Recent advances in proteolytic stability for peptide, protein, and antibody drug discovery

Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=iedc20

Expert Opinion on Drug Discovery ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/iedc20

Recent advances in proteolytic stability for peptide, protein, and antibody drug discovery

Xianyin Lai, Jason Tang & Mohamed E.H. ElSayed To cite this article: Xianyin Lai, Jason Tang & Mohamed E.H. ElSayed (2021): Recent advances in proteolytic stability for peptide, protein, and antibody drug discovery, Expert Opinion on Drug

Discovery, DOI: 10.1080/17460441.2021.1942837 To link to this article: https://doi.org/10.1080/17460441.2021.1942837

Published online: 30 Jun 2021.

Submit your article to this journal Article views: 50

View related articles View Crossmark data REVIEW Recent advances in proteolytic stability for peptide, protein, and antibody drug discovery

Xianyin Lai, Jason Tang and Mohamed E.H. ElSayed Biotechnology Discovery Research, Lilly Research Laboratories, Eli Lilly and Company, Indianapolis, IN, USA

ABSTRACT Introduction: To discover and develop a peptide, protein, or antibody into a drug requires overcoming multiple challenges to obtain desired properties. Proteolytic stability is one of the challenges and deserves a focused investigation.

Areas covered: This review concentrates on improving proteolytic stability by engineering the amino acids around the cleavage sites of a liable peptide, protein, or antibody. Peptidases are discussed on three levels including all peptidases in databases, mixtures based on organ and tissue types, and individual peptidases. The technique to identify cleavage sites is spotlighted on mass spectrometry- based approaches such as MALDI-TOF and LC-MS. For sequence engineering, the replacements that have been commonly applied with a higher chance of success are highlighted at the beginning, while the rarely used and more complicated replacements are discussed later. Although a one-size-fits-all approach does not exist to apply to different projects, this review provides a 3-step strategy for effectively and efficiently conducting the proteolytic stability experiments to achieve the eventual goal of improving the stability by engineering the molecule itself.

Expert opinion: Improving the proteolytic stability is a spiraling up process sequenced by testing and engineering. There are many ways to engineer amino acids, but the choice must consider the cost and properties affected by the changes of the amino acids.

ARTICLE HISTORY Received 18 December 2020 Accepted 10 June 2021

KEYWORDS Proteolytic stability; peptide; protein; antibody; peptidase; cleavage site; amino acid replacement

1. Introduction Peptide, protein, and antibody drugs have been applied in various diseases and syndromes due to their distinct biochem­ ical and therapeutic characteristics, having the advantages of higher efficacy and specificity compared to small molecular drugs [1]. During 2019, the U.S. Food and Drug Administration (FDA) approved 48 new drugs, including three peptides, two proteins, five monoclonal antibodies (mAbs), and three anti­ body drug conjugates (ADCs), which represents 27% of the total approved new drugs [2]. Globally, more than 500 peptide drugs are under clinical developments and over 60 are approved for clinical use [3–5]. Therapeutic monoclonal anti­ bodies have become the best-selling drugs, and their market value was at approximately US$115.2 billion in 2018 and is expected to generate revenue of 300 USD billion by 2025 [6].

Despite the significant achievements of peptide, protein, and antibody drugs in clinical use and market, the challenges for peptide, protein, and antibody drug discovery and devel­ opment are to obtain the right properties for desired pharma­ cokinetics, pharmacodynamics, and safety [7–9]. One of the shared challenges among peptides, proteins, and antibody drugs is proteolytic stability. Proteolytic degradation of pep­ tide drugs is a major weakness in systemic therapeutic appli­ cations, especially when peptides reach the bloodstream containing various proteases [10]. The loss of therapeutic pro­ teins may be caused by proteolytic enzymes in the dermal, subcutaneous, and lymphatic tissues [11]. Although antibodies are generally considered as highly resistant to proteolytic degradation, studies have shown that some antibodies are liable to multiple proteases [12].

Recently, oral peptides, proteins, and antibody drugs have drawn attention from pharmaceutical companies. The latest oral peptide drug approved by FDA is semaglutide oral tablets to treat patients with type 2 diabetes, but it has a low and variable bioavailability value of about 1% [13]. While oral peptides are approved and used in the clinic, oral delivery systems for protein and antibody drugs are still under research and development phases [14–18]. Besides the enzymes that the oral drugs encounter after their absorption, enzymes in gastrointestinal tract are another set of contributors for pro­ teolytic degradation that is one of the many barriers for oral peptide, protein, and antibody drug delivery [19,20].

To overcome the proteolytic degradation, strategies have been taken based on the administration of the peptide, pro­ tein, and antibody drugs. For oral delivery, various approaches have been applied to address the proteolytic stability chal­ lenge. Formulation approaches such as liposomes provide a technique to physically separate the drugs from enzymes to avoid proteolytic degradation [18]. Altering the local pH at the delivery site by formulation is able to inhibit the resident peptidases since the optimal pH for the enzymes is changed

CONTACT Xianyin Lai xianyin.lai@lilly.com; Mohamed E.H. ElSayed mohamed.elsayed@lilly.com

Biotechnology Discovery Research, Lilly Research Laboratories, Lilly Technology Center - North, Eli Lilly and Company, Indianapolis, IN 46221, USA

EXPERT OPINION ON DRUG DISCOVERY https://doi.org/10.1080/17460441.2021.1942837

© 2021 Informa UK Limited, trading as Taylor & Francis Group [21]. Directly using proteinase inhibitors to co-formulate with peptides provides localized protection of peptides [22]. To avoid complex formulations and safety concerns about enzyme inhibition, changing amino acids around the cleavage sites provides a solution for increasing peptide intrinsic pro­ teolytic stability to avoid degradation [23]. For subcutaneous, intravenous, and other non-oral drug delivery approaches, modifying the liable amino acids is the only way to improve their proteolytic stability.

In this review, we focus on enzymes related to proteolytic degradation of peptides, proteins, and antibody drugs, meth­ ods for cleavage site identification, and approaches to chemi­ cally engineer sequences of drugs to improve their proteolytic stability. Compared to proteins and antibodies, peptides have a more flexible structure, which makes peptides more vulner­ able to enzymatic attacks. However, proteins and antibodies are also made of amino acids that are potentially susceptible to proteinases and peptidases (EC 3.4). Therefore, protein and antibody drugs are included in this review, although most examples are given using peptides. The aim of the present review is to present quick guidance for scientists who work on peptides, proteins, and antibody drugs to increase their pro­ teolytic stability during the structure-activity relationship (SAR) optimization process.

2. Peptidases Peptidases (alternatively termed proteinases, proteases, or proteolytic enzymes) are enzymes that hydrolyze peptide bonds. In the current review, peptidase is chosen to represent other alternative terms. In the human body, cells produce numerous endogenous peptidases with var­ ious functions. Peptidases are necessary for living crea­ tures, but peptides, proteins, and antibody drugs with natural amino acids are intrinsically liable to peptidases.

Meanwhile, exogenous peptidases from bacteria, parasites, and viruses that exist within our body might also affect the proteolytic stability of peptide, protein, and antibody drugs [24]. To gain the knowledge of which peptidases potentially hydrolyze peptide bonds, two databases are introduced in this review. One database contains a full list of peptidases from different species and the other database provides detailed expression levels at specific intracellular and intercellular locations. With the combina­ tion of the two databases, scientists can select proper peptidases for specific proteolytic stability studies. Except for a few enzymes allowing highly predictable cleavage sites, most predictions for each unique peptide sequence from the databases are only partially true because the surrounding amino acids affect the liability of the cleavage sites. Therefore, bench experiments are critical to deter­ mine the proteolytic stability of a given sequence. During experiments, peptidase mixtures from certain parts of our body or purified enzymes are both often utilized, so both peptidase mixture and individual peptidases were reviewed. The individual peptidases were further divided into two sections. One section covers enzymes in the gastrointestinal tract that are unique for oral delivery of peptide, protein, and antibody drugs. The other section includes other enzymes that are encountered by non-oral delivery drugs.

2.1. Peptidase database The MEROPS database (www.ebi.ac.uk/merops/) is an informa­ tion resource for peptidases and the proteins that inhibit them [25]. The database includes about 4,000 individual peptidases and inhibitors. More than 1816 out of the 4,000 are unique peptidases. The peptidase database can be reviewed by name, identifier, gene name, organism, and substrates. In the current database (Release 12.2), the total number of organisms that have been classified is 30,053.

Three orthogonal approaches have been used to classify peptidases: the chemical mechanism of catalysis, the position of the reaction the peptidases catalyze, and the molecular structure and homology. Based on the chemical mechanism of catalysis, peptidases are classified as aspartic (A), cysteine (C), glutamic (G), metallo (M), asparagine (N), mixed (P), serine (S), threonine (T), or unknown (U) catalytic type [26,27]. Based on the position of the reactions that they catalyze, peptidases are grouped into exopeptidases that cleave a single amino acid, dipeptide, or dipeptide from the ends of a polypeptide sequence and endopeptidases that cleave peptide bonds between two amino acids within a polypeptide sequence.

The exopeptidases are further divided into carboxypeptidases, aminopeptidases, dipeptidyl-peptidases, and tripeptidyl- peptidases. Based on the molecular structure and homology, peptidases under the nine chemical mechanisms mentioned above are further classified into clans that contain all

Article highlights ●Despite the significant achievements of peptides, proteins, and anti­ body drugs in clinical use and marketing, one of the shared chal­ lenges among peptides, proteins, and antibody drugs is proteolytic stability that could compromise desired pharmacokinetics, pharma­ codynamics, and safety.

●In the human body, cells produce numerous endogenous peptidases with various functions. Peptidases are necessary for living creatures, but peptides, proteins, and antibody drugs with natural amino acids are intrinsically liable to peptidases.

●While peptidase databases provide a full list of peptidases from different species and detailed expression levels at specific intracellu­ lar and intercellular locations for in silico prediction, peptidase mix­ tures, and individual peptidases are the ones used in vitro for experimental analysis.

●Although many sources and software provide in silico cleavage pre­ diction, in vitro experiments are more meaningful than in silico prediction to obtain cleavage sites of peptide, protein, and antibody drugs under biophysical conditions.

●After the identification of cleavage sites, engineering the amino acids around the sites are primarily from P1 and P1ʹ to P2 and P2ʹ, and possibly from P3 and P3ʹ to P4 and P4ʹ. The replacement might change the efficacy, toxicity, and many other properties, while the proteolytic stability is increased. Scientists must balance the proteo­ lytic stability and other desired properties.

●After revealing the key components of proteolytic stability, a 3-step strategy was proposed to conduct the proteolytical stability experi­ ment without overwhelming resources and with balancing under- or over-engineering the sequences.

This box summarizes key points contained in the article.

2 X. LAI ET AL. peptidases from a single evolutionary origin. The clans are then further categorized into families that have a set of homo­ logous proteolytic enzymes [28].

For researchers who focus on proteolytic stability studies during peptide, protein, and antibody drug research and development, the most interesting information is about pep­ tidases in specific species. In MEROPS database, 609, 739, 814,

662, and 815 known and putative peptidases are collected from human (Homo sapiens), cynomolgus monkey (Macaca fascicularis), pig (Sus scrofa), rat (Rattus norvegicus), and mouse (Mus musculus), respectively. Among human proteins encoded by 19,587 genes, the peptidase from 609 genes counts for 3.1% of all the encoded proteins, which makes peptidases one of the most abundant classes of enzymes [27].

The MEROPS database provides a large list of peptidases that potentially degrade peptide, protein, and antibody drugs.

However, it is impossible to test over 600 peptidases for drug candidates during the SAR optimization. Based on the delivery route, in vivo trafficking, and intracellular path of the peptides, proteins, and antibodies, we can exclude many peptidases for testing based on the knowledge of their location. However, the location information is not provided by the MEROPS data­ base. Fortunately, another database, the Human Protein Atlas (www.proteinatlas.org), provides such information.

