Gamma immunoglobulins are the most represented type of antibodies in the plasma. Three out of four subclasses have a half-life of around 21 days in humans thanks to their size and FcRn-mediated recycling.138 The IgG sites responsible for the binding
to the receptor are the CH2 and CH3 domains of the 50 kDa Fc region. As consequence, several Fc fusion proteins have been developed since 1980s for half-life extension using this region as platform for the incorporation of proteins and peptides.117 Several studies
demonostrated that the biological activity of the Fc fusion proteins can be modulated i) dimerizing the sequence or the Fc region for avidity effects, ii) inserting specific linker sequences, or iii) changing the orientation of the fused molecules.139 The first Fc-fusion
protein to enter in the market was etanercept (Embrel, Amgen), used for the treatment of rheumatoid arthritis. Etanercept is composed of the extracellular region of the TNF receptor 2 fused to a Fc domain. This molecule binds with high affinity the natural ligand, TNF-D, blocking the activation of its receptor in chronic inflammatory diseases.117
In the following decades, at least ten other Fc-fusion proteins were approved by the FDA, and many others are in clinical trials, showing half-lives ranging from five to 13 days.139 Among them, only two are peptide-Fc proteins (“peptibodies”) (Fig. 19): romip-
lostim (Nplate, Amgen) and dulaglutide (Trulicity, Eli Lilly), approved for the treatment of immune chronic thrombocytopenia and type 2 diabetes, respectively.93,117 Romiplostim
is composed of two tandem copies of a linear thrombopoietin mimetic peptide recombi- nantly fused to the C-terminus of an IgG1 Fc fragment using a Gly linker. This peptide was initially developed and further optimized by phage display selection, and it showed an IC50 value of 500 pM when tested as dimer.140,141 Dulaglutide is an optimized analogue
of GLP-1 expressed to the N-terminus an IgG4 Fc region. The peptide has been engi- neered to improve the stability against DPP-4. Further modifications were introduced on the Fc region to reduce its affinity for the FcRn and module the effector properties.142
Both peptibodies have half-lives of approximately three to five days when s.c. injected in human, allowing a single administration per week.
Figure 19. Comparison of the structure of (left) IgG antibodies, (center) a small therapeutic peptide, and
(right) a peptibody. Reprinted (open access) from 139,copyright © 2013 Wiley Periodicals, Inc. and the Amer- ican Pharmacists Association.
Other peptibodies have reached different preclinical and clinical stages with op- posing results.139 The most advanced is blisibimod (AMG-623), a B cell-activating factor
(BAFF) antagonist in phase 3 trial for the treatment of systemic lupus erymatosus.143 It
is composed of two equally active parts, each carrying four BAFF-binding peptides, fused to an IgG1 Fc fragment. The BAFF-binding peptide was selected by phage display and cross reacts with both soluble and cell surface BAFF with extremely high affinity (1 pM), interrupting its interaction with the three natural receptors.144 The BAFF-peptibody has a
half-life in human of eight to ten days, shorter than the other anti-BAFF antibodies, but in contrast it can be efficiently expressed in bacteria.145
Another peptibody is trebananib (AMG-386), in which the active peptide was iso- lated by phage display as inhibitor of angiopoietin-1 and -2, and it blocks their interactions with the natural receptor Tie2.146,147 As romiplostim, trebananib presents two tandem
copies of the peptide fused to an IgG1 Fc fragment and showed a half-life of three to six days in human.148 Because of its potent anticancer activity it reached phase 3 trial. How-
ever, these studies were discontinued in 2014 from Amgen without further information.149
Two other promising peptibodies were developed and entered in clinical trials, but no data has recently been reported. The first product was a GLP-1 analogue, named CNTO-736, with a significantly prolonged half-life and in vivo efficacy in mice.150,151 The
second peptibody was CNTO-528, an erythropoietin mimetic peptide discovered from phage peptide libraries.152 The peptide was again fused to an IgG1 Fc fragment as either
a single or a double copy (CNTO-528 and CNTO-530, respectively), and displayed a much better activity and half-life than erythropoietin.153,154 In a phase 1 trial, CNTO-528
