proteins

STRUCTURE O FUNCTION O BIOINFORMATICS

Probing protease sensitivity of recombinant

human erythropoietin reveals a3a4
inter-helical loop as a stability determinant
Jesse Sebastian Samuel,1 Deepak Kumar,1 Sathi Babu Chodisetti,1 Javed N. Agrewala,1
Balvinder Singh,1 Purnananda Guptasarma,2 and Dibyendu Sarkar1*
1 CSIR- Institute of Microbial Technology, Sector-39A, Chandigarh 160036, India
2 Department of Biological Sciences, Indian Institute of Science Education & Research (IISER) Mohali, Punjab 140306, India

ABSTRACT
Although unglycosylated HuEpo is fully functional, it has very short serum half-life. However, the mechanism of in vivo
clearance of human Epo (HuEpo) remains largely unknown. In this study, the relative importance of protease-sensitive sites
of recombinant HuEpo (rHuEpo) has been investigated by analysis of structural data coupled with in vivo half-life measurements. Our results identify a3-a4 inter-helical loop region as a target site of lysosomal protease Cathepsin L. Consistent
with previously-reported lysosomal degradation of HuEpo, these results for the first time identify cleavage sites of rHuEpo
by specific lysosomal proteases. Furthermore, in agreement with the lowered exposure of the peptide backbone around the
cleavage site, remarkably substitutions of residues with bulkier amino acids result in significantly improved in vivo stability.
Together, these results have implications for the mechanism of in vivo clearance of the protein in humans.
Proteins 2015; 83:18131822.
C 2015 Wiley Periodicals, Inc.
V

Key words: cathepsin L; circulating half-life; inter-helical loop region; in vivo clearance; proteolysis; recombinant HuEpo.

INTRODUCTION
Human erythropoietin (HuEpo), a glycoprotein hormone produced primarily in the adult kidneys, is the key
regulator of red blood cell production and differentiation.1,2 The cognate receptor (EpoR) along with the
ligand Epo regulates erythropoiesis3,4 by triggering
intra-cellular signaling events including receptor phosphorylation, and activation of JAK-STAT, RAS, and PI3
kinase pathways.5,6 Together, these signaling events control gene expression to trigger cells to undergo proliferation and differentiation. Thus, recombinant HuEpo
(rHuEpo) is used worldwide for treatment of anemia
resulting from chronic kidney diseases, chemotherapy
and AIDS7,8 (for a review see Ref. 9).
The human epo gene encodes a 165 amino acid mature
protein comprising four long a-helices connected by two
long cross-over loops and one short loop.10 The primary
sequence contains three N-linked (N24, N38, and N83)
and one O-linked glycosylation site (S126), all of which are
normally glycosylated in vivo.11,12 Although glycosylation
of HuEpo has been shown to play an important role in the
stability, solubility, and in vivo biological activity of the

C 2015 WILEY PERIODICALS, INC.

V

protein,1317 it is not required for in vitro biological activity

of rHuEpo. In fact, glycosylation appears to interfere with
the binding of HuEpo to its receptor EpoR.2 Consistently
Epo variants with higher sialic acid content were shown to
have lower-affinity for EpoR but a longer serum half-life
and higher efficiency for stimulating the production of red
blood cells in vivo.18,19 However, what remains unclear is
how HuEpo is cleared off from circulation and degraded in
the body. An earlier investigation indicated that HuEpo is
degraded by receptor-mediated uptake and degradation
occurs via the lysosomal pathway.20 This study compared
Abbreviations: NESP, novel erythropoiesis stimulating protein; WT, wild-type.
Grant sponsors: Department of Biotechnology (DBT), Government of India,
CSIR-IMTECH; Grant sponsors: ICMR (Indian Council of Medical Research) and
CSIR (to J.S.S. and S.B.C); Grant sponsor: CSIR-IMTECH (to D.K.).
Deepak Kumar current address is Department of Experimental Medicine and
Biotechnology, Postgraduate Institute of Medical Education and Research
(PGIMER), Chandigarh 160012, India
Sathi Babu Chodisetti current address is Department of Microbiology and Immunology, Pennsylvania State University College of Medicine, Hershey, Pennsylvania
17033
*Correspondence to: Dibyendu Sarkar, CSIR-Institute of Microbial Technology,
Sector-39A, Chandigarh 160036, India. E-mail: dibyendu@imtech.res.in
Received 18 April 2015; Revised 7 July 2015; Accepted 21 July 2015
Published online 27 July 2015 in Wiley Online Library (wileyonlinelibrary.com).
DOI: 10.1002/prot.24865

PROTEINS

1813

J. Sebastian Samuel et al.

Table I
Oligonucleotide Primers Used in this Study
Primersa

Description

FPrHuEpo
RPrHuEpo
FPepoR55Q
RPepoR55Q
FPepoR133Q
RPepo133Q
D10A
D10E
FPM1b
RPM1
FPM2c
RPM2

TTGTTGCATATGAAAGCCCCACCACGC
TTGTTGGAATTCTCTGTCCCCTGT
GCCTGGAAGCAGATGGAGGTC
GACCTCCATCTGCTTCCAGGC
GCTCCACTCCAGACAATCACT
AGTGATTGTCTGGAGTGGAGC
TTGTTGCATATGAAAGCCCCACGCCTCATCTGTGCAAGCCGAGTC
TTGTTGCATATGAAAGCCCCACGCCTCATCTGTGAAAGCCGAGTC
CCTCCAGATGTTCTTACTCTTGTTCCACTCCGA
TCGGAGTGGAACAAGAGTAAGAACATCTGGAGG
CCTCCAGATCTTGTTACTGTTCTTCCACTCCGA
TCGGAGTGGAAGAACAGTAACAAGATCTGGAGG

Reference
This
This
This
This
This
This
This
This
This
This
This
This

work
work
work
work
work
work
work
work
work
work
work
work

FP, forward primer; RP, reverse primer.

