Добавил:
kiopkiopkiop18@yandex.ru t.me/Prokururor I Вовсе не секретарь, но почту проверяю Опубликованный материал нарушает ваши авторские права? Сообщите нам.
Вуз: Предмет: Файл:
Ординатура / Хирургия / Библиотека им академика М.И. Перельмана / Книга_5886_Библиотеки_им_академика_М_И_Перельмана.pdf
Скачиваний:
0
Добавлен:
02.09.2026
Размер:
21 Мб
Скачать
420
self-interactions and weak nonspecic interactions [154]. Calculation of the pI on the basis of the mAb sequences is offered by multiple software tools as MassLynx [], Vector NTI [155], and EMBOSS [156], and the calculated pIs have been shown to be within a range of 15% compared to experimentally determined ones [157]. Tools that predict the pI on basis of the protein structure can provide a more accu­rate result, since the underlying residue pKa values are calculated by taking into account the residual microenvironments, e.g., shifts in pKa values in case the side chain forms a salt-bridge. A case study conrmed that structure-based pI calcula­tions (via PROPKA) [158] corresponded to the measured values at least by ±1 pI unit, whereby the remaining discrepancies result from protein interactions with salt ions that leave room for improvements [11, 159].
J. Bauer et al.
14.3.6 Viscosity andDiffusion Interaction Parameter (kD)
Predictors of the concentration-dependent viscosity of formulations need to con­sider the pairwise and higher orderself-association of antibodies [19]. Consequently, the main driving forces behind viscosity, i.e., electrostatics and hydrophobicity [20] have been linked to predictable characteristics of the Fv sequence and structure [159, 160]. While the viscosity of mAb formulations increases with hydrophobicity, charge dipole distribution, and aggregation propensity, it decreases with net charge [159, 160]. Considering these principles, the in silico tool SCM (spatial charge map) offers to detect highly viscous antibodies on the basis of the Fv structure [161]. The high-throughput approach makes it especially attractive for the lead identication and optimization performed at discovery. An alternative mathematical model considers again the hydrophobicity and charges of antibody regions to pre­dict concentration-dependent viscosity curves [162]. Charge-related in silico descriptors (e.g., pI, net charge, charge on Fv, zeta potential) could be used to cal­culate the diffusion interaction parameter kD [163] that itself is widely-used in developability assessments to predict the viscoelastic behavior of mAb solutions [164166]. Several case studies conrmed that these computational tools as well as their underlying principles are powerful to rationally (re-)design antibodies with decreased viscosity [137, 138, 141, 142, 167171]. Further studies indicated that MD simulations can be conducted to additionally estimate the effect of formulation components on the viscosity [137, 170, 172176].
14.3.7 Aggregation andSelf-Association
Aggregation in biotherapeutics can nucleate from a wide array of origins ranging from reversible self-association, conformational change, chemical modication, hydrophobic and electrostatic complementarity among the interacting partners. Despite this complexity, aggregation could be directly associated to characteristics
14 Biopharmaceutical Informatics: A Strategic Vision for Discovering Developable…
421
such as presence of aggregation prone regions, hydrophobicity [177], electrostatics [178], and dipole moments [179], paving the way for both sequence- and structure­based computational predictions. The in silico tools are particularly powerful in guiding the design of mAb candidates with high colloidal stability, since the effect of single and multiple amino acid exchanges on the aggregation propensity can be predicted. A selection of these predictors is summarized below, and this aspect has been described in several reviews in more details [147, 151, 180].
A vast majority of aggregation predictors are based on the sequence as they cal­culate aggregation propensity values for sequence stretches. These methods are based on the fact that natural amino acids differ in their physicochemical properties, and individually inuence the aggregation of peptides and proteins [181]. Therefore, techniques as Zyggregator and Pag use intrinsic properties of the amino acids such as hydrophobicity, charge and secondary-structure propensity to identify aggregation- promoting regions and predict the rate of growth [182, 183]. Similarly, a method called AGGRESCAN identies aggregation-prone segments via the aggregation propensity of natural amino acids, whereby this method was trained on empirical data [184]. Alternative tools like TANGO, PASTA, PASTA2, FoldAmyloid, SALSA; and AggreRATE-Pred detect aggregation-prone regions and impact of mutations on aggregation kinetics via the physicochemical properties of secondary structure elements, in particular the ability to form intermolecular cross-β-structures [133, 150, 185188]. In accordance, the frequency of β-sheets was amongst others (e.g., pI and atom-based hydrophobic moment) carved out by a neural network approach that led to the development of APPNN [189]. Another predictor named WALTZ consults a database of experimentally determined amyloid-forming hexa­peptides to predict amyloid-forming sequences [190, 191]. Following a consensus prediction algorithm, 5 and later on 11 other methods (AGGRESCAN, Pag, Tango, Waltz, and other conformational predictors) have been combined to produce AMYLPRED and AMYLPRED2 [192, 193]. Similarly, a statistical approach com­bined existing tools (SALSA, PAFIG, FoldAmyloid, and Waltz) to develop a meta­predictor for amyloid proteins called MetAmyl [194]. Alternatively, articial neuronal networks (ANN) were applied to predict the aggregation temperature (T
agg
) of antibodies based on the amino acid composition, of which charged residues and Met affected the temperature induced aggregation the most [195]. Machine-learning techniques were conducted to develop CISI, which predicts the cross-interaction or self-interaction of mAbs on the basis of tripeptide sequences [196]. Interestingly, this study indicated that tripeptides comprising Tyr, Ser and Ala residues were amongst the most prominent antibody-binding sites.
