Polymer- and Lipid-Based Systems for Parenteral Drug Delivery
4.2 The Chemistry of PEGylation
A third late-stage PEGylated therapeutic is PEGylated Interferon β1a (PEG-IFN- β1a; BIIB017) being developed by Biogen Idec for multiple sclerosis (MS) (Baker et al. 2010 ). This chronic autoimmune disease affects the central nervous system and patients suffer progressive neurological disability. Although there are several products approved for treatment of MS, a signifi cant number of patients choose not to initiate treatment due to perceived side effects, perceived lack of effi -cacy or avoidance of too frequent injections (1–3 times per week depending on therapy). A PEGylated version of IFN-β1a is being developed to meet the needs of this patient population. Preclinical studies showed improved PK and effi cacy. In fact, in a mouse model, a single dose of PEG-IFN-β1a was more effi cacious than nine daily doses of the unmodifi ed IFN-β1a. Two Phase I studies supported the further development of PEG-IFN-β1a and this conjugate is currently in Phase III clinical trials for subcutaneous (sc) dosing every 2 or 4 weeks.
These advanced clinical candidates as well as those in early clinical trials show great potential and appear to provide an advantage over current medicines. The suc-cess of these conjugates shows the value that PEGylation has in enabling bioactive molecules to become medicines.
Table 4.4 Linker chemistries used for commercial and clinical PEGylated biotherapeutics Biotherapeutic MW of PEG Linker chemistry Site of attachment Commercial products Adagen (Pegadamase) Linear 5 kDa NHS ester 11–17 random sites: Lys, Ser, Tyr, His Oncaspar (Pegaspargase) Linear 5 kDa NHS ester Nonspecifi c Lys, Ser, Tyr, His Peg-Intron (Peg-IFN α2b) Linear 12 kDa NHS carbonate ester His 34 (major) Pegasys (Peg-IFN α2a) 2-Branch 40 kDa NHS ester Lys 31, 121, 131 or 134 (Bailon et al. 2001 ) Neulasta (Pegfi lgrastim) Linear 20 kDa Aldehyde reductive amination N-terminal Met Somavert (Pegvisomant) Linear 5 kDa NHS ester 4–6 random sites: Lys 38, 41, 70, 115, 120, 140, 145, 158 or N-terminal Phe Macugen (Pegaptanib) 2-Branch 40 kDa (Lys branching) Active ester coupling to pentylamino linker Pentylamino linker on 5′-phosphodiester terminus on a 28- oligonucleotide aptamer Mircera (Peg-EPO) Linear 30 kDa NHS ester Lys CIMZIA (Certolizumab pegol) 2-Branch 40 kDa Maleimide C-terminal Cys Krystexxa (Pegloticase) Linear 10 kDa p -Nitrophenyl-carbonate ester 10–11 random sites, Lys Clinical products Peginesatide (Hematide) Branched 40 kDa Possibly p -nitrophenyl- carbonate ester Dipeptide amino linker (MacDougall 2008 ) ADI-PEG 20 Linear 20 kDa Succinimidyl succinate 11 random sites, Lys (Holtsberg et al. 2002 ) PEG-IFN beta 1a (BIIB017) Linear 20 kDa Aldehyde reductive amination N-terminus (Baker et al. 2010 ) PEG-Hemoglobin (Hemospan) Linear 5.5 kDa 2-Iminothiolane and maleimide ~7–8 random sites: Cys 93 and thiolated (alpha chain) Lys 7, 16, 40 or (beta) Lys 8, 17, 59, 66, 95, 132 (Vandegriff et al. 2008 ) PEG-SN38 (EZN-2208) 4-Branch 40 kDa Active ester coupling to amine linker on SN38 4-Branch to four SN38 molecules (Zhao et al. 2008 ) PEG-hGH (ACP-001) Unknown TransCon Releasable Linker Unknown PEG-hGH (ARX-201) Linear 30 kDa PEG-aminooxy and p -acetyl- phenylalanine Y35pAcP mutation (Cho et al. 2011 ) PEG-Naloxol (NKTR-118) Unknown Unknown Unknown (continued)
Biotherapeutic MW of PEG Linker chemistry Site of attachment PEG-Irinotecan (NKTR-102) Unknown Unknown Unknown (Nektar website) PEG-Interferon lambda (PEG-rIL- 29)(BMS-914143) Linear 20 kDa Unknown Unknown PEG-PAL Linear 5 kDa NHS ester Random sites: Lys (Sarkissian et al. 2008 , 2011 ) PEG-rFIX (N9-GP) 40 kDa GlycoPEGylation N -glycans PEG-rFVIII Unknown Unknown Unknown PEG-BDD-rFVIII (BAY 94-9027) Single large branched Unknown Specifi c amino acid (Ivens et al. 2010 ) PEG-Glutaminase (GlutaDON) Preferred 1–10 kDa Unknown Preferred 1–5 sites (Bausch et al. 2007 )
Table 4.4 (continued)
4.2.1 Conjugation via Amino Groups
The most common linker chemistry used by the commercial PEG products has been an amino-reactive N -hydroxysuccinimidyl (NHS) ester (Fig. 4.4 ) group that reacts with lysines and N-terminal amino groups but can also react to a lesser extent to other nucleophilic side chain groups (serines, tyrosines, histidines, etc.). When this active ester reacts with an amino group, the resulting linkage is a stable amide bond.
