Project Topics
Get Complete Project Material File(s) Now! »
The SUMO protein
Human Small Ubiquitin-like Modifiers (SUMOs) are ~10kDa proteins that have a three-dimensional structure close to Ubiquitin, a protein used in one of the most studied protein modification system that will be described later (Figure 16). Studies found different isoforms for SUMO: SUMO1 (also known as Smt3c, PIC1, GMP1, Sentrin or Ubl1), SUMO2 (also known as Smt3a or Sentrin3), SUMO3 (also known as Smt3b or Sentrin2) and SUMO4. Ubiquitin and SUMO only share around 20% sequence identity at the protein level and display different charge distribution. All SUMO isoforms have an unstructured stretch of 10-25 amino acids at their N-terminus that is not found in any other ubiquitin related protein. It is probably used for the formation of SUMO chains (Tatham et al. 2001) (Figure 16). SUMO proteins are expressed in all eukaryotes and, in vertebrates, all SUMO isoforms are expressed in all tissues except for SUMO4 who is mainly expressed in the kidney, lymph node and spleen (Guo et al. 2004). However, it is not clear whether SUMO4 is present at the active protein level in-vivo especially since it can not be conjugated (Sinha et al. 2016; Geiss-Friedlander & Melchior 2007; Owerbach et al. 2005). Very recently, this protein family got a little bit bigger with the discovery of SUMO5, an isoform highly homologous to SUMO1, essential for PML NB formation and stability through its conjugation on PML at K160 (Liang et al. 2016).
Enzymatic cascade involved in SUMOylation
SUMOylation results in the formation of an isopeptide bond linking the C-terminal glycine (Gly, G) residue of the modifier protein and the lysine (Lys, K) residue of the acceptor protein. This process involves an enzymatic cascade comprised of three classes of enzymes that are very well conserved and unique to this pathway (Geiss-Friedlander & Melchior 2007). In the first step, SUMO precursor protein is processed by cysteine-specific SUMO proteases, ULPs in yeast but called SENPs (SENtrin specific Proteases) in mammals. This exposes the di-glycine motif mentioned earlier which will then be linked to the unique E1 activating heterodimer enzyme AOS1–UBA2, also known as SAE1-SAE2 dimer. The E1 will catalyze the covalent attachment of SUMO to a reactive cysteine (Cys, C) residue in SAE2 through an ATP-dependent thioesterification reaction. Through a thioester linkage, SUMO is then transferred to the cysteine residue of the unique SUMO E2 conjugating enzyme: UBC9. In vitro, the E2 enzyme is sufficient for conjugating SUMO to a lysine residue on the substrate. However, it is likely receiving help from E3 SUMO ligases in vivo. E3 ligases can serve as scaffold proteins that will bring the SUMO-charged UBC9 and the substrate in close proximity providing in the process the efficiency and specificity to the SUMOylation reaction or just stimulating the E2 enzyme (Enserink 2015; Geiss-Friedlander & Melchior 2007) (Figure 17) . There are relatively few E3 ligases identified, nine to date, the best-known class being the Protein Inhibitor of Activated STAT protein (PIAS) family. For example, PIASy was reported to interact with p53 protein and to be involved in the regulation of cellular senescence and apoptosis (Bischof et al. 2006; Nelson et al. 2001). RANBP2 is also a very well-known E3 ligase targeting SP100 and involved in nuclear import (Pichler et al. 2002). E3 ligases are very interesting to study since they provide substrate specificity to the pathway making them good candidates for drug development. An important aspect of SUMOylation is that it is a very dynamic and reversible process.
SUMOylated proteins can be deSUMOylated through SENPs activity, the same enzymes used for SUMO maturation (Figure 17). These enzymes have an important functional role in the turnover and spatial regulation of SUMO (Mukhopadhyay & Dasso 2007) and are essential for many cellular processes like chromosome cohesion, mitosis or transcription (Enserink 2015).
