Z.-Y. Guo, Institute of Protein Research, College of Life Sciences and Technology, Tongji University, Shanghai 200092, China Fax: +86 21 6598 8403 Tel: +86 21 6598 8634 E-mail: firstname.lastname@example.org Z.-G. Xu; Y.-L. Liu Central Laboratory, East Hospital, Tongji University School of Medicine, Shanghai 200120, China Fax: +86 21 5879 8999 Tel: +86 21 3880 4518 E-mail: email@example.com; firstname.lastname@example.org
Hepcidin is a small disulfide-rich peptide hormone that plays a key role in the regulation of iron homeostasis by binding and mediating the degradation of the cell membrane iron efflux transporter, ferroportin. Since it is a small peptide, chemical synthesis is a suitable approach for the preparation of mature human hepcidin. However, oxidative folding of synthetic hepcidin is extremely difficult due to its high cysteine content and high aggregation propensity. To improve its oxidative folding efficiency, we propose a reversible S-modification approach. Introduction of eight negatively charged sulfonate moieties into synthetic hepcidin significantly decreased its aggregation propensity and, under optimized conditions, dramatically increased the refolding yield. The folded hepcidin displayed a typical disulfide-constrained β-sheet structure and could induce internalization of enhanced green fluorescent protein (EGFP) tagged ferroportin in transfected HEK293 cells. In order to study interactions between hepcidin and its receptor ferroportin, we propose a general approach for site-specific labeling of synthetic hepcidin analogues by incorporation of an l-propargylglycine during chemical synthesis. Following efficient oxidative refolding, a hepcidin analogue with Met20 replaced by l-propargylglycine was efficiently mono-labeled by a red fluorescent dye through click chemistry. The labeled hepcidin was internalized into the transfected cells together with the EGFP-tagged ferroportin, suggesting direct binding between hepcidin and ferroportin. The labeled hepcidin was also a suitable tool to visualize internalization of overexpressed or even endogenously expressed ferroportin without tags. We anticipate that the present refolding and labeling approaches could also be used for other synthetic peptides.
Hepcidin (also known as LEAP-1) is a liver-secreted small disulfide-rich peptide hormone that plays a central role in the regulation of iron homeostasis [1–3]. It was first identified in 2000 as a defensin-like antimicrobial peptide [4,5]. Soon after its discovery, the key regulatory role of hepcidin in iron homeostasis was revealed [6–9]. Hepcidin binds to the cell membrane iron transporter ferroportin (Fpn) at an evolutionarily conserved region, resulting in Fpn internalization and degradation, therefore inhibiting iron efflux [10–16]. Hepcidin is synthesized in vivo as a precursor comprising an N-terminal signal peptide, a pro-peptide and a mature peptide. After a series of in vivo processing, mature human hepcidin, composed of 25 amino acids and four disulfide linkages, is released and secreted. The complex disulfide linkages of hepcidin have recently been elucidated using chemical modification and from studying its crystal structure .
Due to its small size, chemical peptide synthesis is a suitable approach for the preparation of mature human hepcidin. Unfortunately, however, oxidative folding of synthetic hepcidin is extremely difficult, since it has a high cysteine content (∼ 30%) and a propensity for high aggregation [17,18]. Currently, the optimal reported oxidative folding yield is only 6% in solution and 12% when assisted by Clear-OX resin . Some researchers have attempted to prepare recombinant hepcidin through expression in Escherichia coli or yeast but have encountered difficulties in obtaining active mature hepcidin in a considerable yield [19–21]. Furthermore, active hepcidin analogues bearing various functional probes will greatly facilitate mechanistic studies of hepcidin–Fpn interactions. Therefore, in this study we have established efficient approaches for both oxidative folding and site-specific labeling of hepcidin and its analogues.
