Báo cáo khoa học: Autophosphorylation of heme-regulated eukaryotic initiation factor 2a kinase and the role of the modification in catalysis doc

Autophosphorylation of heme-regulated eukaryotic
initiation factor 2a kinase and the role of the
modification in catalysis
Jotaro Igarashi
1,
*, Takehiko Sasaki
1,
*, Noriko Kobayashi
2
, Shinji Yoshioka
2
, Miyuki Matsushita
2
and Toru Shimizu
1
1 Institute of Multidisciplinary Research for Advanced Materials, Tohoku University, Sendai, Japan
2 Bio-Medical Center, R&D Division, Nanotechnology Product Business Group, Hitachi High-Technologies Corporation, Hitachinaka, Ibaraki,
Japan
Introduction
Inhibition of protein synthesis or translation is impor-
tant to ensure cell survival during stress or emergency
states [1,2]. Four eukaryotic initiation factor 2a
(eIF2a)
3
kinases, heme-regulated eIF2a kinase (HRI),
PKR, PERK and GCN2, inhibit translation upon
recognition of stress or emergency states, such as heme
shortage, virus infection, accumulation of denatured
proteins or amino acid shortage, respectively. After
Keywords
autophosphorylation; eIF2a kinase; heme;

mass spectrometry; mutation
Correspondence
J. Igarashi, Institute of Multidisciplinary
Research for Advanced Materials, Tohoku
University 2-1-1 Katahira, Aoba-ku, Sendai
980-8577, Japan
Fax: +81 22 217 5605
Tel: +81 22 217 5605
E-mail:
*These authors contributed equally to this
work
(Received 2 September 2010, revised 23
November 2010, accepted 7 January 2011)
doi:10.1111/j.1742-4658.2011.08007.x
Heme-regulated eukaryotic initiation factor 2a (eIF2a) kinase (HRI),
functions in response to heme shortage in reticulocytes and aids in the
maintenance of a heme:globin ratio of 1:1. Under normal conditions, heme
binds to HRI and blocks its function. However, during heme shortage,
heme dissociates from the protein and autophosphorylation subsequently
occurs. Autophosphorylation comprises a preliminary critical step before
the execution of the intrinsic function of HRI; specifically, phosphorylation
of Ser-51 of eIF2a to inhibit translation of the globin protein. The present
study indicates that dephosphorylated mouse HRI exhibits strong intramo-
lecular interactions (between the N-terminal and C-terminal domains)
compared to phosphorylated HRI. It is therefore suggested that autophos-
phorylation reduces the intramolecular interaction, which induces irrevers-
ible catalytic flow to the intrinsic eIF2a kinase activity after heme
dissociates from the protein. With the aid of MS, we identified 33 phos-
phorylated sites in mouse HRI overexpressed in Escherichia coli. Phosphor-
ylated sites at Ser, Thr and Tyr were predominantly localized within the

kinase insertion region (16 sites) and kinase domain (12 sites), whereas the
N-terminal domain contained five sites. We further generated 30 enzymes
with mutations at the phosphorylated residues and examined their catalytic
activities. The activities of Y193F, T485A and T490A mutants were signifi-
cantly lower than that of wild-type protein, whereas the other mutant pro-
teins displayed essentially similar activity. Accordingly, we suggest that
Tyr193, Thr485 and Thr490 are essential residues in the catalysis.
Abbreviations
CID, collision-induced dissociation; ECD, electron-capture dissociation; eIF2a, eukaryotic initiation factor 2a; Ga-IMAC, gallium immobilized
metal ion affinity chromatography; GCN2, general control nonderepressible 2; HRI, heme-regulated eIF2a kinase or heme-regulated inhibitor;
KD, C-terminal kinase domain containing amino acids 145-619; KI, kinase insert with amino acids 244-371; NTD, N-terminal domain
containing amino acids 1-144; P-eIF2a, phosphorylated-eIF2a ; PERK, PKR-like endoplasmic reticulum-related kinase; PKR, double strand
RNA-dependent protein kinase.
918 FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS
sensing stress, these eIF2a kinases phosphorylate
a common substrate, eIF2a at Ser51, leading to the
termination of protein synthesis [3]. HRI regulates the
translation of hemoglobin proteins in reticulocytes in
response to heme availability [4–10]. The protein
remains inactive in the presence of sufficient heme,
thus inhibiting catalysis under normal conditions,
although it becomes active when heme dissociates from
the protein during heme shortage.
HRI is composed of an N-terminal domain (NTD)
(amino acids 1–144) (Fig. 1, red area) and a C-termi-
nal kinase domain (KD) (amino acids 145–619)
(Fig. 1, green area) including a kinase insert (KI)
region (amino acids 244–371) (Fig. 1, yellow area).
Note that the amino acid numbering is based on the
mouse HRI sequence from the present study (Fig. 1).

Heme binding ⁄ recognition sites are located both within
the NTD (His119 ⁄ His120) and the KD (Cys409), sug-
gesting that significant global structural changes (and
not simply heme binding to a protein surface patch)
are required for heme sensing by HRI [11]. The cata-
lytic activation process of HRI comprises four steps.
Specifically, in response to heme deficiency or low
heme concentration: (a) heme dissociates from the
HRI protein; (b) HRI is autophosphorylated at key
residues; (c) HRI is autophosphorylated at multiple
sites; and (d) subsequent phosphorylation at Ser51 of
eIF2a is triggered. Therefore, it is important to
explore the role of autophosphorylation of the HRI
protein in the intramolecular protein–protein interac-
tion (between the NTD and KD) and catalysis upon
heme binding. Identification of autophosphorylated
sites should further facilitate our understanding of the
molecular mechanism underlying HRI function.
As with other protein kinases, the activities of eIF2a
kinases are regulated by the initial autophosphoryla-
tion step. In addition to Ser ⁄ Thr residues, Tyr residues
are phosphorylated in the eIF2a kinases, PKR and
PERK [12,13]. To date, only a few autophosphoryla-
tion sites have been identified for eIF2a kinases. In
mouse HRI, Thr483 and Thr485 phosphorylation has
been detected using
32
P-labeling and peptide sequenc-
ing analysis, although only phosphorylated Thr485
was implicated in protein function [14]. In addition,

