Lipidation and PEGylation Strategies to Prolong the in Vivo Half-Life of a Nanomolar EphA4 Receptor Antagonist

Abstract

The EphA4 receptor tyrosine kinase plays a role in neurodegenerative diseases, inhibition of nerve regeneration, cancer progression and other diseases. Therefore, EphA4 inhibition has potential therapeutic value. Selective EphA4 kinase inhibitors are not available, but we identified peptide antagonists that inhibit ephrin ligand binding to EphA4 with high specificity. One of these peptides is the cyclic APY-d3 (βAPYCVYRβASWSC-NH₂), which inhibits ephrin-A5 ligand binding to EphA4 with low nanomolar binding affinity and is highly protease resistant. Here we describe modifications of APY-d3 that yield two different key derivatives with greatly increased half-lives in the mouse circulation, the lipidated APY-d3-laur8 and the PEGylated APY-d3-PEG4. These two derivatives inhibit ligand induced EphA4 activation in cells with sub-micromolar potency. Since they retain high potency and specificity for EphA4, lipidated and PEGylated APY-d3 derivatives represent new tools for discriminating EphA4 activities in vivo and for preclinical testing of EphA4 inhibition in animal disease models.

Graphical Abstract

INTRODUCTION

EphA4 is a receptor tyrosine kinase of the Eph family that is highly expressed in the developing and adult nervous system, in the immune system, and in some tumors [1-7]. Furthermore, EphA4 is increasingly recognized as a potential therapeutic candidate against multiple diseases, including Alzheimer’s disease and other neurological disorders, spinal cord injury, traumatic brain injury and stroke, depression, osteoarthritis and cancer [8-25]. However, EphA4-specific inhibitors suitable for testing in clinical trials are not yet available.

The domain structure of EphA4 includes an extracellular region containing the ligand-binding domain (LBD), a transmembrane helix, and an intracellular region containing the kinase domain (Fig. 1A, left). EphA4 tyrosine kinase activity is increased upon interaction with its ligands, which include both the GPI-linked ephrin-A and the transmembrane ephrin-B ligands [26]. This interaction typically occurs at sites of cell-cell contact and also promotes signaling by the ephrin (known as reverse signaling; Fig. 1A, left), resulting in bidirectional signals that affect both EphA4 and ephrin expressing cells [27]. Examples of EphA4 downstream pathways affecting neuronal synaptic connections, and thus possibly involved in neurodegeneration, include regulation of integrin signaling, Rho family GTPases, Abl and Src family kinases, and amyloid precursor protein (APP) processing [10, 24, 25, 28-32]. EphA4-mediated inhibition of neural repair processes could also exacerbate neurodegeneration [32-38]. In addition, inhibiting interaction of EphA4 in neuronal dendritic spines with ephrin-A3 in perisynaptic astrocytic processes can enhance removal of extracellular glutamate by glutamate transporters, thus potentially limiting glutamate toxicity associated with neurodegeneration and stroke [3, 39-41].

Figure 1. Arg7 and the C terminus of APY derivatives are potential sites for derivatization.

(A) Left: Ephrin ligand binding induces EphA4 oligomerization (not shown), tyrosine phosphorylation and activation leading to physiological signals regulating developmental processes and adult tissue homeostasis. Ephrin ligands can also signal (known as reverse signaling). Dysregulated signaling due to increased EphA4 or ephrin expression or other factors can lead to disease processes, including neurodegeneration, cancer progression and others. Middle: Peptide antagonists targeting the ephrin-binding pocket inhibit both EphA4 and ephrin signaling. Right: Kinase inhibitors inhibit EphA4 kinase activity and downstream signaling. LBD, ligand-binding domain; EGF domain, epidermal growth factor-like domain; FNIII domain, fibronectin type III domain; SAM domain, sterile alpha motif domain. (B) A model of the structure of peptide 2 (orange) in complex with the EphA4 LBD (grey) shows that the side chain of Arg7 and the C-terminal GGKG linker (light orange) are exposed to the solvent and that the GGKG linker at the peptide C-terminus is not expected to sterically hinder binding to EphA4. Arrows point to Arg7 and the C-terminus of the APY-d2 core. Modeling was done using the structure of APY-d2 in complex with the EphA4 LBD [9] (PDB 4W4Z). (C) Model of peptide 2 (orange) in complex with the EphA4 LBD (grey) highlighting the key EphA4 residues that interact with the peptide. Polar interactions between peptide 2 and EphA4 are shown as green dashed lines. The EphA4 DE, GH and JK loops are also indicated. (D, E, F) ELISAs showing inhibition of the binding of ephrin-A5 fused to alkaline phosphatase (ephrin-A5 AP) to the immobilized EphA4 extracellular domain fused to the Fc portion of an antibody (EphA4 Fc) by different peptide concentrations. Peptide 1 was derived by replacing Arg7 of APY-d2 with Lys (D), peptide 3 by replacing Arg7 of APY-d3 with Lys (E), and peptide 2 by adding the GGKG sequence to the C-terminus of APY-d2 (F). The curves for peptides 1-3 are compared to the curves for APY-d2 (dark blue) or APY-d3 (gray) obtained in parallel in the same ELISA experiments (which are a subset of the experiments used to calculate the overall average IC₅₀ values shown in Table 1). The graphs show averages ± SE of data from n=3 independent experiments, each including triplicate measurements. Values are normalized to no peptide treatment (plotted at 0.1 nM). Average IC₅₀ values ± SE (nM) were calculated from multiparameter curve fitting of the combined data. The significance for the comparison of the best fit IC₅₀ values for the two curves shown in the same panel was determined using the extra sum-of-squares F test. ****, P<0.0001 ; the IC₅₀ values for peptide 2 and APY-d2 in (F) are not significantly different.

A variety of agents have been used to modulate EphA4 signaling in vitro and in preclinical mouse disease models, including peptides (Fig. 1A, middle) [10, 11, 42-45], nanobodies [46], ephrin-A4 Fc [21] and chemical compounds [43, 47-50] targeting the EphA4 LBD, kinase inhibitors (which typically target the ATP-binding pocket in the kinase domain; Fig. 1A, right) [51, 52] and the EphA4 extracellular region fused to Fc (which can bind ephrins and interfere with their ability to activate EphA4 and other Eph receptors) [53]. Each strategy has different advantages and disadvantages relating to the mechanism of action (Fig. 1A), potency, specificity, stability, toxicity and other properties of each particular inhibitor.