The Human Protein Atlas is a database collecting human protein expression levels in blood, cells, tissues, and organs with specific locations based on various technologies, such as antibody-based imaging, mass spectrometry-based proteo­ mics, transcriptomics, and systems biology. The program was initiated in 2003 with the Tissue Atlas and expanded to six sub-databases, each focusing on a specific aspect of the human proteins. The Tissue Atlas provides the expression and localization of human proteins across major tissues and organs [29,30], the Brain Atlas expands brain regions from a limited number of human brain regions in the Tissue Atlas to 10 major well-defined mammalian brain regions [31], the

Cell Atlas shows high-resolution images of protein subcellular localization within cells [32], the Blood Atlas has proteins from blood cells and proteins secreted from other tissues into blood [33,34], the Metabolic Atlas shows protein function and gene expression in the context of the human metabolic network, and the Pathology Atlas includes the impact of protein levels for survival of patients with cancer [35].

Four out of the six atlases are relevant for proteolytic stability studies. Knowing the peptidase expression level in blood, specific organs, tissues, cells, and organelles, scientists can select the proper peptidases to test peptide, protein, and antibody drug candidates according to their delivery route, in vivo trafficking, and intracellular path.

2.2. Peptidase mixture Except some well-studied peptidases like pepsin (EC 3.4.23.1), trypsin (EC 3.4.23.1), dipeptidyl peptidase 4 (DPP4, EC 3.4.14.5), and others, it is very challenging to select certain peptidases for proteolytic stability optimization of peptide, protein, and antibody drug candidates. Therefore, before specific pepti­ dases are identified, a peptidase mixture of bodily fluids or tissues is an alternative approach to start the optimization.

Because peptidases exist either outside or inside of cells, the strategy to obtain peptidase mixtures is to collect bodily fluids and homogenate cells from various tissues. Major human bodily fluids include gastric fluid, intestinal fluid, blood, serum, plasma, urine, cerebrospinal fluid, saliva, bronchoalveo­ lar lavage fluid, synovial fluid, nipple aspirate fluid, tear fluid, and amniotic fluid [36,37]. There are 11 major organ systems in the human’s organism. A specific part of an organ system is counted as a tissue type. Depending on the tissue collection, human tissue types could be as many as 44 kinds [30]. We will focus on some bodily fluids and tissues to discuss the impact of the peptidases from them on peptide, protein, and anti­ body drug candidates during SAR optimization.

The primary function of the stomach is to initiate a chemical and enzymatic breakdown of food using its enzymes in acidic conditions [38]. In this harsh environment, there is no exception for peptides, proteins, and antibody drugs to be broken down. When they are administered orally and the absorption happens in the stomach, the first exam­ ination is the proteolytical stability in the stomach fluid.

Seventeen peptide drugs were investigated in gastric fluids from both humans and pigs. The results showed that peptides with more enzymatically susceptible bonds and higher struc­ tural flexibility were more vulnerable in gastric fluids due to enzymatic cleavages [39].

The small intestine is another major part of the human digestive system and further processes macromolecules into tri-, di-, and mono-peptides [38]. In small intestinal fluids, proteins showed greater stability than peptides. The reason might be that proteins have more complex secondary and tertiary structures, leading to less easily accessible peptide bonds [40]. Therefore, peptides were more often evaluated in small intestinal fluids than proteins were. Oxytocin and vasopressin are pituitary neuropeptides, functioning as an integrated, adaptive system, allowing the mammalian body to survive, maintain homeostasis, and reproduce [41]. The two molecules are potential drugs to treat social deficits in neurodevelopmental disorders, including autism spectrum dis­ order [42]. A modified vasopressin, 1- deamino-8-D-arginine- vasopressin (dDAVP), has been successfully intranasally used in patients with central diabetes insipidus. Buccal administra­ tion of oxytocin in pregnant women at term is followed by uterine contractions and increased plasma concentrations of oxytocin. However, their oral bioavailability was only 0.1–0.2% in healthy volunteers. Arginine-vasopressin, oxytocin, and their synthetic analogues were incubated in human small intestinal fluid from healthy volunteers. The quantification of degrada­ tion was carried out by reversed-phase high performance liquid chromatography (RP-HPLC). Arginine-vasopressin was completely degraded in 30 min, while dDAVP remained 50% of the intact after 35 min. Oxytocin showed much more sta­ bility than dDAVP and arginine-vasopressin [43].

Plasma and serum are obtained from whole blood using different processing methods. Plasma is produced by add­ ing anticoagulants (such as EDTA or heparin) that might inhibit some enzymatic activities of peptidases, while serum is the supernatant obtained after blood coagulation, which might lose some peptidases. The collection differ­ ence among blood, serum, and plasma might lead to

EXPERT OPINION ON DRUG DISCOVERY 3 different protease compositions and proteolytic activities for some peptide, protein, or antibody drugs, but all the drugs are potentially liable to peptidases in blood, plasma, or serum. In a study on proteolytic degradation of peptide- based drugs, eight peptides were incubated with blood, plasma, and serum. Except one stabilized peptide, all seven other peptides showed degradation products in blood, plasma, and serum [10]. Compared to peptides, proteins, and antibodies have more complex secondary and tertiary structures, leading to increased proteolytic resistance [44].

However, not all proteins and antibodies have a strong proteolytic stability. Their proteolytic stability is deter­ mined by the amino acid sequence and their structure.

In addition to peptidase mixtures from bodily fluids, crude protein homogenates from different organs and tis­ sues provide peptidase mixtures for proteolytic stability evaluation of peptide, protein, and antibody drugs.

Although there are at least 44 types of tissues, the enzy­ matic breakdown of peptide, protein, and antibody mainly happens at the liver and kidney [45]. The proteolytic sta­ bility of exendin-4 was examined in rat liver and kidney homogenates. The results showed that the half-life of exendin-4 in the liver and kidney homogenate was 100.9 and 7.8 min, respectively. The enzymes most likely to be involved in the degradation of exendin-4 were aminopep­ tidases, serine proteases, and metalloproteases [46].

Without knowing the organs that the peptide, protein, and antibody drugs distribute in and how the drugs get inside and outside of cells, using the entire tissue runs into a risk that might identify cleavage sites that might not happen in vivo, leading to over-engineering the amino acid sequence. Scientists must be aware of that risk and keep a balance between increasing proteolytic stability and over engineering the sequences.

To balance the risk, one of the choices is to use organelle fractions instead of the entire tissue. An example is lysosomes that have been focused on their role in protein degradation in mammalian cells [47,48]. If peptides, proteins, and antibody drugs enter the endolysosome, a study of their endolysosome degradation can be carried out with an isolated endolysosome fraction.

One peptidase mixture is so special that it has its own name, pancreatin. Pancreatin is a mixture containing various enzymes, principally amylases to catalyze the hydrolysis of starch into sugars, lipases to catalyze the hydrolysis of fats, and peptidases. It is obtained from the pancreas of swine or ox by precipitating the crude protein extraction from the pan­ creas homogenates using alcohol [49,50]. Five proteins, bovine serum albumin, ovomucoid, β-lactoglobulin A, β-lactoglobulin

B, and lysozyme, were incubated with 10 mg/mL pancreatin in simulated intestinal fluid prepared as described in the United

States Pharmacopoeia.

The results showed that β- lactoglobulin A and β-lactoglobulin B were completely degraded within 2 min, while bovine serum albumin, ovomu­ coid, and lysozyme have 17.7%, 23.4%, and 22.8% degrada­ tion, respectively [51].

2.3. Individual peptidase Among the 600–800 peptidases from different animal species used in pharmaceutical studies, some of them have been intensively studied on their degradation function to amino acid bonds. Using well-known peptidases rather than the peptidase mixture from bodily fluids or tissues is an approach to avoid over identification of cleavage sites. Two subsections are created to describe peptidases for oral delivery and non- oral delivery.

2.3.1. Peptidases for oral delivery The stomach secretes a mixture of compounds, including water, mucus, hydrochloric acid, pepsin, and intrinsic factors.

Among them, pepsin is the principal enzyme involved in pep­ tide, protein, and antibody digestion, generating smaller pep­ tides and amino acids [52]. Pepsin is an endopeptidase that breaks down peptide bonds within a polypeptide sequence through aspartic catalysis mechanism at a low pH (1.5–2.0) [53]. The proteolytic stabilities of the bovine serum albumin, ovomucoid, β-lactoglobulin A, β-lactoglobulin B, and lysozyme were evaluated with 3.2 mg/mL pepsin in simulated gastric fluids prepared as described in the

United States Pharmacopoeia. Bovine serum albumin and ovomucoid were observed to be rapidly digested within 0.25 min at the begin­ ning of the incubation time. Lysozyme was completely degraded in 30 min. β-lactoglobulin A and β-lactoglobulin

B had degradation percentages of 34.3% and 17.2% after

60 min of incubation [51].

The pancreas produces trypsinogen, an inactive precursor of trypsin. Trypsinogen is then secreted into the duodenum through the pancreatic duct. Once in the small intestine, the trypsinogen is activated by the enzyme enteropeptidase via proteolytic cleavage [54,55]. Trypsin is an endopeptidase and breaks peptide bonds down via a serine catalysis mechanism at an optimal pH of 7.5–8.5 [56]. The salmon calcitonin is a small peptide of 32 amino acids with a molecular weight of

3,432 Da and is a more potent analog of human calcitonin [57]. The salmon calcitonin was incubated with 0.5 μM tryp­ sin in a sodium acetate (50 mM) buffer and it was comple­ tely degraded in 15 min [58]. The human granulocyte colony stimulating factor (G-CSF) is a major cytokine regulator of neutrophilic granulocytes and is able to stimulate the growth of neutrophil colonies from human bone marrow progenitor cells with a molecular weight of about

30,000 Da [59]. The incubation of 100 ug/ml of G-CSF with

1 ug/ml of trypsin revealed that G-CSF was susceptible to trypsin digestion and had about 30% intact protein at 4 h incubation [60].

The second most abundant peptidase in pancreatic secre­ tions is the chymotrypsinogen that becomes chymotrypsin in the duodenum [61]. Although both chymotrypsin (EC 3.4.21.1) and trypsin are serine proteases and endopeptidases with high sequence and structural similarities, they have different substrate specificity [62]. In the studies of salmon calcitonin and G-CSF in trypsin incubation mentioned above, their pro­ teolytic stability in chymotrypsin was examined as well. The

4 X. LAI ET AL. salmon calcitonin was incubated with 0.1 μM chymotrypsin in sodium acetate (50 mM) buffer and it completely degraded in

20 min [58]. The incubation of 100 ug/ml of G-CSF with 1 ug/ ml of chymotrypsin showed that G-CSF was not susceptible to chymotrypsin digestion and had no measurable degradation at 4 h of incubation, showing G-CSF had different proteolytic stabilities against trypsin vs. chymotrypsin [60].

The third major peptidase in pancreatic juice is elastase (EC

3.4.21.36). Like trypsin and chymotrypsin, elastase is from an inactive precursor named proelastase that is stored in the acinar cells of the pancreas and activated in the duodenum by trypsin [63]. Elastase is a serine endopeptidase as well but with its own specificity [64,65]. The incubation of calcitonin with elastase resulted in complete degradation at 40 min, showing elastase had a weaker enzymatic activity than chy­ motrypsin and trypsin for calcitonin [58].

2.3.2. Peptidases for non-oral delivery Carboxypeptidases (EC 3.4.16–3.4.18) are peptidases that cleave a peptide bond at the carboxy-terminal (C-terminal) end of a peptide or protein. Carboxypeptidases are named with a combination of one letter and one number following the word carboxypeptidase at the beginning. There are at least 17 carboxypeptidases, belonging to two major groups.

