Design of bio-inspired ionic liquids for protein stabilisation

Ionic Liquids (ILs) consist in organic salts that are liquid at/or near room temperature. Since ILs are entirely composed of ions, the formation of ion pairs is expected to be one essential feature for describing solvation in ILs. In recent years, protein - ionic liquid (P-IL) interactions have been the subject of intensive studies mainly because of their capability to promote folding/unfolding of proteins. However, the ion pairs and their lifetimes in ILs in P-IL thematic is dismissed, since the action of ILs is therefore the result of a subtle equilibrium between anion-cation interaction, ion-solvent and ion-protein interaction. The work developed in this thesis innovates in this thematic, once the design of ILs for protein stabilisation was bio-inspired in the high concentration of organic charged metabolites found in cell milieu. Although this perception is overlooked, those combined concentrations have been estimated to be ~300 mM among the macromolecules at concentrations exceeding 300 g/L (macromolecular crowding) and transient ion-pair can naturally occur with a potential specific biological role. Hence the main objective of this work is to develop new bio-ILs with a detectable ion-pair and understand its effects on protein structure and stability, under crowding environment, using advanced NMR techniques and calorimetric techniques. The choline-glutamate ([Ch][Glu]) IL was synthesized and characterized. The ion-pair was detected in water solutions using mainly the selective NOE NMR technique. Through the same technique, it was possible to detect a similar ion-pair promotion under synthetic and natural crowding environments. Using NMR spectroscopy (protein diffusion, HSQC experiments, and hydrogen-deuterium exchange) and differential scanning calorimetry (DSC), the model protein GB1 (production and purification in isotopic enrichment media) it was studied in the presence of [Ch][Glu] under macromolecular crowding conditions (PEG, BSA, lysozyme). Under dilute condition, it is possible to assert that the [Ch][Glu] induces a preferential hydration by weak and non-specific interactions, which leads to a significant stabilisation. On the other hand, under crowding environment, the [Ch][Glu] ion pair is promoted, destabilising the protein by favourable weak hydrophobic interactions , which disrupt the hydration layer of the protein. However, this capability can mitigates the effect of protein crowders. Overall, this work explored the ion-pair existence and its consequences on proteins in conditions similar to cell milieu. In this way, the charged metabolites found in cell can be understood as key for protein stabilisation.

Stabilisation of dehydrated nanoemulsions using sugar - protein matrices

There are numerous review papers discussing liquid nanoemulsions and how they compare to other emulsion systems. Little research is available on dried nanoemulsions. The objectives of this research were to (i) study the effect of varying the continuous phase of nanoemulsions with different carbohydrate/protein ratios on subsequent emulsion stability, and (ii) compare the physicochemical properties, lactose crystallisation properties, microstructure, and lipid oxidation of spray dried nanoemulsions compared to spray dried conventional emulsions having different water and sugar contents. Nanoemulsions containing sunflower oil (10% w/w), β-casein (2.5–10% w/w) and lactose or trehalose (10–17.5%) were produced following optimisation of the continuous phase by maximising and minimising viscosity and glass transition temperature (Tg’) using mixture design software. Increasing levels of β-casein from caused a significant increase in viscosity, particle size, and nanoemulsion stability, while resulting in a decrease in Tg’. Powders were made from spray drying emulsions/nanoemulsions consisting of lactose or a 70:30 mixture of lactose:sucrose (23.9%), sodium caseinate (5.1%) and sunflower oil (11.5%) in water. Nanoemulsions, produced by microfluidisation (100 MPa), had higher stability and lower viscosity than control emulsions (homogenization at 17 MPa) with lower solvent extractable free fat in the resulting powder. Partial replacement of lactose with sucrose decreased Tg and delayed Tcr. DVS and PLM showed that in powdered nanoemulsions, lactose crystallises faster than in powdered conventional emulsions. Microstructure of both powders (CLSM and cryo-SEM) showed different FGS in powders and different structure post lactose crystallisation. Powdered nanoemulsions had lower pentanal and hexanal (indicators of lipid oxidation) after 24 months storage due to their lower free fat and porosity, measured using a validated GC HS-SPME method, This research has shown the effect of altering the continuous phase of nanoemulsions on microstructure of spray dried nanoemulsions, which affects physical properties, sugar crystallisation, and lipid oxidation.

