ADD3 Deletion in Glioblastoma Predicts Disease Status and Survival

Loss of heterozygosity (LOH) on chromosome 10 frequently occurs in gliomas. Whereas genetic loci with allelic deletion often implicate tumor suppressor genes, a putative tumor suppressor Adducin3 (ADD3) mapped to chromosome 10q25.2 was found to be preferentially downregulated in high-grade gliomas compared with low-grade lesions. In this study, we unveil how the assessment of ADD3 deletion provides clinical significance in glioblastoma (GBM). By deletion mapping, we assessed the frequency of LOH in forty-three glioma specimens using five microsatellite markers spanning chromosome 10q23-10qter. Data were validated in The Cancer Genome Atlas (TCGA) cohort with 203 GBM patients. We found that allelic loss in both D10S173 (ADD3/MXI1 locus) and D10S1137 (MGMT locus) were positively associated with tumor grading and proliferative index (MIB-1). However, LOH events at only the ADD3/MXI1 locus provided prognostic significance with a marked reduction in patient survival and appeared to have diagnostic potential in differentiating high-grade gliomas from low-grade ones. Furthermore, we showed progressive loss of ADD3 in six out of seven patient-paired gliomas with malignant progression, as well as in recurrent GBMs. These findings suggest the significance of ADD3/MXI1 locus as a promising marker that can be used to refine the LOH10q assessment. Data further suggest the role of ADD3 as a novel tumor suppressor, whereby the loss of ADD3 is indicative of a progressive disease that may at least partially account for rapid disease progression in GBM. This study revealed for the first time the downregulation of ADD3 on the genetic level resulting from copy number deletion.


INTRODUCTION
Glioblastoma (GBM), the most malignant primary brain tumor in adults, is often characterized by high levels of genomic instability. Loss of heterozygosity (LOH) of chromosome 10q is among the most frequent genetic alterations that occur in over 80% of de novo cases and 60%-70% of secondary cases, but it is less frequent in anaplastic astrocytoma (40%) and rare in low-grade gliomas (1)(2)(3)(4). Other frequent abnormalities in primary GBM include EGFR amplification (36%), p16 INK4a deletion (31%), TP53 mutation (28%), and PTEN mutation (25%) (5,6). The assessment of LOH10q genotype is not included in the recent WHO classification for gliomas (WHO 2016), as it is believed that the assessment of both chromosome 7+/10q− and TERTp mutations is perhaps more informative than assessing EGFR amplification for the diagnosis of GBM (7). Moreover, most studies to date have identified LOH of 10q as a poor prognostic marker for high-grade gliomas (8)(9)(10). Allelic deletion on chromosome 10q has also been observed in various advanced human malignancies (11)(12)(13)(14). This suggests that such allelic regions might harbor tumor suppressor genes and be involved in the development of cancer.
Allelic deletions can occur in the entire copy or part of chromosome 10. Several commonly deleted loci have been identified on chromosome 10q in the region spanning 10q23 to 10qter, encompassing the allelic loci of a number of genes with well-established tumor-suppressive roles (e.g., MMAC/PTEN, SUFU, FGFR2, and DMBT1) or putative tumor-suppressive roles (e.g., ADD3/MXI1, LGI1, and BTRC) (9,(15)(16)(17). When comparing cohorts with primary and secondary gliomas, the acquisition of a malignant phenotype with marked proliferative activity was observed in a deletion mapping on 10q25-qter (18). This further suggests the importance of tumor suppressor genes in 10q25-qter apart from PTEN that are likely to be involved in the malignant progression of glioma.
Among these putative tumor suppressor genes, ADD3 was found to be downregulated in high-grade gliomas when compared with its less malignant counterpart in several gene expression profiling studies (19)(20)(21). ADD3 is located on chromosome 10q25.1-25.2, which is known to be a functional tumor suppressor region. While it primarily functions as a cytoskeleton protein, ADD3 is characterized as a negative regulator of tumor growth and is negatively associated with malignant phenotypes such as angiogenesis in GBM (22). However, the basis behind ADD3 downregulation has not been identified for which both genetic and epigenetic modifications should be taken into consideration when studying its functional characteristics in cancer cells. Moreover, although PTEN locus is often tested for chromosome LOH10q status, the frequency of allelic deletion on other 10q loci has not been adequately addressed in previous studies.
Here, we performed deletion mapping analysis in a glioma cohort to determine the regions of allelic loss in chromosome 10q and to establish correlations between their accumulation and different pathological phenotypes. By correlating patient characteristics to the occurrence of allelic loss, we determined whether the aberrant expression of ADD3 in malignant glioma results from LOH10q, which may be involved in tumor relapse and malignant progression. We will also evaluate and highlight the potential application of ADD3 as a novel biomarker for diagnostic, prognostic, and future therapeutic implications.

