Background
Melanoma is the most aggressive type of skin cancer. While the advent of targeted therapies (i.e.,
BRAF and
MEK inhibitors) and immune checkpoint inhibitors (i.e., ipilimumab, pembrolizumab, nivolumab) significantly improved recurrence and survival outcomes for melanoma patients, metastatic melanoma and therapy resistance still represent major challenges for clinicians [
1]. Currently, there is no reliable prognostic marker for melanoma.
p53, also known as the “guardian of the genome”, is a critical tumour suppressor that plays a role in cell cycle regulation, DNA repair, senescence, and apoptosis, through transactivation of target genes and protein interactions with key components of cellular pathways [
2]. As such,
TP53 has been found to be mutated in around half of all human cancers. However, its mutation rate in melanoma is much lower than the average, sitting below 20% (cBioPortal database for cancer genomics [
3]). With the importance p53 plays in cancer suppression and maintenance of genome integrity [
2], it is likely that its pathway activity is deregulated even in the absence of mutations. In fact, previous research from our laboratory has shown that p53’s transcriptional activity is deregulated in metastatic melanoma samples and cell lines regardless of p53 mutation status and that p53 may in fact contribute to proliferation of tumour cells, rather than contributing to their senescence and apoptosis [
4]. Other groups have also confirmed that wild-type p53 fails to act as a tumour suppressor in melanoma [
5,
6]. Several hypotheses that may explain impaired p53 pathway activity in p53 wild-type melanoma have been proposed, including mutations of
CDKN2A (encoding p14
ARF, which inhibits p53 degradation by HDM2) and overexpression of HDM2 (human double minute 2) or anti-apoptotic proteins (iASPP and BCL-2) [
7]. Yet, the induction of wild-type p53 in melanoma in response to genotoxic stress suggests regulatory mechanisms upstream of p53 remain intact at least in some cases [
5,
8]. Further, intrinsic apoptotic mechanisms have been found to be operational in melanoma [
9], negating the hypothesis for overexpression of anti-apoptotic proteins as the driving mechanism for p53’s inability to function as a tumour suppressor in melanoma. An alternative hypothesis is that the aberrant regulation of p53 pathway activity in melanoma may be driven by the differential expression of p53 isoforms. However, this has not been extensively studied.
p53 may be expressed as full-length p53 (p53α from herein) or as 12 shorter isoforms bearing a combination of N- (Δ40p53, Δ133p53, Δ160p53) and C-terminal (β, γ, Ψ) truncations. These isoforms may be generated through alternative splicing (Δ40p53, p53β, p53γ), alternative promoter usage (Δ133p53, Δ160p53), alternative initiation of translation (Δ40p53, Δ160p53), or post-translational degradation of p53 via the 20S proteasome (Δ40p53) [
10]. The isoforms have been found to be aberrantly expressed in various cancers, including breast cancer [
11,
12], squamous cell carcinoma of the head and neck [
13], and neuroblastoma [
14] among others. p53 isoforms are known to both enhance and inhibit p53’s canonical function in a cell- and context-specific manner [
14‐
17], highlighting their potential to contribute to dysregulated p53 pathway activity in melanoma.
Previous data from our laboratory highlighted that p53β and Δ40p53 were expressed at both the mRNA and protein level in melanoma cell lines and primary melanoma cultures, while undetectable or expressed at low levels in normal melanocytes and fibroblasts [
17]. Contrastingly, in metastatic melanoma Δ40p53β mRNA expression was found to be lower in tumour compared to normal adjacent tissue, while Δ133p53α and Δ160p53α protein expression was greater compared to normal adjacent tissue. However, gene expression findings did not always correspond with protein findings and vice versa [
18,
19]. In melanoma cell lines, exogenous p53β was found to enhance p53 target gene expression in response to chemotherapeutic treatment with cisplatin, while Δ40p53 was found to impair the upregulation of the same p53 target genes following cisplatin treatment [
17]. However, Takahashi and colleagues reported that exogenous Δ40p53 enhanced apoptosis in cancerous and noncancerous cells, though this was in the absence of chemotherapeutic agents [
20] and supports previous data from our laboratory, showing that at the basal level Δ40p53 acts similarly to p53α and functions as a tumour suppressor [
21]. Δ160p53 isoforms have also been found to promote proliferation and possibly migration when transfected into melanoma cell lines [
19]. Further, endogenous p53 isoforms have been linked to treatment resistance in melanoma cell lines, with BRAF-inhibitor resistant cell lines harbouring increased expression of Δ40p53β and decreased expression of TAp53β [
19].