Mitotic Arrest Deficiency Protein 2 (MAD2) and Histone Deacetylase 6 (HDAC6) Present a Complex Relationship in Their Regulation and Expression and Subsequent Impact on Chemoresponsiveness

Introduction:
Ovarian cancer patients presenting with advanced stage (III/IV)
canceraretreatedwithcarboplatinumincombinationwithpaclitaxel.Despitea
significant initial response rate, fewer than 20% of patients become long-term
survivors. We have published that low MAD2 expression levels associate with
reduced progression free survival (PFS) in patients with high-grade serous
epithelial ovarian cancer (EOC). Moreover, we have demonstrated that MAD2
expressionisdown-regulatedbythemicroRNAmiR-433(
Furlong et al., 2011
).
Interestingly, miR-433 also down-regulates HDAC6 (
Simon et al., 2010
), which
uniquely deacetylates
a
-tubulin prior to HDAC6s binding to
b
-tubulin.
In vitro
studies have shown that HDAC6 inhibition in combination with paclitaxel
treatment enhances chemoresistant cancer cell death. To date, an interaction
between MAD2 and HDAC6 has not been reported.
Experimental design:
MAD2 and HDAC6 immunohistochemistry (IHC) and
Western blot analyses were performed to investigate the role of HDAC6 and
MAD2 in chemoresistance to paclitaxel in high-grade serous EOC.
Results and Discussion:
In vitro
experiments demonstrated that overex-
pression of pre-miR-433, which targets MAD2, resulted in down-regulation
of HDAC6 in EOC cell lines. High levels of HDAC6 are co-expressed with
MAD2 in the paclitaxel resistant UPN251 and OVCAR7 cell lines. While, all
4 paclitaxel resistant EOC cell lines express higher levels of miR-433 than
the paclitaxel sensitive A2780 cells, only ovca432 and ovca433 demonstrated
down-regulation of both HDAC6 and MAD2. Paclitaxel binds to
b
-tubulin and
causesmicrotubulepolymerizationinpaclitaxelsensitivecellsasdemonstrated
by tubulin acetylation in A2780 cells. However, paclitaxel failed to cause a
significant acetylation of
a
-tubulin and microtubule stabilisation in the resistant
UPN251 cells. Therefore resistance in this cell line may be mediated by
aberrantly high HDAC6 activity. We have previously shown that MAD2 knock-
down cells are resistant to paclitaxel (
Furlong F., et al., 2011; Prencipe M.,
et al., 2009
). We measured HDAC6 protein expression in MAD2 knockdown
cells and showed that MAD2 knockdown is associated with concomitant
up-regulation of HDAC6. We hypothesise that the up-regulation of HDAC6
by MAD2 knockdown renders cancer cells more resistant to paclitaxel and
increases the invasive potential of these cells. On-going experiments will test
this hypothesis. Lastly we have observed differential MAD2 and HDAC6 IHC
staining intensity in formalin fixed paraffin embedded EOC samples.
In conclusion
, we have reported on a novel interaction between MAD2 and
HDAC6 which may have important consequences for paclitaxel resistant EOC.
Moreover, understanding chemo-responsiveness in ovarian tumours will lead
to improved patient management and treatment options for women diagnosed
with this disease

Stabilisation of MAD2 expression through inhibition of miR-433; A novel mechanism for restoring chemo-responsiveness in ovarian tumours

Epithelial ovarian carcinoma (EOC) is characterised by late diagnosis and recurrences, both of which contribute to the high morbidity and mortality of this cancer. Unfortunately, EOC has an innate susceptibility to become chemo-resistant. Specifically, up to 30% of patients may not respond to current standard chemotherapy (paclitaxel and platinum in combination) and of those who have an initial response, some patients relapse within a few months. Therefore, in order to improve patient outcome it is crucial to establish what factors influence a patients' individualised response to chemotherapy. We analysed MAD2 protein expression in a patient cohort of 35 ovarian tumours and a panel of 5 ovarian cancer cell lines. We have demonstrated that low nuclear MAD2 expression intensity was significantly associated with chemo-resistant ovarian tumours (p=0.0136). Moreover, in vitro studies of the 5 ovarian cancer cell lines revealed that reduced MAD2 expression was associated with paclitaxel resistance. In silico analysis identified a putative miR-433 binding domain in the MAD2 3′UTR and expression profiling of miR-433 in the ovarian cancer cell lines showed that low MAD2 protein expression was associated with high miR-433 levels. In vitro over-expression of miR-433 attenuated MAD2 protein expression with a concomitant increase in cellular resistance to paclitaxel. Over-expression of a morpholino oligonucleotide that blocks miR-433 binding to MAD2 3′UTR stabilised MAD2 protein expression and protects from miR-433 induced degradation. Furthermore, miR-433 expression analysis in 35 ovarian tumour samples revealed that high miR-433 expression was associated with advanced stage presentations (p=0.0236). In conclusion, ovarian tumours that display low nuclear MAD2 intensity are chemo-resistant and stabilising MAD2 expression by antagonising miR-433 activity is a potential mechanism for restoring chemo-responsiveness in these tumours.