AASAA130 to 126VLTLV130.
AASAA130 to 126LVTLV130.
Note that the position of each amino acid is assigned according to the rHuEpo primary sequence shown in Figure 2A.
b126
c126

glycosylated, WT rHuEpo with Darbepoietin and concluded that the latter was degraded less owing to lower
level of internalization due to its overall lower-affinity with
the Epo-R.21 Furthermore, whatever amount of Darbepoietin was internalized got degraded lesser compared to
the WT molecule due to higher number of carbohydrate
moieties protecting the protein part of the molecule. Keeping this in view, we hypothesized that identifying proteasesensitive sites of rHuEpo might be of insight into mechanisms of degradation of the protein.
rHuEpo, expressed from mammalian cells has a halflife of 413 hours following intravenous administration,
necessitating frequent administration for optimum therapeutic benefit.22 Thus, it was of interest to understand
the mechanisms of in vivo clearance of HuEpo. Although
how and where Epo gets degraded remains a matter of
much debate,23 it is generally believed that in vivo degradation of HuEpo might involve more than one mechanism.20 We hypothesized that a major mechanism for
degradation of Epo may require that the protein has
structural weak points, which most likely provide initial
nucleation sites for cleavage leading to complete degradation. We reasoned that the proteolytic analyses with trypsin might provide important insights into the structural
determinants that intrinsically control degradation of
rHuEpo. To this end, independent of glycosylation
rHuEpo was expressed and purified from E. coli. As a
primary proof-of-concept, we used trypsin to carry out
limited proteolysis of the refolded rHuEpo. We extended
this line of thought to ascertain whether specific
region(s) of rHuEpo is susceptible to proteolytic cleavage. We had already seen this to be true in case of trypsin. Next, we sought to investigate major cleavage sites of
rHuEpo by lysosomal proteases. The results reported
here identify a3a4 inter-helical loop region as a major
structural weak point of the molecule. Using structureguided rational mutagenesis coupled with in vitro cell

1814

PROTEINS

proliferation assays and in vivo stability assays, we show

that mutant rHuEpo proteins carrying amino acid substitutions within the a3a4 inter-helical loop region have
significantly improved in vivo stability compared to the
WT protein. The results have significant implications for
the mechanism of in vivo clearance of the protein in
humans.
MATERIALS AND METHODS
Preparation, amplification, and cloning of
DNA, site-directed mutagenesis

The gene encoding HuEpo was obtained from Origene

Technologies and cloned into expression vector pET23a
(Novagen). For overproduction and purification of
rHuEpo in E. coli, the entire epo gene was PCR-amplified
using oligonucleotide pairs epoStart and epoStop
(Table I) and the wild-type (WT) epo gene as template,
and introduced between an NdeI and a XhoI site of
pET23a (Novagen). The cloning strategy using standard
protocols as described24 resulted in full-length HuEpo
with natural N-termini (residues Met and Lys were
added at the beginning of the sequence to aid expression
in bacteria) and a C-terminal His6-tag.Site-directed
mutagenesis of individual rHuEpo residues (described
under Results section) was carried out in the epo gene
(pET-epo) by PCR-based two-stage overlap extension
method the mutant constructs were verified by DNA
sequence analysis.
Protein purification

The WT and mutant rHuEpo proteins were overexpressed in E. coli BL21(DE3)star cells (Invitrogen) as
insoluble proteins. Cells carrying pET-epo plasmid was
grown to an A600 of 0.6 and protein expression was

Proteolysis of rHuEpo by Lysosomal Proteases

induced with 1 mM IPTG. Subsequently, the cells were

grown for 4 hours at 378C before they were harvested by
centrifugation. After adding an equal weight of ice-cold
buffer A (20 mM Na1-phosphate buffer, pH 7.20, plus
70 mM NaCl) containing 0.1 mg/mL of lysozyme (5 mL
buffer for final 1 g of wet weight of cells) and incubated
on ice for 30 min. All subsequent procedures were performed at 48C. The lysates were sonicated, and soluble
supernatant was removed by centrifugation for 30 min at
15,000 g. The insoluble pellet containing rHuEpo as
inclusion body was re-suspended overnight in buffer A
containing 8M Urea (buffer B). The suspension was clarified at 15,000 g for 30 min, the supernatant was decanted and equilibrated Ni-NTA beads (1 mL of 50%
slurry) were added to it. After incubating the beads for
60 min on a rocking platform, the slurry was loaded
onto a column and washed with buffer B containing
20 mM imidazole. The bound protein was eluted stepwise in buffer B containing imidazole ranging from 100500 mM. Fractions containing denatured protein was
collected, pooled and diluted in buffer C (buffer A plus
5 mM b-cyclodextrin) drop-wise to a final concentration
of 80 mM urea under gentle stirring. On completion of
this process, 1 mL pre-equilibrated Ni-NTA slurry was
added to the solution and incubated overnight under
gentle stirring. The solution containing refolded rHuEpo
bound to Ni-NTA was loaded onto a column and washed
with two column volumes of buffer A containing 0.5M
NaCl, followed by five column volumes of buffer A to
remove b-cyclodextrin. The bound rHuEpo was then
eluted step-wise in buffer A containing imidazole ranging
from 100500 mM. After analysis of polypeptide composition by 12% SDS-PAGE, appropriate column fractions
were pooled, concentrated, and buffer exchanged against
buffer A containing 0.3M NaCl. The molecular mass
determined by mass spectrometry was in agreement with
the values predicted from primary sequence. The purity
of proteins was 95% as judged by Coomassie blue
staining of over-loaded SDS/polyacrylamide gels. Protein
concentration was determined with Bradford reagent
using bovine serum albumin as the standard, and is
expressed in equivalents of protein monomers.
Bioassay of rHuEpo proteins

