Download PDF - University of California, San Francisco

THEJOURNAL
OF BIOLOGICAL
CHEMISTRY
01985 by The American Society of Biological Chemists, Inc.
Val. 260, No. 1, Issue of January 10, pp. 522-530,1985
Printed in U.S.A.
Domain Structure of 3-Hydroxy-3-methylglutarylCoenzyme A
Reductase, a Glycoprotein of the Endoplasmic Reticulum*
(Received for publication, June 26,
1984)
Laura LiscumSQ,
Janet Finer-Moorell, RobertM. Stroudv, Kenneth L. LuskeySII, MichaelS. Brown$,
and Joseph L. Goldstein$
From the $Departments of Molecular Geneticsand InternalMedicine, University of Texas Health Science Center a t Dallas,
Southwestern Medical School, Dallas, Texas 75235 and the llDepartment of Biochemistry and Biophysics, Schoolof Medicine,
University of California, San Francisco, San Francisco, California 94143
We present and evaluate a model for the secondary
structure and membrane orientation of 3-hydroxy-3methylglutaryl coenzyme A reductase, the glycoprotein of the endoplasmic reticulum that controls the rate
of cholesterol biosynthesis. This model is derived from
proteolysis experiments that separate the 97-kilodalton enzyme into two domains, an NHz-terminal membrane-bounddomainof 339 residues and a COOHterminal water-soluble domain of 548 residues that
projects into the cytoplasm and contains the catalytic
site. These domains were identified by reaction with
antibodies against synthetic peptides corresponding to
specific regions in the molecule. Computer modeling of
the reductase structure, based on the amino acid sequence as determined by molecular cloning, predicts
that the NHz-terminal domain contains 7 membranespanning regions. Analysis of the gene structure reveals that each proposed membrane-spanningregion is
encoded in a separate exon and is separated from the
adjacent membrane-spanningregion by an intron. The
COOH-terminal domain of the reductase is predicted
to contain two @-structuresflanked by a series of amphipathic helices, which together may constitute the
activesite. The NH2-terminal membrane-bounddomain of the reductase bears some resemblance to rhodopsin, the photoreceptor protein of retinal rod disks
and the only other intracellular glycoprotein whose
amino acid sequence is known.
findings suggest that the active siteof the reductase is contained within a water-soluble 53-kDa domain that isexposed
to proteases and thus mustproject into the cytoplasm. This
domain is presumablycontiguous with a hydrophobic domain
that fixes the reductase to the ER membrane.
The nucleotide sequence of a cloned cDNA for hamster
reductase was recently established, and the complete amino
acid sequence of the enzyme was deduced (3). This sequence
revealed that the NH2-terminal one-third of the 97-kDa reductase is extremely hydrophobic, while the COOH-terminal
two-thirds is more hydrophilic and typical of a water-soluble
protein (3). These findings
suggest that thehydrophobic NH2terminal end of the reductase binds the molecule to the ER
membrane. The hydrophilic COOH-terminal portion would
thencorrespondtothe
53-kDawater-solublecatalytically
active domain thatprojects into the cytoplasm.
In the current paper,
we test this hypothesis through structural analysis of the reductase protein. For this purpose, we
have prepared antibodies against syntheticpeptides that correspond to specific regions of the reductase sequence, using
the techniques pioneered by Lerner (6) and by Walter and
Doolittle (7). These antibodies have
been used to mapregions
of the reductase that are releasedfrom the membrane by
proteases. The informationderived from these studies isused
in conjunction with computer-based analytic techniques (810) to develop a model for the secondary structure and membrane orientation of the reductase.
EXPERIMENTALPROCEDURES
HMG-CoA reductase,’ the rate-controlling enzyme of cholesterol biosynthesis (l),is a 97-kDa transmembrane glycoprotein that resides in the ER of animal cells (2-4). An
enzymatically active 62-kDa fragment can be released from
ER membrane vesicles by cleavage with an endogenous protease (45). This 62-kDa fragment can
be reduced to a soluble
53-kDafragment by further digestion with a n exogenous
proteasewithout loss of enzymaticactivity(4,5).These
* This research was supported by Grants HL20948 and GM 24485
from the National Institutes of Health. The costs of publication of
this article were defrayed in part by the payment of page charges.