Peptides and peptidomimetics can bind to the broad and shallow ephrin-binding pocket in the extracellular EphA4 LBD with high affinity and specificity, thus antagonizing ephrin binding [9, 37, 44, 49] (Fig. 1A middle). We previously identified using phage display the cyclic peptide APY (APYCVYRGSWSC), which targets the ephrin-binding pocket of EphA4 with low micromolar affinity [42]. Replacement of Gly8 with βAla and amidation of the C-terminus of APY yielded the 50-fold more potent APY-d2 (APYCVYRβASWSC-NH₂) [9]. Replacement of Ala1 in APY-d2 with βAla yielded the highly protease resistant APY-d3 (βAPYCVYRβASWSC-NH₂) [37]. APY-d3 inhibits ephrin-A5 ligand binding to EphA4 with low nanomolar binding affinity (K_d=30 nM) and has a half-life of more than 72 hours when incubated in plasma in vitro [37]. However, once in the blood circulation, small peptides such as APY-d3 (molecular weight 1,403 kDa) are typically excreted through the kidneys within a few minutes due to their small size [54, 55]. A number of modifications can increase effective peptide hydrodynamic volume to slow down excretion of therapeutic peptides through the kidneys and increase their in vivo half-life. For example, lipidation can be used to promote non-covalent peptide binding to the abundant serum protein albumin, which has a molecular weight of 67 kDa. Conjugation to polyethylene glycol (PEG) polymers of sizes ≥ 30 kDa, which are non-immunogenic and highly water soluble, also increases the hydrodynamic volume of a peptide [54-62]. These modifications are present in some FDA-approved peptide therapeutics [55, 62, 63]. Here we describe strategies for lipidation and PEGylation of APY-d3 that dramatically improve the in vivo half-life of the peptide while preserving its high inhibitory potency and selectivity. These new APY-d3 derivatives are suitable for characterization of EphA4 activities in vivo and as potential therapeutic leads for EphA4 inhibition.

RESULTS

Arg7 and the C-terminus of APY derivatives can be modified with minimal loss of potency

The N-terminal amino group of peptides is often used for derivatization. However, we previously showed that modifications affecting the N-terminal positive charge of APY derivatives lead to a dramatic loss in their inhibitory potency [37]. The crystal structures of APY derivatives in complex with EphA4 show that Arg7 and the C-terminus of these peptides are not involved in EphA4 binding or in intramolecular peptide interactions [9, 37] (Fig. 1B,C). Thus, we examined the effects of modifications at these sites for peptide derivatization to optimize pharmacokinetic properties without detrimental effects on potency.

We replaced Arg7 in APY-d2 and APY-d3 with Lys (peptides 1 and 3 in Table 1), since the amino group in the Lys side chain can be used for solution-phase derivatization. This modification decreased potency (IC₅₀) by ~2 folds when comparing peptide 1 to APY-d2 and peptide 3 to APY-d3 in ELISA experiments measuring inhibition of ephrin-A5 AP binding to the immobilized EphA4 extracellular region fused to the Fc portion of an antibody (EphA4 Fc; Fig. 1D,E). As an alternative, we added a Lys to the C-terminus of APY-d2 via a Gly-Gly spacer (peptide 2 in Table 1). Modeling of the structure of peptide 2 in complex with the EphA4 LBD suggests that the C-terminal GGKG sequence extends away from the receptor and thus should not interfere with EphA4 binding (Fig. 1B,C). We indeed found that peptide 2 inhibits binding of ephrin-A5 to EphA4 with similar potency as the unmodified APY-d2 (Fig. 1F; Table 1). Thus, APY derivatives with Lys7 or with the GGKG C-terminal sequence should both be suitable for derivatization. Furthermore, C-terminal modifications may have lower effects on inhibitory potency than modifications within the macrocycle.

Effects of lipid modifications on APY-d3 inhibitory potency

One FDA-approved strategy to increase the half-life of a peptide therapeutic in vivo is “lipidation”, which generally involves the conjugation of a C8 to C20 hydrocarbon chain to the peptide [55, 60]. Lipidation can prolong peptide half-life by promoting non-covalent association with serum albumin, which due to its large size (67 kDa) increases the effective hydrodynamic volume of the peptide, reducing the rate of renal filtration. Lipids bind to serum albumin with low affinity, but the high concentration of albumin in the blood (40 mg/ml) ensures a high level of complexation.

We generated a series of lipidated APY-d3 derivatives and examined the effects of various lipidation strategies on inhibition of EphA4-ephrin-A5 interaction in ELISAs (Fig. 2 and Table 1). Conjugation of the C8 lipid octanoic acid to the side chain of Lys7 alleviated the loss in potency caused by the Arg to Lys substitution (compare peptide 4 with peptide 3). Furthermore, the C-terminal GGRG sequence added to peptide 4 to yield peptide 5 may enhance solubility without detrimental effects on potency. Replacement of the octanoic acid in peptide 4 with octanyl-glycine in peptide 6 did not substantially affect potency. APY-d3 was also modified with octanoic acid conjugated to the lysine side chain of an added C-terminal GGKG sequence in peptide 7, which has similar potency as the other peptides containing octanoic acid. The C12 lipid lauric acid, or lauryl-glycine, attached to Lys7 in peptides 8 and 9, respectively, dramatically decreased potency by >10-fold compared to APY-d3 similarly modified with a C8 lipid, possibly due to decreased solubility. Indeed, introduction of a negatively charged linker such as β-aspartic acid (βAsp or βD) or γ-glutamic acid (γGlu or γE) to increase solubility improved potency by several fold (compare peptides 10 and 11 with peptide 8). Addition of lauric acid to the APY-d3 C-terminus in peptide 12 also resulted in substantial loss of potency compared to the corresponding octanoyl peptide 7. However, the βAsp and γGlu linkers in peptides 13 and 14, respectively, greatly improved the potency of the peptides with C-terminal lauric acid. Finally, a longer linker consisting of two γGlu residues (Fig. 3A) resulted in better potency than a single γGlu (compare peptide 14 with peptide 15, re-named APY-d3-laur8, which has a potency comparable to APY-d3).

Figure 2. Inhibitory potency of APY-d3 derivatives that bind to serum albumin.

ELISAs comparing the ability of the peptides to inhibit binding of ephrin-A5 AP to immobilized EphA4 Fc. The graphs show averages ± SE from n independent experiments, each including triplicate measurements. IC₅₀ values ± SE (nM) were calculated from multiparameter curve fitting of the combined data. The curves for the albumin-binding peptides (orange) are compared to the curves for APY-d2 (dark blue) and/or APY-d3 (gray) obtained in parallel in the same ELISA experiments (which are a subset of the ELISA experiments used to calculate the overall average IC₅₀ values shown in Table 1).

Figure 3. Chemical structures of lipidated and PEGylated APY-d3 derivatives.

(A) Lipidated APY-d3-laur8. (B-E) PEGylated peptides. The APY-d3 core sequence is indicated in black, linker sequences are in red, and the lipid lauroyl or the 30 kDa polyethylene glycol (PEG) are in blue. G, glycine; K, lysine; γE, γ-glutamic acid; βA, β-alanine; Ahx, aminohexanoic acid; n indicates multiple repeats of ethylene glycol.

An increase in the size of the C12 lipid in APY-d3-laur8 to the C14 lipid myristoyl acid attached through a γGlu-γGlu linker in peptide 16, further decreased potency by 2-3-fold. Finally, the C16 lipid palmitic acid proved to be very detrimental to peptide potency even with a γGlu-γGlu linker, particularly when attached to Lys7 (peptides 17 and 18). As observed for the corresponding lauroyl peptides, a longer linker consisting of two γGlu residues resulted in a ~3 fold better potency compared to a single γGlu (peptide 19 versus peptide 18), although peptide 19 remains 5-6-fold less potent than APY-d3-laur8. A further increase in the length of the spacer by including an additional Gly-Ser in the C-terminal linker of peptide 19 to obtain peptide 20 did not further improve potency.

