4880? The?Journal?of?Clinical?Investigation? ? ? http://www.jci.org? ? ? Volume 121? ? ? Number 12? ? ? December 2011
Minihepcidins are rationally designed
small peptides that mimic hepcidin activity
in mice and may be useful for the treatment
of iron overload
Gloria C. Preza,1 Piotr Ruchala,2 Rogelio Pinon,2 Emilio Ramos,3 Bo Qiao,2 Michael A. Peralta,4
Shantanu Sharma,5 Alan Waring,2,6 Tomas Ganz,1,2 and Elizabeta Nemeth2
1Department of Pathology, 2Department of Medicine, and 3Department of Chemistry and Biochemistry, UCLA, Los Angeles, California, USA.
4Department of Chemistry, Columbia University, New York, New York, USA. 5Materials and Process Simulation Center, California Institute of Technology,
Pasadena, California, USA. 6Department of Physiology and Biophysics, School of Medicine at the University of California, Irvine, Irvine, California, USA.
Hepcidin is a peptide hormone that mediates systemic iron
homeostasis in vertebrates (1). Hepcidin controls plasma
iron concentrations by inhibiting dietary iron absorption and
release of recycled iron from macrophages. The hormone acts
by inducing the endocytosis of its receptor ferroportin, the sole
known cellular exporter of iron. Hepcidin deficiency causes
or contributes to iron overload in several diseases, including
hereditary hemochromatosis (2, 3), β-thalassemia (4, 5), and
chronic hepatitis C (6, 7).
Current treatment modalities for iron overload include phle-
botomy and iron chelation. Phlebotomy is a relatively inexpen-
sive and effective treatment for hereditary hemochromatosis but
is not satisfactory for all patients. Some patients do not tolerate
phlebotomy because of concurrent anemia, poor vascular access,
adverse physiological responses to phlebotomy, fears, religious
beliefs, and the lack of long-term convenient access to phlebot-
omy centers (8). Even the patients who are compliant with phle-
botomy treatment may prefer oral therapy if it were available.
In β-thalassemia, phlebotomy is generally not feasible because it
worsens anemia, and iron overload is treated by regular chelation
(9). Iron chelators have side effects ranging from mild to very seri-
ous, and compliance is often suboptimal. Iron overload remains a
major cause of morbidity and mortality in β-thalassemia.
Additional therapeutic options are clearly desirable. Given the
causal role of hepcidin deficiency in the development of iron
overload, replacement of hepcidin would be a rational approach
to the prevention and treatment of iron overload in these dis-
orders (10). However, the use of natural hepcidin as a potential
replacement therapy in hepcidin-deficient conditions has major
limitations. Bioactive hepcidin is 25–amino acids long and has
4 disulfide bonds, rendering the production of a correctly folded
full-length hepcidin expensive. The half-life of natural hepcidin
is very short (11) due to its rapid renal excretion. Furthermore,
peroral absorption of hepcidin would be low due to its large size
(~2.7 kDa). In search of alternatives to full-length hepcidin, we
carried out a structure-function analysis of the hepcidin-fer-
roportin interface. Based on this information, we developed
a series of 7– to 9–amino acid peptides, “minihepcidins,” that
mimic the activity of hepcidin in cell-based bioassays and in
mice. Our findings establish the feasibility of small, drug-like
hepcidin agonists for the treatment of iron overload disorders
due to hepcidin deficiency.
Conflict?of?interest: Tomas Ganz and Elizabeta Nemeth are officers and sharehold-
ers in Intrinsic LifeSciences, a company engaged in the development of iron-related
Citation?for?this?article: J Clin Invest. 2011;121(12):4880–4888. doi:10.1172/JCI57693.
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 12 December 2011
Definition of the extracellular loop of ferroportin
involved in hepcidin binding
We previously showed that an extracellular ferroportin residue,
C326, is essential for hepcidin binding (12), but the extent of the
extracellular loop surrounding C326 was not known. In the 2 pub-
lished models of ferroportin topology, the extracellular loop con-
taining C326 spans the region from D325 to S342 (13) or F316 to
T334 (14). We subjected the proposed ends of the extracellular loop
to cysteine-scanning mutagenesis. In WT ferroportin, C326 and pos-
sibly C205 have extracellular location (13). Because of that, a double
mutant (C205S/C326S) of human ferroportin-GFP was used as a
template for site-directed mutagenesis to introduce cysteines in
multiple positions between M312 and L345. To determine which
of the newly introduced cysteines were located on the cell surface,
HEK293T cells were transfected with plasmids encoding control
or mutant ferroportin-GFP and biotinylated using nonpermeable,
sulfhydryl (SH)-reactive biotin-maleimide. Streptavidin-peroxidase
probing of SDS-PAGE blots of immunoprecipitated ferroportin-
GFP showed that, as expected, WT ferroportin-GFP but not the
C205S/C326S double mutant was biotinylated (Figure 1A). Ferro-
portin mutants with cysteine substitutions in the region between
M312 and L322 and between I344-L345 were expressed on the plas-
ma membrane (Figure 1B) but were not biotinylated, suggesting
that these residues are located in a transmembrane (or intracellular)
region. In contrast, G323, F324, T334C, S338C, S340C, and S343C
mutant ferroportins were all biotinylated, demonstrating their
extracellular location and indicating that the extracellular loop con-
taining C326 likely extends from G323 to S343.
A previous study (15) described the F324-S343 ferroportin region
as the hepcidin-binding domain (HBD) and reported that a syn-
thetic peptide corresponding to the HBD interacts with hepcidin
in a manner similar to that of the membrane-associated cellular
ferroportin. However, in our studies, the synthetic HBD peptide
(RR-FDCITTGYAYTQGLSGSILS-RR) did not bind hepcidin
specifically. As analyzed by surface plasmon resonance, synthetic
human hepcidin interacted equally with the WT and C326S HBD
peptide (Supplemental Figure 1A; supplemental material available
online with this article; doi:10.1172/JCI57693DS1) or with unal-
kylated and alkylated HBD (data not shown). In contrast, similar
modifications to ferroportin have dramatic effects on hepcidin
binding (12). Any interactions that were observed between human
hepcidin and HBD were nonspecific and probably due to hepcidin
aggregating readily on a variety of surfaces (Supplemental Figure
1B). The reverse configuration, in which hepcidin was coated on
the chip and HBD was in fluid phase, did not result in any binding
(Supplemental Figure 1C). The results indicate that the synthetic
F324-S343 peptide does not bind hepcidin specifically and that
the segment does not function as an autonomous HBD.
