Ligand Design with Accelerated Disulfide Formation with Serum Albumin to Extend Blood Retention

Abstract

We proposed a novel ligand for the interaction with human serum albumin (HSA) to extend the blood half-life of small molecular weight therapeutics. The ligand features an alkyl chain and an activated disulfide to allow binding to the hydrophobic pockets of HSA and the formation of disulfide to Cys34 of HSA, thereby minimizing the initial renal clearance. The dual nature of the ligand–HSA bonding was expected to give the ligand long blood retention. After 1 min of mixing with HSA, the ligand showed higher binding (1.7 times) than that of a control ligand (containing only activated disulfide). After intravenous injection to mice, the ligand half-lives were 1.6 and 9.2 times longer than those of control ligands with the active disulfide alone and with the alkyl chain alone, respectively. The proposed ligand has the potential to act as a platform for extending the half-life of small therapeutics in vivo.

One of the most important challenges in the use of small therapeutics (<70 kDa) is their short residence time in the blood because of fast renal clearance.¹ The interaction of small therapeutics with human serum albumin (HSA) has been used to extend the blood retention of these compounds because of the abundance of HSA in the blood (∼600 μM). HSA also has a long half-life of more than 20 days² because its ability to bind with the neonatal fragment crystallizable receptor (FcRn) allows it to be recycled through the endosomal pathway rather than being broken down.³ To bind therapeutics with HSA in blood, hydrophobic interactions between an alkyl chain and the hydrophobic pockets of HSA have been used.^4,5 Analogs of insulin and glucagon-like peptide-1 (GLP-1), which are applied to the treatment of diabetes, utilize this strategy to extend their blood half-life.^4,6 Despite the high affinity of the alkyl chain for HSA, the half-life extension achieved by this reversible hydrophobic interaction remains limited. In contrast, covalent binding with Cys34 of HSA has the potential to extend the half-life considerably.

Covalent binding of maleimide-functionalized therapeutics with Cys34 was used to extend the half-life of anticancer agents.⁷ Disulfide bond formation with Cys34 was used with GLP-1, which contains activated disulfide, to form a disulfide exchange reaction with Cys34.⁸ However, these strategies targeting Cys34 with thiol-reactive groups may have a relatively slow rate of reaction with Cys34 of HSA in the blood, leading to an undesired initial renal clearance.

In this study, we proposed a novel ligand design to facilitate the formation of disulfide bonds with Cys34 (Figure 1). The ligand is comprised of an alkyl chain and an active disulfide group. First, the alkyl chain rapidly binds to the hydrophobic pocket of HSA in blood, which minimizes the initial renal clearance of the ligand. The binding of the ligand with HSA accelerates the disulfide exchange reaction because of the preconcentration of the active disulfide around Cys34. The resulting covalent complex between the ligand and HSA is stabilized by both hydrophobic interaction and the disulfide bond, leading to a blood half-life that is longer than those of ligands that use a single binding interaction. Figure 2 shows the structure and expected interaction of designed ligand 1 with HSA. 2-Mercaptopyridine was selected as the leaving thiol because of its effective leaving nature in the disulfide exchange reaction. Palmitic acid (C15 alkyl chain) was selected as the hydrophobic entity. In tests of hydrophobic binding of ligands with HSA,⁹ palmitic acid was observed to show higher binding efficiency than myristic acid (C13) or stearic acid (C17). Of the seven pockets of HSA, pockets 2 and 5 can bind the palmitic acid moiety under physiological conditions.¹⁰ We targeted pocket 2 because it is the closest pocket to Cys34. In the case of ligand binding to pocket 5, the connection of the pocket and Cys34 with a linking group would disturb the binding of HSA with FcRn because of overlap with the FcRn binding region of HSA.¹¹ To avoid any obstruction of the area between pocket 5 and Cys34 by the ligand, the length of the linking moiety between the activated disulfide and the alkyl-modified lysine was designed to match the distance between pocket 2 and Cys34 (Figure S1). An oligoethylene glycol linker was chosen to match the distance between pocket 2 and Cys34.

Figure 1.

Proposed mechanism of accelerated disulfide formation of the ligand with Cys34 of human serum albumin (HSA) via an alkyl chain (red rectangle) binding to the hydrophobic pocket.

