A Helix-Stabilizing Linker Improves Subcutaneous Bioavailability of a Helical Peptide Independent of Linker Lipophilicity

Abstract

Stabilized peptides address several limitations to peptide-based imaging agents and therapeutics such as poor stability and low affinity due to conformational flexibility. There is also active research in developing these compounds for intracellular drug targeting, and significant efforts have been invested to determine the effects of helix stabilization on intracellular delivery. However, much less is known about the impact on other pharmacokinetic parameters such as plasma clearance and bioavailability. We investigated the effect of different fluorescent helix-stabilizing linkers with varying lipophilicity on subcutaneous (SC) bioavailability using the glucagon-like peptide-1 (GLP-1) receptor ligand exendin as a model system. The stabilized peptides showed significantly higher protease resistance and increased bioavailability independent of linker hydrophilicity, and all subcutaneously delivered conjugates were able to successfully target the islets of Langerhans with high specificity. The lipophilic peptide variants had slower absorption and plasma clearance than their respective hydrophilic conjugates, and the absolute bioavailability was also lower likely due to the longer residence times in the skin. The ease and efficiency of double-click helix stabilization chemistries is a useful tool for increasing the bioavailability of peptide therapeutics, many of which suffer from rapid in vivo protease degradation. Helix stabilization using linkers of varying lipophilicity can further control SC absorption and clearance rates to customize plasma pharmacokinetics.

Introduction

Peptides acting as hormones and growth factors serve selective and crucial signaling roles via binding to specific cell surface receptors¹. These characteristics make peptides and peptidomimetics attractive candidates as therapeutics, but their rapid renal clearance and poor stability from in vivo proteases have limited their broader use in the clinic^{2, 3}. Recently, improvements in synthetic and analytical techniques, high throughput screening, and helix stabilization chemistries have generated a resurgence in the research and development of therapeutic peptides, as demonstrated by the increasing number of peptide therapeutics in clinical trials^{1, 4, 5}.

Peptide therapeutics are often delivered through subcutaneous injection. Although oral administration is the preferred route of delivery for therapeutics, it is generally not applicable to peptides as the harsh environment of the gastrointestinal tract is abundant with proteases. Peptides also suffer from permeability limitations across the intestinal epithelium^{6, 7}. Studies that tested stapled peptides (chemically cross-linked helices) for oral administration showed increases in oral availability, and while promising, further improvements are needed before clinical translation⁸. Alternatively, IV delivery requires technically trained personnel (with associated risks^9–11) and is particularly cumbersome for peptides with fast clearance that require frequent dosing. Given the limitations of intravenous and oral routes, subcutaneously administered peptide therapeutics have seen commercial success with multiple FDA approved molecules and more in clinical trials, notably cancer vaccines, GLP-1 agonists used to treat type 2 diabetes, and anti-inflammatory agents^{1, 12, 13}. Despite the importance of this route of administration, the mechanisms controlling subcutaneous bioavailability are incompletely understood¹².

SC doses are given as a bolus, and the injected therapeutic is absorbed from the site of injection into the blood or lymph capillaries to enter systemic circulation^{12, 14}. These injections are readily self-administered and exhibit a slower absorption rate when compared to the instantaneous systemic delivery of an IV bolus. This provides effective, long-term treatment and may alleviate side effects attributed to high serum concentrations¹⁵. The longer residence time in the body is especially important for peptide therapeutics given their rapid clearance, and SC administration allows the molecule to circumvent the short residence time from renal filtration through absorption-limited pharmacokinetics¹⁶.

