PRL-3 expression, interacting proteins and regulation
PRL-3 mRNA was found predominantly in the skeletal muscle and at moderate levels in the heart, as shown in mouse  and human  tissues, and, in both studies, PRL-3 was also detected in other organs at lower levels. Interestingly, it was reported that the expression in the heart only occurs during development, and not in the human adult organism, as demonstrated at the mRNA and protein levels . This finding could have important implications for any potential drug discovery against PRL-3 because inhibition of PRL-3 in the adult heart could lead to cardiotoxic effects [2,20]. Furthermore, PRL-3 was found to be expressed in the developing blood vessels and pre-erythrocytes , suggesting that PRL-3 plays an important role in embryogenesis. In addition, Zeng et al.  observed that PRL-3 is present in differentiated villus epithelial cells of the small intestine in mice.
The upregulation of PRL-3 in cancer has received the most study with respect to the three PRLs, and was identified in colon [1,58], breast , gastric  and ovarian  carcinomas. In addition, high levels of PRL-3 appear to be associated with a poor prognoses and, particularly for colon cancer, high levels of PRL-3 were shown to be predictive for the development of liver metastatis . These findings are reviewed in detail in Bessette et al. . In addition, PRL-3 was reported to be elevated in oral and cervix squamous cell carcinomas [63,64]. Furthermore, it was found to be overexpressed in haematological malignancies, namely in a subset of multiple myelomas [65,66] and in acute myeloid leukaemia (AML) .
A few substrates have been suggested for PRL-3, namely ezrin [68,69], elongation factor 2 (EF-2) , keratin 8  and integrin β1 , all four of which have been reviewed , as well as stathmin  and nucleolin . Recently, we described phosphatidylinositol(4,5)bisphosphate [PI(4,5)P2] as a potential natural substrate. Although no in vivo activity against PI(4,5)P2 has yet been demonstrated, a correlation between differences of in vitro activity and phenotype in the cell migration of wild-type PRL-3 and three PRL-3 mutants was demonstrated. This correlation was only true for activity against PI(4,5)P2 and not against the unnatural substrate ortho-methylfluorescein phosphate . Of the putative substrates, direct dephosphorylation was demonstrated in the case of ezrin and PI(4,5)P2, whereas, for EF-2, keratin 8, nucleolin and stathmin, PRL-3-dependent downregulation of the phosphorylation level was shown in vivo, and integrin β1 is now considered to be indirectly affected by PRL-3 [2,73]. In independent experiments, however, the influence of PRL-3 overexpression on ezrin phosphorylation could not be confirmed, which may be a result of the use of different cell lines [70,74]. Stathmin, nucleolin and keratin 8 were shown to co-immunoprecipitate with ectopic (inactive) PRL-3, but no direct interaction with EF-2 was reported. Other direct interaction partners have been identified: integrin α1 , cadherin CDH22  and the peptidyl prolyl cis/trans isomerase FK506-binding protein 38 (FKBP38) , all discovered in yeast two-hybrid screens, and PRL-3 itself through potential oligomerization [10,33]. Most of the proposed directly interacting proteins are related to the role of PRL-3 in cell migration and invasion and are connected in some way to the plasma membrane [ezrin, PI(4,5)P2, integrin α1, CDH22] or to the cytoskeleton [keratin 8, stathmin]. Noteworthy, nucleolin is localized to the cytoplasm and nucleus and is involved in cell proliferation  and FKBP38 is a cytosolic protein that regulates PRL-3 protein levels and proteasomal degradation in MCF-7 and HCT116 cell lines .
In addition, an unbiased mass spectrometry-based approach revealed 110 potential interacting proteins when PRL-3 was used as a bait, 38 of which were considered to be of high confidence . The identified proteins have not yet been followed up by experimental validation. It is striking that none of the proposed binding partners from other studies were identified, showing how difficult it is to validate substrates and interacting proteins of PRL-3 (and phosphatases in general).
