University of Groningen Target-based drug discovery: from protein structure to small-molecules by MCR chemistry Wang, Yuanze

(1)

University of Groningen

Target-based drug discovery: from protein structure to small-molecules by MCR chemistry Wang, Yuanze

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2018

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Wang, Y. (2018). Target-based drug discovery: from protein structure to small-molecules by MCR chemistry. University of Groningen.

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

Target-based Drug Discovery:

From Protein Structure to

Small-Molecules by

MCR Chemistry

(3)

PhD Thesis

Target-based Drug Discovery: From Protein Structure to Small-Molecules by MCR Chemistry

The research presented in this thesis was performed in the research group of Drug Design of Groningen Research Institute of Pharmacy at the University of Groningen, the Netherlands. The author thanks the financial support from the Chinese Scholarship Council (CSC).

ISBN: 978-94-034-0958-0 (PDF without DRM) ISBN: 978-94-034-0959-7 (Printed book)

Layout: Yuanze Wang

Cover design: Yuanze Wang

Printed by: Ipskamp Printing

(4)

Target-based Drug Discovery: From

Protein Structure to

Small-Molecules by MCR Chemistry

PhD thesis

to obtain the degree of PhD at the University of Groningen

on the authority of the Rector Magnificus Prof. E. Sterken

and in accordance with

the decision by the College of Deans. This thesis will be defended in public on Monday 10 September 2018 at 09.00 hours

by

Yuanze Wang born on 5 April 1989

(5)

Supervisors

Prof. A.S.S. Dömling

Co-supervisors

Dr. M.R. Groves

Assessment Committee

Prof. P.H. Elsinga Prof. A.J. Minnaard Prof. L. Banfi

(6)

Chapter 1

(9)

1. Phenotypic-based Drug Discovery

Drug discovery can generally be viewed as a challenging process that includes the identification of active substances with a desirable therapeutic effect on certain diseases and optimization of those substances to increase their efficacy, pharmacokinetics and safety. Historically, phenotypic-based screening strategies, also known as classical pharmacology or forward pharmacology, were the main drug discovery paradigms used in both academic research centers and the pharmaceutical industry (Figure 1.1). The phenotypic approach could be specifically defined as the screening of a large number of randomly selected molecules in a system-based and target-agnostic approach that monitors the desirable change in phenotype. Thus, the active ingredient was obtained without any knowledge of the biological target. More often, an effort was made to identify the target after the active substance was identified. 1-3

Figure 1.1 Evolution of drug screening and lead discovery (adapted from reference 1).

2. Target-based Drug Discovery

Since the late 1980s, technological advances in molecular biology and sequencing of the human genome, which allowed rapid production of large quantities of biological targets initiated a new era of molecular target-based screening approach for modern drug discovery.

Target-based or hypothesis-based drug discovery is an approach in which small molecules are synthesized to target a known pathological/physiological pathway, which is hypothesized to be involved in the treatment of a particular disease. This idea came from the conclusion that a drug has to interact with biological macromolecules of the human body (enzymes, receptors, channels, etc.). Thus, biological macromolecules are used by the scientist in the lead discovery

(10)

screening, rational design and optimization. After that, the selected lead compounds were tested in cell and animal experiments to prove their efficacy. It was believed that this approach would result in an increased productivity and improved efficiency toward development of novel treatments for a validated target as it avoids the mass screening of banks of stored compounds. Moreover, according to a recent analysis, 78 of 113 first-in-class drugs approved by FDA from 1999 to 2013 were discovered through the target-based approach.4

2.1 High Throughput Screening

The drug-discovery process usually involves high throughput screening (HTS), wherein a large number of compound libraries are tested for identifying a few ‘active hits’ to modify the target. These ‘hits’ provide starting points for drug design and are then followed up to undergo several iterative screening runs to evaluate their properties, including off-target toxicity. The ‘validated hits’ are then used as ‘lead compounds’ for further studies. HTS has become one of the dominant paradigms for lead identification and has gained widespread popularity among both the academic researchers and industrial scientists over the last three decades.5_{The main goal of the HTS technique is to accelerate the drug discovery}

process and reduce the costs of drug development. Meanwhile, it is of vital importance that HTS also takes advantage of the changes in chemical synthesis strategy. It is obvious that combinatorial chemistry and parallel synthesis enable the efficient generation of a vast number of novel compounds.

With the increasing popularity of HTS, the technology required for HTS has also evolved rapidly. Using robotics, liquid handling devices, data processing software and high content imaging, an Ultra High-Throughput Screening process allows the researcher to conduct 100,000 assays per day. Initially, the libraries were arrayed into 96 micro-well plates, but now these have been replaced by high-density microplates with up to 3456 wells per plate.

The quality of the compounds in the HTS library can be assessed by several parameters. One of the most important parameters is the Lipinski's Rule of Five: the cLogP not greater than 5, molecular mass less than 500 Daltons, no more than 5 hydrogen bond donors and no more than 10 hydrogen bond acceptors. Some other parameters, such as lipophilicity and ligand efficiency are also used to assess the drug-likeness of the compounds.6

(11)

2.2 Fragment based drug discovery

HTS has considerably expanded the number of high quality lead molecules that could be evaluated for many drug discovery programs. However, the drug development process still suffered from high attrition rates as optimization of the screening hits could be problematic due to large size and inappropriate physical properties. Additionally, hit rates were generally low when some intractable biological targets were screened. In the context of the great demand to find small molecule drugs more efficiently, Fragment based drug discovery (FBDD) has emerged as a complementary strategy for our drug discovery arsenal.7-11As a new promising approach, FBDD has several advantages. In FBDD, each atom of the fragment is involved in the desired binding interactions with the target. In other words, although the binding affinity of the fragment is much weaker than the HTS hit, it actually forms interactions with higher quality. Thus, the optimization of the fragment into a higher affinity lead is much easier than optimizing the HTS hit which contains some mismatched groups. In addition, the size and other physical properties such as lipophilicity and polar surface area can be better controlled.

Obviously, the successful design of a fragment library is the basis of the fragment-based approach.12_{The library should be designed on the basis of several}

considerations such as sample relevant chemical space, appropriate complexity, synthetic tractability, fragment hit rate and knowledge of binding target. More importantly, fragments in the library typically obey a rule of three (RO3): molecular mass < 300 Da, the number of hydrogen bond donors < 3, the number of hydrogen bond acceptors < 3 and cLogP ≤ 3. Another consideration is the number of the fragments included in the library, normally on the order of 103_,

because these fragments cover better chemical space than the compounds included in the HTS libraries.

As the fragment normally forms relatively weak interaction with the target macromolecules (100 µM to 10 mM range), sensitive biophysical techniques need to be used. According to a report by S. D. Pickett and coworkers, surface plasmon resonance (SPR), NMR spectroscopy and fluorescence-based thermal shift assay are the most popular techniques used for fragment-based drug discovery, followed by X-ray crystallography, mass spectrometry (MS) and isothermal titration calorimetry (ITC). Other useful techniques include capillary electrophoresis (CE), biolayer interferometry (BLI) and microscale thermophoresis (MST). In our lab,

(12)

the fragment screening strategy involves a three-stage cascade of biophysical screens: 1) preliminary screening using medium throughput technique: differential scanning fluorimetry (DSF), 2) validation of the binding of fragment hits with target protein by ITC or MST, 3) characterization of the fragment-target interaction by X-ray crystallography (Figure 1.2).

Figure 1.2 Three-stage cascade of biophysical screens for fragment based drug discovery.

To improve the potency of the validated fragment hits, three main approaches have been successfully established with the knowledge of the fragment binding mode: fragment linking, fragment merging and fragment growing.13Fragment linking requires covalently joining two or more fragments that are known to bind in different but close pockets to form a novel-scaffold ligand on the same target. Fragment merging involves the incorporation of the structural features of other overlapping ligand or substrates into the fragment. Fragment growing refers to the modification of the fragment by chemical synthesis to incorporate more interactions. More frequently, fragment growing was used as the preferred fragment-optimization strategy, as fragment linking and merging are more challenging.