Human leukemia cells F-36E (RIKEN, Japan) which

proliferate only in the presence of biologically active Epo
were used to test bioactivity of WT and mutant rHuEpo
proteins. F-36E cell line was maintained in RPMI-1640
containing 10% FBS (Fetal Bovine Serum), and 5U/mL
of commercial rHuEpo (Calbiochem). First, cells were
washed three times in RPMI-1640 to remove residual
Epo present in the cultures. Viable cells were counted by
trypan blue exclusion method. The F-36E cells (1 3 104
cells/100 mL) were cultured in 96-well plate in RPMI1640 1 10% FBS lacking Epo for 24 hours. Following

this, graded doses of Epo were added and cells were

incubated at 378C for 48 hours. Finally, 3[H]-thymidine
(0.5 mCi/well) was added and the cells were incubated
for another 5 hours. Subsequently, the cells were harvested and radioactivity incorporated was measured by
liquid scintillation counting.
Trypsin digestion of rHuEpo

Reaction mixtures (15 lL) containing 5 lg of rHuEpo

was incubated in the digestion mixture containing
20 mM Na1-phosphate (pH 7.2), 70 mM NaCl, and 400
ng of trypsin at 378C for indicated time points. Proteolysis was terminated by addition of SDS, the aliquots from
the digestion mixtures were boiled for 1-2 min and then
resolved by 12% SDS/PAGE. To determine the Nterminal sequence, resolved polypeptides from a SDS/
polyacrylamide gel were transferred electrophoretically
onto a PVDF membrane (Millipore) and visualized by
Amido Black (Sigma) staining. Membrane slices containing individual proteolytic products were excised and subjected to automated Edman sequencing using an Applied
Biosystem protein sequencer.
Size-exclusion chromatography and CD
spectroscopy

Gel filtration was performed at 258C on a Protein KW

803 (8 x 300 mm; Shodex) using a HPLC system (Shimadzu). Samples of rHuEpo were eluted at a flow rate of
0.5 mL/min in 50 mM Na1-phosphate (pH 7.2) and
300 mM NaCl. The column was pre-equilibrated with
the same buffer prior to each run, and calibrated by elution of standard globular proteins of known molecular
weight.CD spectroscopy was carried out with a JASCO
spectropolarimeter (J-810; Japan). Residue molar ellipticity [h] was defined as 100 hobs (lc)21, where hobs was
observed ellipticity; l was length of the light path in centimeters, and c was residue molar concentration of each
protein. Scan speed was 50 nm/min and multiple scans
were averaged to increase signal-to-noise ratio.
In vivo stability assays

Seven weeks old BALB/c mice were kept in a room

maintained at 248C and a relative humidity of 55%
throughout the experiment. Regulatory approval for animal experiments was granted by the Institutional Animal
Ethics Committee (No.5/1999/CPCSEA). The animals
were allowed to acclimatize to these conditions for 1
week before starting the experiment. A standard rodent
feed in pellet form and drinking water ad libitum were
available throughout the study. Proteins were labeled
using Na-125I in iodination tubes (Thermo Scientific) as
per the manufacturers recommended protocol. After
labeling for 10 min the reaction was stopped by using
NaI; labeled protein was separated from unincorporated
PROTEINS

1815

J. Sebastian Samuel et al.

label using a P6 quick-spin purification column (BioRad). A 5 mL aliquot was TCA precipitated to assess the
radioactive counts for the pellet and supernatant fractions, and was used to calculate the efficiency of incorporation
of the label by
the equation:
%
incorporation 5 [(P S)/(P 1 S)] * 100Where P stands
for counts-per-minute in the pellet and S stands for
counts-per-minute in the supernatant. 125I labeled WT
and mutant rHuEpo proteins with >95% radioactive
incorporation were administered at a dose of 0.48 mg/kg
polypeptide equivalent and a volume of 2 mL/kg via the
tail vein. Solutions of 125I-rHuEpo (0.24 mg/mL) were
prepared in PBS containing 0.05% (w/v) Tween 20 and
0.05% (w/v) mouse serum albumin. Blood samples were
withdrawn via the retro-orbital sinus using a heparinized
capillary into heparin coated tubes at indicated time
points. The blood was centrifuged at 15,000 rpm for 3
min to isolate the plasma; 50 mL of plasma, along with
300 mg of BSA, was precipitated using 200 mL of 25%
TCA. The mixture was incubated on ice for 10 min and
then centrifuged at 13,000 rpm for 10 min. The resulting
pellet was washed with 200 mL of ice-cold acetone and
centrifuged again at 13,000 rpm for 5 min; acetone was
aspirated out from the tube and the pellet was dried.
Counts per minute (cpm) were noted from the pellets
and the amount of protein back-calculated from it. The
remaining labeled protein at varying time points when
plotted represent decline in labeled rHuEpo concentration according to its plasma half-life.
Figure 1
Structural modeling

X-ray crystal structure of HuEpo (PDB ID: 1EER; 25)

served as the template for structural models. The mutant
molecules (M2 and M3) carrying substitutions at the a3a4 loop region were generated by predicting the mutant
structures with optimized side chains of mutant amino
acid residues followed by overall verification of the stereochemistry by PROCHECK.26 In all cases, structural
models were constructed using PYMOL27 and accessible
surface area of all (backbone plus side-chain) atoms in
A2 for amino acid residues spanning 126-130 were evaluated as described previously.28
RESULTS AND DISCUSSIONS
Cloning, expression and purification of
rHuEpo