In comparison to these sequence-based methods, the structure-based ones con­sider the three-dimensional context of the protein fold to predict aggregation. Therefore, methods as AGGRESCAN3D (A3D) or Solubis combine distinct aggre­gation propensity scales (AGGRESCAN or TANGO) with the protein structure [148, 197, 198]. The A3D server further allows to include dynamic uctuations of the protein structure in solution that might affect the aggregation tendency [197]. It was further pointed out that atomic level characteristics might be relevant for
422
J. Bauer et al.
understanding aggregation, which is offered by ANuPP [145]. Another method called AggScore identies aggregation hotspots on the protein surface via calcula­tion of hydrophobic and electrostatic patches [199]. Orthogonal to these technolo­gies, SAP (spatial aggregation propensity) calculates dynamically exposed hydrophobicity of protein patches [123, 200]. The Developability Index (DI) has complemented the SAP concept with additional information about the net charge of the full-length mAb to account for hydrophobicity and electrostatic interactions [201]. Recently, a machine learning approach combined SAP with other tools (Spatial negative charge map (SCM neg), Spatial positive charge map (SCM pos)) that extract molecular features from MD simulations to train predictive models for the aggregation rate at high mAb concentrations [202].
14.3.8 Prediction ofParatope
Once an experimentally determined or accurately modeled structure of the thera­peutic candidate is available, the mAb can be further analyzed to identify the para­tope comprising the amino acid residues that mediate the binding of its cognate antigen epitope. Oversimplied approaches assume that the paratope equals the six CDR regions comprising together 50–60 residues on average, but statistic evalua­tion of Ab/Ag complexes showed that the actual paratope is only formed by 18–19 residues on average [203]. Therefore, gathering a detailed understanding of the exact paratope is crucial to guide an individually tailored engineering and develop­ment strategy that avoids costly and time-consuming failures. However, especially at early research stages, the Ab/Ag complex might not have been experimentally characterized. In case information about the antigen sequence and/or structure are known, bioinformatic tools offer the prediction of the mAb interaction with the Ag structure/model that can complement experimental paratope studies and streamline conventional R&D workows [204]. Similar to homology modelling, great advances have also been made in the eld of paratope prediction [204, 205]. In the last years, several paratope predictors that pursue individual approaches were established. The most common technique considers the docking of Ab/Ag for the prediction of the paratope [206208]. Another method identies the paratope via circular patches of the antibody surface that exhibit certain physicochemical properties [209]. Deep neural networks are not only suitable for the paratope prediction but outperformed classical methods [205, 210]. Accurate prediction of paratopes is essential to iden­tify potential physicochemical liabilities motifs that may overlap with antigen­binding regions. Removal of such motifs via mutations in paratope, might lead to signicant modication of the biologic drug candidate’s afnity toward its cognate receptor.
14 Biopharmaceutical Informatics: A Strategic Vision for Discovering Developable…
423
14.3.9 Post-Translational Modications (PTM)
Many different approaches have been pursued to develop predictive tools for sites of post-translational modications of which oxidation, deamidation, isomerization, and glycation are the most prominent ones.