One uses serine as an active site residue named as serine carboxypeptidases, and the other uses zinc at the active site, known as metallocarboxypeptidases, which are further sepa­ rated into two subgroups based on amino acid sequence similarities [66]. Among the carboxypeptidases, some of them have low tissue specificity with expression in many tissues and are secreted into bodily fluids with a low concen­ tration, such as carboxypeptidases D and X2. Some are in specific tissues without leaking into bodily fluids or with a low concentration in bodily fluids, such as carboxypeptidase

A4 in the esophagus and skin, carboxypeptidase A6 in the intestine, prostate, and retina, carboxypeptidase E in the brain, carboxypeptidase M in the adipose, carboxypeptidase O in the intestine, X1 in the placenta, and carboxypeptidase Z in the ovary. And others are produced in specific tissues and then secreted into either small intestinal fluid or blood.

Carboxypeptidases A1, A2, and B1 are produced in the pan­ creas and activated in the duodenum. All three proteins are zinc-containing metallopeptidases [67]. Compared to the high abundance of trypsin, chymotrypsin and elastase in the small intestine, carboxypeptidases such as A1, A2, and B1, have a much lower concentration.

Before they are secreted into blood, carboxypeptidase A3 is expressed in multiple tissues, carboxypeptidase A5 is expressed in the testis, carboxypeptidases B2, N1, and N2 are expressed in the liver, and carboxypeptidase Q is expressed in the thyroid gland. Other peptidases in blood are carboxypep­ tidases by the mechanism of how they cleave the amino acids and the definition of carboxypeptidase, but they do not have the word carboxypeptidase in their name, such as cathepsin

Z that has a low tissue specificity and a low concentration in blood. Evaluating the proteolytic stability against the indivi­ dual carboxypeptidases in blood does not have a high priority due to a few reasons. First, most blood carboxypeptidases have a low concentration, causing fewer concerns about their enzymatic activities. Second, using blood, serum, and plasma rather than individual peptidase is more efficient because of the easy availability and enzymatic diversity of blood. Third, C-terminal protection has been applied in the original peptide sequence design due to the finding that many bioactive peptide hormones naturally have a C-terminal amide group from the posttranslational oxidative cleavage of a C-terminal glycine-extended precursor by α- amidating enzyme [68]. Oxytocin, vasopressin, and calcitonin are in widespread clinical use are amidated peptides.

In contrast to carboxypeptidases, aminopeptidases cleave peptide bonds at the N-terminus of proteins. They are widely distributed in many organs and found in many subcellular organelles, cytoplasms, and as membrane components.

Aminopeptidases are classified by multiple criteria.

Depending on the number of amino acids cleaved from the

N-terminus, enzymes that remove one NH2-terminal amino acid are called aminopeptidases, enzymes that remove two

NH2-terminal amino acids are called dipeptidyl peptidases, and enzymes that remove three NH2-terminal amino acids are called tripeptidyl peptidases. Based on which residues are removed, aminopeptidases are classified as alanyl, arginyl, aspartyl, cystinyl, leucyl, glutamyl, and methionyl aminopepti­ dase. Other factors, such as location, pH, susceptibility to inhibition, metal ion content, etc. are applied to category aminopeptidases [69].

Among the aminopeptidases, DPP4 gained the most atten­ tion during peptide drug SAR engineering because it is well known to cleave many biological substrates including growth factors, chemokines, neuropeptides, and vasoactive peptides, leading to loss of their activity in most cases [70]. Specially,

DPP4 is responsible for the degradation of incretins such as gastric inhibitory peptide (GIP) and Glucagon-like peptide-1 (GLP-1) that are targets for novel therapies in the pharmaceu­ tical industry [71]. GIP and GLP-1 are incretins that bind to specific receptors on pancreatic beta cells and increases insu­ lin release before blood glucose levels are elevated, delay gastric emptying, and suppress glucagon secretion, having beneficial effects in type 2 diabetes. However, both hormones are rapidly cleared from the blood by DPP4, resulting in a shorter half-life [72].

The zymogen of plasmin (EC 3.4.21.7) is a plasminogen synthesized in the liver and many organs and circulate in blood. Plasminogen is activated into a plasmin that consists of two polypeptide chains, the heavy chain, and the light chain, connected by two interchain disulfide bridges. Plasmin degrades many blood plasma proteins, including fibrinogen, factors V, VIII, and X, components of the complement system, the hormones adrenocorticotropic hormone, glucagon, and somatotropin [73,74].

Furin (EC 3.4.21.75) is ubiquitously expressed in many tis­ sues and secreted into blood. It is a calcium-dependent serine endoprotease and the best-characterized representative of the mammalian subtilisin-like family of proprotein convertases.

Furin cleaves precursors of a wide variety of proteins, includ­ ing hormones, growth factors, serum proteins, proteases of the blood-clotting system, matrix metalloproteinases, recep­ tors, viral envelope glycoproteins, and bacterial exotoxins [75,76].

EXPERT OPINION ON DRUG DISCOVERY 5 Many more ubiquitous peptidases like plasmin and furin circulate in blood. Thimet oligopeptidase (EC 3.4.24.15) is a metalloprotease associated with peptide processing in ner­ vous system structures. It degrades several peptides, including bradykinin, amyloid beta, and the major histocompatibility complex class I molecules [77]. Nardilysin (3.4.24.61) is a metalloendopeptidase and promotes ectodomain shedding of the precursor forms of various growth factors and cytokines [78]. Neprilysin (neutral endopeptidase, NEP, EC 3.4.24.11) is a membrane-bound zinc-metallopeptidase enzyme and cleaves various peptides, including natriuretic and bombesin- like peptides, endothelin-1, and substance P [79]. Numerous peptidases are expressed in various tissues and secreted into plasma. Whether the peptidases are concerns for peptide, protein, and antibody drugs depends on the peptidase con­ centration in the blood.

For peptidases inside cells of various organs and tissues, many of them act on specific peptides and proteins. However, cathepsins are a special group of peptidases and have high concentrations in endosomes and lysosomes to breakdown peptides and proteins. Cathepsins provide a delicate balance of expression, targeting, zymogen activation, inhibition by protein inhibitors and degradation [80]. There are 15 cathe­ psins encoded in the human genome, including serine pro­ teases A and G, cysteine proteases B, C, F, H, K, L, O, S, V, W, and X (or Z), and aspartyl proteases D and E. Among the cathepsins, cathepsin D (EC 3.4.23.5) is the most intensively studied and is the main endolysosome enzyme [81,82].

Therefore, cathepsin D is highly relevant to the evaluation of peptide, protein, and antibody drug proteolytic stability.

Cytochrome P450 (CYP450s) are a superfamily of enzymes containing heme as a cofactor, acting as monooxygenases to oxidize steroids, fatty acids, xenobiotics, and numerous small molecules. They are very important for the clearance of var­ ious compounds, especially for small molecular drugs, playing a key role in the metabolism of drugs [83]. Although CYP450s have more than 50 enzymes, six of them metabolize 90% of the drugs, with the two most significant enzymes being

CYP3A4 (EC 1.14.14.55) and CYP2D6 (EC 1.14.14.1). Genetic variability of the enzymes may influence a patient’s response to commonly prescribed drugs [84]. However, most peptides, proteins, and antibodies are mainly metabolized by peptidases rather than CYP450s [85].

3. Cleavage During the proteolytic stability studies, one key question is to identify cleavage sites. Currently, two approaches are available to know the cleavage sites. One way is in silico to predict the cleavage sites based on peptide, protein, and antibody sequences using informatics tools. The other approach is in vitro to find the actual cleavage sites by conducting an in vitro experiment.

Many sources and software provide in silico cleavage pre­ diction. Most of them use subsite nomenclature adopted from a scheme created in the 1960s to describe the enzyme specifi­ cities [86,87].

In this model, the amino acid residues of a substrate in the

N-terminal direction from the cleaved bond are named from

P1, P2, P3, P4, to Pn. Likewise, the residues in the C-terminal direction are called from P1ʹ, P2ʹ, P3ʹ, P4ʹ to Pm.’ The peptidase database MEROPS mentioned earlier has cleavage site specifi­ city pattern for many individual peptidases whose specificity has has been well-characterized. In MEROPS, a summary page shows a cleavage site sequence logo to show the strength of the specificity preference using letters with different font sizes.

At the same time, it provides numerical specificity for each cleavage pattern at the positions P4-P4ʹ. For example, the cleavage site specificity of trypsin (S01.174) in MEROPS has both logo and number to indicate the specificity. The logo shows a cleavage pattern with a larger R to show that it is more likely to be cleaved and a smaller K to show that it is less likely to be cleaved at the P1 position. The specificity of the arginine is also indicated with 16 at P1, while the specificity of lysin is indicated with 6 at P1. The remaining residues have 0 at P1. There are multiple other in silico tools to predict clea­ vage sites in peptide, protein, and antibody sequences, such as PeptideCutter, PoPS, SitePrediction, CASVM, Cascleave,

Pripper, Cascleave 2.0, PROSPER, PROSPERous, GraBCas,

CaSPredictor, GPS-CCD, and iProt-Sub [88]. The cleavage sites of the most common peptidases are listed in Table 1.

However, we must be aware that the prediction is based on the sequence of peptides, proteins, and antibodies in a denatured condition, which is not the same as the biophy­ sical environments in which peptide, protein, and antibody drugs stay with a secondary, tertiary, and quaternary structure that will affect the interaction between peptidases and the substrates.

To obtain cleavage sites of peptides, proteins, and antibody drugs under biophysical conditions, in vitro experiments are more meaningful than in silico prediction. In a tube, peptide, protein, and antibody drugs can be incubated with pure single peptidase, a mixture of purified peptidases, or a crude extrac­ tion from a specific tissue in biophysical buffers with a desired pH at 37°C for a certain time duration for cleavage site

Table 1. The cleavage sites of the most common peptidases.

Peptidase Location Type Cleavage site specificity Pepsin

Stomach fluid Endopeptidase Leu, Phe, Ala, Tyr, Asp,

Glu Trypsin Small intestine fluid Endopeptidase Arg, Lys

Chymotrypsin Small intestine fluid Endopeptidase Phe, Tyr, Leu, Trp

Elastase Small intestine fluid Endopeptidase Val, Ile, Ala, Thr, Leu

Carboxypeptidase A1 Small intestine fluid Exopeptidase

Gly, Phe, Trp*, Val*, Ile* Carboxypeptidase B2 Small intestine fluid

Exopeptidase Lys*, Arg*, Gly, Ala, Leu, Tyr Carboxypeptidase

N Blood Exopeptidase Lys*, Arg*, Ala, Gly Aminopeptidase N Blood

Exopeptidase Ala, Leu, Arg, Trp*, Leu*, Phe* DPP4 Blood

Exopeptidase Pro, Ala Plasmin Blood Endopeptidase Arg, Lys

Furin Blood Endopeptidase Arg, Ser*, Lys at P2 NEP

Blood Endopeptidase Phe*, Leu*, Try*, Ile*, Val* Cathepsin D

Endolysosome Endopeptidase Leu, Phe, Ala, Tyr, Val*

Note: The data from the Human Protein Atlas (www.proteinatlas.org) and the

MEROPS database (www.ebi.ac.uk/merops/). * shows the cleavage site at P1ʹ.

6 X. LAI ET AL. identification. During the incubation, peptidases cleave the amino acid sequence from the most exposed sites and further cut the truncated peptide sequence even smaller. Although matrix assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometer (MS) have been applied to identify the cleavage sites in the past, the most effective approach currently is liquid chromatography coupled with mass spectrometry (LC–MS). A generic workflow is provided in Figure 1 for identifying the cleavage sites to optimize the sequence for protease resistance. Analysis of intact peptides, proteins, and antibodies using LC–MS has been extensively applied in pharmaceutical industry [89,90].