To facilitate purification, full-length rHuEpo was overexpressed as C-terminal His6-tagged fusion protein using
pET23a (Novagen) expression plasmid. IPTG induction
resulted in high level expression of a major protein of
18.5 kDa, which was absent in IPTG-treated cells containing the vector plasmid lacking an insert [compare
lane 2 and lane 1; Fig. 1(A)]. However, major portion of

1816

PROTEINS

Expression, purification, and activity assay of rHuEpo.(A) WT rHuEpo

was expressed in E. coli BL21DE3star cells as a fusion protein containing a C-terminal polyhistidine (His6) tag as described in the Materials
and methods. Approximately 5 mg of protein from cell extracts of different stages of expression and purification steps were analyzed by SDSPAGE and visualized by Coomassie blue staining. Lane 1, uninduced
cell extract; lane 2, induced cell extract; lane 4, supernatant of the sonicated cell extract; lane 5, pellet fraction of the sonicated cell extract;
lane 7, purified rHuEpo; lane 8, molecular mass marker and sizes of the
proteins (in kDa) are indicated to the right. (B) Bioactivity of indicated
concentrations of WT rHuEpo (empty columns) or commercially available HuEpo (filled columns) was compared by F-36E cell proliferation
assays and quantitated by 3[H]-thymidine incorporation as described in
the Materials and methods. The data represent average values 6 SEM
derived from at least three independent measurements. [Color figure
can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

the over-produced rHuEpo was found as the insoluble

pellet in the crude lysate [compare lane 5 and lane 4;
Fig. 1(A)]. Finally, rHuEpo comprising residues Met-1 to
Glu-170 was purified by refolding from the insoluble
fraction as described in the Materials and methods. The
purity of the protein preparations was  95% as judged
by SDS-PAGE analysis and subsequent staining with
Coomassie blue [see Fig. 1(A)]. As the purification protocol involved refolding of the insoluble protein, secondary structure of rHuEpo was monitored by far-UV-CD
spectroscopy. Consistent with the solution structure,10

Proteolysis of rHuEpo by Lysosomal Proteases

Figure 2
Proteolysis of rHuEpo by trypsin. (A) shows all of the probable trypsin cleavage sites within the primary sequence of rHuEpo. (B) Trypsin cleavage
sites were identified by N-terminal sequencing of peptides from lane 2 (see Results for a description of proteolytic fragments) as described under
Materials and methods; lane 1 represents undigested rHuEpo; sizes of marker proteins in kDa are indicated to the left. (C) Structural model of
rHuEpo showing trypsin cleavage sites between peptide bonds Arg55-Met56 of the a1-a2 inter-helical loop, and Arg133-Thr134 of the a3-a4
inter-helical loop, respectively. Note that in this study the position of each amino acid is assigned according to the rHuEpo primary sequence
shown in Figure 2A. (D) Far UV-CD spectra of 10 mM of WT and mutant rHuEPo proteins in 50 mM Na-phosphate buffer pH 7.20 and 300 mM
NaCl. The detail of data acquisition was as described in the Materials and methods. Inset shows SDS-PAGE analysis of 2 lg/lane of indicated
rHuEpo proteins (lanes 24), expressed and purified as described in the Materials and methods; sizes of marker proteins (lane 1) in kDa are shown
to the left. (E) 4 mg of WT and indicated mutant rHuEpo proteins were digested with 400 ng of trypsin for 15, 30 and 45 min, respectively. The
undigested proteins, as well as the digested samples as indicated on the figure, were resolved by SDS-PAGE and visualized by Coomassie blue staining. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

spectrum of the refolded protein displayed minima at

208 nm and 220 nm (data not shown), characteristic of
proteins with a high a-helical content. Thus, we surmise
that the E. coli-derived purified rHuEpo is most likely in
the native form of its structure.
To assess quaternary organization of rHuEpo, samples
of rHuEpo was subjected to gel filtration chromatography, and compared to protein standards of known
molecular mass. rHuEpo sample showed a single-peak
elution corresponding to a molecular mass of 20.9 kDa
(data not shown). This is in good agreement with the
primary sequence-derived molecular mass of the mono-

meric form of rHuEpo (19.8 -kDa). Thus, under the

conditions examined, rHuEpo appears to be monomeric
in solution when purified from E. coli.
We next examined biological activity of purified
rHuEpo. To this end, human leukemia cells (F36E) which
proliferate only in presence of biologically active Epo
were utilized to determine bioactivity of rHuEpo as
described in the Materials and methods. In 1989, the
pleural fluid of a male patient with RAEB (refractory
anemia with excess blasts), a subtype of Myelodysplastic
syndrome was used to establish the F36E cell line, and its
subtype, the F36E. If a mutant rHuEpo molecule fails to
PROTEINS

1817

J. Sebastian Samuel et al.

retain its native structure, and therefore activity, when

added it will be unable to support proliferation of F36E
cells. Cell proliferation in the presence of indicated concentrations of either rHuEpo (empty bars) or commercially available HuEpo (filled bars) was monitored by
3
H-thymidine incorporation and measured via liquid
scintillation counting [Fig. 1(B)]. The results show that
rHuEpo share comparable bioactivity as that of the commercially available HuEpo.
Probing protease-sensitive sites of rHuEpo