phosphoamino acid analysis revealed phosphorylation
at Tyr residues in mouse HRI [15]. Because eIF2a
kinases are activated via multiple autophosphorylation
steps, it is important to identify other phosphorylation
sites in HRI and to establish the contribution of the
autophosphorylation process to heme sensing and
catalysis. The results obtained should aid the clarifica-
tion of the general role of autophosphorylation in
catalysis by eIF2a kinases as well as the molecular
mechanism of stress sensing.
In the present study, we overexpressed mouse HRI
protein in Escherichia coli and examined the effects of
phosphorylation on intramolecular protein–protein
interactions, catalysis and heme sensing using both
autophosphorylated and dephosphorylated HRI
proteins. We aimed to identify autophosphorylation
sites of HRI by MS. Based on the 33 phosphorylated
sites identified using LC-MS ⁄ MS analysis, we gener-
ated 30 mutant proteins deficient in phosphorylation
and assessed their catalytic activities aiming to
determine the role of phosphorylation in catalysis.
Notably, phosphorylation at Tyr193, Thr485 and
Thr490 appeared to be critical for intrinsic eIF2a
kinase activity.
Fig. 1. Amino acid sequence and domain
architecture of HRI. HRI is composed of
an NTD (amino acids 1–144) (red area) and
a KD (amino acids 145–619) (green area)
including a KI region (amino acids 244–371)
(yellow area). The peptide fragments used

to determine phosphorylated sites are
underlined, and the phosphorylated sites
identified after trypsin digestion, Ga-IMAC
purification and LC-MS ⁄ MS detection are
indicated in red. Previously identified
phosphorylated sites are shown in bold.
J. Igarashi et al. Autophosphorylation of an HRI
FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS 919
Results and Discussion
Dephosphorylation of HRI with k phosphatase
and autophosphorylation in the presence of ATP
Because purified mouse HRI overexpressed in E. coli is
already phosphorylated, it is possible that the auto-
phosphorylation of HRI observed in the present study
was induced by expression in a heterologous host but
did not occur in eukaryotic cells. To address this issue,
we attempted to generate dephosphorylated HRI pro-
tein by treatment with k phosphatase [16]. Initially,
HRI was co-expressed with k phosphatase in E. coli.
However, the yield of dephosphorylated HRI was only
10% of that of phosphorylated HRI as a result of the
formation of inclusion bodies. Subsequently, we suc-
cessfully obtained dephosphorylated HRI protein with
k phosphatase and examined its autophosphorylation
activity. Dephosphorylation of HRI was evident from
differences in protein mobility on SDS ⁄ PAGE when
purified without k phosphatase and when k phospha-
tase-treated HRI proteins were compared (Fig. S1).
No marked differences in oligomerization status were
observed between autophosphorylated and k phospha-

tase-treated HRI. As shown in Fig. S2 (upper panel),
dephosphorylated HRI protein per se autophosphory-
lates within 1–2 min after mixing with ATP in the
absence of heme. Intrinsic HRI catalytic activity, in
terms of the phosphorylation of eIF2a, was observed
after autophosphorylation in the absence of heme
(Fig. S2, lower panel). We conclude that dephospho-
rylated HRI protein has properties similar to the
native counterpart in terms of autophosphorylation
and phosphorylation of eIF2a.
Effects of dephosphorylation on intramolecular
protein–protein interactions
We employed a pulldown assay to examine the
effects of dephosphorylation on protein–protein inter-
actions between the isolated NTD and KD using
His-tagged NTD (His6-NTD) and both phosphory-
lated and dephosphorylated KD (Fig. 2). Interactions
between NTD and KD were stronger in the presence
than the absence of heme (compare lanes 1 and 2
versus lanes 3 and 4; Fig. 2), which is consistent
with previous studies [17]. However, the interactions
between His6-NTD and dephosphorylated KD were
stronger than those between His6-NTD and the
phosphorylated domains (Fig. 2B). Analogous results
were obtained with a D440N mutant of KD (the cat-
alytic residue is mutated, thus abrogating phosphory-
lation). This finding indicates that the interactions
between NTD and KD are not attributable to the
effects of k phosphatase but rather to phosphoryla-
tion per se. In particular, heme-induced enhancement

of protein–protein interactions involving dephospho-
rylated KD was significantly higher than that
observed with phosphorylated KD. These results sug-
gest that multiple phosphorylations inhibit protein–
protein interactions between isolated NTD and KD
domains, leading to the suppression of intramolecu-
lar interactions (between N-terminal and C-terminal
domains) within the HRI protein. Moreover, the
heme-sensing ability of HRI is markedly inhibited by
multiple phosphorylations.
Effects of dephosphorylation on heme sensitivity
Next, we examined the catalytic activities of dephos-
phorylated HRI, and compared the catalytic and
heme-sensing properties of phosphorylated and
dephosphorylated HRIs. The effects of heme on
the kinase activities of dephosphorylated HRI proteins
are shown in Fig. 3. Specific activity of the eIF2a
kinase reaction of dephosphorylated HRI in the
absence of heme was 8.0 nmol phosphorylated-eIF2a
(P-eIF2a)Æmin
)1
Æmg
)1
HRI, which is significantly
lower than that of phosphorylated HRI (18 nmol
P-eIF2aÆmin
)1
Æmg
)1
HRI) [17], implying an important

role for multiple phosphorylations in HRI catalysis.
The half-inhibition constant (IC
50
s ± SD) of enzyme
activity upon heme binding to dephosphorylated HRI
1.0
Phosphatase
Heme
*
0.8
0.6
0.4
0.2
0.0
KD recovery
1234
AB
Fig. 2. Pulldown assay to detect interactions between His-tagged
NTD (His
6
-NTD) and KD. (A) SDS ⁄ PAGE band patterns reveal His
6
-
NTD and KD interactions. Lane M is the marker band. Only in the
presence of heme (lanes 2 and 4) were interactions between His
6
-
NTD and KD detected. In the absence of heme (lanes 1 and 3),
interactions between His
6