To determine whether albumin binding to the lipidated peptides affects their ability to bind EphA4, we examined the effects of delipidated human serum albumin (HSA) on the inhibitory potency of a subset of the peptides. By comparing peptides tested in parallel in the presence and in the absence of 40 mg/ml HSA, we found that HSA has only a very small effect on the potency of the non-lipidated peptides 1, 2 and APY-d3 examined as controls, as expected (Fig. S1A and Table 1). HSA decreased the potency of the octanoyl peptides 4, 5, 6 and 7 by 2-3 fold, suggesting that the binding of serum albumin to these peptides somewhat interferes with EphA4 binding (Fig. S1A and Table 1). HSA strongly decreased the potency of the lauroyl APY-d3-laur8 and the myristoyl peptide 16 (Fig. S1A and Table 1), which may be explained by the higher HSA binding affinity of the longer lipids and steric interference of albumin. An impact of albumin binding on the receptor binding ability of a peptide is a common effect of peptide lipidation [55]. We did not examine the effects of HSA on the palmitoylated peptides, given their low potency and poor solubility (data not shown). Overall, our data suggest that C-terminal lipidation may be better tolerated in terms of inhibitory potency than lipidation on Lys7, that a shorter lipid is better tolerated than a longer lipid, and that βAsp or γGlu linkers can alleviate the decrease in potency caused by longer lipids.

Lipidation can substantially increase the in vivo half-life of APY-d3

To examine the effects of lipid modifications on peptide in vivo half-life, we devised a method to determine the concentration of active (EphA4 inhibitory) peptide in mouse blood by taking advantage of the ELISA measuring inhibition of EphA4-ephrin-A5 interaction. The method involves deducing the concentration of active peptide at various time points after intravenous or intraperitoneal administration by comparing the apparent IC₅₀ value for the peptide remaining in plasma with the IC₅₀ value of a control obtained using mouse plasma spiked in vitro with a known peptide concentration (Fig. S2 and Methods).

Using this method, we determined the in vivo half-life of representative lipidated peptides (Fig. 4A,B and Table 1). For peptide 5 modified on Lys7 with octanoic acid (the shortest lipid used), only 5% of the peptide remained in the blood 10 min after intravenous injection of the peptide (which was dissolved in PBS supplemented with 5% DMSO). Since peptides modified with longer lipids had poor solubility in PBS containing 5% DMSO, we used a formulation including 40 mg/ml bovine serum albumin (BSA). For peptides modified at the C-terminus with lauric acid, we measured half-lives of 30-40 minutes for peptides 12 and 14 injected intravenously and 40-50 minutes for peptides 11, 14 and APY-d3-laur8 injected intraperitoneally. Myristoyl acid in peptide 16 prolongs to ~100 minutes the in vivo half-life after intraperitoneal injection. Formulation of peptide 16 with methyl-β-cyclodextrin, a biocompatible cyclic oligosaccharide used to improve the solubility and bioavailability of hydrophobic drugs [64], yielded similar in vivo half-life as the formulation with serum albumin (not shown). Finally, peptide 18 with palmitic acid had a half-life of ~50 minutes after intravenous injection and ~100 minutes after intraperitoneal injection and peptide 19, also with palmitic acid, had a half-life of ~70 min after intraperitoneal injection.

Figure 4. Half-life in the blood circulation of APY-d3 derivatives.

(A) Lipidated peptides administered intravenously (green). (B) Lipidated peptides or serum albumin-binding peptide 21 administered intraperitoneally (black). (C) PEGylated peptides administered intraperitoneally. The data points show the percentage of injected peptide remaining in the mouse blood at different times (minutes in A,B or hours in C) after administration. The 10-hour time point is highlighted in C with a red circle. The estimated concentration of the injected peptide at time 0 was calculated based on a blood volume of 2.5 ml and taken as 100% (empty circle in graphs for intraperitoneally injected mice) (see also Fig. S2). Approximate peptide half-lives were calculated using a one phase decay equation. The theoretical 100% concentration at time 0 min was included in the fit for the intraperitoneally injected peptides but not for the intravenously injected peptides, for which there appeared to be a large loss of peptide in the first few minutes after injection. The graphs show averages ± SE from n=2-7 mice. The data for APY-d3-laur8 and APY-d3-PEG4 are shown in both a linear (left) and a logarithmic (right) scale, with the right panel for each also showing the peptide concentrations in the mouse blood at different times after injection of 170 μg APY-d3-laur8 and 0.8 mg APY-d3-PEG4.

Overall, APY-d3-laur8 seems to be the most promising of the lipidated peptides, given its high potency and considerable in vivo half-life. We confirmed that APY-d3-laur8, even at high concentrations, selectively inhibits only EphA4 among the Eph receptors (Fig. 5A). In addition, APY-d3-laur8 inhibits EphA4 tyrosine phosphorylation (indicative of activation; Fig. 1A left) in HEK293 cells expressing EphA4 and stimulated with the ephrin-A5 ligand (Fig. 5B). Its inhibitory potency (IC₅₀ = 290 nM) is only ~2 fold lower than that of unmodified APY-d3 (IC₅₀ = 130 nM). Since intraperitoneal administration of 170 μg APY-d3-laur8 per mouse yielded blood concentrations of ~5 μM at 2 hours and 0.7 μM at 4 hours after injection (Fig. 4B), based on its potency in cell culture APY-d3-laur8 could be periodically administered intraperitoneally to achieve blood concentrations capable of inhibiting EphA4 for a few hours.

Figure 5. APY-d3-laur8 selectively and potently inhibits ligand-induced EphA4 activation in cells.

(A) Eph receptor selectivity was determined by measuring ephrin-A5 AP binding to EphA Fc receptors and ephrin-B2 AP binding to EphB Fc receptors in the presence of 2.5 μM APY-d3-laur8 (a concentration ~100 fold higher than the IC₅₀ value in Fig. 2). Values are normalized to ephrin binding without peptide. The graph shows averages and SEs from 3 measurements (with each measurement shown as a dot). (B) ELISA measuring EphA4 activation in cells. Dose-response curves obtained from quantifications of EphA4 phosphorylation on Y596 induced by ephrin-A5 Fc in HEK293 cells stably expressing FLAG-tagged EphA4 (HEK-EphA4 cells). The cells were treated with the indicated concentrations of APY-d3-laur8 or APY-d3 for 20 min and then with 0.5 μg/ml ephrin-A5 Fc for 10 min, or with Fc as a control. Values are normalized to no peptide treatment (plotted at 1 nM). The graph shows averages ± SE from n independent experiments. IC₅₀ values ± SE were calculated from multiparameter curve fitting of the combined data. The significance for the comparison of the best fit IC₅₀ values for the APY-d3-laur8 and APY-d3 curves was determined using the extra sum-of-squares F test; * P<0.05.