Ferroportin residues in the vicinity of C326 also participate
in hepcidin binding
To determine whether ferroportin residues other than C326 con-
tribute to hepcidin binding, we mutagenized to alanine each resi-
due in the C326 extracellular loop of WT human ferroportin (Fpn)-
GFP except for A332, which was substituted with D (aspartate).
HEK293T cells were transfected with the ferroportin mutants, and
the uptake of radiolabeled hepcidin was measured. To account for
the variability of transfection efficiency, radioactive counts were
normalized to the expression of ferroportin-GFP as determined
by Western blotting. D325A protein was expressed at very low
levels, and we therefore tested the mutant D325N, which had a
normal level of expression. Remarkably, substitutions in the first
half of the extracellular loop markedly reduced hepcidin binding,
whereas mutations in the second half did not interfere with hepci-
din binding (Figure 1C). However, a reduction in hepcidin binding
by a specific mutation may occur not only because the residue is
involved in contacting hepcidin but because of mistrafficking of
the mutant protein or because of a conformational change, lead-
ing to reduced accessibility of the binding site containing C326. To
address these possibilities, we used SH-reactive biotin-maleimide
to biotinylate cells transfected with the ferroportin-GFP mutants
from Figure 1C. The mutations that reduced hepcidin binding
without affecting the accessibility of C326 were F324A and Y333A
but not D325N, I327A, and A332D (Figure 1D). Thus, F324 and
Y333 likely make direct contact with hepcidin. The fluorescence
microscopy images of the mutants’ cellular localization are shown
in Supplemental Figure 2.
Identification of hepcidin residues critical for binding to ferroportin
We showed previously that deletion of the 5 N-terminal resi-
dues of hepcidin ablated the ability of the peptide to cause fer-
roportin degradation (16). By substituting individual N-terminal
residues (residues 1–6 and 8), we also showed that H3, F4, and I6
were important for the peptide activity and that those positions
required hydrophobic side chains for interaction with ferropor-
tin (17). Here, we mutagenized the next large hydrophobic resi-
due, F9. Substitutions of F9 with alanine or cyclohexylalanine (a
nonaromatic hydrophobic residue) resulted in 100- and 10-fold
reduced activity respectively, suggesting that position 9 requires
an aromatic side chain (Figure 2A).
Because our previous study on the ferroportin thiol residue C326-
SH (12) indicated that disulfide bond formation may occur during
hepcidin-ferroportin interaction, we next focused on the effect of
individual disulfide bond substitutions on hepcidin peptide activ-
ity. Although we previously partially explored this question (16),
the hepcidin structure has since been revised (18), and we generated
synthetic hepcidin mutants in which we substituted pairs of cyste-
ines with alanines according to the revised connectivity (C7A/C23A,
C10A/C13A, C11A/C19A, C14A/C22A). A dose-response analysis
of the mutant peptide activity (Figure 2B) showed that all these
mutants had a similar decrease in activity (up to 100 fold). Circular
dichroism spectroscopy indicated that the mutant peptides had
slightly more disordered structure than the native peptide (Supple-
mental Figure 3), perhaps partially accounting for the diminished
activity. Given that none of the mutants displayed a complete loss
of activity, the result suggests that if disulfide exchange is involved
in hepcidin binding, more than one disulfide bond could partici-
pate in exchange with C326-SH on ferroportin.
RosettaDock computer modeling
To gain further insight into the hepcidin-ferroportin interaction, we
performed computer modeling. The input structures were the full-
length hepcidin (PDB 2KEF), with the most recent disulfide bond
assignment (18), and a region of ferroportin (residues 306–362)
encompassing the C326 extracellular loop and flanking transmem-
brane helices, with structure based on a recently developed theo-
retical model of ferroportin molecule (19). Ferroportin and hepci-
din structures were refined using HyperChem 7.5 and GROMACS
4882? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 12 December 2011
before docking studies. The initial stage docking program ZDOCK
(20) was first used to identify the lowest energy conformer set, and
this ensemble was optimized using the RosettaDock program
(21). The RosettaDock refinement run generated 1,000 structures
whose energy scores were plotted versus the root mean square
distance from the starting input conformation (Figure 3A). The
plot showed the optimal energetic “funnel” of low-energy struc-
tures. The 10 best-scoring low-energy structures all had the
N terminus of hepcidin as the part forming the interface with fer-
roportin. Of those 10 structures, the conformation most compat-
Defining the hepcidin-binding loop on ferroportin.
(A) To detect the boundaries of the C326-containing
extracellular loop, certain residues between M312
and L345 on the human Fpn-GFP C205S/C326S
mutant were substituted with cysteine. Mutant con-
structs were transfected into HEK293T cells, and
cell-surface biotinylation was performed using a
nonpermeable, SH-reactive biotin-maleimide. Pro-
tein lysates were immunoprecipitated with anti-GFP
antibody, analyzed by SDS-PAGE, blotted, and
probed with streptavidin-peroxidase to determine
which mutated residues had extracellular localiza-
tion. The bottom row indicates the amount of Fpn-
GFP that was immunoprecipitated. Each mutant
was tested in at least 3 separate experiments. (B)
Confocal microscopy of HEK293 cells transfected
with cysteine substitution mutants. Mutant proteins
M312C, L314C, F316C, Y318C, L322C, I344, and
L345 were all displayed on the cell surface similar
to WT ferroportin-GFP, indicating that mistraffick-
ing was not the reason for the absence of cell-sur-
face biotinylation in A. Original magnification, ×63.
(C) To define the contribution of residues in the
C326 extracellular loop to hepcidin binding, ala-
nine-scanning mutagenesis and other substitutions
were performed as indicated. After transfection of
HEK293T cells, cellular uptake of 125I-hepcidin was
determined and normalized to the Fpn-GFP levels
in each sample, determined by Western blotting.