(13)

3. Proteins as drug target

Target-based drug discovery normally starts with the identification and validation of a drug target which can be used for lead screening. It is very important to have enough evidence to support the drug target selected before it is used for the drug discovery. A good drug target includes three core factors: 1) it participates in a crucial biological process linked to diseases; 2) it contains binding sites which can interact with drug-like molecules; 3) its structure and function has been clearly characterized and is different with known targets. The vast majority of the approved drugs use proteins as their targets. In a recent survey, the number of protein targets for all classes of available marketed drugs was found to be 324, including both human proteins and proteins from pathogenic organisms.14 In terms of potential drug targets, it was estimated that there are 2000-3000 druggable proteins in humans.

3.1 Protein expression and purification

The structural biology studies, such as the ligand-protein binding assay and the X-ray crystallization experiment, need large amount of proteins. In general, for the production of recombinant protein, Escherichia coli (E. coli) is used as the preferred host because it has several advantages: 1) simple and convenient; 2) rapid and cheap; 3) labeling methods for NMR are well established; 4) a wide variety of commercial products are available. The protein expression starts with introducing the plasmid that encodes the protein of interest into the E. coli cell, then the expression is induced and the presence of the protein is finally checked by SDS-PAGE analysis. It is important to establish the optimal growth and expression conditions with small-scale cultures before the protein is produced in large-scale. Most recombinant proteins can be expressed at high levels in E. coli when the host strain and vectors are carefully chosen, and the culture conditions are properly controlled.

If the produced protein is stable and soluble, the purification can be easily achieved by the affinity tag column, ion-exchange chromatography and size exclusion chromatography. However, many polypeptide gene products are expressed in an insoluble form that lack functional activity. This happens especially when the protein is expressed at high levels, leading to aggregation and formation of inclusion bodies. The formation of inclusion bodies can also be influenced by the vector and host cell selected, the nature of the protein, as well as the growth and induction conditions. The inclusion bodies need to be purified

(14)

under denaturing conditions instead of standard purification methods, which are based on the protein’s native solubility. It is usually necessary to renature the inclusion bodies to reconstruct its three dimensional structure before it is used for structural studies. In this case, the refolding experiments need to be designed empirically for each individual protein.15-18

3.2 Protein crystallization

The understanding of the molecular mechanism of the protein function by its three-dimensional structure is a very important tool for the modern biotechnological research.19 Protein crystallization is one of the most powerful ways for structure determination, which have significant impact on the rational drug design. Therefore, the field of protein crystallography has undergone an enormous expansion in recent decades, which was indicated by the rapid growth of protein structures deposited in the Protein Data Bank (PDB).

To crystallize a protein, we first need to purify large quantities of proteins (5-50 mg) with high purity and homogeneity. There are several methods to check the quality of the protein. The CD spectroscopy can help us to detect if the protein is correctly folded. The presence of aggregates could be determined by techniques such as dynamic light scattering (DLS). These techniques are crucial because sometimes the unfolded protein might also exist in a soluble form. In addition, additives like salts, metals and co-factors are sometimes necessary for the protein stabilization, which can be identified by the established DSF buffer screening methods. It is vital to keep in mind that protein quality is one of the key factors for the success of the crystallization.

Even when soluble protein with good quality is available, finding crystallization conditions for a new protein is still sometimes like searching for a needle in a haystack, because it is a complex and multi-parametric process. The most basic factors for crystallization include protein concentration, the presence of salt, precipitate and the pH of the buffer. Among these factors, the role of the precipitate is to bind with water molecules in order to bring the solution into supersaturation. Ammonium sulfate and polyethylene glycol are two commonly used precipitates.

From the crystallization phase diagram, we can see that there are four areas distinguished (Figure 1.3). The precipitation zone, where the protein is super

(15)

.

Figure 1.3 Protein crystallization phase diagram (adapted from reference 19).

nucleation will not take plate but it is good for crystals to grow; the metastable zone is the best area, where large and well-ordered crystals might form and the under-saturated zone is an area, where the protein will never crystallize, because the concentration is too low. The process of the protein crystallization proceeds in two steps: the formation of nuclei and crystal growth. In an ideal experiment, once a critical nucleus has formed, the system might go into the metastable zone naturally as the protein concentration drops. If this is not the case, there are several methods like (i) microbatch, (ii) vapor diffusion, (iii) dialysis and (iv) free interface diffusion which can help to reach the nucleation zone and the metastable zone.

4. Small molecules synthesized by MCR chemistry

4.1 Ugi reaction

Multicomponent reactions (MCRs) are generally defined as a one-pot process during which three or more starting materials react in a single chemical step to form a product that incorporates all or most of the atoms of the reactants. Through the years, MCRs have attracted considerable interest in organic chemistry due to their convergence, atom economy, efficiency, shortened reaction time and

(16)

simplicity. Meanwhile, MCRs have also met renewed interest for medicinal chemists, because they are also powerful synthetic strategies for the construction of biologically interesting scaffolds with huge chemical diversity and molecular complexity. Among MCRs, the Ugi four-component reaction (Ugi-4CR) is one of the most widely used reactions which based on the peculiar reactivity of isocyanides. The Ugi-4CR involves a condensation of a carbonyl component (aldehyde or ketone), an amine, a carboxylic acid and an isonitrile to afford the peptide-like α-acylaminoamides with a newly created stereogenic center.

4.2 Mechanism of Ugi reaction

After the pioneer work of Ivar Ugi on the discovery of the Ugi reaction in 1959, the first mechanistic proposal was postulated by Ugi himself.20_{Almost all the}

synthetic developments around this reaction rely on this mechanistic assumption as depicted in Scheme 1.1, path A. The initial step of the reaction is the formation of an imine 3 by condensation of the amine 1 with the oxo-component (aldehydes/ketones) 2 with loss of one equivalent of water. The imine 3 is then activated by proton exchanging with carboxylic acid 4. Subsequent nucleophilic addition of the isocyanide 6 with its terminal carbon to the iminium ion 5 forms intermediate nitrilium ion 7. This intermediate then undergoes a second nucleophilic addition by the carboxylate anion gives intermediate imidate 9. The last step is an irreversible Mumm rearrangement with transfer of the acyl imidate to generate the final product 10. In this mechanism, all the reaction sequence are in equilibrium except for the last step, which is considered as the driving force for the total reaction sequence.

(17)

Recently, an alternative mechanism raised the debate by proposing the formation of hemiaminal as a key intermediate (See Scheme 1.1, path B). The first step of this pathway is still the formation of imine which is activated by acid component. However, instead of isocyanide addition, the carboxylate anion first attacks the iminium ion to form intermediate hemiaminal 8, which is followed by isocyanide insertion to furnish the same imidate intermediate 9.