Having defined the bioactivity of E. coli-derived

rHuEpo, purified protein was subjected to proteolysis by
trypsin to investigate protease-sensitive site(s) of the protein. Although we found 21 possible trypsin cleavage
sites within the primary sequence of rHuEpo [Fig. 2(A)],
rHuEpo on proteolytic digestion yielded two specific
fragments of approximately 15.5 and 11.2 kDa, respectively [Fig. 2(B)]. To probe the cleavage sites, both peptide fragments were transferred onto PVDF (Millipore)
membrane and subjected to N-terminal sequencing. Our
results show unambiguously that the top and the bottom
fragments were generated by specific cleavage of the peptide bonds between Arg55-Met56, and Arg133-Thr134 of
rHuEpo primary sequence, respectively. Notably, we
found that the former cleavage site maps within a1-a2
inter-helical region, whereas the latter site lies within the
inter-helical region of a3-a4 [Fig. 2(C)].
To examine importance of these residues of rHuEpo in
trypsin sensitivity, we next performed point mutations at
the cleavage sites and their effects on trypsin sensitivity
were evaluated. We introduced specific Arg to Gln substitutions in the epo gene residing in plasmid pET-epo.
Consistent with our expectations WT and the mutant
proteins showed largely comparable secondary structures
as the point mutations were within the unstructured
loops connecting helices [Fig. 2(D)]. Together, these
results suggest that these mutations do not appear to
influence secondary structure of the protein. Importantly,
proteolysis experiments revealed that under the conditions examined and at comparable protein concentrations, both purified mutants displayed 25 fold lower
sensitivity to trypsin relative to the WT rHuEpo [Fig.
2(E)]. Although the results we showed on trypsin sensitivity of mutant rHuEpo proteins are not surprising and
are even expected, without having shown the mutants
lowered sensitivity to trypsin under identical conditions
examined, we could not rule out the possibility of any
other putative cleavage sites [see Fig. 2(A)] of the
mutants showing sensitivity to trypsin. Together, in conjunction with identification of the trypsin-sensitive sites
and lower trypsin sensitivity of the mutant rHuEpo molecules (carrying mutations at the trypsin-sensitive sites),
we could surmise that the amino acid residues Arg55
and Arg133 of the unstructured loop connecting a1a2,

1818

PROTEINS

Figure 3
Probing cleavage sites of rHuEpo by lysosomal proteases, Cathepsin D
and Cathepsin L, respectively. Proteolysis of rHuEpo was carried out
with Cathepsin D and Cathepsin L (Figs A and C, respectively); samples
were resolved by SDS-PAGE and visualized by Coomassie blue staining.
Protease cleavage sites were identified by N-terminal sequencing of the
indicated peptides from lane 3 (see Results for a description of fragments) as described under Materials and methods. Lane 2 shows undigested rHuEpo; sizes of marker proteins (lane 1) in kDa are indicated
to the left. Panels B and D show structural models of rHuEpo indicating the cleavage sites (in yellow) by Cathepsin D and Cathepsin L,
respectively.

and a3a4 helices of rHuEpo, respectively appear to

constitute the primary cleavage sites of rHuEpo.
Probing cleavage sites of rHuEpo by human
lysosomal proteases

Having verified the proof-of-concept using trypsin, we

were interested in identifying cleavage sites of rHuEpo
using physiologically more relevant human proteases. A
previous report had shown that HuEpo is degraded via
receptor-mediated endocytosis and degradation takes place
after fusion with lysosome by the lysosomal proteases.20
In agreement with this study, lysosomal cysteine proteases
of the papain family have been implicated in terminal
degradation of the protein in the lysosomal compartment29 (for a review see Ref. 30). Thus, we set out our
objective to identify the cleavage sites of rHuEpo, and
introduce mutations at the newly identified cleavage sites
to examine whether rHuEpo variants are more resistant to
degradation by lysosomal proteases. To this end, we optimized the cleavage conditions of rHuEpo using Cathepsin
D and Cathepsin L. Upon cleavage by respective proteases
the products were subjected to N-terminal sequencing.
Our results showed that for Cathepsin D the fragments
were generated by specific cleavage of the peptide bonds

Proteolysis of rHuEpo by Lysosomal Proteases

Figure 4
Structural organization of rHuEPo (PDB ID: 1EER). (A) shows HuEpo structure with the a3-a4 inter-helical loop region (comprising residues
126
AASAA130 shown in sticks) within a box. (B) and (C) display expanded view of amino acid residues 126VLTLV130 and 126LVTVL130 (shown in
sticks) of the loop region as described for mutants M2 and M3, respectively (see Results for a description of the mutants). (D) shows accessible
surface area (A2) of N- and Ca- backbone atoms of indicated residues for WT and mutant proteins (M2 and M3) as measured by structural analyses (see Materials and methods for details). Note that Ser to Thr substitution at the 128th residue do not significantly influence accessibility of the
backbone atoms to the solvent. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

between Leu110-Leu111 of the a3, and Leu143-Phe144 of

the a4 helices of rHuEpo, respectively (Fig. 3). However,
fragments generated by Cathepsin-L involved cleavage of
the peptide bonds between Cys9-Asp10, preceding the a1
helix and between Ala126 and Ala127, which turns out to
be part of the a3a4 inter-helical loop region, respectively. In fact, our proteolysis experiments using other
proteases suggest that there exists more cleavage sites
interspersed over the rHuEpo molecule including other
helices. However, absence of clean peptide sequences from
the gel-derived fragments made it extremely difficult to
identify these sites unambiguously.
In vivo clearance of rHuEpo proteins

