The Conse ed Helix C Region in he Supe amily o In e e on-
␥
/
In e leukin-10- ela ed Cy okines Co esponds o a High-a ini y
Binding Si e o he HSP70 Chape one DnaK*
Recei ed o publica ion, Ma ch 27, 2002
Published, JBC Pape s in P ess, Ap il 22, 2002, DOI 10.1074/jbc.M202984200
Koen Vandenb oeck‡§, I aide Alloza‡, Di k B ehme ¶, Al ons Billiau储, Paul P oos 储**,
Neil McFe an‡‡, S e an Ru¨ dige §§¶¶, and B ian Walke ‡
F om he ‡Biomolecula Sciences Resea ch G oup, McClay Resea ch Cen e o Pha maceu ical Sciences,
and he ‡‡Cen e o P o ein and Pep ide Enginee ing, Queen’s Uni e si y o Bel as , Bel as BT9 7BL, Uni ed Kingdom,
he ¶Ins i u e o Biochemis y and Molecula Biology, Uni e si y o F eibu g, F eibu g 79104, Ge many, he 储Rega
Ins i u e o Medical Resea ch, Uni e si y o Leu en, Leu en B-3000, Belgium, and he §§Camb idge Cen e o P o ein
Enginee ing, Medical Resea ch Council Cen e, Camb idge CB2 2QH, Uni ed Kingdom
HSP70 chape ones media e p o ein olding by ATP-de-
penden in e ac ion wi h sho linea pep ide segmen s
ha a e exposed on un olded p o eins. The mode o ac ion
o he Esche ichia coli homolog DnaK is ep esen a i e o
all HSP70 chape ones, including he endoplasmic e icu-
lum a ian BiP/GRP78. DnaK has been shown o be e -
ec i e in assis ing e olding o a wide a ie y o p oka y-
o ic and euka yo ic p o eins, including he
␣
-helical
homodime ic sec e o y cy okine in e e on-
␥
(IFN-
␥
). We
sc eened solid-phase pep ide lib a ies om human and
mouse IFN-
␥
o iden i y DnaK-binding si es. Conse ed
DnaK-binding si es we e iden i ied in he N- e minal hal
o helix B and in he C- e minal hal o helix C, bo h o
which a e loca ed a he IFN-
␥
dime in e ace. Soluble
pep ides de i ed om helices B and C bound DnaK wi h
high a ini y in compe i ion assays. No DnaK-binding
si es we e ound in he loops connec ing he
␣
-helices. The
helix C DnaK-binding si e appea s o be conse ed in
mos membe s o he supe amily o in e leukin (IL)-10-
ela ed cy okines ha comp ises, apa om IL-10 and
IFN-
␥
, a se ies o ecen ly disco e ed small sec e o y p o-
eins, including IL-19, IL-20, IL-22/IL-TIF, IL-24/MDA-7
(melanoma di e en ia ion-associa ed gene), IL-26/AK155,
and a numbe o i al IL-10 homologs. These cy okines
belong o a ela i ely small g oup o homodime ic p o-
eins wi h highly in e digi a ed in e aces ha exhibi
he s ongly hyd ophobic cha ac e o he in e io co e o
a single-chain olded domain. We p opose ha binding o
DnaK o helix C in he supe amily o IL-10- ela ed cy o-
kines may cons i u e he hallma k o a no el conse ed
egula o y mechanism in which HSP70-like chape ones
assis in he o ma ion o a hyd ophobic dime ic “ olding”
in e ace.
In e e on-
␥
(IFN-
␥
)
1
is a small p o ein sec e ed by ac i a ed
T lymphocy es and na u al kille cells. I displays a numbe o
immuno egula o y ac i i ies such as ac i a ion o mac ophages
and clonal expansion o CD4
⫹
Th1 lymphocy es, oge he e-
sul ing in a Th1-biased immune esponse (1). IFN-
␥
exe s
hese ac i i ies h ough in e ac ion wi h and agg ega ion o a
cellula high-a ini y ecep o consis ing o IFN-
␥
ecep o -
␣
and ecep o -

chains (2, 3). Ealick e al. (4) and Samudzi e al.
(5) de e mined he c ys al s uc u es o ecombinan o ms o
human and abbi IFN-
␥
, espec i ely. These s udies e ealed
␣
-helical homodime s, he domains o which consis o six
igh ly packed
␣
-helices connec ed by ela i ely sho u ns
wi h no

-s uc u e. Fou o hese helices a e dona ed by one
pep ide chain and wo om he second chain, oge he cons i-
u ing a highly in e wined homodime ha is eminiscen o
domain swapping (6–8). In pa icula , he segmen helix B-
loop II-helix C om one monome in e locks wi h ex ensi e
in e digi a ion wi h he helix E-loop V-helix F segmen o he
o he monome (4, 5). This dime in e ace comp ises 35–40%
o he o al su ace a ea o IFN-
␥
and is p edominan ly hyd o-
phobic (4, 5, 9). Taken oge he , hese s uc u al ea u es clas-
si y IFN-
␥
in o a small g oup o homodime ic p o eins wi h
hyd ophobic, so-called “ olding” in e aces, which a e p esumed
o adop hei s uc u e in one s ep, i.e. bo h chains old di ec ly
in o he dime (10).
I is no known whe he he de no o olding o IFN-
␥
in o his
complex in e wined homodime s uc u e is a spon aneous
p ocess, a he han being s ic ly dependen on he assis ance
o chape ones. Ci cums an ial obse a ions p o ide suppo o
an assis ed p ocess. Fi s , IFN-
␥
is s uc u ally uns able in
ha i displays a well documen ed endency o agg ega e upon
exposu e o low pH o ele a ed empe a u es (11, 12). In i o
agg ega ion o IFN-
␥
is an i e e sible p ocess ha ollows
i s -o de kine ics and is hough o occu h ough a pa ially
olded in e media e s uc u e (12) o ia a ansien ly ex-
panded na i e species (9, 13). I is likely ha such agg ega ion-
p one con o me s occu also du ing olding in i o, hus in e -
ing he need o a biological mechanism o acili a ing co ec
olding and o limi ing agg ega ion. Second, when exp essed in
Esche ichia coli, he p o ein o ms inclusion bodies om which
bioac i e p o ein can be e olded a he expense o compe ing
* This wo k was suppo ed in pa by he Allen J. McClay T us
(Bel as , Uni ed Kingdom) and by he Fund o Scien i ic Resea ch o
Belgium (Fonds oo We enschappelijk Onde zoek-Flande s). The
32
-
Q132-Q144-C-IAANS compe i ion assay was pe o med by D. B. in he
labo a o y o Be nd Bukau (suppo ed by he Deu sche Fo schungsge-
meinscha , G aduie enkolleg Biochemie de Enzyme). The cos s o
publica ion o his a icle we e de ayed in pa by he paymen o page
cha ges. This a icle mus he e o e be he eby ma ked “ad e isemen ”
in acco dance wi h 18 U.S.C. Sec ion 1734 solely o indica e his ac .
§ To whom co espondence should be add essed: Cy okine Biology
and Gene ics P og amme, McClay Resea ch Cen e, School o Pha -
macy, QUB, 97 Lisbu n Rd., Bel as BT9 7BL, UK. Tel.: 44-2890-
272214; Fax: 44-2890-247794, E-mail: [email p o ec ed].
** Recipien o a pos doc o al esea ch ellowship om he Fonds oo
We enschappelijk Onde zoek-Flande s.
¶¶ Suppo ed by a Ma ie Cu ie ellowship.
1
The abb e ia ions used a e: IFN-
␥
, in e e on-
␥
; HSP70, gene ic
e m o all membe s o he 70-kDa hea shock p o ein amily; IL,
in e leukin; PVDF, poly inylidene di luo ide; RCMLA, educed and
ca boxyme hyla ed lac albumin; IAANS, 2-(4-iodoace amido)anilinon-
aph halene-6-sul onic acid; HIV, human immunode iciency i us.
THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 277, No. 28, Issue o July 12, pp. 25668–25676, 2002
© 2002 by The Ame ican Socie y o Biochemis y and Molecula Biology, Inc. P in ed in U.S.A.
This pape is a ailable on line a h p://www.jbc.o g25668
This is an Open Access a icle unde he CC BY license.
i e e sible agg ega ion (14, 15). In i o e olding condi ions
ha esemble he physicochemical en i onmen o he na u al
olding en i onmen o IFN-
␥
,i.e. he endoplasmic e iculum,
a e no pe missi e o assembly o IFN-
␥
dime s, bu esul in
agg ega ion (16). Thus, physiological empe a u es and high
concen a ions o olding in e media es o IFN-
␥
appea o
coun e ac spon aneous dime o ma ion. Thi d, suppo o a
ole o he endoplasmic e iculum DnaK homolog BiP/GRP78 in
he de no o olding o IFN-
␥
comes om expe imen s in which
IFN-
␥
-p oducing Chinese hams e o a y cells we e ans ec ed
wi h a BiP ATPase mu an (T37G) ha was p e iously shown
o be able o bind pep ide, bu o be unable o elease i upon
binding o ATP (17). In co-immunop ecipi a ion expe imen s,
mu an (bu no wild- ype) BiP o med a s able complex wi h
IFN-
␥
. In addi ion, IFN-
␥
sec e ion was educed om cells
ans ec ed wi h T37G BiP.