Oxidation occurs most frequently at the side chains of Trp and Met that differ in their atomistic nature (aromatic ring; sulfur), sensitivity toward different oxidative stress (light exposure, metal-catalyzation and free radicals), and oxidized products [211]. Therefore, different in silico tools have been developed for the prediction of Trp and Met oxidation sites. For the prediction of Met oxidation, correlations between the experimental oxidation data of stressed mAbs and the solvent accessi­ble surface area (SASA) of the Met side chain have been seen [212]. Recently, several different machine learning approaches have been conducted to train and t different classication models such as MetODeep for the prediction of oxidation­sensitive Met sites [213217]. Some of the resulting models used random forest algorithm and consider characteristics about sequence, secondary andtertiary struc­tures, and dynamics to predict Met oxidation. Interestingly, again the SASA of the Met side chain was amongst the most important discriminators. Furthermore, the number of residues between the analyzed Met and the next Met toward the N-terminus and the spatial distance between the sulfur atom and the closest aro­matic residue affected the oxidation propensity [215]. Typically, MD simulations are conducted to capture the dynamics of the structure and improve the prediction of critical oxidation sites [218, 219].
Like the Met oxidation sites, high solvent accessibility of the Trp residue is a prerequisite for oxidation [159, 220, 221]. However, it is hypothesized that in case of Met residues the solvent exposure of the sulfur atom is more important than that of the entire side chain [27, 215, 219]. In contrast, the entire side chain is considered for the prediction of Trp sites sensitive to oxidation. The oxidation susceptibility does not linearly correlate to the SASA of the Trp residue, indicating that other structural characteristics as side chain orientation and/or surrounding structural ele­ments might play a role [221]. The predictive power of such tools remains to be veried for molecules that underwent different types of stress (light, reactive oxy­gen, metal).
Sites sensitive to deamidation and isomerization have been historically identied via sequence motifs (e.g., NG), but this approach was shown to be less accurate and prone to overprediction [222]. Therefore, the risk predictions could be improved for mAbs by not only considering the motif but also the CDR region [223]. Later algo­rithms as NGOME were developed under consideration of simple predictions of secondary structure elements [224]. Finally, the complex conformational environ­ment of the reactive groups, particularly the local structure, structure exibility and solvent accessibility, has been considered in several other studies [159, 225228]. Furthermore, a predictor for Asn deamidation was recently trained on structural features via machine learning algorithms [229]. Thereby, Asn torsion, exibility,
424
J. Bauer et al.
solvent exposure, structural elements, and the distance between C-N for the nucleo­philic attack were identied as the most relevant descriptors [229].
14.4 Conclusion andFuture Directions
Computational approaches are nding increasing applications toward discovery and development of biotherapeutics in recent years. This chapter has attempted to pro­vide an overview of this emerging eld, we call as biopharmaceutical informatics. However, several major challenges remain toward increasing the contribution of biopharmaceutical informatics andthereby increasingthe probability ofsuccessful translation of biologic drug candidates into medicines available in the clinic. One of the most signicant hurdles for greater acceptance for biopharmaceutical informat­ics is the predictability and validation of the computational tools. Validation via self-consistent experimental data is important for the development of these emerg­ing computational tools, but self-consistent experimental data sets on biologic mac­romolecules are rarely available in public domain. A second hurdle is that the use of antibody-based biologics as medicines is more recent when compared to the small molecule drugs. Therefore, the number of biologic drug products currently available in the market is small and our overall pharmaceutical development experience with these macromolecules is not as extensive as in thecase of the small molecule drug products. This situation is further complicated by the pervasive nature of empiri­cism inherent to both experimental and computational tools used to study biologic drug candidates, and with the arrival of novel biotherapeutic formats such as bi­specic and multi-specic antibodies. A third hurdle for the progress of biopharma­ceutical informatics is created if the computational tools are not available easily to everyone, and the underlying core programming codes are not published. A fourth hurdle is the general unavailability of the data in digital format in central reposito­ries within and outside the biopharmaceutical companies. The need for digital trans­formation of biopharmaceutical industry cannot be understated. However, digital transformation of biopharmaceutical industry is hampered by the lack of digitiza­tion of the experimental data. Overcoming the digitization hurdle that makes the experimental data available to computational biophysicists, data scientists, and machine learners is the foundational step toward digital transformation. Once this is accomplished, the scientists shall then be able to use the data and connect them with sequence and structural characteristics of biologic drug candidates. Discovery of correlations among “microscopic” (properties/descriptors derived from amino acid sequence, protein structure and molecular simulations) and “macroscopic” (experi­mental biochemical and biophysical experiments) attributes of biologic drug candi­dates will prove crucial for navigating developability of biologic drugs and will truly enable biopharmaceutical informatics. As summarized in this article, these are still early days for this eld as computational as well as hybrid studies containing computational and experimental data are beginning to emerge. Overall, this eld is expected to grow exponentially in the coming years.