Figure 2.

Molecule design of the ligand and its expected interaction with HSA.

Figure 3 shows the structures of the fluorescence-modified ligand 1 and control ligands 2 and 3. First, we evaluated the rate of disulfide bond formation between carboxyfluorescein-modified ligands 1a–3a (100 μM) and fatty acid free HSA (Sigma A3782) (500 μM) in phosphate-buffered saline (pH 7.4) at 37 °C. During the incubation of each ligand and HSA, the disulfide exchange reaction was quenched at designated times by the addition of excess N-ethylmaleimide (NEM).¹² The resulting complexes (HSA-1a and HSA-2a) were isolated by reversed-phase high-performance liquid chromatography (RP-HPLC). Figure 4a,b shows the chromatograms after 1 min of reaction. The complex peaks, which were identified by MALDI-TOF-MS (Figure S5), were detected as positive peaks with both FAM fluorescence and protein absorption (280 nm) detection. Evidence of the formation of small amounts of homodimers (1a–1a, 2a–2a) was also observed (Figure S4). To quantify the ligand that formed a disulfide bond with HSA, the complexes fractionated by RP-HPLC were treated with tris(2-carboxyethyl)phosphine (TCEP)¹³ to cleave the disulfide bond, and the released ligand was quantified by RP-HPLC. Figure 4c shows the time course of complex formation. In the first 1 min, ∼36 μM of 1a bound to HSA (36% of 1a bound to HSA based on the initial concentration of the ligand 100 μM), which is 1.7 times more than that of 2a (∼19 μM). The rapid disulfide bond formation of 1a (faster than 2a) is a reflection of the molecular design of 1a; that is, the rapid initial binding with the alkyl chain allows accelerated formation of the disulfide linkage with Cys34. During the 60 min of reaction, 1a always showed a higher amount of complex than 2a.

Figure 3.

Chemical structures of ligands 1–3. For the a series, R = carboxyfluorescein; for the b series, R = sulfo-cyanine7 (Cy7).

Figure 4.

Detection of covalent complexes after 1 min of reaction with ligand 1a (a) and 2a (b), and the time course of ligands 1a and 2a binding to HSA (c). HSA (500 μM) and ligand (100 μM) were incubated in PBS at 37 °C, followed by quenching with 10 mM NEM. The HSA–ligand complex was isolated by HPLC, reduced with 40 mM TCEP to obtain a free ligand for quantification by HPLC. Data plotted in panel (c) are mean ± standard deviation (n = 3) peaks marked with an asterisk are NEM and its derivatives.

As the natural carrier of fatty acids in blood, circulating HSA carries between 0.3 and 1 fatty acid per albumin molecule, rarely exceeding 2.¹⁴ To investigate the effect of prebound fatty acid in the disulfide formation with ligands 1a and 2a, we used HSA isolated from human blood (Sigma A1653) with an average fatty acid content of approximately 1.99 per HSA molecule.¹⁵ There was no significant difference in the amount of covalent complex formation of 1a and 2a between fatty acid-free and fatty acid-bound HSA after 30 min of the reaction (Table S1). Bound fatty acids to HSA were reported to dissociate from HSA in around ten seconds.¹⁶ It will be the reason why there is no effect of prebound fatty acid on accelerated disulfide formation of 1a with HSA than that of 2a. These results indicate that accelerated disulfide formation of 1a with HSA bound to fatty acids would occur in the blood.

Previous studies have shown that the disulfide bonds formed at Cys34 can be exchanged with low molecular weight thiols in the blood to release thiol compounds,¹⁷ suggesting that our system also possesses sustained release properties. We observed that the addition of 10 mM l-cysteine to the solution after the reaction of HSA with 1a for 30 min caused the release of 1a from the complex (Figure S3).