-NTD and KD were not observed. (B)
Densitometric analysis demonstrates that interactions between
His
6
-NTD and KD are influenced by phosphorylation of KD because
this binding was significantly (*; P < 0.05) stronger for dephospho-
rylated HRI (lane 4) than phosphorylated HRI (lane 2).
Autophosphorylation of an HRI J. Igarashi et al.
920 FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS
was 2.4 ± 1.4 lm, which is similar to that of phos-
phorylated HRI (2.1 lm) [17]. Because His119⁄ His120
and Cys409 are the heme-binding sites of mouse
HRI, we measured the IC
50
values of heme for the
dephosphorylated H119A ⁄ H120A and C409S mutant
proteins (Fig. 3). The IC
50
value obtained for the
dephosphorylated H119A ⁄ H120A mutant protein
(0.39 ± 1.3 lm) was 10-fold lower than that of the
phosphorylated enzyme (3.7 lm) [17]. By contrast, the
IC
50
value of the dephosphorylated C409S mutant was
5.5 ± 1.3 lm, which is comparable to that (5.1 lm)of
the phosphorylated mutant protein. Notably, the opti-
cal absorption spectra of heme-bound dephosphoryl-
ated HRI proteins were essentially similar to those of
heme-bound phosphorylated HRI (Fig. S3).

In HRI activation, the sensing of heme concentration
is critical for function. We did not detect differences in
heme sensitivity between proteins with mutations at the
heme-binding sites (His119 ⁄ His120 and Cys409) and
wild-type full-length enzyme [17]. However, when KD
(an N-terminal deleted mutant protein) was used to
examine heme sensitivity, marked differences in
heme-binding ability were evident between wild-type
and the C409S mutant protein. Specifically, the IC
50
values of wild-type and C409S KD proteins for heme
were 0.25 lm and > 10 lm, respectively [17]. In the
present study, we observed significant differences in
heme sensitivity between dephosphorylated wild-type
and mutant proteins (Fig. 3), and the IC
50
values for
dephosphorylated wild-type, H119A ⁄ H120A and
C409S HRI proteins were 2.4, 0.39 and 5.5 lm, respec-
tively. The high heme sensitivity (IC
50
= 0.39 lm)of
dephosphorylated H119A ⁄ H120A HRI is similar to
that of wild-type KD (IC
50
= 0.25 lm), suggesting that
the C-terminal domain itself has sensitivity in the sub-
micromolar range. It appears that Cys409 is necessary
for heme binding and heme sensitivity in sub-micromo-
lar range. His119 or His120 would modulate the sensi-

tivity in the low micromolar range because the IC
50
value of the dephosphorylated wild-type HRI was
increased to 2.4 lm. On the other hand, the IC
50
value
(5.5 lm) of the dephosphorylated C409S HRI was
distinct from that (> 10 lm) of the C409S KD. These
results suggest that His119 ⁄ His120 may partially
compensate for Cys409 in the dephosphorylated C409S
HRI. In addition, His119 or His120 alone is possibly
sufficient, in the dephosphorylated form, for heme bind-
ing and modulation of heme sensitivity at 5 lm, even in
the absence of Cys409.
Identification of the autophosphorylation sites of
HRI overexpressed in E. coli
To further clarify the details of autophosphorylation,
we aimed to identify phosphorylation sites in HRI
using MS. The HRI protein was initially digested with
trypsin and phosphorylated peptides recovered with
gallium immobilized metal ion affinity chromatography
(Ga-IMAC). Phosphorylated peptides were subjected
to LC linked to MS. The sequence coverage of MS
analysis was 35.8% (222 residues ⁄ 619 amino acids). A
mascot search () was
conducted to identify peptides (Fig. S4). The phosphor-
ylated sites of mouse HRI, together with the peptide
sequences employed to identify these sites and the LC-
MS ⁄ MS data, are summarized in Fig. 1 and Table 1.
In total, 33 phosphorylated sites, including 23 Ser,

seven Thr and three Tyr residues, were identified. The
number of phosphorylated sites was almost consistent
with the quantification of phosphates using BIOMOL
Specific activity
(nmol P-elF2α·min
–1
·mg
–1
HRI)
Heme (µM)
[Heme] (µ
M)
A
B
Fig. 3. (A) Phosphorylation of eIF2a (black triangle) by wild-type,
H119A ⁄ H120A and C409S dephosphorylated HRI proteins detected
via Coomassie staining of a SDS-acrylamide gel containing Phos-tag
acrylamide and manganese. The reaction was terminated at 4 min,
and phosphorylated eIF2a proteins were compared with various
heme concentrations. (B) The dose-response curve of eIF2a
kinase activities of wild-type, H119A ⁄ H120A and C409S dephos-
phorylated HRI versus log [heme]. The kinase activities of dephos-
phorylated HRI were 50% of those of phosphorylated HRI. Data
were analyzed using the equation: Y = bottom + (top ) bottom) ⁄
(1 + 10
X ) logIC
50
). The IC
50
values of wild-type, H119A ⁄ H120A and