14 Biopharmaceutical Informatics: A Strategic Vision for Discovering Developable…
425

References

1. Kaplon H, Chenoweth A, Crescioli S, Reichert JM (2022) Antibodies to watch in 2022. MAbs 14:2014296. https://doi.org/10.1080/19420862.2021.2014296
2. Yamaguchi S, Kaneko M, Narukawa M (2021) Approval success rates of drug candidates based on target, action, modality, application, and their combinations. Clin Transl Sci 14:1113–1122. https://doi.org/10.1111/cts.12980
3. Ecker DM, Jones SD, Levine HL (2015) The therapeutic monoclonal antibody market. MAbs 7:9–14. https://doi.org/10.4161/19420862.2015.989042
4. Bailly M, Mieczkowski C, Juan V etal (2020) Predicting antibody developability proles through early stage discovery screening. MAbs 12:1743053. https://doi.org/10.1080/1942086
2.2020.1743053
5. Saxena V, Panicucci R, Joshi Y, Garad S (2009) Developability assessment in pharmaceuti­cal industry: an integrated group approach for selecting developable candidates. J Pharm Sci 98:1962–1979. https://doi.org/10.1002/jps.21592
6. Xu Y, Wang D, Mason B etal (2018) Structure, heterogeneity and developability assessment of therapeutic antibodies. MAbs 11:1–26. https://doi.org/10.1080/19420862.2018.1553476
7. Bui LA, Hurst S, Finch GL etal (2015) Key considerations in the preclinical development of biosimilars. Drug Discov Today 20:3–15. https://doi.org/10.1016/j.drudis.2015.03.011
8. Vulto AG, Jaquez OA (2017) The process denes the product: what really matters in bio­similar design and production? Rheumatology 56:iv14–iv29. https://doi.org/10.1093/
rheumatology/kex278
9. Rathore AS, Winkle H (2009) Quality by design for biopharmaceuticals. Nat Biotechnol 27:26–34. https://doi.org/10.1038/nbt0109- 26
10. Carrara SC, Ulitzka M, Grzeschik J etal (2020) From cell line development to the formu­lated drug product: the art of manufacturing therapeutic monoclonal antibodies. Int J Pharm 594:120164. https://doi.org/10.1016/j.ijpharm.2020.120164
11. Jarasch A, Koll H, Regula JT etal (2015) Developability assessment during the selection of novel therapeutic antibodies. J Pharm Sci 104:1885–1898. https://doi.org/10.1002/jps.24430
12. Wang Q, Chen Y, Park J et al (2019) Design and production of bispecic antibodies. Antibodies 8:43. https://doi.org/10.3390/antib8030043
13. Furtmann N, Schneider M, Spindler N etal (2021) An end-to-end automated platform pro­cess for high-throughput engineering of next-generation multi-specic antibody therapeutics. MAbs 13:1955433. https://doi.org/10.1080/19420862.2021.1955433
14. Cocco P, Ayaz-Shah A, Messenger MP et al (2020) Target Product Proles for medical tests: a systematic review of current methods. BMC Med 18:119. https://doi.org/10.1186/
s12916- 020- 01582- 1
15. Baumann A (2006) Early development of therapeutic biologics– pharmacokinetics. Curr Drug Metab 7:15–21. https://doi.org/10.2174/138920006774832604
16. Thomas F (2019) Considering alternative dosage forms in biologics. BioPharm Int 32:16–18
17. Kumar S, Plotnikov NV, Rouse JC, Singh SK (2018) Biopharmaceutical Informatics: support­ing biologic drug development via molecular modelling and informatics. J Pharm Pharmacol 70:595–608. https://doi.org/10.1111/jphp.12700
18. Chiu ML, Goulet DR, Teplyakov A, Gilliland GL (2019) Antibody structure and function: the basis for engineering therapeutics. Antibodies 8:55. https://doi.org/10.3390/antib8040055
19. Kuroda D, Tsumoto K (2020) Engineering stability, viscosity, and immunogenicity of anti­bodies by computational design. J Pharm Sci 109:1631–1651. https://doi.org/10.1016/j.
xphs.2020.01.011
20. Tomar DS, Kumar S, Singh SK etal (2016) Molecular basis of high viscosity in concen­trated antibody solutions: strategies for high concentration drug product development. MAbs 8:216–228. https://doi.org/10.1080/19420862.2015.1128606
426
21. Hebditch M, Warwicker J (2019) Charge and hydrophobicity are key features in sequence­trained machine learning models for predicting the biophysical properties of clinical-stage antibodies. PeerJ 7:e8199. https://doi.org/10.7717/peerj.8199
22. Jain T, Sun T, Durand S etal (2017) Biophysical properties of the clinical-stage antibody landscape. Proc Natl Acad Sci USA 114:944–949. https://doi.org/10.1073/pnas.1616408114
23. Lecerf M, Kanyavuz A, Lacroix-Desmazes S, Dimitrov JD (2019) Sequence features of vari­able region determining physicochemical properties and polyreactivity of therapeutic anti­bodies. Mol Immunol 112:338–346. https://doi.org/10.1016/j.molimm.2019.06.012
24. Raybould MIJ, Marks C, Krawczyk K etal (2019) Five computational developability guide­lines for therapeutic antibody proling. Proc Natl Acad Sci USA 116:4025–4030. https://doi.