The glucoincretin hormone GLP-1(7–36)amide is limited as a drug because it is rapidly metabolized by DPP4, resulting in a very short half-life of less than 2 min [91]. To increase its proteolytic stability, replacing its second N-terminal amino acid alanine with serine produced a new peptide [Ser8]GLP-1 (7–36)amide. Both GLP-1(7–36)amide and [Ser8]GLP-1(7–36) amide were incubated with human plasma and analyzed by

LC-MS. The LC-MS successfully identified the N-terminally trun­ cated GLP-1 metabolite GLP-1(9–36)amide with a molecular mass of 3092.5 Da from the intact GLP-1(7–36)amide with a molecular mass of 3297.6 Da, while [Ser8]GLP-1(7–36)amide with a mass of 3317.7 did not result in any significant degra­ dation of the peptide. Multiple sets of data from the study demonstrated that the replacement of the second N-terminal amino acid alanine by serine improved the plasma stability of

GLP-1(7–36)amide [92]. Contrary to the previous findings that

GLP-1(7–36)amide metabolites are pharmacologically inactive, recent studies have showed that GLP-1(7–36)amide was cleaved by NEP, resulting in two metabolites of GLP-1(9–36) amide and GLP-1(28–36)amide. The two new peptides have cardioprotective and insulin-like effects in both animals and humans. To examine the proteolytic stability of GLP-1(9–36) amide and GLP-1(28–36)amide, an LC-MS method was devel­ oped for the quantitation of the intact peptides and identifica­ tion of their metabolites in hepatocyte incubations. A total of

9 and 5 metabolites were identified from the GLP-1(9–36) amide and GLP-1(28–36)amide incubation in human hepato­ cytes, respectively. The MS data showed that the 14 metabo­ lites are GLP-1(15–36), GLP-1(12–36), GLP-1(13–36), GLP-1(16–

36), GLP-1(18–36), GLP-1(19–36), GLP-1(20–36), GLP-1(21–36),

GLP-1(27–36), GLP-1(28–36), C6H10O5-GLP-1(28–36), C6H10O5- GLP-1(29–36), C6H10O5-GLP-1(31–36), GLP-1(29–36), and GLP- 1(31–36) [93].

A peptide named Pin2 (FWGALAKGALKLIPSLFSSFSKKD) is an α- helical polycationic antimicrobial peptides with high antimicrobial activity against a broad range of both Gram- positive and Gram-negative bacteria. However, its pharmaceu­ tical use is limited due to its proteolytic degradation. In a study, a D-diastereomer of Pin2, D-Pin2, was synthesized to examine whether D-Pin2 was able to improve its ability to resist proteolytic cleavage. Both Pin2 and D-Pin2 were incu­ bated with either bovine pancreatic trypsin (ratio 20:1) or

Figure 1. A generic workflow for identifying the cleavage sites and engineering the sequence to improve proteolytic stability. Initially, a peptide, protein, or antibody is selected based on potency for further optimization, including improving proteolytic stability. The selected molecule is incubated with a peptidase or a mixture of peptidases at certain pH, temperature, and duration. The samples are processed and analyzed by LC-MS. From the acquired MS data, the percentage of the intact molecule at various time points is calculated comparing to time 0. Meanwhile, the degradants are identified to provide cleavage site information. Based the cleavage sites, replacement of amino acids around the cleavage sites is performed, generating a new molecule for next round of testing and engineering.

EXPERT OPINION ON DRUG DISCOVERY 7 human elastase from leukocytes (ratio 20:1) for 4 h at 37°C.

The samples were analyzed using the RP-HPLC and LCQ DUO ion trap mass spectrometer. The results showed that D-Pin2 was not cleaved by either bovine pancreatic trypsin or human elastase up to 4 h, while Pin2 was rapidly digested by the enzymes. MS identified two degradants, LIPSLFSSFSK (1,225.4

Da) and FWGALAK (791.9 Da), from the trypsin incubation and three metabolites, WGALAKGA (919.7 Da), FWGALA (663.4 Da) and FWGALAKGALKLIPS (1,571.8 Da), from the elastase incuba­ tion [94]. The cleavage site identification by MS provides engineering guidance to increase proteolytic stability by repla­ cing the amino acids around the cleavage sites.

Generally, proteins and antibodies are resistant to pro­ teases because of their native conformation, unless they con­ tain loop regions that are accessible and flexible enough for a cleavage or they are unfolded [95]. It has been reported that bovine pancreatic ribonuclease A (RNase A) is cleaved by subtilisin Carlsberg even at 25°C. The speculation of cleavage site is between the residues Ala20 and Ser21 is that they are in a well-accessible loop between two helices based on the cleavage pattern and mechanism of ribonuclease S, proteinase

K, and elastase. To examine the hypothesis, a study has been carried out to compare the wild-type and mutated RNase

A (Ala20Pro) on the proteolytic resistance of the enzyme to proteinase K, subtilisin Carlsberg, and elastase under native conditions using an MALDI mass spectrometry and a time-of- flight mass spectrometer. To proteinase K and subtilisin

Carlsberg, A20P-RNase A was extremely stable, while wild- type RNase A was cleaved at Ala20 and Ser21. For the degra­ dation by elastase, A20P-RNase had two complementary frag­ ments, but MALDI-MS identified them as Lys1–Ser21 and

Ser22–Val124, indicating Ser21–Ser22 as the main primary cleavage site in A20P-RNase A. Using MS technique, it was found that the Ser21–Ser22 instead of Ala20–Ser21 was the primary cleavage site for the A20P-RNase A. In this study, mass spectrometry technique was demonstrated as a powerful tool to identify accurate cleavage sites that had only one amino acid difference. Meanwhile, the single mutation Ala20Pro strongly stabilized RNase A against the unspecific proteases of proteinase K and subtilisin Carlsberg, demonstrating a great potential of rational strategies of protein engineering to improve protein stability property [96].

4. Engineering Based on the identified cleavage sites, engineering should focus on amino acids around the sites, which are primarily at P1/P1ʹ and P2/P2ʹ, and possibly at P3/P3ʹ and P4/P4ʹ.

The replacement might change the efficacy, toxicity, and many other properties, while the proteolytic stability is increased. Scientists must balance the proteolytic stability and other desired properties. In the past decades, researchers have developed many strategies to improve peptide drug proteolytic stability by chemical modifica­ tions. For protein and antibody drugs, fewer approaches are available since they rely on cell culture rather than chemical syntheses. Methods discussed below are also exampled in Table 2.

4.1. Replacing the cleavage site amino acids with other proteinogenic amino acids

Replacing the cleavage site amino acids with other proteino­ genic amino acids might eliminate the cleavage. A typical example is the cleavage between positions 2 and 3 of the

GLP-1 peptides by DPP4. Native GLP-1 has Ala2 that is suscep­ tible to DPP4 and has a half-life of less than 2 min, which makes it impractical for therapeutic applications. Replacing the Ala2 with Gly2 in exenatide, albiglutide, albenatide, and lixisenatide eliminated the cleavage by DPP4. With the repla­ cement, exenatide and lixisenatide achieved 2.4 h and 2–4 h of plasma half-life, respectively. With the replacement plus albumin-binding capability, albiglutide and albenatide obtained 6–8 d and 8 d of plasma half-life, respectively [97].

The simple replacement significantly extends half-life by removing the cleavage site.

4.2. Replacing the cleavage site amino acids with other non-proteinogenic amino acids

In addition to the 20 proteinogenic amino acids, more than

800 natural amino acids are known in the literature, while thousands of synthetic amino acids have been reported [98].

The incorporation of non-proteinogenic amino acids that are not found in natural polypeptide chains into a peptide sequence likely will increase the metabolic stability of an analog because the new groups cannot be recognized by the same enzymes. The approach has been applied to many peptide drugs which are on the market [99]. Non- proteinogenic amino acids can be classified into many

Table 2. Examples of the most common engineering techniques to increase proteolytic stability.

Method Engineering* Molecule** Enzyme*** Ref.

Replacing the cleavage site amino acids with other proteinogenic amino acids

A2G GLP-1 DPP4 [97] Replacing the cleavage site amino acids with other non- proteinogenic amino acids

T25t MUC2 Carboxypeptidase [100] A4TfI P2-XaaFA Pepsin [103]

G2Aib GLP-1 DPP4 [104] Terminal modification N-acetylation

MART-1 Aminopeptidases [113] Cyclization Thioether bridge

Angiotensin Angiotensin- converting enzyme [119] N-Methylation of amide backbone

N-methylation iAβ5p Plasma [121] Glycosylation O-glycosylation G-CSF

Elastase [123] Note: * The first amino acid was replaced by the second amino acid at the position indicated by the number in the middle. **The engineering was carried out for the listed molecule. *** The engineering was performed to increase the molecule’s proteolytic stability against specific enzymes. A2G, replacing the Ala2 with Gly2. GLP-1, glucagon-like peptide-1. DPP4, dipeptidyl peptidase-4. T25t, replacing the L-Thr25 with D-Thr25. MUC2, Mucin 2. A4TfI, replacing the Ala4 with Tfl. P2-XaaFA, A peptide name. G2Aib, replacing the

Gly2 with Aib2. MART-1, melanoma-associated antigen recognized by T cells. iAβ5p, 5-amino acid β-sheet breaker peptide. G-CSF, granulocyte colony stimulating factor.

8 X. LAI ET AL. categories with various terms based on chirality (L vs. D), specific atoms introduced (such as F, S), specific groups intro­ duced (such as CH3), the backbone length (α- vs. β-amino acids), and their combinations. In this section, the most com­ monly used non-proteinogenic amino acids are discussed as examples.

High-molecular-weight MUC2 mucin glycoproteins are expressed in elevated levels of human colonic carcinoma. An epitope (18PTGTQ22) within the 1PTTTPITTTTTVTPTPTPTGTQT23 tandem-repeat unit of MUC2 glycoprotein is considered as synthetic vaccines in active specific immunotherapy to stimu­ late the B cell-specific immune responses of patients with

MUC2 related disease. Eleven peptides with the sequence of

15TPTPTGTQTPT25 were produced containing the PTGTQ epi­ tope and systematical substitution by one to three D-amino acids at C- and N-terminal regions. The peptides were incu­ bated with human serum and rat liver lysosomal preparation.

The results showed that the presence of D-amino acids only at the C terminus does not improve the degradation kinetics, the

D-amino acid(s) present only at the N terminus improved the stability, and replacement simultaneously at both ends of the peptide lead to compounds with almost full protection against serum degradation [100]. Others also demonstrated that the replacement of D-amino acid residues in linear and multichain polypeptide antigens improved their stability to proteoly­ sis [101].

The decoration of organic compounds with fluorine sub­ stituents is a well-known approach to medicinal chemists.

Fluorinated amino acids have also been used on a routine basis to improve peptide proteolytic stability. Common non- proteinogenic amino acids with fluorine are Abu, DefGly,

TfeGly, HfVal, Hfleu, TfIIe, 4-F-Phe, 3-F-Phe, 2-F-Phe, 2,4-F2- Phe, 4-CF3-Phe, 5-F-Tr, and 3-F-Try [102]. A study systemati­ cally investigated the effects of leucine, isoleucine, and their side-chain fluorinated variants on the proteolytic stability of 13 distinct peptides against α-chymotrypsin, pepsin, proteinase K, and elastase using LC-MS. Some of the replacements excep­ tionally increased the proteolytic stability upon the position of the fluorine substituent. 5,5,5-trifluoroisoleucine can signifi­ cantly protect peptides from proteolysis by all enzymes in the study when positioned N-terminal to the cleavage site, providing evidence for the application of fluorinated amino acids in the design of proteolytically stable peptide-based pharmaceuticals [103].