Fusion of APY-d3 to a serum albumin binding peptide

As an alternative strategy to promote peptide complexation with serum albumin, we fused the C-terminus of APY-d3 with the N-terminus of the albumin-binding cyclic peptide SA21 (RLIEDICLPRWGCLWEDD, where the two underlined Cys residues form a disulfide bond) through a βAla-βAla linker [65, 66]. The SA21 peptide has been reported to bind albumin with higher affinity than lipids, likely because it interacts with a distinct binding site in serum albumin, and to prolong in vivo half-life more than lipidation [65, 66]. The resulting bicyclic peptide 21 appears to be as potent as APY-d3 (Fig. 2 and Table 1). However, the potency was decreased ~8-fold in the presence of 20 mg/ml BSA (Fig. S1C), suggesting that serum albumin binding to the SA21 moiety interferes with the binding of the APY-d3 moiety to EphA4. The in vivo half-life of peptide 21 (~40 min) is similar to that of the lipidated APY-d3-laur8 (Fig. 4B), despite the expected stronger serum albumin binding of peptide 21. The half-life of peptide 21 might be shortened due to proteolytic cleavage by serum proteases at Arg15 in the linker between the two cyclic portions of the peptide (Table 1), leading to rapid excretion of the released APY-d3 moiety. Regardless, since the longer peptide 21 including two disulfide bonds is more difficult to synthesize than the lipidated peptides and does not appear to have obvious advantages, we did not further pursue it.

PEGylation increases the in vivo half-life of APY-d3 more effectively than lipidation

In cases where persistent EphA4 inhibition is beneficial, peptides that display even longer half-lives than the lipidated APY-d3 derivatives would be desirable. We have previously shown that a short peptide targeting another Eph receptor has a half-life of ~11 hours in the mouse circulation when coupled to high molecular weight PEG [67]. We therefore conjugated 30 kDa linear PEG to the C-terminus of APY-d3 using an amide forming reaction and two linkers of different reactivity to yield APY-d3-PEG1 and APY-d3-PEG2 (Fig. 3B,C). The peptide-PEG conjugate with the longer linker (APY-d3-PEG2) proved more potent (Fig. 6A and Table 1), but with an IC₅₀ of ~80 nM it is several folds less potent than non-PEGylated APY-d3. However, we were encouraged by the observation that the in vivo half-life of APY-d3-PEG2 is longer than 6 hours (Fig. 4C).

Figure 6. APY-d3-PEG4 selectively and potently inhibits ligand-induced EphA4 activation in cells.

(A) ELISAs comparing PEGylated APY-d3 derivatives with APY-d3 for their ability to inhibit binding of ephrin-A5 AP to EphA4 Fc. The graphs show averages ± SE from n independent experiments, each including triplicate measurements. IC₅₀ values ± SE (nM) were calculated from multiparameter curve fitting of the combined data. The curves for the PEGylated peptides (red) are compared to the curves for APY-d3 (gray) obtained in parallel in the same ELISA experiments (which are a subset of the experiments used to calculate the overall IC₅₀ values shown in Table 1). (B) Eph receptor selectivity was determined by measuring ephrin-A5 AP binding to EphA Fc receptors and ephrin-B2 AP binding to EphB Fc receptors in the presence of 1.9 μM APY-d3-PEG4 (a concentration ~100 fold higher than the IC₅₀ value in (A)). Values are normalized to ephrin binding without peptide. The graph shows averages and SEs from 3-6 measurements (with each measurement shown as a dot). (C) ELISA measuring EphA4 activation in cells. Dose-response curves obtained from quantifications of EphA4 phosphorylation on Y596 induced by ephrin-A5 Fc in HEK293 cells stably expressing FLAG-tagged EphA4 (HEK-EphA4 cells). The cells were treated with the indicated concentrations of APY-d3-PEG4 or APY-d3 for 20 min and then with 0.5 μg/ml ephrin-A5 Fc for 10 min, or with Fc as a control. Values are normalized to no peptide treatment (plotted at 1 nM). The graph shows averages ± SE from n independent experiments. IC₅₀ values ± SE were calculated from multiparameter curve fitting of the combined data. The significance for the comparison of the best fit IC₅₀ values for the APY-d3-PEG4 and APY-d3 curves was determined using the extra sum-of-squares F test; ** P<0.01. The APY-d3 curve is the same shown in Fig. 5B.

The approach used to generate APY-d3-PEG1 and APY-d3-PEG2, which involves amide formation, is not optimal because it requires deprotection of the APY-d3 N-terminus after PEG conjugation. This late stage deprotection, which does not reliably proceed to completion, presents a problem because the purity of the final conjugate cannot be accurately determined by standard analytical methods due to the heterogeneity of the high molecular weight PEG reagent. Since, as noted above, the peptide N-terminal positive charge is critical for EphA4 inhibition, incomplete deprotection results in less active peptide. We therefore developed two alternative strategies for PEG conjugation, both of which can be carried out with a deprotected N-terminus, facilitating the synthesis of large amounts of PEGylated peptide. The first strategy involves conjugating the PEG to APY-d3 through a stable oxime linkage (Fig. 3D; APY-d3-PEG3). The aldehyde and aminooxy bioconjugation groups are known to react quantitatively [68] and because the PEG conjugation is the final step in the synthesis, a conjugate of increased purity can be produced. In addition, with this strategy most of the synthetic steps occur on the solid phase, which eliminates the need for repeated intermediate HPLC purifications resulting in increased yields. The second strategy involves conjugating APY-d3 to 30 kDa metoxy-PEG-azide through a linker containing 3 amino-hexanoic acid groups and propargyl-glycine using click chemistry (Fig. 3E; APY-d3-PEG4). This conjugation can also be carried out in solution.

Remarkably, both APY-d3-PEG3 and APY-d3-PEG4 inhibit EphA4–ephrin-A5 interaction with potency similar to the non-PEGylated APY-d3 (Fig. 6A and Table 1) and their in vivo half-lives after intraperitoneal injection are >10 hours (Fig. 4C). Given the variability we observed in the potency of some batches of APY-d3-PEG3, we chose APY-d3-PEG4 for experiments to measure inhibition of ligand-induced EphA4 activation in cells. We confirmed that APY-d3-PEG4, even at high concentrations, selectively inhibits only EphA4 among the Eph receptors (Fig. 6B). Furthermore, APY-d3-PEG4 inhibits EphA4 activation in cells stimulated with ephrin-A5 ligand with a potency (IC₅₀ = 530 nM) ~4 fold lower than APY-d3 (IC₅₀ = 130 nM; Fig. 6C). Since intraperitoneal administration of 0.8 mg APY-d3-PEG4 per mouse yielded blood concentrations of ~3 μM at 20 hours after injection and 0.4 μM at 48 hours (Fig. 4C), based on its potency in cell culture APY-d3-PEG4 could be administered intraperitoneally every other day in order to maintain a concentration sufficient to considerably inhibit EphA4.