Each bar represents the average of 3 to 4 separate
experiments, and error bars indicate standard devi-
ation. (D) Cell-surface biotinylation of constructs
from C. Of the mutants that showed decreased
hepcidin uptake, only F324A and Y333A mutants
had thiol-specific biotinylation similar to that of WT
Fpn-GFP, whereas others (D325N, I327A, A332D)
had decreased cell-surface thiol-biotinylation, sug-
gesting that these mutants may be mistrafficked or
that a conformational change may be blocking the
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 12 December 2011
ible with our mutagenesis data (second-lowest energy structure)
was used as the final hypothetical model of the ferroportin-hepci-
din complex (Figure 3B and Supplemental Figure 4).
The model highlighted that aromatic and hydrophobic side
chain interactions can drive the hepcidin-ferroportin interaction
(Supplemental Table 1). In the model, H3, F4, and I6 of hepcidin
formed a hydrophobic pocket for the interaction with Y333 of fer-
roportin, and F9 of hepcidin interacted with F324 of ferroportin
through pi-stacking, in agreement with our mutagenesis data. The
model also positioned the C326 residue in close proximity to the
hepcidin disulfide framework.
Minihepcidins display agonist activity
The mutagenesis approach as well as RosettaDock modeling sug-
gested that hepcidin residues important for binding to ferroportin
are located at the highly conserved N terminus of the peptide and
prominently include residues H3, F4, I6, and F9. We thus synthe-
sized short peptides (minihepcidins), consisting of up to 9 N-ter-
minal amino acids of hepcidin (Supplemental Table 2), and tested
their ability to cause ferroportin-GFP degradation in the cellular
bioassay. These first generation minihepcidins showed substantial
agonist activity (Figure 4), although they were less potent than the
full-length hepcidin (EC50 = 9 nM). The minihepcidin containing
9 N-terminal residues (hep9, DTHFPICIF; EC50 = 76 nM; Supple-
mental Figure 5) was more active than shorter minihepcidins.
Thiol cysteine is required for minihepcidin activity
Because our previous study showed that a thiol form of the C326
residue on ferroportin was necessary for hepcidin binding and
that the mutant with isosteric substitution C326S did not bind
hepcidin in vitro (12) and caused hepcidin resistance in vivo (22),
we hypothesized that a disulfide exchange between C326 and the
disulfide framework of hepcidin may occur upon hepcidin bind-
ing to ferroportin. To determine whether disulfide bonding could
also be involved in minihepcidin binding to ferroportin, we tested
the requirement for cysteine on minihepcidin by mutating the C7
residue of hep9. We found that the activity of hep9 depended on
its ability to participate in disulfide exchange. Isosteric substitu-
tion with serine (C7S) or blocking of C7 thiol with a t-butyl group
(C7-S-t-but), which prevents disulfide exchange, ablated the ago-
nist activity of hep9 (Figure 5A). However, the activity was consid-
erably preserved when the blocking group was disulfide linked to
C7 (C7-SS-t-but) and was therefore capable of disulfide exchange.
These results strongly suggest that the minihepcidin-ferro-
portin interaction is dependent on the thiol-thiol interaction or
Hepcidin amino acid residues important for binding to ferro-
portin. Full-length 25–amino acid hepcidin variants were pre-
pared by chemical synthesis and oxidatively refolded. The
peptide activity was measured by a flow cytometry–based
assay detecting ferroportin-GFP degradation. (A) F9 substi-
tution with nonaromatic cyclohexylalanine (cha) or alanine
caused a 10- and 100-fold decrease in peptide activity. (B)
Pairwise substitutions of cysteines with alanines to remove
individual disulfide bonds reduced bioactivity to a similar
extent. Each data point is the average of 3 to 4 individual
experiments and error bars indicate SD.
A RosettaDock model of the interaction between hepcidin and the fer-
roportin loop surrounding C326. (A) Plot of the full set of 1,000 struc-
tures generated in the RosettaDock refinement run. The energy score
versus the root mean square (rms) distance from the starting structure
shows the optimal energetic funnel of low-energy structures clustered
around a single position. The plot indicates that the docking algorithm
has reached a good local minimum and that the lowest scoring (low-
est free energy) structures are the most stable. The second-lowest
energy structure was most compatible with our mutagenesis data and
is presented in B. (B) Hepcidin sequence is shown in pink, with yellow
disulfide connectivities. Ferroportin extracellular loop is shown in cyan,
transmembrane helices are shown in gray, and the sulfur of C326 is
shown in yellow. The strongly interacting side chain pairs are displayed
in thicker lines. Hydrogens are not shown. The model identified hydro-
phobic interactions between H3, F4, and I6 on hepcidin and Y333 on
ferroportin and pi-stacking interaction between F9 and F324. The inter-
action of aromatic residues on the hepcidin and ferroportin surfaces
brings C326 close to the disulfide network of hepcidin (disulfide cage),
raising the possibility of transient disulfide exchange.
4884? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 12 December 2011
disulfide formation and that the hepcidin-ferroportin interac-
tion could involve disulfide exchange. To further examine this,
we tested whether the hepcidin-ferroportin complex is affected
by reducing agents. In our previous experiments, we noted that
125I-hepcidin forms a complex with ferroportin that does not
dissociate during protein extraction, immunoprecipitation, and
SDS-PAGE. However, this complex readily dissociated in the pres-
ence of DTT (Figure 5B), supporting the possibility of hepcidin-
ferroportin disulfide exchange.
Design and in vitro bioassay of minihepcidins
In order to optimize the agonist activity, we designed additional
minihepcidins of varying lengths and structural modifications
(Supplemental Table 2). All peptides were synthesized as carboxy-
amides (-CONH2) to make their C terminus similar to the corre-
sponding intact hepcidin segment. The peptide activity was tested
in a cellular bioassay in vitro, measuring their ability to cause deg-
radation of ferroportin-GFP (16).
Substitution or stabilization of the C7 thiol group in hep9
decreases bioactivity. To increase the stability of minihep-
cidins, we synthesized circular peptides by introducing
an internal disulfide bond with C7. We also modified
the size and reactivity of the amino acid correspond-
ing to C7 by introducing various cysteine analogs at
that position. All of these cysteine analogs showed
decreased bioactivity relative to that of hep9.
