The Oncogenic Potential of the Centromeric Border Protein FAM84B of the 8q24.21 Gene Desert

FAM84B is a risk gene in breast and prostate cancers. Its upregulation is associated with poor prognosis of prostate cancer, breast cancer, and esophageal squamous cell carcinoma. FAM84B facilitates cancer cell proliferation and invasion in vitro, and xenograft growth in vivo. The FAM84B and Myc genes border a 1.2 Mb gene desert at 8q24.21. Co-amplification of both occurs in 20 cancer types. Mice deficient of a 430 Kb fragment within the 1.2 Mb gene desert have downregulated FAM84B and Myc expressions concurrent with reduced breast cancer growth. Intriguingly, Myc works in partnership with other oncogenes, including Ras. FAM84B shares similarities with the H-Ras-like suppressor (HRASLS) family over their typical LRAT (lecithin:retinal acyltransferase) domain. This domain contains a catalytic triad, H23, H35, and C113, which constitutes the phospholipase A1/2 and O-acyltransferase activities of HRASLS1-5. These enzymatic activities underlie their suppression of Ras. FAM84B conserves H23 and H35 but not C113 with both histidine residues residing within a highly conserved motif that FAM84B shares with HRASLS1-5. Deletion of this motif abolishes FAM84B oncogenic activities. These properties suggest a collaboration of FAM84B with Myc, consistent with the role of the gene desert in strengthening Myc functions. Here, we will discuss recent research on FAM84B-derived oncogenic potential.


Introduction
Tumorigenesis is a complex pathological process. It is affected by sophisticated genetic networks and even more complex epigenetic modifications. The multiplex nature of oncogenesis underlies our continuous effort in the search for cancer etiology. One of the classical genetic events of oncogenesis is amplification of the Myc (c-Myc) oncogene [1,2]. Myc is the most commonly amplified oncogene across all cancer types [3]. Despite its powerful oncogenic nature, Myc's oncogenic potential cannot be fulfilled without direct contributions from other oncogenes. For instance, BMI1 (B lymphoma Mo-MLV insertion region 1 homolog) was identified during screenings for potential collaborators for c-Myc-initiated leukemogenesis [4,5]. c-Myc mediates BMI1 gene transcription to ensure BMI1 availability during oncogenesis for leukemia, neuroblastoma, and nasopharyngeal carcinoma [6][7][8]. Among numerous Myc collaborators, Ras is arguably the most classic one. Their collaboration results in transformation of primary fibroblasts [9] and the activation of cyclin D-and E-dependent kinases [10,11].
The Myc gene resides on 8q24.21 and is surrounded by regions known as "gene deserts" as they lack protein coding genes. Downstream (telomeric end) of Myc is the PVT1 gene encoding for a long   After reviewing FAM84B's contributions to oncogenesis, we will propose a model to discuss FAM84B's oncogenic roles in the context of Myc-derived tumorigenesis, i.e., a potential mechanistic pathway for which FAM84B collaborates with Myc during tumor formation.
The 8q24.21 gene desert contributes to cervical cancer as a frequent site of viral integration by human papilloma virus (HPV), and evidence in support of this concept has been briefly reviewed by Huppi et al. in 2012 [13]. Built on the seminal detection of HPV16 DNA in 61.1% (11/18) of cervical cancers in 1983 [61], it became clear that infection by HPV is the primary etiology of cervix carcinoma, particularly with the high-risk HPV types 16 and 18 [62]. Besides HPV infection, alterations in cellular oncogenic events are also required for cervical cancer [63]. The integration of both HPV16 and HPV18 at the gene desert suggests that HPV coordinately affects oncogene alterations for cervical cancer formation [13,[64][65][66][67]. The integration hot spots in the gene desert include CASC8, CASC21, and POU5F1B [68,69]. Among 3667 breakpoints of HPV integration detected in cervical carcinoma (n =
The 8q24.21 gene desert contributes to cervical cancer as a frequent site of viral integration by human papilloma virus (HPV), and evidence in support of this concept has been briefly reviewed by Huppi et al. in 2012 [13]. Built on the seminal detection of HPV16 DNA in 61.1% (11/18) of cervical cancers in 1983 [61], it became clear that infection by HPV is the primary etiology of cervix carcinoma, particularly with the high-risk HPV types 16 and 18 [62]. Besides HPV infection, alterations in cellular oncogenic events are also required for cervical cancer [63]. The integration of both HPV16 and HPV18 at the gene desert suggests that HPV coordinately affects oncogene alterations for cervical cancer formation [13,[64][65][66][67]. The integration hot spots in the gene desert include CASC8, CASC21, and POU5F1B [68,69]. Among 3667 breakpoints of HPV integration detected in cervical carcinoma (n = 104), cervical intraepithelial neoplasia (n = 26), and 5 cervical cancer cell lines, POU5F1B is the top site of integration (9.7%) [69].