2
To un a el he mechanis ic basis o how chape ones may
assis olding o IFN-
␥
, we ha e ecen ly conduc ed a se ies o
expe imen s using he E. coli chape ones G oEL/ES and DnaK/
DnaJ/G pE (16, 18). Bo h hese chape one sys ems ha e been
ho oughly cha ac e ized, and he de e minan s o in e ac ion
wi h p o ein subs a es a e known. The DnaK chape one and
i s HSP70 homologs, including he endoplasmic e iculum a -
ian BiP/GRP78, bind p e e en ially o s e ches o hyd ophobic
amino acids ha a e usually bu ied in he co e o comple ely
olded p o eins, bu exposed on un olded o pa ially olded
p o eins ( o e iew, see Re . 19). This in e ac ion p e en s
“o -pa hway”mis olding no mally leading o agg ega ion.
Folding esumes a e ATP-media ed subs a e elease om
he chape one, a eac ion ha is ca alyzed by co ac o s such as
he nucleo ide exchange ac o G pE. The mechanism o ac ion
o G oEL is di e en in ha exposu e o domains made up by
␣
- olds wi h ex ensi e hyd ophobic su aces is hough o be
he d i ing o ce o in e ac ion (20).
We assessed in e ac ion o IFN-
␥
wi h bo h chape one sys-
ems using ecombinan IFN-
␥
samples ha we e p e iously
dena u ed by ei he acid o hea (50 °C). Fo ma ion o e ia y
s uc u e was sco ed by a combina ion o me hods, including
enzyme-linked immunoso ben assay, an i i al bioassay, and
SDS-PAGE. Unde nonpe missi e condi ions o egain o e -
ia y s uc u e (low ionic s eng h, 35 °C), he yield o co ec ly
olded dime ic IFN-
␥
inc eased ⬃30- old wi h G oEL/ES and
⬃20- old wi h DnaK/DnaJ/G pE abo e he le el o spon aneous
e olding s a ing om acid-un olded p o ein (16, 18). In addi-
ion, G oEL/ES was shown o p o ec IFN-
␥
agains he mal
inac i a ion a 50 °C by binding o and a es ing he agg ega-
ion o a he mal un olding in e media e. This sugges s ha
bo h E. coli chape one sys ems can assis olding o IFN-
␥
in
much he same way as hey do wi h physiological subs a es,
i.e. by in e ac ing wi h agg ega ion-p one con o me s.
To u he model he in e ac ion be ween HSP70 chape ones
and IFN-
␥
, we ha e now sc eened cellulose-bound pep ide li-
b a ies de i ed om human and mouse IFN-
␥
o iden i ica ion
o DnaK-binding si es. This app oach has been used success-
ully in he iden i ica ion o DnaK-binding mo i s om a a i-
e y o p o eins (21, 22). Pu a i e binding si es we e con i med
by compe i ion assays using soluble pep ides. We epo he
iden i ica ion o conse ed unc ional DnaK-binding si es in
␣
-helices B and C o human and mouse IFN-
␥
. These mo i s a e
loca ed a he dime in e ace o IFN-
␥
. The helix C mo i is
speci ically conse ed among membe s o he supe amily o
in e leukin (IL)-10- ela ed cy okines. Ru¨dige e al. (22, 23)
only a ely ound DnaK-binding si es wi hin sequences o
dime iza ion in e aces ha adop helical s uc u es in he
na i e s a e. Ou da a he e o e indica e ha binding o DnaK
o he helix C mo i may be ep esen a i e o an uncommon
conse ed egula o y mechanism in which HSP70-like chape -
ones media e o ma ion o he in e digi a ed hyd ophobic old-
ing in e ace in his class o homodime ic cy okines.
EXPERIMENTAL PROCEDURES
Sc eening o DnaK Binding o IFN-
␥
-de i ed Cellulose-bound Pep-
ides by F ac iona ed Elec oblo ing—Pep ide lib a ies consis ing o 44
and 41 13-me pep ides co e ing he en i e ma u e p o ein po ions o
human and mouse IFN-
␥
, espec i ely, we e p epa ed by au oma ed
spo syn hesis (24, 25). Pep ides we e a ached h ough hei C e mini
o a cellulose memb ane ia a (

-Ala)
2
space (Je ini Bio Tools GmbH,
Be lin, Ge many). Analysis o DnaK binding was ca ied ou essen ially
as desc ibed p e iously (21, 22). B ie ly, pep ide memb anes we e al-
lowed o eac wi h 100 nMDnaK (S essgen Bio ech Co p.) in bu e
con aining 31 mMT is-HCl, pH 7.6, 170 mMNaCl, 6.4 mMKCl, 0.05%
( / ) Tween 20, and 5.0% (w/ ) suc ose o 1ha 25°C wi h gen le
shaking. Unbound DnaK was emo ed by washing h ee imes wi h
T is-bu e ed saline (31 mMT is-HCl, pH 7.6, 170 mMNaCl, and 6.4 mM
KCl) a 4 °C. Elec o ans e o pep ide-bound DnaK on o poly inyli-
dene di luo ide (PVDF) memb anes was pe o med using a semid y
blo e . Cellulose and PVDF memb anes we e sandwiched be ween
blo ing pape soaked in anode bu e (anode bu e I: 30 mMT is base
and 20% MeOH; anode bu e II: 300 mMT is base and 20% MeOH) and
ca hode bu e (25 mMT is base, 40 mM6-aminohexane acid, and 20%
MeOH supplemen ed wi h inc easing concen a ions o SDS). Elec o-
ans e was pe o med a a cons an powe o 0.8 mA/cm
2
. Ca hode
bu e was supplemen ed successi ely wi h 0.01, 0.1, and 0.5% SDS. A
each ca hode bu e change, new PVDF memb anes we e used. T ans-
e ed DnaK was de ec ed wi h a mouse monoclonal an i-DnaK an i-
body (clone 8E2/2, S essgen Bio ech Co p.) using an ECL ki (Ame -
sham Biosciences). Cellulose memb anes con ained h ee e e ence
pep ides (C1, C2, and C3), all o which ha e p e iously been shown o
cons i u e good DnaK binde s (21, 22), i.e. om hemagglu inin (T p
436
–
Asn
448
), pep ide WTYNAELLVLLEN (indica ed as C1); om alkaline
phospha ase (Gly
382
–His
394
), pep ide GNTLVIVTADHAH (indica ed as
C2); and om
32
(Se
192
–Lys
204
), pep ide SHAMAPVLYLQDK (indi-
ca ed as C3).
Pep ide Syn hesis—Pep ides we e syn hesized on a Pionee
®
ins u-
men using s anda d solid-phase Fmoc (N-(9- luo enyl)me hoxyca -
bonyl) syn hesis p o ocols. Al e na i ely, solid-phase pep ide syn hesis
was pe o med on a Model 433A pep ide syn hesize wi h condi ional
double coupling s eps (Applied Biosys ems, Fos e Ci y, CA). Each
pep ide was p epa ed as a C- e minal amide and was “capped”using
ace ic anhyd ide. All pep ides we e cha ac e ized using liquid ch oma-
og aphy/ andem mass spec ome y and pu i ied by high p essu e
liquid ch oma og aphy.
Pep ide Compe i ion o Complex Fo ma ion be ween DnaK and Re-
duced and Ca boxyme hyla ed Lac albumin (RCMLA)—Compe i ion
binding assays we e pe o med as desc ibed (26, 27). 70 nMDnaK was
incuba ed wi h 40 mMRCMLA o 2ha 37°C in bu e con aining 25
mMT is-HCl, 20 mMHEPES, pH 7.15, 47.5 mMKCl, and 2.25 mM
Mg(OAc)
2
in he p esence o compe ing pep ide (0–500
M). F ee DnaK
and DnaK䡠pep ide complexes we e sepa a ed om DnaK䡠RCMLA com-
plexes by na i e PAGE on 6% ac ylamide gels and isualized by immu-
noblo ing wi h an i-DnaK an ibody 8E2/2 using he ECL ki . All ex-
pe imen s we e epea ed a leas h ee imes.