Molecular characteristics impacting interstitial transport include shape, molecular weight, charge, and stability in the presence of extracellular proteases^{11, 14, 17, 18}. The bioavailability of a subcutaneously injected therapeutic, defined as the fraction of the dose that makes it to the systemic circulation, has been shown to be dose and injection site-dependent for several different antibody treatments^{19, 20}. Further complicating drug design is a lack of robust tissue and cellular models to explain SC absorption kinetics. For certain antibody treatments, the SC route has received FDA approval, but the wide observed ranges in bioavailability remain unexplained, although factors such as interstitial pressure, applied heat, protein size, formulation, and in the case of IgGs, FcRn binding, are thought to play a major role^21–24. Assuming negligible cellular uptake/trapping at the site of injection, SC bioavailability is theoretically 100% for a catabolically stable molecule²⁵. However, peptide and protein drugs exhibit highly variable SC bioavailability, suggesting degradation during the absorption process^{14, 22, 26}. Pretreatment of the site of injection with protease inhibitor-containing ointments increases the bioavailability of subcutaneously injected insulin²⁷. Others have found evidence of degradation products of injected proteins and peptides using in vitro assays of subcutaneous tissue homogenates or by examining the site of injection^{28, 29}. Given the role of proteases in hindering absorption, the design of protease-resistant peptides is crucial for the development of highly bioavailable peptide and protein-based therapeutics. Using stabilization strategies that can impart varying lipophilic and hydrophilic properties can further fine-tune the absorption rate as shown by favorable pharmacokinetics of transdermally applied stabilized exendin³⁰, but the quantitative impact on bioavailability for the subcutaneous route and effects of peptide stability are not well understood.

One increasingly common way to improve peptide stability is through helix stabilization. Although several techniques for generating stabilized alpha helices have been reported, including olefin metathesis, lactam ring formation, and disulfide bond formation among others, another promising alternative is to use a double copper (I)-catalyzed click reaction between two azidohomoalanine (AHA) amino acid substitutions along the peptide sequence^31–36. Azidohomoalanine is easily incorporated during solid phase peptide synthesis (SPPS) and has been previously used to generate fluorescent imaging agents via alkyne fluorophore conjugation as well as stabilized fluorescent and non-fluorescent peptides such as exendin and inhibitors of Mdm2^{31, 32, 37, 38}. Based on past results focused on the glucagon-like peptide-1 receptor (GLP-1R) ligand exendin and FDA approval of multiple subcutaneously administered exendin analogues for treating type 2 diabetes including exenatide, liraglutide, and dulaglutide, exendin was chosen as the model system for investigating the effect of lipophilicity and stability on SC bioavailability^{39, 40}. Helix stabilization and lipophilicity impact the absorption and bioavailability in complex ways. Increased protein binding and/or membrane interaction slows absorption and can sequester the peptide from proteolytic enzymes but simultaneously increases exposure time to epithelial proteases. Many lipophilic agents have been utilized to increase plasma protein binding of peptides including fatty acids⁴¹, diphenyl cyclohexanol⁴², and lipophilic fluorescent dyes^{43, 44}. Alternatively, albumin-binding peptides can be used for a similar purpose ^45–47, although these peptides would also be subject to protease degradation following subcutaneous delivery. Alpha helix stabilization may alter absorption and clearance rates in addition to improving protease stability.

Due to the interplay of these factors, we studied the impact of helix stabilization and lipophilicity on SC absorption and bioavailability of stabilized and non-stabilized exendin derivatives. Conjugation of either Alexa Fluor 680 (AF680) or Cy7 fluorophores on the alkyne-functionalized linkers increased or decreased the overall lipophilicity. Here we report increased bioavailability of these stabilized alpha helical peptides compared to their non-stabilized counterparts and slowed plasma clearance using lipophilic dye linkers. A stronger understanding of the impact of these modifications on bioavailability and clearance is crucial for generating predictive models to aid in clinical translation.

Results

To investigate the effects of peptide stabilization on in vivo subcutaneous (SC) absorption using linkers with different lipophilicity, two series of fluorescently labeled exendin probes were synthesized (Fig. 1). Based on crystallography data for the binding pocket of glucagon-like peptide-1 receptor (GLP-1R) for exendin⁴⁸, AHA substitutions were made at the 14^th position for single mutant exendin and both 14^th and 21^st positions for double mutant exendin. These substitutions allow for a direct comparison between stabilized and non-stabilized peptide. A more hydrophilic set of peptides was generated by first functionalizing Alexa Fluor 680 NHS ester with either 1 or 2 reactive alkynes followed by a single-click cycloaddition for non-stabilized peptide or a double-click cycloaddition for stabilized peptide. Similar syntheses were performed using Cy7 NHS ester to generate a lipophilic pair of exendin conjugates. Careful consideration was taken when choosing dyes, as dye structure and molecular charge can greatly impact non-specific cellular and plasma protein interactions^{49, 50}. Cy7 and Alexa Fluor 680 were chosen based on plasma protein binding of the free dye as determined by rapid equilibrium dialysis as well as previously published data on Cy7 exendin and s-AF680 exendin^{32, 51}. All fluorescent peptides were purified with moderate yield (40–60%) and the exact mass confirmed by ESI-MS (SI). Isotopic spacing corresponded to singly stabilized peptides, and no branched peptide multimers were detected. Alexa Fluor 680 conjugates were ionized in negative mode while Cy7 conjugates were ionized in positive mode for ESI-MS. Purity was determined on RP-HPLC while monitoring both 214 nm and peak fluorophore absorbance (SI). The logD was experimentally measured for each conjugate to confirm the increased lipophilicity of the Cy7 conjugates (SI).