2-Aminoisobutyric acid (Aib, α-aminoisobutyric acid, α- methylalanine, or 2-methylalanine) is a commonly used non- proteinogenic amino acid to replace cleavage site amino acids. Due to the cleavage between positions 2 and 3 by

DPP4, native GLP-1 has a very short half-life. Besides, A2G was used in the engineering of exenatide, A2Aib was used to engineer albiglutide, albenatide, lixisenatide, traspoglutide, and semaglutide to eliminate the cleavage by DPP4.

Taspoglutide showed outstanding resistance to DPP4 and half-life of 13 h in human plasma upon subcutaneous admin­ istration without an acyl chain to bind albumin [104].

Semaglutide containing a fatty acid chain connected to

Lys26 through a miniPEG spacer achieved a half-life of 6–7 d in plasma, enabling a once-weekly subcutaneous adminis­ tration [105].

Aib is an example of the α-methylation of amino acid strategy. All amino acids can be α-methylated to provide an advantage against enzymatic attacks. Another example is α- MePhe that increases proteolytic stability. Tridecaptin A1 is a peptide that kills multidrug resistant bacteria. However, it can be rapidly hydrolyzed by peptidases such as BogQ and

TriF. In a study the tridecaptin analogues were used as model substrates to identify chemical modifications that can prevent peptide hydrolysis by BogQ and TriF. Tridecaptin A1 analogues were synthesized with α-MePhe9 to compare their proteolytic stability. In the BogQ incubation, Tridecaptin 16 with the α- MePhe9 showed 39% degradation, while Tridecaptin 2 with­ out the α-MePhe9 showed more than 90% degradation. In the

TriF incubation, Tridecaptin 23 with the α-MePhe9 had 0% degradation, while its analog Tridecaptin 2 had 100% degra­ dation. The α-MePhe9 significantly enhanced tridecaptin pro­ teolytic stability against the two peptidases BogQ and

TriF [106]. β-Amino acids are less prevalent in mammalian systems [107], while they commonly exist in natural products produced by plants [108], microorganisms [109], and marine organisms [110]. Compared to α-amino acids, β-amino acids have differ­ ent configurations, decreasing recognition for proteases [111].

Therefore, introducing β-amino acids into peptide sequences has a chance to improve peptide proteolytic stability. Each of the seven peptides including one peptide with only α-amino acids and the other six peptides with only β-amino acids or a mixture of α- and β-amino acids were incubated with multi­ ple peptidases such as pepsin, trypsin, chymotrypsin, elastase, proteinase K, subtilisin, and pronase. The results showed that only the peptide consisting of all α-amino acids was degraded by all tested peptidases except trypsin, while the other six peptides were not cleaved by all tested peptidases [112].

4.3. Terminal modification Since one of the common peptide degradations occurs through the action of aminopeptidases or carboxypeptidases, the proteolytic stability can be increased by blocking the N- or

C-terminal ends. Although many chemical modifications have been applied, the most popular ones are N-acetylation and

C-amidation.

MART-1 peptide (AAGIGILTV) is a melanoma peptide and commonly used in peptide-based vaccine trials for melanoma, but it is very unstable in human plasma with a half-life of 22 s. Two N-acetylated peptides, Cap-MART-1, and N-MART-1, were synthesized to compare with MART-1 in the incubation with plasma for 20 h at 37°C. The data confirmed that the native MART-1 peptide is unstable in human plasma, while

Cap-MART-1 and N-MART-1 showed significantly increased half-lives, showing that N-acetylation is capable of increasing peptide proteolytic stability [113].

About 50% of the bioactive peptides found in the nervous and endocrine systems possess a C-terminal α-amide group (-X-NH2). For most of these peptides, the presence of the α- amide moiety is important to their bioactivities. Peptide tyr­ osine tyrosine (PYY), amylin, calcitonin, oxytocin, vasopressin, and many others naturally have α-amidation at the C-terminal [114,115].

Besides the bioactivity benefits from the EXPERT OPINION ON DRUG DISCOVERY

9 C-amidation, the modification prevents degradation by pro­ teases. In a study to examine the effect of C-amidation against a periplasmic endoprotease Tsp, two pairs of peptides were used. Peptides BAS7 and BAS9 differed only in whether the

C-terminus was a carboxyl or a carboxamide group. The only difference between Peptides BAS8 and BAS10 was that BAS8 was C-amidated. After the peptides were incubated with Tsp for 5 h, BAS7, and BAS8 had less than 5% cleavage, while

BAS9 and BAS10 showed 50% and 45% cleavage, respec­ tively [116].

4.4. Cyclization Some peptides have a cyclic structure that enhances peptide proteolytic stability [117]. By cyclization location, it can be classified into head to tail, head to side chain, side chain to side chain, and side chain to tail [118]. Depending on the chemical type of the newly formed bond, it can be categorized as an amide, disulfide, ester, or thioether bridge bond.

Angiotensin (Ang)-(1–7) is an endogenous heptapeptide that plays an important role in the renin-angiotensin system, reduces liver fibrosis, and inhibits the growth of human lung cancer cells in vivo, making Ang-(1–7) an ideal therapeutic agent. However, it has less potential as a drug due to its rapid in vivo catabolism by angiotensin-converting enzymes and other proteases. cAng-(1–7) was made by forming a thioether bridge from positions 4 to 7 by the nisin biosynth­ esis machinery. The in vitro and in vivo assays showed that the proteolytic stability of cAng-(1–7) is strongly enhanced com­ pared with Ang-(1–7) [119]. This is a side chain to side-chain cyclization via a thioether bond.

4.5. N-Methylation of amide backbone N-methylation is a special type of chemical modification because methylation occurs at the N atom on the backbone.

This phenomenon has been observed in several macrocyclic peptides, such as enniatin, bouvardin, echinomycin, omphalo­ tin, and cyclosporine A [120]. A 5-amino acid β-sheet breaker peptide (iAβ5p)amide induce a dramatic reduction of amyloid deposition in two different transgenic Alzheimer’s models.

Although iAβ5p showed high blood–brain barrier permeabil­ ity, low toxicity, low immunogenicity, and high solubility, its half-life in vivo was short. To increase its proteolytic stability, a series of N-methylation modifications were incorporated into iAβ5p. The data showed that the N-methylation of one amide bond increased stability in vitro and achieved a 10-fold higher in vivo half-life [121].

4.6. Glycosylation Glycosylation is the reaction between a carbohydrate and another molecule. For example, glycans are linked to proteins and antibodies, mainly including N- and O-glycosylation. The

N-glycosylation is defined as N-acetylglucosamine (GlcNAc) attached by a β1-glycosidic linkage to the nitrogen atom of the amino group of Asn (N) at the consensus glycosylation motif Asn-X-Ser/Thr (in which X denotes any amino acid except for Pro). The O-glycosylation is used for glycans attached to the oxygen atom of the hydroxyl groups of Ser (S) or Thr (T) residue [105]. Glycosylation is critical for physio­ logical and pathological cellular functions. Glycosylation is rarely used for peptides to increase their proteolytic stability.

This approach is mainly for protein and antibody drugs [122].

Granulocyte-colony stimulating factor (G-CSF) is an 18–20 kDa hematopoietic cytokine that stimulates proliferation, dif­ ferentiation, and activation of committed precursors and mature neutrophils and is used to treat neutropenia. The natural G-CSF has one O-linked carbohydrate chain attached to Thr133. The recombinant G-CSF expressed in Escherichia coli is a non-glycosylated (NG) protein, while the one expressed in yeast or Chinese hamster ovary (CHO) cells have the

O-glycans. It has been noticed that NG G-CSF has reduced biological activity following incubation with serum, and the glycosylated form of the cytokine is relatively resistant to serum inactivation. An experiment was conducted by incubat­ ing both forms with elastase in a time- and concentration- dependent manner. The data showed that glycosylated G-CSF was more resistant to elastase degradation.

The O-glycosylation of G-CSF provided protection against the enzymatic effects of elastase [123].

Immunoglobulins G (IgGs) are N-glycosylated in the CH2 domain of the Fc with various glycans. These glycans provide numerous protections against oxidation, aggregation, pH, thermal, and proteolytic degradation [124]. Recent studies were conducted to compare multiple forms of Fc glycan structures against papain. Sialic acid (G2S2), β-galactose (G2),

N -acetylglucosamine (G0), and non-glycosylated IgGs were incubated with papain. The samples at different time points during the incubation were analyzed by MALDI-TOF MS and

HPLC. The results showed that non-glycosylated IgGs had a faster rate of papain-mediated degradation of the Fc domain. It was also observed that the G0 glycoform was at least two times more resistant to papain digestion than the G2 and G2S2 glycoforms. These data suggest that N-glycosylation play important roles in antibody proteolytic stability [125,126].

4.7. Other chemical modification Polyethylene glycol (PEG) is a water-soluble polymer with a structure of repeated units of polyether diols (either linear or branched) chemically formulated as H-(OCH2CH2)n-OH. It has a relative molecular weight of 200 to 8000 or above, but it exists as a mixture of different molecular weights with 44 Da difference. PEGylation is covalent binding with one or more

PEG chains with proteins, peptides, or small organic mole­ cules. The aim of most PEGylation is to increase the molecular weight of drugs to reduce kidney filtration for increasing the circulation half-life, but it also brings in multiple improved properties for peptide and protein drugs, including decreasing the accessibility for proteolytic enzymes [127–129]. However,

PEGylation has some disadvantages, such as immunogenicity of the PEG, and heterogeneity of the conjugated molecules [130,131].

Acylation with fatty acids is mainly used as an approach to extend peptide half-life via non-covalent association of the fatty acid alkyl moiety with albumin for peptide drugs

10 X. LAI ET AL. such liraglutide [132] and semaglutide [133]. Native GLP-1 and GLP-1 derivatives with various fatty acids at different positions were incubated in DPP4 solution with or without the existence of human serum albumin. Without HSA, some derivatives showed slower degradation rates than native

GLP-1, indicating that the acyl chain did improve the pro­ teolytic stability. In the presence of HSA, the results showed that the derivatives had improved stabilities and that the stabilities were correlated with the length of the carbon chain in fatty acids. The C-16 fatty acid conjugates were more stable than the C-12 and C-8 conjugates [134]. The data suggest that the fatty acid acylation improves proteo­ lytic stability via acting like other modification to affect enzymatic recognition to avoid peptidase attacks and bind­ ing albumin to increase physical obstacles for peptidase to access the cleavage sites.

Many chemical modifications of amino acids have been published for various purposes [135–137]. Since biotin has a very strong affinity for streptavidin and avidin, biotin- conjugated peptides and proteins are commonly used in immunoassays to enrich the target molecules [138].

Dansylation and fluorescein conjugation are used in fluores­ cence-based assays to study protein–protein interaction, pro­ tein localization, or protein quantification [139,140]. In addition to the desired purposes, the modifications might block the enzymatic access to the cleavage sites, increasing the proteolytic stability as an additional benefit. However, they are not commonly used in peptides, proteins, and antibody drugs during the SAR process with the specific goal of increas­ ing proteolytic stability. Therefore, they are not detailed in this review.

5. Conclusion In this review, we focus on the proteolytic stability of peptides, proteins, and antibodies for drug discovery and development.

Proteolytic stability is one of the few key properties that enable a molecule to become a drug. To improve the proteo­ lytic stability, we need to know what enzymes are involved, how to identify the cleavage sites, and how to engineer the sequences. A big picture of peptidases was discussed through existing databases, specific peptidase mixtures were intro­ duced based on organ and tissue types, and some well- known individual peptidases were reviewed for their applica­ tions. The approaches to identify cleavage sites were spot­ lighted on mass spectrometry-based techniques since it is the most accurate and convenient technique to find the answer. The methods to replace amino acids around the clea­ vage sites were summarized according to a sequence of prac­ tical usages. The most common and successful methods were discussed at the beginning.