Human Glioma Tissue Specimens
Fresh tumor tissues were snap-frozen in liquid nitrogen immediately after surgical resection at our institution. All specimens were obtained with informed consent from the patients. The study protocol was approved by our Institution Review Board. Diagnosis and histological classification were confirmed by specialists according to the WHO's brain tumor classification system before the release of the WHO 2016 classification on gliomas. Forty-three glioma specimens with WHO grades 2-4 (8 grade 2, 12 grade 3, and 23 GBM/grade 4) were collected for clinical and molecular analyses in the study. Specimens were obtained at the time of surgery between 2009 and 2016 and were stored at −80°C until prior study use. All diagnoses were made radiologically via MRI before surgery and confirmed histologically by a certified pathologist. Tumors from another cohort of ten patients (P1-P10) who developed recurrence after the initial surgical treatment of the disease were also collected during the second operation for longitudinal comparison in our study.

Clinical Data
The records of forty-three patients were retrieved from their medical records from initial consultation to the latest follow-up. The data included the patient's age, gender, date of diagnosis, pathological diagnosis, initial tumor location, MRI-confirmed tumor size, and nature of the specimen types (primary or recurrent tumor). Other characteristics were MGMT promoter methylation status, cycles of temozolomide treatment, survival time in days after treatment, and patient status at the last followup (dead or alive). Missing data and patients who were lost to follow-up were not counted in the study. The study cutoff date was November 30, 2020, and patients who survived through this date were defined as the last follow-up date.

PCR and PCR-Based Microsatellite Loss of Heterozygosity Analysis
Genomic DNA was extracted from frozen tumor tissue using PureLink Genomic DNA mini kit (Life Technologies), according to the manufacturer's protocol. After DNA extraction, 50 ng of genomic DNA was amplified using AmpliTaq gold 360 master mix (Applied Biosystems) by PCR. Reaction mix measuring 25 ml was subjected to an initial denaturation cycle of 95°C for 7 min, followed by 35 cycles of denaturation at 95°C for 30 s, annealing at 58°C-62°C for 30 s, and extension at 72°C for 30 s, with a final extension at 72°C for 7 min.
LOH on chromosome 10q was studied by PCR-based microsatellite analysis of the frequently deleted regions on chromosome 10q. Five microsatellite markers spanning across 10q23-qter were selected according to the genetic loci on the UCSC Table Browser on Human (GRCh37/hg19) assembly. Primer sequences were purchased from Life Technologies. Microsatellite marker D10S579 flanked PTEN gene (10q23.2); D10S198 is an intragenic marker within CNNM1 gene (10q24.2); D10S173 is intragenic within MXI1 and is 50 kb adjacent to ADD3 gene (10q25.2); D10S1483 is intragenic of FGFR2 gene (10q26.13); and D10S1137 flanked MGMT gene (10q26.3). The cytogenetic localization and the approximate genetic distances are shown in Figure 1. Tumor DNA samples were analyzed by capillary gel electrophoresis. DNA fragments from the PCR reaction were separated by a Fragment Analyzer automated CE system (Agilent). Allelic loss for each microsatellite locus was determined by the evaluation of peak intensity using PROSize 3.0 software (Agilent).

