Development of Esterase-Resistant and Highly Active Ghrelin Analogs via Thiol–Ene Click Chemistry

Abstract

The orexigenic peptide ghrelin exerts important functions in energy metabolism and has therapeutic potential to treat certain diseases. Native ghrelin carries an essential O-fatty acyl moiety; however, this post-translational modification is susceptible to hydrolysis by certain esterases in circulation, representing a major route of its in vivo inactivation. In the present study, we developed a novel approach to prepare various esterase-resistant ghrelin analogs via photoinduced thiol–ene click chemistry. A recombinant unacylated human ghrelin mutant was reacted with commercially available terminal alkenes; thus, various alkyl moieties were introduced to the side chain of its unique Cys3 residue via a thioether bond. Among 11 S-alkylated ghrelin analogs, analog 11, generated by reacting with 2-methyl-1-octene, not only acquired much higher stability in serum but also retained full activity compared with native human ghrelin. Thus, the present study provided an efficient approach to prepare highly stable and highly active ghrelin analogs with therapeutic potential.

The orexigenic peptide ghrelin is an endogenous agonist of the growth hormone secretagogue receptor type 1a (GHSR1a), an A-class G protein-coupled receptor that was first identified as the receptor of certain synthetic growth hormone secretagogues, hence its name.^1,2 In vivo, mature ghrelin is derived from prepropeptide precursors after a series of post-translational processes, including a critical O-acylation,¹ in which a fatty acyl moiety, typically n-octanoyl, is covalently attached to the side chain of a Ser residue at the third position via an ester bond (Figure 1A). This special post-translational modification is catalyzed by ghrelin O-acyltransferase (GOAT), also known as membrane-bound O-acyltransferase domain containing 4 (MBOAT4).^3,4 Recent studies have demonstrated that liver-expressed antimicrobial peptide 2 (LEAP2 or LEAP-2) functions as a competitive antagonist of the ghrelin receptor GHSR1a in mammals and fish.⁵⁻¹¹ The ghrelin system plays important functions in energy metabolism and cellular homeostasis, and thus ghrelin and its analogs have therapeutic potential to treat certain diseases, such as anorexia, cancer cachexia, and growth hormone deficiency.¹²⁻¹⁵

Figure 1.

Schematic presentation of in vivo octanoylation of UAG by GOAT (A) and in vitro S-alkylation of [S3C]UAG by the photoinduced thiol–ene click chemistry (B). The amino acids in UAG and [S3C]UAG are shown as one-letter code, and the side chain of their third residue is shown.

Unacylated ghrelin (UAG), also known as des-acyl ghrelin (DAG), has no detectable binding with the ghrelin receptor GHSR1a, although some reports suggested it might have certain biological functions.¹⁶⁻¹⁸ Thus, the special O-fatty acyl modification is essential for ghrelin binding to, and activation of, its receptor GHSR1a. However, the O-fatty acyl moiety of ghrelin is susceptible to hydrolysis by certain esterases in circulation,¹⁹⁻²² such as butyryl cholinesterase, carboxylesterase, α₂ macroglobulin, and acyl-protein thioesterase 1 (also known as lysophospholipase 1). Hydrolysis of the fatty acyl moiety represents a major route of in vivo inactivation of ghrelin; therefore, the development of esterase-resistant ghrelin analogs with high activity is needed for the therapeutic application of ghrelin.

In a previous study,²³ a chemically synthesized ghrelin analog carrying an n-octyl moiety via a thioether bond displayed only moderately lower activity compared with native ghrelin, implying that S-alkylation might be a suitable approach to develop esterase-resistant ghrelin analogs. To conveniently prepare various S-alkylated ghrelin analogs, in the present study, we employed photoinduced thiol–ene click chemistry, which has been used in peptide modifications in recent years.²⁴⁻²⁶ For this purpose, we designed an analog of human UAG, designated as [S3C]UAG, by replacing Ser3 of human UAG with a Cys residue (Figure 1B). Thereafter, [S3C]UAG was reacted with some commercially available terminal alkenes under exposure to 365 nm UV light, and different alkyl moieties were introduced to the side chain of its unique Cys residue via a thioether bond (Figure 1B). Using this approach, we generated 11 S-alkylated ghrelin analogs and identified a novel analog with full activity as well as markedly higher stability compared with native human ghrelin. Thus, the present study provided an efficient approach to prepare novel highly stable and highly active ghrelin analogs with therapeutic potential.