PI(4,5)P2 as a substrate for PRL-3 offers a connection to another substrate, ezrin. Ezrin forms part of the ERM (ezrin–radixin–moesin) complex, which connects the plasma membrane with the actin cytoskeleton and is implicated in tumour metastasis . Ezrin requires PI(4,5)P2 binding and Thr567 phosphorylation to become active at the plasma membrane , so that it can exert its multiple functions in cell adhesion, motility, morphogenesis and signalling pathways [79,80]. In addition to potential PI(4,5)P2 depletion by PRL-3, PRL-3 is assumed to dephosphorylate Thr567 , meaning that PRL-3 could inactivate ezrin in multiple ways. Considering that the binding of ezrin by PI(4,5)P2 is required for the phosphorylation of Thr567, the lower phosphorylation level of Thr567 could also be an indirect effect as a result of the prevention of ezrin binding to the plasma membrane . On the other hand, PRL-3 has been reported to upregulate Src kinase activity  (see also below), and Src can phosphorylate Tyr477 in ezrin, which is required for anchorage-independent growth and cell invasion in a 3D environment . Tyr477 phosphorylation was crucial for the correct localization of ezrin to submembraneous patches in the 3D culture. The influence of Thr567 phosphorylation and PI(4,5)P2 binding was not studied in this context; however, other factors aside from the latter two are important for proper activity and membrane localization of Ezrin, depending on the functional context. This shows that the activity of PRL-3 takes place in a very complex environment, which in itself remains incompletely understood.
PRL-3 is subject to complex regulatory mechanisms. It is known that PRL-3 mRNA levels do not necessarily correspond to protein levels  and that PRL-3 abundance is controlled at the transcriptional and translational levels, as well as through degradation mechanisms  (see above).
PRL-3 is a direct transcriptional p53 target gene in mouse (mouse embryonic fibroblast; MEF) and human (H1299 human lung adenocarcinoma, SK-Hep-1 hepatocellular carcinoma) cells [30,84], and ectopic expression of p53 and p73 increases PRL-3 transcription in H1299 nonsmall cell lung cancer cells . In other cancer cells, such as SNU-475, Hep3B and HeLa cells, the transcriptional level of PRL-3 did not increase upon ectopic p53 expression , suggesting that this interaction is cell type specific (although two PRL-3 introns harbour a p53 consensus sequence that can bind the p53 protein) . PRL-3 transcription is also activated by the vascular endothelial growth factor (VEGF) through the transcription factor myocyte enhancer factor 2C (MEF2C) in human umbilical vein endothelial cells (HUVEC) . MEF2C binds the promoter region of PRL-3 in vitro and in vivo, and notably, the presence of MEF2C is critical in heart and skeletal muscle where PRL-3 is abundant. This, together with the distinct expression pattern in human healthy tissues, suggests that transcription of PRL-3 could be controlled by tissue specific transcription factors . However, an equal enhancement of PRL-3 protein amounts in the presence of MEF2C was not observed. Interestingly, in PRL-3-positive nonsmall cell lung cancer cells (NSCLC), elevated levels of VEGF and its isoform VEGF-C were found, and high levels of both were correlated with micro and lymphatic vessel density , demonstrating that the expression of PRL-3 facilitates angiogenesis .
Snail is a transcription factor involved in the epithelial–mesenchymal transition (EMT). EMT is an important process during development and metastasis and, in this process, cells lose cell–cell adhesion and gain motility. Snail is known to repress the expression of E-cadherin, resulting in the disassembly of cell–cell adhesion junctions and an increase of invasiveness . The overexpression of PRL-3 was demonstrated to promote EMT, and it was suggested that the action of PRL-3 leads indirectly to the deinhibition of Snail [75,88]. Recently, however, Zheng et al.  reported that the PRL-3-encoding gene contains three potential binding sites of Snail in the promoter region, and that the transcriptional activity of the PRL-3 promoter was abolished after the mutation of one Snail binding site. Snail was suggested to regulate promoter activity and protein expression of PRL-3 in colorectal cancer cell lines, which appears to be contradictory to the earlier reports. Thus, the interaction between PRL-3 and Snail requires further investigation.
Recently, Jiang et al.  reported that PRL-3 is a direct regulatory target of transforming growth factor (TGF)β signalling in colon cancer metastasis. TGFβ signalling suppresses the metastasis of colon cancer cells potentially by inducing stress-induced apoptosis. It was demonstrated that TGFβ signalling inhibited the expression of PRL-3 in a mothers against decapentaplegic homologue (Smad) 3-dependent manner. Because a loss of TGFβ signalling occurs in 30–50% of colon cancers, this could be a feasible mechanism for explaining PRL-3 upregulation in colon cancer .