Scheme 1.2 The Ugi reaction started form electrophilic iminium species

(Structure adapted form reference 21-24)

The formation of the imine as the first step was generally admitted in both mechanisms. This was also demonstrated by the direct use of electrophilic iminium species in the Ugi reaction (Scheme 1.2). In 1982, Nutt and co-workers first reported the use of substituted 1-pyrrolines instead of the amine and aldehyde components to produce substituted prolyl peptides by an Ugi-type three-component reaction (U-3CR).21In 2004, N-acylazinium salt formation in situ was reported by Lavilla and co-workers as a new source of iminium ion equivalents.22

In 2007, the use of the 3,4-dihydroisoquioline dehydrogenated by in situ oxidation of the corresponding secondary amine tetrahydroisoquinoline in U-3CR was explored by Zhu and co-workers.23 Recently, Orru’s group reported stereoselective synthesis of highly functionalized, optically pure 3,4-substituted prolyl peptides by Ugi reaction starting from optically active 1-pyrrolines.24

(18)

Scheme 1.3 Imidate intermediate isolated from Ugi reaction

(Structure adapted form reference 25-27)

In fact, the existence of imidate intermediate was also proved by several studies (Scheme 1.3). The first successful isolation of the imidate 11 and 12 has been reported by Ugi in 1971.25_{In a research to explore thiols in the Ugi–Smiles}

reaction, Barthelon and co-workers surprisingly obtained thioimidate 13 in good yields when using methyl mercaptosalicylate as acid component.26In 2009, Faggi and co-workers isolated a stable imidate 14 in the tautomeric enediamine form, which was further demonstrated by X-ray crystallography.27

Figure 1.4 Energy profile of intermediate in Ugi reaction

(Structure adapted form reference 28)

(19)

including ZPE corrections, based on Ugi’s proposal presented in Scheme 1.1, path A.28 Both Ugi-Mumm and Ugi-smiles reaction were studied with a realistic model and the calculation was computed separately in two solvents: methanol and toluene. According to the energy profile, optimization of the transition state (TS) for the insertion of the isocyanide in the hemiaminal leads to the TS of subsequent isocyanide addition to the iminium. Thus, path B proceeds through hemiaminal first fragmentation into the iminium and carboxylate anion and then the isocyanide addition. This was also confirmed by the intrinsic reaction coordinate (IRC) approach, which is a valuable tool to check if the given transition state is the expected transition state for the reaction of interest. Therefore, path A in Scheme 1.1 was considered as the privileged mechanistic pathway for the Ugi reaction. More importantly, two commonly accepted features in the Ugi mechanistic pathway were challenged by this theoretical research. Firstly, although the formation of imine as the first step is obviously confirmed, the involvement of iminium during isocyanide insertion was questioned by their calculation. Instead of activation of imine by a proton transfer, their computed results suggested that the formation of a hydrogen bonded complex between imine and acid substrate is more favorable. This conclusion was also in accordance with a recent NMR experimental study in imine activation reported by Fleischmann and co-workers. Secondly, according to the energy profile, the imidate solvated by a methanol dimer 9a lies at -33.6 kcal mol-1_{, whereas the highest TS 15 for}

Mumm rearrangement lies at -32.5 kcal mol-1_{, which means this final}

rearrangement can evolve easily with a barrier of only 0.9 kcal mol-1_{(Figure 1.4).}

On the contrary, in the isocyanide addition step, it requires 19.8 kcal mol-1 of activation energy to reach a stable nitrilium-acetate ion pair. Therefore, they proposed that Ugi reaction should not be considered as an equilibratory reaction sequence driven by a final irreversible step as previously stated. Instead, the isocyanide addition step where a new stereogenic center was formed was the only rate-determining step.

In 2014, the mechanism of the Ugi reaction was investigated by Eberlin and co-workers using electrospray ionization mass spectrometry (ESI-MS) with alternatively two imidazolium charge-tagged reagents (a carboxylic acid or an amine).29 The Ugi’s original mechanistic proposal (Scheme1.1, path A) was consolidated as the key intermediate 7 has been isolated and characterized. In addition, the energetics of the final Mumm rearrangement were calculated by Density Functional Theory (DFT) studies. It predicted a very low energy barrier from transient imidate 9 to the final product 10, which is consistent with both

(20)

ESI-MS/MS and TWIM-MS data. In the same year, Angelis and co-workers developed finely selected reaction conditions for ESI-MS characterization of Ugi reaction mechanism which can avoid the impact of ion tagged reactants on the reaction pathway. Remarkably, their data demonstrated that the formation of nitrilium ion

7 is kinetically favored and its formation is the rate-determining step. This

conclusion not only strongly supported the original hypothesis of Ugi, but also agreed with the theoretical findings predicted by Fleurat-Lessard and co-workers30

4.3 Ugi reaction and its post-cyclizations

Although Ugi reaction stands out as a powerful method for the construction of compounds with great diversity, the backbone of these compounds are normally linear which thus lack the conformational constriction. In this context, the combination of Ugi reaction with a subsequent secondary transformation, typically a ring-forming process, has been proven to be an extremely powerful strategy for the synthesis of structurally diverse complex molecules, especially highly functionalized heterocyclic compounds. Inevitably, a large variety of reactions has been introduced for the post-Ugi transformations strategy such as acid/base-catalyzed cyclizations, cycloadditions, condensations, Ugi-deprotection-cyclization (UDC), SNAr reactions, macrolactonizations, SN2

reactions, aryl couplings, ring closing metathesis, radical cyclization etc. Therefore, all kinds of functionalized skeletons have been successfully constructed by this two-step procedure, as concluded in Scheme 1.4.

(21)

Scheme 1.4 Functionalized skeletons obtained by Ugi reaction and its post-cyclization.

(22)

5. Aim and scope of this thesis

Refolding of proteins derived from inclusion bodies is very promising as it can provide a reliable source of target proteins of high purity. However, inclusion body-based protein production is often limited by the lack of techniques for the detection of correctly refolded protein. Thus, the selection of the refolding conditions is mostly achieved using trial and error approaches and is thus a time-consuming process. In chapter 2, we use the latest developments in the differential scanning fluorimetry guided refolding approach as an analytical method to detect correctly refolded protein. We describe a systematic buffer screen that contains a 96-well primary pH-refolding screen in conjunction with a secondary additive screen. Our research demonstrates that this approach could be applied for determining refolding conditions for several proteins. In addition, it revealed which “helper” molecules, such as arginine and additives are essential. In chapter 3, we describe the fast and efficient synthesis of libraries of positional isomeric 1H-tetrazoles and 5H-tetrazoles for the purpose of testing binding hypothesis of isomeric tetrazoles in fragment-based drug discovery.

Isocyanide-based multicomponent reactions (IMCR) are by far the most versatile reactions that can construct relatively complex molecules by one-pot synthesis. More importantly, the development of post IMCR modifications significantly improves the scaffold’s diversity. In chapter 4, we describe the use of N-Boc protected hydrazine together with -amino acid derived isocyanides in the Ugi-tetrazole reaction and its post-cyclization under both acidic and basic conditions. The cyclization in acidic conditions was conducted in a one pot fashion, which give 7-aminotetrazolopyrazinone and tetrazolotriazepinone cyclic products. The post cyclization under basic conditions could selectively afford Boc-protected 7-aminotetrazolopyrazinone products in yield from 38 to 87%.

In chapter 5, the Pomeranz-Fritsch reaction was for the first time successfully applied in the Ugi post-cyclization strategy by using aminoacetaldehyde diethyl acetal and electron rich aldehydes as starting materials. Isoquinoline derivatives and benzo[d]azepinone scaffolds with great diversity were constructed in moderate to good yield. In addition, the isoquinoline-tetrazoles and an alkaloid-like tetrazole-fused tetracyclic compound were synthesized by this method in a very efficient manner.

(23)

dispensing-enabled scouting enabled a pipeline of fast and efficient nL scale scouting to mg to gram scale synthesis. Isoquinoline derivatives and heterocyclic compounds were constructed by this method in moderate to excellent yield. These synthetic approaches were unprecedented, simple and efficient. Meanwhile, the substrate scope and functional group tolerance were proved to be exceptional.

(24)

References

[1]Zheng, W., Thorne, N., & McKew, J. C. (2013). Phenotypic screens as a renewed approach for drug discovery.Drug discovery today,18(21-22), 1067-1073.

[2]Swinney, D. C. (2013). Phenotypic vs. target-based drug discovery for first-in-class medicines. Clinical Pharmacology & Therapeutics,93(4), 299-301.

[3]Brown, D. (2007). Unfinished business: target-based drug discovery.Drug Discovery Today,

12(23-24), 1007-1012.