Preparation of the S-Alkylated Ghrelin Analogs via the Photoinduced Thiol–Ene Click Chemistry

To prepare the small [S3C]UAG (28 amino acids) via bacterial overexpression, we designed a larger precursor by fusing [S3C]UAG to the C-terminus of the expected mature peptide (61 amino acids) of human C4ORF48 via a Met residue (Supporting Information Figure S1). As expected, the larger precursor could be efficiently overexpressed in Escherichia coli (E. coli) as inclusion bodies. To obtain the mature [S3C]UAG peptide from the recombinant precursor, we developed an efficient procedure to purify and process the precursor (Figure S2A,B). After the inclusion bodies were solubilized via an S-sulfonation approach, the S-sulfonated precursor was purified using an immobilized metal ion affinity chromatography and then subjected to CNBr cleavage. As monitored by HPLC (Figure S2C), the S-sulfonated [S3C]UAG could be quickly released from the precursor by chemical cleavage. Finally, the reversibly modified S-sulfonate moiety was removed via dithiothreitol treatment, and mature [S3C]UAG was obtained at a considerable yield. From 1 L of the E. coli culture broth, typically 5–10 mg of mature [S3C]UAG peptide could be obtained within a week.

To introduce various alkyl moieties into the mature [S3C]UAG, we optimized the reaction conditions of the photoinduced thiol–ene click chemistry.^27,28 We selected the hydrophilic 2-hydroxy-4′-(2-hydroxyethoxy)-2-methylpropiophenone as the photoinitiator, because its elution peaks were well separated from those of the S-alkylated ghrelin analogs on HPLC (Figure 2). To initiate the photoreaction, we used a 365 nm UV–LED lamp, which has a sharp spectrum around 365 nm and an adjustable power output. To minimize the oxidized byproducts, we conducted all photoinduced reactions in an AtmosBag filled with nitrogen gas.

Figure 2.

Preparation of the S-alkylated ghrelin analogs by photoinduced thiol–ene click chemistry. (A) HPLC analyses of the reaction mixture of [S3C]UAG reacting with 1-octene after UV light exposure for different times. The peak of [S3C]UAG is indicated by an octothorpe, and that of the S-octylated product is indicated by an asterisk. (B, C) HPLC analyses of the reaction mixture of [S3C]UAG reacting with various alkenes after UV light exposure for 45 min. The peak of the S-alkylated analogs is indicated by an asterisk. (D) Purity analyses of the S-alkylated ghrelin analogs by HPLC. In panels B–D, WT indicates the chemically synthesized native human ghrelin and numbers 1–11 indicate S-alkylated analogs listed in Tables 1 and 2

When [S3C]UAG was reacted with 1-octene, its elution peak (indicated by an octothorpe) on HPLC decreased quickly after exposure to UV light, meanwhile a new peak (indicated by an asterisk) appeared and increased correspondingly (Figure 2A). As measured by mass spectrometry (Table 1), the new peak was the expected S-octylated ghrelin analog. Estimated from their elution peaks, approximately 60% of [S3C]UAG could be converted to the S-octylated product (Figure 2A and Table 1).

Table 1. Summary of the Measured Molecular Masses and Yields of the S-Alkylated Ghrelin Analogs.

^aAverage molecular mass measured by electrospray mass spectrometry.

^bThe conversion yields of [S3C]UAG to S-alkylated analogs in the photoinduced thiol–ene reaction calculated according to the peak areas of the unreacted [S3C]UAG and the S-alkylated analogs when the reaction mixture was analyzed by HPLC as shown Figure 2.

When [S3C]UAG was reacted with other terminal alkenes (Table 1), S-alkylated ghrelin analogs were obtained with the yields ranging from ∼20% to ∼80% (Figure 2B,C and Table 1). We prepared 11 S-alkylated ghrelin analogs in total via the thiol–ene click chemistry, which all displayed the expected molecular mass as analyzed by mass spectrometry (Table 1) and a symmetrical elution peak as analyzed by HPLC (Figure 2D). Judged from their elution peaks, purity of these peptides was at least over 90%. Thus, it seemed that various S-alkylated ghrelin analogs could be conveniently prepared using the present approach.

Activity Measurement of the S-Alkylated Ghrelin Analogs

To measure activity of these S-alkylated ghrelin analogs, we employed a receptor binding assay and a receptor activation assay (Figure 3). We measured their binding activity with human GHSR1a via the NanoBiT-based homogeneous binding assay using two different tracers,^6,11,29,30 and measured their activation potency toward human GHSR1a via a CRE-controlled NanoLuc reporter.^6,11,29,30 Although GHSR1a signals through the Gq pathway, it also activates the transcription factor CREB through kinases such as Ca²⁺/calmodulin kinase IV and protein kinase C.³¹ Thus, a CRE-controlled luciferase reporter can be used to monitor GHSR1a activation.^6,11,29−32

Figure 3.