32
-Q132-Q144-C-IAANS Compe i ion Assay—The K
D
alues o pep-
ides H10-9, M10-9, H44-9, and Rc2 o DnaK we e de e mined by
compe i ion wi h an IAANS-labeled pep ide (E. coli
32
-Q132-Q144-C-
IAANS) o binding o DnaK (21). DnaK (0.1
M) and
32
-Q132-Q144-
C-IAANS (0.5
M) we e incuba ed wi h inc easing concen a ions o
pep ides H10-9 (0–20
M), M10-9 (0–20
M), H44-9 (0–60
M), and Rc2
(0–40
M)in25mMHEPES, pH 7.6, 50 mMKCl, and 10 mMMgCl
2
o
90 min a 30 °C. The luo escence in ensi ies (F⫺F
max
) a an emission
maximum o 452 nm (
ex
⫽335 nm) we e plo ed agains he co e-
sponding pep ide concen a ions and i ed (GRAPHIT 5.0.1) by a quad-
a ic equa ion using he K
D
o binding o
32
-Q132-Q144-C-IAANS o
DnaK (78 nM) (see Re . 28). All expe imen s we e epea ed a leas h ee
imes.
RESULTS
Sc eening o DnaK Binding o Pep ide Lib a ies De i ed om
Human and Mouse IFN-
␥
—In a p elimina y expe imen , we
analyzed he binding o DnaK o cellulose-bound pep ides de-
2
K. Vandenb oeck and J. R. Gau , unpublished da a.
Helix C DnaK-binding Si e in IFN-
␥
and Rela ed Cy okines 25669
i ed om he hyd ophobic C helices om human, mouse, and
po cine IFN-
␥
. Pep ide-bound DnaK was eleased by elec o-
elu ion using ca hode bu e supplemen ed successi ely wi h
0.01, 0.1, and 0.5% SDS. Unde he condi ions o he assay,
elease o DnaK om speci ic IFN-
␥
pep ides appea ed o be
in luenced by he concen a ion o SDS (Fig. 1). This is exem-
pli ied by he di e en ial elu ion beha io o DnaK om pep-
ides FYLRLFEVLKDNQ ( om mouse IFN-
␥
; indica ed by
black a ows in Fig. 1) and FYFKLFKNFKDDQ (homologous
pep ide om human IFN-
␥
; indica ed wi h whi e a ows in Fig.
1). Whe eas ca hode bu e supplemen ed wi h 0.01% SDS
o ced elu ion o ⬎60% o DnaK om he i s pep ide, 0.5%
SDS was equi ed o quan i a i e elu ion o DnaK om he
second pep ide. These indings we e aken in o accoun in all
u he expe imen s in ha elec oelu ion was ou inely ca -
ied ou in h ee s eps using ca hode bu e supplemen ed
successi ely wi h 0.01, 0.1, and 0.5% SDS. Ru¨dige e al. (22)
obse ed e ec s simila o hose ound in his s udy du ing he
ac iona ed elec o ans e o DnaK om solid-phase bound
pep ides, e en a cons an SDS concen a ion. Howe e , hey
op imized he condi ions o hei s udy o supp ess his e ec
so as o ob ain an a e aged image in one blo . The e o e, he
da a ob ained in bo h s udies a e compa able.
Cellulose-bound pep ide lib a ies we e used o scan human
and mouse IFN-
␥
o he p esence o DnaK-binding si es. The
pep ide scans consis ed o 13-me pep ides o e lapping by 10
amino acids, s a ing wi h he N e minus o he ma u e se-
quences. 44 pep ides co e ed he human sequence, whe eas he
mouse sequence, which lacks he 9 C- e minal amino acids,
was co e ed by 41 pep ides. Because he hyd ophobic co e
egion in he DnaK-binding mo i has a leng h o 4–5 amino
acids, all po en ial linea DnaK-binding si es a e hus logically
ep esen ed in hese lib a ies. Fig. 2 shows he esul s o he
sc een. Amoun s o DnaK elu ed s epwise wi h 0.01, 0.1, and
0.5% SDS-supplemen ed ca hode bu e we e app oxima ed by
densi ome y and added up. Th ee e e ence pep ides (C1, C2,
and C3) p e iously shown o cons i u e good DnaK binde s (21,
22) we e included in each memb ane. To ensu e ha he bind-
ing o DnaK o any o he cellulose-bound pep ides was speci ic,
an addi ional expe imen was ca ied ou in which a soluble
compe i o subs a e pep ide, i.e. pep ide BH801 (sequence
YWWNLLQ) de i ed om HIV gp160, was added du ing incu-
ba ion o he memb ane wi h DnaK. This pep ide has p e i-
ously been shown o display high a ini y o DnaK (27). I s
inclusion in he incuba ion bu e se e ely educed binding o
DnaK o memb ane-bound pep ides (da a no shown). This
sugges s ha he in e ac ion o DnaK wi h cellulose-bound
pep ides occu s mechanis ically h ough he subs a e ecogni-
ion si e.
The Amoun o DnaK Elu ed om IFN-
␥
-de i ed Cellulose-
bound Pep ides Co ela es wi h DnaK-Pep ide A ini y Meas-
u ed in he RCMLA Assay—Upon conside a ion o he o al
amoun o DnaK elu ed, a leas wo majo and se e al mino
DnaK-binding si es in bo h human and mouse IFN-
␥
can be
in e ed (Fig. 2). Some pep ides appea ed o bind limi ed
amoun s o DnaK, al hough subsequen elu ion equi ed high
FIG.1.E ec o SDS concen a ion on elec oelu ion o solid-
phase pep ide-bound DnaK. DnaK was incuba ed wi h a cellulose
memb ane con aining pep ides de i ed om he C helices o human,
mouse, and po cine IFN-
␥
. Pep ide-bound DnaK was ans e ed on o
PVDF memb anes by ac iona ed elec oblo ing using ca hode bu e
supplemen ed successi ely wi h 0.01, 0.1, and 0.5% SDS and isualized
wi h an i-DnaK an ibodies and chemiluminescence. A ows indica e
he di e en ial elu ion beha io o DnaK om wo pep ides: FY-
FKLFKNFKDDQ (whi e a ows) and FYLRLFEVLKDNQ (black
a ows). The cellulose memb ane con ained h ee e e ence pep ides
(C1, C2, and C3; see “Expe imen al P ocedu es” o de ails), all o which
ha e p e iously been shown o cons i u e good DnaK binde s (22). SDS
concen a ions in ca hode bu e and he co esponding ime o each
elu ion s ep a e indica ed on he igh . The cumula i e ime o elu ion
is indica ed on he le .
FIG.2.Iden i ica ion o DnaK binding si es in IFN-
␥
.aand b,
analysis o he binding o DnaK o solid-phase pep ide lib a ies o
human (huIFN-
␥
) and mouse (muIFN-
␥
) IFN-
␥
, espec i ely. Resul s o
one expe imen o h ee pe o med a e shown. Pep ides (13-me s) s a
wi h he N e mini o he ma u e IFN-
␥
po ions and scan he en i e
sequences owa d he C e minus, mo ing along by 3 amino acids, hus
yielding an o e lap window o 10 esidues. Memb anes we e incuba ed
wi h DnaK, ollowed by ac iona ed elec o ans e o pep ide-bound
DnaK o PVDF memb anes. Ca hode bu e was supplemen ed succes-
si ely wi h 0.01, 0.1, and 0.5% SDS. The amoun o elu ed DnaK was
quan i ied by densi ome y o immunoblo s. Each cellulose memb ane
con ained he h ee e e ence pep ides (C1, C2, and C3), desc ibed in
he legend o Fig. 1. As e isks deno e solid-phase pep ides om which
9-me pep ides we e de i ed o analysis in he RCMLA and/o
32
-
Q132-Q144-IAANS compe i ion assays. (9-me s we e de i ed om he
o e lap windows o DnaK-binding (i.e. pep ides H10, H18, H32, M18,
and M39) o homologous (i.e. pep ides M10 and H39) egions; pep ides
H43 and H44 did no show appa en binding o DnaK, bu we e heo-
e ically p edic ed o con ain a DnaK-binding si e (see Fig. 6).) Rele an
␣
-helices a e indica ed a he op (B, C, E, and F). c, a po ion o human
IFN-
␥
encompassing helices B and C, shown aligned wi h co espond-
ing 13-me pep ides om he pep ide scan. The a e age o he pooled
amoun o DnaK elu ed wi h 0.01, 0.1, and 0.5% SDS-supplemen ed
ca hode bu e om hese pep ides in h ee expe imen s is app oxi-
ma ed by he shade o he ba s. P edic ed DnaK-binding co e si es in
he human sequence a e unde lined in ed. Fo e e ence, he co e-
sponding sequence om mouse IFN-
␥
(mIFN-
␥
) is also gi en aligned
wi h ha om human IFN-
␥
(hIFN-
␥
). DSU, densi ome ic scanning
uni .