Retro-inverso minihepcidins are bioactive. To increase the
resistance of minihepcidins to proteolysis in vivo, we
introduced unnatural amino acids (chemical structures
shown in Supplemental Table 3) or synthesized retro-
inverso analogs. Retro-inverso peptides are composed of
D-amino acids assembled in the reverse order from that
of the parent L-peptide sequence, making them steri-
cally similar to the original but much more resistant to
proteolysis. Retro-inverso hep9 analog (ri-hep9, FICIP-
FHTD) and retro-inverso hep3-9 (ri-hep3-9, FICIPFH)
caused ferroportin degradation in vitro, although with
several fold lower activity than that of hep9.
Effects of other peptide modifications. Additional peptide
variants were pegylated in order to increase solubility,
reduce immunogenicity, and increase circulatory time
by reducing renal clearance. To increase the intestinal
uptake of minihepcidins via enterohepatic circulation
of bile acids, the pegylated retro-inverso hep9 was con-
jugated with chenodeoxycholic or ursodeoxycholic bile
acids (cheno-ri-hep9 and urso-ri-hep9). Cheno-ri-hep9
had similar activity in vitro to that of unmodified hep9, while urso-
ri-hep9 had 10-fold lower activity. We also synthesized retro-inver-
so hep9 conjugated to palmitoyl groups because palmitoylation
of peptides could increase intestinal absorption, reduce renal
excretion through increased binding to albumin, and potentially
increase affinity for cell membranes (23). Singly palmitoylated
retro-inverso hep9 (palmitoyl-ri-hep9) was as active as hep9, but
the peptide with 2 palmitoyl groups had no activity. Beta-propel-
ler and collagen domain modifications, designed to multimerize
minihepcidins and thereby reduce their renal excretion, resulted
in inactive minihepcidins.
Minihepcidins induce hypoferremia in mice
Several of the peptides that were most active in vitro were selected
for testing in mice. Although our minihepcidins were derived from
the human sequence, the human hepcidin is active in the mouse
presumably because of the high degree of identity between mouse
and human sequences of hepcidin and ferroportin. Selected ana-
Minihepcidins are effective hepcidin agonists. The activ-
ity of synthetic minihepcidins was determined by flow
cytometric quantitation of ferroportin-GFP degradation in
stably transfected HEK293 cells. Each point is an aver-
age of 5 to 6 experiments and error bars indicate SD.
hep8, DTHFPICI; hep7, DTHFPIC; hep3-9, HFPICIF;
hep3-8, HFPICI; Hep3-7, HFPIC; Hep4-7, FPIC.
Evidence for the role of disulfide exchange in the hepcidin-ferroportin interaction.
(A) The agonist activity of minihepcidin depends on its thiol reactivity. Hep9 was
modified at residue C7; C7-SS-tbut contains an exchangeable tert-butyl group
disulfide-linked to C7; C7-S-tbut contains a tert-butyl group in a thioether linkage
to C7, which prevents disulfide exchange; and C7S contains a serine in posi-
tion 7 instead of a cysteine. Each data point is the average of 3 to 4 individual
experiments and error bars indicate SD. (B) Reducing agents disrupt 125I-hepcidin
association with Fpn-GFP. 125I-hepcidin (125I-hep) was added to cells expressing
ferroportin-GFP for 15 minutes, and total cellular protein was immunoprecipitated
using anti-GFP antibody. The sample was split into 2 identical aliquots, and DTT
was added to 1 of the aliquots (final concentration 100 mM). The amount of
immunoprecipitated Fpn-GFP was determined by Western blotting. Similar experi-
ments were performed at least 5 times.
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 12 December 2011
logs were injected intraperitoneally into mice, and a decrease in
serum iron was measured 4 hours after injection. Control mice
were treated with the diluent or with 20 nmoles human hep25
(~50 μg per mouse). Even at 10-times higher concentrations,
minihepcidin hep9 had no effect on serum iron levels, presumably
because it was rapidly degraded, but a more stable retro-inverso
hep9 showed significant activity (P = 0.028; Figure 6A). Mini-
hepcidins conjugated to fatty or bile acids were even more active.
Injection of 20 nmoles palmitoyl-ri-hep9 caused almost as large
a decrease in serum iron levels as the equivalent dose of hep25
(Figure 6B). Importantly, when the cysteine residue in palmitoyl-
ri-hep9 was substituted with serine, no serum iron decrease was
observed (Supplemental Figure 6). Injection of a higher dose of
palmitoyl-ri-hep9 (200 nmoles) caused more profound hypofer-
remia than did injection of hep25, with an 80% drop in serum iron
(Figure 6B). Even a 7–amino acid retro-inverso peptide (ri-hep3-9;
Figure 6B) was bioactive after parenteral injection, causing a 70%
decrease in serum iron at 200 nmoles. The decrease in serum iron
was not a consequence of the induction of endogenous hepcidin,
as murine hepatic hepcidin-1 mRNA levels measured by qRT-PCR
were not increased in agonist-treated mice compared with those in
controls (Supplemental Figure 7).
Selected minihepcidins are bioactive when administered by gavage
In many situations, the most desirable method of drug delivery
is by the peroral route. To test whether minihepcidins adminis-
tered orally would exert a hypoferremic effect, we gavaged mice
with 200 nmoles retro-inverso hep9 that was palmitoylated or
conjugated to bile acids (chenodeoxycholic or ursodeoxycholic
acid). All 3 minihepcidins caused a significant decrease in serum
iron compared with that in controls gavaged with diluent or
with Hep25 (Figure 6C).
Minihepcidins prevent liver iron overload in hepcidin-1
In a proof-of-principle experiment, we examined the chronic
effect of minihepcidin injections on liver iron accumulation in a
mouse model of hereditary hemochromatosis. We chose hepci-
din-1 knockout mice to eliminate any possibility of endogenous
hepcidin contributing to the regulation of iron loading. Hepc1–/–
animals were injected parenterally with palmitoyl-ri-hep9 (100
nmoles; n = 7) or diluent (n = 6) daily for 2 weeks. The end point
hemoglobin (15 ± 2 g/dl and 15.9 ± 1.5 g/dl in palmitoyl-ri-hep9
and control group, respectively) and other hematological param-
eters were similar between the 2 groups of mice, indicating that
minihepcidin injections did not cause iron-restricted erythropoi-
esis. However, the nonheme iron concentration of the liver was sig-
nificantly lower in animals treated with the minihepcidin (13.7 ± 4
μmoles/g wet weight) than that in those treated with diluent only
(27.1 ± 13.7 μmoles/g wet weight) (P = 0.03; Figure 6D).