Activity assays of the S-alkylated ghrelin analogs. (A, D) Binding activity with human GHSR1a measured by the NanoBiT-based binding assay using ghrelin-SmBiT as a tracer. (B, E) Binding activity with human GHSR1a measured by the NanoBiT-based binding assay using LEAP2-SmBiT as a tracer. (C, F) Activation potency toward human GHSR1a measured by a CRE-controlled NanoLuc reporter. In this figure, all binding and activation data are expressed as mean ± SD (n = 3).

Compared to native human ghrelin, analog 1 carrying a short n-pentyl moiety displayed ∼15-fold lower receptor binding activity (Figure 3A,B and Table 2) and ∼7-fold lower receptor activation potency (Figure 3C and Table 2). When the chain length of the linear alkyl moiety was increased from n-pentyl (−C₅H₁₁) to n-nonyl (−C₉H₁₉), both the receptor binding activity and the receptor activation potency of analogs 2–5 increased gradually, with analog 5 carrying an n-nonyl moiety having full activity compared with native human ghrelin (Figure 3A–C and Table 2). However, introduction of a longer n-decyl (−C₁₀H₂₁) moiety caused a slight decrease in activity (Figure 3A–C and Table 2), suggesting that further increase of the chain length of the linear alkyl moiety likely has some detrimental effects.

Table 2. Summary of the Measured Activity Values of the S-Alkylated Ghrelin Analogs.

	IC50 (nM)a/pIC50b		EC50 (nM)a/pEC50b
ghrelin analog	ghrelin-SmBiT	LEAP2-SmBiT	NanoLuc reporter
WTc	2.19 ± 0.15	3.55 ± 0.34	1.95 ± 0.21
	(8.66 ± 0.03)b	(8.45 ± 0.04)b	(8.71 ± 0.05)b
1	33.9 ± 2.3	49.0 ± 4.3	13.8 ± 1.8
	(7.47 ± 0.03)	(7.31 ± 0.04)	(7.86 ± 0.06)
2	11.5 ± 0.8	20.4 ± 2.5	6.76 ± 0.83
	(7.94 ± 0.03)	(7.69 ± 0.05)	(8.17 ± 0.05)
3	3.90 ± 0.25	8.51 ± 0.57	3.31 ± 0.58
	(8.42 ± 0.03)	(8.07 ± 0.03)	(8.48 ± 0.07)
4	2.51 ± 0.17	4.27 ± 0.41	2.40 ± 0.31
	(8.60 ± 0.03)	(8.37 ± 0.04)	(8.63 ± 0.05)
5	1.91 ± 0.09	3.24 ± 0.22	1.58 ± 0.17
	(8.72 ± 0.02)	(8.49 ± 0.03)	(8.80 ± 0.05)
6	2.24 ± 0.15	4.57 ± 0.33	1.62 ± 0.24
	(8.65 ± 0.03)	(8.34 ± 0.03)	(8.79 ± 0.06)
7	6.03 ± 0.28	10.0 ± 0.9	6.61 ± 0.80
	(8.22 ± 0.02)	(8.00 ± 0.04)	(8.18 ± 0.05)
8	50.1 ± 3.3	70.8 ± 4.7	17.4 ± 1.6
	(7.30 ± 0.03)	(7.15 ± 0.03)	(7.76 ± 0.04)
9	1.58 ± 0.10	2.69 ± 0.18	1.38 ± 0.17
	(8.80 ± 0.03)	(8.57 ± 0.03)	(8.86 ± 0.05)
10	1.66 ± 0.11	2.19 ± 0.15	1.51 ± 0.16
	(8.78 ± 0.03)	(8.66 ± 0.03)	(8.82 ± 0.05)
11	0.93 ± 0.07	1.45 ± 0.10	1.00 ± 0.11
	(9.03 ± 0.03)	(8.84 ± 0.03)	(9.00 ± 0.05)

^aThe IC₅₀ and EC₅₀ values are converted from the pIC₅₀ and pEC₅₀ values and expressed as mean ± SD (n = 3).

^bThe pIC₅₀ and pEC₅₀ values are shown in parentheses, calculated from the binding and activation curves using the SigmaPlot software, and expressed as mean ± SD (n = 3).

^cThe chemically synthesized native human ghrelin.