C409S mutant proteins were 2.4, 0.39 and 5.5 l
M, respectively.
The IC
50
values of wild-type, H119A ⁄ H120A and C409S mutants
of phosphorylated HRI were reported as 2.1, 3.7 and 5.1 l
M,
respectively [17].
J. Igarashi et al. Autophosphorylation of an HRI
FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS 921
Green (Enzo Life Sciences International, Inc., Plym-
outh Meeting, PA, USA) in k phosphatase-treated HRI
protein (28 sites). In general, phosphorylated sites were
located within three domains or regions; specifically,
NTD, KD and KI. NTD contained only five sites at
Ser5, Ser6, Ser34, Ser41 and Ser125. We identified 12
sites in KD at Ser144, Tyr145, Thr160, Ser161, Tyr163,
Ser495, Thr532, Thr535, Ser545, Ser547, Ser609 and
Ser612. KI had 16 phosphorylation sites; specifically,
Ser252, Ser257, Ser273, Ser274, Ser275, Ser276,
Thr283, Ser293, Ser304, Tyr305, Thr306, Ser317,
Ser319, Ser320, Thr329 and Ser332. Interestingly, many
of the phosphorylated sites are located in KI, whereas
only five sites are phosphorylated in NTD. However,
we failed to identify Thr483 and Thr485 as phosphory-
lation sites, which is inconsistent with previous data
obtained from
32
P-labeling experiments [14]. We sus-
pect that the cysteine residues were not adequately pro-

tected in our examination of peptides containing
Thr483 and Thr485.
Multiple phosphorylations of HRI are essential for
generation of the active form. However, Thr485 is the
Table 1. Phosphorylation sites of mouse HRI identified by fragmentation analysis using LC-MS ⁄ MS.
Position Peptide sequence P-site m ⁄ zzM
r
(experimental) M
r
(calculated)
Ion
score
1–11 MLGG
SSVDGER Ser5 634.2251 2 1266.4356 1266.4353 67
Ser6
12–37 DTDDDAAGAVAAPPAIDFPAEV
SDPK Ser34 879.0528 3 2634.1366 2634.1374 52
38–51 YDE
SDVPAELQVLK Ser41 843.3845 2 1684.7544 1684.7597 70
93–100
a
LLCQTFIK Ser97
b
551.7808 2 1101.5470 1101.5294 16
123–129 AI
THLMR Thr125 461.2209 2 920.4272 920.4303 31
137–148
a
QDPCQDNSYMQK Ser144 845.2712 2 1688.5278 1688.5249 12
Tyr145

153–162 EIAFEAQ
TSR Thr160 656.2490 2 1310.4834 1310.4945 25
Ser161
163–175
YLNEFEELAILGK Tyr163 809.8862 2 1617.7578 1617.7691 89
188–196 LDGQH
YAIK Tyr193
b
562.7601 2 1123.5056 1123.5063 20
246–263 VPIQLP
SLEVLSEQEGDR Ser252 1084.9967 2 2167.9788 2167.9803 83
Ser257
270–286 DNE
SSSSIVFAELTPEK Ser273 1086.8777 2 2171.7408 2171.7391 42
Ser274 1086.8788 2 2171.7430 2171.7391 42
Ser275
Ser276
Thr283
287–296 EKPFGE
SEVK Ser293 615.2781 2 1228.5416 1228.5377 36
297–312 NENNNLV
SYTANLVVR Ser304 1030.4192 2 2058.8238 2058.8214 54
Tyr305
Thr306
313–338 NSSE
SESSIELQEDGLTDLSVRPVVR Ser317 1082.7552 3 3245.2438 3245.2257 72
Ser319
Ser320
Thr329
Ser332

488–507
a
VGTCLYASPEQLEGSQYDA Thr490
b
845.9799 3 2534.9179 2534.8756 24
Ser495
Ser502
b
Tyr504
b
531–538 ATVLTGVR Thr532 448.7290 2 895.4434 895.4528 35
Thr535 448.7307 2 895.4468 895.4528 38
542–548 IPE
SLSK Ser545 427.2018 2 852.3890 852.3994 39
542–548 IPESL
SKR Ser547 505.2566 2 1008.4986 1008.5005 27
607–615 QL
SLLSQDR Ser609 1139.5358 1 1138.5285 1138.5383 32
Ser612 570.2753 2 1138.5360 1138.5383 36
a
Detected only carbamidomethylated sample using iodoacetamide.
b
These residues are unreliable when considering the low ion score of
phosphorylated peptides and few fragment ions adjacent to phosphorylated residues; thus, we could not clearly determine whether these
residues were phosphorylated.
Autophosphorylation of an HRI J. Igarashi et al.
922 FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS
only phosphorylated residue identified to date that
might be critical for HRI activation [14]. In the present
study, we identify 33 phosphorylated sites in HRI. The

reaction with autophosphorylated HRI protein and
k phosphatase showed that 28 nmol of phosphates
were released from 1 nmol of HRI for 1 h. This indi-
cated that most of the phosphorylated sites in HRI
were identified in the present study, although the
sequence covereage was not high (36%). Interestingly,
KI contained 16 phosphorylation sites, whereas NTD
and KD had five and 12 phosphorylation sites, respec-
tively. However, mutation of phosphorylated residues
in KI resulted in proteins with catalytic activities essen-
tially similar to those of wild-type enzyme (described
below). Thus, we propose that KI is phosphorylated
principally to increase the solubility or stability of par-
ticular peptide regions in the HRI protein, although
such phosphorylation is not associated with catalytic
function. Cys409 is one of the heme binding ⁄ sensing
sites located near KI. Thus, multiple phosphorylations
of KI after heme dissociation possibly prevent protein
interactions with heme and promote the intrinsic eIF2a
kinase activity.
Effects of mutations at phosphorylated Ser, Thr
and Tyr on catalysis
To determine the detailed significance of HRI auto-
phosphorylation with respect to catalysis, we generated
30 mutations at phosphorylated Ser, Thr and Tyr resi-
dues and examined the catalytic activities of the
mutant proteins (Fig. 4). Interestingly, the activities of
Y193F, T485A and T490A mutant proteins were sig-
nificantly lower than that of wild-type protein (Fig. 4).
Although the K196R (lacking ATP binding) and

