Some breast cancer cells acquire metastatic potential in a very early stage, which gains the capability to spread from the breast tissue, enter the blood or lymphatic vessels, and disseminate to distal organs, preferentially to the lung, liver, brain, and bone. Metastasis represents the multistep cascade of cancer cell alterations accompanied by structural and functional changes. It is well recognized that distant metastasis colonization consists of sequential steps ( Figure 1 ) (2, 5), including cell detachment from the primary tumor site involving epithelial–mesenchymal transition (EMT) (6); migration and invasion into surrounding tissue; penetration of the basal membrane (trans-endothelial intravasation) into the vasculature of blood and/or lymphatic vessels to be circulating tumor cells (CTCs); extravasation of CTCs to secondary sites as disseminated tumor cells (DTCs) (7); dissemination to distant organs; and formation of a micro-metastatic niche and construction of macrometastases. In addition, to survive and initiate the secondary cancer foci, cells need the capability of evading immune defenses, delivering to distant sites, adapting to supportive niches such as angiogenesis (8), and inducting retro-differentiation to gain stemness (9).

Ubiquitination is a post-translational modification of proteins, which is essential for nearly all biological processes, including cell growth, autophagy, apoptosis, and differentiation. It is a three-step enzymatic cascade: i) A ubiquitin-activating enzyme (E1) mediates the activation of the carboxyl-terminal glycine residue of ubiquitin in an ATP-dependent manner. ii) The activated ubiquitin is then transferred to E1 followed by the transfer of ubiquitin to a thiolester of a ubiquitin-conjugating enzyme (E2) to format the thiolester linkage. iii) A ubiquitin protein ligase (E3) confers substrate specificity by recognizing the target proteins and mediating the conjugation of ubiquitin molecules to a lysine residue on the targeted protein via an isopeptide bond (10). E3’s ability to specifically recognize and target substrates makes it a key regulator in the ubiquitin process.

The fate of ubiquitinated proteins is dependent on the different types of ubiquitin linkage. It has been characterized as mono-ubiquitination or poly-ubiquitinated chains. Mono-ubiquitination involves the transfer of a single ubiquitin to a substrate. E3 ligases can also connect several ubiquitin molecules together using the C-terminus of one subunit and one of the seven lysine (K) residues (K6, K11, K27, K29, K33, K48, K63) or the N-terminal methionine (M1) on the other, to form a homotypic or branched ubiquitin chain (11). The recruitment of polyubiquitinated proteins to the proteasome is a classic protein turnover pathway ( Figure 2A ). Ubiquitin-dependent proteasomal degradation involves the polyubiquitination of substrate catalyzed by E1, E2, and especially E3 (12, 13). Subsequently, polymers linked through K11 or K48 (14, 15) trigger substrate protein degradation by the ATP-dependent 26S proteasome (16).

Besides, different ubiquitin topologies adopt distinct structural conformations. Signals can be transduced by different types of ubiquitin modification, including mono-ubiquitination, multi-monoubiquitination, homotypic ubiquitin chains, and heterotypic ubiquitin chains, which send various ‘codes’ to precisely exert degradative and non-degradative functions including modification of protein trafficking, interaction, and signal transduction (17, 18). For example, linkages of homotypic K11 and K48 can drive proteasomal degradation; K27 linkages have been implicated in regulating DNA repair (19) and autoimmunity (20); K33 linkages were proposed to regulate trafficking; and mono-ubiquitination can prevent protein interactions (21). SMAD4 (SMAD family member 4) engages its signal partner SMAD2 (SMAD family member 2) after removing its own mono-ubiquitination (18). In addition, the mono-ubiquitination of histone H2A promotes transcriptional silencing (22), while the mono-ubiquitination of histone H2B mediates transcriptional elongation (23).

The canonical E3s are classified into two types: Homologous to the E6AP Carboxyl Terminus (HECT) family (24) and Really Interesting New Gene (RING) finger family (25). HECT E3 ligases were named accordingly to identify protein E6AP (E6-associated protein), and they are characterized by a conserved C-terminal ~350 aa HECT domain and various N-terminal substrate-binding domains (26). HECT E3 ligases mediate a two-step ubiquitin transfer process, in which ubiquitin is firstly transferred from the E2-Ub intermediate to the E3 active cysteine residue before transferring to the substrate lysine residue ( Figure 2B ) (27). In contrast, The RING core is a small domain of 40–70 residues. A cross-braced pattern of conserved cysteine and histidine residues coordinating two zinc ions maintains the native fold. The RING domain brings the E2∼Ub conjugate into the proximity of the substrate bound via substrate-recognition domains ( Figure 2B ). RING E3s are further divided into two subtypes: single RING and multi-unit RING family. The Cullin-RING ubiquitin ligase (CRL) family is composed of a multi-unit. CRLs utilize Cullin proteins as a central scaffold which binds to a RING-box protein (Rbx) and an adaptor protein–substrate receptor complex through its C- and N-termini. Most recognized CRLs are known as the SCF (Skp-Cullin-F-box) complex, in which Cullin interacts with Skp (S-phase kinase associated protein) proteins and utilizes various F-box proteins, to recruit substrates and initiate ubiquitin ligation (28). Besides, U-box E3s are also categorized as RING-type E3s, but their molecular structure subtly differs in that zinc-bound sites are replaced by a hydrophobic core (29) ( Figure 2B ).