Upregulation of Myc as A Mechanism underlying the Gene Desert-Derived Oncogenic Activities
In light of the well-established and powerful oncogenic functions of Myc, it is expected that research exploring the oncogenic impact of those non-coding genes within the gene desert ( Figure 1) has been largely focused on the regulation of Myc. HPV integration in the 8q24.21 gene desert upregulates Myc [69]. CCAT2 expression is upregulated in CRC and lnRNA CCAT2 enhances Myc expression, which likely contributes to CCAT2-facilitated CRC metastasis [70].
The major mechanistic action in enhancing Myc expression is through regulation of chromatin structure. By examination of chromatin interactions using chromosome conformation capture (3C)-based technologies, the prostate, breast, and colon cancer risk regions within the 8q24.21 gene desert display long-range physical interaction with the Myc locus in a tissue-specific manner [71,72]. These non-coding risk regions contain super-enhancer elements and TCF-4 (transcription factor 4) binding sites that enhance Myc transcription [71,73]. The long-range association of these regions with the Myc locus thus stimulates Myc transcription, which is facilitated by Wnt/β-catanin signaling through TCF-4. These enhancers are functionally important. Mice deficient in an enhancer element Myc-335 that lies 335kb upstream of Myc are protected from APC (Adenomatous polyposis Coli) mutation-induced intestinal cancer [74]. Mice deficient in multiple Myc enhancers, including Myc-196, Myc-335, and Myc-540, within 538kb upstream of Myc, exhibit >50% reductions of Myc expression in colon and prostate. Importantly, these mice are more protected from APC mutation-induced intestinal cancer compared to mice deficient in only Myc-335 [75]. Both CCAT1 and CCAT2 interact with Myc via the formation of DNA loops, which strongly enhances Myc expression in CRC [17,70,76]. Interestingly, long-range physical associations with Myc also facilitate the transcription of lnRNAs. For instance, the physical association allows the Myc enhancer to upregulate the transcription of CARLo-5, a short form of CCAT1 [77].
Variants in the 8q24.21 gene desert also display long-range association with the non-coding PVT1 locus that lies downstream of the Myc locus and thus outside of the gene desert bordered by FAM84B and Myc (Figure 1). A prostate cancer risk variant within the gene desert was reported to facilitate PVT1 transcription through physical association [78]. The lnRNA PVT1 plays a critical role in Myc-driven CRC. PVT1 is co-amplified with Myc in CRC. High levels of lnRNA PVT1 helps to maintain high levels of Myc protein expression in CRC, and ablation of PVT1 prevents Myc from inducing HCT116 cell-derived tumorigenesis [79]. In addition to Myc, PVT1 also activates β-catenin and Cyclin D1 [80]. The interplay between PVT1 and Myc has been intensively studied and reviewed [15].

The Contributions of FAM84B to Oncogenesis
FAM84B is the only second protein coding gene bordering the 1.2 Mb gene desert (Figure 1). It is an interesting disparity considering the relatively unknown status of FAM84B in tumorigenesis compared to the well-studied oncogenic functions of Myc. Nonetheless, emerging evidence suggest the need for a closer examination of FAM84B's involvement in oncogenesis. In this section, we will review the data related to FAM84B's roles in cancer.

FAM84B Facilitates Esophageal Cancer
FAM84B plays a role in esophageal cancer. In a small cohort study (n = 59), increases in FAM84B expression were observed in 39 (66%) cases [81]. Amplification of the FAM84B gene and increases in its expression at the protein level occur in both preclinical lesions and esophageal squamous cell carcinomas (ESCC) [82,83]. Reductions in serum FAM84B protein expression predict pathological complete response (PCR) in ESCC patients treated with neoadjuvant chemoradiation [83]. Knockdown of FAM84B in two ESCC cell lines KYSE150 and TE-1 reduced their proliferation, migration, and invasion in vitro [82], and knockdown of FAM84B in ESCC CE81T/VGH cells significantly delayed xenograft growth in vivo [83]. Upregulations of FAM84B were also reported in melanoma [84]. However, the role of FAM84B in tumorigenesis may be complex. While downregulation of serum FAM84B protein is associated with PCR in ESCC treated with neoadjuvant chemoradiation, high levels of serum FAM84B mRNA were also observed in ESCC with PCR [83]. Downregulation of FAM84B was observed in gastroesophageal junction cancer cell lines and xenograft tumors [85]. Increases in the expression of lnRNA FAM84B-AS (antisense) transcribed from the antisense strand of the FAM84B gene were reported to reduce FAM84B expression in gastric cancer. lnRNA FAM84B-AS facilitates gastric cancer tumorigenesis and predicts poor prognosis [86].