As our objective was to study effect of site-specific

mutations and alteration of the residues within the

a-helices might destabilize overall structure and/or fold

of rHuEpo, we tested multiple mutants generated against
Cathepsin L cleavage sites and not against Cathepsin D
cleavage sites. To this end, we constructed a collection of
plasmids encoding a series of mutant rHuEpo proteins.
First, we replaced the negatively charged Asp at the 10th
position of the primary sequence of rHuEpo with Ala
(D10A, referred to as M1) to examine if the change in
charge distribution at this site causes alteration of cleavage by Cathepsin L. We did not mutate Cys9 because it
is part of a highly conserved Cys9-Cys163 disulfide bond,
which remains essential for bioactivity of rHuEpo.31 The
other cleavage site was the peptide bond within the a3a4 inter-helical loop involving residues Ala126 and
Ala127 (Fig. 3), next to the O-linked glycosylation site
(Ser128). We next performed in silico analyses to identify
likely substitutions within AASAA which are expected to
PROTEINS

1819

J. Sebastian Samuel et al.

contribute to in vivo stability of the molecule. The objective was to introduce uncharged bulkier side-chains
replacing Ala residues with the anticipation that they may
protect particularly labile peptide bonds from proteases.
Thus, mutant M2 and M3 comprising amino acid
sequence of 126VLTLV130 and 126LVTVL130, respectively
were constructed based on structural analyses of mutant
proteins. For both mutant proteins, the structural models
were generated using WT HuEpo structural coordinates
(PDB ID: 1EER; 25) (Fig. 4). The accessible surface area
(A2) of backbone atoms of indicated amino acids (spanning residues Ala126 to Ala130) strongly indicate a striking reduction in protease accessibility of the peptide
backbone as a result of mutations (for mutants M2 and
M3) [Fig. 4(D)].Note that the accessible surface area of
the backbone atoms did not display a significant change
when Ser was replaced with Thr residues. Since there are
two consecutive Ala residues (Ala129 and Ala130) next to
Ser128, which we thought could easily serve as the substitute cleavage site if we just mutated Ala126 and Ala127
(as they are equally solvent accessible in unglycosylated
rHuEpo), we also mutated these two Ala residues with
Leu and Val, or vice versa. These two mutants which we
referred to as M2 and M3 were over-expressed and purified to homogeneity to examine if jumbling the sequence
of same amino acids will have any particular advantage
(or otherwise).
Importantly, during in vitro cell proliferation assays,
over a range of protein concentrations examined,
mutants M2 and M3 were found to be comparably active
relative to WT rHuEPo [Fig. 5(A)]. Thus, we conclude
that residues spanning Ala126 to Ala130 of the a3-a4
inter-helical loop region do not appear to contribute to
structural stability and/or bioactivity of rHuEpo. Next,
we examined in vivo stability of mutant rHuEpo proteins
by injecting I125-labeled proteins into mice as described
in the Materials and methods. Quantification of the
experimental data suggests that under the conditions
examined labeled WT rHuEpo is degraded with a halflife of 2.11 6 0.3 hours [Fig. 5(B)]. Under identical conditions examined, mutant M1 (D10A) showed a comparable half-life of 2.03 6 0.2 hours. In contrast, mutant
M2 showed a strikingly longer half-life of 3.11 6 0.3
hours, reflecting 41 6 3% rise in circulating half-life of
the mutant protein (see inset to Fig. 5(B,C)]. Also,
mutant M3 (with similar substitutions within a3-a4
inter-helical loop but an altered sequence of amino
acids) showed 15 6 2% higher half-life compared to the
WT protein. As a control, under identical experimental
condition a double mutant (M4; R55QR133Q) carrying
substitutions at Arg55 of a1a2 inter-helical loop and
Arg133 of a3a4 inter-helical loop region, respectively
showed an insignificant variation (5% higher) of circulating half-life compared to WT rHuEpo. Together, these
results suggest that a3a4 inter-helical loop spanning
residues 126AASAA130 constitute a primary Cathepsin

1820

PROTEINS

Figure 5
Circulating half-lives of WT and mutant rHuEpo proteins in a mouse
model. (A) Bioactivity of varying concentrations of WT and mutant
rHuEpo proteins as compared by F-36E cell proliferation assay and
quantitated by 3[H]-thymidine incorporation as described in the Materials and methods. Average counts due to effective 3[H]-thymidine
incorporation are derived from triplicate wells and presented with SEM.
The data are representative of three independent experiments. (B)
Kinetics of decline of I125-labeled WT rHuEpo concentration in mouse
blood plasma. Values shown here represent average counts of I125labeled rHuEpo (WT) from three independent experiments. In all cases,
the radioactive labeling efficiency of rHuEPo proteins were 95% or
above. Inset shows half-lives of WT and mutant rHuEpo proteins as
measured by time-course of decline in I125-labeled rHuEpo concentration from mouse blood plasma (see Materials and methods for details).
Panel (C) shows percent enhancement of circulating half-lives of
mutant rHuEpo proteins relative to the WT rHuEpo.

L-cleavage site of rHuEpo and substitutions herein significantly influences circulating half-life of the protein in
mice.