Helix C DnaK-binding Si e in IFN-
␥
and Rela ed Cy okines25670
concen a ions o SDS (ⱖ0.1%) in ca hode bu e , e.g. all pep-
ides co esponding o he C e mini o human and mouse
IFN-
␥
(pep ides H41–H44 o human IFN-
␥
and pep ides M40
and M41 o mouse IFN-
␥
). All o hese pep ides con ain poly-
basic sequences (KRKR in pep ides H41–H43, RGRR in pep ide
H44, and RKRK in pep ides M40 and M41), aising he possi-
bili y ha lack o elu ion o DnaK wi h 0.01% SDS migh be due
o ein o cemen o hyd ophobic in e ac ions by added elec o-
s a ic e ec s. Indeed, eg ession analysis o he ac ion o
DnaK elu ed wi h ⱖ0.1% SDS agains he isoelec ic poin s o
all 88 pep ides om Fig. 2 e ealed a linea co ela ion ( ⫽
0.81, ⫽12.69, deg ees o eedom ⫽86, p⬍10
⫺10
) (Fig. 3).
The ac ion o o al quan i y o DnaK elu ed wi h ⱖ0.1% SDS
did no co ela e wi h alipha ic index o g and a e age o
hyd opa hici y o hese pep ides (da a no shown).
Fu he indica ions ha he amoun o DnaK elu ed wi h
ⱖ0.1% SDS is o be aken in o accoun when es ima ing a ini y
on he basis o he solid-phase assay came om subsequen
analysis o soluble pep ides in he RCMLA assay (Table I and
Fig. 4). This is a compe i ion assay in which soluble pep ides
a e es ed o hei abili y o compe e wi h an un olded o m o
lac albumin (RCMLA) (see Re . 26) o binding o DnaK o e a
ange o concen a ions (0–500
M). Nine 9-me pep ides we e
syn hesized ha we e de i ed om he o e lap windows o
weak, in e media e, and s ong binde s as in e ed om he
solid-phase assay. Appa en a ini y o DnaK (K
app
) (see Re .
27) was measu ed as he pep ide concen a ion yielding 50%
inhibi ion o binding o RCMLA o DnaK. HIV gp160-de i ed
pep ide BH801 (YWWNLLQ), he K
app
o which in he RCMLA
assay has p e iously been de e mined o be 50
M(27), was
used as a posi i e con ol and was included in each assay
(Table I and Fig. 4). The amoun o DnaK elu ed om pep ides
in he solid-phase assay was ound o be indica i e o he
a ini y measu ed in he RCMLA assay (see below o co ela-
ion analysis).
Howe e , a disc epancy was obse ed be ween he in e ac-
ions o pep ide M10-9 (LFLDIWRNW) wi h DnaK in he RC-
MLA e sus solid-phase assay (in he la e o which his pep-
ide cons i u es he o e lap o pep ides M9 and M10). Al hough
his pep ide displayed a s ong a ini y o DnaK in he o me
assay (K
app
⫽10
M), i bound only a small quan i y o DnaK
in he solid-phase assay. The homologous pep ide in human
IFN-
␥
, H10-9 (LFLGILKNW), did no display his anomalous
beha io , as i showed s ong a ini ies o DnaK in bo h as-
says. To cla i y his issue, we used ano he compe i ion assay
o de e mina ion o a ini y. In his assay, he K
D
alues o he
binding a ini y o bo h pep ides M10-9 and H10-9 o DnaK
we e measu ed by compe i ion wi h
32
-Q132-Q144-C-IAANS
o binding o DnaK (21). Up o 12 concen a ions o compe ing
pep ide we e used cen e ing a ound he inhibi o y concen a-
ion (IC
50
) (Fig. 5). This assay yielded K
D
alues o 0.12 ⫾0.02
M o pep ide H10-9 and 0.25 ⫾0.02
M o pep ide M10-9,
hus iden i ying bo h pep ides as uly e y s ong DnaK
binde s.
We in es iga ed whe he he ela ionship be ween he
amoun s o DnaK elu ed om pep ides in he solid-phase assay
and he a ini ies o hei co e mo i s o DnaK as de e mined in
he RCMLA assay could be o malized by linea eg ession
analysis. O he nine IFN-
␥
-de i ed pep ides es ed (Table I),
we omi ed pep ide M10-9 because o i s anomalous beha io .
We obse ed a signi ican in e se co ela ion be ween he K
app
ob ained o each o he emaining 9-me pep ides and he
pooled amoun o DnaK elu ed wi h 0.01, 0.1, and 0.5% SDS
om hei solid phase coun e pa s ( ⫽⫺0.91, p⫽0.005).
Upon conside a ion o he amoun o DnaK elu ed wi h ⱖ0.1%
SDS only, he co ela ion wi h K
app
alues became sligh ly
s onge ( ⫽⫺0.92, p⫽0.004).
␣
-Helices B and C o Mouse and Human IFN-
␥
Con ain
Conse ed High-a ini y DnaK-binding si es—The combined
da a om he solid-phase sc eening and om he RCMLA and
32
-Q132-Q144-C-IAANS compe i ion assays indica e he loca-
ion o conse ed high-a ini y DnaK-binding si es in he N-
e minal hal o helix B and he C- e minal hal o helix C o
human and mouse IFN-
␥
(Table I). We used an algo i hm based
on he expe imen al sc eening o he binding o DnaK o 4360
pep ides o localize mo e p ecisely he DnaK-binding co e mo-
i s in hese helices (22). The mos p obable mo i o in e ac-
ion o DnaK wi h helix B was con ined o a 6-amino acid
s e ch consis ing o wo o e lapping 5- esidue co es (L
28
FL-
GIL) (ene gy alue less han ⫺10; alues less han ⫺5 a e
likely o be DnaK binde s (see Re . 22)). In helix C, a 12-amino
acid s e ch consis ing o six o e lapping co e mo i s wi h en-
e gy alues less han ⫺6 was iden i ied (I
49
VSFYFKLFKNF)
(Fig. 2c). Wi hin his s e ch, he co e mo i s I
49
VSFY and
Y
53
FKLF had he lowes ene gy alues o –9 and –10, espec-
i ely, and hus he highes p edic ed a ini y o DnaK. As
shown abo e, hese helix B and C mo i s as pa o 9-me
pep ides bound DnaK wi h high a ini y (Table I and Figs. 4
and 5). In each o hese homologs, he helix B and C DnaK
subs a e egions a e sepa a ed by a s e ch o 15 amino acids.
An addi ional DnaK mo i was iden i ied in helix F o mouse
IFN-
␥
(R
114
VVHQLLPE; K
app
⫽80
M) ha seems no o be
unc ionally conse ed in he human o m (homologous posi ion
Q
115
VMAELSPA; K
app
⫽400
M).
In gene al, he expe imen al indings co obo a e he heo-
e ical p edic ion o DnaK si es in human and mouse IFN-
␥
.
Fo ins ance, as shown in Fig. 6 he conse ed DnaK-binding
si es in helices B and C as well as he helix F mo i in mouse
IFN-
␥
we e co ec ly p edic ed. A binding si e was heo e ically
p edic ed o occu in he C e minus o human IFN-
␥
(Q
133
MLFRGR; ene gy alue o ⫺10) ha did no show up in
he solid-phase assay (pep ides H42 and H43) (Fig. 2) and
nei he in he RCMLA assay (K
app
⬎500
M o pep ide H44-9)
(Table I and Fig. 4). Howe e , he K
D
o his pep ide as meas-
u ed in he
32
-Q132-Q144-C-IAANS compe i ion assays was
3.6
M. This alue is smalle han he ecen ly de e mined
cu o alue o 7
M o he K
D
, abo e which pep ide subs a es
a e an icipa ed no o s imula e ATP hyd olysis and hence no
o bind o DnaK (28). Thus, wi h he p esen da a, i is di icul
o exclude a ole o his si e in HSP70 chape one-assis ed
olding o IFN-
␥
.
Implica ions o Folding o IFN-
␥
—Wi h he high- esolu ion
(1.4 Å) h ee-dimensional s uc u e o human IFN-
␥
being
a ailable (code 1 g9; P o ein Da a Bank), we a emp ed o in e
he implica ions o he localiza ion o he wo conse ed DnaK-
binding si es in helices B and C o he olding o he dime
FIG.3. Plo o he ac ion o DnaK elu ed wi h >0.1% SDS
agains he isoelec ic poin s o all 88 pep ides om Fig. 2 ( ⴝ
0.81, ⴝ12.69, deg ees o eedom ⴝ86).
Helix C DnaK-binding Si e in IFN-
␥
and Rela ed Cy okines 25671
(Figs. 7 and 8). The helix B and C DnaK mo i s exhibi signi -
ican sol en accessibili y (helix B mo i , 617 Å(monome ) and
327 Å(dime ); and helix C mo i , 625 Å(monome ) and 410 Å
(dime )). The dec ease in going om he monome o he dime
is due o he p esence o helices E and F o he second subuni
packing agains helices B and C o he i s . I is in e es ing o
no e ha he helix B and C po ions encompassing DnaK si es
ha e a e y simila pa ch o sol en -accessible su ace (Fig. 8).