Impact d’un épisode asphyxique périnatal associé à des convulsions sur le développement des interneurones corticaux : rôle de l’adenosine monophosphate-activated protein kinase (AMPK)

L’encéphalopathie hypoxique-­‐ischémique cause des milliers de victimes à travers le monde chaque année. Les enfants survivants à un épisode hypoxique-­‐ischémique sont à risque de développer des problèmes neurologiques incapacitants comme une paralysie cérébrale, un retard mental, une épilepsie ou des troubles d’ordre comportemental. Les modèles animaux ont amélioré nos connaissances sur les mécanismes sous-­‐jacents aux dommages cérébraux, mais elles sont encore trop incomplètes pour être capables de prévenir les problèmes neurologiques. Ce projet vise à comprendre l’impact d’un épisode asphyxique périnatale associé à des convulsions ainsi que l’activation de l’adenosine monophosphate-­‐activated protein kinase (AMPK) sur les circuits GABAergiques inhibiteurs en développement chez la souris. Dans le but d’investiguer le sort des neurones inhibiteurs, appelés interneurones, suite à un épisode asphyxique périnatal associé à des convulsions avec des animaux transgéniques, nous avons pris avantage d’un nouveau modèle d’hypoxie permettant d’induire des convulsions chez la souris. Deux populations d’interneurones représentant ensemble environ 60% de tous les interneurones corticaux ont été étudiées, soit les cellules exprimant la parvalbumine (PV) et les cellules exprimant la somatostatine (SOM). L’étude stéréologique n’a montré aucune mort neuronale de ces deux populations d’interneurones dans l’hippocampe chez les souris hypoxique d’âge adulte. Par contre, le cortex des souris hypoxiques présentait des zones complètement ou fortement dépourvues de cellules PV alors que les cellules SOM n’étaient pas affectées. L’utilisation d’une lignée de souris transgénique exprimant une protéine verte fluorescente (GFP) dans les cellules PV nous a permis de comprendre que les trous PV sont le reflet de deux choses : 1) une diminution des cellules PV et 2) une immaturité des cellules PV restantes. Puisque les cellules PV sont spécifiquement affectées dans la première partie de notre étude, nous avons voulu étudier les mécanismes moléculaires sous-­‐jacents à cette vulnérabilité. L’AMPK est un senseur d’énergie qui orchestre le rétablissement des i niveaux d’énergie cellulaire dans le cas d’une déplétion énergétique en modulant des voies de signalisation impliquant la synthèse de protéines et l’excitabilité membranaire. Il est possible que l’activation d’AMPK suite à un épisode asphyxique périnatal associé à des convulsions soit néfaste à long-­‐terme pour le circuit GABAergique en développement et modifie l’établissement de l’innervation périsomatique d’une cellule PV sur les cellules pyramidales. Nous avons étudié cette hypothèse dans un modèle de culture organotypique en surexprimant la forme wild-­‐type (WT) de la sous-­‐unité α2 d’AMPK, ainsi qu’une forme mutée dominante négative (DN), dans des cellules PV individuelles. Nous avons montré que pendant la phase de formation synaptique (jours post-­‐natals équivalents EP 10-­‐18), la surexpression de la forme WT désorganise la stabilisation des synapses. De plus, l’abolition de l’activité d’AMPK semble augmenter le nombre de synapses périsomatiques faits par la cellule PV sur les cellules pyramidales pendant la phase de formation et semble avoir l’effet inverse pendant la phase de maturation (EP 16-­‐24). La neurotransmission GABAergique joue plusieurs rôles dans le cerveau, depuis la naissance jusqu’à l’âge adulte des interneurones, et une dysfonction des interneurones a été associée à plusieurs troubles neurologiques, comme la schizophrénie, l’autisme et l’épilepsie. La maturation des circuits GABAergiques se fait majoritairement pendant la période post-­‐natale et est hautement dépendante de l’activité neuronale et de l’expérience sensorielle. Nos résultats révèlent que le lourd fardeau en demande énergétique d’un épisode asphyxique périnatal peut causer une mort neuronale sélective des cellules PV et compromettre l’intégrité de leur maturation. Un des mécanismes sous-­‐ jacents possible à cette immaturité des cellules PV suite à l’épisode hypoxique est l’activation d’AMPK, en désorganisant leur profil d’innervation sur les cellules pyramidales. Nous pensons que ces changements dans le réseau GABAergique pourrait contribuer aux problèmes neurologiques associés à une insulte hypoxique.