org/10.1073/pnas.1810576116
25. Starr CG, Tessier PM (2019) Selecting and engineering monoclonal antibodies with drug-like specicity. Curr Opin Biotechnol 60:119–127. https://doi.org/10.1016/j.copbio.2019.01.008
26. Zhang Y, Wu L, Gupta P etal (2020) Physicochemical rules for identifying monoclonal anti­bodies with drug-like specicity. Mol Pharm 17:2555–2569. https://doi.org/10.1021/acs.
molpharmaceut.0c00257
27. Grassi L, Cabrele C (2019) Susceptibility of protein therapeutics to spontaneous chemi­cal modications by oxidation, cyclization, and elimination reactions. Amino Acids 51:1409–1431. https://doi.org/10.1007/s00726- 019- 02787- 2
28. Alam ME, Barnett GV, Slaney TR etal (2019) Deamidation can compromise antibody colloi­dal stability and enhance aggregation in a pH-dependent manner. Mol Pharm 16:1939–1949.
https://doi.org/10.1021/acs.molpharmaceut.8b01311
29. Aswad DW, Paranandi MV, Schurter BT (2000) Isoaspartate in peptides and proteins: for­mation, signicance, and analysis. J Pharm Biomed Anal 21(6):1129–1136. https://doi.
org/10.1016/s0731- 7085(99)00230- 7
30. Geiger T, Clark S (1987) Deamidation, isomerization, and racemization at asparaginyl and aspartyl residues in peptides. J Biol Chem 262:785–794
31. Füssl F, Trappe A, Cook K etal (2018) Comprehensive characterisation of the heterogeneity of adalimumab via charge variant analysis hyphenated on-line to native high resolution Orbitrap mass spectrometry. MAbs 11:116–128. https://doi.org/10.1080/19420862.2018.1531664
32. Gervais D (2016) Protein deamidation in biopharmaceutical manufacture: understand­ing, control and impact. J Chem Technol Biotechnol 91:569–575. https://doi.org/10.1002/
jctb.4850
33. Yan B, Steen S, Hambly D etal (2009) Succinimide formation at Asn 55in the complemen­tarity determining region of a recombinant monoclonal antibody IgG1 heavy chain. J Pharm Sci 98:3509–3521. https://doi.org/10.1002/jps.21655
34. Rehder DS, Chelius D, McAuley A etal (2008) Isomerization of a single aspartyl residue of anti-epidermal growth factor receptor immunoglobulin γ2 antibody highlights the role avidity plays in antibody activity. Biochemistry 47:2518–2530. https://doi.org/10.1021/bi7018223
35. Wakankar AA, Borchardt RT, Eigenbrot C et al (2007) Aspartate isomerization in the complementarity- determining regions of two closely related monoclonal antibodies. Biochemistry 46:1534–1544. https://doi.org/10.1021/bi061500t
36. Cacia J, Keck R, Presta LG, Frenz J (1996) Isomerization of an aspartic acid residue in the complementarity-determining regions of a recombinant antibody to human IgE: identica­tion and effect on binding afnity. Biochemistry 35(6):1897–1903. https://doi.org/10.1021/
bi951526c
37. Harris RJ, Kabakoff B, Macchi FD etal (2001) Identication of multiple sources of charge heterogeneity in a recombinant antibody. J Chromatogr B Biomed Sci Appl 752:233–245.
https://doi.org/10.1016/s0378- 4347(00)00548- x
38. Sreedhara A, Cordoba A, Zhu Q etal (2012) Characterization of the isomerization products of aspartate residues at two different sites in a monoclonal antibody. Pharm Res 29:187–197.
https://doi.org/10.1007/s11095- 011- 0534- 2
J. Bauer et al.
14 Biopharmaceutical Informatics: A Strategic Vision for Discovering Developable…
39. Sakhnini LI, Greisen PJ, Wiberg C etal (2019) Improving the developability of an anti­gen binding fragment by aspartate substitutions. Biochemistry 58:2750–2759. https://doi.