Consensus motifs for SUMOylation
SUMOylation of substrates usually occurs on lysine residues in a canonical SUMO consensus motif ΨKx(D/E), in which Ψ is a large hydrophobic residue and x any amino acid followed by an acidic residue (Rodriguez et al. 2001). The hydrophobic and acidic residues promote stability of the interaction between the substrate and the E2 conjugating enzyme (Lin et al. 2002; Enserink 2015). Different variations of this motif have been identified including Negatively charged amino acid-Dependent SUMO Motifs (NDSMs) and Phosphorylation Dependent SUMO Motif (PDSMs). This last motif is an extended canonical motif in which phosphorylation, by proline-directed kinases, increase SUMOylation efficiency. PDSM and NDSM probably increase SUMOylation efficiency by increasing the stability of the interaction between UBC9 and the substrate because of the negatively charged amino acid (NDSM) or phosphate (PDSM) which will interact with charged amino acids of UBC9 (S.-H. Yang et al. 2006; Yang & Grégoire 2006). It is also important to note that although there is a canonical consensus motif, non-consensus SUMOylation sites are a relatively common event (Enserink 2015).
Chain formation and SIM
SUMO is capable to form polymeric chains (only with SUMO isoforms 2 and 3) through its K11 site found in a canonical SUMO consensus motif. However SUMO1 does not contain K11 and is conjugated to its substrate once or at the end of a poly-SUMO chain (Hay 2005; Tatham et al. 2001). These chains were mostly studied and characterized because of their role as an indirect degradation signal. SUMO chains can recruit conserved enzymes known as SUMO Targeted Ubiquitin ligases (STUbls). Theses E3 ubiquitin ligases can then ubiquitinate polySUMOylated substrates leading them to proteasomal degradation as it is the case for PML under arsenic trioxide induced stress (Enserink 2015; Lallemand-Breitenbach et al. 2008).
As shown, SUMO can be used as signal relay for protein degradation but most importantly, it is creating a new interface for protein-protein interactions through a particular motif called the SUMO Interacting Motif (SIM). NMR spectroscopic characterization of the interaction of SUMO and peptides derived from known substrates identified a hydrophobic core with the following consensus to identify SIM: [V/I]x[V/I] [V/I] (Song et al. 2004). Following studies confirmed the essential role of this hydrophobic core and showed that a hydrophobic pocket on SUMO was interacting with hydrophobic side chains of the SIM (Song et al. 2005). This site is also often flanked by acidic amino acid or in some cases phosphorylated residues that will interact with the lysine residue of the SUMO protein to stabilize the interaction (Stehmeier & Muller 2009). This also gives a way for the cell to control protein SUMOylation spatially and temporally as phosphorylated SIM might add specificity for appropriate substrates (Enserink 2015). SUMO can also serve as glue in a complex, stabilizing it through a cooperation of multiple weak SUMO-SIM interactions that would significantly increasing stability (Enserink 2015; Psakhye & Jentsch 2012).
SUMO plays an important role under normal conditions by maintaining cell homeostasis, promoting cell growth and proliferation whereas under stress conditions such as heat shock, DNA damage, oxidative stress but also viral infections, it is a key component in the cellular response by activating pro-survival pathways (Saitoh & Hinchey 2000; Enserink 2015).
PML SUMOylation and SIMs
Ran GTPase-activating protein 1 (RanGAP1) was the first SUMO target identified (Matunis et al. 1996) and was closely followed by another heavily SUMOylated protein that is PML. PML possess one SUMO-Interacting Motif (SIM) at the C-terminus of its sequence (Shen et al. 2006) and is also heavily SUMOylated by SUMO1 and SUMO2/3 (Cheng & Kao 2012). Three main SUMOylation sites on PML were identified at K65, K160 and K490 (Kamitani et al. 1998) (Table 2). Three more sites were identified for poly-SUMOylation under arsenic trioxide treatment at K380, K400 and K497 (Galisson et al. 2011). Two more potential polySUMO sites were discovered through quantitated proteomics at K226 and K616 but further confirmation is needed (Vertegaal et al. 2006) (Table2).
Regulation of PML through the SUMO pathway
PML SUMOylation seems to be dependent on the cell cycle: its SUMOylation is elevated during interphase and declines during mitosis (Everett et al. 1999). SUMOylation of PML can also be triggered through DNA damage induced chemically, for example with Adriamycin, a chemotherapeutic agent (Gresko et al. 2009). PML NBs regulate transcription through sequestration or dissociation of transcription factors. Therefore, PML SUMOylation might have both direct and indirect effects on transcriptional regulation. One example of such indirect regulation is the release from PML NBs of Signal Transducer and Activator of Transcription 3 (STAT3), due to PML deSUMOylation by SENP1 previously activated through Interleukin-6 treatment (Kawasaki et al. 2003; Ohbayashi et al. 2008). PML SUMOylation also plays a role in apoptosis regulation through PML NBs interaction with DAXX or P53 proteins (Meinecke et al. 2007; Cheng & Kao 2012).