Figure 1.

Design and synthesis of stabilized and non-stabilized exendin conjugates. The modified residues are highlighted in red. Fluorophores are first functionalized with one or two alkynes using amine-NHS chemistry, followed by either single or double click reactions to generate the fluorescent conjugates.

The binding affinity of each exendin conjugate to GLP-1R was experimentally determined to quantify the impact of stabilization and dye lipophilicity. In vitro twelve-point affinity curves indicate all conjugates maintain the low nanomolar affinity unmodified exendin exhibits^52–55. The data also indicate stabilization resulted in a small increase in affinity (lower K_d) when compared to the non-stabilized counterpart for each dye (Fig. 2). The affinity improvement is not surprising given helix stabilization is known to lessen entropic penalties to binding through conformational constraints⁵⁶.

Figure 2.

Peptide characterization includes binding affinity (A), trypsin digest (B), and circular dichroism measurements (C). All conjugates maintain strong binding affinity to GLP-1R. Stabilized exendin conjugates demonstrate superior protease resistance compared to single click peptides. Affinity, digest half-lives, and helicity are tabulated (D).

Changes in proteolytic stability due to helix stabilization were investigated through a trypsin digest adapted from a previously published protocol³² as well as serum and plasma stability measurements (SI). Digests of stabilized and non-stabilized peptides (Fig. 2) and unreacted double mutant exendin (SI) in 5 ng/µl trypsin indicate the double-click stabilization across the i, i+7 residues significantly improves protease resistance. Non-stabilized AF680 and Cy7 exendin demonstrated digest half-lives of 4.9 ± 0.2 h and 9 ± 1 h, respectively. The stabilized counterparts showed improvements with digest half-lives of 13 ± 1 h and 14 ± 1 h, respectively. Unreacted double mutant exendin digested with a half-life of 0.8 ± 0.2 h (SI).

CD spectra of peptides were collected in 1:1 water:acetonitrile to quantify helicity. Although CD spectra of peptides are typically collected in a mild potassium phosphate buffer, the Cy7 conjugates absorbance spectra in aqueous solutions showed a broad shoulder near 700 nm. This shoulder is inconsistent with the original fluorophore absorbance spectra (SI), and suggests H-aggregates of the dye⁵⁷, which are common with lipophilic cyanine dyes in aqueous solutions at the high concentrations needed for CD measurements. This absorbance anomaly for Cy7 was absent in serum (SI). Therefore, when bound to plasma proteins and/or at lower (physiologic) concentrations, the molecule is unlikely to be aggregated. Since the absorbance spectra in 1:1 water:acetonitrile was normal for Cy7 and had no impact on the CD measurements of the non-aggregation prone AF680 peptides (SI), the disaggregated state in 1:1 water:acetonitrile may more accurately reflect the physiologic state of the molecule. Though less common, organic solvents have been used for collecting CD spectra of poorly soluble peptides^{31, 38}. The CD spectra for Cy7 in 5 mM potassium phosphate buffer showed lower helicity similar to the unmodified peptide (SI)³². Cy7 conjugates displayed higher helicity in the disaggregated form, regardless of stabilization (χ^helix=0.84) possibly due to side chain interactions with the lipophilic dye. The 221 nm absorbance on CD indicated a minor increase in helicity due to stabilization for AF680 conjugates. AF680 conjugate helicity and the small increase upon stabilization agree with previously published values for exendin helicity and are consistent with the significant helicity of wild-type exendin.