Thus, there is evidence of p53 isoform deregulation in melanoma cell lines and metastatic melanoma [
17,
18], and in vitro studies suggest that p53 isoforms may affect melanoma aggressiveness and treatment response [
17,
20]. Altogether this indicates that p53 isoforms may harbour biomarker potential in melanoma. However, studies of p53 isoforms in primary melanoma specimens are still lacking and this represents a crucial next step to using p53 isoforms expression in a clinical setting. Melanomas are frequently preserved through formalin fixation and paraffin embedding, which impedes the assessment of p53 isoforms at the mRNA level due to nucleic acid degradation [
22]. In the present study we investigated the expression of p53 isoforms by immunohistochemistry (IHC), using a suite of C- and N-terminal p53 isoform-specific antibodies in a retrospective melanoma cohort and evaluated their prognostic biomarker potential.
Discussion
Here we demonstrated that p53 isoforms detected through IHC in FFPE melanoma samples using a suite of C- and N-terminal p53 isoform-specific antibodies (Fig.
1) harbour biomarker potential. KJC40 staining (Δ40p53) correlated with less aggressive melanoma but worse prognosis and DO-1 (TAp53) and KJC8 (p53β) staining correlated with more advanced and aggressive melanoma and worse prognosis (Table
1, Figs.
2 and
3). Composite biomarkers comprised of the expression of multiple p53 isoforms revealed a potentially complex interplay between the different isoforms and their relevance to prognosis. Such composite biomarkers highlight the need to consider p53 isoforms, not in isolation but as a connected network of redundant, synergistic, and antagonistic players [
27,
43,
44]. In this context, elevated cytoplasmic DO-1 (TAp53) and KJC8 (p53β) staining in the presence of low KJC40 and KJC133 staining was associated with the worst survival outcomes (Fig.
3, Table
2).
Contrary, to our breast cancer study [
12], we identified staining ranging from weak to strong across all isoforms (Fig.
1B). All isoforms were found to be more highly detected in the nucleus (Fig.
1), yet nuclear and cytoplasmic staining generally correlated (Fig.
1D). This is consistent with our previous findings in breast cancer [
12] and with the isoforms exerting control over the nuclear transcriptional activity of p53 [
15]. P53β isoforms did not correlate with any of the assessed N-terminal variants (TAp53, Δ40p53, or Δ133p53) in the nucleus (Additional file
1: Fig. S2A). Observed differences and similarities between our IHC study in breast cancer [
12] and the findings described herein may be related to differences in
TP53 mutation status or other genetic variants within
TP53 that regulate p53 isoform expression such as mutation in the internal ribosome entry sites, introns, or in splice sites [
45‐
48]. Additionally, with different tissue types known to express varying levels of p53 isoforms [
15], the different cellular origins (melanocytes and epithelial cells) may also contribute to the different p53 isoform expression patterns of melanoma and breast cancer. Finally, other factors such as pathogenic infections, the expression of transcription and splicing factors involved in isoform expression, and the stress context of cells may determine p53 isoform expression patterns in different tissues [
34,
49‐
53].
With KJC8 staining being the predominant nuclear staining in melanoma (Fig.
1A, B), questions arise around their potential regulation of the p53 transcription factor by p53β. Herein, we have shown that elevated nuclear KJC8 staining is associated with worse probability of survival (Fig.
2E), potentially indicating a role for the p53β isoforms in driving tumour progression, which has also been reported by others [
35]. Yet increased KJC8/DO-1 ratios were associated with reduced Breslow thickness (Table
1), reduced probability of metastasis (Fig.
2B), and a better probability of survival (Fig.