[4]Eder, J., Sedrani, R., & Wiesmann, C. (2014). The discovery of first-in-class drugs: origins and evolution.Nature Reviews Drug Discovery,13(8), 577.

[5]Hajare, A. A., Salunkhe, S. S., Mali, S. S., Gorde, S. S., Nadaf, S. J., & Pishawikar, S. A. (2013). Review on: high-throughput screening is an approach to drug discovery.Am. J.

PharmTech. Res,4, 112-129.

[6]Lipinski, C. A. (2004). Lead-and drug-like compounds: the rule-of-five revolution.Drug

Discovery Today: Technologies,1(4), 337-341.

[7]Chessari, G., & Woodhead, A. J. (2009). From fragment to clinical candidate - a historical perspective. Drug discovery today,14(13-14), 668-675.

[8]Murray, C. W., & Rees, D. C. (2009). The rise of fragment-based drug discovery.Nature

chemistry, 1(3), 187.

[9]Sledz, P., Silvestre, H. L., Hung, A. W., Ciulli, A., Blundell, T. L., & Abell, C. (2010). Optimization of the interligand Overhauser effect for fragment linking: application to inhibitor discovery against Mycobacterium tuberculosis pantothenate synthetase. Journal

of the American Chemical Society, 132(13), 4544-4545.

[10]Scott, D. E., Coyne, A. G., Hudson, S. A., & Abell, C. (2012). Fragment-based approaches in drug discovery and chemical biology. Biochemistry, 51(25), 4990-5003.

[11]Mashalidis, E. H., Śledź, P., Lang, S., & Abell, C. (2013). A three-stage biophysical screening cascade for fragment-based drug discovery.Nature protocols,8(11), 2309. [12]Keserű, G. M., Erlanson, D. A., Ferenczy, G. G., Hann, M. M., Murray, C. W., & Pickett,

S. D. (2016). Design principles for fragment libraries: maximizing the value of learnings from pharma fragment-based drug discovery (FBDD) programs for use in academia.Journal

of medicinal chemistry,59(18), 8189-8206.

[13]Hung, A. W., Silvestre, H. L., Wen, S., Ciulli, A., Blundell, T. L., & Abell, C. (2009). Application of fragment growing and fragment linking to the discovery of inhibitors of Mycobacterium tuberculosis pantothenate synthetase. Angewandte Chemie International

Edition, 48(45), 8452-8456.

[14] Bakheet, T. M., & Doig, A. J. (2009). Properties and identification of human protein drug targets. Bioinformatics,25(4), 451-457.

[15]Yang, Z., Zhang, L., Zhang, Y., Zhang, T., Feng, Y., Lu, X., ... & Wang, X. (2011). Highly efficient production of soluble proteins from insoluble inclusion bodies by a two-step-denaturing and refolding method. PloS one, 6(7), e22981.

[16]Baneyx, F., & Mujacic, M. (2004). Recombinant protein folding and misfolding in Escherichia coli. Nature biotechnology, 22(11), 1399.

[17]Hartl, F. U., & Hayer-Hartl, M. (2009). Converging concepts of protein folding in vitro and in vivo. Nature Structural and Molecular Biology, 16(6), 574.

[18]Biter, A. B., Andres, H., Thapar, R., Lin, J. Z., & Phillips, K. J. (2016). DSF guided refolding as a novel method of protein production. Scientific reports, 6, 18906.

[19]Chayen, N. E., & Saridakis, E. (2008). Protein crystallization: from purified protein to diffraction-quality crystal. Nature methods, 5(2), 147.

(25)

[20] Ugi, I.; Meyr, R.; Fetzer, U.; Steinbrückner, C. (1959). Angewandte Chemie International

Edition, 71, 386-386.

[21]Nutt, R. F., & Joullie, M. M. (1982). Four-component condensation: a new versatile method for the synthesis of substituted prolyl peptides. Journal of the American Chemical Society,

104(21), 5852-5853.

[22]Díaz, J. L., Miguel, M., & Lavilla, R. (2004). N-Acylazinium salts: A new source of iminium ions for Ugi-type processes. The Journal of organic chemistry, 69(10), 3550-3553. [23]Ngouansavanh, T., & Zhu, J. (2007). IBX‐Mediated Oxidative Ugi‐Type Multicomponent Reactions: Application to the N and C1 Functionalization of Tetrahydroisoquinoline.

Angewandte Chemie International Edition, 46(30), 5775-5778.

[24]Znabet, A., Zonneveld, J., Janssen, E., De Kanter, F. J., Helliwell, M., Turner, N. J., ... & Orru, R. V. (2010). Asymmetric synthesis of synthetic alkaloids by a tandem biocatalysis/Ugi/Pictet-Spengler-type cyclization sequence. Chemical Communications,

46(41), 7706-7708.

[25]Gokel, G., & Ludke, G. (1971). I. Ugi in Isonitrile Chemistry.

[26]Barthelon, A., El Kaim, L., Gizolme, M., & Grimaud, L. (2008). Thiols in Ugi‐and Passerini–Smiles‐Type Couplings. European Journal of Organic Chemistry, 35, 5974-5987. [27]Faggi, C., García-Valverde, M., Marcaccini, S., & Menchi, G. (2010). Isolation of Ugi

four-component condensation primary adducts: A straightforward route to isocoumarins.

Organic letters, 12(4), 788-791.

[28]Chéron, N., Ramozzi, R., Kaïm, L. E., Grimaud, L., & Fleurat-Lessard, P. (2012). Challenging 50 years of established views on Ugi reaction: A theoretical approach. The

Journal of organic chemistry, 77(3), 1361-1366.

[29]Medeiros, G. A., da Silva, W. A., Bataglion, G. A., Ferreira, D. A., de Oliveira, H. C., Eberlin, M. N., & Neto, B. A. (2014). Probing the mechanism of the Ugi four-component reaction with charge-tagged reagents by ESI-MS (/MS). Chemical Communications, 50(3), 338-340.

[30]Ramozzi, R., Chéron, N., El Kaim, L., Grimaud, L., & Fleurat‐Lessard, P. (2014). Predicting New Ugi–Smiles Couplings: A Combined Experimental and Theoretical Study.

Chemistry-A European Journal, 20(29), 9094-9099.

[31]Zohreh, N., Alizadeh, A., Bijanzadeh, H. R., & Zhu, L. G. (2010). Novel approach to 1, 5-benzodiazepine-2-ones containing peptoid backbone via one-pot diketene-based ugi-4CR.Journal of combinatorial chemistry,12(4), 497-502.

[32]Banfi, L., Basso, A., Guanti, G., Kielland, N., Repetto, C., & Riva, R. (2007). Ugi multicomponent reaction followed by an intramolecular nucleophilic substitution: convergent multicomponent synthesis of 1-sulfonyl 1, 4-diazepan-5-ones and of their benzo-fused derivatives.The Journal of organic chemistry,72(6), 2151-2160.

[33] Azuaje, J., Pérez-Rubio, J. M., Yaziji, V., El Maatougui, A., González-Gomez, J. C., Sánchez-Pedregal, V. M., ... & Sotelo, E. (2015). Integrated Ugi-based assembly of functionally, skeletally, and stereochemically diverse 1, 4-benzodiazepin-2-ones.The

Journal of organic chemistry,80(3), 1533-1549.

[34] Gunawan, S., Ayaz, M., De Moliner, F., Frett, B., Kaiser, C., Patrick, N., ...& Hulme, C. (2012). Synthesis of tetrazolo-fused benzodiazepines and benzodiazepinones by a two-step protocol using an Ugi-azide reaction for initial diversity generation.Tetrahedron,68(27-28), 5606-5611.

[35]Mossetti, R., Saggiorato, D., & Tron, G. C. (2011). Exploiting the Acylating Nature of the Imide-Ugi Intermediate: A Straightforward Synthesis of Tetrahydro-1, 4-benzodiazepin-2-ones. The Journal of organic chemistry, 76(24), 10258-10262.