Given the specificity of these exendin conjugates for pancreatic beta cell targeting, the subcutaneous bioavailability of the compounds was investigated. One set of C57BL/6 mice was given 1 nmol of Cy7 or AF680 exendin with or without helix stabilization intravenously (Fig. 3). A second set of mice was given the same dose subcutaneously. AUC calculated from the subcutaneous delivery divided by the intravenous plasma concentration AUC provided the absolute bioavailability. The s-AF680 exendin demonstrated complete bioavailability (103 ± 6%) when compared to AF680 exendin (65 ± 6%). Similarly, stabilization via a lipophilic Cy7 linker yielded improved bioavailability of 82 ± 12% vs. 57 ± 9% for stabilized and non-stabilized peptides, respectively. The differences in bioavailability between stabilized and non-stabilized peptides are statistically significant (p value = 0.0004 and 0.02 for AF680 and Cy7 conjugates, respectively). The fluorescent non-stabilized exendin bioavailability results agree well with previously reported values for exendin in rats^58–60. SC injection of non-stabilized Cy7 exendin resulted in higher plasma concentrations than stabilized Cy7 due to slower clearance, although overall bioavailability increased through helix stabilization. SDS-PAGE was used to verify that experimentally measured plasma intensities were a result of intact peptide absorbed from the skin rather than degradation products (SI). Degradation byproducts are fluorescent as well and can lead to errors while quantifying bioavailability, but no fluorescent degradation byproducts were detected in the plasma. Fluorescent exendin peptides subject to a trypsin digest were also analyzed using SDS-PAGE to verify that fragments differ significantly from intact, control peptide. To verify blood stability, conjugates were incubated at 37°C for 24 h in both mouse plasma (EDTA added) and fresh mouse serum. Degradation rates in both serum and plasma were generally low (<10% degraded) over 24 h (SI), consistent with previously published values for human serum stability⁶¹.

Figure 3.

Fluorescent exendin with and without helix stabilization were administered either intravenously or subcutaneously in mice. The plasma concentration was quantified and plotted versus time to determine the effect of helix stabilization on SC bioavailability. Stabilized AF680 and Cy7 exendin displays significantly higher bioavailability than their non-stabilized counterparts.

Following the last blood sample (24 h), each animal was sacrificed, and the pancreas was removed for macroscopic imaging to confirm beta cell targeting specificity. As expected, all conjugates successfully targeted the islets of Langerhans, visualized as punctate white spots scattered across the organ (Fig. 4) and consistent with previous work showing overlap with insulin staining and MIP-GFP^{32, 62}. The combination of in vitro and in vivo sample analysis on SDS-PAGE and specific beta cell targeting demonstrates that intact and functional peptide is absorbed into the plasma.

Figure 4.

Macroscopic pancreas images demonstrate islet targeting whether peptide is administered subcutaneously or intravenously for all exendin conjugates. Beta cells are located in the islets of Langerhans and appear as distinct punctate spots.

To quantify the impact of stabilization and lipophilicity on absorption, degradation and bioavailability, a simple three-compartment model was used to calculate first-order absorption and degradation rate constants (Fig. 5). The exchange rates between the central and peripheral compartment and the clearance rate were fit using the intravenous plasma clearance. These values were then fixed, and the absorption rate and extracellular degradation rate were fit to the subcutaneous absorption data.

Figure 5.

A three-compartment analysis of transient peptide concentration in plasma. Experimental plasma concentrations were measured after IV administration of each peptide and used to fit the rates shown with blue arrows. These values were then fixed, and the rates for the green arrows were fit to the data following SC administration to quantify absorption and degradation rates. Fitted half-lives suggest increased proteolytic resistance resulting from helix stabilization and slowed absorption due to linker lipophilicity determine the in vivo absorption profile and absolute SC bioavailability.