This is cen e ed on a lysine esidue in bo h cases (L
30
–K
34
N
35
o helix B and F
54
–K
58
N
59
o helix C). Taking in o accoun
s uc u al cons ain s imposed by he second domain o IFN-
␥
ha is adjacen o he helix C mo i , i seems ha , in he olded
dime , he helix B (bu no he helix C) chape one mo i migh
be exposed o mac omolecula access. This could be o impo -
ance o in oking a ole o HSP70-like chape ones in he
p e en ion o i e e sible he mal inac i a ion h ough “ea ly”
in e ac ion wi h he helix B mo i , i.e. be o e he mally induced
un olding o he dime exposes he helix mo i C o in e ac ion
wi h chape ones. Figs. 7 and 8 show ha , owa d he in e io o
he dime , he helix B and C DnaK-binding si es in e ac a
hei N e mini wi hin each subuni (F
29
–F
52
Y
53
), showing he
no mal “edge- o- ace”alignmen o a oma ic ings, and wi h
Leu
113
in helix F o he second subuni . The hyd ophobic su -
ace c ea ed by F
52
Y
53
(dona ed by he helix C mo i ) and Leu
28
( om he helix B mo i ) om he i s subuni o ms a small
pocke illed by he C-

o Ala
109
om helix F o he second
subuni (29). Helix F shows a p onounced bend o ⬃55–60°
cen e ed on Glu
112
, which is unique o IFN-
␥
, IL-10, and e-
la ed cy okines (29–31). Leu
28
, Phe
52
, Ty
53
, Ala
109
, and Glu
112
a e conse ed in IL-10 and IFN-
␥
and clus e wi h ex ensi e
con ac s in he co e o he dime a ound he bend in helix F.
These 5 amino acids a e hough o ep esen key esidues in
s abilizing he IFN-
␥
/IL-10 ypical s uc u al old (29). The
dona ion o he i s 3 esidues by high-a ini y DnaK-binding
si es (Leu
28
by he helix B mo i and F
52
Y
53
by he helix C
mo i ) ad oca es a ole o HSP70-like chape ones in de no o
posi ioning, s abiliza ion, o p e en ion o lawed alignmen o
his IFN-
␥
/IL-10 ypical old.
The DnaK-binding Mo i in
␣
-Helix C o IFN-
␥
Is Conse ed
in he Supe amily o In e leukin-10- ela ed Cy okines—A
numbe o no el IFN-
␥
/IL-10 homologs ha e ecen ly been
iden i ied, including IL-19 (32), IL-20 (33), IL-22 (34, 35), and
FIG.4.Compe i ion o complex o ma ion be ween DnaK and
RCMLA by soluble pep ides. 70 nMDnaK (50 ng/assay) was incu-
ba ed wi h 40
MRCMLA in he absence o p esence o inc easing
concen a ions o compe ing pep ide (0–500
M). F ee DnaK and
DnaK䡠pep ide complexes we e sepa a ed om DnaK䡠RCMLA complexes
by na i e PAGE (6% polyac ylamide) and isualized by blo ing, ol-
lowed by de ec ion wi h an i-DnaK an ibodies. Pep ide BH801 (YWWN-
LLQ) de i ed om HIV gp160 was used as a e e ence pep ide (27) and
was included in each expe imen . The esul s o one o h ee expe i-
men s pe o med a e p esen ed. The sequences o he pep ides es ed
a e gi en in Table I.
TABLE I
In e ac ions be ween DnaK and pep ides de i ed om IFN-
␥
, IL-20, and IL-24/FISP
Pep ide P o ein Loca ion Pep ide
sequence
DnaK binding
DnaK
elu ed
a
K
appb
K
Dc
DSU
M
M
H10–9 huIFN-
␥
d
Helix B LFLGILKNW 77,000 ⬍10 0.12 ⫾0.02
M10–9 muIFN-
␥
Helix B LFLDIWRNW 19,000 10 0.25⫾0.02
H18–9A huIFN-
␥
Helix C IVSFYFKLF 66,000 40 ND
H18–9B huIFN-
␥
Helix C FYFKLFKNF 66,000 30 ND
e
M18–9 muIFN-
␥
Helix C FYLRLFEVL 85,000 ⬍10
ND
e
H32–9 huIFN-
␥
Helix E KLTNYSVTD 45,000 400 ND
H39–9 muIFN-
␥
Helix F QVMAELSPA 25,000 400 ND
M39–9 muIFN-
␥
Helix F RVVHQLLPE 48,000 80 ND
H44–9 huIFN-
␥
CT RSQMLFRGR 5,000 ⬎500 3.6 ⫾1.2
4
␣
HC huIL-20 Helix C HLLRLYLDRVF ND 15 ND
Rc1 muIL-24/FISP Helix C SLLKFYLNTVF ND 60 ND
Rc2 muIL-24/FISP Helix C LLKFYLNTV ND 40 1.03 ⫾0.4
BH801
a
HIV gp160 CP YWWNLLQ ND 40 ND
a
To al amoun o DnaK elec oelu ed wi h 0.01, 0.1, and 0.5% SDS om solid-phase pep ides app oxima ed by an i-DnaK immunoblo ing and
densi ome ic scanning.
b
K
app
is de ined as he concen a ion o pep ide necessa y o hal -maximum compe i ion o DnaK binding o RCMLA.
c
K
D
was de e mined by compe i ion o unlabeled pep ides wi h
32
-Q132-Q144-C-IAANS o binding o DnaK (21).
d
hu, human; mu, mu ine; CT, C e minus; CP, cy oplasmic domain; ND, no de e mined; DSU, densi ome ic scanning uni .
e
K
D
alues o pep ides H18–9B and M18–9 could no be de e mined due o agg ega ion.
Pep ide M18–9 agg ega ed/caused agg ega ion o DnaK a concen a ions abo e 20
Min he RCMLA assay.
g
BH801 was used as e e ence pep ide. (K
app
in he DnaK/RCMLA assay was p e iously de e mined a 50
M(see Re . 27)).
Helix C DnaK-binding Si e in IFN-
␥
and Rela ed Cy okines25672
IL-26/AK155 (36). Also MDA-7 (37), a C49a/MOB-5 (38, 39),
and mouse in e leukin-4 induced sec e ed p o ein (40) belong
o his supe amily and we e ecen ly enamed as IL-24 (41).
Mo e dis an ly ela ed i al IL-10 homologs ha e been iden i-
ied in he genomes o hesus, simian, and human cy omegalo-
i us (42, 43); lumpy skin disease i us (44); and Yaba-like
disease i us (45).
Alignmen o 55 membe s o his supe amily un eiled wo
egions o ex ensi e conse a ion (Fig. 9). The i s egion o
conse a ion was ound in helix C a ound he DnaK subs a e
egion iden i ied in IFN-
␥
(70% consensus mo i , phlpFYLc);
he second egion is composed o amino acids a ound he p o-
nounced bend in helix F seen in he c ys al s uc u es o IFN-
␥
and IL-10 (70% consensus mo i , KAhpEh). The likelihood o
occu ence o DnaK si es in helices C and F was calcula ed
using he DnaK si e p edic ion algo i hm (22). No DnaK-bind-
ing si es we e p edic ed in he conse ed helix F egion. In
helix C, a conse ed 6- esidue s e ch (consensus, hlpFYL) was
p edic ed o cons i u e he co e o a DnaK-binding si e in
IFN-
␥
, IL-19, IL-20, IL-22, IL-24, and IL-26 (Fig. 9). Howe e ,
he co esponding sequences in IL-10 and mos i al IL-10
homologs we e p edic ed o be non-binde s, essen ially due o
he clus e ing o nega i ely cha ged esidues in he egions
lanking he co e mo i , which a e likely o dis a o binding o
DnaK (22, 23).
To expe imen ally e i y he alidi y o his calcula ion, we
es ed in e ac ion be ween DnaK and he heo e ically p e-
dic ed helix C mo i s in wo membe s o his amily, i.e. mo i s
HLLRLYLDRVF om human in e leukin-20 and SLLK-
FYLNTVF p esen in a and mouse IL-24, by he RCMLA
compe i ion assay (Table I and Fig. 4). Bo h pep ides we e
indeed iden i ied as s ong DnaK binde s, yielding K
app
alues
o 15 and 60
M, espec i ely, sugges ing unc ional conse a-
ion o he helix C chape one-binding si e in his supe amily o
cy okines. The nonapep ide co e mo i LLKFYLNTV de i ed
om a and mouse IL-24 bound o DnaK wi h s ong a ini y
(K
D
⫽1
M)in he
32
-Q132-Q144-C-IAANS compe i ion assay,
u he con i ming his si e as a po en ial na u al a ge o
HSP70 chape one in e ac ion.