DISCUSSION

Peptides are gaining increasing interest as research tools and therapeutics [69-73]. They combine advantages of both biologics and small molecules. In particular, like biologics, peptides are suitable as inhibitors of protein-protein interactions and can exhibit high affinity and specificity with low toxicity. Furthermore, the production costs and immunogenicity of synthetic peptides are typically lower than most biologics. We previously developed the APY-d3 peptide as a potent, selective and highly protease resistant EphA4 inhibitor [9, 37, 42, 74] (Table S1]. As a small cyclic peptide with an extracellular target, APY-d3 is well-suited for further development [61, 70, 75, 76]. The next hurdle for use of the peptide to inhibit EphA4 in vivo, was to increase peptide persistence in the blood circulation. To this end, we examined the effects of lipidation, a widely used modification that causes peptide complexation with serum albumin [55], and PEGylation, another widely used modification that involves increasing the molecular mass of a peptide through conjugation with high molecular weight PEG [54, 56-58, 60, 61].

We investigated a series of lipidated peptides modified at two different positions with lipids of different lengths between 8 and 16 carbons, according to the lipidation strategies used in peptide therapeutic development [55]. As expected, we found that longer lipids such as myristoyl (14 carbons) and palmitoyl (16 carbons) most effectively increase peptide half-life in the circulation. However, the longer lipids greatly decrease the solubility of APY-d3 derivatives and cause a greater decrease in peptide inhibitory potency in ELISAs than shorter lipids. This was alleviated by using a γGlu-γGlu linker to increase the distance between the lipid and the peptide and improve serum albumin binding [55]. Our data identify lauroyl APY-d3-laur8 as the lipidated peptide that best combines acceptable solubility, high potency (IC₅₀ = 25 nM), and persistence in the blood (Table S1). Although its 40 min half-life in the blood may enable EphA4 inhibition for only a few hours after each administration, transient levels of a bioactive peptide in the blood can still be effective while reducing potential toxicities [77]. It should also be noted that the in vivo half-lives of drugs, including peptides, can be longer in human compared to mouse [55, 78, 79] and that lipidated peptides that have penetrated into tissues or tumors exhibit a lower rate of clearance [55].

We also found that binding of serum albumin to the lipidated peptides and to the serum albumin binding moiety of peptide 21 impacts the ability of APY-d3 to bind EphA4, a frequent effect of albumin on the binding of lipidated peptides to their target proteins [55]. In some cases, this could be an advantage because it would keep APY-d3 in an inactive state in the blood, thus avoiding potential unwanted side effects, for example due to inhibition of EphA4 in immune cells such as T cells and NK cells, where it is highly expressed [4]. The peptide should regain most of its activity once it reaches target tissues, where the concentration of serum albumin is much lower than in the blood [55]. Another potential advantage is that lipidation may facilitate peptide penetration into tissues and possibly across the blood-brain barrier [55, 80, 81]. Furthermore, in some neurological disorders the blood-brain barrier appears to be partially compromised, which could facilitate peptide penetration into the central nervous system [22, 82].

Since several PEGylated peptide drugs are in use in the clinic [54, 56-58, 61], we also explored PEGylation as a strategy to increase peptide persistence in the mouse blood after intraperitoneal injection. We chose to use 30 kDa linear PEG to improve the in vivo half-life of APY-d3. PEG has been widely used to increase peptide half-life in vivo because it is a highly soluble, generally non-toxic and biologically inert material, although PEG and its metabolites can accumulate in some organs with prolonged use [54, 57]. PEGylation in APY-d3-PEG4 resulted in an in vivo half-life of more than 10 hours with minimal loss of EphA4 inhibitory activity, making it our lead candidate (Table S1) for evaluating the effects of EphA4 inhibition in cancer, immunotherapy, nerve injury models and other disease models not requiring penetration across the blood-brain barrier. It will also be interesting in future studies to evaluate APY-d3 incorporation into nanoparticles for in vivo EphA4 targeting and/or imaging.

The lipidated APY-d3-laur8 and PEGylated APY-d3-PEG4 have a number of important advantages compared to other available EphA4-targeting agents, including (1) their specificity for EphA4 compared to most chemical compounds targeting the ephrin-binding pocket or the ATP-binding pocket of EphA4; (2) their ability to inhibit pathological effects mediated not only by EphA4 signaling but also by ephrin reverse signaling compared to the more limited effects of kinase inhibitors; (3) their low immunogenicity and ease of production compared to antibodies, nanobodies and Fc fusion proteins; and (4) their long in vivo half-life compared to unmodified peptides and peptidomimetics. In conclusion, the lipidated APY-d3-laur8 and the PEGylated APY-d3-PEG4 that we have developed and characterized represent promising candidates for studies of EphA4 inhibition in preclinical disease models.

METHODS

Molecular modeling of peptide 2 bound to the EphA4 LBD.

Chains C and G from our previous LBD crystal structure of the APY-d2-EphA4 complex (PDB ID 4W4Z [9]) were used as template to generate the model of the peptide 2-EphA4 LBD complex. The GGKG amino acid sequence was added at the C-terminus of APY-d2 using Coot [83] and the structure idealized in Refmac [84]. The FlexPepDock [85] server was used for subsequent refinement to generate the final displayed model.

Peptide synthesis

APY-d2, APY-d3, peptides 1-3 and peptides synthesized in the Dawson lab to be coupled to lipids or PEG were obtained through solid phase synthesis as previously described [37], with some modifications. Briefly, peptides were synthesized using manual synthetic cycles for Fmoc solid phase synthesis. Typically, syntheses were performed on a 0.2 mmol scale using either Rink amide aminomethyl resin (0.69 mmol/g, Novabiochem) or Rink amide ChemMatrix resin (0.4 mmol/g, Biotage) using standard Fmoc side chain protecting groups (Tyr/Ser, tBu; Cys, Trt; Arg, Pbf; Trp, Boc) unless otherwise noted. The bicyclic peptide 21 was synthesized with Acm protection on Cys4 and Cys12 and with Trt protection on Cys21 and Cys27. Couplings were performed with Fmoc protected amino acids (1.1 mmol) and neat N,N-diisopropylethylamine (DIEA, 1.5 mmol) dissolved in 2.5 ml of 0.4 M HCTU in DMF (1.0 mmol) for 20 min. The final residue (βAla1) was coupled as Boc-βAla-OH, except for precursor 1 (Fmoc-βAPYCVYRβASWSCβAβAK-NH2) used to generate APY-d3-PEG1 and APY-d3-PEG2, where βAla1 was coupled as Fmoc-βAla-OH. Fmoc deprotection was facilitated with 20% 4-methylpiperidine in DMF for a total of 7 min. After removal of the Fmoc group, N-terminal protection of precursor 1 was performed with 2-acetyldimedone (Dde; 2.0 mmol) at 1.0 M in DMF. Protection was complete after 3 hours as indicated by qualitative ninhydrin test.

