An integral component of understanding E3 ligase function is identifying the enzyme's cognate substrate(s) and/or binding proteins. Previous binding partners of BCA2 included Rab7, isolated through yeast-II-hybrid screening [17
], tetherin, which was found though brute-force GST-pulldowns [18
], and ubiquitin and UBC9 from bacteria-II- hybrid screening [7
]. Bacteria and yeast screening systems were used in this study to identify additional potential binding partners of BCA2 (Table ). Of the potential partners found, 14-3-3σ and hHR23a were chosen for further investigation in the context of BCA2 activity and expression. Through binding experiments we confirmed that the BCA2 protein bound both to 14-3-3σ and hHR23a (Figure and ). hHR23a and BCA2 were co-expressed in a mammalian system, while 14-3-3σ was expressed in a bacterial system and incubated with recombinant purified BCA2. The expression levels of wild-type BCA2 and the S132, 133A mutant in HEK293T cells were much lower than those of the RING-mutant and BZF mutant, which is a possible confounding factor in determining domains necessary for binding hHR23a. The overexpression of BZF and RING mutants likely resulted from steric hinderance of BCA2 autoubiquitination activity from the GST-expression tag. Both the wild-type BCA2 protein and the RING mutant, but not the BZF or S132, 133A mutants of BCA2 were pulled down by 14-3-3σ, indicating that sequences in the BZF domain or AKT domain may be necessary for interaction.
Protein interaction was confirmed in an endogenous system. Breast cancer cell lines were probed for expression of hHR23a and 14-3-3σ, along with BCA2 (Figure ). BCA2 has previously been reported as co-expressing with estrogen receptor [2
], thus high expression was expected in ER-positive cell lines ZR751, BT474 and MCF7, while lower expression was expected in the ER-negative cell line MDA MB 231. Our findings were consistent with these expectations. hHR23a was expressed in all cell lines, but had slightly higher expression in the ER-positive cell lines, which correlated with the co-expression of hHR23a and ER in breast cancer TMAs. Immunofluorescence studies confirmed that endogenous proteins co-localized.
Protein expression was further investigated in the context of breast cancer analyzed though IHC. Tumor cores which stained highly positive for BCA2 also stained highly positive for hHR23a in both the nucleus and the cytoplasm. This correlation was determined to be significant by Fisher's Exact T-Test. From this we inferred that not do these proteins interact, but they are co-expressed endogenously in the same cells and tissues. While not statistically significant, in the tumors BCA2 and 14-3-3σ trended towards an inverse relationship in expression. The majority of samples which stained highly for BCA2 were scored as having low expression of 14-3-3σ.
A problem which emerged over the course of studying BCA2, was the unstable nature of the protein due to intrinsic autoubiquitination activity. Initial investigations of BCA2 stability, found that degradation via autoubiquitination was equally possible with all UbcH5 family members, but not with other E2's such as the UbcH3 ubiquitin conjugating enzyme. Moreover, the presence of members of the UbcH5 family was responsible for the rapid and complete degradation of wild-type BCA2 protein in vitro and in vivo (Figure ) [2
As a molecular chaperone, hHR23a regulates the lifespan of its binding partners in both agonisticly and antagonistically. In the proteasome system, hHR23a acts as a shuttle to the proteasome, by simultaneously binding both the target protein and the lid of the 26 s proteasomal subunit. In contrast to this function, hHR23a also regulates protein half-life through sequestering nascent ubiquitin moieties in the emerging polyubiquitin chain. Thereby preventing chain elongation and de-ubiquitinase activity and stabilizing the targeted protein [25
]. This dual nature of hHR23a is largely due to the presence of its N-terminal Ubiquitn-Like domain (UBL) and two C-terminal Ubiquitin-Associating (UBA) domains [26
]. The UBL domain may be a key component of hHR23a's interaction with BCA2. The significance of this interaction is such that it increases the stability of BCA2. Increased concentration of the partner protein correlated with the elevated levels of BCA2 protein (Figure ). We inferred that BCA2 was stabilized through the prevention of ubiquitin chain elongation by hHR23a (Figure ). This was consistent with reported behaviour of hHR23a as an ubiquitin receptor. These observations were strengthened since hHR23a was also shown to increase the half-life of BCA2 over a 24 hours time period (Figure ). hHR23a was not a substrate for BCA2-mediated degradation (Figure ). This does not exclude hHR23a mono- or multi-ubiquitination from being catalyzed by BCA2. Figure showed an increased stability of hHR23a in the presence of high BCA2 expression, which may result from ubiquitin modification of hHR23a, which has been previously reported [22
]. Taken together, data presented here argues that hHR23a presence may be necessary to modulate and regulate BCA2 in a cancer setting.