DISCUSSION
In his wo k, we p o ide e idence o he p esence o wo
high-a ini y DnaK-binding si es in
␣
-helices B and C o in e -
e on-
␥
. In addi ion, nonconse ed lowe a ini y binding si es
we e pu a i ely iden i ied in helix F o mouse IFN-
␥
and in he
C e minus o human IFN-
␥
. Upon conside a ion o all pep ides
showing a leas some deg ee o DnaK binding in he solid-
phase assay (Fig. 2), i seems ha he occu ence o si es
sus aining a minimal le el o in e ac ion wi h DnaK is en-
iched in and la gely con ined o only hose
␣
-helices ha
con ibu e o he in e locking dime in e ace (cons i u ed by
he helix B-loop II-helix C and helix E-loop V-helix F segmen s)
(4, 5). Thus, helices A and D can easonably be excluded as
po en ial a ge s o chape one in e ac ion. O no able in e es
is he inding ha he helix C mo i appea s o be conse ed in
he majo i y o cy okines belonging o he IL-10 supe amily.
We used h ee complemen a y app oaches o iden i ica ion
o hese si es, i.e. sc eening o lib a ies o solid-phase bound
pep ides scanning human and mouse IFN-
␥
, quan i ica ion o
supp ession by soluble IFN-
␥
-de i ed pep ides o complex o -
ma ion be ween DnaK and RCMLA, and analysis o he com-
pe i ion o some o hese pep ides wi h
32
-Q132-Q144-C-
IAANS o binding o DnaK. An in e es ing inding is ha he
amoun o DnaK elu ed om solid-phase pep ides co ela ed
in e sely wi h he K
app
o hei co e pep ides in he RCMLA
assay ( ⫽⫺0.91, p⫽0.005). Fu he mo e, o he ou pep ides
ha we e p ocessed in he
32
-Q132-Q144-C-IAANS assay, he
K
D
alues co ela ed wi h hei K
app
alues as measu ed in
he RCMLA assay ( ⫽0.98, p⫽0.02). Al hough, because o he
limi ed numbe o pep ides es ed, cau ion is wa an ed when
in e p e ing hese esul s, his sugges s ha he quan i y o
DnaK elu ed om pep ides in he solid-phase assay could be
FIG.5. De e mina ion o he a ini y o pep ides o DnaK. a
and b, i a ion o pep ides H10-9 (helix B om human IFN-
␥
) and
M10-9 (helix B om mouse IFN-
␥
), espec i ely, agains DnaK and
32
-Q132-Q144-C-IAANS ( ull sequence, QRKLFFNLRKTKQ) o de-
e mina ion o K
D
. The luo escence (F⫺F
max
) was plo ed agains he
co esponding pep ide (H10-9 and M10-9) concen a ion. Complex o -
ma ion be ween DnaK and
32
-Q132-Q144-C-IAANS in he absence o
compe ing pep ide yielded he highes luo escence in ensi y (F
max
).
Values a e a e aged o e h ee independen expe imen s. Pep ides
H44-9 and Rc2 (see Table I) we e also p ocessed in his assay (no
shown).
FIG.6.P edic ion o DnaK-binding si es in he p ima y s uc-
u es o he ma u e po ions o human (a) and mouse (b) IFN-
␥
.
Each o he ba s ep esen s he p edic ed sco e o binding o DnaK o
a single 13-me pep ide om he o iginal pep ide scans shown in Fig. 2.
Sco es we e calcula ed wi h he algo i hm desc ibed by Ru¨dige e al.
(22). Pep ide numbe ing co esponds o ha shown in Fig. 2 (con ol
pep ides C1, C2, and C3 we e excluded). The do ed lines indica e he
cu o alue se a –5 (co esponding o an expe imen ally e i ied
co ec le el o p edic ion o binding si es o 82%). huIFN-
␥
, human
IFN-
␥
;muIFN-
␥
, mouse IFN-
␥
.
Helix C DnaK-binding Si e in IFN-
␥
and Rela ed Cy okines 25673
applied as a p edic i e measu e o es ima ing he a ini y o
in e ac ion o hese pep ides wi h DnaK. Howe e , i should be
no ed ha he in e ac ion o DnaK wi h pep ide M10-9 (LFLDI-
WRNW) in soluble o m di e ed quan i a i ely om ha seen
in he solid-phase assay (o e lap o pep ides M9 and M10).
Al hough his pep ide was iden i ied as a s ong DnaK binde
in bo h RCMLA and
32
-Q132-Q144-C-IAANS compe i ion as-
says, i bound only a small amoun o DnaK in he solid-phase
assay. These indings a e cong uen wi h ea lie obse a ions
ha quan i a i e di e ences be ween pep ides in solu ion and
cellulose-bound pep ides migh occu . An explana ion o his
di e gence migh be sough in he unusually high ac i a ion
en opy o subs a e binding o DnaK (46). The en opy o a
gi en pep ide in soluble o m di e s om ha in solid-phase
o m and is also di e en om he posi ional con ex in he
whole p o ein. Howe e , he en halpic e ms should be compa-
able, hus subs an ia ing he no ion ha he obse ed di e -
ences a e quan i a i e, bu no quali a i e.
The DnaK mo i s iden i ied in his s udy con o m o he
p oposed binding mo i consis ing o 4–5 hyd ophobic amino
acids en iched o Leu, Ile, Val, Phe, and Ty wi h basic esi-
dues adjacen o he hyd ophobic co e (22, 47). The high-a in-
i y DnaK mo i s iden i ied in his s udy a e all loca ed in
pep ide segmen s ha adop an
␣
-helical con o ma ion in he
olded p o ein. This is ema kable in iew o he inding ha
helical s uc u es we e epo ed no o i in o he subs a e-
binding cle o DnaK (47). The majo i y o good DnaK-binding
si es localize p e e en ially o

-s ands o co esponding olded
p o eins, e en i i is known ha pep ide segmen s wi h in in-
sic p opensi y o o m
␣
-helical s uc u es can bind DnaK as
long as hey occu in an ex ended (i.e. un olded) con o ma ion
(22, 23). IFN-
␥
may he e o e cons i u e an uncommon example
o a high-a ini y all-
␣
-helical HSP70 subs a e. Especially he
iden i ica ion o a unc ional DnaK si e in helix C o IFN-
␥
and
FIG.8.Molecula modeling o DnaK-binding si es in
␣
-helices B and C o IFN-
␥
.Shown is a s e eo iew o he sol en -accessible a ea,
loca ion, and in e ac ion o DnaK-binding mo i s in
␣
-helices B and C o human IFN-
␥
. Only he side chains o Leu
28
–T p
36
(helix B, uppe egion;
B) and Phe
52
–Phe
60
(helix C, lowe egion; C) a e shown om one subuni o he dime in he c ys al s uc u e o human IFN-
␥
(code 1 g9; P o ein
Da a Bank). The diagonal g ay lines di ide he wo mo i s, and he pu ple a ows show he o ien a ion o he wo helices (diamonds a he
N- e minal ends). The helix old is ep esen ed as a ibbon (pu ple). A semi anspa en sol en -accessible su ace is also displayed, wi h
hyd ophobic egions in ed and hyd ophilic egions in blue. Indi idual esidues a e labeled in one o h ee colo s: ed o sol en -accessible
hyd ophobic, blue o sol en -accessible hyd ophilic, and black o ei he bu ied esidues o an exposed glycine.
FIG.7. Loca ion o DnaK-binding
si es in he IFN-
␥
dime . S e eo ibbon
diag ams show he loca ions o high-a in-
i y DnaK-binding mo i s in
␣
-helices B
(blue)andC(cyan) in he IFN-
␥
dime
(code 1 g9; P o ein Da a Bank). The lowe
pa s ep esen magni ied diag ams o
he in e ac ion be ween Phe
29
, Phe
52
, and
Ty
53
o hese mo i s in he i s subuni
and Ala
109
and Leu
113
loca ed a ound he
bend in helix F in he second subuni o
IFN-
␥
(code 1 g9). Yellow and ed ibbons
co espond o he dis inc monome s in
he dime .