(26)

[36] Keating, T. A., & Armstrong, R. W. (1996). A remarkable two-step synthesis of diverse 1, 4-benzodiazepine-2, 5-diones using the Ugi four-component condensation.The Journal of

organic chemistry,61(25), 8935-8939.

[37] Sañudo, M., García-Valverde, M., Marcaccini, S., Delgado, J. J., Rojo, J., & Torroba, T. (2009). Synthesis of benzodiazepine β-turn mimetics by an Ugi 4CC/Staudinger/Aza-Wittig sequence. Solving the conformational behavior of the Ugi 4CC adducts.The Journal of

organic chemistry,74(5), 2189-2192.

[38] Xiang, Z., Luo, T., Lu, K., Cui, J., Shi, X., Fathi, R., ...& Yang, Z. (2004). Concise synthesis of isoquinoline via the Ugi and Heck reactions.Organic Letters,6(18), 3155-3158. [39] Che, C., Li, S., Yu, Z., Li, F., Xin, S., Zhou, L., ...& Yang, Z. (2013). One-pot syntheses of isoquinolin-3-ones and benzo-1, 4-diazepin-2, 5-diones utilizing Ugi-4CR post-transformation strategy.ACS combinatorial science,15(4), 202-207.

[40] Tyagi, V., Khan, S., Giri, A., Gauniyal, H. M., Sridhar, B., & Chauhan, P. M. (2012). A Ligand-Free Pd-Catalyzed Cascade Reaction: An Access to the Highly Diverse Isoquinolin-1 (2 H)-one Derivatives via Isocyanide and Ugi-MCR Synthesized Amide Precursors.Organic letters,14(12), 3126-3129.

[41]Ma, Z., Xiang, Z., Luo, T., Lu, K., Xu, Z., Chen, J., & Yang, Z. (2006). Synthesis of functionalized quinolines via Ugi and Pd-catalyzed intramolecular arylation reactions.Journal of combinatorial chemistry,8(5), 696-704.

[42] Li, Y., Meng, J. P., Lei, J., Chen, Z. Z., Tang, D. Y., Zhu, J., ... & Xu, Z. G. (2017). Efficient synthesis of fused oxazepino-isoquinoline scaffolds via an Ugi, followed by an intramolecular cyclization.ACS combinatorial science,19(5), 324-330.

[43] Ji, F., Yi, W. B., Sun, M., Lv, M. F., &Cai, C. (2013). Synthesis of novel isoquinolinone and 1, 2-dihydroisoquinoline scaffolds via Ugi reaction and ring opening reaction of furans.Molecular diversity,17(2), 295-305.

[44] Pandey, S., Khan, S., Singh, A., Gauniyal, H. M., Kumar, B., & Chauhan, P. M. (2012). Access to Indole-And Pyrrole-Fused Diketopiperazines via Tandem Ugi-4CR/Intramolecular Cyclization and Its Regioselective Ring-Opening by Intermolecular Transamidation.The Journal of organic chemistry,77(22), 10211-10227.

[45] Golebiowski, A., Jozwik, J., Klopfenstein, S. R., Colson, A. O., Grieb, A. L., Russell, A. F., ... & Chen, J. J. (2002). Solid-supported synthesis of putative peptide β-turn mimetics via Ugi reaction for diketopiperazine formation.Journal of combinatorial chemistry,4(6), 584-590.

[46] Kennedy, A. L., Fryer, A. M., & Josey, J. A. (2002). A new resin-bound universal isonitrile for the Ugi 4CC reaction: preparation and applications to the synthesis of 2, 5-diketopiperazines and 1, 4-benzodiazepine-2, 5-diones.Organic letters,4(7), 1167-1170. [47] Cho, S., Keum, G., Kang, S. B., Han, S. Y., & Kim, Y. (2003). An efficient synthesis of 2,

5-diketopiperazine derivatives by the Ugi four-center three-component reaction.Molecular

diversity,6(3-4), 283-286.

[48] Stucchi, M., Cairati, S., Cetin-Atalay, R., Christodoulou, M. S., Grazioso, G., Pescitelli, G., ...& Lesma, G. (2015). Application of the Ugi reaction with multiple amino acid-derived components: synthesis and conformational evaluation of piperazine-based minimalist peptidomimetics.Organic & biomolecular chemistry,13(17), 4993-5005.

[49] Santra, S., & Andreana, P. R. (2011). A rapid, one-pot, microwave-influenced synthesis of spiro-2, 5-diketopiperazines via a cascade Ugi/6-exo-trig aza-Michael reaction.The Journal

of organic chemistry,76(7), 2261-2264.

[50] Abdessalem, A. B., Abderrahim, R., Agrebie, A., Dos Santos, A., El Kaïm, L., & Komesky, A. (2014). Formal [3+2] cycloaddition of Ugi adducts towards pyrrolines.Chemical

(27)

[51] Moni, L., Banfi, L., Basso, A., Carcone, L., Rasparini, M., & Riva, R. (2015). Ugi and Passerini reactions of biocatalytically derived chiral aldehydes: application to the synthesis of bicyclic pyrrolidines and of antiviral agent telaprevir.The Journal of organic

chemistry,80(7), 3411-3428.

[52] Banfi, L., Basso, A., Cerulli, V., Guanti, G., & Riva, R. (2008). Polyfunctionalized Pyrrolidines by Ugi Multicomponent Reaction Followed by Palladium-Mediated SN2 ‘Cyclizations.The Journal of organic chemistry,73(4), 1608-1611.

[53] Polindara-García, L. A., & Miranda, L. D. (2012). Two-step synthesis of 2, 3-dihydropyrroles via a formal 5-endo cycloisomerization of Ugi 4-CR/propargyl adducts.Organic letters,14(21), 5408-5411.

[54] C Polindara-García, L. A., & Vazquez, A. (2014). Combinatorial synthesis of nicotine analogs using an Ugi 4-CR/cyclization-reduction strategy.Organic & biomolecular

chemistry,12(36), 7068-7082.

[55]Znabet, A., Blanken, S., Janssen, E., de Kanter, F. J., Helliwell, M., Turner, N. J., ... & Orru, R. V. (2012). Stereoselective synthesis of N-aryl proline amides by biotransformation-Ugi-Smiles sequence.Organic & biomolecular chemistry,10(5), 941-944.

[56] Gulevich, A. V., Balenkova, E. S., & Nenajdenko, V. G. (2007). The first example of a diastereoselective thio-Ugi reaction: A new synthetic approach to chiral imidazole derivatives.The Journal of organic chemistry,72(21), 7878-7885.

[57] Yan, Y. M., Rao, Y., & Ding, M. W. (2016). One-Pot Synthesis of Multisubstituted Benzimidazoles via Sequential Ugi and Catalytic Aza-Wittig Reaction Starting from 2-Aminobenzoyl Azides.The Journal of organic chemistry,81(3), 1263-1268.

[58] Guchhait, S. K., & Chaudhary, V. (2014). Desilylative activation of TMSCN in chemoselective Strecker-Ugi type reaction: functional fused imidazoles as building blocks as an entry route to annulated purines.Organic & biomolecular chemistry,12(34), 6694-6705.

[59] Tyagi, V., Khan, S., Bajpai, V., Gauniyal, H. M., Kumar, B., & Chauhan, P. M. (2012). Skeletal diverse synthesis of N-fused polycyclic heterocycles via the sequence of Ugi-type MCR and cui-catalyzed coupling/tandem Pictet–Spengler reaction.The Journal of organic

chemistry,77(3), 1414-1421.

[60] Guchhait, S. K., Chaudhary, V., & Madaan, C. (2012). A chemoselective Ugi-type reaction in water using TMSCN as a functional isonitrile equivalent: generation of heteroaromatic molecular diversity.Organic & biomolecular chemistry,10(46), 9271-9277.