Proteolysis of rHuEpo by Lysosomal Proteases

Much of the effort to extend half-life of HuEpo thus

far utilized chemical modifications of the protein. Notably, a hyperglycosylated analog of Epo called novel erythropoiesis stimulating protein (NESP), comprising two
additional N-linked carbohydrate chains, was shown to
have a lower-affinity for EpoR and a 23 fold longer
serum half-life as compared to HuEpo.18,32 Similarly, a
recombinant Epo dimer construct, comprising two complete Epo sequences in tandem has been shown to have a
longer half-life.33,34 We imagined that a potential way
of obviating the need for glycosylation could be to introduce mutations that would replace smaller side-chains
around the labile peptide bonds with bulkier ones. Thus,
the next challenge was to identify labile peptide bonds
which may be considered as structural weak points of
rHuEpo. In vitro proteolysis of rHuEpo mapped the specific trypsin cleavage sites within the inter-helical linkers
(Fig. 2), suggesting that these two unstructured loops
most likely represent flexible structures of the protein,
and are therefore, readily accessible to proteases.35,36
Importantly, alterations at the cleavage sites were shown
to have clearly reduced trypsin sensitivity of the mutant
rHuEpo molecules (Fig. 2). Thus, we extended this line
of thought to identify primary cleavage sites of rHuEpo
by more relevant human lysosomal proteases, Cathepsin
D and Cathepsin L (Fig. 3).
Next, we used structural bioinformatics to evaluate
exposure of backbone atom in every peptide bond of
126
AASAA130 region of rHuEpO to aqueous solvent (Fig.
4). The measured values of surface accessible area indicate that the small sizes of side-chains and the folding of
the polypeptide chain in the relevant regions of the protein have together worked to expose these peptide bonds
to the solvent. We also show that substitutions with
bulkier residues critically influence the exposure of the
backbone in the vicinity of the O-linked glycosylation
site [Fig. 4(D)]. Notably, Cathepsin L cleaved the peptide
bond between Ala124 and Ala125 (adjacent to the Olinked glycosylation site at Ser128) of the a3a4 interhelical region (Fig. 3) justifying our designing of mutants
within the region. Consistent with data from structural
analyses (Fig. 4), mutant M2 carrying substitutions
within the a3a4 inter-helical loop showed a striking
improvement of in vivo half-life of rHuEpo (41 6 3%)
(Fig. 5). Also, mutant M3 carrying similar substitutions
(with identical composition but with a different
sequence) showed an approximately 15% elevated halflife compared to the WT protein. Thus, the structurederived exposure of the backbone atoms for relevant
peptide bonds [Fig. 4(D)] account for the difference in
stability of mutant proteins as determined in a mouse
model (Fig. 5). As a control, mutant M4 carrying mutations at the trypsin cleavage sites failed to influence in
vivo stability of rHuEpo, suggesting that specific amino
acid substitutions (around the Cathepsin L cleavage site)

within the a3a4 inter-helical loop contributes to

improved stability of rHuEpo.
Of all of the mutants, M2 showed the longest half-life.
Although a 40% increase in half-life appears relatively
insignificant, particularly in the light of pre-existing engineered rHuEpo-derivatives, most strikingly in this study
we establish the specific a3-a4 inter-helical loop of
rHuEpo as a target site of a specific lysosomal protease
Cathepsin L, a novel result of unusual significance with
implications on the mechanism(s) of in vivo clearance of
HuEpo. Consistent with a previous study suggesting lysosomal degradation of HuEpo,20 this study, for the first
time provides direct evidence showing lysosomal degradation of HuEpo by specific lysosomal protease. However, we cannot rule out the possibility of cleavage of
HuEpo by Cathepsin L at other sites or other enzymes
degrading HuEpo, particularly because the effect of
Cathepsin D could not be verified in our experiments as
mutants likely to be resistant to Cathepsin D cleavage,
either were difficult to purify or did not express at all.
However, this knowledge may lay a foundation to alter
the protein backbone of HuEpo to obtain variant molecules to our advantage.
ACKNOWLEDGMENTS
The authors thank Yukio Makamura of Cell Bank,
RIKEN (Japan) for F36E cell line, Subramanian Karthikeyan for analyses of structural data, Sharanjeet Kaur for
circular dichroism studies, Paramjeet Kaur, Neha Rana
and Girish Sahni for N-terminal sequencing of the peptides, Neeraj Khatri for his help with animal studies, and
Renu Sharma for technical assistance and for her help
with the preparation of the manuscript.
REFERENCES
1. Graber SE, Krantz SB. Erythropoietin and control of red-cell production. Annu Rev Med 1978;29:5166.
2. Krantz SB, Goldwasser E. Specific binding of erythropoietin to
spleen-cells infected with the anemia strain of friend-virus. Proc
Natl Acad Sci USA 1984;81:75747578.
3. Wu H, Klingm
uller U, Besmer P, Lodish HF. Interaction of the
erythropoietin and stem-cell-factor receptors. Nature 1995;377:242
246.
4. Yamaguchi K, Akai K, Kawanishi G, Ueda M, Masida S, Sasaki R.
Effects of site-directed removal of N-glycosylation sites in human
erythropoietin on its production and biological properties. J Biol
Chem 1991;266:2043420439.
5. Youssoufian H, Longmore G, Neumann D, Yoshimura A, Lodish
HF. Structure, function, and activation of the erythropoietin receptor. Blood 1993;81:22232236.
6. Tong W, Zhang J, Lodish HF. Lnk inhibits erythropoiesis and Epodependent JAK2 activation and downstream signaling pathways.
Blood 2005;105:46044612.
7. Markham A, Bryson HM. Epoetin alpha. A review of its pharmacodynamic and pharmacokinetic properties and therapeutic use in
non renal applications. Drugs 1995;49:232254.
8. Mocini D, Leone T, Tubaro M, Santini M, Penco M. Structure, production and function of erythropoietin: implications for

PROTEINS

1821

J. Sebastian Samuel et al.

9.
10.

11.

12.

13.
14.

15.

16.

17.

18.

19.

20.

