Increased Expression of the Δ133p53β Isoform Enhances Brain Metastasis

The Δ133p53β isoform is increased in many primary tumors and has many tumor-promoting properties that contribute to increased proliferation, migration and inflammation. Here we investigated whether Δ133p53β contributed to some of the most aggressive tumors that had metastasized to the brain. Δ133p53β mRNA expression was measured in lung, breast, melanoma, colorectal metastases and, where available, the matched primary tumor. The presence of Δ133p53β expression was associated with the time for the primary tumor to metastasize and overall survival once the tumor was detected in the brain. Δ133p53β was present in over 50% of lung, breast, melanoma and colorectal metastases to the brain. It was also increased in the brain metastases compared with the matched primary tumor. Brain metastases with Δ133p53β expressed were associated with a reduced time for the primary tumor to metastasize to the brain compared with tumors with no Δ133p53β expression. In-vitro-based analyses in Δ133p53β-expressing cells showed increased cancer-promoting proteins on the cell surface and increased downstream p-AKT and p-MAPK signaling. Δ133p53β-expressing cells also invaded more readily across a mock blood–brain barrier. Together these data suggested that Δ133p53β contributes to brain metastases by making cells more likely to invade the brain.


Introduction
Wild-type p53 is well known for its role in tumor suppression; however, TP53 encodes at least 12 isoforms and some of these promote tumor progression. The ∆133p53 isoforms (∆133p53α, ∆133p53β and ∆133p53γ) have been associated with cancer, with the ∆133p53β isoform being the most frequently increased in cancer, including brain, prostate, breast and colorectal tumors [1][2][3][4][5]. ∆133p53β-expressing tumors resist treatment, invade, recur and contain high numbers of tumor-supporting cells, such as tumor-associated macrophages (TAMs) and those that prevent cytotoxic T-cell anti-tumor responses [3,6]. Analysis of ∆133p53β was largely restricted to primary tumors. Considering cancer mortality is largely associated with metastases, with brain tumor metastases being a significant cause of mortality for 20-40% of cancer patients [7], ∆133p53β may be more prominent in more aggressive tumors, such as those that have metastasized to the brain. In an analysis of 18 brain tumor metastases from the breast, lung and colon, ∆133p53β protein aggregates were present in 72% of tumors, suggesting that ∆133p53β is present in the majority of brain metastases [8,9]. Here we investigated the role of ∆133p53 isoforms in a larger cohort of 140 brain metastases from tumor types most likely to go to the brain, including tumors from the breast, lung, colon and skin. We report that ∆133p53β was a key feature of brain metastases from all tumor types. ∆133p53β had properties that may make the tumors more likely to enter the brain, providing a possible explanation for why it was associated with a reduced time for primary tumors to spread to the brain.

∆133p53β Was Present in Many Brain Tumor Metastases
Breast, colorectal, lung and melanoma brain metastases (n = 140) were included in the study ( Table 1). All tumors were subjected to RNAscope using a probe toward ∆133p53β mRNA ( Figure 1 and Supplementary Figure S1) [2,3]. All tumor sections showed strong staining for ubiquitin C (UBC), except in areas with necrosis, indicating that each tumor had good RNA preservation (Figure 1a). Tumors were classified as having no (0%), low (2-10%), moderate (11-25%) or high (>25%) ∆133p53β mRNA expression based on the percentage of positive cells (Figure 1b). All tumors with low, moderate or high ∆133p53β were considered positive for ∆133p53β expression, provided the DapB negative control showed no staining, and all tumors with no ∆133p53β staining were considered negative for ∆133p53β mRNA expression.

∆133p53β Was Increased in Metastases Compared with the Primary Tumor
The matched primary tumor was available for 24 brain tumor metastases. Using RNAscope, 11 (46%) primary tumors expressed ∆133p53β mRNA. Of the primary tumors with no ∆133p53β mRNA expression, 10 showed ∆133p53β mRNA expression in the metastases, suggesting that ∆133p53β mRNA was increased in more aggressive tumors. Many primary tumors did not show ∆133p53β -positive staining across the entire tumor; instead, regions of positive staining were present. The percentage of positive cells were lower in primary tumors compared with the matching brain metastases (mean percentage of positive cells: 2.6 in primary tumors versus 17.8 for brain metastases, p < 0.0001, Figure 1e and Supplementary Figure S2).