[61] Song, G. T., Xu, Z. G., Tang, D. Y., Li, S. Q., Xie, Z. G., Zhong, H. L., ... & Chen, Z. Z. (2016). Facile microwave-assisted synthesis of benzimidazole scaffolds via Ugi-type three-component condensation (3CC) reactions.Molecular diversity,20(2), 575-580.

[62] Garrido, M., Corredor, M., Orzáez, M., Alfonso, I., & Messeguer, A. (2016). Regioselective Synthesis of a Family of β-Lactams Bearing a Triazole Moiety as Potential Apoptosis Inhibitors.ChemistryOpen,5(5), 485-494.

[63] Zidan, A., Garrec, J., Cordier, M., El-Naggar, A. M., Abd El‐Sattar, N. E., Ali, A. K., ... & El Kaim, L. (2017). β‐Lactam Synthesis through Diodomethane Addition to Amide Dianions.Angewandte Chemie International Edition,56(40), 12179-12183.

[64] Santra, S., & Andreana, P. R. (2007). A one-pot, microwave-influenced synthesis of diverse small molecules by multicomponent reaction cascades.Organic letters,9(24), 5035-5038.

[65] Hulme, C., Ma, L., Cherrier, M. P., Romano, J. J., Morton, G., Duquenne, C., ...& Labaudiniere, R. (2000). Novel applications of convertible isonitriles for the synthesis of mono and bicyclic γ-lactams via a UDC strategy.Tetrahedron Letters,41(12), 1883-1887.

(28)

[66] Tye, H., & Whittaker, M. (2004). Use of a Design of Experiments approach for the optimisation of a microwave assisted Ugi reaction.Organic & biomolecular chemistry,2(6), 813-815.

[67] Polindara-García, L. A., Montesinos-Miguel, D., & Vazquez, A. (2015). An efficient microwave-assisted synthesis of cotinine and iso-cotinine analogs from an Ugi-4CR approach.Organic & biomolecular chemistry,13(34), 9065-9071.

[68] El Kaïm, L., Grimaud, L., Miranda, L. D., Vieu, E., Cano-Herrera, M. A., & Perez-Labrada, K. (2010). New xanthate-based radical cyclization onto alkynes.Chemical

Communications,46(14), 2489-2491.

[69] Schneekloth, J. S., & Sorensen, E. J. (2009). An interrupted Ugi reaction enables the preparation of substituted indoxyls and aminoindoles. Tetrahedron, 65(16), 3096-3101. [70] Lu, K., Luo, T., Xiang, Z., You, Z., Fathi, R., Chen, J., & Yang, Z. (2005). A Concise and

Diversity-Oriented Strategy for the Synthesis of Benzofurans and Indoles via Ugi and Diels-Alder Reactions. Journal of combinatorial chemistry, 7(6), 958-967.

[71] Gordillo-Cruz, R. E., Rentería-Gómez, A., Islas-Jácome, A., Cortes-García, C. J., Dí az-Cervantes, E., Robles, J., & Gámez-Montaño, R. (2013). Synthesis of 3-tetrazolylmethyl-azepino [4, 5-b] indol-4-ones in two reaction steps:(Ugi-azide/N-acylation/SN2)/free radical

cyclization and docking studies to a 5-Ht 6 model. Organic & biomolecular chemistry,

11(38), 6470-6476.

[72] Kumar, A., Li, Z., Sharma, S. K., Parmar, V. S., & Van der Eycken, E. V. (2013). Switching the regioselectivity via indium (III) and gold (I) catalysis: a post-Ugi intramolecular hydroarylation to azepino-and azocino-[c, d] indolones. Chemical

Communications, 49(60), 6803-6805.

[73] Zhang, L., Zhao, F., Zheng, M., Zhai, Y., & Liu, H. (2013). Rapid and selective access to three distinct sets of indole-based heterocycles from a single set of Ugi-adducts under microwave heating. Chemical Communications, 49(28), 2894-2896.

[74]Modha, S. G., Vachhani, D. D., Jacobs, J., Van Meervelt, L., & Van der Eycken, E. V. (2012). A concise route to indoloazocines via a sequential Ugi-gold-catalyzed intramolecular hydroarylation. Chemical Communications, 48(52), 6550-6552.

[75] Lesma, G., Cecchi, R., Crippa, S., Giovanelli, P., Meneghetti, F., Musolino, M., ...& Silvani, A. (2012). Ugi 4-CR/Pictet-Spengler reaction as a short route to tryptophan-derived peptidomimetics. Organic & biomolecular chemistry, 10(45), 9004-9012.

[76] Sinha, M. K., Khoury, K., Herdtweck, E., & Dömling, A. (2013). Various cyclization scaffolds by a truly Ugi 4-CR. Organic & biomolecular chemistry, 11(29), 4792-4796. [77] Cano-Herrera, M. A., & Miranda, L. D. (2011). Expedient entry to the

piperazinohydroisoquinoline ring system using a sequential Ugi/Pictet-Spengler/reductive methylation reaction protocol. Chemical Communications, 47(38), 10770-10772.

[78] Yugandhar, D., Kuriakose, S., Nanubolu, J. B., & Srivastava, A. K. (2016). Synthesis of alkaloid-mimicking tricyclic skeletons by diastereo-and regioselective Ugi/ipso-cyclization/aza-Michael cascade reaction in one-pot. Organic letters, 18(5), 1040-1043. [79] Cheng, G., He, X., Tian, L., Chen, J., Li, C., Jia, X., & Li, J. (2015). Ugi/Himbert

arene/allene Diels–Alder cycloaddition to synthesize strained polycyclic skeleton. The

Journal of organic chemistry, 80(21), 11100-11107.

[80] Kumar, A., Vachhani, D. D., Modha, S. G., Sharma, S. K., Parmar, V. S., & Van der Eycken, E. V. (2013). Post-Ugi gold-catalyzed diastereoselective domino cyclization for the synthesis of diversely substituted spiroindolines. Beilstein journal of organic chemistry, 9, 2097.

(29)

(30)

Chapter 2 A Systematic Protein Refolding Screen Method

using the DGR Approach Reveals that Time and

Secondary TSA are Essential Variables

This work has been published:

Wang, Y., Oosterwijk, N., Ali, A. M., Adawy, A., Anindya, A. L., Dömling, A. S., & Groves, M. R. (2017). Scientific Reports, 7(1), 9355.

(31)

Introduction

The demand for innovative protein-based therapeutics to address drug-resistant diseases is ever more pressing. Although the Human Genome Project completed in 2004 has provided a wealth of biological targets for exploration by structural biology and therapeutic drug design, the number of human protein structures determined by NMR spectroscopy, X-ray crystallography and cryoelectron microscopy has not increased dramatically. One of the main obstacles is the availability of large amounts of the target protein. Although Escherichia coli (E.

coli) is a cheap and readily accessible recombinant expression host, up to 70% of

eukaryotic proteins expressed in E. coli are insoluble as stated by reports from the Center for Eukaryotic Structure Genomics (CESG).1-5_{Despite this major attrition}

and the availability of a large number of well-established expression systems (both eukaryotic and bacterial), E. coli is still the most widely used expression platform for protein production, due to advantages such as high growth rate in inexpensive medium, rapid biomass accumulation, ease of scale-up and high productivity.6-8_{In addition, a great number of expression vectors, engineered}

strains and many cultivation strategies are well characterized for the high-level production of heterologous proteins in this organism. Therefore, the development of a systematic refolding method that can be robustly applied for a wide range of protein candidates to efficiently recover correctly folded and biologically active recombinant proteins from inclusion bodies is highly attractive.