Results and Discussion
Reversible S-modification and purification of synthetic human hepcidin
Previous reports showed that synthetic linear hepcidin is prone to aggregation [17,18]. In this work, we have also observed that crude synthetic hepcidin powder was difficult to dissolve in various solvents, including trifluoroacetic acid (TFA), acetonitrile, acetic acid and 6 m guanidine chloride solution. We therefore sought to identify an approach that would increase the solubility of the crude peptide. S-sulfonation is known to introduce negatively charged sulfonate moieties () onto a peptide chain containing cysteine residues [22,23], resulting in increased peptide solubility in neutral or basic solutions. Since this modification is reversible, it is compatible with subsequent oxidative folding. The linear hepcidin has eight cysteine residues and therefore carries eight sulfonate moieties after S-sulfonation (Fig. 1A). As expected, the turbid hepcidin suspension (10 mg·mL−1 of crude peptide) became clear following overnight reaction, indicating that the S-sulfonated peptide had a higher solubility due to the presence of eight negatively charged sulfonate moieties. When the S-sulfonated peptide was applied to RP-HPLC and eluted with an acidic acetonitrile gradient, no peak was observed. This was probably caused by the low solubility of the S-sulfonated hepcidin in the acidic elution solvent since the highly negatively charged S-sulfonated peptide had a low isoelectric point. However, with a neutral acetonitrile gradient, a sharp peak (indicated with a star) was eluted from the C18 reverse-phase column (Fig. 2). The measured molecular mass (3433.2 Da) of the peak was consistent with the expected value (3437.4 Da) for the fully S-sulfonated hepcidin. Using a standard curve of purified, folded hepcidin, the linear hepcidin content in the crude peptide powder was around 15–20% estimated from the peak area of the S-sulfonated hepcidin (shaded part in Fig. 2). The S-sulfonated hepcidin aggregated after lyophilization, so the eluted S-sulfonated hepcidin fraction was directly subjected to oxidative folding as described below.
Oxidative folding of the S-sulfonated human hepcidin
Previous reports have illustrated that the oxidative folding yield of synthetic hepcidin is extremely low . We deduced that its high aggregation propensity was a major cause for its low folding yield. Since S-sulfonation significantly decreases the aggregation propensity of the linear peptide, we folded the S-sulfonated hepcidin directly in an oxidative redox solution. The oxidative redox potential was able to catalyze disulfide reshuffling to form intramolecular disulfide bonds in a stepwise manner (Fig. 1A), enabling the sulfonate moieties to be spontaneously removed from the peptide chain. As shown in Fig. 3A, the S-sulfonated hepcidin prepared using a semi-preparative C18 column was homogeneous as analyzed by analytical RP-HPLC. In this study we used this S-sulfonated hepcidin to carry out oxidative folding through disulfide reshuffling. In Tris/HCl solution, the folding yield was quite low. However, this was significantly increased in l-arginine solution which has been previously reported to improve folding efficiency through the inhibition of aggregation [24,25]. A major peak appeared following oxidative folding in l-arginine solution, as shown in Fig. 3B. Interestingly, the folding yield at 37 °C was higher than at 25 °C, although further temperature increase had little effect. The acetonitrile concentration also significantly affected the oxidative folding, as shown in Fig. 3C: 30% acetonitrile resulted in a higher yield than 20%, although a further increase in acetonitrile concentration had little effect. Under optimal folding conditions [0.5 m l-arginine-HCl, 1.0 mm EDTA, pH 7.5, 30% acetonitrile, 0.5 mm reduced glutathione (GSH), 0.5 mm oxidized glutathione (GSSG), ∼ 0.1 mg·mL−1 linear peptide, at 37 °C for 4 h], the folding yield could reach ∼ 80%, as estimated from the peak area before and after refolding. In sample preparations (Fig. 4A), typically ∼ 1.3 mg of lyophilized folded hepcidin could be obtained from 10 mg of crude synthetic hepcidin powder. As analyzed above, the linear hepcidin content in the crude powder was 15–20%, suggesting that the estimated ∼ 80% folding yield was a reasonable conclusion. Using the crude peptide as a starting material, the overall yield of the purified folded hepcidin could reach ∼ 13% through the present two-step procedure. During oxidative folding, we also tried to treat the S-sulfonated hepcidin with dithiothreitol to remove the sulfonate moieties and then oxidatively refold the fully reduced peptide by adding oxidant GSSG; however, the folding yield was significantly lower than that of direct oxidative folding, probably due to higher aggregation propensity of the fully reduced hepcidin. Thus we thought that both S-sulfonation and the arginine solution contributed to the efficient oxidative folding of hepcidin by inhibition of aggregation.