An orthogonal protection strategy was used to attach the alkyl chain of the lipid to Lys7 or the C-terminal Lys during solid phase synthesis, according to two different methods. In the first method, the alkane was attached by a simple acylation reaction, as in the development of the FDA-approved liraglutide [86]. In the second alkylation method, the Lys was first acylated with bromoacetate and then a 1-aminoalkane displaced the bromine through an S_N2 reaction. This generates an alkylated Gly and a secondary amine at the proximal end of the attached alkane, which would be protonated at physiological pH and should help with solubility. Acyl and bromoacetyl group coupling to the Lys side chain was accomplished using orthogonal protection of the Lys side chain with ivDde (Fmoc-Lys(ivDde)-OH, Novabiochem). After primary chain elongation was completed, ivDde was removed from the Lys side chain by exposing the peptide to 4% hydrazine (8% hydrazine hydrate in DMF) in 4x 5 min incubations. Acyl groups were coupled to the Lys ε-amine using the corresponding acid chloride (2.0 mmol) at 1.0 M in DMF for 10 min. Bromoacetate coupling was achieved by pre-activating bromoacetic acid (1.0 mmol) with NHS (1.0 mmol) and neat DIC (1.0 mmol) for 45 min in DMF and adding the resultant mixture to the peptide-resin for 10 min. The secondary amine was formed by adding 1-octylamine or 1-dodecylamine (1.0 mmol) at 1.0 M in DMF to Lys-ε-bromoacetate peptide-resin for 30 min.

βAsp and γGlu couplings to the Lys ε-amine were performed as described above using Fmoc-Glu(OH)-OtBu (Novabiochem). βAsp and γGlu were coupled via their side chain carboxyl group in order to maximize linker length and to position the extra carboxylate group, which has a negative charge at neutral pH, close to the main chain of the peptide. The lipid acyl chain was then attached to the βAsp and γGlu, according to the strategy previously used in the design of glucagon-like peptide 1 derivatives [87]. Following all couplings, APY-d2, APY-d3 and peptides 1-14, 17, 18 and 21 were deprotected and cleaved from the resin with TFA:TIS:EDT:H₂O (37:1:1:1 v/v) for 2 hours at room temperature. The TFA was evaporated under a N₂ stream. The peptide was precipitated with ice-cold Et₂O, filtered, and washed with Et₂O. The crude peptides were dissolved in 10% acetic acid, 45% acetonitrile in water and lyophilized. Samples were analyzed by reversed phase high performance liquid chromatography (HPLC) and electrospray ionization mass spectrometry (ESI-MS). Peptides with significant side products were purified by HPLC.

Oxidation to form the disulfide bond was achieved by dissolving the peptides at 0.1 mg/ml in 1:1 acetonitrile:0.2 M NH₄HCO₃, pH 8 and stirring overnight open to air. The oxidized peptides were purified by HPLC. For peptide 21, the first disulfide bond was formed by dissolving the peptide at 0.1 mg/ml in 0.1 M NH₄HCO₃, pH 8 and stirring overnight open to air, and purified by HPLC. The second disulfide was formed by redissolving the peptide at 0.3 mg/ml in 1:1 acetonitrile: water, 0.1% TFA with approximately 2 equivalents I₂ dissolved at 1.0 M in 1:1 acetonitrile:water, 0.1% TFA for 15 min. The reaction was quenched with sodium ascorbate and the peptide was purified by HPLC. For APY-d3-laur8 synthesized in the Dawson lab and peptides 16, 19, and 20 and precursor 1, the disulfide bond was formed on the resin. Fmoc-Cys(Acm)-OH was introduced at position 4 and, following chain elongation, the resin was equilibrated in MeOH:CH₂Cl₂:H₂O (60:25:4 v/v), 50 ml per mmol of peptide-resin (20 mM resin-bound peptide). To this mixture, an equal volume of 80 mM I₂ in CH₂Cl₂:MeOH (8:1.5 v/v) was added for 15 min at room temperature. The I₂ solution was drained, and the remaining reagent was quenched with saturated sodium ascorbate:DMF (1:5 v/v) until no color remained. The peptides were then deprotected and cleaved from the resin in TFA:TIS:H₂O (38:1:1 v/v) and purified by HPLC as described above.

APY-d2, APY-d3, peptides 1-21 and precursor 1 were purified via semi-preparative HPLC on a Waters Prep LC 4000 system with a Jupiter 10 μm Proteo 90 Å (250 x 21.2 mm) column (Phenomenex) at flow rates of 15 ml/min using linear gradients of 0.5% buffer B/min (buffer A: 0.05% TFA in H₂O; buffer B: 0.045% TFA in 9:1 CH₃CN:H₂O). The molecular weight of each purified peptide was determined by ESI-MS (API 2000, PE/Sciex). Purity was determined by analytical reversed phase HPLC and integration of the signal at 214 nm. Analytical HPLC was performed on either a Dynamax Rainin SD-200 system or an Agilent 1100 system using linear gradients of 0.5% buffer B/min on a Jupiter 4 μm Proteo 90 Å (150 x 4.6 mm) column (Phenomenex) at flow rates of 1 ml/min. Peptides were obtained at a non-optimized yield of 5-20%. Final purities were >95% for most peptides, and >90% for all peptides (Fig. S5).

The key APY-d3-laur8 peptide was also synthesized by Vivitide/Biosynth according to the following protocol. The peptide base sequence was synthesized at a 0.25 mmole scale using Gyros Rink amide resin (0.35mmol/g) on a CEM Liberty Blue instrument. Standard protecting groups were used for all amino acids, except for Lys, for which Fmoc-Lys(ivDde)-OH was used. All amino acids were coupled at a 5-fold excess using DIC/Oxyma activators; the first 8 amino acids were single coupled, while the remaining amino acids were double coupled. Upon sequence completion, the N-terminal Fmoc was removed, and the peptide was capped with a Boc protecting group. The Lys(ivDde) was deprotected using 2% hydrazine in DMF (2 x 15 min). After rinsing, both γGlu residues were coupled using 6 equivalents of γGlu/PyAOP/HOAt and 12 equivalents of DIPEA for 30 min at 40°C. Lauric acid was coupled at 40°C overnight, using 6 equivalents of HBTU and 12 equivalents of DIPEA. Cleavage and global deprotection were performed at room temperature for 3 hours, using trifluoroacetic acid (TFA)/water/thioanisole/EMS/EDT (20:1:1:1:1). The cleaved peptide was precipitated using ether, and the precipitated peptide was rinsed with ether to remove scavengers, after which it was lyophilized. The crude material was dissolved in DMSO and TEA was added to induce disulfide formation; the reaction was allowed to proceed overnight. LC/MS was used to confirm disulfide formation. The peptide was lyophilized and then purified by reversed phase HPLC (column: C18, 120Å, 10 μm, 25 x 250 mm) using a gradient of 30-50% buffer B over 60 min (buffer A: 0.1% TFA in water, buffer B: 0.08% TFA in acetonitrile). The peptide was analyzed by ultra-high performance liquid chromatography (UHPLC)/MS to verify purity and determine the molecular weight (Fig. S3). The average calculated mw of the peptide was 2,142.0 (average theoretical mw 2,142.5) (Fig. S3A) and its purity was >95% (Fig. S3B). Pure fractions were lyophilized to yield the desired product.