We tested the blood half-life of sulfo-cyanine7 (Cy7)-modified ligands 1b–3b in mice. After intravenous administration, ligands in the collected plasma were quantified using a plate reader. Figure 5 shows the decline of the ligand concentration in blood over time. The decay curves were analyzed by a two-compartment model, and calculated parameters are summarized in Table 1. Ligands 1b and 3b showed initial plasma concentrations higher than those of 2b (1.5-fold), which was attributed to the rapid interaction of their alkyl group with mouse serum albumin in the blood. Ligand 1b had the longest half-life, which was 9.2-fold longer than that of 3b and 1.6-fold longer than that of 2b. When compared with 3b, the longer half-lives of 1b and 2b suggest that conjugation with HSA via disulfide bonds is more effective in prolonging the half-life than hydrophobic interaction with the pocket. In a comparison of the half-lives of 1b and 2b, the longer half-life of 1b is attributed to the facile formation of the disulfide of 1b and the slower release of 1b from HSA because of the combined bonding effect of hydrophobic interaction and the disulfide linkage. The release of 1b in blood is expected to occur via the disulfide exchange reaction with endogenous low molecular weight thiols because the half-life of 1b in mice (7.62 ± 0.49 h) is significantly shorter than that of mouse serum albumin (28.8 h).¹⁸

Figure 5.

Time course of blood retention of Cy7-modified ligands in mice after an intravenous injection of 17.1 nmol of ligand per mouse (1.18 mg/kg, 1b; 0.93 mg/kg, 2b; and 1.03 mg/kg, 3b). Plotted data given as mean ± standard deviation (n = 3).

Table 1. Parameters of the Pharmacokinetic Study of Ligands in Mice^a.

	1b	2b	3b
t1/2 α (h)	1.51 ± 0.11	1.26 ± 0.15	n.d.
t1/2 β (h)	7.62 ± 0.49	4.73 ± 0.24	0.83 ± 0.02
AUCt (mg h L–1)	145.94 ± 3.73	50.48 ± 0.99	30.85 ± 3.47
CL (mL h–1 kg–1)	8.09 ± 0.21	18.43 ± 0.36	33.80 ± 3.64
VD (mL kg–1)	28.43 ± 1.09	41.26 ± 1.33	42.73 ± 4.23

^aAUC, area under curve; CL, clearance; n.d., not determined; V_D, volume of distribution; n = 3 per peptide. Data given as mean ± standard deviation.

Our study of novel ligand compounds, we found that ligands containing an alkyl chain and active disulfide rapidly formed disulfide links with Cys34 of HSA, leading to suppression of the initial renal clearance. Ligands that were bound to HSA via hydrophobic interactions and disulfide bonding showed longer retention in the blood than ligands bound through one of these interactions alone. The ligand bound through dual interactions could be released from HSA by the addition of a thiol compound, which displaced the ligand via a disulfide exchange reaction. The ligand proposed in this work has the potential to improve the bioavailability and blood retention of low molecular weight therapeutics during systemic circulation. The ligand would be applicable to the delivery of GLP-1 and insulin for the treatment of type 2 diabetes and FXIIa inhibitors for antithrombotic therapy.⁹

Glossary

Abbreviations

AUC

area under curve

CL

clearance

Cy7

sulfo-cyanine7

FcRn

fragment crystallizable receptor

GLP-1

glucagon-like peptide-1

HSA

human serum albumin

MALDI-TOF-MS

matrix-assisted laser desorption/ionization time-of-flight mass spectrometry

n.d.

not determined

NEM

N-ethylmaleimide

RP-HPLC

reverse phase high performance liquid chromatography

TCEP

tris-2-carboxyethyl phosphine

V_D

volume of distribution

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.4c00503.

• Materials, molecular docking, ligand and HSA binding assay by RP-HPLC and MALDI-TOF-MS data (PDF)

Author Contributions

^# S.Q. and Z.X.L. contributed equally to this work. S.Q.: conceptualization, methodology, investigation, writing–original draft, and visualization. Z.X.L.: investigation, writing–original draft, and visualization. K.S. and T.N.: resources. Y.T., I.I., and T.H.: formal analysis. N.Y. and A.S.: software usage and investigation. J.C., K.Q.P., T.M., and A.K.: writing–review and editing. T.M.: conceptualization, writing–review and editing, and funding acquisition. Y.K.: writing–review and editing and supervision.

This research was supported by AMED under Grant Number 22ym0126816j0001 and the Center for Clinical and Translational Research of Kyushu University Hospital

The authors declare no competing financial interest.