Analytical RP-HPLC analysis showed that our folded hepcidin was homogeneous (Fig. 4B). Our folded hepcidin and a human hepcidin sample purchased from Bachem (Heidelberg, Germany) were co-eluted on analytical RP-HPLC, suggesting they had identical disulfide linkages. As shown in Fig. 4C, the measured molecular mass (2790.0 Da) of the folded human hepcidin was consistent with the theoretical value (2789.4 Da), suggesting the formation of four intramolecular disulfide bonds. We have therefore established an efficient approach for the oxidative folding of synthetic hepcidin. In previous methods, the oxidative folding of the synthetic hepcidin was typically carried out in solution containing a denaturant, such as guanidine chloride, in order to prevent peptide aggregation. However, denaturants disturb native protein structure and may lead to the formation of isomers with non-native disulfide linkages. Using our approach, oxidative folding is carried out in denaturant-free solution, which favors the formation of native disulfide linkages. The folding temperature, pH value and redox potential in this approach were also similar to in vivo conditions. Hepcidin contains four disulfide bonds; however, the role of these disulfide bonds for hepcidin activity is still controversial. In some reports, hepcidin analogues with pairwise disulfide deletion all retained quite high activity (over 50% of wild-type hepcidin) [26,27], while in another paper these analogues showed a significant activity decrease (∼ 10-fold decrease) . Thus, the role of each disulfide bond for hepcidin activity needs to be further studied in the future using more sensitive and more reliable activity assays.
The conformation of folded human hepcidin in solution was studied using circular dichroism (Fig. 5). Similar to previously reported data , the spectrum of folded hepcidin had a maximum negative peak around 200–205 nm, suggesting that the peptide was dominated by a disulfide-stabilized β-sheet conformation. As estimated from the spectrum, the folded hepcidin was composed of ∼ 60%β-sheet and ∼ 40% random structure.
A general approach for site-specific labeling of synthetic hepcidin analogues
Active hepcidin analogues labeled with functional probes may facilitate the study of hepcidin–Fpn interactions. In this study, we established a general approach for site-specific labeling of synthetic hepcidin analogues with various functional probes. During chemical synthesis, an l-propargylglycine bearing an alkyne moiety on its side-chain was incorporated into an appropriate position of a designed hepcidin analogue (Fig. 1B). After oxidative refolding using the S-sulfonation approach, the folded hepcidin analogue could be site-specifically labeled using click chemistry by reacting with a functional probe bearing an azido moiety (Fig. 1B). As shown in Fig. 6A, a synthetic hepcidin analogue in which Met20 was replaced by l-propargylglycine was refolded well using the S-sulfonation approach, with mass spectrometry analysis showing the correct molecular mass (measured value 2754.0 Da, theoretical value 2753.3 Da). Met20 is variable among hepcidins from different species. After it was replaced by a tyrosine, the resultant hepcidin analogue retained high activity and could be efficiently labeled by radionuclide iodine-125 . Thus we expected that its replacement with l-propargylglycine had minimal disturbance on both oxidative folding and activity of the resultant hepcidin analogue. As shown in Fig. 6B, a sharp peak (indicated by a star) with strong absorbance at 280 nm appeared after reaction with excess modification reagent, a red fluorescent dye carrying an azido moiety (Fig. 1B). The measured molecular mass (3413.0 Da) of this peak was consistent with the expected value (3413.1 Da) of the fluorescent-dye-labeled analogue. Therefore, a homogeneous fluorescent-dye-labeled hepcidin analogue was obtained using click-chemistry-based site-specific labeling. Other functional probes, such as biotin or metal ion chelators, could also be introduced into this or other hepcidin analogues via this approach.