∆133p53β Was Associated with a Reduced Time to Metastasize to the Brain
The finding of increased ∆133p53β in the metastasis compared with the primary tumor was consistent with a role for ∆133p53β in more aggressive tumors. To investigate this further, a positive or negative ∆133p53β status in the brain metastasis was compared with the time for the primary tumor to metastasize to the brain and the overall patient survival once the tumor was diagnosed in the brain. A significant reduction in the time for the primary tumor to metastasize to the brain was found for breast (61 versus 23 months, p = 0.0028; ratio 2.652, 95% CI 1.91-5.904), melanoma (48 versus 14.5 months, p = 0.0114; ratio 3.319, 95% CI 1.715-6.388) and lung (16 versus 7 months, p = 0.0466; ratio 2.286, 95% CI 1.226-4.261) tumors that were positive for ∆133p53β compared with negative tumors (Figure 2a). No significant difference in time was found for positive colorectal tumors (46 versus 23 months, p = 0.559; ratio 1.186, 95% CI 0.4999-2.816) compared with ∆133p53βnegative tumors). No difference in overall patient survival from when the tumor was diagnosed in the brain was found between ∆133p53β-positive and -negative tumors across all tumor types (p > 0.05, Figure 2b). These findings suggested that increased ∆133p53β in brain metastasis identified tumors that spread faster to the brain, but once the tumor was in the brain, ∆133p53β did not affect overall patient survival.
We next investigated whether there was an association between ∆133p53β expression and mutant p53, as mutant p53 promotes metastasis [10]. Immunohistochemistry was used as a surrogate for identifying tumors with mutant p53 based on mutations in p53 increasing the stability of the p53 protein. Strong widespread expression of p53 consistent with mutant p53 was assigned to 56% of breast metastases (n = 15/27), 86% of colorectal metastases (n = 18/21), 39% of lung metastases (n = 18/46) and 41% of melanoma metastases (n = 19/46). All tumors designated as mutant for p53 showed strong nuclear staining in almost 100% of the tumor cells (Figure 3a). There was no correlation between the presence of mutant p53 and ∆133p53β using RNAscope (Figure 3b). Mutant p53 in the brain metastases was not associated with reduced time for the primary tumor to metastasize to the brain or overall patient survival once the tumor was detected in the brain (Figure 3c). when the tumor was diagnosed in the brain was found between Δ133p53β-positive and negative tumors across all tumor types (p > 0.05, Figure 2b). These findings suggested tha increased Δ133p53β in brain metastasis identified tumors that spread faster to the brain but once the tumor was in the brain, Δ133p53β did not affect overall patient survival. Overall survival time once the tumor was diagnosed in the brain. Δ133p53 was measured using RNAscope in breast, melanoma, lung and colorectal brain metastases and th patient outcomes were compared between those with Δ133p53β-expressing (high, medium, or low tumors and those with no Δ133p53β expression. * p < 0.05; ** p < 0.01; ns, not significant. We next investigated whether there was an association between Δ133p53β expression and mutant p53, as mutant p53 promotes metastasis [10]. Immunohistochemistry wa used as a surrogate for identifying tumors with mutant p53 based on mutations in p53 Overall survival time once the tumor was diagnosed in the brain. ∆133p53β was measured using RNAscope in breast, melanoma, lung and colorectal brain metastases and the patient outcomes were compared between those with ∆133p53β-expressing (high, medium, or low) tumors and those with no ∆133p53β expression. * p < 0.05; ** p < 0.01; ns, not significant. metastases (n = 18/21), 39% of lung metastases (n = 18/46) and 41% of melanoma metast (n = 19/46). All tumors designated as mutant for p53 showed strong nuclear stainin almost 100% of the tumor cells (Figure 3a). There was no correlation between the prese of mutant p53 and Δ133p53β using RNAscope (Figure 3b). Mutant p53 in the b metastases was not associated with reduced time for the primary tumor to metastasiz the brain or overall patient survival once the tumor was detected in the brain ( Figure   Figure 3. Correlation between mutant p53 and Δ133p53β expression in brain metastases. (a) Mu p53 was estimated as intense staining in over 80% of tumor cells using immunohistochemistry Across brain metastases, mutant p53 was not correlated with Δ133p53β mRNA expression f RNAscope analyses. (c) Mutant p53 was not associated with altered patient progression time fo primary tumor to metastasize to the brain or with overall survival when the tumor was diagn in the brain across different tumor types.