2H), suggesting that it is not the p53β isoforms in isolation that are important for correlation with clinicopathological features, but their expression relative to other isoforms [
27]. p53β has been previously found to enhance p53 transcriptional activity of key target genes involved in cell cycle regulation and apoptosis [
17,
34], which may indicate that elevated p53β may enhance p53’s function as a tumour suppressor even when p53 expression is low (high p53β:TAp53 expression).
Nonetheless, elevated KJC8 staining may also harbour oncogenic potential underlying its association with worse survival outcomes, particularly when DO-1 staining is also elevated (Fig.
3) and p53 is potentially mutated (Fig.
4C). We hypothesise that the p53β isoforms may contribute to melanoma progression by promoting dedifferentiation. A finding from our previous study showed that primary melanoma cultures with elevated p53β expression formed non-adherent spheres, while cells from the same patient with low p53β expression grew in an adherent monolayer [
17]. With sphere-forming capacity frequently used as a surrogate for a stemness phenotype [
54], p53β isoforms may be involved in cancer stem cell (CSC) regulation. CSCs have previously been linked to enhanced cancer recurrence and treatment evasion, driving poorer survival outcomes [
55], thus p53β may not only be a marker of worse prognosis, but may contribute to cancer aggressiveness by positively regulating dedifferentiation into CSCs. In support of this, p53β expression was associated with serous and poorly differentiated ovarian cancers and correlated with worse recurrence-free survival in patients with functional p53 [
47]. Further, a recent study highlighted that Δ133p53β was associated with an increased probability of melanoma recurrence as well as a reduction in the time for the primary tumour to metastasise to the brain [
35], highlighting worse prognosis linked to p53β isoforms that may guide the need to select more aggressive treatment approaches such as systemic treatments in addition to surgical resection to prolong survival [
56]. Given that p53 isoform function is cell- and context-specific [
27] and p53β isoforms with different N-terminal truncations are likely to have varying effects [
27], it is difficult to ascertain how exactly p53β isoforms contribute to worse survival and perhaps increased cancer stem cell potential. However, we do know that isoforms may interact, shorter p53 isoforms can oligomerise with p53α and modulate its capacity to transactivate target genes [
34,
53]. The exact effect p53β has likely depends on the cellular with previous studies having shown that in the absence of treatments, p53β isoforms may inhibit cell growth and induce senescence in normal human fibroblasts, T cells, and MCF-7 breast cancer cells [
34,
57,
58], it may drive proliferation following treatment with TG003 (a CDK inhibitor) [
34]. We hypothesise that p53β in the presence of DNA-damaging therapies, such as chemotherapeutic agents, impairs the transcriptional activity of p53α, driving the cancer cell towards survival and perhaps dedifferentiation and thus contributing to worse patient survival outcomes following treatment. Future studies should investigate whether systemic treatment of melanoma expressing high p53β is specifically associated with this worse survival and aim to determine the molecular mechanisms underpinning this potential observation.
DO-1 staining was the predominant stain in the cytoplasm of melanomas (Fig.
1A and C) and greater staining of cytoplasmic DO-1 was associated with reduced metastasis-free survival and reduced melanoma-specific survival (Fig.
2A and K). Elevated nuclear DO-1 staining was also associated with worse survival outcomes (Fig.
2D). Similarly, in a cohort of 140 benign and malignant melanocytic lesions, p53 expression, detected via IHC with the DO-7 antibody which detects a similar epitope to DO-1, was found to increase from benign nevi to malignant melanomas [
59]. Further, p53 mutation was found to correlate with greater p53 expression [
59], as found in this study (Fig.
4C). In these cases, p53 may have lost its function or acquired a gain-of-function mutation, enabling the mutant protein to drive cancer progression as opposed to suppressing it [
60].
KJC40 staining was found to correlate with less advanced melanoma (Table
1), yet also with worse survival outcomes (Fig.