Hepcidin-induced Fpn internalization
The activity of folded human hepcidin prepared using the S-sulfonation approach was assessed by the measurement of its ability to induce enhanced green fluorescent protein (EGFP) tagged Fpn internalization in transfected HEK293 cells (Fig. 7A). The EGFP-tagged Fpn was expressed on the cell membrane as previously reported (panel a). After treatment with either purchased hepcidin (from Bachem) or by our hepcidin, EGFP-tagged Fpn internalization was observed according to the green fluorescence of tagged EGFP (panels b and c), suggesting that our hepcidin was active. To confirm that the Fpn internalization was caused by direct binding with hepcidin, the red-fluorescent-dye-labeled hepcidin analogue was used to treat transfected HEK293 cells (Fig. 7B). The labeled hepcidin analogue could also induce internalization of the EGFP-tagged Fpn according to the green fluorescence of the tagged EGFP (panel a), suggesting that the fluorescent-dye-labeled analogue was also active. The labeled hepcidin was also internalized into the transfected cells according to the red fluorescence of the labeled hepcidin (panel b). The internalized green fluorescent dots and the internalized red fluorescent dots were well colocalized (panel c), suggesting that hepcidin directly bound with Fpn and induced its internalization. Since the fluorescent-dye-labeled hepcidin could be internalized into cells together with Fpn, we used the labeled hepcidin to visualize the internalization of untagged Fpn in transfected HEK293 cells (Fig. 7C). After treatment by the labeled hepcidin, internalized red dots were observed in some cells (presumably transfected cells), while no red dots were found in other cells (presumably non-transfected cells). For some cells, their membrane was also stained red by the labeled hepcidin probably because these cells expressed more Fpn at the cell membrane. Next, we used the labeled hepcidin to treat the isolated mice macrophages that express endogenous Fpn. As shown in Fig. 7D, internalized red dots were observed in some cells (presumably macrophages), while no red dots were observed in other cells, suggesting that internalization of the fluorescent-dye-labeled hepcidin was most probably mediated by cell membrane Fpn rather than by non-specific endocytosis. So, the fluorescent-dye-labeled hepcidin is a valuable tool to study hepcidin–Fpn interactions in future work.
Reversible S-modification strategy for efficient oxidative folding of other disulfide-rich peptides with high aggregation propensities
There are many natural small disulfide-rich peptides, such as conotoxins  and cyclotides . Currently, these small peptides are prepared using solid-phase peptide synthesis, which is a practical option since it is fast, cheap and compatible with unnatural amino acid incorporation. After chemical synthesis, an oxidative folding step is required to form the correct disulfide linkages of these synthetic peptides. The oxidative folding yields of different peptides vary significantly. In general, there are two major factors that affect the oxidative folding yield: foldability  and aggregation propensity . Foldability refers to the ability to form the correct disulfide linkages during oxidative folding, while aggregation propensity means the tendency to form aggregates during oxidative folding. Through prevention of aggregation, the oxidative folding yield can be increased significantly. A commonly used strategy is the addition of soluble additives (such as l-arginine) [24,25] or insoluble porous resins (such as Clear-OX resin) . In this study we present an alternative approach for aggregation prevention involving reversible S-modification of the peptide chain. Apart from the sulfonate moiety, other small hydrophilic moieties such as cysteamine () may be applicable for this strategy. The cysteamine moieties can be introduced into the peptide chain by modification of the linear peptide with an excess of cystamine. The cysteamine moiety is highly hydrophilic and positively charged in neutral or acidic solutions, and we would expect that it will play a similar role to the sulfonate moiety in improving the oxidative folding efficiency of some disulfide-rich peptides with high aggregation propensities.