SUMOylation of PML is controlled in part by the E3 SUMO ligases targeting it. The first identified PML E3 SUMO ligase was RAN Binding Protein 2 (RanBP2) which mediated the K490 SUMOylation, essential for PML NBs maintenance (Tatham et al. 2005; Satow et al. 2012; Saitoh et al. 2006). Another E3 ligase recently discovered for PML is Protein Inhibitor of Activated STAT1 (PIAS1) that would be responsible for the SUMOylation of K65 and K160 residues. These two sites also appear to influence Casein Kinase 2 activity on PML leading to S565 phosphorylation and subsequent degradation of PML (Rabellino et al. 2012). Histone Deacetylate 7 (HDAC7) was proposed as an E3 ligase for PML since its presence is required to keep PML SUMOylated, but there is no evidence so far that it is acting directly as a SUMO E3 ligase (Gao, Ho, et al. 2008). Recently a new SUMO2/3 E3 ligase for PML was discovered, ZNF451-1, modifying PML at established SUMOylation sites (K65/K160/K490) and involved in its stability (Koidl et al. 2016). A more indirect approach can also alter PML SUMOylation status: for example Beta-catenin was shown to prevent RanBP2 from interacting with PML thus preventing its SUMOylation (Satow et al. 2012). Finally another way to prevent PML from interacting with its E3 ligases would be by restricting the localization of the substrate or the enzymes involved (Cheng & Kao 2012). Another way to regulate the SUMOylation of PML is through the involvement of specific deSUMOylases (SENPs). In Humans, there are six SENPs, SENP-1, -2,-3,-5 and -6. The nuclear SENP1 was shown to specifically remove SUMO1 from PML (Gong et al. 2000). SENP2 isoform (SuPr-1) has also been involved on PML deSUMOylation, causing c-Jun, transactivation (Best et al. 2002). PML PolySUMO chains formed by SUMO2/3 are removed by SENP3 under mild oxidative stress causing PML NBs disruption and stimulating cell proliferation (Han et al. 2010). Just as SENP3, SENP5 deconjugates SUMO2/3 at the SUMOylation sites K160 and K490 as well as all SUMO isoforms present on K65 (Gong & Yeh 2006). SENP6 was reported to specifically remove SUMO2/3 and a loss of this SENP causes an increase of cell death as well as an increase in PML NBs (Hattersley et al. 2011; Mukhopadhyay et al. 2006). In line with MEFs experiment, removing SENP increases PML NBs, whereas knocking out SUMO1 in MEFs causes a decrease. It is rather obvious that SENPs, just as E3 ligases, are involved in PML SUMOylation dynamics however, it is still poorly understood under which conditions each component play a role and how they are coordinated (Cheng & Kao 2012) (Figure 18). Besides SENPs, other deSUMOylases exist such as Ubiquitin-specific peptidase-like protein 1 (USPL1) (Schulz et al. 2012).
The SKP-Cullin-F-Box containing Complex (SCF)
Substrate specific degradation is a key component of the Ubiquitin Proteasome System (UPS), it governs very diverse cellular processes such as cell cycle progression, apoptosis, transcription or cell proliferation. Therefore, it is not surprising that the human genome encodes for more than 300 proteins containing a RING domain, each targeting proteins more or less specifically. In order to add another layer of selectivity in this process, multi-unit complexes can be formed in which RING ubiquitin ligase associates with other components that will bring substrate specificity to the newly formed complex.