Discussion

Owing to their low off-target effects, ease of synthesis via SPPS, and availability of tools for engineering including phage display, bacterial surface display, one-bead-one-compound selection and others, peptides hold clinical relevance both as imaging agents and therapeutics. There has also been a recent resurgence in investigating peptides for targeting intracellular protein-protein interactions. However, significant challenges such as poor affinity, low in vivo stability, and rapid systemic clearance lower the potential for clinical efficacy. One strategy to address these shortcomings involves secondary structure stabilization^{4, 8, 63, 64}, which has been shown to improve protease stability and affinity by decreasing the entropic penalty upon binding. This works by constraining the helix and confers protease resistance by increasing helicity⁶⁵ and sterically hindering digestive enzymes with non-natural amino acids and crosslinks. Significant effort has been spent investigating the impact of helix stabilization on binding affinity and intracellular delivery, but much less is known about the impact on subcutaneous bioavailability. This work used a model exendin system to quantify the interplay of helix stabilization and peptide lipophilicity on the absolute SC bioavailability and absorption profile of stabilized and non-stabilized alpha helical peptides.

Currently, multiple peptide therapeutics have FDA approval for SC delivery. However, abundant epidermal proteases (e.g. in the extracellular matrix⁶⁶ and on immune cells⁶⁷) can degrade the peptide, often rendering a large fraction of the injected dose ineffective, especially if the dose reaches systemic circulation slowly²¹. Hydrophilic peptides that are absorbed more rapidly can escape the degradation pathway, but rapid plasma clearance of a quickly absorbed bolus dose is equally unwanted since the systemic dose may be too low to achieve a sustained therapeutic effect. In contrast, lipophilic compounds are known to interact with plasma proteins, and plasma protein binding is an established method for slowing the plasma clearance of molecules by conjugating lipophilic moieties^{39, 43, 44, 51, 68}. The increase in lipophilicity may lower solubility in aqueous formulations compared to the hydrophilic counterparts and can result in self-association, aggregates, and/or precipitates. Consequently, poor solubility is a potential problem in protein pharmaceuticals^{17, 69, 70}. Although free Cy7 dye is poorly soluble in aqueous solutions, the peptide conjugate is much more soluble due to exendin’s high aqueous solubility. To further avoid aggregation, fluorescent conjugates were stored in a 1:1 dilution of water:acetonitrile and diluted in PBS prior to SC injection; no aggregation is visible prior to injection. Given the complex interactions between self-association, non-specific interactions, proteolysis, and absorption rates, the degradation and absorption rates must be considered together. Peptide stability and absorption rates were independently changed to quantify the impact on absolute bioavailability. A quantitative and mechanistic understanding of these rates is critical since variability in animals makes it challenging to scale to the clinic.

Peptides were synthesized using either single alkyne or dialkyne linkers functionalized with Alexa Fluor 680 or Cy7 to independently modify the protease stability and lipophilicity (Fig. 1). As demonstrated by in vitro cellular affinity assays, all peptides maintained low nanomolar affinity (Fig. 2). Helix stabilization was able to improve the affinity 1.4–1.6 fold, and this small increase for an already helical native peptide agrees well with a three-state thermodynamic model of binding³². CD spectra were collected to compare with the differences in binding affinity (Fig. 2). Interestingly, both Cy7 exendin conjugates displayed high helicity regardless of helix stabilization. Non-covalent side chain interactions (such as phenylalanine-phenylalanine interactions at i, i+4 residues^{71, 72} and phenylalanine-lipid interactions⁷³) have been shown to contribute to the overall peptide helix stability. The CD measurements presented here indicate that a fluorescent pharmacokinetic modifier (Cy7) may have a similar impact as non-fluorescent modifiers (lipids) by inducing helicity of peptides with hydrophobic side chains as shown by various lipidated GLP-1 peptides⁷³. Intrahelix associations with a nearby F22 residue may explain both the increased helicity for the Cy7 variants and reduced affinity. The F22 residue is in the center of the hydrophobic binding pocket⁴⁸, where steric hindrance could lower the affinity despite the higher helicity.

To observe how the improved affinity and increased protease resistance of the stabilized peptides impact SC bioavailability, 1 nmol of each of the four fluorescent exendin conjugates was delivered intravenously or subcutaneously to mice (Fig. 3). The lipophilic Cy7 conjugates were absorbed more slowly with the maximum plasma concentration (C_max) occurring at ~2 h compared to the hydrophilic AF680 peptides, which reached C_max at ~30 min. This is expected due to increased non-specific interactions and protein binding when compared to hydrophilic molecules⁴⁰. CD spectra also indicate higher helicity for Cy7 conjugates, and the physical state of the peptide at the site of injection is known to affect the pharmacokinetics of absorption¹⁷. Although helical forms can be readily absorbed from the SC site, increased non-specific interactions from the lipophilic dye likely dominate compared to helicity, hence the slower absorption rates.