A translational regulator of PRL-3 is polyC-RNA-binding protein 1 (PCBP1) . PCBP1 overexpression inhibited PRL-3 expression via interaction with a GC-rich motif at the 5′ UTR of PRL-3 mRNA. In clinical samples of normal and cancerous epithelia, an inverse correlation between protein levels of PRL-3 and PCBP1 was observed, and knockdown of endogenous PCBP1 in HCT-116 cells inhibited tumourigenesis in mice, indicating that PCBP1 acts as a tumour suppressor in vivo .
Signalling pathways affected by PRL-3
The signalling pathways affected by PRL-3 have been reviewed by Bessette et al.  and Al-Aidaroos and Zeng . Therefore, we only briefly describe the key signalling effects of PRL-3 and add data that have appeared subsequent to these reviews.
By demonstrating that PRL-3 upregulates mesenchymal markers and downregulates epithelial markers, it was shown that PRL-3 promotes EMT [75,88]. It promotes EMT and cell survival by acting upstream of phosphatidylinositol 3-kinase (PI3K) [74,89]. PI3K signalling promotes many processes, such as cell survival, cell proliferation or cell motility, and PI3K is an oncogene [90,91]. PRL-3 was reported to post-transcriptionally downregulate PTEN protein levels . PTEN counteracts PI3K activity by converting phosphatidylinositol triphosphate PI(3,4,5)P3 into PI(4,5)P2; thus, its downregulation leads to the activation of PI3K signalling. In addition, PRL-3-mediated activation of PI3K could relieve the inhibition of the mesenchymal marker Snail (see above) by inhibition of glycogen synthase kinase (GSK)-3β [75,88]. Furthermore, PRL-3 was reported to promote cell survival under growth factor deprivation stress by activating and maintaining the activity of the PI3K/Akt pathway .
PRL-3 was suggested to either reduce the number of focal adhesions and/or increase focal adhesion turnover to mediate cell invasion and motility . Focal adhesion complexes are multi-component sites where integrins mediate the contact between the cell and the extracellular matrix . Levels of PI(4,5)P2 at the cell membrane are crucial for regulating the dynamics of focal adhesion complexes , and focal adhesion kinase (FAK) is a key component of focal adhesion complexes . FAK integrates external signals to promote cell motility via many different pathways involving the regulation of (or interaction with) proteins such as cadherins, Src, p130Cas, Rho-family GTPases and ezrin , many of which were shown to be affected by PRL-3. Integrin α1 and cadherin-22 were reported to be direct interactors of PRL-3 (see above), E-cadherin was shown to be downregulated by PRL-3 [75,88] and PRL-3 signalled via integrin β1 in LoVo colon cancer cells leading to extracellular signal-regulated kinase (ERK)1/2 activation . Src kinase was activated by PRL-3 via translational downregulation of C-terminal Src kinase (Csk), which is a negative regulator of Src [81,94]. Src activation by PRL-3 led to the phosphorylation of downstream proteins such as signal transducer and activator of transcription (STAT) 3 and p130CAS, and, in agreement with Peng et al. , ERK1/2. In further studies, PRL-3 activated RhoC, downregulated Rac-GTP [28,88] and had no effect on Cdc42 . RhoA activity was reduced by PRL-3 overexpression in the earlier study  and enhanced in the later study . These findings demonstrate that the Rho family of GTPases act downstream of PRL-3, and also show that the regulation is complex. An interesting context in this regard is that active ezrin recruits both positive and negative regulators of the Rho family of GTPases . Upon inactivation of ezrin by PRL-3, these regulators could be released, which would contribute to maintaining the active form of the Rho GTPases and may explain the activation of RhoA and RhoC when overexpressing PRL-3 .
The activity of PRL-3 against PI(4,5)P2  offers the intriguing possibility of PRL-3 regulating all of the noted proteins upstream of FAK. This regulation, however, is very complex and highly dynamic, with the activation of Src and Rho GTPases on the one hand and deactivation of ezrin and Rho GTPases on the other. With PRL-3 being membrane bound and PI(4,5)P2 being the highest abundant phosphoinositide and a crucial part of the membrane in many respects [92,96,97], this regulation essentially needs to be highly dynamic and tightly regulated. Nevertheless, this interaction would destabilize focal adhesions and could regulate focal adhesion turnover, leading to enhanced motility and invasiveness. Further studies are necessary to evaluate this hypothesis.