org/10.1021/acs.biochem.9b00251
40. Zhang B, Yang Y, Yuk I etal (2008) Unveiling a glycation hot spot in a recombinant human­ized monoclonal antibody. Anal Chem 80:2379–2390. https://doi.org/10.1021/ac701810q
41. Wei B, Berning K, Quan C, Zhang YT (2017) Glycation of antibodies: modication, meth­ods and potential effects on biological functions. MAbs 9:586–594. https://doi.org/10.108
0/19420862.2017.1300214
42. Miller AK, Hambly DM, Kerwin BA etal (2011) Characterization of site-specic glyca­tion during process development of a human therapeutic monoclonal antibody. J Pharm Sci 100:2543–2550. https://doi.org/10.1002/jps.22504
43. Manning MC, Liu J, Li T, Holcomb RE (2018) Rational design of liquid formulations of pro­teins. Adv Protein Chem Struct Biol 112:1–59. https://doi.org/10.1016/bs.apcsb.2018.01.005
44. Oyama H, Koga H, Tadokoro T etal (2019) Relation of colloidal and conformational stabili­ties to aggregate formation in a monoclonal antibody. J Pharm Sci 109:308–315. https://doi.
org/10.1016/j.xphs.2019.10.038
45. Rizzo JM, Shi S, Li Y etal (2015) Application of a high-throughput relative chemical stabil­ity assay to screen therapeutic protein formulations by assessment of conformational sta­bility and correlation to aggregation propensity. J Pharm Sci 104:1632–1640. https://doi.
org/10.1002/jps.24408
46. Kalonia CK, Heinrich F, Curtis JE et al (2018) Protein adsorption and layer formation at the stainless steel–solution interface mediates shear-induced particle formation for an IgG1 monoclonal antibody. Mol Pharm 15:1319–1331. https://doi.org/10.1021/acs.
molpharmaceut.7b01127
47. Brader ML, Estey T, Bai S et al (2015) Examination of thermal unfolding and aggrega­tion proles of a series of developable therapeutic monoclonal antibodies. Mol Pharm 12:1005–1017. https://doi.org/10.1021/mp400666b
48. Fleischman ML, Chung J, Paul EP, Lewus RA (2017) Shipping-induced aggregation in thera­peutic antibodies: utilization of a scale-down model to assess degradation in monoclonal antibodies. J Pharm Sci 106:994–1000. https://doi.org/10.1016/j.xphs.2016.11.021
49. Nejadnik MR, Randolph TW, Volkin DB etal (2018) Post-production handling and adminis­tration of protein pharmaceuticals and potential instability issues. J Pharm Sci 107:2013–2019.
https://doi.org/10.1016/j.xphs.2018.04.005
50. Cromwell MEM, Hilario E, Jacobson F (2006) Protein aggregation and bioprocessing. AAPS J 8(3):E572–E579. https://doi.org/10.1208/aapsj080366
51. Yadav S, Laue TM, Kalonia DS etal (2012) The inuence of charge distribution on self­association and viscosity behavior of monoclonal antibody solutions. Mol Pharm 9:791–802.
https://doi.org/10.1021/mp200566k
52. Yadav S, Sreedhara A, Kanai S etal (2011) Establishing a link between amino acid sequences and self-associating and viscoelastic behavior of two closely related monoclonal antibodies. Pharm Res 28:1750–1764. https://doi.org/10.1007/s11095- 011- 0410- 0
53. Geng SB, Cheung JK, Narasimhan C etal (2014) Improving monoclonal antibody selec­tion and engineering using measurements of colloidal protein interactions. J Pharm Sci 103:3356–3363. https://doi.org/10.1002/jps.24130
54. Nishi H, Miyajima M, Nakagami H etal (2010) Phase separation of an IgG1 antibody solu­tion under a low ionic strength condition. Pharm Res 27:1348–1360. https://doi.org/10.1007/
s11095- 010- 0125- 7
55. Ghazvini S, Kalonia C, Volkin DB, Dhar P (2016) Evaluating the role of the air-solution interface on the mechanism of subvisible particle formation caused by mechanical agitation for an IgG1 mAb. J Pharm Sci 105:1643–1656. https://doi.org/10.1016/j.xphs.2016.02.027
56. Duerkop M, Berger E, Dürauer A, Jungbauer A (2018) Impact of cavitation, high shear stress and air/liquid interfaces on protein aggregation. Biotechnol J 13:1800062. https://doi.