Lysine tends to be the most abundant and accessible amino acid in proteins and in reactions with NHS-activated PEG reagents, a variety of lysines are possible sites of attachment. For example, in PEG-Interferon α2a (IFN-α2a, PEGASYS), four pos-sible lysines (Lys 31, 121, 131, or 134) are known to be the site of PEG attachment using a branched 40 kDa NHS-activated PEG (Bailon et al. 2001 ). In Mircera, a 30 kDa NHS-activated PEG binds to either Lys 52 or Lys 46. When smaller size PEG is used, it is possible to attach multiple PEG polymers on the same parent protein. For example, Adagen is composed of multiple PEG polymers of 5 kDa molecular weight (Alconcel 2011 , Booth 2009 ).
Another amino-reactive linker that has been used commercially in a PEG- conjugate is the p -nitrophenyl carbonate ester (Fig. 4.4 ). This functional group reacts with accessible lysines on the surface of the protein and forms urethane bonds.
PEG N
O
O C O O
Biologic H2N
PEG C
O
Biologic HN
PEG C
O
O NO2
Biologic H2N
PEG C
O
Biologic HN
PEG O
Biologic H2N
PEG N Biologic
Reducing Agent NaCNBH3
PEG HN Biologic
Schiff Base Intermediate
Reduction to Secondary amine N-Hydroxysuccinimidyl Ester (NHS)-PEG
Aldehyde-PEG para-Nitrophenyl ester (PNP)-PEG
Fig. 4.4 N -Hydroxysuccinimidyl ester (NHS), p -nitrophenyl ester (PNP), and reductive amina-tion (aldehyde-PEG) chemistries for PEGylaamina-tion to an amine-containing biologic. In the case of reductive amination, the main target is the amino-terminus of a protein or peptide whereas the NHS and PNP reagents will typically react with available Lys
In the case of Krystexxa (pegloticase aka peg-uricase), approximately 9 strands of 10 kDa PEG were attached to each uric acid tetrameric enzyme via the 12 accessible lysines (Sherman et al. 2008 ). Interestingly, in developmental studies on the uricase enzyme, a conjugate containing 6 strands of 10 kDa PEG per subunit were found to provide a signifi cantly longer half-life in mice than conjugates with the same total mass of PEG (60 kDa per subunit) but fewer strands of PEG (i.e., 3 strands of 20 kDa or 2 strands of 30 kDa PEG).
Finally, the last amino-reactive PEG conjugation chemistry used commercially is the reductive amination on the N-terminal amino acid. One example of this chemis-try is Neulasta (pegfi lgrastim) where an aldehyde-functionalized 20 kDa PEG reacts under slightly acidic conditions (pH 5) to react more specifi cally at the α-amino group of the N-terminal methionine residue of GCSF (fi lgrastim) (Piedmonte and Treuheit 2008 ; Kinstler et al. 1996 ; Molineux 2004 ). The specifi c attachment at the N-terminus is afforded by the lower p K a of this α-amino group (p K a ~7.6–8.0) rela-tive to ε-amino groups of lysines (p K a ~10.0–10.2) (Wong 1991 ). PEG-aldehyde reacts with the α-amine and readily forms an imine bond which is subsequently reduced with sodium cyanoborohydride (or other selective reducing agent) to form a very stable secondary amine (Fig. 4.4 ).