T490A mutants showed very low activity, the effects of
the Y193F and T485A mutations were less substantial.
This appeared to reflect differences in HRI autophos-
phorylation status, in that some (but not all) Y193F
and T485A mutant protein could autophosphorylate in
a manner similar to that of wild-type protein (Fig. S5).
In an effort to mimic phosphorylated Thr490, the cata-
lytic activity of the T490D mutant protein was
assayed. The activity was almost the same as that of
T490A, and thus significantly lower than that of wild-
type protein. This indicates that phosphorylation of
Thr490 is an early key event in the autophosphoryla-
tion of HRI.
Tyr193 lies within catalytic subdomain II and is
highly conserved among all eIF2a kinases (Tyr293 in
PKR and Tyr-615 in PERK) (Fig. 5A) [18–20]. Given
the function of highly conserved Tyr residues in cataly-
sis, it is reasonable to assume that Tyr193 plays a criti-
cal role in HRI activity. On the basis of a structural
model, we speculate that Tyr193 is located at the
dimer interface and is thus trans-phosphorylated by
the kinase domains of other subunits of HRI (Fig. 5B).
Unfortunately, because of the low ion score (i.e. a
measure of how well the observed MS ⁄ MS spectrum
matches that of the relevant peptide) of the phosphory-
lated peptide, as well as the availability of only few
fragment ions adjacent to Tyr193 (Fig. S4), Tyr193
phosphorylation status in mouse HRI remains unclear.
Thr485 and Thr490 are critical for HRI activation.
On the basis of amino acid alignments (Fig. 5A), we

propose that Thr485 and Thr490 are located within
the activation loop between subdomains VII and VIII.
Thr485 is well conserved in the eIF2a kinases and cor-
responds to Thr446 of PKR, Thr981 of PERK and
Thr899 of GCN2 [18–20]. By contrast to earlier studies
Specific activity (nmol
P-elF2α·min
–1
·mg
–1
HRI)
Fig. 4. In vitro kinase activities of wild-type and mutant proteins. SDS ⁄ PAGE of Phos-tag acrylamide and manganese used for the evaluation
of phosphorylated eIF2a protein (Fig. S6A, B). The K196R mutation is within the ATP binding site (Fig. 5B), and thus the mutant protein dis-
plays very low activity. K196R activity was used as a negative control. Experiments were repeated at least three times, and are shown as
the mean ± SD. Note that the S502A mutant protein showed very low activity (Fig. S6C). However, the protein band on the SDS ⁄ PAGE
was largely downshifted as a result of proteolysis (Fig. S6D). By contrast, the S502D mutant was expressed normally and displayed activity
similar to the wild-type protein.
J. Igarashi et al. Autophosphorylation of an HRI
FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS 923
showing that Thr483 and Thr485 are phosphorylated
[14], we did not detect phosphorylation at both sites.
However, the T485A mutant protein showed signifi-
cantly lower activity than the wild-type counterpart
(Fig. 4), which is consistent with the previous proposal
that Thr485 of HRI is critical for catalysis [14].
Thr490 of mouse HRI is also well conserved and cor-
responds to Thr451 of PKR, Thr986 of PERK and
Thr904 of GCN2 (Fig. 5A) [18–20]. The results
obtained in the present study suggest that Thr490 is
phosphorylated and essential in catalysis (Table 1 and

Fig. 4).
Regulatory mechanism of HRI
Figure 6 shows a schematic representation of the regu-
latory mechanism of HRI. On the basis of results
obtained in a previous study, we propose that global
protein rearrangements, including intramolecular
protein–protein interactions, are required for heme
sensing because the axial ligands of the heme, His119 ⁄
His120 in the N-terminal domain and Cys409 in the
C-terminal domain, are heme-sensing sites [17]. This
heme-sensing function is operative under normal con-
ditions when heme concentrations are physiologically
sufficient. However, under conditions of heme short-
age, heme dissociates from the HRI protein, followed
by protein autophosphorylation. It is currently unclear
why autophosphorylation is required, although an
increase in protein solubility and structural alterations
induced by autophosphorylation may be important in
catalysis. Data from the present study confirm a previ-
ous suggestion that Thr485 is involved in catalysis. We
show, for the first time, that Tyr193, Thr485 and
Thr490 are key residues for multiple phosphorylations,
and that dephosphorylation enhances intramolecular
interactions between the N- and C-terminal domains
of HRI critical for heme sensing, and between heme
and protein. Therefore, multiple autophosphorylation
steps are possibly important for preventing heme
rebinding, inhibiting interactions between heme and
protein after dissociation, and shifting the equilibrium
of the biochemical reaction irreversibly towards phos-