showed a dose-dependent half-life ranging from 1.5 to 7.5 days.153 Preclinical studies
with CNTO-530 gave similar results in mice.154
In our laboratory, Angelini et al. developed a peptibody composed of the bicyclic peptide UK18 fused to a Fc fragment in the linear format and the cyclized in situ, skipping the individual steps of peptide production, cyclization, purification, activation and chemi- cal conjugation.85 They initially fused UK18 on both Fc monomers, replacing the hinge
region with a short Gly-Ser linker, and they cyclized the peptide with TBMB (Fig. 20a). However, they experienced the formation of different dimers due to a cross-reaction be- tween the two displayed peptides, and this resulted in a 60-fold reduction in the inhibitory activity. Thus, they extended the short Gly-Ser linker testing hinge regions of different lengths and they expressed UK18 only on one of the Fc monomers (Fig. 20b). This new format proved that the inhibition potency was strictly related to the linker size since longer hinge regions had Ki values in the same order of magnitude as the peptide alone.85 The
half-life of the best heterodimeric peptibody was tested in mice upon intravenous injec- tion and resulted to be approximately 1.5 days, 72-fold longer than the bicyclic peptide alone.71,85 The ELISA assay used for the quantification demonstrated that the peptide
was fully stable. These results proved that Fc fusion of peptides that need to be chemi- cally cyclized can be efficiently achieved if appropriate optimizations are introduced in the system.85
Figure 20. Chemical macrocyclization of UK18-Fc fusion protein. (a) Schematic representation of UK18-Fc
homodimer and cyclization reaction of the peptide with TBMB. Two of the nine possible regioisomers are shown, including the desired one (lower drawing). (b) Schematic representation of UK18-Fc heterodimer and cyclization reaction based of the peptide with TBMB. Only the desired isomer can be obtained. Linkers with different lengths were inserted between the peptide and the Fc fragment, using the amino acid sequence of the human IgG1 hinge region wherein Cys were replaced by Ser. Adapted with permission from 85, copy- right © 2012 American Chemical Society.
In addition to recombinant expression, peptibodies can be created through a chem- ical conjugation of peptides to both Lys and Cys residues present or recombinantly added in the hinge part of the Fc region.139 Mezo et al. demonstrated the feasibility of
this approach with the atrial natriuretic peptide (ANP). ANP is a natural peptide hormone involved in the control of pressure and fluid in the body, characterized by a short half- life. The peptide was modified with terminal thioester moieties and coupled by native
chemical ligation to an IgG1 Fc fragment bearing two free N-terminal Cys residues (Fig. 21). Similar to other peptibodies previously reviewed, the variation of linker length be- tween peptide and Fc region resulted in molecules with altered in vitro activity. The de- veloped monomeric and dimeric conjugates improved the half-life of ANP more than 30- fold in rats. Nevertheless, the degradation of the linear peptide was much faster than the effective half-life of the Fc fragment to which it was conjugated, confirming the superior stability of cyclic peptides over linear ones.155
Figure 21. Schematic representation of the semisynthetic method used to generate the ANP-Fc fusion pro-
teins. Reprinted with permission from 155, copyright © 2012 American Chemical Society.
Despite the high success, the Fc fusion protein strategy still has several limitations which makes it less attractive117,138, such as:
i) difficulties in recombinantly introducing multiple unnatural amino acids in pep- tide active moieties or for site specific conjugation;
ii) low protein yield due to the complex production and purification steps, further reduced with homo or bispecific dimers, as shown in one of the reported ex- amples (few milligrams of pure heterodimer product before cyclization)85;
iii) low tissue penetration for targeting solid tumors due to a relatively large size; iv) high cost of production and issues in storage and/or handling;
v) recombinant engineering often required to remove unwanted ancillary activa- tions of the immune system when targeting soluble proteins, or to restore the proper sites of glycosylation.