4.2.2 Conjugation via Thiol Groups
More recently, there have been efforts toward more specifi c attachment of PEG to provide less heterogeneity, decrease interference with enzyme active sites or protein binding sites and increase reproducibility of manufacturing. One route commonly used is to design into the protein a cysteine mutation in the amino acid sequence at a site distal from that involved in its biological activity. Cysteine is typically chosen for this because its thiol moiety can be much more reactive than amino groups depending on reaction conditions (pH, linker, etc.) affording some measure of spec-ifi city. This specspec-ifi city is increased by the fact that cysteine residues are seldom found in protein sequences, and if they are present, many times they are oxidized and involved in a disulfi de bridge and are less accessible and less reactive.
There are several thiol linker reagents that are reactive and somewhat more selec-tive toward cysteine residues, such as maleimides, haloacetamides, haloalkyls, vinylsulfones, and disulfi de reagents (Hermanson 2008 ). These reagents form stable thioether bonds after reaction with cysteine, although there are reports of retro- Michael addition reactions occurring for the maleimide reagents (Baldwin and Kuck 2011 ). The disulfi de reagents undergo interchange reactions with the free thiol of the cysteine resulting in a mixed disulfi de product bound to the biotherapeu-tic. This mixed disulfi de bond is susceptible to further thiol interchange or to cleav-age via reducing cleav-agents or reducing conditions in vivo. These disulfi de recleav-agents may have some role as a releasable linker for certain conjugates.
There are a number of clinical conjugates as examples for thiol PEG conjugation but the one commercial product is CIMZIA (Veronese and Mero 2008 ). This human-ized anti-tumor necrosis factor (TNF)-α antibody fragment (Fab′) has an accessible
cysteine in the hinge region that is conjugated to a 40 kDa branched PEG maleimide reagent. Since the reacting thiol resides in the hinge region away from the antigen- binding pocket, there is essentially no loss in biological activity for the PEGylated antibody fragment compared to the native antibody.
4.2.3 Other Conjugation Types
The number of PEG reagents and attachment chemistries has increased over the past decade. Reagents are now readily available with PEG chains of molecular weights ranging up to 80 kDa and confi gured in various geometries from linear to multi-arm branched and pendants. Reagents range from PEGs with a single reactive group to PEGs containing homo- and hetero-multifunctionalities (Monfardini et al. 1995 ; Jevševar et al. 2010 ). Monodispersed PEGs with discrete MWs up to 4 kDa are also available (Quanta Biodesigns).
A number of alternate approaches to amine and thiol chemistries have been devel-oped in order to improve attachment site selectivity. Directing PEG to a single site or location, regiospecifi city, is likely to increase process yields, minimize the complexity of product purifi cation and characterization, and potentially reduce PEG interference with bioactivity (Kinstler et al. 1996 , 2002 ; Chapman 2002 ; Cox et al. 2007 ; Cazalis et al. 2004 ; Finn 2009 ; Buckley et al. 2008 and Greenwald et al. 2003b ).
Groups have reported success in this area through the development of new PEG reagent functionalities. One approach (PolyTherics Ltd.) utilizes mono- and bis-sulfone- activated PEGs designed for conjugation to accessible disulfi de bridges.
Following reduction of disulfi de thiols, these reagents undergo bis-alkylation with the two disulfi de sulfur atoms to form a stable three carbon bridge in the location of the original disulfi de to which the PEG is attached. An advantage of this technology is that it allows for the selectivity of other cysteine reagents when PEGylating bio-therapeutics that do not possess free thiols and maintains the structural integrity of the original disulfi de bridge (Balan et al. 2007 ; Brocchini et al. 2008 ).
4.2.3.1 Chemical Modifi cations
Many groups have examined chemical modifi cations, such as periodate oxidation of oligosaccharides and amino acid side chains, as a means to generate new reactive sites for PEGylation (Zalipsky 1995 ; Wolfe and Hage 1995 ; Wilchek and Bayer 1987 ; Dorwald 2007 ); others have reported site-selectivity through reversible block-ing of possible side reaction sites (Tsunoda et al. 2001 ).
4.2.3.2 Enzymatic Approaches
Site-selective PEGylation of hGH was achieved through the enzymatic insertion of reactive handles via Carboxypeptidase Y (CPY) transpeptidation. Ketone or azide moieties were separately incorporated onto the hGH C-terminus for PEG coupling
using either oxime ligation or copper (I) catalyzed (2 + 3) cycloaddition reactions (Peschke et al. 2007 ). The term “GlycoPEGylation” refers to a novel PEGylation in which specifi c glycosyltransferases are used for conjugations of sialic acid- modifi ed PEG reagents to serine and threonine residues and other potential O-glycosylation sites. This technology has been demonstrated in several clinically relevant biothera-peutics such as GCSF, GMCSF, interferon α2b, Factor VIIa, and Factor IX (DeFrees et al. 2006 ; DeFrees et al. 2007 ; Klausen et al. 2008 ). Another enzymatic approach utilizes transglutaminases (TG-ases) for PEGylation targeting glutamine residues.