To generate APY-d3-PEG1, precursor 1 N-terminally protected with 1-(4,4-dimethyl-2,6-dioxocyclohex-1-ylidene)ethyl (Dde) during solid phase synthesis was dissolved at 0.5 mM in 0.1 M borate buffer and 45 mg of 30 kDa methoxy PEG succinimidyl carboxymethyl ester (JenKem Technology USA) were added to the dissolved peptide for 30 min. To generate APY-d3-PEG2, precursor 1 was dissolved at 0.5 mM in 0.1 M borate buffer and 45 mg of 30 kDa methoxy-PEG-(CH₂)₅COO-NHS (#ME-300HS, Sunbright, NOF America Corporation) were added to the dissolved peptide for 6 hours. The PEGylated peptides were purified by HPLC using a 2% per minute gradient. The protective Dde group at the peptide N terminus was removed under mild hydrazine conditions in aqueous solution by treating the peptides dissolved at 0.5 mM in milli-Q H₂O with hydrazine (0.1 M final concentration) for 45 min.

APY-d3-PEG3 was generated by attaching 30 kDa PEG to APY-d3 without protecting its N-terminus by taking advantage of an oxime-based bioconjugation reaction using substituted anilines as nucleophilic catalysts [68]. APY-d3-βAla-glyoxyl was generated by forming the glyoxyl group from a Ser incorporated during solid phase synthesis and then conjugated with 30 kDa PEG containing an amino-oxy reactive group (methoxy-PEG-CONH(CH₂)₂-ONH₂; #ME-300CA, Sunbright, NOF America Corporation). Briefly, pure peptide-glyoxyl (1 eq) and amino-oxy-PEG 30 kDa (1.1 eq) were dissolved to a concentration of 1 mM peptide in acetate buffer (25 mM pH 4.5). Aniline was then spiked in to a concentration of 50 mM and the reaction mixture was allowed to stir for 2-6 hours. An aliquot of the reaction was used for HPLC analysis and once it was confirmed that the reaction had proceeded to completion (based on total consumption of the starting peptide), the reaction mixture was transferred to a 10 kDa dialysis bag. The crude reaction mixture was dialyzed against pure water for 48 hours to remove any salts, catalyst, and unreacted peptide that might still be in the reaction mixture and lyophilized.

The key APY-d3 PEG4 peptide was synthesized by Vivitide/Biosynth according to the following steps. Solid phase synthesis. The H-βAla-Pro-Tyr(tBu)-Cys(Trt)-Val-Tyr(tBu)-Arg(Pbf)-βAla-Ser(tBu)-Trp(Boc )-Ser(tBu)-Cys(Trt)-Ahx-Ahx-Ahx-Pra-OH peptide was synthesized manually at 0.5 mmol scale, using Fmoc/tBu chemistry and Wang resin as solid support. The first amino acid, Fmoc-Pra-OH, was loaded on the Wang resin using DMAP/DCI (sub: 0.60 mmol/g, Lot: 009166L). The coupling of the amino acids (3 eq) was carried out with DIC (3 eq), Oxyma (3 eq) and 20% DIPEA in DMF activation chemistry for 6 hours or overnight at room temperature. The Fmoc group was removed by treating the peptide resin with 20% piperidine/DMF (2x, 2 min and 20 min each). Start resin weight: 0.83 g; end resin weight: 1.6 g. Deprotection and cleavage. The peptide resin was washed with DMF, IPA, DCM, and ether and dried under vacuum for 24 hours. Then, the peptide deprotection and cleavage from the resin was performed by treatment with 30 ml TFA/triisopropylsilane/water/anisole/thioanisole/DODT (87.5/2.5/5.0/2.5/1.5/1) v/v¾) for 2 hours at room temperature. The TFA was partially evaporated, and the crude product was precipitated and washed three times with ice-cold ether, redissolved in 20% CH3CN/H2O and lyophilized using a VirTis Sentry 2.0 Lyophilizer with no lyophilization cycle (full vacuum) for 24 hours. Theoretical crude: 0.92 g; actual crude: 0.55 g; yield: 59.78%. Disulfide bond formation (oxidation) and purification. The linear crude product (0.55 g) was purified using a reversed phase HPLC column (Luna Cl8(3), 100 A, 5.0 cm diameter, Phenomenex Inc., Torrance, USA). The gradient used to purify the peptide was 15% to 45% B in 75 min. Buffer A was 0.1 % TFA in H₂O and Buffer B was 100% acetonitrile and the flow rate used was 100 ml/min. The wavelength used to detect the product was 220 nm. The product eluted around 21.7% B and the fractions were analyzed by mass spectrometry (ZEVO G2, Waters) and analytical HPLC (Shimadzu) using a C-18 column (4.5 x 250 mm, particle size 5 μm, 200 A, No: l 19YA90009, YMC-Pack-ODS-A, YMC, Japan). Pure fractions were combined to yield pure pools. The pH of the solution was adjusted to 7.9 and 200 μl of H₂O₂ were added. After stirring the solution for 24 hours, the reaction was quenched by lowering the pH with 50% acetic acid. The solution was loaded into the HPLC column, and the cyclized product was purified as above. LC/MS was used to confirm disulfide formation. Click reaction and purification. The oxidized product (1 eq, 24.1 mg, 0.0131 mmol) and metoxy-PEG-azide (2.1 eq, 828 mg, 0.0276 mmol, Creative Works, Cat. WXH0lOl 7) were dissolved in 7 ml of degassed H₂O and 3.6 ml of degassed tert-butanol. Then, 1 g of copper powder was added, and the solution was stirred for 24 hours at room temperature. The completion of the reaction was monitored by HPLC. The solvents were removed by evaporation and the impurities were removed by dialysis using H₂O. Then, the PEGylated peptide was dissolved in 20% acetonitrile/H₂O and the solution was lyophilized. Yield: 665 mg. The average calculated molecular weight of the oxidized peptide before PEGylation was 1,837.87 as determined by mass spectrometry (average theoretical molecular weight 1,838.2;Fig. S4A) and its purity was >95%, as determined by HPLC (Fig. S4B).

For biological assays, peptide stock solutions of ~10 mM were made in H₂O, PBS or DMSO, and the concentration was determined by measuring absorbance at 280 nm. It should be noted that since PEG does not absorb at this wavelength, free PEG remaining in the final preparations of the PEGylated peptides was not detected.

ELISA measuring peptide inhibitory potency and specificity for EphA4

To determine IC₅₀ values for inhibition of EphA4-ephrin-A5 interaction by the peptides, human EphA4 Fc (# CJ75, Novoprotein) or mouse EphA4 Fc (#641-A4, R&D Systems/BioTechne) at 1 μg/ml in TBST (150 mM NaCl, 50 mM Tris-HCl, pH 7.5, 0.01% Tween 20) were immobilized on protein A–coated 96-well plates (# 15132, Pierce/Thermo Scientific) for 1 hour at room temperature. The plates were then washed 3 times with TBST and incubated for ~1 hour at room temperature with 0.05 nM ephrin-A5 alkaline phosphatase (AP) and different peptide concentrations in 40 μl of TBST/well. In some cases (see Fig. S1 and Table 1), the TBST contained 40 mg/ml delipidated HSA or 20 mg/ml BSA. To remove unbound ephrin-A5 AP, the plates were washed 3 times with TBST and the amount of bound ephrin-A5 AP was quantified by adding 1 mg/ml p-nitrophenyl phosphate substrate (#34045, Pierce/Thermo Scientific) diluted in SEAP buffer (105 mM diethanolamine, 0.5 mM MgCl₂, pH 9.8). After ~1 hour incubation at 37 °C, absorbance was measured at 405 nm, and the absorbance from wells coated with Fc alone (#55911, MP Biomedicals) was subtracted as background.