In hereditary hemochromatosis, hepcidin deficiency is a conse-
quence of destructive mutations in the genes encoding hepcidin
(24) or one of its regulators, HFE (3) (most common), transferrin
receptor 2 (25), or hemojuvelin (26). Hepcidin deficiency results
Bioactivity of parenteral and oral minihep-
cidins in mice. (A) Full-length hepcidin
(20 nmoles), minihepcidins (200 nmoles),
or diluent were injected intraperitoneally
(n = 6 animals per treatment), mice were
euthanized after 4 hours, and nonheme iron
levels were determined in serum. ri-hep9,
retro-inverso hep9; Cyc-1, minihepcidin with
a disulfide bond (CDTHFPICIF). *P = 0.003
and **P = 0.035 in comparison with control
mice. (B) Full-length hepcidin (20 nmoles),
minihepcidins (20 and 200 nmoles), or dilu-
ent were injected intraperitoneally into mice
(n = 5–7 per treatment), and their serum
iron was measured after 4 hours. ri-hep3-9,
retro-inverso HFPICIF. *P = 0.02; **P = 0.002;
†P = 0.015; #P < 0.001. (C) Peptides (200
nmoles) or diluent were administered orally
by gavage, and serum iron measured was
after 4 hours. The results are expressed as
the percentage of decrease in serum iron
as compared with the serum iron levels in
PBS-treated mice to normalize for the varia-
tions in absolute serum iron levels between
experiments. *P = 0.006; **P = 0.003;
***P = 0.01. (D) Hepcidin-1 knockout mice
were injected intraperitoneally with palm-
ri-hep9 (100 nmoles per mouse; n = 7) or
diluent (n = 6) daily for 2 weeks. Liver iron
content was measured on day 15. *P = 0.03.
Error bars represent SD.
4886? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 12 December 2011
in excessive intestinal iron absorption and maldistribution of
tissue iron. If severe and untreated, the disease can progress to
cirrhosis, hepatocellular carcinoma, and endocrine and cardiac
problems. In the US, about 0.4% of people of mixed European
descent have hereditary hemochromatosis mutations that put
them at risk for iron overload (27).
Iron overload also occurs in β-thalassemia, a disease that affects
millions of patients whose origins are in the Mediterranean basin
and Asia, regions historically affected by endemic malaria (28). In
β-thalassemia, hepcidin is suppressed by increased erythropoietic
activity through mechanisms that are still incompletely defined. In
the absence of transfusions (β-thalassemia intermedia), hepcidin
levels are very low (4, 5), and patients develop iron overload similar
to that of severe hereditary hemochromatosis and may succumb
to iron overload if untreated (29). In β-thalassemia major, trans-
fusions rather than dietary iron absorption are the predominant
cause of iron overload. Here, hepcidin levels are higher, because
ineffective erythropoiesis is suppressed by transfusions and also
because the additional iron load stimulates hepcidin production.
However, in the intervals between transfusions, hepcidin concen-
trations progressively decrease toward the end of the interval (4, 5,
30), and this recurrent lowering of hepcidin may also contribute
to iron overload in β-thalassemia major.
Milder hepcidin insufficiency is seen in chronic liver disease,
including viral hepatitis (6, 7, 31). The cause of hepcidin suppres-
sion in these diseases is not yet clear, but the resulting chronic iron
loading in the liver worsens the prognosis (32).
Therapeutic augmentation of hepcidin levels would be expected
to curb hyperabsorption of dietary iron in these patients. The proof
of this concept was achieved in mouse model of HFE hereditary
hemochromatosis, in which the introduction of a hepcidin trans-
gene into Hfe–/– mice prevented the development of iron overload
(33). Hepcidin also causes redistribution of iron when iron over-
load is already established. Hepcidin induction in HFE-null mice
carrying a Tet-inducible hepcidin construct did not acutely reverse
the iron overload but did shift the excess iron from hepatocytes
and other parenchymal cells to macrophages (34). Macrophages
are relatively resistant to the toxic effects of iron (35), as demon-
strated by the relatively benign course of “ferroportin disease,” in
which iron accumulates in macrophages. Most affected patients
have no clinical manifestations, despite often severe iron overload
(36). Thus, hepcidin-mediated redistribution of iron from paren-
chyma to macrophages in iron-overloaded patients could poten-
tially limit iron toxicity in the heart, the pancreas, and the liver.
In addition to the beneficial effect of hepcidin on iron balance
and distribution, recent studies suggest that hepcidin may also
improve disordered erythropoiesis in β-thalassemia. Although
the mechanism is still not understood, transgenic expression of
hepcidin in a mouse model of β-thalassemia intermedia increased
hemoglobin and decreased extramedullary erythropoiesis (37).
Using full-length hepcidin for the treatment of iron overload con-
ditions would be expected to be expensive, not only because the syn-
thesis and refolding of hepcidin with 4 disulfide bonds allows for a
large number of alternative folds but also because of the relatively
high dose required for its biological effect. Even assuming that
hepcidin could be produced at a cost comparable to that of recom-
binant insulin, a typical dose of hepcidin would likely be many
fold higher than that of insulin. In human patients with type 1
diabetes mellitus (and mouse models), a typical dose of insulin is
0.2–0.7 U/kg/d or 9–32 μg/kg/d (1 U is the biological equivalent
of 45.5 μg pure crystalline insulin). A typical dose of hepcidin, on
the other hand, is 50 μg/mouse/d or 2 mg/kg/d, which is nearly
100-times higher than the typical dose of insulin. Given the current
limitations of peptide synthetic technology, designing more potent
and less expensive hepcidin analogs would be advantageous.
In this study, we developed small peptides, minihepcidins, that
act as hepcidin agonists. Their rational design was facilitated by
the identification of the region on hepcidin and ferroportin mol-
ecules that is critical for their binding. Together with our previous
structure-function studies (16, 17), we showed that hydrophobic
contacts dominate the ligand-receptor interaction and that several
N-terminal amino acids of hepcidin, up to residue 9, are critical for
its activity. We therefore used the first 9 residues as the scaffold for
minihepcidin design. We also found that the ability to participate
in thiol-thiol interactions is essential for minihepcidin activity.
