par Genevois, Philippe
Président du jury Langer, Ingrid
Promoteur Amighi, Karim
Co-Promoteur Wauthoz, Nathalie
Publication Non publié, 2023-10-24
Thèse de doctorat
Résumé : The aim of this work was to obtain a proof-of-concept for the administration of monoclonal antibody- ((m)Ab) based therapeutics directly to the lungs using dry powder for inhalation formulations (DPIf) to treat asthma more efficiently than current marketed treatments. Indeed, while effective mAb treatments for asthma exist, they are all administered subcutaneously, which reduces patient comfort and compliance because of their invasiveness. Moreover, the inhalation route is currently used as the gold standard for treating pulmonary diseases and has proven to be the best choice, whenever possible, to achieve both better outcomes and fewer side effects. Furthermore, DPIf are usually more stable than liquid formulations because of a decrease in the free Gibbs energy and lower kinetics of degradation. Regarding the target, we chose the human (h) interleukin- (IL) 13 as the inflammatory mediator based on a literature review conducted at the beginning of this thesis. Indeed, as of 2014, the most promising therapeutics, besides IL-5 inhibiting treatments, were without a doubt the IL-13 inhibiting treatments with two anti-IL-13 treatments and an anti-IL-4/IL-13 dual treatment in phase III randomized clinical trial (RCT).The first part of the work was, therefore, about developing the anti-hIL-13 molecules from scratch. Two strategies were considered: (i) generating a conventional Ab antigen-binding fragment (FAB) and (ii) generating an antigen-binding fragment from atypic Camelidae immunoglobulin (Ig) G that lacks light chains, also called nanobodies (Nb). Fragments were considered rather than complete mAb because these molecules have a simpler structure, which comes with several advantages: they are easier to formulate and characterize, they require lower mass doses, and are also generally easier and cheaper to produce. Furthermore, they are much less immunogenic by nature and lack the Fc portion responsible for unwanted immune system activations. The generation of the anti-hIL-13 fragments itself was outsourced and then we performed the characterization ourselves. In addition to these active substances, a commercially available murine mAb described as neutralizing the hIL-13 was also characterized. The purpose was to use it as a positive control for our future in vitro and in vivo studies (and thereby confirm that the test did work the way it was supposed to do) on generated mAb but also as a point of comparison during further formulations steps. The commercial anti-hIL-13 mAb was first characterized and confirmed to be a murine IgG1 with kappa light chains. The samples were received in two batches of correct purity (96 ± 2% for batch 1 and 90 ± 3% for batch 2). However, the charge variants’ analysis revealed many subspecies with an isoelectric point (pI) situated between 6.4 and 6.7.Finally, the mAb was revealed to be both a poor binder and a neutralizing mAb as its thermodynamic affinity constant (KD) was estimated at around 1.0 ± 0.8μM for batch 1 and 3 ± 2μM for batch 2, while no inhibitory activity was observed up to 10μM using an in vitro test. Regarding the FAB generation, two out of four immunized mice with hIL-13 showed mAb binding to IL-13. After an initial screening, three mAb samples (8C2H6F11, 3H10A6, and 18D4C3) coming from individual wells were further characterized and identified as IgG1-κ. Further characterization allowed 8C2H6F11 and 3H10A6 to be identified as being the same, or at least very close, variants because of their identical mass spectrometry (MS) analysis and charge variant (pI between 6.35 and 6.93) profiles. On the other hand, 18D4C3 had completely distinct MS and charge variant (pI between 5.23 and 5.44) profiles. The mAb were natively good IL-13 binders (low nanomolar range) but only 18D4C3 displayed a moderate neutralizing activity towards IL-13 with an estimated 50% inhibitory concentration (IC50) at 300 ± 40nM. None of these mAb were further developed but 18D4C3 was sequenced in anticipation of any potential project continuation. Regarding the Nb generation, two llamas were immunized with hIL-13 but only one of them gave a sufficient immune response, generating 38 potential hIL-13 binders, which could be categorized in five complementary-determining region (CDR) 3 regions, which is the binding part the most responsible to the binding to the antigen (one family is defined by identical CDR3 sequence). Among these 38 binders, 30 Nb could be produced and were characterized in terms of binding potency, which was dependent on the CDR3 family but rather constant within the same family. Family 2 displayed a KD between 52 ± 8nM and 150 ± 50nM and family 3 between 2.3 ± 0.2nM and 15 ± 8nM. Families 1, 4, and 5 all displayed KDs above 1,000nM. Nb were then selected from each CDR3 family and functionally tested in vitro. Despite their higher binding potency, the Nb inhibitory potency from CDR3 family 3 (starting inhibitory concentration between 5μM and 12μM) were of the same magnitude as family 2 (starting inhibitory concentration between 3μM and 12μM). Family 4 also had a weak but existing inhibitory activity above 40μM. Although they neutralized IL-13 to some extent, IC50 from the Nb would be several tens higher than those that were reported in the literature for anti-hIL-13 mAb entering RCT. Bivalent and biparatopic dimers were, therefore, generated from families 2, 3, and 4 to increase the inhibitory activity. The biparatopic dimers made from families 2 and 3 were particularly efficient as, despite a KD of the same magnitude as monomers, they were 100-fold more potent on a molar basis (IC50 between 0.60 ± 0.06μM and 0.73 ± 0.08μM, respectively) compared to the extrapolated IC50 (based on the hypothesis on a complete inhibition) of their monomeric