FAM84B-Mediated Enhancement of Prostate Cancer
Evidence supports FAM84B-mediated promotion of prostate cancer (PC). FAM84B locus lies within a 2Mb region that is associated with PC risk [87]. We observed a significant upregulation of FAM84B expression in DU145 PC cell-derived prostate cancer stem cells (PCSCs) [12]. This observation is in accordance with a report showing that a risk region of prostate and colon cancer in the 8q24.21 desert was able to direct reporter expression in prostate luminal stem-like cells of transgenic mice and in prostate cancer stem cells [88]. PCSCs play critical roles in PC initiation and progression, including metastasis and therapy resistance [89]. PC mainly metastasizes to the bone [90]. The standard of care for metastatic PC (mPC) is androgen deprivation therapy (ADT). While the therapy shows remarkable response in more than 80% of cases, castration-resistant metastatic PCs (mCRPCs) commonly develop [91], to which effective therapy remains challenging. In this regard, bone metastasis and CRPC are considered major progression with poor prognosis. Of note, in comparison to prostate (n = 181), FAM84B mRNA was elevated in PC (n = 343) and further increased in metastasis in two populations (primary PC, n = 131, versus metastatic PC, n = 19; primary PC, n = 181, versus metastatic PC, n = 37) [12]. In vivo, FAM84B protein was expressed at higher levels in PCSCs-generated xenografts compared to non-PCSCs-produced xenografts, in lung metastasis compared to subcutaneous xenografts, and in CPRC produced in castrated prostate-specific PTEN -/mice compared to PC generated in intact PTEN -/mice [12]. Amplification of the FAM84B gene occurs more frequently in mCRPC (121/467 = 26%) compared to primary PCs (26/546 = 4.8%, p < 0.0001), and the amplification associates with reductions in disease-free survival (DFS) [12]. Additionally, increases in FAM84B mRNA expression contribute to the biomarker potential of a multigene panel in stratification of the risk of PC biochemical recurrence [92]. Collectively, a comprehensive set of evidence supports the association of FAM84B with PC tumorigenesis, metastasis, and CRPC development.
Functionally, FAM84B overexpression enhances DU145 cell invasion in vitro, subcutaneous xenograft tumor growth in vivo, and lung metastasis in a tail-vein mouse model [93]. In comparison to empty vector (EV) cell-produced xenograft tumors, those generated from FAM84B overexpression DU145 cells display elevations in AKT activation and reductions in BAD (BCL2 associated agonist of cell death) expression. Furthermore, RNA sequencing (RNA-seq) analysis revealed a large number of differentially expressed genes (DEGs) in DU145 FAM84B xenografts compared to DU145 EV tumors. These DEGs affect cell cycle progression, Golgi to ER (endoplasmic reticulum) process, mitochondrial events, and translation regulations [93]. A multigene signature (SigFAM) was derived from these DEGs, which robustly stratifies the risk of PC biochemical recurrence [93].