Helix C DnaK-binding Si e in IFN-
␥
and Rela ed Cy okines25674
IL-10- ela ed cy okines may u he he unde s anding o he
ole o chape ones in p o ein olding. Helix C is he longes and
mos hyd ophobic helix and is bu ied in he co e o he dime (4,
5). Essen ially, he IFN-
␥
dime in e ace is cen e ed on a pai
o wo symme y- ela ed C helices (4). In hei su ey o ho-
modime ic p o eins, La sen e al. (10) iden i ied a small class o
in e aces ha ha e subuni s ha in e digi a e ex ensi ely
and ha exhibi he s ongly hyd ophobic cha ac e o he
in e io co e o a single-chain olded domain. This led hem o
conclude ha , o hese p o eins, he en i e dime o ms in one
olding s ep, a he han om associa ion o wo indi idually
olded subuni s. This ype o p o ein olding is mos accu a ely
ep esen ed wi h a wo-s a e olding model in which he wo
chains old coope a i ely, and s able in e media e monome s
end no o occu (48). La sen e al. (10) dis inguished wo
unc ional classes in his ca ego y o p o eins, i.e. DNA-binding
p o eins (e.g. a c and p ep esso s) and cy okines (e.g. IFN-
␥
and IL-10). Analysis o he olding o he
␣
-helical p ep esso
suppo s he wo-s a e model o his ca ego y o p o eins, as
his p o ein exhibi s a olding pa hway in which he in e sub-
uni hyd ophobic co e i s o ms and only subsequen ly he
lanking s uc u es (49, 50). The mal dena u a ion s udies o
IFN-
␥
subs an ia e a dime o med as a single coope a i e
he modynamic domain, wi h dime iza ion being manda o y
o gene a ion o e ia y s uc u e (51). To ou knowledge, he
ole o chape ones in olding o his ca ego y o p o eins has no
been in es iga ed in de ail. The unc ional conse a ion o a
DnaK-binding mo i in helix C a he dime in e ace sugges s
ha , a leas in he cy okines IFN-
␥
, IL-19, IL-20, IL-22, IL-24,
and IL-26, HSP70-like chape ones migh be di ec ly in ol ed in
assembly o he dime by inhibi ing unp oduc i e side eac ions
a his s age. In suppo o his concep a e he indings o
Kend ick e al. (9), who, using solu ion he modynamic ap-
p oaches, ound ha only a sligh expansion o he su ace a ea
o he IFN-
␥
dime is needed o o m an agg ega ion-p one
con o me . Wi h he IFN-
␥
dime in e ace accoun ing o up o
40% o he o al su ace a ea, his inding implies a ansien
na i e-like dime ( a he han a dime -monome ) ansi ion
in ol ed in o -pa hway agg ega ion. I ele an o de no o
olding, his would in oke a ole o HSP70 chape ones in
supp essing agg ega ion o his dime o m o IFN-
␥
. This
model is in con as wi h he majo i y o dime s ha old
h ough a h ee-s a e model (48), o which subuni olding and
o ma ion o hyd ophobic nuclei and hence concomi an in e -
ac ion wi h HSP70 chape ones a e likely o be comple ed be o e
he dime iza ion s ep.
An addi ional DnaK-binding si e was pu a i ely iden i ied in
he C e minus o human IFN-
␥
. The K
D
o his mo i (RSQML-
FRGR) was 3.6
M. This alue is in he same o de o magni ude
as ha p e iously ound o he hea shock ansc ip ion ac o
32
, a physiological subs a e o DnaK (21). Because he cu o
alue o binding o DnaK has been de e mined a a K
D
o 7
M
(28), i is hus possible ha his si e ep esen s a ue subs a e
o HSP70 chape ones. Howe e , i so, his mo i may be o lesse
signi icance, as i was no e olu iona ily conse ed in oden
IFN-
␥
due o he absence o he 9 C- e minal amino acids. The C
e minus o human IFN-
␥
was no obse ed in he o iginal c ys-
al s uc u e (4), bu is supposed o be lexible in solu ion (52).
Landa e al. (53) de e mined he c ys al s uc u e o a single-
chain mu an o IFN-
␥
wi h an in ac C e minus (IFN
␥
SC1). The
C- e minal sequence QMLFRG was modeled in a hyd ophobic
binding pocke o med by wo non-c ys allog aphically ela ed
IFN-
␥
SC1 molecules. Al hough i is no ce ain whe he his
in e ac ion is ep esen a i e o he soluble s a e, he inding ha
his mo i coincides exac ly wi h a DnaK-binding si e may a gue
o a bu ied ( a he han sol en -exposed) loca ion o he C
e minus.
In he na u al olding compa men o IFN-
␥
, he endoplas-
mic e iculum, i is likely ha i is he DnaK homolog BiP/
GRP78 ha will in e ac wi h any o he mo i s de e mined in
his s udy. E en hough mino di e ences in subs a e ecog-
ni ion ha e been sugges ed (26, 54), BiP and DnaK binding
pa e ns a e gene ally conse ed, and bo h chape ones ecog-
nize simila side chains (55). A emp s in ou labo a o y o
sc een he IFN-
␥
pep ide lib a ies o binding o BiP ailed
(da a no shown), p obably due o he ac ha he pep ide
a ini y o BiP has been shown o be 1–2 o de s o magni ude
lowe han ha o DnaK (56, 57). Howe e , by co-immunop e-
cipi a ion expe imen s, we ha e been able o demons a e in-
e ac ion o BiP/GRP78 wi h IFN-
␥
,
2
sugges ing a unc ional
ole o his chape one in de no o olding o IFN-
␥
.
This wo k may con ibu e o a be e unde s anding o he
p ocesses egula ing olding and sec e ion o his he apeu i-
cally impo an class o cy okines. Josephson e al. (58) ecen ly
p esen ed compelling e idence o a no el ligand- ecep o ec-
FIG.9.DnaK-binding si es a e p edic ed o occu in he s uc-
u ally conse ed helix C (bu no F) o cy okines belonging o
he IFN-
␥
/IL-10 supe amily. The loca ion o he 6- esidue DnaK-
binding co e mo i in helix C is indica ed wi h a ho izon al g ay ba .As
he p obabili y o binding o DnaK is de e mined by he di e en ial
ene gy con ibu ions o each amino acid in a 5- esidue co e and wo
4- esidue lanking egions, combined ene gy alues we e calcula ed o
he sequences shown, aking in o accoun addi ional lanking amino
acids (no shown). Sco es a e ep esen ed in bold ace i less han o
equal o ⫺5 (cu o co esponding o an expe imen ally e i ied co ec
p edic ion le el o binding o 82%). Conse ed amino acid sequences a
consensus le els o 100, 90, 80, and 70% a e based on a o al o 55
p o eins, including hose shown and, in addi ion, 18 mammalian and
a ian IFN-
␥
p o eins and 18 mammalian and i al IL-10 p o eins. Fou
o he 5 conse ed amino acids (Phe
52
, Ty
53
, Ala
109
, and Glu
112
in
human IFN-
␥
) clus e ing a ound he bend in helix F o iginally iden i-
ied in he h ee-dimensional s uc u es o human IFN-
␥
and IL-10 by
Wal e and Nagabhushan (29) a e indica ed wi h as e isks. The emain-
ing esidue (Leu
28
) is loca ed in helix B (no shown). Colo ing e lec s
physicochemical p ope ies o esidues: c, cha ged; h, hyd ophobic; l,
alipha ic; p, pola ; s, small; ⫺, nega i ely cha ged. hu, human; bo,
bo ine; po, po cine; mu, mu ine; ch, chicken; YLDV, Yaba-like disease
i us; HCMV, human cy omegalo i us; SCMV, simian cy omegalo i-
us; RCMV, hesus cy omegalo i us; LSDV, lumpy skin disease i us;
FISP, in e leukin-4 induced sec e ed p o ein.
Helix C DnaK-binding Si e in IFN-
␥
and Rela ed Cy okines 25675
ogni ion pa adigm in which IL-10 ecep o -1 and ecep o -2
sha e he same binding si e on IL-10. Rema kably, 3 amino
acids o he con ac in e ace wi h he ecep o appea o be
conse ed in he whole supe amily o IL-10- ela ed cy okines,
including IFN-
␥
. Two o hese a e he Lys and Glu esidues
ound nea he bend in helix F (Fig. 9). The s uc u e o he
IFN-
␥
䡠IFN-
␥
ecep o -1 complex e ealed ha IFN-
␥
ecep o -1
in e ac s in a i ually iden ical way wi h his conse ed esi-
due clus e on IFN-
␥
(59). Ou wo k indica es ha he amino
acid sequence a ound and including he highly conse ed
F
52
Y
53
mo i in helix C o his supe amily is p obably manda-
o y o chape one-assis ed gene a ion o he IFN-
␥
/IL-10 yp-
ical old. Thus, pending u u e in es iga ion, we p opose ha
bo h he mechanisms de e mining egula ion o olding and
ligand- ecep o in e ac ion occu along he same lines in his
class o cy okines.