Loss of Heterozygosity Evaluation
Allelic loss was evaluated according to the peak-height ratio as previously described (23). A high level of LOH was assumed when the signal intensity of the allele in tumor tissue was less than half of that in the reference DNA (i.e., DNA from the normal brain). LOH ratio ≤0.5 or LOH ratio ≥1.5 is defined as a high level of LOH; LOH ratio >0.5 or <1.5 is defined as low level or no LOH. LOH ratio of <0.5 indicates significant loss of allele 2 in the glioma tissue, and a ratio of >1.5 indicates the significant loss of allele 1 in the glioma tissue. LOH10q was defined when at least three informative loci were deleted.

Bioinformatics Analysis
Gene expression data, DNA copy number information, and survival data were downloaded from cbioportal for cancer genomics (https://www.cbioportal.org/). The Cancer Genome Atlas (TCGA) project used for analyses consisted of clinical and gene expression data from 206 GBM patients (24).

Western Blotting
Tissue protein lysates were prepared from clinical specimens snap-frozen in liquid nitrogen and stored at −80°C until use. (B) Assessment of allelic loss in fragment analysis on an automated capillary electrophoresis system (from seven representative cases). Allelic status was calculated based on the peak intensity ratios in gliomas relative to normal brain tissue. LOH ratio of >0.5 or < 1.5 is defined as a low level of LOH, and a ratio of ≤0.5 or ≥ 1.  Protein lysates were extracted using radioimmunoprecipitation assay (RIPA) lysis buffer with protease inhibitor cocktail. Detailed procedures of immunoblotting were performed as previously described (25). Anti-ADD3 antibody was purchased from Sigma-Aldrich (#HPA035696); anti-GAPDH antibody (#2118) was purchased from Cell Signaling Technologies. Protein band intensities were quantified by ImageJ software. Band intensities were measured as the total volume under the peak of interest, with background intensity subtracted under the peak of interest, and normalized to that of the reference protein (GAPDH).

Immunohistochemical Staining
Immunohistochemical (IHC) staining of formalin-fixed paraffinembedded (FFPE) sections was performed on consecutive 5-mmthick sections. Tissue sections were subjected to deparaffinization by xylene and rehydration in serial dilutions of ethanol, followed by heat-induced antigen retrieval in 10 mM of sodium citrate (pH 6.0). Endogenous peroxidase was quenched by treatment with 3% hydrogen peroxide for 30 min, and non-specific protein binding was blocked with 10% normal goat serum (Dako) for 1 h. Sections were incubated with primary antibodies at appropriate dilutions at 4°C overnight in a moist chamber. After incubation, sections were washed with Tris-buffered saline three times, followed by incubation with horseradish peroxidase (HRP)conjugated secondary antibodies (Dako) for 30 min. DAKO EnVision System (Dako) was used to detect signals from DAB chromogen substrate. Finally, sections were counterstained with hematoxylin (Vector Laboratories) and mounted in DPX mounting solution (BDH Laboratory). All IHC sections were quantified and evaluated as the mean value from five random ×200 microscopic fields. All calculations on tissue sections were processed and analyzed by ImageJ software.

Statistical Analysis
The goals of the statistical analysis were to uncover the clinical association of 10qLOH by ascertaining molecular genetics and pathologic and clinical parameters. The association between variables was tested with Pearson's chi-squared (c 2 ) test. The Kaplan-Meier method was used to estimate overall survival. The time of an event was calculated as the date of the initial pathological diagnosis until the time to death or the time of the last contact if the patient was alive or the last day of the study period. Differences in survival distributions were evaluated using a log-rank test. All the results were considered statistically significant when the two-sided p was <0.05.

Patient Demographics
This is a retrospective analysis of 43 glioma cases evaluating the association between allelic loss at chromosome 10q and the clinicopathological features.