Inclusion bodies are composed of aggregates of unfolded, partially folded and misfolded protein. They are often formed due to a lack of chaperones, failing to reach a correct conformation in the reducing environment of the cytoplasm and potentially undergoing proteolytic degradation.9 Although the protein found in inclusion bodies cannot be directly used for studies due to lack of biological activity, they provide a highly enriched source of target proteins with high purity. Consequently, several refolding methods have been extensively reported (e.g. dilution, dialysis, chromatography and microfluidic chips.10-14_{However, as many}

proteins can only be refolded under very specific conditions, the development of systematic screening methods that can screen multiple refolding conditions in parallel is still challenging.

Several fractional factorial refolding kits (QuickFold [AthenaES], FoldIt [Hampton research], iFOLD [Novagen] and QuickFoldTM Protein Refolding Kit [Molecular Dimensions Limited]) designed to identify optimal refolding conditions are available commercially. Despite this, refolding efforts are still

(32)

limited by the lack of analytical assays to monitor multiple refolding experiments in parallel. More often, the refolding process is detected by surrogate assays such as turbidity or absorbance that cannot readily discriminate between properly folded and misfolded proteins. Similarly, SDS-PAGE, size exclusion chromatography (SEC) and reversed-phase HPLC assays can solve this problem but are time-consuming and are not compatible with high-throughput methods.15 Recently, Biter et al. demonstrated that differential scanning fluorimetry (DSF) guided protein refolding (DGR) allows for discrimination between folded, misfolded and unfolded states of refolded protein.16_{However, the pH, Anion,}

Cation crystallization trial (PACT) screen described is optimized for screening protein crystallization conditions rather than protein refolding. The PACT screen DGR based assay evaluates the effect of pH, anions and cations on a sparse matrix basis and does not address widely used refolding agents, such as chaotropes (urea) or aggregation inhibitors (arginine).

In this study, we have developed a systematic buffer screen that contains a dedicated 96-well primary pH refolding screen, combined with a secondary additive screen to assess other refolding agents (e.g. chaotropes, stabilizing agents, counter-ions and reducing agents) as an extension of the DGR approach. Our research demonstrates that DGR can be applied for determining refolding conditions of proteins that contain more specific and complex buffer compositions. To establish and validate the refolding screen, four different proteins were used: the receptor binding domain of hemagglutinin (HA-RBD), interleukin-17A (IL-17A), mouse double minute 2 homolog (MDM2) and the inhibitory receptor ligand of programmed death receptor 1 (PD-L1). We describe the development of a dedicated and systematic protocol that is able to evaluate the impact of pH, buffer system, time, redox couples as well as a range of additives on the protein refolding process. We also show that a secondary thermal shift assay (TSA) screen performed on the refolded material provides improved dialysis conditions for the removal of refolding agents.

(33)

Figure 1: Process flow chart: The inclusion bodies of the protein of interest were first purified

and solubilized. This was followed by a primary pH refolding screen in 96-well plate, detected the refolding by DGR at different time points. From this screen, initial refolding conditions (pH/Buffer system/Arginine) can be identified. The initial refolding conditions are then combined with an additive screen and assayed using the same procedure to identify the final refolding buffer. Subsequently, protein refolding was scaled up. The first batch of refolded protein was concentrated and directly used for TSA to obtain an optimal protein stability buffer. Finally, the protein refolding was scaled up again and the refolded protein was dialyzed to the protein stability buffer, concentrated and used for characterization.

(34)

Figure 2: Composition of the primary pH screen; Buffer conc.: 50 mM; Salt conc.: 100 mM;

Arginine conc.: 0.4 M; MIB: sodium malonate, imidazole and boric acid; CHC: citric acid, HEPES and CHES; MMT: DL-malic acid, MES and Tris-HCl; ***: pH 4.5-5.5 (Citric acid), pH 6-6.5 (MES), pH 7-7.5 (HEPES), pH 8-8.5 (Tris-HCl), pH 9-10 (CHES); C12, G12: Citric acid at pH 4.

(35)

Methods

Protein expression

For protein expression experiments, 5 mL of an E. coli overnight culture was used to inoculate 1 L Luria–Bertani broth. The cells were grown at 37 ºC with shaking (180 rpm), 1 mM isopropyl-β-D-1-thiogalactopyranoside (IPTG) was added when the OD600 reached 0.6. After induction, the temperature of the culture was lowered

to 18 ºC and the cells were incubated overnight. The cells were subsequently harvested by centrifugation at 5,000 g for 20 min.

Isolation of inclusion bodies from cell pellets

The cell pellets were resuspended in lysis buffer (50 mM Tris-HCl, pH 8.0; 300 mM NaCl; 5% glycerol; 3 mM β-mercaptoethanol (β-ME) and 5 mM ethylenediaminetetraacetic acid (EDTA)). The cell suspension was sonicated for 2 min and 2 mM MgSO4, 0.01 mg/mL DNAse and 0.1 mg/mL lysozyme were

added. The solution was incubated at room temperature for 15 min. Cell debris was removed by centrifugation (19,000 g). The resuspension, sonication and centrifugation procedure was repeated 4 times using lysis buffer with 0.5% Triton 100. The inclusion bodies were incubated in lysis buffer with 0.5% Triton X-100 overnight. Finally, the inclusion bodies were washed twice in wash buffer (50 mM Tris-HCl, pH 8.0; 300 mM NaCl; 5% glycerol; 3 mM β-ME and 5 mM EDTA).

Solubilisation/denaturation of proteins from inclusion bodies

Inclusion bodies were resuspended in either 8 M urea or 6 M guanidine-HCl. Insoluble material was removed by centrifugation at 19,000 g for 20 min.

Refolding screen in 96-well format

The denatured target protein was refolded by shock dilution at a ratio of 1:20. 10 µL of denatured protein solution at ~5 mg/mL was added into each well of a 96-well plate. Subsequently, 190 µL of refolding buffer from a pre-prepared master plate was transferred to the plate containing the denatured protein. Each well contained a final volume of 200 µL. After set up, the plate was incubated with gentle shaking at room temperature or 4 ºC.

Detection of a refolding buffer by DGR

(36)

Orange stock (Invitrogen) was prepared in water. For each well, 5 µL of this diluted SYPRO Orange and 45 µL of the refolding sample was added. The plate was covered with sealing tape and centrifuged at 600 g for 1 min to remove any air bubbles. A BioRad CFX96 RT-PCR machine was programmed with a 3 min equilibration time at 22 ºC. Subsequently, the plate was heated from 22 to 90 ºC with stepwise increments of 0.5 ºC per 30 s, followed by the fluorescence reading optimized for SYPRO Orange at 485 nm (Excitation) and 530 nm (Emission) after each step. The 200 µL refolding sample prepared allows for the refolding experiments to be analyzed by DGR at four different time points, aiding the identification of the optimal incubation time for the refolding process.

Scale-up refolding for protein stability buffer screen

The subsequent step uses a TSA analysis to screen for dialysis conditions to remove refolding agents. 2.5 mL of denatured protein sample was added dropwise into 50 mL of the refolding buffer identified above, under continuous stirring. The mixture was incubated under stirring for a period of time (as identified from the time points tested in the previous screen). The optimal buffer for subsequent purification and/or concentration could then be identified using a more standard thermal shift assay approach (DSF/TSA).17_{Briefly, samples from the up-scaled}

refolding step were concentrated to a maximum of 1 mg/mL and spun to remove any precipitation (5 minutes at 18,000g). SYPRO Orange (Invitrogen) was prepared by diluting 2.5 µL of the stock solution in DMSO into 500 µL of the concentrated refolded protein sample. For each well in the 96-well plate, 5 µL of protein/SYPRO Orange solution was diluted with 45 µL of the buffer screen formulation. Then the plate was centrifuged at 600 g for 1 min before the start of the assay to remove any air bubbles. The TSA analysis was performed as described above.