To assess Eph receptor selectivity, 1 μg/ml Eph receptor Fc fusion proteins (R&D Systems/BioTechne) were immobilized on protein A-coated wells and incubated with 0.05 nM ephrin-A5 AP (for EphA receptors) or ephrin-B2 AP (for EphB receptors) in the presence or absence of peptide. The ephrin AP fusion proteins were generated as previously described [88].

Measurement of active APY-d3 derivatives in mouse blood

The method we developed to measure the concentration of active (EphA4 inhibitory) APY-d3 derivatives in the blood and calculate their in vivo half-life, is based on measuring peptide-mediated inhibition of EphA4-ephrinA5 interaction in ELISA. This method has considerable advantages compared to other more commonly used strategies to quantify peptides in blood or tissues, which typically use mass spectrometry and/or HPLC. Our ELISA-based method is straightforward and cost effective, and the same protocol can be applied to any modified version of APY-d3, without requiring development of different protocols to detect and measure different modified peptides. The method is also very sensitive, allowing measurement of nanomolar peptide concentrations using only a few microliters of blood, and detects the biologically relevant, active form of the peptide.

For intravenous administration, peptide 5 was dissolved in DMSO and diluted in 100 μl PBS (to a 5% final DMSO concentration). The other lipidated peptides were dissolved in 100 μl PBS supplemented with 40 mg/ml BSA for intravenous or intraperitoneal administration. In one experiment, peptide 16 was dissolved in PBS supplemented with 6.5 mM methyl-β-cyclodextrin, which enabled peptide solubilization and yielded similar half-life as the peptide dissolved in BSA. Peptide 21 was dissolved in 250 μl PBS and the PEGylated peptides were dissolved in 250–300 μl PBS for intraperitoneal injection. Peptides were sterile filtered before administration. Blood from each mouse was collected at 3 different time points, the first 2 times by retro-orbital bleeding and the third time by terminal cardiac bleeding under deep anesthesia. Red blood cells were removed by centrifugation from blood collected in heparinized microcentrifuge tubes (#365965, Microtainer) and the resulting plasma was frozen in aliquots.

To measure the amount of active peptide present in the plasma, we used the ELISA measuring peptide inhibitory potency (see above). For each time point, we obtained dose-response curves by using different plasma dilutions (Fig. S2A). We deduced the concentration of active peptide remaining in the blood at various time points after administration by: (1) measuring the “0 min (control)” IC₅₀ value for a peptide diluted in plasma at the concentration expected in the mouse blood, assuming that all the peptide reaches the circulation and a mouse blood volume of 2.5 ml and (2) comparing the apparent IC₅₀ values for the peptide recovered in plasma at different time points to the “0 min (control)” IC₅₀ value (Fig. S2). For example, the ~5-fold increase in the apparent IC₅₀ value observed using plasma obtained 120 min after peptide administration indicates a 5-fold decrease in peptide concentration in the blood, and therefore that ~20% of the injected peptide remains in the blood after 120 min (Fig. S2B). The in-vivo half-life of each peptide was determined using 2-7 mice per peptide.

ELISA to measure EphA4 activation in cells

To detect EphA4 activation in cells, we developed a method involving capture of FLAG-tagged EphA4 from cell lysates in ELISA wells coated with an anti-FLAG antibody and detection of EphA4 phosphorylation on tyrosine 596 (pY596) using an antibody to the conserved pY588 motif of EphA2, followed by an HRP-conjugated secondary antibody.

Specifically, HEK293AD cells stably transfected to express FLAG-tagged EphA4 (HEK-EphA4 cells) [9] were cultured at 37°C in Dulbecco’s Modified Eagle Medium (DMEM; # 10-013-CV, Corning/Thermo Fisher Scientific) containing 10% fetal bovine serum (FBS) with 1% antibiotic-antimycotic solution (#30-004-Cl, Corning). To assess inhibition of EphA4 activation in cells, HEK-EphA4 cells were plated in 48-well plates. Once they reached 60-70% confluence, the cells were starved for 1 hour in culture medium without FBS, incubated for 20 min with peptide at different concentrations, and then treated for an additional 10 min with 0.5 μg/ml ephrin-A5 Fc (#374-EA-200, R&D Systems/Bio-Techne) to activate EphA4 in the continued presence of the peptide. After treatment, cells were rinsed once with ice-cold PBS containing Ca⁺ and Mg⁺ (#17-513F, Lonza Bioscience) and collected in 250 μl TX100 buffer (1% Triton X-100, 10% glycerol, 2 mM EDTA in PBS) containing phosphatase and protease inhibitors (#78443, Thermo Fisher Scientific). Lysates were incubated for 30 min on a rocking platform at 4°C, and centrifuged for 10 min at 16,000g.

For ELISAs, high binding capacity polystyrene 96-well plates were coated overnight at 4°C with 1 μg/ml mouse anti-FLAG antibody (#F3165, Sigma-Aldrich) in 50 μl/well borate buffer (0.1 M boric acid, 0.1 M Na borate, pH 8.7). The plates were then washed 5 times with TBST, incubated for 1.5 hours with 200 μl/well 5 mg/ml BSA (#3116964001, Roche) in PBS, washed 5 times with TBST, incubated overnight at 4°C with 200 μl cell lysate/well, washed 3 times with TBST, incubated for 1.5 hour at room temperature with 100 μl/well anti-EphA2 pY588 antibody (#12677S, Cell Signaling Technology), diluted 1:4,000 in TBST. Wells were then washed 5 times with TBST, incubated for 1 hour at room temperature with 100 μl/well anti-rabbit horseradish peroxidase (#A16110, Life Technologies) diluted 1:1,000 in TBST, washed 5 times with TBST, and incubated with 100 μl/well tetramethylbenzidine (TMB) substrate solution (#7004P6, Cell Signaling Technology) for ~10 min before addition of 100 μl/well stop solution (#DY994, R&D Systems). OD450 was then measured.

Highlights.

• The EphA4 receptor tyrosine kinase plays a role in multiple pathological processes

• The cyclic APY-d3 peptide is a highly specific and potent EphA4 antagonist

• APY-d3 is resistant to proteolysis but has a very short half-life in the blood

• Lipidated APY-d3-laur8 and PEGylated APY-d3-PEG4 have increased in vivo half-life

• The two new APY-d3 derivatives show promise for use in preclinical disease models

ABBREVIATIONS:

AP

alkaline phosphatase

APP

amyloid precursor protein

BSA

bovine serum albumin

EGF

epidermal growth factor

ELISA

enzyme-linked immunosorbent assay

Fc

fragment crystallizable

FNIII

fibronectin type III

GPI

glycosylphosphatidylinositol

LBD

ligand-binding domain

HSA

human serum albumin

PEG

polyethylene glycol

SAM domain

sterile alpha motif domain