Blocking the only cysteine (C7) with a protective group abrogated
the peptide activity unless the protective group was added in such
a way that disulfide exchange was possible. Furthermore, in con-
trast to its Cys-containing counterpart, minihepcidin bearing the
Cys to Ser substitution did not cause serum iron decrease in mice
after intraperitoneal injection.
To improve agonist activity, we developed the minihepcidins
with structural modifications that addressed some undesirable
physicochemical properties of the minihepcidin scaffold, such as
thiol instability, hydrophobicity, poor gastrointestinal absorption,
susceptibility to proteolysis, and instability in the bloodstream.
We succeeded in designing minihepcidins of 7– or 9–amino acids
in length that were active in mice in vivo. Intraperitoneal injec-
tion of proteolysis-resistant retro-inverso minihepcidins caused
hypoferremia similar to that caused by native hepcidin. In a proof-
of-principle experiment, chronic administration of minihepcidins
significantly decreased iron loading in a mouse model of heredi-
tary hemochromatosis. Hepcidin-1 knockout mice, which received
intraperitoneal injections of a retro-inverso minihepcidin daily for
2 weeks, had significantly lower liver iron content that hepcidin-1
knockout mice injected with solvent.
Importantly, minihepcidin conjugated to fatty or bile acids
caused hypoferremia after oral administration by gavage also.
Given the number and rich variety of peptides involved in biologi-
cal processes, relatively few have been FDA approved as therapeu-
tics, and their efficient delivery by the oral route is still an unmet
goal. Development of oral hepcidin analogs would represent a
major advance in peptide pharmacology.
The mainstay of treatment for patients with iron overload
is phlebotomy if they are not anemic and iron chelation if they
have anemia. Although these measures are effective at reducing
excess iron, they are frequently not well tolerated by patients, and
compliance with iron-depleting therapy is suboptimal. If hepci-
din therapy proves to be effective and relatively free of side effects,
it could represent a major improvement over existing therapies,
either alone or in combination with current approaches to allow
modifications that would make the treatment less burdensome
and better accepted by patients.
Site-directed mutagenesis. For cysteine scanning of the ferroportin region
M312-L345, the human ferroportin-GFP C205S/C326S mutant (12) was
used as a template for site-directed mutagenesis. For alanine scanning of
the region F324-S343, WT human ferroportin-GFP was used as a template.
Primers were selected using the QuikChange Primer Design (https://www.
? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 12 December 2011
genomics.agilent.com/), and mutagenesis was performed using Quik-
Change II XL Mutagenesis Kit (Agilent Technologies). Mutant plasmids
were sequenced to verify the presence of the mutations.
Transient transfection and microscopy. WT and mutant ferroportin-GFP
plasmids were transiently transfected into HEK293T cells using either Lipo-
fectamine 2000 (Invitrogen) or Amaxa nucleofection (Lonza) (12). Cells
were visualized with an epifluorescence microscope (Nikon Eclipse), and
images were acquired with a SPOT camera and SPOT Advanced Imaging
Software (Diagnostic Instruments). Confocal microscopy was performed
at the UCLA CNSI Advanced Light Microscopy Facility, using a Leica
TCS-SP MP Confocal and Multiphoton Inverted Microscope equipped
with an argon laser (488 nm blue excitation, JDS Uniphase). The images
were acquired using a ×63 oil-immersion objective lens and LCS Lite Soft-
ware (Leica Confocal Software, Leica Microsystems).
Cell-surface biotinylation. Cells were treated with a nonpermeable SH-reac-
tive, maleimide-PEG2-biotin (12) (0.5 mM final concentration) according
to the manufacturer’s instructions (Pierce). Each ferroportin-GFP con-
struct was analyzed in at least 3 independent experiments.
125I-hepcidin internalization assay. The assay was performed as described
previously (12, 38). Cells were treated with 125I-hepcidin for 1 hour at 37°C.
The radioactivity of untransfected cells was subtracted as background
for each point. The counts were then normalized to the ferroportin-GFP
expression of each sample as determined by quantitation of Western blots
using Image Lab Software (Bio-Rad).
Immunoprecipitation and Western blotting. Cell lysis, immunoprecipitation,
and Western blotting were performed as previously described (12). Polyclonal
ab290 anti-GFP antibody (Abcam) was used for immunoprecipitation, bio-
tinylated proteins were detected with streptavidin-HRPO (Pierce), and a
monoclonal anti-GFP antibody (clones 7.1 and 13.1, Roche) was used to
determine the amount of ferroportin-GFP that was immunoprecipitated.
Surface plasmon resonance. Surface plasmon resonance was performed on
a BIAcore 3000 System (BiaCore AB) as previously described (39). Peptides
were immobilized on a CM5 sensor chip by the amine-coupling protocol at a
range of response units (1,100–6,000 response units). Analyte solutions were
prepared at concentrations of 1–10 μg/ml in HBS-EP buffer (10 mM HEPES,
pH 7.4, 150 mM NaCl, 3 mM EDTA, and 0.005% surfactant P20). Analytes were
introduced into the flow cells at 50 μl/min for 3 minutes at 37°C. Resonance
signals were corrected for nonspecific binding by subtracting the signal of the
control flow cell and analyzed using BIAevaluation 4.1 software (Biacore).
Peptide synthesis and purification. All peptides were synthesized using the
standard solid-phase fmoc chemistry (16) and were purified using reverse-
phase HPLC. Minihepcidins were synthesized as carboxyamides (-CONH2),
which created a charge-neutral end more similar to a peptide bond than the
negatively charged –COOH. The list of minihepcidins with their sequences
is shown in Supplemental Table 2, and the structures of unnatural amino
acids are displayed in Supplemental Table 3.
Circular dichroism spectroscopy. Circular dichroism spectra (185–260 nm) of
the hepcidin peptides were recorded on a JASCO 720 Spectropolarimeter
(Easton) as previously described (16). Estimates of the secondary structure
contributions were made using the SELCON algorithm (40) and spectral
basis set SP43 in the Olis Global Works software suite (Olis).
Modeling of hepcidin-ferroportin interaction. The structure for hepcidin (NP_
066998, http://www.ncbi.nlm.nih.gov/protein/NP_066998) was modeled
by the coordinate set for the lowest energy conformer from the Protein
Data Bank (PDB 2KEF) (18). The structure for ferroportin (NP_055400.1,
http://www.ncbi.nlm.nih.gov/protein/NP_055400.1) was approximated
using the coordinate set provided by Nathan Subramaniam (19). The fer-
roportin region spanning residues 306–362 and the hepcidin structure
were individually minimized in the HyperChem environment to obtain
the lowest energy conformer.