Site-specific labeling of other synthetic peptides using click chemistry
The introduction of a functional probe can facilitate the study of peptide function, such as ligand–receptor interactions and in vivo trafficking. In recent years, click chemistry based on cycloaddition between an azido moiety and an alkyne moiety has been widely used for labeling of biomolecules due to its high selectivity [33,34]. In this study, we obtained an active hepcidin analogue by replacing residue Met20 with l-propargylglycine. The analogue could be refolded normally using the S-sulfonation approach, retaining activity following site-specific mono-labeling with a red fluorescent dye. In a recent paper, fluorescent dye rhodamine-labeled hepcidin was prepared through a random labeling approach and used to probe the internalization of Fpn , in which an N-succinimidyl ester activated rhodamine green was used to react with the primary amine moieties of hepcidin. Hepcidin has three primary amine moieties (one N-terminal α-amine and two internal side-chain ε-amines) that all react with the modification reagent, so we deduced that it was quite difficult, or even impossible, to obtain site-specifically mono-labeled product by using this random labeling approach. In contrast, our present site-specific labeling approach resulted in a homogeneous mono-labeled product with high efficiency. In other synthetic peptides, l-propargylglycine could also be incorporated during chemical synthesis and the resultant analogue could be conveniently mono-labeled using various functional probes through click chemistry. However, the position of the l-propargylglycine residue should be carefully selected to minimize its disturbance on both activity and refolding of the resultant peptide.
Materials and methods
Agilent reverse-phase columns (analytical column: Zorbax 300SB-C18, 4.6 mm × 250 mm; semi-preparative column: Zorbax 300SB-C18, 9.4 mm × 250 mm, Agilent Technologies, Santa Clara, CA, USA) were used in the experiments. The peptide was eluted from the C18 reverse-phase column using either a neutral or acidic acetonitrile gradient composed of solvent A and solvent B. For the neutral gradient, solvent A was aqueous 20 mm triethylamine (pH 7.0), and solvent B was 90% acetonitrile containing 20 mm triethylamine (pH 7.0). For the acidic gradient, solvent A was 0.1% aqueous TFA, and solvent B was acetonitrile containing 0.1% TFA. The elution gradient was as follows: 0 min, 10% solvent B; 2 min, 10% solvent B; 52 min, 60% solvent B; 54 min, 100% solvent B; 55 min, 100% solvent B; 56 min, 10% solvent B. The flow rate for the analytical column was 0.5 mL·min−1, and that for the semi-preparative column was 1.0 mL·min−1. The eluted peptide was detected by UV absorbance at both 214 and 280 nm. All animal experiments were approved by the Committee on the Use and Care of Laboratory Animals of Tongji University.
Chemical synthesis of human hepcidin and its analogue
The linear human hepcidin and its analogue were chemically synthesized at GL Biochem (Shanghai) Ltd (Shanghai, China) according to a standard solid-phase peptide synthesis protocol using Fmoc methodology. After synthesis, the linear peptides were chemically cleaved from the resin, washed and lyophilized.
S-sulfonation and purification of synthetic hepcidin and its analogue
The powder of crude synthetic hepcidin or its analogue was resuspended in the sulfonation solution (50 mm Tris/HCl, 6 m guanidine chloride, pH 8.5) at a final concentration of ∼ 10 mg·mL−1. Solid sodium sulfite (Na2SO3) and solid sodium tetrathionate (Na2S4O6) were then added to give a final concentration of 100 and 80 mm, respectively. The S-sulfonation reaction was carried out at 25 °C overnight with gentle shaking. The reaction mixture was applied to RP-HPLC after adjusting the pH to 7.0 with NaOH solution. The S-sulfonated peptide was eluted from the C18 reverse-phase column using a neutral acetonitrile gradient. The eluted S-sulfonated peptide fraction was stored at room temperature for subsequent oxidative refolding.
Oxidative folding of the S-sulfonated hepcidin and its analogue
The purified S-sulfonated hepcidin or its analogue was diluted to a final concentration of ∼ 0.2 mg·mL−1 by adding neutral mobile phase solvents. The oxidative folding was carried out by mixing one volume of the pre-warmed diluted S-sulfonated hepcidin fraction and one volume of a pre-warmed folding solution (1.0 m l-arginine-HCl, 2.0 mm EDTA, pH 7.5, with 1.0 mm GSH and 1.0 mm GSSG). At various time points, an aliquot of the folding mixture was removed, acidified to pH 3.0 by adding TFA, mixed with an equal volume of acidic solvent A, applied to RP-HPLC, and eluted by an acidic acetonitrile gradient. The eluted folded hepcidin fraction was manually collected, lyophilized and analyzed by mass spectrometry.