The higher bioavailability of the stabilized peptides correlates with improved proteolytic resistance. Although subcutaneous absorption is a complex process and incompletely understood¹², the epidermis contains abundant proteases necessary for maintaining homeostasis of the skin⁶⁶. These enzymes have been implicated in degrading peptide therapeutics before reaching the systemic circulation, reducing their efficacy. Proteases almost universally cleave amide bonds in an extended beta strand conformation, rendering the alpha helix conformation more stable than linear peptides⁷⁴. The non-stabilized fluorescent conjugates were not able to achieve total bioavailability although the amount was high (65% and 57% for AF680 and Cy7, respectively). Cy7 stabilization also showed improved bioavailability over non-stabilized Cy7 exendin (82% vs. 57%), and complete bioavailability was possible through helix stabilization using AF680 (103%). All peptides maintained efficient targeting of islets following subcutaneous delivery (Fig. 4). These improvements are likely due to the reduced proteolysis from the stabilized secondary structure. Improved absorption has been demonstrated using protease inhibitors concurrent with treatment, but these approaches may cause unwanted side effects given that defective proteolytic enzymes in the epidermis cause multiple skin disorders⁶⁶. Engineering peptides that are more resistant to these degradation pathways avoids special formulation requirements.

The absorption rate of the peptides in the skin inversely correlated with the plasma AUC following intravenous injection. For example, the non-stabilized Cy7 peptide had the highest AUC (SI) and also the slowest absorption rate, while the AF680 peptides had the lowest AUC and fastest absorption rates. This likely results from the amount of interaction with skin and plasma proteins. These data are also consistent with non-specific cellular interactions measured in vitro, where non-stabilized Cy7 had the highest level of cellular uptake, correlating to the highest plasma protein binding (⁵⁰ and SI) .

Interestingly, when comparing liraglutide and semaglutide, Lau et al. found that the slower clearing peptide (semaglutide) also had the highest subcutaneous bioavailability (94% versus 66%)⁴⁰. They argued that the higher protein binding of semaglutide reduced clearance (due to binding plasma proteins) and made the molecule less susceptible to subcutaneous protease degradation (by binding proteins in the skin). For the fluorophore-clicked peptides, the slower clearing (and presumably more protein bound in the subcutaneous space) unstabilized Cy7 peptide would then be expected to have increased bioavailability relative to the s-Cy7 peptide (in contrast to the results) in the absence of helix-stabilizing effects. Therefore, the more rapidly clearing s-Cy7 peptide (which is presumably less bound to proteins in the skin) likely had to overcome increased protease susceptibility in the subcutaneous space to still attain improved bioavailability over the unstabilized Cy7 peptide. The increased protease susceptibility of the s-Cy7 peptide could explain the lower magnitude of improved bioavailability upon stabilization for the Cy7 peptides than with AF680 peptides (where s-AF680 and AF680 peptides had similar plasma clearance, absorption rates, and presumably protease exposure in the skin). The AF680 labeled peptides would be the most susceptible to proteases, but their rapid absorption reduces the time of exposure. These results provide guidance for designing stabilizing linkers with increased plasma protein binding to achieve maximum bioavailability^{30, 40}. Helix stabilization can increase subcutaneous bioavailability while increased plasma protein binding improves the AUC and has competing effects (increased exposure time but decreased protease susceptibility) on bioavailability (SI).

In situ measurement of the peptide degradation and absorption rates is challenging, which contributes to the incomplete understanding of subcutaneous bioavailability. To estimate these values, a 3-compartmental model was modified from Lu et al. with parameters fitted using MATLAB (Fig. 5)⁷⁵. The fitted degradation rates are qualitatively consistent with the in vitro degradation rates (SI); slower absorption rates with the lipophilic exendin conjugates are also consistent with the compartmental model. The factors affecting delivery are complex and include size, pKa, solubility, concentration, injection depth, body movement, blood supply, injection site, among many others^{12, 17}. There is a limited understanding how these factors quantitatively impact bioavailability, hence the lack of strong predictive models for SC delivery. The current model lumps these factors into overall absorption and degradation mechanisms, but a more detailed and quantitative understanding of the underlying steps in SC drug delivery will allow for better predictions and scaling to the clinic in the future.