counterparts (around 80-90 μM). As the biparatopic Nb from families 2 and 4 also saw its IC50 decreased (42 ± 3μM), a triparatopic Nb composed of those three Nb was made (2IL43/2IL172/3ILT82) and its IC50 was further decreased to 200 ± 100nM.The triparatopic construct (TPC) was then evaluated in a hIL-13-induced pulmonary inflammation in vivo (murine) model. The lung functional parameters were evaluated using the forced oscillation technique and inflammatory mediators and cells were also quantified. The trimer significantly improved the lung condition for parameters affected by hIL-13, with a dose-dependent efficacy observed from 0.3mg/kg to 1.0mg/kg. However, from 3.0mg/kg to 10mg/kg, lung remodeling and inflammation were observed, which could be due to a negative immune response to the foreign nature of the trimer (from a llama). This effect could potentially override the therapeutic benefit of the trimer. Finally, the overall potency of the TPC treatment had a statistically significant in vivo effect on the lung function at a comparable dose (1mg/kg) to tralokinumab (an anti-IL-13 monoclonal antibody which was evaluated in a phase 3 randomized clinical trial (RCT)) within a comparable in vivo study, even though the latter was administered intraperitoneally.The second part of the work was about generating DPIf on model proteins, similar to the ones generated during the first part of the work (mAb and Nb). Spray-drying was chosen because it is a one-step process which allows the particle distribution size to be easily engineered.First, a bovine polyclonal Ab (pAb) was used because of its high availability and low cost. The pAb was very stable over time at room temperature (RT) up to 10 months and at 40°C up to three months, independently of the excipients that were used (most of the formulations had above 95% soluble protein and monomer recoveries). The formulations that showed a high level of degradation (up to 50% monomer or soluble protein content lost) were correlated to (i) the ability of the stabilizing excipients to make H-bonds and (ii) their crystallization propensity. The combination of low m.w. sugars/polyols and high m.w. polysaccharides (i.e., dextran 10kDa) allowed a better stabilization of the pAb. Because the residual moisture (RM) in the DPIf was very high (up to 16%w/w), several strategies were considered to reduce the RM to reduce the crystallization propensity by increasing the glass transition temperature (Tg). The goal was mostly achieved using a secondary drying at 40°C under vacuum with a RM of around 5% being obtained. Finally, a technology transfer from the Büchi B-290 minispray-drier to the ProCept 4M8-Trix was performed. The simple transfer of the parameters between each piece of equipment did not work as lower fine particle fractions (FPFs) were obtained with the ProCept 4M8-Trix (52-59%) compared to the Büchi B-290 (66-67%). Nebulization was identified as a critical process parameter for the ProCept 4M8-Trix and the particles’ size was optimized based on this parameter. Similar FPFs (63-68%) were obtained using the highest nebulization pressure tested for each nozzle (3.0 bars for 0.2mm and 0.6mm and 2.23bars for 0.4mm).Second, the anti-IL-13 murine mAb characterized during the first part of the work was formulated as a DPIf. Contrary to the pAb, more degradations were observed, especially over storage. Adding a polyol or a small sugar provided better protection compared to the mAb spray-dried alone and, similarly for the pAb, combining them with either a higher-order saccharide (pullulan) or L-leucine (L-Leu) allowed the mAb to be completely preserved just after the spray-drying formulation step (T0). On the other hand, none of the strategies which were considered were useful to prevent the degradations, which were mostly identified as oxidation although other degradations unrelated to oxidation were also observed. Finally, storing the formulations under dinitrogen (N2) permitted the mAb to be preserved from aggregation and chemical modifications over a mid-term period (two months). The FPFs obtained from the formulations including the mAb were, however, lower than those obtained with the pAb as the highest FPF was observed for the DPIf formulated only with the buffer (57%) compared to the same formulation with the pAb (68 ± 4%).Finally, an anti-beta-lactamase (BC) II (from Bacillus cereus (B. cereus)) monomeric Nb was spray-dried according to the same strategies as for the anti-IL-13 mAb. The anti-BCII Nb was used rather than the TPC because the expression yield for the TPC was too low to obtain enough material for spray-drying purposes. The anti-BCII Nb was more resistant than the mAb to the spray-drying process but the degradation profile over storage was similar. The degradations occurring over storage were (almost exclusively) identified as oxidation, leading to aggregation, and were highly potentiated by the storage temperature (up to 100% degradation under ambient air (AA) at 40°C for two months) while low-to-no degradations were observed under N2 at 2-8°C for two months. However, even with high rates of degradation, the Nb affinity did not change much as binding was still observed with a KD of 1.5 ± 0.3nM at 40°C under AA versus (vs.) 0.18 ± 0.03nM at 40°C under N2 and vs. 0.16 ± 0.02nM natively. The FPF which was obtained for the Nb was comparable to the same formulation containing the mAb (44 ± 2% vs. 43 ± 3%).To conclude, this thesis has globally demonstrated that the pulmonary administration of an anti-IL-13 Nb provided an equivalent therapeutic effect on a mass basis as a well-known reference injected intraperitoneally. The project has also demonstrated the feasibility of obtaining DPIf by spray-drying for mAb fragments as well as complete mAb with adequate mid-term storage, although much work remains to be done to obtain commercially viable therapeutic options.

Documents en relation

DI-fusion