RNA Sequencing Showed That Δ133p53β Was Associated with Altered Gene Expression
To provide more insight into how Δ133p53β contributes to metastasis, R sequencing (RNAseq) was performed on five clonal cell lines that stably expres Δ133p53α (Δ133p53α1, Δ133p53α2 and Δ133p53α8) or Δ133p53β (Δ133p53β5 Δ133p53β10), which were cell lines created by adding individual Δ133p53 isoforms to p53 null cell line H1299 [2]. H1299 vector control cells were included as a control (Fig  4). A principal component analysis (PCA) and unsupervised hierarchical cluste analysis revealed the p53 null cell line clustered distinctly away from the Δ133p53α Δ133p53β cell lines. Differences between the Δ133p53α and Δ133p53β cell lines w subtle, though they were more evident in the PCA compared with the clustering anal (Figure 4a,b). No biological processes were enriched in differentially expressed ge using an FDR cutoff of <0.05. (b) Across brain metastases, mutant p53 was not correlated with ∆133p53β mRNA expression from RNAscope analyses. (c) Mutant p53 was not associated with altered patient progression time for the primary tumor to metastasize to the brain or with overall survival when the tumor was diagnosed in the brain across different tumor types.

RNA Sequencing Showed That ∆133p53β Was Associated with Altered Gene Expression
To provide more insight into how ∆133p53β contributes to metastasis, RNA sequencing (RNAseq) was performed on five clonal cell lines that stably expressed ∆133p53α (∆133p53α1, ∆133p53α2 and ∆133p53α8) or ∆133p53β (∆133p53β5 and ∆133p53β10), which were cell lines created by adding individual ∆133p53 isoforms to the p53 null cell line H1299 [2]. H1299 vector control cells were included as a control ( Figure 4). A principal component analysis (PCA) and unsupervised hierarchical clustering analysis revealed the p53 null cell line clustered distinctly away from the ∆133p53α and ∆133p53β cell lines. Differences between the ∆133p53α and ∆133p53β cell lines were subtle, though they were more evident in the PCA compared with the clustering analysis (Figure 4a,b). No biological processes were enriched in differentially expressed genes using an FDR cutoff of <0.05.
An analysis of individual transcripts showed that consistent transcripts were increased or decreased in ∆133p53β and ∆133p53α cells ( Figure 5). SOX9, NDGR1, RIP4, RTL6, LDHAL6FP, ADRPH and SPATC1 were decreased in all ∆133p53α and ∆133p53β cell lines and MGMT was decreased in ∆133p53β cell lines. ANAX3 and CTP32 were increased in all ∆133p53α and ∆133p53β cell lines, with STAT6, CALHM5, NR0B1, TPM2, CRACR2A, GDNF and ZNF22 increased in ∆133p53β cell lines and ERC2 increased in ∆133p53α cell lines.  Transcripts that encoded the tumor suppressors NDGR1 and RIP4 were decreased and transcripts that encoded the tumor-promoting proteins GDNF, ZNF22 and ANXA3 were increased in ∆133p53 cells, particularly ∆133p53β, which is consistent with a role for ∆133p53β in promoting cancer; however, some results were inconsistent with a role for ∆133p53β in promoting cancer, with increased expression of the tumor suppressor TPM2 and decreased expression of SOX9, which was described as having both tumor suppressor and tumor-promoting properties [11][12][13][14][15][16][17][18].
The RNAseq analysis showed subtle changes in gene expression with the introduction of ∆133p53α or ∆133p53β. To determine whether growth-promoting signaling was increased overall in ∆133p53 cells, immunofluorescence for p-MAPK, p-AKT and major downstream effector pathways in oncogenic signaling was performed. ∆133p53 cells showed marked increases in p-MAPK-and p-AKT-positive cells (Figure 6a,b).
These results suggested that despite subtle gene expression differences in ∆133p53 cells, the oncogenic signaling pathways were activated.