This article must therefore be hereby marked “advertisement” in
accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ Recipient of Postdoctoral Fellowship HL 06533 from the National Institutes of Health.
I( Established Investigator of the American Heart Association.
The abbreviations used are: HMG-CoA reductase, 3-hydroxy-3methylglutaryl coenzyme A reductase; ER, endoplasmic reticulum;
kDa, kilodalton; KLH, keyhole limpet hemocyanin; EGTA, ethylene
glycol bis(6-aminoethyl ether)-N,N,N’,N’-tetraacetic acid; NaDodSO,, sodium dodecyl sulfate.
‘
Materials-Purified Ca2+-activatedproteasefrom bovineheart (11)
was kindly provided by George N. DeMartino of the Department of
Physiology, University of Texas Health Science Center at Dallas.
Compactin was kindly provided by Akira Endo, Tokyo Noko University, Tokyo, Japan. A polyclonal antibody directed against the 53kDa water-soluble fragment of rat liver HMG-CoA reductase was
prepared in rabbits as previously described (12).Other materials were
obtained from previously described sources (4, 12).
Antibodies against Synthetic Peptides-Peptides corresponding to
amino acids (354-368), (379-393),
and (874-887)of thehamster
reductase (see Fig. 1A) were synthesized by solid-phase chemical
methods (13)in the laboratory of Richard Lerner, Department of
Molecular Biology, Scripps Clinic and Research Foundation, La Jolla,
CA. Peptide (379-393)had a cysteine residue added to the COOHterminal end. The composition of each peptide was confirmed by
amino acid analysis. Each peptide was coupled to KLH with maleimidobenzoyl-N-hydroxysuccinimideester (14).New Zealand White
rabbits were injected subcutaneously on days 0 and 14 with synthetic
peptide (0.2mg) coupledto KLH. The peptide-KLH conjugates were
emulsified with Freund’s complete adjuvant (day 0) or incomplete
adjuvant (day 14) in a total of 1.5 ml. Intraperitoneal injections of
synthetic peptide (0.2mg) coupled to KLH were given on days 28
and 35 in a total of 1 nil containing 4 mg of alum. Rabbits were bled
522
Domains of HMG-CoA Reductase
523
first on day 35 and then 6-9 days after each booster injection. y20 @
30 50
40
60 0
In
Globulin fractions of the immune sera were prepared by precipitation
I V G T V T L T I CW S M N M F T G NN K I C G Y N Y E CP K F E E O V L S SO l l l L T l T R C
MLSRLFRllHGLNASHPUEV
with 50% ammonium sulfate (15).
0
110 100
@ 140
90
80 130 120
Cells and Microsomes-UT-1 cells, a compactin-resistant clone of
I A I L Y I Y F Q F QNLRQLGSKY I L G I A G L F TFI S S N F S T V VI H F L D K E L T GL N E A L P F F L L I D L S R A S A L
Chinese hamster ovary cells, were grown in monolayer in Ham's F160
150 200
190
180
@ 170
@ 210
12 medium supplemented with 25 mM N'-2-hydroxyethylpiperazineAKFALSSNSQOEVRENIARGMAILGPTFTLOALVECLVIGVGTMSQVRQLEIMCCFCMSVLANVFVNT
N'-2-ethanesulfonic acid (pH 7.4), 10% (v/v) newborn calf lipopro280
270
260
250
240
230
220