Deletion Mapping Revealed Diagnostic Implication of D10S173 in Gliomas
Five microsatellite markers spanning 10q23 to 10q26 were selected for LOH analyses in 43 cases of gliomas. Figure 1A illustrates the corresponding chromosomal loci and the markers flanking or intragenic of the gene loci. LOH with three or more loci in all examined chromosomes was considered to amount to the entire loss of the long arm (55.8% in all cases, 37.5% in grade 2, 50% in grade 3, and 69.5% in GBM); 18/43 (41.8%) showed partial or interstitial allelic losses, and only one grade 2 glioma A#3 was intact. Of the gliomas, 97% harbor allelic loss in at least one locus. The peak intensity from normal brain tissue (N#1) was used as a reference for LOH analysis as shown in Figure 1B. It shows the representative peak from capillary electrophoresis, where all the five alleles were lost in one of the grade 3 gliomas (AA#1). The frequency of loss in gliomas of different malignancy grades is represented as a deletion map according to their genetic loci ( Figure 1C). Among the five microsatellite markers, D10S173 at ADD3/MXI1 locus demonstrated frequent deletion in high-grade gliomas (83.3% in grade 3 and 82.6% in grade 2) and was less common in low-grade cases (12.5%) (  D10S173 had poorer survival than patients with low LOH (*p = 0.047) (Figure 2A). Our finding was validated using a larger GBM cohort in the TCGA database, which showed copy number variations of both ADD3 and MXI1 in GBM. Survival analysis of GBM with ADD3 copy number loss, 46 diploid vs. 157 deletion ( Figure 2B) and MXI1 copy number loss, and 21 diploid vs. 69 deletion ( Figure 2C) showed a significant correlation with an unfavorable outcome (log rank **p < 0.01), with a median survival of 12.95 months in ADD3-deleted tumors compared with 15.65 months in non-deleted (diploid) tumors (*** p < 0.001) and median survival of 12.95 months in MXI1 deleted tumors compared with 28.47 months in non-deleted (diploid) tumors (*** p < 0.001).

ADD3 Loss Is Implicated in Disease Progression
While the tumor-suppressive role of MXI1 in glioma has been reported, less is known about the role of ADD3 in glioma. In this study, we showed that LOH at D10S173 (ADD3/MXI1) was frequently seen in high-grade gliomas compared with low-grade gliomas and with a high proliferative index, and the results were suggestive of the role of these genes in regulating tumor progression. ADD3 protein expression was determined by Western blotting and IHC in another cohort with 10 patientpaired primary-recurrent gliomas (P1-10). In line with our hypothesis, further losses of ADD3 protein expression were observed in 6 out of 7 gliomas progressing to a higher malignancy grade (P1-P7, Figure 3A). ADD3 was also further downregulated in recurrent GBM compared with its primary lesion (P8-P10) ( Figure 3B).