The effect of osmolytes on protein fold stability at the single-molecule level

By pulling and releasing the tension on protein homomers with the Atomic Force Miscroscope (AFM) at different pulling speeds, dwell times and dwell distances, the observed force-response of the protein can be fitted with suitable theoretical models. In this respect we developed mathematical procedures and open-source computer codes for driving such experiments and fitting Bell’s model to experimental protein unfolding forces and protein folding frequencies. We applied the above techniques to the study of proteins GB1 (the B1 IgG-binding domain of protein G from Streptococcus) and I27 (a module of human cardiac titin) in aqueous solutions of protecting osmolytes such as dimethyl sulfoxide (DMSO), glycerol and trimethylamine N-oxide (TMAO). In order to get a molecular understanding of the experimental results we developed an Ising-like model for proteins that incorporates the osmophobic nature of their backbone. The model benefits from analytical thermodynamics and kinetics amenable to Monte-Carlo simulation. The prevailing view used to be that small protecting osmolytes bridge the separating beta-strands of proteins with mechanical resistance, presumably shifting the transition state to significantly higher distances that correlate with the molecular size of the osmolyte molecules. Our experiments showed instead that protecting osmolytes slow down protein unfolding and speed-up protein folding at physiological pH without shifting the protein transition state on the mechanical reaction coordinate. Together with the theoretical results of the Ising-model, our results lend support to the osmophobic theory according to which osmolyte stabilisation is a result of the preferential exclusion of the osmolyte molecules from the protein backbone. The results obtained during this thesis work have markedly improved our understanding of the strategy selected by Nature to strengthen protein stability in hostile environments, shifting the focus from hypothetical protein-osmolyte interactions to the more general mechanism based on the osmophobicity of the protein backbone.

Involvement of tissue transglutaminase in the stabilisation of biomaterial/tissue interfaces important in medical devices

Tissue transglutaminase (tTG) has recently been established as a novel cell surface adhesion protein that binds with high affinity to fibronectin in the pericellular matrix. In this study, we have made use of this property to enhance the biocompatibility of poly(epsilon-caprolactone) (PCL), a biomaterial currently used in bone repair. Poly(epsilon-caprolactone) discs were first coated with fibronectin and then tTG. The surface localisation of the two proteins was confirmed using ELISA and the tTG shown to be active on the surface by incorporation of biotin cadaverine into the fibronectin coating. When human osteoblasts (HOBs) were seeded onto the coated polymer surfaces in serum free medium, the surface coated with fibronectin and then tTG showed an increase in the spreading of the cells as compared to the surface coated with fibronectin alone, when analysed using environmental scanning electron microscopy. The presence of tTG had no effect on HOB cell differentiation when analysed by determining alkaline phosphatase activity. The use of tTG as a novel adhesion protein in this way may therefore have considerable potential in forming a stable tissue/biomaterial interface for application in medical devices.

Stabilising suppressor of cytokine signalling 3 (SOCS3) protein levels to limit neointimal hyperplasia