Scale-up refolding for protein production

After identification of a suitable dialysis buffer, scale-up refolding was repeated and the refolded sample was dialyzed against the dialysis buffer identified overnight. The dialyzed sample was then concentrated to 20 mL using an Amicon 6050 concentrator and further concentrated by a Vivaspin 20 centrifugal concentrator (Sartorius) at 2,000 RCF at 4 ºC.

(37)

The concentrated proteins were purified via SEC using a Superdex 200 10/300 or HiLoad 16/60 Superdex 75 column (GE Healthcare) pre-equilibrated with the identified dialysis buffer using an NGC liquid-chromatography system (Bio-Rad). Where necessary, the molecular masses of the proteins in solution were determined using static light scattering (SLS). A mini DAWN TREOS from Wyatt technology was connected in line with the NGC system. Proteins were concentrated to 2 mg/mL before injection.

(38)

Results

Development of a Dedicated Refolding Screen

Our refolding buffer screen is composed of a “primary pH screen” (Fig. 2) and a “secondary additive screen” (Fig. 3). The first screen identifies a “basic buffer” with the optimal pH in the best buffer system for protein refolding; the second screen uses the “basic buffer” in combination with a wide range of additives (mostly protein stabilizers).18-22

Primary pH Screen

The first part (row A-D) contains a variety of buffer systems in combination with 100 mM NaCl. Similarly, the second part (row E-H) contains these buffers combined with 0.4 M L-arginine hydrochloride, the most commonly used aggregation inhibitor to suppress aggregation and enhance protein refolding.23-29 This screen covers a pH range from 4.0 to 10.0 (the pH of MMT buffer was 4.0-9.0). Importantly, we use 4 different buffer systems: MIB buffer, CHC buffer, MMT buffer and single buffers. These buffer systems are composed of several reagents, ensuring pH stability over a large range. The different buffer systems not only allow us to explore the effect of pH, but also the impact that different buffer compositions may have on the protein refolding process. The initial DGR screen (Fig. 2) focuses on the choice of pH and buffer system in the presence or absence of arginine. Information from this screen is used to perform the secondary screen (Fig. 3), which focuses on identifying further buffer components that will improve refolding efficiency (i.e. chaotropes, stabilizing agents, counter ions and reducing agents).

Secondary Additive Screen

The optimal condition found in the primary pH screen was used as a standard composition to prepare a secondary additive screen, by diluting the additives into the identified conditions of the primary screen. Conditions A1-A9 are used to analyze the effect of 3 different sugars: glucose, sucrose and lactose at three different concentrations. Conditions A10-A12 contain 3 different detergents, which can bind to protein surfaces to prevent aggregation. Conditions B1-B8 contain 2 different chaotropic agents (0.5-2 M): guanidine-HCl and urea, which are commonly used for dissolving protein inclusion bodies and may promote

(39)

D12 compare the effect of different concentrations of arginine (0.4-1 M) and 3 amino acids (alanine, glycine and proline). Conditions E1-E12 are included to analyze protein refolding in the presence of NaCl/KCl over a wide concentration range (50-1000 mM). This allows exploration of whether low or high ionic strength will contribute to protein refolding process. Several metal salts with different valences are contained in conditions F1-F9. EDTA was also included in this group to investigate the refolding process in the absence of divalent metals. The buffer found in the primary pH screen in triplicate was contained in F10-F12 to ensure that the results from the assay are robust and reliable. G1-G12 is used to evaluate the effect of 12 different salts. Finally, two different reducing agents (β-ME and Dithiothreitol (DTT)) in several concentrations are included in H1-H12.

HA-RBD

(40)

Figure 4: DGR of HA-RBD. (a) Melting transition curves of samples in single buffer at

different pH values. (b) The impact of arginine on refolding. (c) Comparison of sample refolded in different buffer systems at pH 8 in the presence of arginine. (d) Melting transition curves of samples in the presence of the additive β-ME. (e) Melting transition curves collected at different time points. (f) Results of TSA buffer screen. (g) Size exclusion chromatogram of refolded HA-RBD. (h) Crystals of HA-RBD and (i) Molecular replacement solution of refolded HA-HA-RBD.

The receptor-binding domain of Hemagglutinin (HA-RBD) is one of the major surface proteins of the influenza virus and plays an important role at the beginning of the viral cycle.30, 31 HA-RBD is located within residues 63-286 of the influenza virus H1N1 and has a molecular weight of approximately 25 kDa, containing 4 cysteine residues. A synthetic gene corresponding to A/H1N1/2009 influenza virus hemagglutinin was cloned into the pETM11 vector32_{and expressed in E.}

coli BL21* (DE3) cells. The inclusion bodies were denatured in Tris-HCl

buffered 8 M urea solution (pH 8) at a protein concentration of 5 mg/mL. The denatured inclusion bodies were first submitted to the pH refolding screen. The results showed that pH and the presence of arginine play a key role in the refolding process. Firstly, when refolding experiments were performed over a pH range in single buffers (Fig. 3, Row H), a clear melting transition signal was seen for pH values above 7.5 (Fig. 4a). Further inspection revealed that the addition of arginine to the refolding buffer resulted in a further improvement of the melting

(41)

pH 8.0 gave the best refolding signal (Fig. 4c). As the TM values estimated from

the curves shown in figures 4a-4c are very similar (~45 ºC), refolding efficiency was judged by the highest fluorescence intensity and lowest initial background. Thus, our initial screen identified an optimal pH, buffer component and additive for the refolding of HA-RBD (Tris-HCl pH 8.0 and 0.4 M arginine).

This buffer was then used as a standard addition in follow-up refolding experiments using the secondary additive screen (Fig. 3). These experiments demonstrated that only the addition of β-ME resulted in an improved signal (Fig. 4d). All other conditions screened by plate 2 resulted in either no effect or a reduction in the refolding process as judged from the melting curves. This includes an analysis of increasing salt concentrations (Fig 3, Row E), suggesting that 100 mM NaCl represented the optimal ionic strength for refolding.

Our results also showed that refolding time has a substantial impact on the refolding process (Fig. 4e). When the protein was refolded over 1h, clear melting transition signals could be measured. Surprisingly, when the refolding time was ~6h, the melting transition signals disappeared. This conclusion is also confirmed by the preparative-scale (50 mL) refolding experiment. Initially, the refolding solution was clear, while after 1 h the refolding solution showed minor precipitation, increasing to larger degrees of precipitation at 6h.

The above analysis resulted in the identification of a suitable refolding buffer (50 mM Tris-HCl pH 8, 100 mM NaCl, 0.4 M arginine, 3 mM β-ME). A second DSF-based screen (TSA buffer screen), similar in approach to that used in other laboratories,17_{was used to identify an dialysis buffer for the removal of refolding}

agents. From this screen, four key factors were found to stabilize the refolded protein: 50 mM Tris buffer at pH 8, 200 mM sodium chloride and 10 mM sodium citrate (Fig. 4f). Surprisingly, while the presence of β-ME was shown to be a positive factor in the refolding of HA-RBD, this second analysis demonstrated that the presence of β-ME was a negative factor on HA-RBD stability once it is refolded. When the sample was dialyzed to this new buffer, the protein could be easily concentrated to 7 mg/mL, whereas the protein dialyzed into 50 mM Tris-HCl, pH 8, 100 mM NaCl and 3 mM β-ME (ie. the refolding buffer omitting arginine) could not be concentrated above 0.5 mg/mL. The increase in ionic strength and addition of citrate ions seem to play a very important role in protein stabilization after the removal of arginine.

(42)

Refolded HA-RBD appeared homogeneous via size-exclusion chromatography and we were able to crystallize the refolded HA-RBD (100 mM Tris-HCl, pH 8.8 and 36% PEG 2000) in the space group P21, with unit-cell parameters a = 39.5 Å;

b = 73.7 Å; c = 73.7 Å and α = 90°; β = 100.72°; γ = 90° (Fig 4g).31_{The structure}

was solved to 1.9 Å resolution (Fig. 4h and 4i).