DISCUSSION
Loss of genetic material at certain chromosomal regions has been considered a major event in tumor development and progression. The assessment of allelic loss is one of the most useful approaches hinting at the loci of potential tumor suppressor genes. Chromosome 10q has been suggested to encompass multiple tumor suppressors apart from PTEN (10q23), on account of the frequent losses observed also at loci in the distal region of 10q (10q25-qter) (5,9,10,16). In an attempt to assess the possible implications of deletion at different chromosome regions, we examined the frequency of 10q LOH in a glioma cohort and evaluated the association with patients' clinicopathological characteristics. The results from deletion mapping were consistent with those of other studies, where most GBM cases demonstrated a high incidence of allelic loss and appeared to have lost the entire arm of chromosome 10q (69.5%). Of the five allelic loci examined, only LOH of D10S173 at the ADD3/MXI1 locus was a predictor of shorter survival and had a significant association with tumor grade and proliferative index. MXI1 is a transcription repressor of MYC (26). Despite that somatic mutations were not found in MXI1, which did not seem to support the two-hit hypothesis for gene inactivation as a tumor suppressor, its growth-suppressive    function has been revealed in different cancers (27,28). Compared with MXI1, less is known about the role of ADD3 in cancer. We reported previously on the putative tumorsuppressive function of ADD3 in GBM (22). It is by chance mapped to chromosome 10q25.2, where LOH frequently occurs.
ADD3 was found to be significantly downregulated in GBM, and such loss was associated with enhanced tumor growth (22). In this study, we were able to demonstrate the mechanism by which ADD3 was downregulated in gliomas as being most likely due to LOH with copy number deletions. Our findings further suggested that ADD3 is a putative tumor suppressor in GBM. However, one should note that the current study was conducted before the release of the fifth edition of WHO classification of gliomas published in 2021, which has grouped tumors according to the genetic changes including IDH mutation (29). The term "glioblastoma" has been classified as a specific entity with an IDH wild-type genotype. It is important to note that the term "glioblastoma or GBM" being used in this study is irrespective of the IDH mutation status. Future studies should further refine the prognostic significance of ADD3 in subgroups of tumors with and without IDH mutation.
ADD3 is a crucial assembly factor in the actin cytoskeleton that functions to recruit and promote the formation of spectrin-actin membrane skeleton to provide physical support in the cell (30,31). Given that the cytoskeleton structures are essential for cell motility, downregulation of ADD3 has been shown to be associated with enhanced migratory and invasive potential in lung cancer cells (32). Alteration of ADD3 may also modulate the tumor microenvironment mediated by changes in focal adhesion as well as cell-cell contacts (32,33). It was found that ADD3deficient GBM cells were able to elicit pro-angiogenic signals to stimulate VEGFR expression in endothelial cells (22). All of these could be the potential mechanisms and the pathophysiological consequences that underlie allelic loss of ADD3 in malignant gliomas, particularly during disease progression.
There is accumulating evidence on the prognostic significance of ADD3 in GBM. A recent study by Navarro et al. reported a strong association between EGFR amplification (or EGFRvIII expression) and ADD3 copy number variations in GBM. Genetic clustering suggested the co-occurrence of both events in a subgroup of GBM, the cluster where patients were presented with the shortest survival (34). In accordance with this finding,  GBMs with EGFR amplification are often accompanied by the complete loss of chromosome 10 and are considered the common phenotypic endpoint of different genetic pathways (35,36). Another notable finding is the progressive event of ADD3 loss during malignant progression in the patient-paired specimens. Indeed, ADD3 deletion appears to occur at a much higher frequency, primarily in the most malignant subtype compared with those in low-grade cases, suggesting that further loss of ADD3 is an event associated with disease progression. In line with this observation, LOH at 10q-qter was associated with increased proliferative activity, together with the acquisition of a morphological transition to an advanced malignant status (18).

CONCLUSION
This study provides novel insights into understanding the pathogenesis of malignant progression in glioma. The identification of ADD3/MXI1 locus carries diagnostic and prognostic implications and can be particularly helpful for making decisions concerning postoperative management. ADD3 is a putative tumor suppressor that may also serve as a promising marker on 10q in predicting survival. While a limited number of allelic loci are being tested for LOH10q, pathological tests could be refined with additional markers at loci with higher prognostic significance. The importance of ADD3 gene assumes a new dimension that evaluation on ADD3 expression may be useful to identify patients who are likely to have a progressive disease. The fact that ADD3 is dysregulated at the genetic level should be further explored in regard to the underlying mechanisms of tumor progression and genetic vulnerabilities in glioma. Future studies may target ADD3 depleted tumors as a novel therapeutic approach utilizing synthetic lethality, which can be developed to target gene defects in tumors with ADD3 deletion, similar to the concept of using PARP inhibitors exclusively for BRCA-mutated breast tumors. Future investigations may also focus on the genetic alterations associated with ADD3 and the possible regulatory mechanisms on the transcriptional and post-transcriptional levels.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Institution Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

AUTHOR CONTRIBUTIONS
KK carried out the experiments and wrote the paper. SS performed the statistical analysis and collected the data. GL conceived the study and contributed to the critical revision. All authors contributed to the article and approved the submitted version.