Due to the complexity of interaction between substrates and enzymes, the prediction of cleavage sites and the selec­ tion of the amino acid replacement need final confirmation from experimental data. The process of trial and error con­ tinues until the desired results are achieved on a case-by-case basis. This review provides the most relevant information to shorten the cycle time and achieve the goal in an effective manner.

6. Expert opinion After reviewing peptidases that have been reported, methods used to identify cleavage sites, and approaches available to engineer the amino acids around the cleavage sites, a practical question is what the best strategy is to apply the learnings from the literature to daily projects. Before peptides, proteins, and antibody drugs are made, how to design the sequence to avoid the well-known peptidase degradation? After they are generated, how to select proper peptidases to practically test them without either under- or over-degradations? In this review, we propose a 3-step strategy for peptidase-focused design and process that is shown in Figure 2.

In the first step, the peptidases highlighted in green, such as trypsin, carboxypeptidases, aminopeptidases, and DPP4, need to be considered in the initial sequence design. For peptides in denatured or biophysical conditions, trypsin very specifically cleaves amino acids of lysine and arginine at P1 position since peptides really do not have complex higher levels of structure. For proteins and antibodies, the higher- level structure does affect the accessibility of trypsin. The prediction must consider the effect from the structure to decide whether a specific arginine or lysine residue is liable to trypsin. Due to the concern from carboxypeptidases,

C-amidation is a great choice to protect the C-terminal against carboxypeptidases. To avoid cleavage by aminopeptidases and DPP4, Aib replacement at position 2 plus other learning from the literature for certain amino acids at position 1 is a good starting point to address the issue. For example, Met,

Ser, Ala, Thr, Val, or Gly at N-terminus typically have stronger proteolytic stability than Phe, Leu, Asp, Lys, or Arg [141].

The second step is to focus on a few individual peptidases highlighted in red in Figure 2, including pepsin, chymotrypsin, elastase, DPP4, and cathepsin D. These peptidases are well- known endopeptidases to cleave peptides, proteins, and anti­ bodies. Due to their high abundance, the easy availability, and convenient assay performance, these peptidases should be examined first.

In the last step, more complex peptidase mixtures high­ lighted in purple in Figure 2 are selected for testing the proteolytic stability. For peptide, protein, and antibody drugs that are absorbed in the stomach, stomach fluid should be selected since it has more complex enzymes than pepsin alone. For the drugs to be absorbed in the small intestine, small intestine fluid is a good choice since it is more complex than trypsin, chymotrypsin, and elastase. For both oral and subcutaneous drugs, peptidases in the blood and tissues are another set of challenges. Because there are too many carbox­ ypeptidases and aminopeptidases to select for individual test­ ing, blood, plasma, or serum is a great source to examine all potential cleavages from a broader group of peptidases. When the drugs are internalized into cells, they encounter pepti­ dases in the endolysosome rather than the cytoplasm.

Therefore, using subcellular fractions of the endolysosome is the primary choice. Whole crude tissue homogenates from the stomach, small intestine, liver, kidney, heart, and others is another resource to use. However, a risk of over degradation exists when the whole homogenate is used since all pepti­ dases are released from all components in a cell. After the

EXPERT OPINION ON DRUG DISCOVERY 11 three steps, the proteolytic stability issue should be addressed well. If the drugs still have a proteolytic stability issue, an in vivo approach should be applied to identify metabolites from animal bodily fluids or tissues.

Improving the proteolytic stability is a spiraling up process sequenced by testing and engineering. There are many ways to engineer amino acids, but the choice must consider the cost and properties affected by the changes of the amino acids. The most practical strategy to be recommended follows a priority of a different proteinogenic amino acid, a D-form of amino acid, an α-methylation of amino acid, a simply modified amino acid by fluorine or other atoms, a β-amino acid,

N-Methylation, and others.

Acknowledgments The authors would like to thank Christopher Reutter for editing and proofreading the manuscript.

Funding This manuscript was not funded.

Declaration of interest X Lai, J Tang, and M ElSayed are all employees of Eli Lilly and Company.

The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures Peer reviewers in this manuscript have no relevant financial or other relationships to disclose.

References Papers of special note have been highlighted as either of interest (•) or of considerable interest (••) to readers.

Figure 2. A 3-step strategy of peptidase-focused design and process to improve proteolytic stability. Based on the delivery approach, different sets of peptides are considered. Compared to subcutaneous injection, oral delivery needs additional testing of peptidases from the gastrointestinal (GI) system. For each delivery approach, peptidases from bodily fluids, such as GI fluid or blood, and tissues are evaluated by using individual peptidases or a mixture of them. In this 3-step strategy, the first one is to think of the peptidases highlighted in green color in the initial sequence design. Trypsin is the most predictable peptidase. Cleavages caused by carboxypeptidases, aminopeptidases, and DPP4 are prevented by C-amidation and N-terminal replacement. The second step is to focus on a few individual peptidases highlighted in red because of their high abundance, the easy availability, and convenient assay performance. The last step is for more complex peptidase mixtures highlighted in purple. Each type of bodily fluid has complex peptidases. When the drugs are internalized into cells in different tissues, they might encounter specific peptidases. Using subcellular fractions such as endolysosome or whole crude tissue homogenate from the stomach, small intestine, liver, kidney, heart, and others is practical to perform. However, a risk of over degradation exists when the whole homogenate is used since all peptidases are release from all components in a cell.

12 X. LAI ET AL.

1. Ibrahim YHY, Regdon G Jr., Hamedelniel EI, et al. Review of recently used techniques and materials to improve the efficiency of orally administered proteins/peptides. Daru. 2020 Jun;28 (1):403–416.

•• Challenges of improving the bioavailability of orally adminis­ tered biopharmaceutical drugs

2. de la Torre BG, Albericio F. The pharmaceutical industry in 2019. An analysis of FDA drug approvals from the perspective of molecules.

Molecules. 2020 Feb 9;25(3):3.

3. Lee AC, Harris JL, Khanna KK, et al. A comprehensive review on current advances in peptide drug development and design.

Int J Mol Sci. 2019 May 14;20(10). 10.3390/ijms20102383.

•• Challenges of peptide drug research and development

4. Lau JL, Dunn MK. Therapeutic peptides: historical perspectives, current development trends, and future directions. Bioorg Med

Chem. 2018Jun1;26(10):2700–2707.

•• Challenges of peptide drug research and development.

5. Usmani SS, Bedi G, Samuel JS, et al., THPdb: database of

FDA-approved peptide and protein therapeutics. PLoS One. 12(7): e0181748. 2017..

•• Comprehensive information on 239 US FDA approved thera­ peutic peptides and proteins and their 380 drug variants.

6. Lu RM, Hwang YC, Liu IJ, et al. Development of therapeutic anti­ bodies for the treatment of diseases. J Biomed Sci. 2020 Jan 2;27 (1):1.

•• Challenges of antibody drug research and development

7. Otvos L Jr., Wade JD. Current challenges in peptide-based drug discovery. Front Chem. 2014;2:62.

8. Lagasse HA, Alexaki A, Simhadri VL, et al. Recent advances in (therapeutic protein) drug development. F1000Res. 2017;6:113.

•• Challenges of protein drug research and development

9. Le Basle Y, Chennell P, Tokhadze N, et al. Physicochemical stability of monoclonal antibodies: a review. J Pharm Sci. 2020 Jan;109 (1):169–190.

10. Bottger R, Hoffmann R, Knappe D. Differential stability of therapeu­ tic peptides with different proteolytic cleavage sites in blood, plasma and serum. PLoS One. 2017;12(6):e0178943.

•• Proteolytic stability of therapeutic peptides 11. Varkhede N, Bommana R, Schoneich C, et al. Proteolysis and oxida­ tion of therapeutic proteins after intradermal or subcutaneous administration. J Pharm Sci. 2020 Jan;109(1):191–205.

•• Proteolytic stability of therapeutic proteins 12. Brezski RJ, Jordan RE. Cleavage of IgGs by proteases associated with invasive diseases: an evasion tactic against host immunity?

MAbs. 2010 May-Jun;2(3):212–220.

•• Proteolytic stability of therapeutic antibodies

13. Brayden DJ, Hill TA, Fairlie DP, et al. Systemic delivery of peptides by the oral route: formulation and medicinal chemistry approaches.

Adv Drug Deliv Rev. 2020 May;29;157:2-36.

14. Yamamoto A, Ukai H, Morishita M, et al. Approaches to improve intestinal and transmucosal absorption of peptide and protein drugs. Pharmacol Ther. 2020 Jul;211:107537.

15. Bai Y, Zhou R, Wu L, et al. Nanoparticles with surface features of dendritic oligopeptides as potential oral drug delivery systems.

J Mater Chem B. 2020 Apr 1;8(13):2636–2649.

16. Bajracharya R, Song JG, Back SY, et al. Recent advancements in non-invasive formulations for protein drug delivery. Comput Struct

Biotechnol J. 2019;17:1290–1308.

17. Madani F, Hsein H, Busignies V, et al. An overview on dosage forms and formulation strategies for vaccines and antibodies oral delivery. Pharm Dev Technol. 2020 Feb;25(2):133–148.

18. Reilly RM, Domingo R, Sandhu J. Oral delivery of antibodies. Clinical

Pharmacokinetics. 1997 Apr;32(4):313–323.

19. Wright L, Barnes T, Prestidge C. Oral delivery of protein-based ther­ apeutics: gastroprotective strategies, physiological barriers and in vitro permeability prediction. Int J Pharm. 2020 Jun;3:119488.

20. Muheem A, Shakeel F, Jahangir MA, et al. A review on the strate­ gies for oral delivery of proteins and peptides and their clinical perspectives. Saudi Pharm J. 2016 Jul;24(4):413–428.

21. Lee YH, Perry BA, Labruno S, et al. Impact of regional intestinal pH modulation on absorption of peptide drugs: oral absorption stu­ dies of salmon calcitonin in beagle dogs. Pharm Res. 1999 Aug;16 (8):1233–1239. .

22. Liu H, Tang R, Pan WS, et al. Potential utility of various protease inhibitors for improving the intestinal absorption of insulin in rats.

J Pharm Pharmacol. 2003 Nov;55(11):1523–1529.

23. Eng J, Kleinman WA, Singh L, et al. Isolation and characterization of exendin-4, an exendin-3 analogue, from Heloderma suspectum venom. Further evidence for an exendin receptor on dispersed acini from guinea pig pancreas. J Biol Chem. 1992 Apr 15;267 (11):7402–7405.

24. Sun C, Chen L, Shen Z. Mechanisms of gastrointestinal microflora on drug metabolism in clinical practice. Saudi Pharm J. 2019 Dec;27 (8):1146–1156.

25. Rawlings ND, Waller M, Barrett AJ, et al. MEROPS: the database of proteolytic enzymes, their substrates and inhibitors. Nucleic Acids

Res. 2014 Jan;42(D1):D503–9.

•• Introduction of the database 26. Hartley BS. Proteolytic enzymes. Annu Rev Biochem. 1960;29 (1):45–72.

27. Klein T, Eckhard U, Dufour A, et al. Proteolytic cleavage- mechanisms, function, and “omic” approaches for a near- ubiquitous posttranslational modification. Chem Rev. 2018 Feb

14;118(3):1137–1168.

28. Rawlings ND, Tolle DP, Barrett AJ. Evolutionary families of pepti­ dase inhibitors. Biochem J. 2004 Mar 15;378(Pt 3):705–716.

29. Uhlen M, Bjorling E, Agaton C, et al. A human protein atlas for normal and cancer tissues based on antibody proteomics. Mol Cell

Proteomics. 2005 Dec 4;4(12):1920–1932. .

• Introduction of the database 30. Uhlen M, Fagerberg L, Hallstrom BM, et al. Proteomics.