Expression and activity of matrix metalloproteinases (MMP) is affected by PRLs. MMPs are extracellular secreted proteins with a key function in tumour metastasis . Increased MMP2 (but not MMP9) activity and expression levels have been found in PRL-3 stably transfected LoVo cells . PRL-3-induced invasion in these cells was dependent on MMP2 upregulation and ERK1/2 activation. PRL-3 also downregulated the expression of the MMP2 inhibitor TIMP2, explaining, at least in part, the activation of MMP2. Recently, Lee et al.  investigated expression levels of several MMPs in PRL-3-overexpressing colorectal DLD-1 cells. MMP2 was also found to be enhanced, and MMP2 knockdown partially inhibited cell migration and invasion. In addition, migration and invasion of DLD-1-PRL3 cells was completely inhibited by small interfering RNA (siRNA) knockdown of MMP7, whereas the overexpression of MMP-7 increased migration. In agreement with earlier studies, PRL-3 acted through oncogenic pathways including PI3K/Akt and ERK1/2 .
A recent study revealed that intermediate-conductance Ca2+-activated K+ (KCNN4) channels were upregulated in ectopically PRL-3 expressing LoVo cells, and this upregulation was nuclear factor-κB (NF-κB)-dependent, revealing a novel pathway that PRL-3 can interfere with. Blocking of KCNN4 channels inhibited PRL-3-induced cell proliferation and arrested the cell cycle at the G2/M phase, indirectly suggesting that PRL-3 facilitates G2/M transition in this setting .
PRL-3 was also described to play a role in cell cycle regulation in normal cells . Tight control of PRL-3 basal expression in MEF cells appears to be important to ensure cell cycle progression by facilitating G1/S transition (as opposed to G2/M transition in LoVo cancer cells). Its overexpression in MEF cells led to G1 arrest downstream of p53 via a PI3K-Akt-mediated negative feedback loop, in which initial levels of PRL-3 activated the PI3K-Akt pathway but subsequent higher levels of PRL-3 correlated with a decrease in activated Akt. A decrease of PRL-3 expression levels also led to cell cycle arrest through increased p53 expression and via cyclin-dependent kinase 2 (CDK2), relying on an intact p53 pathway. Interestingly, in the global study by Ewing et al. , CDK2 was found to be a PRL-3 interacting protein.
These results appear to contradict the role of PRL-3 in cancer; however, Basak et al.  suggested that, as a result of multiple mutations in cancer cells, particularly in later (metastatic) stages, and with p53 loss of function being a very common mutation, high expression levels of PRL-3 might not succeed in inducing cell cycle arrest, and other functions of PRL-3 might prevail. It is now important to dissect the primary role of PRL-3 in healthy cells, whether it is related to cell migration, cell cycle regulation or both, and whether the activity in cancer is a malfunction or hyperactivity of a normal function. Min et al.  addressed the ability of PRL-3 to regulate p53 in cancer cells. In agreement with the results obtained in MEF cells, PRL-3 upregulation in HCT116 colorectal cancer cells led to a decrease in p53 expression; however, it did not lead to cell cycle arrest but to inhibition of p53-mediated apoptosis . An earlier, more detailed study on PRL-1 investigated the mechanism of action of PRL-1 on p53 , and this is discussed below. PRL-3 was described to act on p53 through the same mechanisms involving mouse double minute 2 (MDM2) stabilization via PI3K/Akt signalling and also increased transcription of protein with a RING-H2 domain (PIRH2), both leading to p53 inactivation .
PRL-3 appears to play a role in drug resistance in AML. PRL-3 was found at elevated levels in AML patients and, in six out of nine patient samples, the overexpression was correlated with internal tandem repeat duplication of fms-like tyrosine kinase 3 (FLT3-ITD), a mutation that occurs in approximately 25% of AML patients. Zhou et al.  reported that, in AML MOLM-14-cells, PRL-3 acts downstream of FLT3-ITD through STAT5 and STAT3 (but not through Akt) activation and upregulation of McI-1, which is known to contribute to a resistance to chemotherapy when it is highly abundant. In addition, PRL-3 was shown to bind histone deacetylase 4 in MOLM-14 cell lysate .