org/10.1002/biot.201800062
427
428
57. Hauptmann A, Podgoršek K, Kuzman D etal (2018) Impact of buffer, protein concentration and sucrose addition on the aggregation and particle formation during freezing and thawing. Pharm Res 35:101. https://doi.org/10.1007/s11095- 018- 2378- 5
58. Koepf E, Eisele S, Schroeder R etal (2018) Notorious but not understood: how liquid­air interfacial stress triggers protein aggregation. Int J Pharm 537:202–212. https://doi.
org/10.1016/j.ijpharm.2017.12.043
59. Fesinmeyer RM, Hogan S, Saluja A etal (2009) Effect of ions on agitation- and temperature­induced aggregation reactions of antibodies. Pharm Res 26:903–913. https://doi.org/10.1007/
s11095- 008- 9792- z
60. Lapidoth GD, Baran D, Pszolla GM etal (2015) AbDesign: an algorithm for combinato­rial backbone design guided by natural conformations and sequences. Proteins Struct Funct Bioinf 83:1385–1406. https://doi.org/10.1002/prot.24779
61. Sirin S, Apgar JR, Bennett EM, Keating AE (2016) AB-Bind: antibody binding muta­tional database for computational afnity predictions. Protein Sci 25:393–409. https://doi.
org/10.1002/pro.2829
62. Robinson J, Barker DJ, Georgiou X etal (2019) IPD-IMGT/HLA database. Nucleic Acids Res 48:D948–D955. https://doi.org/10.1093/nar/gkz950
63. Swindells MB, Porter CT, Couch M etal (2017) abYsis: integrated antibody sequence and structure—management, analysis, and prediction. J Mol Biol 429:356–364. https://doi.
org/10.1016/j.jmb.2016.08.019
64. Kovaltsuk A, Leem J, Kelm S etal (2018) Observed antibody space: a resource for data min­ing next-generation sequencing of antibody repertoires. J Immunol 201:2502–2509. https://
doi.org/10.4049/jimmunol.1800708
65. Dunbar J, Krawczyk K, Leem J etal (2014) SAbDab: the structural antibody database. Nucleic Acids Res 42:D1140–D1146. https://doi.org/10.1093/nar/gkt1043
66. Berman HM, Westbrook J, Feng Z etal (2000) The Protein Data Bank. Nucleic Acids Res 1:235–242. https://doi.org/10.1093/nar/28.1.235
67. Ehrenmann F, Kaas Q, Lefranc M-P (2010) IMGT/3Dstructure-DB and IMGT/ DomainGapAlign: a database and a tool for immunoglobulins or antibodies, T cell recep­tors, MHC, IgSF and MhcSF.Nucleic Acids Res 38:D301–D307. https://doi.org/10.1093/
nar/gkp946
68. Tiller T, Schuster I, Deppe D etal (2013) A fully synthetic human Fab antibody library based on xed VH/VL framework pairings with favorable biophysical properties. MAbs 5:445–470. https://doi.org/10.4161/mabs.24218
69. Pan X, Kortemme T (2021) Recent advances in de novo protein design: principles, methods, and applications. J Biol Chem 296:100558. https://doi.org/10.1016/j.jbc.2021.100558
70. Amimeur T, Shaver JM, Ketchem RR etal (2020) Designing feature-controlled humanoid anti­body discovery libraries using generative adversarial networks. bioRxiv 2020.04.12.024844.
https://doi.org/10.1101/2020.04.12.024844
71. Baran D, Pszolla MG, Lapidoth GD etal (2017) Principles for computational design of binding antibodies. Proc Natl Acad Sci USA 114:10900–10905. https://doi.org/10.1073/
pnas.1707171114
72. Chowdhury R, Allan MF, Maranas CD (2018) OptMAVEn-2.0: de novo design of variable antibody regions against targeted antigen epitopes. Antibodies 7:23. https://doi.org/10.3390/
antib7030023
73. Nimrod G, Fischman S, Austin M etal (2018) Computational design of epitope-specic func­tional antibodies. Cell Rep 25:2121–2131.e5. https://doi.org/10.1016/j.celrep.2018.10.081
74. Sheng Z, Bimela JS, Katsamba PS etal (2022) Structural basis of antibody conformation and stability modulation by framework somatic hypermutation. Front Immunol 12:811632.
https://doi.org/10.3389/mmu.2021.811632
75. Liu X, Taylor RD, Grifn L etal (2017) Computational design of an epitope-specic Keap1 binding antibody using hotspot residues grafting and CDR loop swapping. Sci Rep 7:41306.