TG-ases catalyze reactions where the γ-carboxamido group on glutamine acts as an acyl donor which can react with amine PEG (PEG-NH 2 ) to form an amide linkage.
Studies with several therapeutic proteins have demonstrated that these PEGylation reactions can be strikingly site-specifi c with PEGylation at only one or two specifi c glutamine residues in many cases (Sato 2002 ; Maullu et al. 2009 ). Analysis of the TG-ase reaction suggests that this selectivity results from enhanced enzymatic attack at fl exible backbone regions (Fontana et al. 2008 ).
4.2.3.3 Incorporation of Non-native Amino Acids
In order to gain better control of site-selectivity, efforts have been made to introduce custom amino acids with orthogonally reactive functional side chains. This would allow for exquisite site-directed bioconjugation with PEG reagents that are nonreac-tive toward nanonreac-tive amino acids but specifi c for the custom amino acid side chains.
One example is the addition of an azido-functionalized methionine analog to the media of an E . coli expression system in order to increase the incorporation of this non-native amino acid into the expressed protein (Cazalis et al. 2004 ). The azido- functional group was used for subsequent site-directed PEGylation using Staudinger ligation with a triarylphosphine-PEG reagent. Previous attempts to make the C-terminal cysteine mutant were unsuccessful while the use of this non-native amino acid method led to the successful generation of a bioactive site-specifi c PEG-conjugate.
In a second approach, Peter Schultz and coworkers have developed a novel tech-nology for genetically incorporating non-native amino acids directly into exoge-nously expressed proteins (Deiters and Schultz 2005 ; Wang et al. 2001 , 2003 ). The group is able to alter the cell’s translational workings by incorporating a new t RNA/ t RNA synthetase pair specifi c for the desired non-native amino acid. In this manner, non-native amino acids with functional groups for site-selective modifi ca-tion may be introduced into a protein sequence. Specifi cally, a keto-amino acid, p -acetyl-phenylalanine, can be incorporated into proteins as a chemical handle for the specifi c linkage of proprietary PEG reagents. Ambrx has used this technology to generate a number of PEGylated proteins. One molecule (ARX201), a PEGylated hGH, was shown to induce weight gains for a single weekly dose similar to that of daily doses of hGH in the hypophysectomized rat model (Cho et al. 2011 ). The molecule is currently in Phase II clinical trials.
4.2.3.4 Releasable PEGylation
Other groups have focused on generating PEG reagents with labile linkers for controlled release of the native biotherapeutic into the circulation or to targeted locations in the body. Releasable or reversible PEGylation has been investigated as a means to circumvent PEG-related protein inactivation through slow release of the PEG moiety to generate the active protein in a pro-drug manner. These reagents utilize labile sites in the linkers such as esters or other hydrolyzable groups. One example of this is a PEG benzyl elimination (BE) linker system consisting of a hydrolyzable ester trigger which initiates either a 1,4- or 1,6-benzyl elimination reaction for releasing the native molecule (Greenwald et al. 2003a ; Zhao et al. 2006 , 2008 ; Filpula and Zhao 2008 ). Ascendis Pharma has developed proprietary auto- hydrolyzing linkers (Transcon Technology) with rate of release controlled by pH and temperature ( www.ascendispharma.com ).
PEG reagents with linkers employing a labile 2-sulfo-9-fl uorenyl-methoxycar-bonyl (FMS) group have been demonstrated to slowly release several proteins and peptides including Interferon α2, hGH, Exendin-4, PYY, and others (Tsubery et al.
2004 ; Shechter et al. 2005 ). One such PEG reagent was demonstrated to slowly release Interferon α2 following subcutaneous administration in rats with active levels peaking at 50 h, with substantial levels still being detected 200 h after admin-istration (Peleg-Shulman et al. 2004 ).
Another example of reversible PEGylation uses a hydrolyzable β-alanine linkage for slow release. A slow release PEG-hGH conjugate developed with this reagent demonstrated a similar growth response in rats with a single dose compared to daily dosing over a week (Pasut et al. 2008 ).