2). These findings are in line with Δ40p53 being a two-faced player in cancer [
21,
31,
33]. Previous findings from our lab, have indicated that at the basal level, i.e., in the absence of chemotherapeutic agents, Δ40p53 isoforms may act similarly to p53, suppressing oncogenic traits [
21]. This would support expression of these isoforms positively correlating with less advanced stages of cancer (Table
1). However, when considering the survival curves, patients are likely to have undergone various treatments for their advanced melanoma prior to succumbing to the disease. In these instances, elevated expression of Δ40p53 isoforms may have adversely affected treatment sensitivity, driving melanoma survival and growth. We have observed this in breast cancer models, where elevated Δ40p53 expression impaired p53’s canonical response to doxorubicin, promoting cancer survival, DNA repair and proliferation, while inhibiting apoptotic signalling [
31]. Similarly, Δ40p53 was found to impair the upregulation of p53 target genes following cisplatin treatment of melanoma models [
17]. In melanoma cell lines Δ40p53β has also been found to be elevated in vemurafenib (
BRAF inhibitor) resistant cell lines [
61], highlighting that Δ40p53 may serve as a potential biomarker to select more individualised and appropriate treatments for melanoma patients. Other studies have also reported dual roles for Δ40p53, which can act independently to p53, and both enhance or inhibit p53’s canonical functions (reviewed in [
62]).
Further nuances in how isoforms may work in concert could be derived from composite biomarker analysis, where DO-1
bright KJC8
bright KJC40
dim/− KJC133
dim/− (HHLL) showed the worst survival outcomes. The complex interplay between different isoforms may also provide a potential explanation for conflicting findings between various studies hoping to uncover the function of individual p53 isoforms (reviewed in [
10]). In this context, our findings in breast cancer have shown that the upregulation of Δ40p53 led to increased stability and levels of other isoforms due to heterotetramer formation and decreased proteasomal degradation [
31,
33]. Thus, the composite biomarker evaluation in tissues may entail the state of activation and interaction of the isoforms. While our composite biomarker analysis lacks statistical power it is an important first step regarding p53 isoforms as an interconnected system.
Pending further validation, the isoforms may support oncologist-patient decision-making in selecting a more aggressive treatment regime (for melanomas with elevated TAp53 and p53β for example) or support the rejection of chemotherapy or
BRAF inhibition for melanomas with elevated Δ40p53, where the isoform may drive unfavourable treatment responses [
31,
33,
61]. However, in the absence of an in-depth understanding of how the different p53 isoforms contribute to potentially more aggressive cancer (TAp53 and p53β) and a lack of therapies that can directly target the various p53 isoforms, their role in melanoma needs to be considered conservatively and future studies should aim to establish the isoforms’ contribution to pathophysiology through in vitro knockout and overexpression studies, and the use of animal models. Only with such studies, would it be possible to identify the specific functions of the p53 isoforms and compensatory mechanisms among these isoforms, which may contribute to another isoform’s expression. While such in-depth molecular studies are ongoing, prognostic biomarkers, such as the putative ones identified within this manuscript still offer value to patients and oncologists, who can make informed life decisions about their prognosis.
There are several limitations of the study, including the retrospective nature of the cohort, the use of FFPE tissues of varying quality and the limited sample size. Future studies should aim to validate these findings in larger, prospective cohorts and consider the use of tissue microarrays for higher throughput or fresh melanoma and control samples, where immunofluorescence may be used as an alternative technique. Additionally, the results need to be considered in the context of possible selection bias given that the annotation areas were selected by researchers, though every effort was made to select annotations representative of the whole melanoma, to minimise such bias (see Additional file
5 for examples of selected annotations across a range of melanomas). Moreover, in the current study, we were only able to look at C- and N- terminal truncations in isolation. Given that p53 isoforms harbour C- and N- terminal truncations simultaneously [
27] and, for example, different N-terminal variants of the p53β isoform are likely to have different biological functions [
27], it is very hard to interpret the pathophysiological contribution of all p53β isoforms together. Future studies should aim to characterise isoforms by looking at both amino acid terminals together, though the lack of specific antibodies has limited such studies to western blotting. The fact that the p53 isoforms likely work in concert with each other further complicates the interpretation of findings and may be the main driver for so many conflicting findings about the role of different p53 isoforms in the literature [
27,
63].