E3s target a broad spectrum of substrates. In cancer, the aberrant functions of E3s are linked to deregulated oncoproteins or tumor suppressors and affect the biological behavior of cells. Many E3s have been reported to be associated with breast cancer metastasis over the past few decades. In this review, we provided an overview of E3 ubiquitin ligases that have been found to be deregulated in breast cancer metastasis and summarized the multistep cascade of breast cancer cell alterations in metastasis, in which these E3 ubiquitin ligases are involved. Furthermore, we classified E3s based on their structures and analyzed the correlation of E3s with the survival of breast cancer patients. Finally, we considered their potency to the therapeutic intervention or prognostic assessment of metastatic breast cancer.

The literatures involved in this study were searched from the databases PubMed and Medline (last search updated on January 1st, 2021). The key words used in the searching were “E3 ubiquitin”, “breast cancer” and “Metastasis” or “dissemination”/“EMT”/“invasion”/“migration”/“intravasation”/“CTC”. All the searching results were imported in the Endnote software to eliminate duplicates.

By scanning the titles and abstracts and further reading the full text, the irrelevant and retracted papers were excluded. The reference list of all selected articles was scanned to identify potentially relevant reports. The search results followed these including and excluding criteria:

Including criteria: 1) research limits to the E3 ubiquitin ligases; 2) research related to human breast cancer; 3) research focused on metastasis or steps in the metastasis mechanisms of breast cancer; 4) studies provided sufficient experimental evidence or clinical data to support thesis; 5) peer-reviewed and formally published original literatures

Excluding criteria: 1) research of the regulation of E3 ubiquitin ligases; 2) research of ubiquitin-like modifiers including SUMOylation; 3) retracted articles; 4) reviews or letter to editors.

The above searches were performed and reviewed by two authors independently. The following items were recorded from each study: E3 name, gene ID, substrate, role in breast cancer metastasis, cellular function, signal pathway, verification by cell biological experiments, breast cancer cell lines, and clinical significance (expression difference and survival analysis). These records were cross-checked and double- checked by another author.

The association between the specific E3s’ ubiquitin ligase expression and survival in breast cancer was first analyzed using the PrognoScan database (30) as in previously mentioned methods (31). Then, we used a Kaplan–Meier plotter (32) to validate and illustrate as a Kaplan–Meier plot, in which the distant metastasis–free survival (DMFS) curves for high (red) and low (black) expression groups dichotomized at the optimal cut-point were plotted. The logrank P-value and the hazard ratio with 95% confidence intervals were calculated. The threshold was adjusted to logrank P‐values at <0.05.

As shown in the brief flow chart ( Figure 3 ), the records on E3 ubiquitin ligases involved in breast cancer metastasis were screened from both the PubMed (N = 199) and Medline (N = 103) databases. After duplicate elimination, 202 records remained. By reading the titles and abstracts of these records, we excluded 91 literatures for the following reasons: studies in other cancer types (N = 23); irrelevant to metastasis (N = 28) or irrelevant to E3 ubiquitin ligase (N = 12); focused on the regulation of E3 ubiquitin ligase (N = 17) or de-ubiquitination (N = 3); and reviews (N = 4) and retracted articles (N = 4). Other studies had been validated by reading the full texts. Finally, a total of 111 literatures were included in our analysis.

We further extracted key information from these literatures, including E3 name, gene ID, substrate, roles in breast cancer metastasis, cellular function (experimentally verified), signal pathway, breast cancer cell lines, and clinical significance. This detailed information of 54 E3 ubiquitin ligases/ligase complexes in total have been summarized in Table 1 . By their molecular roles’ participation in the multiple steps of breast cancer metastasis, we briefly presented them as follows.

Ubiquitination is a ubiquitous form of post-translational modification of proteins. In this process, E3s specifically bind to the substrate proteins, mediate the ligation of ubiquitin molecules and affect the specific proteins' turnover and functions. Thus, it is important to clarify the mechanism in target drug research and development (129). Do the mechanisms of these identified breast cancer metastasis E3 ligases have commonalities?

We classified the previously summarized E3 ligases of classical families based on their structures. The results suggested that the E3s involved in breast cancer metastasis belong to diversified classes, such as the HECT family, RING family, and U-box family ( Figure 2B ). There are eight E3 ligases that belong to the HECT family and are further classified into three subfamilies, including SMAD-specific E3 ubiquitin protein ligase 1 (Smurf1), SMAD-specific E3 ubiquitin protein ligase 2 (Smurf2), itchy E3 ubiquitin protein ligase (Itch) and WW domain–containing E3 ubiquitin protein ligase 1 (WWP1), NEDD4, belonging to the NEDD4 family, which contains a WW domain, C2 domain, and HECT domain; HECT and RLD domain which contains E3 ubiquitin protein ligases 2 and 4 (HERC2 and HERC4), as the HERC family, which contains the common RCC1-like domain (RLD) and HECT domain; and UBR5, as the “other” family, which contains a UBA domain, zinc finger domain, and HECT domain.