Potential Collaboration between FAM84B and MYC during Tumorigenesis
HRASLS1-5 possess enzymatic activities: phospholipase A 1/2 (PLA) and O-acyltransferase (AT) activities [94] with the catalytic site being formed by histidine 23 (H23), H35, and cysteine 113 (C113) (Figure 3) [94,95]. HRASLS members can suppress H-Ras-derived tumorigenesis, in which the catalytic activities play a role [94]. HRASLS1/A-C1 inhibits the proliferation of H-Ras-transformed NIH3T3 cells (Table 2) [96]. Ectopic expression of HRASLS2 suppresses the colony formation of HCT116 (colon cancer) and HeLa (cervical cancer) cells, and reduces the active Ras (Ras-GTP) and Ras expression in HtTA cervical cancer cells (Table 2) [97]. HRASLS3 (H-rev107) was most thoroughly studied in the HRASLS family for suppression of Ras activity. H-rev107 was identified for reversal of H-Ras-derived transformation of rat fibroblasts [98], the PLA/AT activities of HRASLS3 suppress H-Ras signaling [99], and HRASLS3 inhibits K-Ras signaling via a physical association ( Table 2) [100]. HRASLS4 (RIG1, TIG3, RARRES3) suppresses Ras activation [101] and the lung metastasis of breast cancer (  While the mechanisms responsible for FAM84B-derived oncogenesis remain unclear, its direct association with the gene desert as the only other protein coding gene suggests that FAM84B contributes to Myc's oncogenic actions. This possibility is consistent with the theme of 8q24.21 gene desert in facilitating Myc actions. This concept is also intriguing considering the similarities shared between FAM84B and the HRASLS family. Among the catalytic triad, FAM84B conserves H23 and H35 but not C113 (Figure 3) [93]. Unlike the HRASLS family, FAM84B does not possess PLA/AT enzymatic activities and will not suppress Ras signaling. To the contrary, FAM84B displays oncogenic activities. With this knowledge, it is an interesting scenario for FAM84B to facilitate Ras signaling via inhibiting the actions of the HRASLS family, and thereby in part contributing to its collaboration with Myc. The collaboration between Myc and Ras was the first demonstration of oncogene collaboration and is the most widely studied relationship [11]. Mechanisms of this collaboration are complex, and the FAM84B concept will be a new avenue in this collaboration considering its genome proximity to the Myc gene locus. Its functional and genetic linkage with Myc would suggest a co-regulatory pattern with Myc. In support of this possibility, mice deficient in the 430kb region encompassing CCAT1, POU5F1B, CCAT2, and CASC8 within the 8q24.21 gene desert ( Figure 1) downregulate both FAM84B and Myc expression in mammary gland and prostate [103]. Deletion of this region delays the growth of luminal, Her2, and basal breast cancer in MMTV-PuVT, MMTV-Neu, and C3(1)-TAg transgenic mouse models for breast cancer, respectively [103]. Furthermore, among 35 cancer types within the cBioPortal database, 20 cancer types show co-amplification of FAM84B and Myc with the rate ≥5%. Ovarian cancer and breast cancer are the first (>40%) and fifth (>20%) cancer type with respect to the prevalence of FAM84B and Myc co-amplification [103]. This co-amplification associates with poor overall survival in breast cancer [103].
The co-amplification has also been detected in acute myeloid leukemia [104], esophageal cancer [105], colorectal carcinoma [106], and prostate cancer [103]. For prostate cancer (PC), we have analyzed gene amplification for Myc and FAM84B in all independent published datasets (n = 12) containing 3546 patients within cBioPortal (http://www.cbioportal.org/) [107,108] (Figure 4A). Among PCs with Myc amplification, 85% of cases have FAM84B concurrently amplified, and among tumors with FAM84B amplification, 96.8% of cases show Myc co-amplification ( Figure 4A). Neuroendocrine and metastatic PCs are well-known for having poor prognosis [89]. Of note, 55.5% (152/274) of PCs with the co-amplification are aggressive PC types (metastatic and neuroendocrine PCs) ( Figure 4A). In line with this evidence, PCs with FAM84B and Myc co-amplification are associated with reductions in overall survival ( Figure 4B). Collectively, evidence supports an intriguing collaboration between FAM84B and Myc.
Genes 2020, 11, x FOR PEER REVIEW 9 of 16 associated with reductions in overall survival ( Figure 4B). Collectively, evidence supports an intriguing collaboration between FAM84B and Myc.

Conclusions
Along with Myc being the most commonly amplified gene in human cancers, the 8q24.21 gene desert bordered by both Myc and FAM84B is also frequently amplified [3]. While the oncogenic functions of the gene desert likely involve complex networks, it promotes tumorigenesis at least in part via facilitating Myc's actions. In this regard, we proposed a collaboration between FAM84B and Myc which may involve Ras. Demonstration of this possibility is straightforward owning to the rich knowledge on the collaboration between Ras and Myc. However, the interaction between FAM84B and Myc is likely not limited to the potential connection of FAM84B and Ras. For instance, FAM84B and Myc may interact via lnRNAs within the 8q24.21 gene desert. This possibility is in accordance with the co-downregulation of both genes in mice with knockout of a 430Kb fragment within the gene desert [103]. Direct interactions between FAM84B and Myc at both the protein and transcriptional levels are also possible, and the latter is intriguing in view of Myc being a

Conclusions
Along with Myc being the most commonly amplified gene in human cancers, the 8q24.21 gene desert bordered by both Myc and FAM84B is also frequently amplified [3]. While the oncogenic functions of the gene desert likely involve complex networks, it promotes tumorigenesis at least in part via facilitating Myc's actions. In this regard, we proposed a collaboration between FAM84B and Myc which may involve Ras. Demonstration of this possibility is straightforward owning to the rich knowledge on the collaboration between Ras and Myc. However, the interaction between FAM84B and Myc is likely not limited to the potential connection of FAM84B and Ras. For instance, FAM84B and Myc may interact via lnRNAs within the 8q24.21 gene desert. This possibility is in accordance with the co-downregulation of both genes in mice with knockout of a 430Kb fragment within the gene desert [103]. Direct interactions between FAM84B and Myc at both the protein and transcriptional levels are also possible, and the latter is intriguing in view of Myc being a transcriptional factor [109]. Regardless of what major routes FAM84B may employ in its interaction with Myc, this interaction is certainly an appealing avenue of investigation. This proposition is based both on the accumulating evidence for an oncogenic role of FAM84B as well as the association of FAM84B and Myc with the 8q24.21 gene desert. While the impact of FAM84B on tumorigenesis has been relatively well-studied in prostate cancer, its oncogenic functions in general and its potential relationship with the HRASLS family should be explored in the future.