https://doi.org/10.1038/srep41306
J. Bauer et al.
14 Biopharmaceutical Informatics: A Strategic Vision for Discovering Developable…
76. Li W, Prabakaran P, Chen W etal (2016) Antibody aggregation: insights from sequence and structure. Antibodies 5:19. https://doi.org/10.3390/antib5030019
77. Roguska MA, Pedersen JT, Keddy CA etal (1994) Humanization of murine monoclonal antibodies through variable domain resurfacing. Proc Natl Acad Sci USA 91(3):969–973.
https://doi.org/10.1073/pnas.91.3.969
78. Townsend S, Fennell BJ, Apgar JR et al (2015) Augmented binary substitution: single­pass CDR germ-lining and stabilization of therapeutic antibodies. Proc Natl Acad Sci USA 112:15354–15359. https://doi.org/10.1073/pnas.1510944112
79. Singh S, Kroe-Barrett RR, Canada KA etal (2015) Selective targeting of the IL23 pathway: generation and characterization of a novel high-afnity humanized anti-IL23A antibody. MAbs 7:778–791. https://doi.org/10.1080/19420862.2015.1032491
80. Teplyakov A, Obmolova G, Malia TJ etal (2016) Structural diversity in a human antibody germline library. MAbs 8:1–19. https://doi.org/10.1080/19420862.2016.1190060
81. Choi Y, Hua C, Sentman CL et al (2015) Antibody humanization by structure-based computational protein design. MAbs 7:1045–1057. https://doi.org/10.1080/1942086
2.2015.1076600
82. Teplyakov A, Obmolova G, Malia TJ etal (2018) Structural insights into humanization of anti-tissue factor antibody 10H10. MAbs 10:269–277. https://doi.org/10.1080/1942086
2.2017.1412026
83. ULC CCG (2021) Molecular Operating Environment (MOE), 2019.01.1010 Sherbooke St. West, Suite #910, Montreal, QC, Canada, H3A 2R7
84. Olimpieri PP, Marcatili P, Tramontano A (2015) Tabhu: tools for antibody humanization. Bioinformatics 31:434–435. https://doi.org/10.1093/bioinformatics/btu667
85. Abhinandan KR, Martin ACR (2007) Analyzing the “degree of humanness” of antibody sequences. J Mol Biol 369:852–862. https://doi.org/10.1016/j.jmb.2007.02.100
86. Gao SH, Huang K, Tu H, Adler AS (2013) Monoclonal antibody humanness score and its applications. BMC Biotechnol 13:55. https://doi.org/10.1186/1472- 6750- 13- 55
87. Seeliger D (2013) Development of scoring functions for antibody sequence assessment and optimization. PLoS One 8:e76909. https://doi.org/10.1371/journal.pone.0076909
88. Lazar GA, Desjarlais JR, Jacinto J etal (2007) A molecular immunology approach to anti­body humanization and functional optimization. Mol Immunol 44:1986–1998. https://doi.
org/10.1016/j.molimm.2006.09.029
89. Choi Y, Verma D, Griswold KE, Bailey-Kellogg C (2017) EpiSweep: computationally driven reengineering of therapeutic proteins to reduce immunogenicity while maintaining function. Methods Mol Biol 1529:375–398
90. Clavero-Álvarez A, Mambro TD, Perez-Gaviro S et al (2018) Humanization of anti­bodies using a statistical inference approach. Sci Rep 8:14820. https://doi.org/10.1038/
s41598- 018- 32986- y
91. van der Kant R, Bauer J, Karow-Zwick AR etal (2019) Adaption of human antibody λ and κ light chain architectures to CDR repertoires. Protein Eng Des Sel 32:109–127. https://doi.
org/10.1093/protein/gzz012
92. Lehmann A, Wixted JHF, Shapovalov MV et al (2015) Stability engineering of anti­EGFR scFv antibodies by rational design of a lambda-to-kappa swap of the VL framework using a structure-guided approach. MAbs 7:1058–1071. https://doi.org/10.1080/1942086
2.2015.1088618
93. Mimoto F, Kuramochi T, Katada H etal (2016) Fc engineering to improve the function of therapeutic antibodies. Curr Pharm Biotechnol 17:1298–1314. https://doi.org/10.217
4/1389201017666160824161854
94. Lord DM, Bird JJ, Honey DM etal (2018) Structure-based engineering to restore high afn­ity binding of an isoform-selective anti-TGFβ1 antibody. MAbs 10:444–452. https://doi.
org/10.1080/19420862.2018.1426421
95. Somani S, Jo S, Thirumangalathu R etal (2021) Toward biotherapeutics formulation compo­sition engineering using site-identication by ligand competitive saturation (SILCS). J Pharm Sci 110:1103–1110. https://doi.org/10.1016/j.xphs.2020.10.051
429