After confirming that p-MAPK and p-AKT signaling were increased in ∆133p53α and ∆133p53β cells, we next investigated whether ∆133p53 cells had functions that were consistent with promoting cancer, including brain metastasis. Using an in vitro blood-brain barrier containing human microvascular endothelial cells (HBMECs) on the apical surface and astrocytes on the basolateral surface of a transwell insert [19], ∆133p53β cell lines and one ∆133p53α cell line showed greater invasion across the barrier compared with control cells (Figure 6c).
These findings suggested that increased ∆133p53 makes tumors more invasive and more likely to enter the brain.  Transcripts that encoded the tumor suppressors NDGR1 and RIP4 were decreased and transcripts that encoded the tumor-promoting proteins GDNF, ZNF22 and ANXA3 were increased in Δ133p53 cells, particularly Δ133p53β, which is consistent with a role for Δ133p53β in promoting cancer; however, some results were inconsistent with a role for Δ133p53β in promoting cancer, with increased expression of the tumor suppressor TPM2 and decreased expression of SOX9, which was described as having both tumor suppressor and tumor-promoting properties [11][12][13][14][15][16][17][18].
The RNAseq analysis showed subtle changes in gene expression with the introduction of Δ133p53α or Δ133p53β. To determine whether growth-promoting signaling was increased overall in Δ133p53 cells, immunofluorescence for p-MAPK, p-AKT and major downstream effector pathways in oncogenic signaling was performed. Δ133p53 cells showed marked increases in p-MAPK-and p-AKT-positive cells (Figure 6a,b).
These results suggested that despite subtle gene expression differences in Δ133p53 cells, the oncogenic signaling pathways were activated.
After confirming that p-MAPK and p-AKT signaling were increased in Δ133p53α and Δ133p53β cells, we next investigated whether Δ133p53 cells had functions that were consistent with promoting cancer, including brain metastasis. Using an in vitro bloodbrain barrier containing human microvascular endothelial cells (HBMECs) on the apical surface and astrocytes on the basolateral surface of a transwell insert [19], Δ133p53β cell

∆133p53β Had Increased Cancer-Promoting Proteins on the Cell Surface
The increased p-AKT and p-MAPK signaling in ∆133p53 cells suggested that ∆133p53 may be associated with increased growth signaling from the cell surface. To determine whether cancer-promoting proteins were increased on the ∆133p53 cell surface, protooncogenes or proteins induced by proto-oncogenes in lung cancer, including c-MET, VEGFR, ITGα5, EGFR, TFRC and ALK, were measured on the cell surface of ∆133p53α and ∆133p53β cell lines using flow cytometry (Figure 7a-c). All proteins were increased on ∆133p53β cells, and c-MET, ITGα5 and TFRC were also increased on ∆133p53α cells. Increased TFRC in ∆133p53α and ∆133p53β cells was also confirmed using immunofluorescence (Figure 7a). With the exception of EGFR, the increased expression of the corresponding genes did not explain the increase in cell surface proteins, as was evident using TaqMan qPCR assays (Figure 7d).
Whether ∆133p53 isoforms could increase cancer-promoting proteins on the surface was tested further using B16 murine melanoma cell lines expressing wild-type p53 or ∆122p53 (a mimic of ∆133p53α function) [20]. Flow cytometry revealed increased c-Met and TRFC in ∆122p53 cells compared with the control (Figure 7e).
These findings suggested that some cancer-promoting properties of ∆133p53β and ∆133p53α occur by increasing growth-promoting receptors on the cell surface.
lines and one Δ133p53α cell line showed greater invasion across the barrier compared with control cells (Figure 6c).
These findings suggested that increased Δ133p53 makes tumors more invasive and more likely to enter the brain.