tein-deficient serum, and 40 PM compactin (12). Cells were seeded in
FFPACVSLVLELSRESREGRPIYQLSHFARVLEEEENKPNPVTQRVMIMSLGLVLVHAHSRYIMPSPQ
Petri dishes (60 X 15 mm) or roller bottles and harvested on day 6
290
300320
310
340@ 330
350
(12). Cells were washed a t 4 "C with Dulbecco's phosphate-buffered
NSTTEHSKVSLGLOEOVSKRIEPSVSLYQFVLSMISMOIEQVVTLSLAFLLAVKYIFFEQAETESTLSL
saline, scraped, collected by centrifugation (1000 X g for 10 min a t
4 "C), and
disrupted inone of three ways: Method 1 , addition of buffer
A (15% NaDodSO,, 8 M urea, 10% sucrose, 62.5 mM Tris-chloride,
490
480
470
460
450
440
430
100 mM dithiothreitol at pH 6.8); Method 2, hypotonic incubation (15
GTSPPVMRTQELEIELPSEPRPNEECLQILESAEKGAKFLSOAElIQLVNAKHIPAYKLETLMETHERG
min at 24 "C) in buffer B (10 mM sodium phosphate, 5 mM EGTA,
560
550
540
530
5 00
520
510
10 mM dithiothreitol, 0.1 mM leupeptin, 0.2 mM phenylmethanesulVSIRRQLLSTKLPEPSSLQYLPYROVNYSLWGACCENVIGYMPIPVGVAGPLCLOGKEYQVWATTEGC
fonyl fluoride at pH 6.5) followed by incubation (15 min at 24 "C) in
570
630
620
600
610
buffer B containing1%(w/v) Zwittergent 3-14; or Method 3,590
nitrogen 580 LVASTNRGCR
AIGLGGGASSRVLAOUITRGPVVRLPRACOSAEVKAMLETPEGFAVIKOAFOSTSRFARL
cavitation (4) in buffer C (0.15 M NaC1, 50 mM potassium phosphate,
670
660
650
640
680
7
690
00
5 mM EGTA, 10 mM dithiothreitol, 0.1 mM leupeptin, 0.2
mM phenQKLHVTWAGRNLYIRFQSKT GOACIGMNMIS KGTEKALLKLQEFFPEMQILAVSGNYCTOKKPAAINYIEG
ylmethanesulfonyl fluoride at pH 7.5). When cells were disrupted by
750
740
730
720
710
760
770
Method 3, a 1000 X g supernatant fraction was centrifuged at 100,000
PGKTVVCEAVIPAKVVREVLKTTTEAMIOVNINKNLVGSAMAGSIGGYNAHAANIVTAIYIACGQOARQN
X g for 60 min at 4 "C to obtain a microsomal pellet.
790
780
800
810
840
820
830
Immunoblotting and Other Assays-Samplesfor
electrophoresis
VGSSNCITLMEASGPTNEOLYISCTRPSIEIGTVGGGTNLLPQQACLPL
GVQGACKONP GENARQLARI
were prepared in buffer A. NaDodS04-polyacrylamide gel electropho880
870
850
860
resis and immunoblotting on nitrocellulose were performed as deVCGTVMAGELSLMAALMGHLVRSHMVHKI:
SKIFWCEL~~-I~LKKJ~~
scribed (4) except that 2% (w/v) hemoglobin was used in place of 5%
(w/v) bovine serum albumin. Gels were calibrated with M, standards
obtained from Bethesda Research Laboratories. Protein was meaSOLUBLE W M A I N
c
sured by a modification of the method of Lowryet al. (16). Enzymatic
53kDa
MEMBRANE BOUND
c
DOMAIN
-662
kDa
activity of reductase was measured as described (12). One unit of
reductase activity represents the formation of 1 nmol of meValonate/ NH 2
COOH
min at 37 "C.
CHO
Methods for Analysis of the Secondary Structure of HMG-GOA
Reductase-Hydrophobicity plots of the amino acid sequence were
FIG. 1. Amino acid sequence and domain map of hamster
generated by the method of Kyte and Doolittle (10). The hydrophoHMG-CoA reductase. Panel A , amino acid sequence. Amino acid
bicity values used for each amino acid were those of Eisenberg et al.