As well as through interaction with hHR23a, BCA2 was also increased through modification by AKT phosphorylation. Phosphorylation has far reaching consequences for the modified proteins, amoung them are conformational changes which effect substrate affinity and specificity [28
]. It had been previously demonstrated that BCA2 was phosphorylated in the presence of AKT [8
]. Here, we showed that BCA2 was more stable when co-expressed with constitutively active AKT as opposed to kinase-dead AKT. The S132, 133A mutant displayed no notable change in stability. This confirmed that these serine residues are the primary sites of AKT-mediated BCA2 phosphorylation. Surrounding S132 and S133 are two 14-3-3σ binding motifs. The 14-3-3σ motif R(S/X)XpSXP [30
] is highly similar to the BCA2 protein sequence 130-RGSS
RP-135 (Figure ). 14-3-3σ is a member of the ubiquitously expressed 14-3-3 family of proteins. The seven proteins of this family are highly conserved and integral to a number of important cellular activities. Specifically, 14-3-3σ is known to be a conspicuous regulator of cell cycle checkpoints, as well as being involved in multiple and diverse cellular pathways through interactions with its copious number of known ligands [20
]. We demonstrated that phosphorylation was an integral element of the interaction between BCA2 and 14-3-3σ. In samples where the AKT domain of BCA2 was disrupted or where wild-type BCA2 was treated with AKT inhibitor, BCA2 was less apt to bind 14-3-3σ. However, since interaction between 14-3-3σ and BCA2-S132, 133A was not completely abolished, it is likely that a second functional AKT-phosphorylation site and 14-3-3σ binding site exists. We predicted this site to be in the adjacent sequence, 133-S
PS-140 (Figure ).
Similar to hHR23a, the outcome of interaction of BCA2 with 14-3-3σ was an increase in BCA2 stability. Presumptively, BCA2 is stabilized as 14-3-3σ sterically or conformationally prevents access of the RING finger domain to the target lysines (K26 and K32) in the BZF domain [7
]. More than stabilizing BCA2, this interaction decreases the protein levels of 14-3-3σ (Figure ). This effect was not seen when the same experiment was repeated with hHR23a. hHR23a levels did not decrease with the increase in BCA2 (Figure ), suggesting that 14-3-3σ but not hHR23a is a substrate of BCA2. Other E3 ligases that have been demonstrated to be stabilized by the binding of their substrate include HOS (homologue of Slimb), where presence of substrate IκBα stabilized HOS turnover [34
]. Another examples being Cdc 4 (cell division cycle 4), which binds ubiquitin to promote its own degradation in the absence of substrate, as the binding of ubiquitin and substrate are mutually exclusive events. However when substrate is present, Cdc4 is stabilized, suggesting that substrate concentration is a method of E3 regulation [35
]. Moreover, the lack of considerable stabilization by the presence of a proteasome inhibitor alone, provides basis for the argument that BCA2 may not be solely degraded in a proteasome-mediated pathway, but may be regulated by other pathways, such as the lysosomal degradation pathway, similar to c-Cbl, which auto-regulates and is degraded via the lysosome [36
Pharmacological pursuits in the area of cancer therapies have in the past been aimed at the non-specific components of the UPS. The most effective and promising of these therapeutics is Bortezomib (Velcade™), which inhibits the chymotryptic activity of the proteasome [39
]. As the components of the UPS are arranged in a pyramidal way, with few E1 enzymes at the top and multiple and specific E3 ligases at the base, they make attractive and optimal components for study and potential therapeutic targeting. In this study, BCA2 was analyzed in the context of its expression in breast cancer tissues and the relationship between the location of BCA2 or hHR23a expression and variables considered being of diagnostic and prognostic value. Of the patient attributes analyzed, only grade had a statistically significant correlation with nuclear BCA2 expression, such that a mid- level (2) Grade classification was correlated with high BCA2 expression in the nucleus. Along with previous studies which correlated overall increased BCA2 expression with ER-positive status, negative nodal status and increased survival over 5 years [2
]; it is more likely that BCA2 is indirectly involved in progression of breast cancer to highly invasive and metastatic disease, rather than a driver of the process. Moreover, when hHR23a expression was analyzed against the same attributes, it was found that hHR23a correlated with ER-positive status, a correlation which had been found previously with BCA2 in a large scale (n = 1000) [2
]. Also while not statistically significant, it is noteworthy that in a majority of samples, high hHR23a expression occurred in cases which were negative for lymph node status. This clinical variable was found to have correlated with high BCA2 in our previous studies [2
]. The small number of significant correlations in this study may be due to the relatively small sample population, as greater emphasis is placed on individual anomalies and deviations. As ligases can be of both tumor suppressive and oncogenic natures, in order to evaluate E3s or their substrates in the context of cancer therapeutics and prognostics; an understanding of the mechanisms in which they are deregulated and/or stabilized is essential.