General structure of Cullin-RING Ligases complex
The best studied example of Ubiquitin ligase complex is Cullin-Ring Ligases (CRLs). These complexes are composed of a Cullin-family protein that serves as a scaffold by interacting directly with the RING domain of the enzyme, through a protein-protein interaction domain on the C-terminus of the protein. In mammals, there are eight different Cullins able to make complexes; some of these can be interchangeable leading to the formation of many different complexes (Petroski & Deshaies 2005; Lee & Diehl 2013). Almost all Cullin proteins including CUL-1, -2, -3, -4a, -4b, -7 and -9, bind the E2 ubiquitin conjugation enzyme (UBER1) through the small RING-box protein 1 (RBX1 also known as ROC1) except for CUL5 which is only recruiting RBX2 (Lydeard et al. 2013) (Figure 27). Cullin-repeat motifs situated at the N-terminus of the Cullin protein allow for a large number of adaptor proteins to interact causing the assembly of more than 200 CRL complexes using the eight Cullin scaffold proteins (Skaar et al. 2014). These adaptor proteins include S phase Kinase-associated Protein 1 (SKP1) for CRLs 1 and 7, Elongin B and C for CRLs 2 and 5, DNA Damage-Binding protein 1 (DDB1) for CRLs 4A and 4B and finally, CRLs 3 and 9 do not use adaptor proteins or are unknown (Figure 27). The purpose of adaptors is to recruit proteins that will be able to specifically recognize a substrate and efficiently recruit it to the complex for its ubiquitination. These substrates vary widely from one complex to the other, and depending on the adaptor protein recruiting it (Figure 27). The best prototype for these CRLs is the Cul1-containing complex also known as the SCF ligase (Lee & Diehl 2013).
Table of contents :
Acknowledgements
Collaborations
Table of Contents
Abbreviations
Introduction
I) Discovery of ProMyelocytic Leukemia protein (PML)
1) Acute Promyelocytic Leukemia (APL)
2) The PML protein
3) PML Nuclear Bodies
II) Post-translational modifications
1) Diversity of post-translational modifications
2) Phosphorylation
3) Acetylation
4) SUMOylation
5) The SUMO/Ubiquitin coupled pathway
6) Ubiquitination
III) The SKP-Cullin-F-Box containing Complex (SCF)
1) General structure of Cullin-RING Ligases complex
2) The SCF complex and the F-Box proteins
3) Regulation of the SCF complex
4) Substrate recognition
5) SCFs and diseases
6) PML and SCF
IV) PML, a tumor suppressor
1) PML physiological functions
2) PML and diseases
3) PML and the apoptotic pathway
4) PML and the P53 pathway
5) PML and transcriptional regulation
6) Role of PML in DNA damage repair
7) PML and the Akt pathway
8) Cytoplasmic PML in tumorigenesis
9) PML in cancers
Thesis Objective
Results
I) Initial Data
1) Screen Design
2) Primary screen results
3) Validation screen
II) Functional study of selected candidates
1) Identification of SKP1a and RBX1
2) Manual validation of screen results
3) Depletion of SKP1a and RBX1 stabilizes PML
4) The overexpression of SKP1a and RBX1 destabilizes PML
5) RBX1 and SKP1a are both interacting with PML
6) Identification of the specific F-Box protein for PML
7) Validation of the interaction between PML and FBXO9
8) FBXO9 interacts with all PML isoforms
9) Localization of FBXO9-PML interaction
10) SUMOylation, arsenic and PML-FBXO9 interaction
11) FBXO9 degrades PML under arsenic trioxide treatment
12) Impact of FBXO9 on the half-life of PML
13) SCFFBXO9 ubiquitinates PML in vitro
14) An attempt to localize the FBXO9 degron on PML
15) Kinase mini-screen to localize the FBXO9 degron on PML
16) Search for physiological stimuli leading to SCFFBXO9-induced PML degradation .
17) Possible links to diseases
Discussion and perspectives
1) SKP1a and RBX1 are members of an ubiquitination complex involved in the degradation of PML
2) The Cullin-1 is involved in SCF complexes
3) FBXO9 specifically interacts with PML
4) FBXO9 is involved in PML stability
5) Is PML degradation dependent on SUMOylation?
6) SCFFBXO9 specifically ubiquitinates PML
7) PML’s degron
8) Kinases phosphorylating PML’s degron
9) PML and FBXO9 in diseases
10) Mouse model Fbxo9 KO
11) PML and cellular differentiation
12) PML and innate immunity
13) The SCF complex: a druggable target
14) Other potential candidates to be studied
Conclusion
Material and Methods
Annexes
Annex 1: The 69 Mammalian F-Box proteins
Annex 2: F-Box protein and E3 ubiquitin ligase implication in cellular pathways.
Annex 3: Validated candidates inducing a morphological change of PML Nuclear Bodies.
Annex 4: Co-immunoprecipitation screen to identify PML interacting F-Box protein.
Annex 5: Cell lines used in mRNA mini screen.
Annex 6: FBXO9 is overexpressed in some types of breast cancers.
Annex 7: Cancer tissue PML antibody staining of Breast and lung cancers.
Bibliographical References