When an aromatic phenyl moiety was introduced, the resultant analogs 7 and 8 displayed significantly lower activity compared with native human ghrelin (Figure 3D–F and Table 2), suggesting that the aromatic ring cannot be well accommodated in the ligand-binding pocket of GHSR1a. Analogs 9 and 11, carrying a branched alkyl moiety with a methyl group at the second carbon atom, displayed significantly higher activity than the corresponding analogs 3 and 4 carrying a linear alkyl moiety (Figure 3D–F and Table 2), suggesting that the methyl branch at the second carbon favors interaction of the analogs with receptor GHSR1a. However, introduction of a branch at the far end of the hydrocarbon chain seemingly had no effects, since analogs 9 and 10 displayed similar activity (Figure 3D–F and Table 2). Among these S-alkylated ghrelin analogs, analog 11 was most active in both the receptor binding assay and the receptor activation assay.

Stability of the S-Alkylated Ghrelin Analogs

As shown above, some S-alkylated ghrelin analogs, such as 11 and 9, were highly active. Next, we wanted to know whether their stability was significantly increased, as expected. To compare their stability with that of native human ghrelin, we incubated them in human serum or fetal bovine serum and then quantified the remaining active peptide via a washing-based ligand–receptor binding assay using NanoLuc-conjugated ghrelin as a bioluminescent tracer (Supporting Information).³³ In the presence of 45% human serum or fetal bovine serum in the binding assay, the standard competition binding curves were typically sigmoidal for these peptides (Figure 4A,B), suggesting that this washing-based binding assay is resistant to interference from a high percentage of serum. In the presence of human serum or fetal bovine serum, analog 11, analog 9, and native human ghrelin displayed similar IC₅₀ values (Table 3), confirming that both analogs are highly active.

Figure 4.

Stability assay of native human ghrelin and some S-alkylated analogs in serum. (A, B) Stability in human serum (A) or fetal bovine serum (B) measured using the washing-based receptor-binding assay. All binding data are expressed as mean ± SD (n = 3). (C, D) Stability of native human ghrelin (C) and analog 9 (D) in human serum measured by HPLC. Trace 1, human serum alone; trace 2, native human ghrelin or analog 9 plus human serum, but without coincubation; trace 3, native human ghrelin or analog 9 plus human serum and coincubation at 37 °C for 15 h; trace 4, synthetic UAG plus human serum, but without coincubation. The elution peak of native human ghrelin or analog 9 is indicated by an asterisk, and that of UAG is indicated by an octothorpe.

Table 3. Summary of Stability of Native Human Ghrelin and Some S-Alkylated Analogs in Human Serum or Fetal Bovine Serum Measured by the Washing-Based Receptor Binding Assay.

	assayed in human serum		assayed in fetal bovine serum
peptide	IC50 (nM)a/pIC50b	left after incubation (%)	IC50a (nM)/pIC50b	left after incubation (%)
WTc	11.7 ± 1.2	∼10	12.3 ± 1.1	∼5
	(7.93 ± 0.05)b		(7.91 ± 0.04)b
analog 9	12.0 ± 1.3	∼70	10.7 ± 1.3	∼40
	(7.92 ± 0.05)		(7.97 ± 0.05)
analog 11	10.5 ± 1.2	∼75	9.1 ± 0.9	∼45
	(7.98 ± 0.05)		(8.04 ± 0.04)

^aThe IC₅₀ values are converted from the pIC₅₀ values and expressed as mean ± SD (n = 3).

^bThe pIC₅₀ values are shown in parentheses, calculated from the binding curves using the SigmaPlot software, and expressed as mean ± SD (n = 3).

^cThe chemically synthesized native human ghrelin.

After native human ghrelin was incubated in 90% human serum at 37 °C for 15 h, its measured binding values were far from the standard binding curve of the native human ghrelin (Figure 4A), suggesting that incubation with human serum caused significant inactivation of native human ghrelin. Calculated according to the standard binding curve, only ∼10% of human ghrelin remained active after incubation with 90% human serum at 37 °C for 15 h (Table 3). In contrast, incubation with human serum had much less effect on analog 11 and analog 9 (Figure 4A): ∼70% of the analogs remained active after incubation with 90% human serum at 37 °C for 15 h (Table 3). A similar phenomenon was observed for these peptides after incubation with fetal bovine serum (Figure 4B and Table 3). Thus, analogs 11 and 9 acquired much higher stability in both human serum and fetal bovine serum compared with native human ghrelin, suggesting that the S-alkylated analogs would have a much longer in vivo half-life than native ghrelin when used as a therapeutic reagent in future studies.