Stabilized alpha helices allow researchers to investigate the impact of physicochemical properties on absorption at the site of injection. Adding lipophilic moieties through side chain substitution or conjugation is an increasingly used strategy to engineer ‘long-acting’ pharmaceutics because patients require fewer doses to maintain an efficacious systemic concentration^{41, 76, 77}. However, long residence times in the skin can lower the bioavailability of these formulations⁷⁸. Stabilizing peptides via double-click reaction using a lipophilic dye-linker molecule simultaneously remedies low bioavailability and promotes plasma protein interactions to slow systemic clearance.

In conclusion, we independently modified the protease stability and lipophilicity of fluorescently labeled exendin molecules using stabilized and non-stabilized variants with either lipophilic or hydrophilic dyes as pharmacokinetic modifiers. Helix stabilization via the double-click linker resulted in improved bioavailability when all peptides were administered subcutaneously in mice likely from improved protease resistance. This result occurred for both hydrophilic and lipophilic derivatives, providing an independent method for improving bioavailability. Lipophilicity of the stabilizing crosslinker also significantly impacted absorption rates and AUC. The combination of imparting lipophilicity to slow systemic clearance and protease resistance to increase bioavailability is an attractive strategy for engineering peptide therapeutics.

Experimental Procedures

Similar to previously published work, either one or two AHA substitutions were made to generate single-mutant and double-mutant exendin^{32, 51}. A methionine residue at the 14^th position was substituted with AHA during solid phase peptide synthesis based on its position pointing away from the binding pocket of GLP-1R⁴⁸. An additional substitution replaced leucine at the 21^st position with AHA to generate a double mutant exendin; i, i+7 residues correspond to positions located two helix turns apart and enabled subsequent helix stabilization.

Materials

Double mutant exendin-4 (HGEGTFTSDLSKQXEEEAVRXFIEWLKNGGPSSGAPPPS) and single mutant exendin-4 (HGEGTFTSDLSKQXEEEAVRLFIEWLKNGGPSSGAPPPS), where is X is the non-natural amino acid azidohomoalanine, were obtained from Innopep (San Diego, CA). Alexa Fluor 680 NHS ester and Cy7 NHS ester were purchased from Life Technologies (Carlsbad, CA) and Lumiprobe (Hallandale Beach, FL), respectively. N-boc-2m2’-(ethylenedioxy)-diethylamine was purchased from Santa Cruz Biotechnology (Dallas, TX). Other reagents were purchased from Sigma-Aldrich (Milwaukee, WI). ESI-MS spectra were collected using an Agilent Q-TOF 1200 series. RP-HPLC was performed on a Shimadzu LC unit using Phenomenex Luna C18(2) analytical and semi-prep columns.

Preparation of Fluorescent Alkyne and Dialkyne

Fluorescent dialkynes are synthesized from a previously published N-(2-(2-(2-aminoethoxy)ethoxy)ethyl)-N-(prop-2-yn-1-yl)prop-2-yn-1-amine linker (1)³². Propargylamine or 1 (5 µmol) was added to 200 µl of 1:1 water:acetonitrile with 20 µl 7.5% sodium bicarbonate. Either Cy7 NHS ester or Alexa Fluor 680 NHS ester (0.5 µmol in DMSO) was added and the reaction was stirred at room temperature for 30 min followed by purification on preparative RP-HPLC (AF680 alkyne: calcd: 895.17, found: [M-H]^- = 894.16; AF680 dialkyne: calcd: 1064.28, found: [M-H]^- = 1063.27; Cy7 alkyne: calcd: 586.38, found: [M] = 586.38; Cy7 dialkyne: calcd: 755.49, found: [M] = 755.49).