Hepcidin was first docked into the ferroportin segment using the initial-
stage docking program ZDOCK (http://zdock.bu.edu/). The lowest energy
conformer set was then used as starting ensemble for a more refined dock-
ing using RosettaDock (http://rosettadock.graylab.jhu.edu/), which gen-
erated a low-energy structure used as a final hypothetical model for the
In vitro bioassay for measuring the activity of hepcidin derivatives. ECR293-
Fpn (38), a cell line stably transfected with the mouse ferroportin-GFP
construct under the control of the ponasterone-inducible promoter, was
used for flow cytometry measurements of GFP fluorescence as previously
described (16). For each peptide, the treatment with the full range of
peptide concentrations was repeated independently 3 to 7 times.
In vivo bioassay for measuring activity of hepcidin derivatives. Animal studies
were approved by the Animal Research Committee at UCLA. Six-week-old
C57BL/6 mice were obtained from The Jackson Laboratory. Mice were
placed on an iron-deficient diet (~4 ppm iron, Harlan Teklad) for 2 weeks
to suppress endogenous hepcidin and to decrease its mouse-to-mouse
variability (16, 41). Mice were subjected to the following treatments: (a)
intraperitoneal injection with 0, 20, or 200 nmoles peptide solubilized
in 100 μl SL220 solubilization agent (NOF) and (b) gavage with 0 or 200
nmoles peptide in 250 μl 1× solvent (Cremophor EL [Sigma-Aldrich]/etha-
nol/PBS = 12.5:12.5:75) (42). Mice were sacrificed 4 hours later, and serum
iron was determined using a colorimetric assay (Diagnostic Chemicals)
(16). To normalize for the variations in absolute serum iron levels among
experiments, the results were expressed as the percentage of decrease in
serum iron as compared with the serum iron levels in PBS-treated mice.
Each peptide concentration was tested in 5 to 7 animals.
For the study of chronic minihepcidin effects, we used hepcidin-1 knock-
out mice, originally provided to our laboratory by Sophie Vaulont (43).
Hamp1–/– mice (5–6 weeks of age, 6 males, and 7 females) were placed on
low-iron diet for 2 weeks (4 ppm iron) to slow down the development of
iron overload prior to injections with minihepcidins. The mice were then
switched to the standard mouse chow (336 ppm iron) and given 100 μl
intraperitoneal injections daily of either 100 nmoles palmitoyl-ri-hep9
solubilized in SL220 or the solvent only. After 2 weeks, the mice were
euthanized, and their nonheme liver iron concentration was measured.
Nonheme iron concentration was determined as described (44), using
acid-based protein precipitation followed by a colorimetric assay for iron
quantitation (Diagnostic Chemicals).
Quantitative real-time PCR. Liver RNA was extracted using TRIzol reagent
(Invitrogen). Gene expression of Hamp1 and the reference gene β-actin
(Actb) was assessed by qRT-PCR using the iScript cDNA Synthesis Kit and
iQ SYBR Green Supermix (Bio-Rad) (45). HAMP1 primer sequences are
as follows: 5′-TTGCGATACCAATGCAGAAGA-3′ and 5′-GATGTGGCTC-
TAGGCTATGTT-3′. β-Actin primer sequences are as follows: 5′-ACCCA-
CACTGTGCCCATCTA-3′ and 5′-CACGCTCGGTCAGGATCTTC-3′.
Statistics. SigmaPlot version 11.0 (Systat Software) was used for statistical
analyses. A P value of less of 0.05 was considered significant.
We would like to acknowledge the Office of Information Technol-
ogy at the University of California, Irvine, and Broadcom Corp. for
providing the Broadcom Distributed/Unified Cluster used for the
molecular dynamics simulation and docking of the protein com-
plex. The UCLA Jonsson Comprehensive Cancer Center and Center
for AIDS Research Flow Cytometry Core Facility was instrumental
for flow cytometry measurements. Matthew Schibler at the UCLA
CNSI Advanced Light Microscopy Facility provided help with con-
focal microscopy. We thank Grace Jung for assistance with Biacore
analysis and Nathan Subramaniam for providing the PDB file of
research article Download full-text
4888? The?Journal?of?Clinical?Investigation http://www.jci.org Volume 121 Number 12 December 2011
his ferroportin model. The work was funded by NIH grants R01
DK 082717 (to E. Nemeth) and R01 DK 065029 (to T. Ganz).
Received for publication February 21, 2011, and accepted in revised
form September 21, 2011.
Address correspondence to: Elizabeta Nemeth, CHS 37-055,
Department of Medicine, David Geffen School of Medicine at
UCLA, 10833 Le Conte Ave., Los Angeles, California 90095-1690,
USA. Phone: 310.825.7499; Fax: 310.206.8766; E-mail: enemeth@
1. Ganz T, Nemeth E. Hepcidin and disorders of iron
metabolism. Annu Rev Med. 2011;62:347–360.
2. Ganz T, Olbina G, Girelli D, Nemeth E, Westerman
M. Immunoassay for human serum hepcidin. Blood.
3. Piperno A, et al. Blunted hepcidin response to oral
iron challenge in HFE-related hemochromatosis.
4. Origa R, et al. Liver iron concentrations and urinary
hepcidin in beta-thalassemia. Haematologica. 2007;
5. Kearney SL, et al. Urinary hepcidin in congeni-
tal chronic anemias. Pediatr Blood Cancer. 2007;
6. Fujita N, et al. Patients with chronic hepatitis C
achieving a sustained virological response to pegin-
terferon and ribavirin therapy recover from impaired
hepcidin secretion. J Hepatol. 2008;49(5):702–710.
7. Girelli D, et al. Reduced serum hepcidin levels in
patients with chronic hepatitis C. J Hepatol. 2009;
8. Barton JC. Chelation therapy for iron overload.
Curr Gastroenterol Rep. 2007;9(1):74–82.
9. Porter JB. Optimizing iron chelation strategies
in beta-thalassaemia major. Blood Rev. 2009;
10. Bartnikas TB, Fleming MD. A tincture of hepcidin
cures all: the potential for hepcidin therapeutics.