phorylation of eIF2a.
Before intrinsic eIF2a kinase activity is acquired,
heme must initially dissociate from the HRI binding site
composed of His119 ⁄ His120 and Cys409 because auto-
phosphorylation and eIF2a kinase activities are com-
pletely inhibited by 10 lm heme (Fig. S2). Once heme is
released from HRI, the intramolecular interactions
A
B
Fig. 5. (A) Amino acid alignment of the
kinase domains of human PKR, PERK,
GCN2 and mouse HRI. Phosphorylated sites
in the subdomain II and activation loop
between subdomains VII and VIII are shown
in bold. Phosphorylation sites determined in
the present study are shown in red, and the
residues important for catalysis identified
in the present study are underlined.
(B) Homology model of the kinase domain
of HRI based on human PKR data (Protein
data bank code: 2A19). HRI sequence
missing in PKR, key autophosphorylated
residues (Tyr193, Thr485 and Thr490), as
well as Lys196 mediating ATP binding, are
shown in magenta, cyan and violet,
respectively.
Autophosphorylation of an HRI J. Igarashi et al.
924 FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS
between NTD and KD become weak (Fig. 2, lanes 1
and 3), after which autophosphorylation takes place as

the second step. Intramolecular interactions are further
decreased and the enzyme becomes active. Activated
autophosphorylated HRI phosphorylates eIF2a at
Ser51 in the fourth step (Fig. 6 and S2).
Large-scale proteomics studies in HeLa cells indicate
that human HRI has eight potential phosphorylated
sites [21,22]. These residues, corresponding to those in
the mouse sequence, are phosphorylated. A number of
these residues contain the phosphorylation motif
specific for casein kinase 1 (S-X-X-S ⁄ T), casein
kinase 2 (S ⁄ T-X-X-E) and glycogen synthase kinase 3
(S-X-X-X-S). Phosphorylation of HRI by casein kinase
2 has been described previously [23]. These findings
suggest that other kinases, including casein kinase 1,
casein kinase 2 and glycogen synthase kinase 3,
phosphorylate HRI, which functions upstream of these
signaling cascades.
Conclusions
We demonstrate that autophosphorylation of mouse
HRI occurring after heme dissociation is important for
weakening intramolecular interactions between the
N-terminal and C-terminal domains critical for heme
sensing, thus confirming the importance of the heme
binding site, His-119, or His-120 and Cys-409, for
heme sensing. We identified 33 autophosphorylated
sites using MS, and identified Tyr193, Thr485 and
Thr490 as important residues in catalysis based on
site-directed mutagenesis of the phosphorylation sites.
Experimental procedures
Materials

Phos-tag acrylamide was purchased from Phos-tag Consor-
tium Co. (Osaka, Japan). Oligonucleotides were obtained
from Nihon Gene Research Laboratories (Sendai, Japan).
k Protein phosphatase was purchased from New England
Biolabs Japan (Tokyo, Japan). Other reagents were pur-
chased from Wako Pure Chemical Industries (Osaka,
Japan). Reagents were of the highest commercial grade
available and used without further purification.
Site-directed mutagenesis, protein expression
and purification
Mutagenesis was conducted using the QuikChange site-
directed mutagenesis kit obtained from Stratagene (La
Jolla, CA, USA). The oligonucleotide sequences are sum-
marized in Table S1. Mutations were confirmed by DNA
sequencing.
Mouse HRI was overexpressed in the E. coli strain
BL21(DE3) Codon Plus RIL (Stratagene). Protein expression
was induced with 50 lm isopropyl-b-d-galactoside, as
described previously [11,17,24]. The protein purification pro-
cedures for full-length and N-terminal truncated KD mutant
(amino acids 1–145 of full-length HRI were deleted) HRI
and NTD proteins were similar to those described previously
[11,17,24]. To generate dephosphorylated protein, HRI was
treated with k protein phosphatase for 3 h at 4 ° C after
cleavage of the His-tag. The term ‘phosphorylated protein
(full-length or KD)’ refers to an enzyme prepared without
the use of k protein phosphatase, whereas ‘dephosphorylated
Fig. 6. Hypothetical regulatory mechanism of HRI. Heme association ⁄ dissociation at the heme-sensing site of HRI regulates the eIF2a
kinase reaction. (1) Heme association with full-length HRI blocks catalysis, whereas heme dissociation exposes the active site and permits
catalysis. (2) Next, HRI is autophosphorylated at key residues, including Tyr193, Thr485 and Thr490. (3) Thereafter, HRI autophosphorylates

multiple residues. (4) Finally, HRI becomes an active eIF2a kinase, phosphorylating the substrate at Ser51.
J. Igarashi et al. Autophosphorylation of an HRI
FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS 925
protein (full-length or KD)’ refers to an enzyme treated with
k protein phosphatase. His6-tagged eIF2a expression and
purification is described elsewhere [11].
Enzyme assay and pulldown assay
In vitro kinase assays were performed in according with pre-
vious studies [17]. Briefly, the reaction mixture, consisting of
20 mm Tris ⁄ HCl (pH 7.7), 60 mm KCl, 2 mm magnesium
acetate, 0.35 lm HRI and 10 lg of eIF2a, was incubated at
15 °C for 5 min, and the reaction initiated by adding 50 lm
ATP at 15 °C. At the indicated times, the reaction was termi-
nated by adding sample buffer, and heat-denatured at 95 °C
for 5 min. Samples were loaded on a 7.5% SDS gel contain-
ing 50 lm Phos-tag acrylamide and 100 lm manganese chlo-
ride. The specific activities were calculated from a time
course experiment. P-eIF2a interacted with the Phos-tag
manganese complex so that mobility was slower than that of
eIF2a. A pulldown assay using His6-tagged NTD and KD
on Ni nitrilotriacetic acid-agarose column (Qiagen KK,
Tokyo, Japan) has been described previously [17]. Briefly,
His6-NTD (amino acids 1–138) and phoshorylated KD or k
protein phosphatase-treated KD (amino acids 146–619) pro-
teins were mixed with each other (100 pmol each) in the pres-
ence or absence of hemin (100 pmol). The mixture was
loaded on to the Ni-Sepharose 6 column (GE Healthcare,
Little Chalfont, UK). Next, the column was washed with
20 mm Tris ⁄ HCl (pH 8.0), 150 mm NaCl and 100 mm imid-
azole, and His6-tagged protein eluted with 20 mm Tris ⁄ HCl