Increased Cell Surface ∆133p53β Protein Phenotype Could Be Used to Target ∆133p53β Cells
The decreased expression of NDGR1 and increased TFRC on the cell surface were consistent with the role of ∆133p53β in promoting cancer (Figure 8a). Since both TFRC and NRDG1 also have opposing functions in iron regulation and increased TFRC can reduce cell viability upon ferroptosis induction with the GPX inhibitor (1S,3R)-RSL3 (RSL3) [21], we investigated whether the increased TFRC on the cell surface would make ∆133p53 cells more susceptible to iron-related cell death. Consistent with ∆133p53β having increased TFRC expression, ∆133p53β cell lines had increased intracellular total iron concentrations relative to the total protein (Figure 8b). Upon ferroptosis induction with RSL3, ∆133p53β cells showed reduced viability and increased cellular lipid peroxides (Figure 8c,d). These results suggested that ∆133p53β cells are more susceptible to ferroptosis and the increased proteins on the cell surface could be manipulated to target ∆133p53β cells.

Discussion
∆133p53β is present in many aggressive primary tumors and here we extended this to some of the most aggressive tumors that metastasize to the brain. ∆133p53 was present across tumor types and increased in metastasized tumors compared with primary tumors. ∆133p53β was associated with a reduced time for the primary tumor to metastasize to the brain. In addition, the introduction of ∆133p53β in a lung cell line increased migration across a transwell assay that mimicked aspects of the blood-brain barrier.
∆133p53β pro-migratory properties may make tumors more likely to enter the brain, as was evident with ∆133p53 tumors showing a reduced time to metastasize to the brain. In primary breast, colon, brain and prostate tumors ∆133p53 isoforms, with a predominance for the ∆133p53β isoform, are associated with poorer clinical outcomes, which is consistent with ∆133p53β being a characteristic of aggressive tumors [2,3,5,22]. In breast cancer, ∆133p53β increased the risk of recurrence and death and was the most significant independent predictor of outcome in tumors that would otherwise be predicted to have a good prognosis [1]. In prostate cancers, ∆133p53β alone predicted aggressive disease with 88% accuracy [2]. In breast, colon and prostate cancer, the association of ∆133p53β with increased risk of recurrence suggests ∆133p53β increases the likelihood of invasion from the primary site [2,3,5,22]. In in vitro corroboration studies, ∆133p53β and a murine mimic of ∆133p53 (∆122p53) increased the migration potential in cells with wild-type and mutant p53 and in those with otherwise poor invasive properties [1,5]. In an orthotopic transplantation model, ∆133p53 has three-fold higher expression in a highly metastatic cell line with reduced time to metastasis [23].
∆133p53β produces highly invasive cells by promoting the loss of adhesive structures and a rounded amoeboid morphology [1]. The findings from this study also found enhanced cancer-promoting functions with ∆133p53β compared with ∆133p53α. ∆133p53β clonal cell lines showed increased migration across a transwell containing endothelial and astrocytes to mimic the more restrictive blood-brain barrier, suggesting that ∆133p53β would enhance movement into the brain. In addition, many well-characterized tumorpromoting proteins, including those with roles in metastasis, were increased on the cell surface, including c-MET, EGFR, VEGFR, ALK and ITGα5.
The relatively subtle changes in gene expression compared with the proportion of cell surface proteins studied that were found to have increased on the ∆133p53β cell surface suggested that a larger pan-proteomic-based approach may identify more pathways that are associated with ∆133p53β function. Increased proteins on the cell surface without changes in the expression of the corresponding genes were reported for mutant p53. Mutant p53 was proposed to promote enhanced receptor recycling of c-MET and ITGα5 to the cell surface in mutant p53 cancers, leading to increased cell surface protein expression, downstream signaling and effector functions of enhanced scattering and invasion [24,25]. Other evidence supports a role for ∆133p53α and ∆133p53β in increasing cell surface proteins; GLUT1 was increased in ∆133p53α-expressing MCF7 cells; and in the murine model to investigate ∆133p53α function, ∆122p53-positive peripheral blood mononuclear cells showed increased alpha-enolase on the cell surface and the associated plasminogen activation [26,27].