21.

therapeutical use in cardiovascular disease. Curr Med Chem 2007;

14:22782287.
Jelkmann W. Erythropoietin after a century of research: younger
than ever. Eur J Haematol 2007;78:83205.
Cheetham JC, Smith DM, Aoki KH, StevensonJL Hoeffel TJ, Syed
RS, Egrie J Harvey TS. NMR structure of human erythropoietin
and a comparison with its receptor bound conformation. Nat Struct
Biol 1998;5:861866.
Lai PH, Everett R, Wang FF, Arakawa T, Goldwasser E. Structural
characterization of human erythropoietin. J Biol Chem 1986;261:
31163121.
Takeuchi M, Takasaki S, Miyazaki H, Kato T, Hoshi S, Kochibe N,
Kobata A. Comparative study of the asparagine-linked sugar chains
of human erythropoietins purified from urine and the culture
medium of recombinant Chinese hamster ovary cells. J Biol Chem
1988;263:36573663.
Dordal MS, Wang FF, Goldwasser E. The role of carbohydrate in
erythropoietin action. Endocrinology 1985;116:22932299.
Takeuchi M, Inoue N, Strickland TW, Kubota M, Wada M, Shimizu
R, Hoshi S, Kozutsumi H, Takasaki S, Kobata A. Relationship
between sugar chain structure and biological-activity of recombinant
human erythropoietin produced in chinese-hamster ovary cells.
Proc Natl Acad Sci USA 1989;86:78197822.
Takeuchi M, Takasaki S, Shimada M, Kobata A. Role of sugarchains in the in vitro biological-activity of human erythropoietin
produced in recombinant chinese hamster ovary cells. J Biol Chem
1990;265:1212712130.
Tsuda E, Kowanishi G, Ueda M, Masuda S, Sasaki R. The role of
carbohydrate in recombinant human erythropoietin. Eur J Biochem
1990;188:405411.
Toyoda T, Arakawa T, Yamaguchi H. Nglycans stabilize human
erythropoietin through hydrophobic interactions with the hydrophobic protein surface: studies by surface plasmon resonance analysis. J Biochem 2002;131:511515.
Egrie JC, Browne JK. Development and characterization of novel
erythropoiesis stimulating protein (NESP). Nephrol Dial Transplant
2001;16(Suppl 3):313.
Macdougall IC, Gray SJ, Elston O, Breen C, Jenkins B, Browne J,
Egrie J. Pharmacokinetics of novel erythropoiesis stimulating protein compared with epoetin alfa in dialysis patients. J Am Soc
Nephrol 1999;10:23922395.
Gross AW, Lodish HF. Cellular trafficking and degradation of erythropoietin and novel erythropoiesis stimulating protein (NESP).
J Biol Chem 2006;281:20242032.
Egrie JC, Dwyer E, Browne JK, Hitz A, Lykos MA. Darbepoetin alfa
has a longer circulating half-life and greater in vivo potency than
recombinant human erythropoietin. Exp Hematol 2003;31:290299.

1822

PROTEINS

22. Flaharty KK, Caro J, Erslev A, Whalen JJ, Morris EM, Bjornsson
TD, Vlasses PH. Pharmacokinetics and erythropoietic response to
human recombinant erythropoietin in healthy men. Clin Pharmacol
Ther 1990;47:557564.
23. Jelkmann W. The enigma of the metabolic fate of circulating erythropoietin (Epo) in view of the pharmacokinetics of the recombinant
drugs rhEpo and NESP. Eur J Haematol 2002;69:265274.
24. Sambrook J, Fritsch EF, Maniatis T. Molecular cloning: a laboratory
manual, 2nd ed. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory; 1989.
25. Syed RS, Reid SW, Li C, Cheetham JC, Aoki KH, Liu B, Zhan H,
Osslund TD, Chirino AJ, Zhang J, Finer-Moore J, Elliot S, Sitney K,
Katz BA, Matthews DJ, Wendoloski JJ, Egrie J, Stroud RM. Efficiency of signalling through cytokine receptors depends critically on
receptor orientation. Nature 1998;395:511516.
26. Laskowski RA, Moss DS, Thornton JM. Main-chain bond lengths
and bond angles in protein structures. J Mol Biol 1993; 231:1049
1067.
27. The PyMOL Molecular Graphics System, Version 1.2r3pre Ed.,
Schr
odinger, L.L.C.
28. Winn MD, Ballard CC, Cowtan KD, Dodson EJ, Emsley P, Evans
PR, Keegan RM, Krissinel EB, Leslie AG, McCoy A, McNicholas SJ,
Murshudov GN, Pannu NS, Potterton EA, Powell HR, Read RJ,
Vagin A, Wilson KS. Overview of the CCP4 suite and current developments. Acta Crystallogr D Biol Crystallogr 2011; 67:235242.
29. Roth W, Deussing J, Botchkarev VA, Pauly-Evers M, Saftig P, Hafner
A, Schmidt P, Schmahl W, Scherer J, Anton-Lamprecht I, Von
Figura K, Paus R, Peters C. Cathepsin L deficiency as molecular
defect of furless: hyperproliferation of keratinocytes and pertubation
of hair follicle cycling. Faseb J 2000;14:20752086.
30. Barrett AJ. Cellular proteolysis: an overview. Ann NY Acad Sci 1992;
674:115.
31. Boissel J-P, Lee W-R, Presnell SR Cohen FE, Bunn HF. Erythropoietin structure-function relationships. Mutant proteins that test a
model of tertiary structure. J Biol Chem 1993;268:1598315993.
32. Chmielewski C. Aranesp (darbepoetin alfa): a new erythropoiesisstimulating protein. Nephrol Nurs J 2002;29:6768.
33. Sytkowski AJ, Lunn ED, Davis KL, Feldman L, Siekman S. Human
erythropoietin dimers with markedly enhanced in vivo activity. Proc
Natl Acad Sci USA 1998;95:11841188.
34. Sytkowski AJ, Lunn ED, Risinger MA, Davis KL. An erythropoietin
fusion protein comprised of identical repeating domains exhibits
enhanced biological properties. J Biol Chem 1999;274:2477324778.
35. Schulz G, Schrimer R. In Principles of protein structure. New York:
Springer; 1979.
36. Janin J, Clothia C. Domains in proteins: definitions, location, and
structural principles. Methods Enzymol 1985;115:420430.