Suppressor of cytokine signalling 3 (SOCS3) is a potent inhibitor of the mitogenic, migratory and pro-inflammatory pathways responsible for the development of neointimal hyperplasia (NIH), a key contributor to the failure of vascular reconstructive procedures. However, the protein levels of SOCS3, and therefore its potential to reduce NIH, is limited by its ubiquitylation and high turnover by the proteasome. I hypothesised that stabilisation of endogenous SOCS3 by inhibiting its ubiquitylation has the potential to limit vascular inflammation and NIH. Consequently, the aim of this PhD was to identify the mechanisms promoting the rapid turnover of SOCS3. Initial experiments involved the identification of residues involved in regulating the turnover of SOCS3 at the proteasome. I assessed the ubiquitylation status of a panel of FLAG tagged SOCS3 truncation mutants and identified a C-terminal 44 amino acid region required for SOCS3 ubiquitylation. This region localised to the SOCS box which is involved in binding Elongin B/C and the formation of a functional E3 ubiquitin ligase complex. However, the single lysine residue at position 173, located within this 44 amino acid region, was not required for ubiquitylation. Moreover, Emetine chase assays revealed that loss of either Lys173 or Lys6 (as documented in the literature) had no significant effect on SOCS3 stability 8 hrs post emetine treatment. As mutagenesis studies failed to identify key sites of ubiquitylation responsible for targeting SOCS3 to the proteasome, LC-MS-MS analysis of a SOCS3 co-immunoprecipitate was employed. These data were searched for the presence of a Gly-Gly doublet (+114 Da mass shift) and revealed 8 distinct sites of ubiquitylation (Lys23, Lys28, Lys40, Lys85, Lys91, Lys173, Lys195, Lys206) on SOCS3 however Lys6 ubiquitylation was not detected. As multiple Lys residues were ubiquitylated, I hypothesised that only a Lys-less SOCS3, in which all 8 Lys residues were mutated to Arg, would be resistant to ubiquitylation. Compared to WT SOCS3, Lys-less SOCS3 was indeed found to be completely resistant to ubiquitylation, and significantly more stable than WT SOCS3. These changes occurred in the absence of any detrimental effect on the ability of Lys-less SOCS3 to interact with the Elongin B/C components required to generate a functional E3 ligase complex. In addition, both WT and Lys-less SOCS3 were equally capable of inhibiting cytokine-stimulated STAT3 phosphorylation upon co-expression with a chimeric EpoR-gp130 receptor. To assess whether SOCS3 auto-ubiquitylates I generated an L189A SOCS3 mutant that could no longer bind the Elongins and therefore form the E3 ligase complex required for ubiquitylation. A denaturing IP to assess the ubiquitylation status of this mutant was performed and revealed that, despite an inability to bind the Elongins, the L189A mutant was poly-ubiquitylated similar to WT SOCS3. Together these data suggested that SOCS3 does not auto-ubiquitylate and that a separate E3 ligase must regulate SOCS3 ubiquitylation. This study sought to identify the E3 ligase and deubiquitylating (DUB) enzymes controlling the ubiquitylation of SOCS3. Our initial strategy was to develop a tool to screen an E3 ligase/DUB library, using an siARRAY, to sequentially knockdown all known E3 ligases in the presence of a SOCS3-luciferase fusion protein or endogenous SOCS3 in a high content imaging screening platform. However, due to a poor assay window (<2) and non-specific immunoreactivity of SOCS3 antibodies available, these methods were deemed unsuitable for screening purposes. In the absence of a suitable tool to screen the si-ARRAY, LC-MS-MS analysis of a SOCS3 co-immunoprecipitate (co-IP) was investigated. I performed a SOCS3 under conditions which preserved protein-protein interactions, with the aim of identifying novel E3 ligase and/or DUBs that could potentially interact with SOCS3. These data were searched for E3 ligase or DUB enzymes that may interact with SOCS3 in HEK293 cells and identified two promising candidates i) an E3 ligase known as HectD1 and ii) a DUB known as USP15. This thesis has demonstrated that in the presence of HectD1 overexpression, a slight increase in K63-linked polyubiquitylation of SOCS3 was observed. Mutagenesis also revealed that an N-terminal region of SOCS3 may act as a repressor of this interaction with HectD1. Additionally, USP15 was shown to reduce SOCS3 polyubiquitylation in a HEK293 overexpression system suggesting this may act as a DUB for SOCS3. The C-terminal region of SOCS3 was also shown to play a major role in the interaction with USP15. The original hypothesis of this thesis was that stabilisation of endogenous SOCS3 by inhibiting its ubiquitylation has the potential to limit vascular inflammation and NIH. Consistent with this hypothesis, immunohistochemistry visualisation of SOCS3, in human saphenous vein tissue derived from CABG patients, revealed that while SOCS3 was present throughout the media of these vessels the levels of SOCS3 within the neointima was reduced. Finally, preliminary data supporting the hypothesis that SOCS3 overexpression may limit the proliferation, but not migration, of human saphenous vein smooth muscle cells (HSVSMCs) is presented. It is expected that multiple E3 ligases and DUBs will contribute to the regulation of SOCS3 turnover. However, the identification of candidate E3 ligases or DUBs that play a significant role in SOCS3 turnover may facilitate the development of peptide disruptors or gene therapy targets to attenuate pathological SMC proliferation. A targeted approach, inhibiting the interaction between SOCS3 and identified E3 ligase, that controls the levels of SOCS3, would be expected to reduce the undesirable effects associated with global inhibition of the E3 ligase involved.