(pH 8.0), 150 mm NaCl and 200 m m imidazole. The input
sample, washed flow-through and elution fractions were ana-
lyzed by SDS ⁄ PAGE. Proteins were visualized with Phastgel
Coomassie Brilliant Blue R350 (GE Healthcare) staining.
Gel images were acquired using LAS-3000 (Fujifilm, Tokyo,
Japan) and quantified with multi-gauge software (Fujifilm).
To calculate specific activity and KD recovery values, the
ratios of P-eIF2a over eIF2a, and KD over His6-NTD were
obtained from densitometric analysis, respectively. Nonlin-
ear fitting to obtain the IC
50
value was conducted using
prism 5 (GraphPad Software, La Jolla, CA, USA).
Trypsin digestion and phosphopeptide
purification
Full-length HRI protein was degraded using immobilized
l-1-tosylamido-2-phenylethyl chloromethyl ketone-treated
trypsin obtained from Pierce (Thermo Scientific, Rockford,
IL, USA) in accordance with the manufacturer’s instruc-
tions. Purified proteins (100 lg) were dissolved in 50 lLof
0.1 m NH
4
HCO
3
(pH 8.0). Immobilized l-1-tosylamido-2-
phenylethyl chloromethyl ketone-treated trypsin (20 lL)
was added to protein samples. The reaction mixture was
incubated at 37 °C for 20 h. Trypsin beads were separated
from the digestion mixture by centrifugation. Cys residues
were protected using iodoacetamide before trypsin digestion.

Phosphopeptide purification was conducted using a
specific phosphopeptide isolation kit from Pierce (Thermo
Scientific) with Ga-IMAC, in accordance with the manufac-
turer’s instructions.
LC
We performed phosphopeptide separation and concentration
with Nano-LC using a capillary column (diameter 75 mm,
length 100 mm; column packing 3 lm) (Nikkyo Technos
Co., Tokyo, Japan) with a flow rate of 200 nLÆmin
)1
. For
separation, solvent A (2% acetonitrile in 0.1% formic acid)
plus 5% solvent B (98% acetonitrile in 0.1% formic acid)
(v ⁄ v), followed by a linear gradient up to 40% solvent B for
60 min, was applied.
Collision-induced dissociation (CID) and
electron-capture dissociation (ECD) LIT q-TOF MS
and MS
⁄
MS analyses
MS was performed at Hitachi High Technologies Corpora-
tion (Tokyo, Japan). Analytical conditions for the CID
measurements were: MS mode, trap mode; ionization, ESI
(positive ionization); nebulizer gas flow, 0.8 LÆmin
)1
; spray
potential, 1500 V; detector potential, 2050 V; scan range,
m ⁄ z 50–2000. Analytical conditions for the ECD measure-
ments were: MS mode, trap mode; ionization, ESI (positive
ionization); nebulizer gas flow, 0.6 LÆmin

)1
; spray potential,
1500 V; detector potential, 2150 V; scan range, m ⁄ z 50–
2000. MS ⁄ MS data were analyzed using mascot software.
Details of the analyses are summarized in Fig. S4.
Acknowledgements
This work was supported in part by Grants-in-Aid for
Scientific Research to J.I. (2177130 and 21117501) and
T.S. (17101002), and Special Education and Research
Expenses from the Ministry of Education, Culture,
Sports, Science, and Technology of Japan.
References
1 Dever TE (2002) Gene-specific regulation by general
translation factors. Cell 108, 545–556.
2 Ron D & Harding HP (2007) eIF2a phosphorylation in
cellular stress responses and disease. In Translational Con-
trol in Biology and Medicine (Mathews MB, Sonenberg N
& Hershey JWB eds), pp. 345–368. Cold Spring Harbor
Laboratory Press, Cold Spring Harbor, NY.
3 Dever TE, Dar AC & Sicheri F (2007) The eIF2a
kinases. In Translational Control in Biology and
Medicine (Mathews MB, Sonenberg N & Hershey JWB
eds), pp 319–344. Cold Spring Harbor Laboratory
Press, Cold Spring Harbor, NY.
Autophosphorylation of an HRI J. Igarashi et al.
926 FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS
4 Chen J-J & London IM (1995) Regulation of protein
synthesis by heme-regulated eIF-2a kinase. Trends
Biochem Sci 20, 105–108.
5 Chen J-J (2000) Heme-regulated eIF2a kinase. In

Translational Control of Gene Expression (Sonenberg N,
Hershey JWB & Mathews MB eds), pp 529–546. Cold
Spring Harbor Laboratory Press, Cold Spring Harbor,
NY.
6 Chen J-J (2007) Regulation of protein synthesis by the
heme-regulated eIF2a kinase: relevance to anemias.
Blood 109, 2693–2699.
7 Berlanga J, Rivero D, Martı
´
n R, Herrero S, Moreno
S & de Haro C (2010) The role of the mitogen-
activated protein kinase Sty1 in regulation of eIF2a
kinases in response to environmental stress in
Schizosaccharomyces pombe. Eukaryot Cell 9, 194–
207.
8 Berlanga JJ, Herrero S & de Haro C (1998) Character-
ization of the hemin-sensitive eukaryotic initiation
factor 2a kinase from mouse nonerythroid cells. J Biol
Chem 273, 32340–32346.
9 de Haro C, Me
´
ndez R & Santoyo J (1996) The eIF-2a
kinases and the control of protein synthesis. FASEB J
10, 1378–1387.
10 Me
´
ndez R, Moreno A & de Haro C (1992) Regulation
of heme-controlled eukaryotic polypeptide chain initia-
tion factor 2 a-subunit kinase of reticulocyte lysates.
J Biol Chem 267, 11500–11507.