Tissue-based map of the human proteome. Science. 2015 Jan

23;347(6220):1260419.

• Introduction of the database 31. Sjostedt E, Zhong W, Fagerberg L, et al. An atlas of the protein-coding genes in the human, pig, and mouse brain.

Science. 2020 Mar 6;367(6482):eaay5947.

• Introduction of the database 32. Thul PJ, Akesson L, Wiking M, et al. A subcellular map of the human proteome. Science. 2017 May 26;356(6340):eaal3321.

• Introduction of the database 33. Uhlen M, Karlsson MJ, Zhong W, et al. A genome-wide transcrip­ tomic analysis of protein-coding genes in human blood cells.

Science. 2019 Dec 20;366(6472):eaax9198.

• Introduction of the database 34. Uhlen M, Karlsson MJ, Hober A, et al. The human secretome. Sci

Signal. 2019 Nov 26;12(609):eaaz0274.

• Introduction of the database 35. Uhlen M, Zhang C, Lee S, et al. A pathology atlas of the human cancer transcriptome. Science. 2017 Aug 18;357(6352):eaan2507.

• Introduction of the database 36. Ulleberg EK, Comi I, Holm H, et al. Human gastrointestinal juices intended for use in in vitro digestion models. Food Digestion. 2011

Dec 2;2(1–3):52–61.

• Introduction of gastrointestinal fluid 37. Hu S, Loo JA, Wong DT. Human body fluid proteome analysis.

Proteomics. 2006 Dec 6(23):6326–6353.

• Introduction of bodily fluid 38. Boland M. Human digestion--a processing perspective. J Sci Food

Agric. 2016 May;96(7):2275–2283.

39. Wang J, Yadav V, Smart AL, et al. Toward oral delivery of biophar­ maceuticals: an assessment of the gastrointestinal stability of 17 peptide drugs. Mol Pharm. 2015 Mar 2;12(3):966–973.

40. Smart AL, Gaisford S, Basit AW. Oral peptide and protein delivery: intestinal obstacles and commercial prospects. Expert Opin Drug

Deliv. 2014 Aug;11(8):1323–1335.

41. Carter CS. The Oxytocin–Vasopressin pathway in the context of love and fear. Front Endocrinol (Lausanne). 2017;8:356.

EXPERT OPINION ON DRUG DISCOVERY 13 42. Baribeau DA, Anagnostou E. Oxytocin and vasopressin: linking pituitary neuropeptides and their receptors to social neurocircuits. Front Neurosci. 2015;9:335.

43. Fjellestad-Paulsen A, Söderberg-Ahlm C, Lundin S. Metabolism of vasopressin, oxytocin, and their analogues in the human gastro­ intestinal tract. Peptides. 19951995 Jan 01;16(6):1141–1147.

44. Ahmad S, Kumar V, Ramanand KB, et al. Probing protein stability and proteolytic resistance by loop scanning: a comprehensive mutational analysis. Protein Sci. 2012 Mar;21(3):433–446.

45. Hall MP. Biotransformation and in vivo stability of protein biother­ apeutics: impact on candidate selection and pharmacokinetic profiling. Drug Metab Dispos. 2014 Nov;42(11):1873–1880.

46. Liao S, Liang Y, Zhang Z, et al. In vitro metabolic stability of exendin-4: pharmacokinetics and identification of cleavage products. PLoS One. 2015;10(2):e0116805.

47. Ciechanover A. Intracellular protein degradation: from a vague idea thru the lysosome and the ubiquitin–proteasome system and onto human diseases and drug targeting. Cell Death Differ. 2005 Sep;12 (9):1178–1190.

48. Pillay CS, Elliott E, Dennison C. Endolysosomal proteolysis and its regulation. Biochem J. 2002 May 1;363(Pt 3):417–429.

49. Ferrone M, Raimondo M, Scolapio JS. Pancreatic enzyme pharmacotherapy. Pharmacotherapy. 2007 Jun;27(6):910–920.

50. Littman A, Hanscom DH. Current concepts: pancreatic extracts.

N Engl J Med. 1969 Jul 24;281(4):201–204.

51. Zheng C, Liu Y, Zhou Q, et al. Capillary electrophoresis with non­ covalently bilayer-coated capillaries for stability study of allergenic proteins in simulated gastrointestinal fluids. J Chromatogr B Analyt

Technol Biomed Life Sci. 2010 Oct 15;878(28):2933–2936.

52. Heda R, Toro F, Tombazzi CR. Physiology, Pepsin. Treasure Island (FL): StatPearls; 2020.

53. Fruton JS. A history of pepsin and related enzymes. Q Rev Biol.

2002 Jun;77(2):127–147.

54. Gray GM, Cooper HL. Protein digestion and absorption.

Gastroenterology. 1971 Oct;61(4):535–544.

55. Kassell B, Kay J. Zymogens of proteolytic enzymes. Science. 1973

Jun 8;180(4090):1022–1027.

56. Lam MPY, Lau E, Liu X, et al. Sample Preparation for Glycoproteins.

In: Pawliszyn J, editor. Comprehensive Sampling and Sample

Preparation. Oxford: Academic Press; 2012; p. 307-322.

57. Chesnut CH 3rd, Azria M, Silverman S, et al. Salmon calcitonin: a review of current and future therapeutic indications.

Osteoporos Int. 2008 Apr;19(4):479–491.

58. Ryan SM, Frias JM, Wang X, et al. PK/PD modelling of comb-shaped

PEGylated salmon calcitonin conjugates of differing molecular weights. J Control Release. 2011 Jan 20;149(2):126–132.

59. Panopoulos AD, Watowich SS. Granulocyte colony-stimulating factor: molecular mechanisms of action during steady state and

‘emergency’ hematopoiesis. Cytokine. 2008 Jun;42(3):277–288.

60. Jensen-Pippo KE, Whitcomb KL, DePrince RB, et al. Enteral bioavailability of human granulocyte colony stimulating factor conjugated with poly(ethylene glycol). Pharm Res. 1996 Jan;13 (1):102–107.

61. Carrere J, Figarella C, Guy O, et al. Human pancreatic chymotrypsi­ nogen A: a non-competitive enzyme immunoassay, and molecular forms in serum and amniotic fluid. Biochim Biophys Acta. 1986 Aug

6;883(1):46-531.

62. Ma W, Tang C, Lai L. Specificity of trypsin and chymotrypsin: loop-motion-controlled dynamic correlation as a determinant.

Biophys J. 2005 Aug;89(2):1183–1193.

63. de Oliveira EB, Salgado MCO. Chapter 584 - Pancreatic elastases. In:

Rawlings ND, Salvesen G, editors. Handbook of proteolytic enzymes. Third Edition ed. Cambridge, Massachusetts: Academic

Press; 2013. p. 2639–2645.

64. Bode W, Meyer E Jr., Powers JC. Human leukocyte and porcine pancreatic elastase: x-ray crystal structures, mechanism, substrate specificity, and mechanism-based inhibitors. Biochemistry. 1989

Mar 7;28(5):1951–1963.

65. Werb Z, Banda MJ, McKerrow JH, et al. Elastases and elastin degradation. J Invest Dermatol. 1982 Jul;79(1):154s–159s. .

66. Fricker L. Carboxypeptidases. In: Enna SJ, Bylund DB, editors. xPharm: the comprehensive pharmacology reference. New York:

Elsevier; 2007. p. 1–4.

67. Clauser E, Gardell SJ, Craik CS, et al. Structural characterization of the rat carboxypeptidase A1 and B genes. Comparative analysis of the rat carboxypeptidase gene family. J Biol Chem. 1988 Nov

25;263(33):17837–17845.

68. Merkler DJ. C-Terminal amidated peptides: production by the in vitro enzymatic amidation of glycine-extended peptides and the importance of the amide to bioactivity. Enzyme Microb

Technol. 19941994 Jun 01;16(6):450–456.

69. Taylor A. Aminopeptidases: structure and function. FASEB J. 1993

Feb 1;7(2):290–298.

70. Mulvihill EE, Drucker DJ. Pharmacology, physiology, and mechan­ isms of action of dipeptidyl peptidase-4 inhibitors. Endocr Rev.

2014 Dec;35(6):992–1019.

71. Cho YM, Kieffer TJ. Chapter Four - K-cells and glucose-dependent insulinotropic polypeptide in health and disease. In: Litwack G, editor. Vitamins & Hormones. Vol. 84. Cambridge, Massachusetts:

Academic Press; 2010. p. 111–150.

•• The proteolytic stability of GIP and GLP-1 against DPP4

72. Tran KL, Park YI, Pandya S, et al. Overview of glucagon-like peptide-1 receptor agonists for the treatment of patients with type 2 diabetes. Am Health Drug Benefits. 2017 Jun;10(4):178–188.

•• The proteolytic stability of GIP and GLP-1 against DPP4

73. Aisina RB, Mukhametova LI. [Structure and functions of plasmino­ gen/plasmin system]. Bioorg Khim. 2014 Nov-Dec;40(6):642–657.

74. Ogston D. Biochemistry of the plasmin system. J Clin Pathol Suppl (R Coll Pathol). 1980;14(Suppl 14):5–9.

75. Kibirev VK, Osadchuk TV, Radavskii IL. [Furin and its biological role].

Ukr Biokhim Zh (1999) 2007 Nov-Dec;79(6):5–18.

76. Thomas G. Furin at the cutting edge: from protein traffic to embry­ ogenesis and disease. Nat Rev Mol Cell Biol. 2002 Oct;3 (10):753–766.

77. Simoes PSR, Visniauskas B, Perosa SR, et al. Expression and activity of thimet oligopeptidase (TOP) are modified in the hippocampus of subjects with temporal lobe epilepsy (TLE). Epilepsia. 2014 May;55 (5):754–762. .

78. Kimura Y, Ikuta K, Kimura T, et al. Nardilysin regulates inflammation, metaplasia, and tumors in murine stomach. Sci Rep. 2017 Feb;23 (7):43052. .

79. Hashimoto S, Amaya F, Oh-Hashi K, et al. Expression of neutral endopeptidase activity during clinical and experimental acute lung injury. Respir Res. 2010 Nov;29(11):164. .

80. Turk V, Stoka V, Vasiljeva O, et al. Cysteine cathepsins: from struc­ ture, function and regulation to new frontiers. Biochim Biophys

Acta. 2012 Jan;1824(1):68–88.

•• Introduction of cathepsins 81. Sevlever D, Jiang P, Yen SH. Cathepsin D is the main lysosomal enzyme involved in the degradation of alpha-synuclein and gen­ eration of its carboxy-terminally truncated species. Biochemistry.

2008Sep9;47(36):9678–9687.

•• Introduction of cathepsins.

82. Oberle C, Huai J, Reinheckel T, et al. Lysosomal membrane permea­ bilization and cathepsin release is a Bax/Bak-dependent, amplifying event of apoptosis in fibroblasts and monocytes. Cell Death Differ.

2010 Jul;17(7):1167–1178. .

•• Introduction of cathepsins 83. McDonnell AM, Dang CH. Basic review of the cytochrome p450 system. J Adv Pract Oncol. 2013 Jul;4(4):263–268.

84. Lynch T, Price A. The effect of cytochrome P450 metabolism on drug response, interactions, and adverse effects. Am Fam Physician.

2007 Aug 1;76(3):391–396.

85. Yao JF, Yang H, Zhao YZ, et al. Metabolism of Peptide Drugs and

Strategies to Improve their Metabolic Stability. Curr Drug Metab.

2018;19(11):892–901. .

86. Schechter I, Berger A. On the size of the active site in proteases. I.

Papain. Biochem Biophys Res Commun. 1967 Apr 20;27(2):157–162.

.

•• Introduction of cleavage patterns.

14 X. LAI ET AL.

87. Schechter I, Berger A. On the active site of proteases. 3. Mapping the active site of papain; specific peptide inhibitors of papain.