J Clin Invest. 2010;120(12):4187–4190.
11. Rivera S, Liu L, Nemeth E, Gabayan V, Sorensen
OE, Ganz T. Hepcidin excess induces the seques-
tration of iron and exacerbates tumor-associated
anemia. Blood. 2005;105(4):1797–1802.
12. Fernandes A, et al. The molecular basis of hepci-
din-resistant hereditary hemochromatosis. Blood.
13. Liu XB, Yang F, Haile DJ. Functional consequenc-
es of ferroportin 1 mutations. Blood Cells Mol Dis.
14. Rice AE, Mendez MJ, Hokanson CA, Rees DC,
Bjorkman PJ. Investigation of the biophysical and
cell biological properties of ferroportin, a mul-
tipass integral membrane protein iron exporter.
J Mol Biol. 2009;386(3):717–732.
15. De Domenico I, et al. The hepcidin-binding site on
ferroportin is evolutionarily conserved. Cell Metab.
16. Nemeth E, Preza GC, Jung CL, Kaplan J, Waring AJ,
Ganz T. The N-terminus of hepcidin is essential for
its interaction with ferroportin: structure-function
study. Blood. 2006;107(1):328–333.
17. Clark RJ, Tan CC, Preza GC, Nemeth E, Ganz T,
Craik DJ. Understanding the structure/activity
relationships of the iron regulatory peptide hepci-
din. Chem Biol. 2011 2011;18(3):336–343.
18. Jordan JB, et al. Hepcidin revisited, disulfide con-
nectivity, dynamics, and structure. J Biol Chem. 2009;
19. Wallace DF, Harris JM, Subramaniam VN. Function-
al analysis and theoretical modeling of ferroportin
reveals clustering of mutations according to pheno-
type. Am J Physiol Cell Physiol. 2010;298(1):C75–C84.
20. Chen R, Li L, Weng Z. ZDOCK: an initial-stage pro-
tein-docking algorithm. Proteins. 2003;52(1):80–87.
21. Lyskov S, Gray JJ. The RosettaDock server for local
proteinprotein docking. Nucleic Acids Res. 2008;
36(Web Server issue):W233–W238.
22. Sham RL, Phatak PD, Nemeth E, Ganz T. Heredi-
tary hemochromatosis due to resistance to hep-
cidin: high hepcidin concentrations in a fam-
ily with C326S ferroportin mutation. Blood. 2009;
23. Resh MD. Palmitoylation of ligands, receptors,
and intracellular signaling molecules. Sci STKE.
24. Roetto A, et al. Mutant antimicrobial peptide hep-
cidin is associated with severe juvenile hemochro-
matosis. Nat Genet. 2003;33(1):21–22.
25. Nemeth E, Roetto A, Garozzo G, Ganz T, Camas-
chella C. Hepcidin is decreased in TFR2 hemochro-
matosis. Blood. 2005;105(4):1803–1806.
26. Papanikolaou G, et al. Mutations in HFE2 cause
iron overload in chromosome 1q-linked juvenile
hemochromatosis. Nat Genet. 2004;36(1):77–82.
27. U.S. Preventive Services Task Force. Screening for
Hemochromatosis: Recommendation Statement.
Ann Intern Med. 2006;145(3):204–208.
28. Cunningham MJ. Update on thalassemia: clini-
cal care and complications. Pediatr Clin North Am.
29. Taher A, Isma’eel H, Cappellini MD. Thalassemia
intermedia: Revisited. Blood Cells Mol Dis. 2007;
30. [No authors listed]. International BioIron society
meeting. Am J Hematol. 2009;84(8):E236–E375.
31. Fujita N, et al. Hepcidin expression in the liver: rel-
atively low level in patients with chronic hepatitis
C. Mol Med. 2007;13(1–2):97–104.
32. Alla V, Bonkovsky HL. Iron in nonhemochromatotic
liver disorders. Semin Liver Dis. 2005;25(4):461–472.
33. Nicolas G, et al. Constitutive hepcidin expression
prevents iron overload in a mouse model of hemo-
chromatosis. Nat Genet. 2003;34(1):97–101.
34. Viatte L, et al. Chronic hepcidin induction causes
hyposideremia and alters the pattern of cellular
iron accumulation in hemochromatotic mice. Blood.
35. Gualdi R, Casalgrandi G, Montosi G, Ventura
E, Pietrangelo A. Excess iron into hepatocytes is
required for activation of collagen type I gene dur-
ing experimental siderosis. Gastroenterology. 1994;
36. Pietrangelo A. The ferroportin disease. Blood Cells
Mol Dis. 2004;32(1):131–138.
37. Gardenghi S, et al. Hepcidin as a therapeutic tool to
limit iron overload and improve anemia in beta-thal-
assemic mice. J Clin Invest. 2010;120(12):4466–4477.
38. Nemeth E, et al. Hepcidin regulates cellular iron
efflux by binding to ferroportin and inducing its
internalization. Science. 2004;306(5704):2090–2093.
39. Wang W, et al. Retrocyclins kill bacilli and germinat-
ing spores of Bacillus anthracis and inactivate anthrax
lethal toxin. J Biol Chem. 2006;281(43):32755–32764.
40. Sreerama N, Venyaminov SY, Woody RW. Estima-
tion of the number of alpha-helical and beta-strand
segments in proteins using circular dichroism spec-
troscopy. Protein Sci. 1999;8(2):370–380.
41. Nemeth E, et al. IL-6 mediates hypoferremia of
inflammation by inducing the synthesis of the iron
regulatory hormone hepcidin. J Clin Invest. 2004;
42. Kim S, et al. Sorafenib inhibits the angiogenesis and
growth of orthotopic anaplastic thyroid carcinoma
xenografts in nude mice. Mol Cancer Ther. 2007;
43. Lesbordes-Brion JC, et al. Targeted disruption of
the hepcidin 1 gene results in severe hemochroma-
tosis. Blood. 2006;108(4):1402–1405.
44. Pak M, Lopez MA, Gabayan V, Ganz T, Rivera S. Sup-
pression of hepcidin during anemia requires eryth-
ropoietic activity. Blood. 2006;108(12):3730–3735.
45. Ramos E, et al. Evidence for distinct pathways of
hepcidin regulation by acute and chronic iron load-
ing in mice. Hepatology. 2011;53(4):1333–1341.