11 Igarashi J, Sato A, Kitagawa T, Yoshimura T, Yamau-
chi S, Sagami I & Shimizu T (2004) Activation of
heme-regulated eukaryotic initiation factor 2a kinase by
nitric oxide is induced by the formation of a five-coordi-
nate NO-heme complex: optical absorption, electron
spin resonance, and resonance raman spectral studies.
J Biol Chem 279, 15752–15762.
12 Su Q, Wang S, Gao H, Kazemi S, Harding HP, Ron D
& Koromilas AE (2008) Modulation of the eIF2a
kinase PERK by tyrosine phosphorylation. J Biol Chem
283, 469–475.
13 Su Q, Wang S, Baltzis D, Qu L-K, Wong AH-T &
Koromilas AE (2006) Tyrosine phosphorylation acts as
a molecular switch to full-scale activation of the eIF2a
RNA-dependent protein kinase. Proc Natl Acad Sci
USA 103, 63–68.
14 Rafie-Kolpin M, Han A-P & Chen J-J (2003)
Autophosphorylation of threonine 485 in the activation
loop is essential for attaining eIF2a kinase activity of
HRI. Biochemistry 42, 6536–6544.
15 Bauer BN, Rafie-Kolpin M, Lu L, Han A & Chen
J-J (2001) Multiple autophosphorylation is essential
for the formation of the active and stable homodimer
of heme-regulated eIF2a kinase. Biochemistry 40,
11543–11551.
16 McKenna SA, Lindhout DA, Shimoike T & Puglisi JD
(2007) Biophysical and biochemical investigations of
dsRNA-activated kinase PKR.
Methods Enzymol 430,
373–396.

17 Igarashi J, Murase M, Iizuka A, Pichierri F,
Martinkova M & Shimizu T (2008) Elucidation of the
heme-binding site of heme-regulated eIF2a kinase
(HRI) and the role of the regulatory motif in heme
sensing by spectroscopic and catalytic studies of mutant
proteins. J Biol Chem 283, 18782–18791.
18 Lu J, O’Hara EB, Trieselmann BA, Romano PR & Dever
TE (1999) The interferon-induced double-stranded
RNA-activated protein kinase PKR will phosphorylate
serine, threonine, or tyrosine at residue 51 in eukaryotic
initiation factor 2a. J Biol Chem 274, 32198–32203.
19 Harding HP, Zhang Y & Ron D (1999) Protein
translation and folding are coupled by an endoplasmic-
reticulum-resident kinase. Nature 397, 271–274.
20 Romano PR, Garcia-Barrio MT, Zhang X, Wang Q,
Taylor DR, Zhang F, Herring C, Mathews MB, Qin J
& Hinnebusch AG (1998) Autophosphorylation in the
activation loop is required for full kinase activity
in vivo of human and yeast eukaryotic initiation factor
2a kinases PKR and GCN2. Mol Cell Biol 18, 2282–
2297.
21 Daub H, Olsen JV, Bairlein M, Gnad F,
Oppermann FS, Ko
¨
rner R, Greff Z, Ke
´
ri G, Stemmann
O & Mann M (2008) Kinase-selective enrichment
enables quantitative phosphoproteomics of the kinome
across the cell cycle. Mol Cell 31, 438–448.

22 Olsen JV, Blagoev B, Gnad F, Macek B, Kumar C,
Mortensen P & Mann M (2006) Global, in vivo, and
site-specific phosphorylation dynamics in signaling
networks. Cell 127, 635–648.
23 Me
´
ndez R & de Haro C (1994) Casein kinase II is
implicated in the regulation of heme-controlled transla-
tional inhibitor of reticulocyte lysates. J Biol Chem 269,
6170–6176.
24 Miksanova M, Igarashi J, Minami M, Sagami I,
Yamauchi S, Kurokawa H & Shimizu T (2006)
Characterization of heme-regulated eIF2a kinase: roles
of the N-terminal domain in the oligomeric state, heme
binding, catalysis, and inhibition. Biochemistry 45,
9894–9905.
Supporting information
The following supplementary material is available:
Fig. S1. SDS ⁄ PAGE to demonstrate homogeneity of
the purified protein and dephosphorylation with k
phosphatase.
Fig. S2. Autophosphorylation of dephosphorylated
HRI monitored using western blotting with anti-HRI
serum and phosphorylation of eIF2a by HRI detected
via Coomassie staining of a SDS-acrylamide gel con-
taining Phos-tag acrylamide and manganese.
J. Igarashi et al. Autophosphorylation of an HRI
FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS 927
Fig. S3. Optical absorption spectra of heme-bound
phosphorylated and dephosphorylated HRI proteins.

Fig. S4. mascot analyses for LS-MS ⁄ MS data for
autophosphorylated peptide fragments.
Fig. S5. Phosphorylation status of HRI detected via
Coomassie staining of an SDS-acrylamide gel contain-
ing Phos-tag acrylamide and manganese.
Fig. S6. SDS ⁄ PAGE containing Phos-tag to demon-
strate catalytic activities towards eIF2a of the mutant
proteins.
Table S1. Oligonucleotides used for the site-directed
mutagenesis experiments.
This supplementary material can be found in the
online version of this article.
Please note: As a service to our authors and readers,
this journal provides supporting information supplied
by the authors. Such materials are peer-reviewed and
may be re-organized for online delivery, but are not
copy-edited or typeset. Technical support issues arising
from supporting information (other than missing files)
should be addressed to the authors.
Autophosphorylation of an HRI J. Igarashi et al.
928 FEBS Journal 278 (2011) 918–928 ª 2011 The Authors Journal compilation ª 2011 FEBS