRNA expression following CN gain is indicative of gain of function of the gene. This is also consolidated by its prognostic value at mRNA level. Of note, TCGA cohort shows ASXL1 is associated with incomplete resection and our cohort indicates ASXL1 is associated with advanced stage. Both findings suggest that ASXL1 gain confers growth advantage to treatment-naïve ACC.

Being the first and only phase III randomized controlled trial, the FIRM-ACT trial provided by far the most solid evidence for chemotherapy regimen. The EDP-M regimen is now recognized as the first line therapy for advanced ACC [16]. Immune checkpoint inhibitors (ICIs) that reinforced last defense in many solid tumors only show moderate effect in ACC and are currently used under salvage and trial settings [17]. In fact, most trials trying to transplant agents effective in other cancers into ACC yield frustrating outcomes [18]. The nature of multi-resistance and rare disease renders drug development stumbling over decades.

Genetic landscape of ACC revealed by TGCA triggers various translational studies on ACC and our findings in the current study also takes advantage of the CNV data of TCGA. We for the first time report association between EDP sensitivity and CN of a certain gene. Gain of function in both ASXL1 and FSCN1 plays critical role in mediating resistance. Interestingly, most positive findings in NCI-H295R cells has also been observed in SW-13 cells that is now considered to originate from small cell carcinoma of adrenal gland. Furthermore, overexpression ASXL3 in SW-13 cells suggests that crosstalk may exist in ASXL family to facilitate resistance. Nonetheless, we find that gain of function of ASXL3 occurs in only 4% of ACC cases, further supporting different cell context of SW-13 and NCI-H295R. The interplay between FSCN1 and ASXL1 has not been reported before. However, overactive FSCN1 has been reported to be pro-tumorigenic in a variety of cancers [19–21]. Detailed mechanism of how FSCN1 mediate ASXL1 expression in the presence of ASXL1 copy number gain, another driver incentive, would be interesting.

Our study also has limitations. Although ASXL1 appears to be a promising target in ACC, we failed to test compound(s) that target ASXL1. Using gene-compound sensitivity dataset, we have identified austocystin D as a potential agent. However, the compound is not commercially available. Austocystin D is an organic heteropentacyclic compound isolated from Aspergillus and Aspergillus ustus and has been identified as a potent cytotoxic agent with in vivo antitumor activity and selectivity for cells expressing the multidrug resistance transporter MDR1, which was a characteristic of ACC as well [22–24]. Selective cytotoxic action of austocystin D arises from its selective activation by cytochrome P450 (CYP) enzymes in specific cancer cell lines, leading to induction of DNA damage in cells and in vitro [25]. The pattern of cytotoxicity of austocystin D was distinct from doxorubicin and etoposide putting austocystin D fairly promising in overcoming chemoresistance. Of note, as mitotane is a potent inducer of CYP3A4 activity which may have reduced the blood levels of doxorubicin and etoposide, both of which being metabolized by CYP3A4 [26, 27], monotherapy of austocystin D may in theory overcome this drawback. Production of austocystin D is now in progress by our group according to reported protocol and we look forward to testing the compound in ACC. Recent studies report that ACC cells are highly susceptible to ferroptosis, potentiating novel drug combination in this aggressive disease [28–30].

To sum up, we have shown that ASXL1 gain is common in ACC. Gain of function of ASXL1 confers worsened prognosis and promotes tumor growth of ACC. ASXL1 overexpression also induces chemoresistance to EDP regimen. Targeting ASXL1 hold promise to combat this rare but aggressive disease.

Supplementary Materials

Supplementary Table 1