Circular dichroism measurement
The lyophilized, purified human hepcidin was weighed and then dissolved in 20 mm phosphate buffer (pH 7.4) to a final concentration of 0.2 mg·mL−1 (72 μm) for circular dichroism measurement, which was performed on a Jasco-715 spectropolarimeter at 25 °C using a quartz cuvette with 1.0 mm path-length. The data were expressed as molar ellipticity. The software j-700 for windows secondary structure estimation, version 1.10.00 (Jasco Inc, Easton, MD, USA), was used for secondary structural content estimation from the circular dichroism spectrum.
Site-specific labeling of the hepcidin analogue through click chemistry
To form the catalyst CuBr/tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine (TBTA) complex, one volume of fresh CuBr (Sigma-Aldrich, St Louis, MO, USA) solution [100 mm in dimethylsulfoxide (DMSO)/t-BuOH solvent, v/v 3 : 1, containing 10 mm ascorbate] was mixed with two volumes of TBTA (Sigma-Aldrich) solution (100 mm in DMSO/t-BuOH solvent, v/v 3 : 1). The CuBr/TBTA complex was freshly prepared for each use. The folded hepcidin analogue with l-propargylglycine was dissolved in DMSO/t-BuOH solvent (v/v 3 : 1) at a final concentration of 5–10 mg·mL−1. The red fluorescent dye azido-PEG4-Fluoro525 (Click Chemistry Tools, Scottsdale, AZ, USA) was dissolved in DMSO/t-BuOH (v/v 3 : 1) at a final concentration of 50 mm. To initiate the reaction, the peptide, CuBr/TBTA complex and azido-PEG4-Fluoro525 were mixed together at a molar ratio of 1 : 5 : 10. The labeling reaction was carried out at 20 °C overnight. Thereafter, the reaction mixture was diluted with water, acidified to pH 3.0 by adding TFA and then applied to C18 RP-HPLC, and eluted using an acidic acetonitrile gradient. The eluted peptide fraction was manually collected, lyophilized and analyzed by mass spectrometry.
Hepcidin-induced Fpn internalization assays
The coding region of human Fpn was amplified by PCR based on the construct pCMV6-XL5/Fpn (Origene, Beijing, China). The amplified DNA fragment was either cloned into the pEGFP-N1 vector that infused an EGFP protein at the C-terminus of human Fpn or cloned into a pcDNA6 vector that resulted in an untagged Fpn. The expression constructs of the EGFP-tagged Fpn (pEGFP-N1/Fpn) and the untagged Fpn (pcDNA6/Fpn) respectively were then transfected into HEK293 cells in glass bottomed cell culture dishes. One day after transfection, the cells were treated with either folded human hepcidin (2 μm) or the fluorescent-dye-labeled analogue in normal cell culture medium with 10% fetal bovine serum. Following continuous culturing for 5–6 h, the cells were visualized using a confocal fluorescence microscope. The fluorescent-dye-labeled hepcidin was washed away before the treated cells were visualized by confocal fluorescence microscopy.
To isolate macrophages, mice were intraperitoneally injected with 1.0 mL of 4% liquid thioglycollate broth. Three to four days later, the mice were sacrificed and the peritoneal cells were collected and cultured in glass bottomed cell culture dishes. The next day, the non-adherent cells were washed away, and the remaining cells (primarily macrophages) were treated with 2.0 μm of the red-fluorescent-dye-labeled hepcidin. Following continuous culturing for 5–6 h, the cells were visualized using confocal fluorescence microscopy after the labeled hepcidin was washed away.
This work was supported by a grant from the National Natural Science Foundation of China (31100561).