∆133p53α and ∆133p53β isoforms may share with mutant p53 an ability to enhance the return of cancer-promoting proteins to the cell surface, suggesting oncogenic variants of p53 may promote metastasis by enhancing signaling from the cell surface. The markedly high p-AKT and p-MAPK staining in ∆133p53α and ∆133p53β cells are consistent with multiple cancer-promoting proteins being increased on the ∆133p53 cell surface. The altered cell surface protein phenotype could be manipulated as a means to target ∆133p53α and ∆133p53β cells, as demonstrated here with increased TFRC, which may have been aided by a corresponding decrease in NDGR1, leading to an increased susceptibility to induced ferroptosis.
Increased expression of cancer-promoting proteins on the ∆133p53α cell surface has not been universally found, with reduced ITGα5 found in MCF10A cells [28]. The overall p53 and p53 family environment may be important for determining the overall effects of ∆133p53α and ∆133p53β on the cell surface, with other studies finding the binding of mutant p53 or ∆133p53 to p63 or p53α responsible for the changes to individual cell surface proteins. Other p53 proteins were not present in H1299 expressing cells used here, but brain metastases did have multiple p53 moieties. Unlike primary brain tumors where ∆133p53 was almost mutually exclusive with mutant p53, many brain metastases harbor both mutant p53 and ∆133p53, suggesting that both may contribute to brain metastases. ∆133p53 was not the only TP53 isoform increased in brain metastases, with ∆40p53α also showing a higher expression. The cancer-promoting functions of ∆40p53 are less thoroughly characterized. However, in response to DNA damage, ∆40p53 inhibited apoptosis [29].
How ∆133p53 isoforms are elevated in cancers is unclear, but possibilities include hypoxia and infection [2,3,30,31]. ∆133p53 contributes to treatment resistance, and investigations of other p53 isoforms found increased expression in response to DNA-damaging cancer treatments, which suggests that ∆133p53 may also increase with treatment. Increased ∆133p53 with chemotherapy and downstream resistance to treatment offers an explanation for why ∆133p53 is elevated in brain metastases compared with the primary tumor [3,32,33].
The limitations of this study were those specific to studying brain tumors, where only tumors that were removed, and potentially not very aggressive inoperable tumors, were represented in the cohort. The survival analyses were not adjusted for confounding factors; these are numerous for patients with brain metastases, and thus, a much larger cohort size is required.
Overall, the results from this study indicate that ∆133p53β is increased in brain metastases. ∆133p53β may allow tumors to more readily enter the brain, reducing the time for primary tumors to metastasize to the brain.

Human Tumors
Brain tumor metastases from primary lung (n = 46), breast (n = 27), melanoma (n = 46) and colorectum (n = 21) tumors were selected from those collected at Dunedin and Hamilton hospitals. Following surgery, the tumors were either fixed in 10% neutral buffered formalin or snap-frozen in liquid nitrogen and stored at −80 • C. Fixed tissue was processed into paraffin wax and embedded. Formalin-fixed paraffin-embedded tumors were cut on a microtome into either 4 µm (immunohistochemistry) or 5 µm (RNAscope) tissue sections that were placed onto coated slides (Leica Bond, Leica Biosystems, Wetzlar, Germany). Paraffin-embedded formalin-fixed tissues were available for all tumors and additional frozen tissue was also available for 51 tumors. Matching primary tumors were available for 24 cases (7 breast, 6 colorectal, 7 lung and 3 melanoma). The clinicopathological data associated with all samples are outlined in Table 1. The inclusion criteria were brain tumor metastases with tumor tissue removed between 2010 and 2016 diagnosed as breast, lung, colorectal or lung metastases with only tissue from the first brain tumor metastasis analyzed in cases with recurrent brain tumors excised. Ethical approval (reference LRS/10/09/037 and MEC/08/02/061) was obtained in New Zealand and all procedures followed institutional guidelines. All individuals provided written informed consent.