REFERENCES
1. Billiau, A. (1996) Ad . Immunol. 62, 61–130
2. Ma s e s, S. A., Pennica, D., Bach, E., Sch eibe , R. D., and Ashkenazi, A.
(1995) P oc. Na l. Acad. Sci. U. S. A. 92, 5401–5405
3. Bach, E. A., Tanne , J. W., Ma s e s, S., Ashkenazi, A., Ague , M., Shaw, A. S.,
and Sch eibe , R. D. (1996) Mol. Cell. Biol. 16, 3214–3221
4. Ealick, S. E., Cook, W. J., Vijay-Kuma , S., Ca son, M., Nagabhushan, T. L.,
T o a, P. P., and Bugg, C. E. (1991) Science 252, 698–701
5. Samudzi, C. T., Bu oni, L. E., and Rubin, J. R. (1991) J. Biol. Chem. 266,
21791–21797
6. Sp ang, S. R., and Bazan, J. F. (1993) Cu . Biol. 3, 815–827
7. Benne , M. J., Schlunegge , M. P., and Eisenbe g, D. (1995) P o ein Sci. 4,
2455–2468
8. Xu, D., Tsai, C.-J., and Nussino , R. (1998) P o ein Sci. 7, 533–544
9. Kend ick, B. S., Ca pen e , J. F., Cleland, J. L., and Randolph, T. W. (1998)
P oc. Na l. Acad. Sci. U. S. A. 95, 14142–14146
10. La sen, T. A., Olson, A. J., and Goodsell, D. S. (1998) S uc u e 6, 421–427
11. A akawa, T., Hsu, Y.-R., and Yphan is, D. A. (1987) Biochemis y 26,
5428–5432
12. Mulke in, M. G., and We zel, R. (1989) Biochemis y 28, 6556–6561
13. Kend ick, B. S., Cleland, J. L., Lam, X., Nguyen, T., Randolph, T. W., Manning,
M. C., and Ca pen e , J. F. (1998) J. Pha m. Sci. 87, 1069–1076
14. A akawa, T., Al on, N. K., and Hsu, Y.-R. (1985) J. Biol. Chem. 260,
14435–14439
15. Vandenb oeck, K., Ma ens, E., D’And e´, S., and Billiau, A. (1993) Eu . J. Bio-
chem. 215, 481–486
16. Vandenb oeck, K., Ma ens, E., and Billiau, A. (1998) Eu . J. Biochem. 251,
181–188
17. Wei, J. Y., Gau , J. R., and Hende sho , L. M. (1995) J. Biol. Chem. 270,
26677–26682
18. Vandenb oeck, K., and Billiau, A. (1998) Biochimie (Pa is)80, 729–737
19. Go esman, M. E., and Hend ickson, W. A. (2000) Cu . Opin. Mic obiol. 3,
197–202
20. Hou y, W. A., F ishman, D., Ecke sko n, C., Lo speich, F., and Ha l, F. U.
(1999) Na u e 402, 147–154
21. McCa y, J. S., Ru¨dige , S., Scho¨n eld, H.-J., Schneide -Me gene , J.,
Nakahigashi, K., Yu a, T., and Bukau, B. (1996) J. Mol. Biol. 256, 829–837
22. Ru¨dige , S., Ge me o h, L., Schneide -Me gene , J., and Bukau, B. (1997)
EMBO J. 16, 1501–1507
23. Ru¨dige , S., Buchbe ge , A., and Bukau, B. (1997) Na u e S uc . Biol. 4,
342–349
24. F ank, R. (1992) Te ahed on 48, 9217–9232
25. K ame , A., Schus e , A., Reineke, U., Malin, R., Volkme -Enge , R.,
Landg a , C., and Schneide -Me gene , J. (1994) Me hods Comp. Me hods
Enzymol. 6, 388–395
26. Fou ie, A. M., Samb ook, J. F., and Ge hing, M.-J. (1994) J. Mol. Biol. 269,
30470–30478
27. Kna , G., Mod ow, S., Todd, A., Ge hing, M.-J., and Buchne , J. (1999) J. Biol.
Chem. 274, 29850–29857
28. Maye , M. P., Sch o¨de , H., Ru¨dige , S., Paal, K., Lau en, T., and Bukau, B.
(2000) Na u e S uc . Biol. 7, 586–593
29. Wal e , M. R., and Nagabhushan, T. L. (1995) Biochemis y 34, 12118–12125
30. Zdano , A., Schalk-HiHi, C., Gus china, A., Tsang, M., Wea he bee, J., and
Wlodawe , A. (1995) S uc u e 3, 591–601
31. Zdano , A., Schalk-HiHi, C., Menon, S., Moo e, K. W., and Wlodawe , A. (1997)
J. Mol. Biol. 268, 460–467
32. Gallaghe , G., Dickenshee s, H., Eskdale, J., Izo o a, L. S., Mi ochni chenko,
O. V., Pea , J. D., Vazquez, N., Pes ka, S., Donnelly, R. P., and Ko enko,
S. V. (2000) Genes Immun. 1, 442–450
33. Blumbe g, H., Conklin, D., Xu, W. F., G ossmann, A., B ende , T., Ca ollo, S.,
Eagan, M., Fos e , D., Haldeman, B. A., Hammond, A., e al. (2001) Cell
104, 9–19
34. Dumou ie , L., Louahed, J., and Renauld, J.-C. (2000) J. Immunol. 164,
1814–1819
35. Xie, H.-H., Agga wal, S., Ho, W.-H., Fos e , J., Zhang, Z., S inson, J., Wood,
W. I., Godda d, A. D., and Gu ney, A. L. (2000) J. Biol. Chem. 275,
31335–31339
36. Knappe, A., Ho , S., Wi man, S., and Fickensche , H. (2000) J. Vi ol. 74,
3881–3887
37. Jiang, H., Su, Z.-Z., Lin, J. J., Golds ein, N. I., Young, C. S. H., and Fishe ,
P. B. (1996) P oc. Na l. Acad. Sci. U. S. A. 93, 9160–9165
38. Soo, C., Shaw, W. W., F eymille , E., Longake , M. T., Be olami, C. N., Chiu,
R., Tieu, A., and Ting, K. (1999) J. Cell. Biochem. 74, 1–10
39. Zhang, R., Tan, Z. J., and Liang, P. (2000) J. Biol. Chem. 275, 24436–24443
40. Schae e , G., Venka a aman, C., and Schindle , U. (2001) J. Immunol. 166,
5859–5863
41. Fickensche , H., Ho¨ , S., Ku¨pe s, H., Knappe. A., Wi mann, S., and S ich , H.
(2002) T ends Immunol. 23, 89–96
42. Ko enko, S. V., Saccani, S., Izo o a, L. S., Mi ochni chenko, O. V., and Pes ka,
S. (2000) P oc. Na l. Acad. Sci. U. S. A. 97, 1695–1700
43. Lock idge, K. M., Zhou, S. S., K a i z, R. H., Johnson, J. L., Sawai, E. T.,
Blewe , E. L., and Ba y, P. A. (2000) Vi ology 268, 272–278
44. Tulman, E. R., A onso, C. L., Lu, Z., Zsak, L., Ku ish, G. F., and Rock, D. L.
(2001) J. Vi ol. 75, 7122–7130
45. Lee, H. J., Essani, K., and Smi h, G. L. (2001) Vi ology 281, 170–192
46. Fa , C. D., Galiano, F. J., and Wi , S. N. (1995) Biochemis y 34,
15574–15582
47. Zhu, X. T., Zhao, X., Bu kholde , W. F., G age o , A., Oga a, C. M., Go esman,
M. E., and Hend ickson, W. A. (1996) Science 272, 1606–1614
48. Tsai, C.-J., and Nussino , R. (1997) P o ein Sci. 6, 1426–1437
49. Gi elman, M. S., and Ma hews, C. R. (1990) Biochemis y 29, 7011–7020
50. Taysaco, M. L., and Ca ey, J. (1992) Science 255, 594–597
51. Dudich, I. V., Za yalo , V. P., Bumyalis, V. A. V., and Paulauskas, D. V. (1992)
Mol. Biol. 26, 342–351
52. G zesiek, S., Dobeli, H., Gen z, R., Ga o a, G., Labha d , A. M., and Bax, A.
(1992) Biochemis y 31, 8180–8190
53. Landa , A., Cu y, B., Pa ke , M. H., DiGiacomo, R., Indelica o, S. R.,
Nagabhushan, T. L., Rizzi, G., and Wal e , M. R. (2000) J. Mol. Biol. 299,
169–179
54. G age o , A., and Go esman, M. E. (1994) J. Mol. Biol. 241, 133–135
55. Flynn, G. C., Pohl, J., Flocco, M. T., and Ro hman, J. E. (1991) Na u e 353,
726–730
56. Flynn, G. C., Chappell, T. G., and Ro hman, J. E. (1989) Science 245, 385–390
57. Blond-Elguindi, S., Cwi la, S. E., Dowe , W. J., Lipshu z, R. J., Sp ang, S. R.,
Samb ook, J. F., and Ge hing, M.-J. H. (1993) EMBO J. 75, 717–728
58. Josephson, K., Logsdon, N. J., and Wal e , M. R. (2001) Immuni y 14, 35–46
59. Wal e , M. R., Windso , W. T., Nagabhushan, T. L., Lundell, D. J., Lunn, C. A.,
Za odny, P. J., and Na ula, S. K. (1995) Na u e 367, 230–235
Helix C DnaK-binding Si e in IFN-
␥
and Rela ed Cy okines25676