Preparation of Stabilized (s-)/non-stabilized Exendin

AF680 alkyne, AF680 dialkyne, Cy7 alkyne, or Cy7 dialkyne (300 nmol) was added to 100 µl of 1:1 water:tert-butanol followed by CuSO₄-TBTA (10 nmol in 1:1 water:DMSO) and sodium ascorbate (300 nmol in water). Lastly, single mutant or double mutant exendin was added (350 nmol) and the reaction was gently stirred at room temperature for 5 h followed by purification using RP-HPLC. The purified peptides were dissolved in methanol and concentrated under reduced pressure three times followed by overnight lyophilization (AF680 exendin: calcd: 5074.22, found: 5074.19; s-AF680 exendin: calcd: 5256.30, found: 5256.26; Cy7 exendin: calcd: 4765.43, found: 4765.41; s-Cy7 exendin: calcd: 4947.51, found: 4947.50).

Cell Culture

NIT-1, a GLP-1R positive mouse beta cell line, were grown in F12K containing 10% (v/v) FBS, 50 U/mL penicillin, 50 µg/mL streptomycin, and 1.5 g/L sodium bicarbonate. Passage number used for cellular assays was between 6–10.

In Vitro Receptor Binding Assay

NIT-1 cells were grown for 48 h before being harvested with 0.05% trypsin-EDTA, washed and resuspended in PBS containing 0.1% bovine serum albumin (BSA). Cells were aliquoted and suspended in binding buffer containing fluorescent exendin conjugates ranging in concentration (0.01 nM – 340 nM) for 3 h on ice. Cells were then washed two times with 0.1% BSA in PBS and immediately analyzed using an Attune Acoustic Focusing Cytometer (Applied Biosystems). The binding affinity (K_d) was calculated using Prism 6.0 software.

Trypsin Digest

To compare the digest half-lives between fluorescent exendin conjugates, both peptides were subject to a trypsin digest adapted from previously published methods³². In short, 0.05% trypsin-EDTA was diluted to a digest concentration of 5 ng/µL (DF100) with PBS (pH 7.4, room temperature). Peptide (45–50 µM) was added and the digest was monitored at 214 nm using HPLC to quantify the AUC of intact peptide (SI). AUC of the intact peptide peak was plotted against time and fit to an exponential decay to determine a degradation half-life.

Circular Dichroism Spectra

To quantify and compare changes in peptide secondary structure, CD spectra were collected on a Jasco-815 CD spectrometer. Peptides were dissolved in 1:1 water:acetonitrile at concentrations determined by amino acid analysis. Scans containing only 1:1 water:acetonitrile were collected and subtracted as background. Helicity values for peptides were quantified using CD absorbance at 221 nm and the maximum ellipticity⁷⁹.

Animals

Animal experiments were in compliance with the University of Michigan Institutional Animal Care and Use Committee (IACUC). All fluorescent exendin conjugates (1 nmol in 100 µL PBS) were injected in triplicate subcutaneously in the dorsal skin between the shoulders or intravenously (1 nmol in 150 µL PBS) via the tail vein of C57BL/6 mice. For SC injections, mice were anesthetized with 2.5% isoflurane. Animals were kept under anesthesia for no longer than 5 min for each blood draw and no heating elements were used to reduce local variability in heating of the skin. Because injections were localized in the dorsal region, animals were oriented on their stomach during the injection and while under anesthesia. Animals were briefly oriented to the side during retro-orbital blood sampling but promptly reoriented after the blood draw. Previous experiments with inconsistencies in orientation, temperature, and isoflurane concentration resulted in inconsistent absorption rates and plasma concentrations (data not shown). For intravenous injections, blood samples were collected at 1, 3, 5, 15, 30, 60, 180, 300, and 1440 min and fluorescent peptide concentration quantified using a LICOR Odyssey CLx scanner (Lincoln, NE). SC injected peptides were quantified similarly through blood samples at pre-determined time points to capture absorption into the blood and plasma clearance. Area under curve (AUC) was calculated by applying the trapezoidal rule to a plot of plasma concentration as a function of time. For each fluorescent compound, the AUC ratio was taken between the SC injection and the intravenous injection to quantify the bioavailability. After 24 h, animals were sacrificed and the pancreas removed and macroscopically imaged using the Odyssey CLx to verify successful islet of Langerhans targeting. Plasma samples were then run on a reducing SDS-PAGE gel and scanned to confirm fluorescent signal from plasma corresponded to intact peptide rather than degraded byproducts (SI).