RING E3 ligases are further divided into two subtypes: single RING and multi-unit RING family. Several E3 ligases belong to single RING type (listed in Figure 2B , single RING family), performing a single-step ubiquitin transfer from the E2-Ub to the substrate, which work as allosteric activators. As for multi-subunit RING type, The Cullin-RING ubiquitin ligase (CRL) family is composed of a multi-unit. Most recognized CRLs are known as the SCF (SKP-Cullin-F-box) complex, in which Cullin1 interacts with Skp1 or Skp2 and utilizes various F-box proteins (shown in Figure 2B , F-box proteins) to recruit substrates and initiate ubiquitin ligation. Cullin7 also forms a SCF complex with SKP1 and F-box and WD repeat domain containing 8 (FBXW8). Cullin3-RBX1 (Ring-box1) E3 ubiquitin ligase complex requires BTB (Bric-a-brac-Tramtrack-Broad complex) domain protein as an adaptor. Two BTB proteins, speckle-type BTB/POZ protein (SPOP) and K⁺ channel tetramerization domain 5 (KCTD5), forms a complex with Cullin3, respectively.

Three E3 ubiquitin ligases, CHIP, NRBE3, and peptidylprolyl isomerase like 2 (PPIL2) belong to the U-box family, which contains a U-box domain. U-box E3s are also categorized as RING-type E3s, but their molecular structure subtly differs in that zinc-bound sites are replaced by a hydrophobic core (29).

There are still many E3 ligases that cannot be characterized into classical types. Ankyrin repeat and SOCS box containing 13 (ASB13) belongs to the ankyrin repeat and suppressor of cytokine signaling (SOCS) box (Asb), which contains six-ankyrin repeat domain (34, 130). The UBR-box is a 70-residue zinc finger domain present in the UBR family of E3 ubiquitin ligases. Unlike UBR5, which also contains an HECT domain, the structures responsible for UBR7 executing its E3 role need to be verified by in vitro ubiquitin assays (113, 131, 132). Intriguingly, UBE2O, an E2/E3 hybrid ubiquitin-protein ligase, displays both E2 ubiquitin conjugating enzyme and E3 ubiquitin ligase activities (116, 133). Compared to classical types, the specific catalytic mechanism of several non-classical E3s will need more efforts to be figured out in the future.

Since the E3s are involved in breast cancer metastasis, are they associated with patients’ survival? Through the analysis of literatures, we found that out of the 54 E3 ligases, 31 have been reported to be related to the survival of patients ( Table 2 , from “ASB13” to “UBE2O”). The other 23 E3 ubiquitin ligases lack clinical data to determine whether they have clinical significance. Although some of them have verified the specific roles involved in metastasis by cellular experiments and animal models, these E3 ligases’ relationships with breast cancer patients’ survival need to be elucidated. Therefore, we used publicly available clinical data to conduct survival analysis (only in which the literature were claimed to have been verified by molecular and animal experiments).

By comparing with molecular mechanism, five E3 ligases have been found to be significantly correlated with the DMFS of breast cancer patients. As illustrated in Kaplan–Meier plots ( Figure 4 ), we conducted the survival analysis of Arkadia ( Figure 4A ), NKLAM ( Figure 4B ), PPIL2 ( Figure 4C ), Smurf1 ( Figure 4D ), and Smurf2 ( Figure 4E ), which is consistent with the laboratory mechanism previously found. Patients with high expression of Arkadia (Hazard Ratio: 0.73; Logrank P: 0.00046), NKLAM (Hazard Ratio: 0.77; Logrank P: 0.00086), and PPIL2 (Hazard Ratio: 0.78; Logrank P: 0.0085), which have an inhibitory effect on metastasis, have higher survival probability. Patients with high expression of Smurf1 (Hazard Ratio: 1.34; Logrank P: 0.0022) and Smurf2 (Hazard Ratio: 1.52; Logrank P: 1.9e-07) have lower survival probability. Combining this analysis results with summarized literature reports, E3 ubiquitin ligases which are significantly correlated with patients’ survival were presented in Table 2 .

Interestingly, we found that some results of survival analysis were contrary to the molecular mechanism. For example, if all subtypes of breast patients were considered together, a higher expression of MDM2 showed better survival (Hazard Ratio: 0.81; Logrank P: 0.015) ( Figure 4F ), which was obviously contrary to known molecular mechanisms. MDM2 has been well proven to be an E3 ubiquitin ligase that simultaneously targets tumor suppressor protein RB and P53 to degradation. We further analyzed different subtypes and found that it was opposite in patients with luminal A (Hazard Ratio: 1.5; Logrank P: 0.0041) ( Figure 4G ) and basal subtype (Hazard Ratio: 0.69; Logrank P: 0.019) ( Figure 4H ). This suggests that the heterogeneity of breast cancer cannot be ignored. In addition, it might be affected by the expression level or mutation of ubiquitin substrates, such as the status of p53 mutation or loss of RB. The context-dependent role of E3s and breast cancer subtypes need to be considered more in future survival analyses.