RNAscope
Probes included a custom probe to the unique region of ∆133TP53 and TP53β made by Advanced Cell Diagnostics named ∆133TP53 and described elsewhere [2,3], as well as the probes ubiquitin C (UBC, positive control) and DapB (negative control, Advanced Cell Diagnostics, Newark, CA, USA). Formalin-fixed paraffin-embedded tumors or cell clots were cut into 5 µm sections. The RNAscope method used the manual assay method with Protease Plus reagent for protein digestion and the 2.5HD reagent kit brown according to the manufacturer's instructions. Following the addition of DAB to the ∆133TP53 and DapB assays, DAB enhancer was added (Leica Biosystems, Wetzlar, Germany). Positive cells and the number of dots per positive cell were identified using the Aperio Scancope CS digital pathology system and quantified using the Aperio RNA ISH Algorithm (Aperio, Vista, CA, USA). The slides were evaluated by two blinded examiners. The percentage of positive cells out of the total cell number was measured. Ten fields (×400 magnification) were chosen at random and the percentage of positive cells out of the total cells counted. Tumors with no positive cells in the 10 random fields were examined over the entire tissue section for the presence of positive cells.

RNAseq
Library generation (DNBSEQ Eukaryotic Strand-Specific mRNA library) and RNA sequencing (read length 100 base pair reads and the DNBseq platform) were performed by BGI Genomics in Hong Kong, China. Raw data was subject to quality control using Fast QC version 0.11.9. Adaptor sequences and low-quality sequences (Phred score < 30) were trimmed using Cutadapt version 2.6 [36]. Reads were aligned against the human reference genome (GRCh38) using HISAT2 version 2.1.0 and annotated with GRCh38.107.gtf from Ensembl [37,38]. Aligned reads were counted by exon and summarized by gene using featureCounts version v1.5.3 and the counts were normalized for differential gene expression using Wald Test from DESeq2 package (version 1.30.1) performed locally on R (version 4.0.2) and RStudio™ (version 1.3.1093) [39][40][41]. The p-values were adjusted for multiple testing using the Benjamini-Hochberg correction. Genes were considered statistically differentially expressed if the criteria of false discovery rate (FDR) < 0.05 and log 2 fold change of ±1.5 (FC ± 1.5) were satisfied (Supplementary File S1).

In Vitro Blood-Brain Barrier
An in vitro blood-brain barrier was established using a co-culture cell model that contained human primary astrocytes and human microvascular endothelial cells (HBMECs, ThermoFisher Scientific, Waltham, MA, USA). Cells were initially cultured using a specialized cell culture medium as per the manufacturer's instructions (HBMECs, Medium 131 with microvascular growth supplement and Gibco Astrocyte Medium for human astrocytes). Cells were maintained at 37 • C in 5% CO 2 and used before passage 6. Once confluent, the cells were harvested for the transwell assay using Corning Transwell-Col collagen-coated membrane inserts, 12 mm, 0.4 µm pore PTFE membrane inserts (Merck KGaA, Darmstadt, Germany). Astrocytes (3.5 × 10 5 cells) were seeded onto the basolateral transwell insert coated with poly-L-Lysine (Millipore, Sigma-Aldrich, St. Louis, MO, USA). Three to four days later, HBMECs (8 × 10 4 cells) were seeded onto the apical side of the transwell insert. The supplemented astrocyte media was placed in the basal chamber and the HBMEC media was placed in the apical chamber. Once the astrocytes and HBMECs reached 90% confluency, the transepithelial resistance (TEER) was measured to determine the integrity of the barrier using the Epithelial Volt/Ohm (TEER) meter (World Precision Instruments, Sarasota, FL, USA). Once the TEER was ≥45 Ω/cm 2 , the ∆133TP53 or control p53 null cells (1.25 × 10 4 cells) were added, and 24 h later, the insert was removed and the cells on the bottom of the 12-well plate were fixed with 4% paraformaldehyde and stained with 3% crystal violet. The plates were then imaged using an inverted research microscope and all cells were counted. Data are presented as the fold change in the total number of cells compared with the control cells. Results: mean ± s.d. from at least three replicates. * p < 0.05, ** p < 0.01, *** p < 0.001, and **** p < 0.0001.