Biochem Biophys Res Commun. 1968Sep6;32(5):898–902.

•• Introduction of cleavage patterns.

88. Song J, Wang Y, Li F, et al. iProt-Sub: a comprehensive package for accurately mapping and predicting protease-specific substrates and cleavage sites. Brief Bioinform. 2019 Mar 25;20(2):638–658.

•• Introduction of cleavage prediction tools 89. Kang L, Weng N, Jian W. LC-MS bioanalysis of intact proteins and peptides. Biomed Chromatogr. 2020 Jan;34(1):e4633.

90. van den Broek I, van Dongen WD. LC-MS-based quantification of intact proteins: perspective for clinical and bioanalytical applications. Bioanalysis. 2015;7(15):1943–1958.

91. Araujo F, Fonte P, Santos HA, et al. Oral delivery of glucagon-like peptide-1 and analogs: alternatives for diabetes control? J Diabetes

Sci Technol. 2012 Nov 1;6(6):1486–1497.

92. Ritzel U, Leonhardt U, Ottleben M, et al. A synthetic glucagon-like peptide-1 analog with improved plasma stability. J Endocrinol.

1998 Oct;159(1):93–102.

93. Sharma R, McDonald TS, Eng H, et al. In vitro metabolism of the glucagon-like peptide-1 (GLP-1)-derived metabolites GLP-1 (9-36) amide and GLP-1(28-36)amide in mouse and human hepatocytes.

Drug Metab Dispos. 2013 Dec;41(12):2148–2157.

•• Understanding the application of LC-MS in cleavage site identification

94. Carmona G, Rodriguez A, Juarez D, et al. Improved protease stabi­ lity of the antimicrobial peptide Pin2 substituted with D-amino acids. Protein J. 2013 Aug;32(6):456–466.

•• Understanding the application of LC-MS in cleavage site identification

95. Hubbard SJ. The structural aspects of limited proteolysis of native proteins. Biochim Biophys Acta. 1998 Feb 17;1382(2):191–206.

96. Markert Y, Koditz J, Mansfeld J, et al. Increased proteolytic resis­ tance of ribonuclease A by protein engineering. Protein Eng. 2001

Oct;14(10):791–796.

97. Alavi SE, Cabot PJ, Moyle PM. Glucagon-Like Peptide-1 Receptor

Agonists and Strategies To Improve Their Efficiency. Mol Pharm.

2019Jun3;16(6):2278–2295.

•• Example of peptide engineering to increase proteolytic stability.

98. Xue YP, Cao CH, Zheng YG. Enzymatic asymmetric synthesis of chiral amino acids. Chem Soc Rev. 2018 Feb 19;47(4):1516–1561.

99. Narancic T, Almahboub SA, O’Connor KE. Unnatural amino acids: production and biotechnological potential. World J Microbiol

Biotechnol. 2019 Apr 8;35(4):67.

100. Tugyi R, Uray K, Ivan D, et al. Partial D-amino acid substitution: improved enzymatic stability and preserved Ab recognition of a MUC2 epitope peptide. Proc Natl Acad Sci U S A. 2005 Jan

11;102(2):413–418.

101. Sela M, Zisman E. Different roles of D-amino acids in immune phenomena. FASEB J. 1997 May;11(6):449–456.

102. Huhmann S, Koksch B. Fine-Tuning the proteolytic stability of pep­ tides with fluorinated amino acids. European J Org Chem.

2018;2018(27–28):3667–3679.

103. Huhmann S, Stegemann A-K, Folmert K, et al. Position-dependent impact of hexafluoroleucine and trifluoroisoleucine on protease digestion. Beilstein J Org Chem. 2017;13:2869–2882.

104. Dong JZ, Shen Y, Zhang J, et al. Discovery and characterization of taspoglutide, a novel analogue of human glucagon-like peptide-1, engineered for sustained therapeutic activity in type 2 diabetes.

Diabetes Obes Metab. 2011 Jan;13(1):19–25.

•• Example of peptide engineering to increase proteolytic stability

105. Madsbad S, Kielgast U, Asmar M, et al. An overview of once-weekly glucagon-like peptide-1 receptor agonists--available efficacy and safety data and perspectives for the future. Diabetes Obes Metab.

2011 May;13(5):394–407.

106. Bann SJ, Ballantine RD, McCallion CE, et al.

A chemical-intervention strategy to circumvent peptide hydrolysis by d-stereoselective peptidases. J Med Chem. 2019

Nov 27;62(22):10466–10472.

107. Griffith OW. Beta-amino acids: mammalian metabolism and utility as alpha-amino acid analogues. Annu Rev Biochem. 1986;55 (1):855–878.

108. Nicolaou KC, Dai W-M, Guy RK. Chemistry and Biology of Taxol.

Angew Chem Int Ed Engl. 1994;33(1):15–44.

109. Minami Y, Yoshida K-I, Azuma R, et al. Structure of an aromatization product of C-1027 chromophore. Tetrahedron Lett. 19931993 Apr

16;34(16):2633–2636.

110. Cardillo G, Tomasini C. Asymmetric synthesis of ß-amino acids and α-substituted β-amino acids [10.1039/CS9962500117]. Chem Soc

Rev. 1996;25(2):117–128.

111. Cabrele C, Martinek TA, Reiser O, et al. Peptides containing beta-amino acid patterns: challenges and successes in medicinal chemistry. J Med Chem. 2014 Dec 11;57(23):9718–9739.

112. Hook DF, Bindschadler P, Mahajan YR, et al. The proteolytic stability of ‘designed’ beta-peptides containing alpha-peptide-bond mimics and of mixed alpha, beta-peptides: application to the construction of MHC-binding peptides. Chem Biodivers. 2005 May;2(5):591–632.

113. Brinckerhoff LH, Kalashnikov VV, Thompson LW, et al. Terminal modifications inhibit proteolytic degradation of an immunogenic

MART-1(27-35) peptide: implications for peptide vaccines.

Int J Cancer. 1999 Oct 29;83(3):326–334.

114. Kim K-H, Seong BL. Peptide amidation: production of peptide hormones in vivo and in vitro. Biotechnol Bioprocess Eng.

20012001 Aug 01;6(4):244–251.

115. Eipper BA, Mains RE. Peptide alpha-amidation. Annu Rev Physiol.

1988;50(1):333–344.

116. Keiler KC, Silber KR, Downard KM, et al. C-terminal specific protein degradation: activity and substrate specificity of the Tsp protease.

Protein Sci. 1995 Aug;4(8):1507–1515.

117. Borchardt RT. Optimizing oral absorption of peptides using pro­ drug strategies. J Control Release. 1999 Nov 1;62(1–2):231–238.

118. Khatri B, Nuthakki VR, Chatterjee J. Strategies to enhance metabolic stabilities. Methods Mol Biol. 2019;2001:17–40.

•• Detailed description of cyclization technique to increase pro­ teolytic stability

119. Kluskens LD, Nelemans SA, Rink R, et al. Angiotensin-(1-7) with thioether bridge: an angiotensin-converting enzyme-resistant, potent angiotensin-(1-7) analog. J Pharmacol Exp Ther. 2009

Mar;328(3):849–854. .

120. Chatterjee J, Rechenmacher F, Kessler H. N-methylation of peptides and proteins: an important element for modulating biological functions. Angew Chem Int Ed Engl. 2013Jan2;52(1):254–269.

•• Detailed description of N-methylation technique to increase proteolytic stability.

121. Adessi C, Frossard MJ, Boissard C, et al. Pharmacological profiles of peptide drug candidates for the treatment of Alzheimer’s disease.

J Biol Chem. 2003 Apr 18;278(16):13905–13911.

122. Sola RJ, Griebenow K. Effects of glycosylation on the stability of protein pharmaceuticals. J Pharm Sci. 2009 Apr;98(4):1223–1245.

123. Carter CR, Whitmore KM, Thorpe R. The significance of carbohy­ drates on G-CSF: differential sensitivity of G-CSFs to human neu­ trophil elastase degradation. J Leukoc Biol. 2004 Mar;75(3):515–522.

124. Zhou Q, Qiu H. The mechanistic impact of N-Glycosylation on stability, pharmacokinetics, and immunogenicity of therapeutic proteins. J Pharm Sci. 2019 Apr;108(4):1366–1377.

125. Raju TS, Scallon BJ. Glycosylation in the Fc domain of IgG increases resistance to proteolytic cleavage by papain. Biochem Biophys Res

Commun. 2006 Mar 17;341(3):797–803.

126. Raju TS, Scallon B. Fc glycans terminated with N-acetylglucosamine residues increase antibody resistance to papain. Biotechnol Prog.

2007 Jul-Aug;23(4):964–971.

127. Mishra P, Nayak B, Dey RK. PEGylation in anti-cancer therapy: an overview. Asian J Pharm Sci. 2016 Jun 01;11(3):337–348.

128. Zhang C, Desai R, Perez-Luna V, et al. PEGylation of lysine residues improves the proteolytic stability of fibronectin while retaining biological activity. Biotechnol J. 2014 Aug;9(8):1033–1043. .

EXPERT OPINION ON DRUG DISCOVERY 15 129. Belen LH, Rangel-Yagui CO, Beltran Lissabet JF, et al. From synth­ esis to characterization of site-selective PEGylated proteins. Front

Pharmacol. 2019;10:1450.

130. Zhang F, Liu MR, Wan HT. Discussion about several potential draw­ backs of PEGylated therapeutic proteins. Biol Pharm Bull. 2014;37 (3):335–339.

131. Shiraishi K, Yokoyama M. Toxicity and immunogenicity concerns related to PEGylated-micelle carrier systems: a review. Sci Technol

Adv Mater. 2019;20(1):324–336.

132. Knudsen LB. Liraglutide: the therapeutic promise from animal models. Int J Clin Pract Suppl. 2010 Oct;64(167):4–11.

133. Knudsen LB, Lau J. The discovery and development of liraglutide and semaglutide. Front Endocrinol (Lausanne). 2019;10:155.

134. Li Y, Wang Y, Wei Q, et al. Variant fatty acid-like molecules

Conjugation, novel approaches for extending the stability of ther­ apeutic peptides. Sci Rep. 2015 Dec;11(5):18039.

135. Boutureira O, Bernardes GJ. Advances in chemical protein modification. Chem Rev. 2015Mar11;115(5):2174–2195.

•• Description of chemical modification of proteins.

136. Basle E, Joubert N, Pucheault M. Protein chemical modification on endogenous amino acids. Chem Biol. 2010Mar26;17(3):213–227.

•• Description of chemical modification of proteins.

137. Chandrashekar C, Okamoto R, Izumi M, et al. Chemical modification of the N termini of unprotected peptides for semisynthesis of modified proteins by utilizing a hydrophilic protecting group.

Chemistry. 2019 Aug 1;25(43):10197–10203.

•• Description of chemical modification of proteins

138. Selo I, Negroni L, Creminon C, et al. Preferential labeling of alpha-amino N-terminal groups in peptides by biotin: application to the detection of specific anti-peptide antibodies by enzyme immunoassays.

J Immunol Methods.

1996 Dec 15;199 (2):127–138.

139. Glukhov E, Stark M, Burrows LL, et al. Basis for selectivity of cationic antimicrobial peptides for bacterial versus mammalian membranes.

J Biol Chem. 2005 Oct 7;280(40):33960–33967.

140. Foerg C, Weller KM, Rechsteiner H, et al. Metabolic cleavage and translocation efficiency of selected cell penetrating peptides: a comparative study with epithelial cell cultures. AAPS J. 2008

Jun;10(2):349–59.141.

141. Di L. Strategic approaches to optimizing peptide ADME properties.

AAPS J. 2015 Jan;17(1):134–143.

•• Description of N-terminal protection 16 X. LAI ET AL.