To confirm above results, we analyzed these peptides by HPLC after incubation with human serum (Figure 4C,D). After 225 μL of human serum was subjected to sequential incubation at 37 °C for 15 h, precipitation by acetonitrile, and HPLC analysis, a lot of high peaks were eluted from an analytical C₁₈ reverse-phase column (Figure 4C,D, trace 1), suggesting that serum contains a lot of peptides or other substances that cannot be removed by organic solvent precipitation. These serum-derived peaks were not overlapped with the elution peak of native human ghrelin and analog 9 (Figure 4C,D, trace 2), but they completely masked the elution peak of analog 11. Thus, the stability of analog 11 could not be analyzed by the HPLC approach, but it was reasonable to deduce that its property is similar to that of analog 9, since both of them carry a branched hydrocarbon chain via a thioether bond.

After 20 μM native human ghrelin was incubated in 90% human serum at 37 °C for 15 h, the peak of intact human ghrelin (indicated by an asterisk) was drastically decreased (Figure 4C, trace 3). Meanwhile, there appeared a new peak (indicated by an octothorpe) whose retention time was identical with that of the synthetic UAG (Figure 4C, trace 4). Thus, it seemed that most of native human ghrelin was converted to UAG via hydrolysis of its fatty acyl moiety by certain esterases after incubation in human serum. In contrast, the peak height of intact analog 9 (indicated by an asterisk) was not changed much after incubation in human serum (Figure 4D, trace 3), implying that its thioether bond is stable in serum. In summary, HPLC analysis confirmed that the S-alkylated ghrelin analogs are much more stable than native ghrelin in serum, since their thioether bond is resistant to enzymatic hydrolysis.

In the present study, we developed an efficient approach to prepare highly stable and highly active ghrelin analogs by S-alkylation of [S3C]UAG with various terminal alkenes via the photoinduced thiol–ene click chemistry. In future studies, more S-alkylated ghrelin analogs could be prepared via this approach. To further improve the activity and stability of the S-alkylated analogs, two approaches might be used in future studies. On the one hand, some mutations might be introduced to other positions of the [S3C]UAG peptide to make the peptide chain more resistant to proteases in circulation. Thereafter, the designed analogs might be prepared either by bacterial overexpression or by solid-phase peptide synthesis. On the other hand, more terminal alkenes with various structures might be used to modify [S3C]UAG or its analogs. Thus, our present study provided a practical approach to prepare various S-alkylated ghrelin analogs with high activity and high stability.

In the present study, analog 11, generated by reacting with 2-methyl-1-octene, retained full activity compared to native human ghrelin. Thus, it seemed that the branched methyl group at the second position of the modified alkyl moiety favors the activity of the S-alkylated analog. In future studies, the effect of a larger branched group, such as an ethyl group, might be tested using the corresponding alkenes. After a terminal alkene with a branch at the second position, such as 2-methyl-1-octene and 2-methyl-1-heptene, was reacted with [S3C]UAG, its second carbon will be converted to a chiral atom with an R-configuration or an S-configuration. Thus, the reaction theoretically produces two S-alkylated analogs carrying an alkyl moiety either with an R-configuration or with an S-configuration at the second carbon atom. However, only one elution peak was identified as the modification product when [S3C]UAG was reacted with these branched alkenes, suggesting that the two expected products could not be separated on HPLC. Thus, it is unclear whether the two products have similar activity or not. In future studies, the corresponding analogs might be chemically synthesized in order to test their activity.

Glossary

Abbreviations

CRE

cAMP-response element

DAG

des-acyl ghrelin

GHSR1a

growth hormone secretagogue receptor type 1a

GOAT

ghrelin O-acyltransferase

HPLC

high-performance liquid chromatography

LEAP2

liver-expressed antimicrobial peptide 2

LED

light emitting diode

MBOAT4

membrane-bound O-acyltransferase domain containing 4

NanoBiT

NanoLuc Binary Technology

SD

standard deviation

sLgBiT

secretory large NanoLuc fragment for NanoBiT

SmBiT

low-affinity complementation tag for NanoBiT

UAG

unacylated ghrelin

UV

ultraviolet

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.2c00339.

• Experimental procedures; (Figure S1) Nucleotide sequence and amino acid sequence of 6 × His-C4ORF48-[S3C]UAG fusion protein overexpressed in E. coli; (Figure S2) preparation of [S3C]UAG by overexpression of a larger precursor in E. coli and subsequent CNBr cleavage (PDF)

Author Contributions

H.-Z.L., X.-X.S., L.-L.S., and N.L. conducted the experiments; Y.-L.L. and Z.-G.X. analyzed the data; Z.-Y.G. conceived and designed the research, and wrote the manuscript.

The authors declare no competing financial interest.