Immunohistochemistry and Immunofluorescence
Four-micrometer sections from formalin-fixed paraffin-embedded tissues were used for IHC. For IHC, p53 staining used the D07 clone from Cell Marque (Rocklin, CA, USA) and the Bond RX system (Leica Biosystems, Wetzlar, Germany) with the Bond Refine Detection kit. Stained cells were identified using the Aperio Scancope CS digital pathology system (Leica Biosystems, Wetzlar, Germany). The slides were evaluated by two blinded examiners. A tumor was considered p53 mutant positive if 80% of the tumor was positive for nuclear p53 staining [42][43][44].
Unstained and stained cells were measured on a Gallios flow cytometer with Kaluza software version 2.2 (Beckman Coulter, Indianapolis, IN, USA). The percentage of positive cells was calculated using FLOWJO software version 10.6.2 (BD Biosciences, Franklin Lakes, NJ, USA). The percentage of positive cells was compared between ∆133p53 and control cells.

Total Iron Measurement
Measurement of total intracellular iron followed a published protocol with modifications [45]. The Iron Assay Kit (Sigma-Aldrich, St. Louis, MO, USA) and Pierce BCA Protein Assay Kit-Reducing Agent Compatible (ThermoFisher Scientific, Waltham, MA, USA) were used according to the manufacturer's instructions.

Cell Viability Measurement
A total of 10,000 cells per well were seeded in 96-well plates and treated with RASselective lethal 3 (RSL3, 0-5 µM, MedChemExpress, Monmouth Junction, NJ, USA) for 24 h. The cell viability was measured using the Cell Counting Kit 8 (WST-8) (Abcam, Cambridge, UK) and incubated with WST-8 solution for 2.5 h at 37 • C and the absorbance was measured at 460 nm according to the manufacturer's instructions.

Measurement of Lipid Peroxidases
Ferroptosis was induced using RSL3 (2 µM) for 24 h; the cells were then incubated with 200 nM C-11-BODIPY (ThermoFisher Scientific, Waltham, MA, USA) for 30 min at 37 • C and measured using flow cytometry.

Statistical Analyses
Results are presented as mean ± s.d. Statistical analyses of continuous variables were performed using Student's t-test or one-way ANOVA corrected for multiple comparisons. Spearman's rank correlation analysis was employed to evaluate correlations between the mRNA levels of pairwise genes. The statistical analyses of categorial variables used Fisher's exact test. Survival analyses were performed using the log-rank (Mantel-Cox) test and corrected for multiple comparisons using the Bonferroni correction. Statistical analyses were performed with Prism 9 software (Graphpad). The survival analysis involved either the time taken for the primary tumor to metastasize (taken from the date of surgery to remove the primary tumor and the date to remove the brain metastasis) or the overall survival once the tumor was in the brain (taken from the date of surgery to remove the brain metastasis and the date of death or last follow-up appointment).

Conclusions
∆133p53β was present in the majority of tumors that metastasized to the brain. ∆133p53 was increased in more aggressive tumors and those that more readily metastasized, suggesting ∆133p53 promoted aggressive tumors by upregulating cell surface signaling and entry into the brain, reducing the overall time to metastasis.

Institutional Review Board Statement:
The study was conducted in accordance with the Declaration of Helsinki and approved by New Zealand Health and Disability Ethics Committees (reference LRS/10/09/037 and MEC/08/02/061).

Informed Consent Statement:
Written informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Raw files for RNA sequences (FASTQ format) were deposited in the Gene Expression Omnibus with accession number GSE221282.