residues are shown in the single-letter code, which translates to the
(17). The secondary structure of the extra-membrane regions of three-letter code as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe;
reductase was analyzed by two procedures: 1) a secondary structure G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q , Gln;
prediction scheme described by Garnier et al. (9); and 2) a Fourier R, Arg; S, Ser; T, Thr;V, Val; W, Trp; and Y, Tyr. The 7 postulated
transform analysis of amino acid hydrophobicity described by Finer- membrane-spanning regions are numbered and indicated bysolid
Moore and Stroud (8). The secondary structure prediction scheme lines above and below the sequence. The regions of the protein
analyzes the amino acid sequence by taking into account 8 flanking sequence selected for synthesis of peptides (354-368), (379-393), and
residues on either side of the subject residue and relies on a data base (874-887) are indicated by dashed underlines. Four pairs of closely
of amino acids as they occur in x-ray crystal structures of soluble spaced dibasic amino acids between residues 360 and 400 are indicated
proteins. The Fourier transform analysis was carried out as previously by brackets. The assignment of phenylalanine as the amino acid at
described ( 8 ) , using a variable window size (number of successive residue 673, which was previously ambiguous on the basis of cDNA
residues in each linear transform) of from 11 to 25 residues. The sequencing (3), has now been established from the DNA sequence of
variable window size allows for a determination of the length of a a genomic clone that encodes this region (31). Panel B , linear map of
given feature, such as an amphipathic a-helix or an amphipathic p- HMG-CoA reductase protein showing structural features in relation
sheet in the structure, since the signal will be highest if the window to proteolytic fragments. The holoenzyme has a molecular weight of
size exactly matches the size of the periodic feature. The transforms, 97,000. The cross-hatched areas in the membrane-bound domain
called amphipathic Fourier transforms, were plotted as two-dimen- represent the 7 putative membrane-spanning regions. The presumed
sional contour maps with frequency (l/periodicity) on the vertical site of carbohydrate addition (Asn 281) is indicated by CHO. The two
axis and residue number on the horizontal axis. This procedure postulated sites at which proteolytic clips generate soluble fragments
detects periodicity in hydrophobicity and, therefore, assays an overall of62 and 53 kDa are indicated. The location of the two synthetic
peptides (residues 354-368 and 379-393) used to define the sites of
property of secondary structures (and tertiary folding patterns) as
they evolved in the extra-membrane folded structure. A repeat fre- proteolysis are indicated by asterisks.
quency of 1/3.6 residues" is characteristic of an amphipathic a-helix,
and a frequency of 1/2 residues" suggests an amphipathic 0-pleated protein) as a result of amplification and enhanced transcripsheet structure. Twisted tertiary structures alter these periodicities tion of the reductase gene (18). In previous studies we showed
somewhat, and peaks can, therefore, be diagnostic of packing arrangethat a fully active fragment of the reductase could be released
ment.
-
-
RESULTS AND DISCUSSION
Identification of the 53-kDa Fragment as the COOH-terminal
Domain of HMG-GOAReductase
Fig. 1A shows the amino acid sequence of HMG-CoA reductase as determined from the nucleotide sequence of cDNA
clones isolated from UT-1 cells (3). UT-1cells are a clone of
compactin-resistant Chinese hamster ovary cells (12) that
express large amounts of HMG-CoA reductase (2% of total
from UT-1 cell microsomes as a 53-kDa water-soluble protein
after digestion with a Ca2+-activatedprotease (4).This 53kDa fragment is equivalent to the fragment that was initially
purified fromratliver microsomes and used to generate a
polyclonal anti-reductase antibody (12).
To determine the portion of the intact 97-kDa reductase
that is represented by the 53-kDa water-soluble fragment, we
obtained a synthetic peptide corresponding to amino acids
874-887 at the extreme COOH terminus of the protein (Fig.
1A). After coupling this peptide to keyhole limpet hemocy-
Domains of HMG-CoA Reductase
524
anin, we injected it into rabbits, and the resultant antibody
was used as a probe to identify the COOH terminus of the
reductase. This COOH-terminal peptide antibody was used
in conjunction with the polyclonal antibody to perform the
immunoblotting experiments shown in Fig. 2.
The left panel of Fig. 2 shows a series of immunoblots using
theantibodydirectedagainsttheentire53-kDafragment
(anti-reductase IgG) (Lanes 1-6). When intact UT-1 microsomes were subjected to NaDodS04-gel electrophoresis and
blotted against the anti-reductase
IgG, a predominant 97-kDa
band was seen (Lane I ). (The less prominent bands a t 62 and
53 kDa are proteolytic fragments that are produced during
solubilization of the cells (Refs. 4 and 5 andsee below).) After
centrifugation of the microsomes a t 100,000 x g, nearly all of
the immunoreactive 97-kDa reductase
was found in the membrane pellet (Lane 3 ) and very little was found in the supernatant (Lane 2).When the microsomes were treated with a
Ca'+-activated protease, the predominant immunoreactive
97kDa reductase was reduced to 53kDa (Lane 4 ) . After centrifugation a t 100,000 x g, most of this fragment remainedin the
supernatant (Lane 5 ) , and little was found in the pellet (Lane
6).
The right panel of Fig. 2 shows aliquots of the same incubation mixtures that were subjected to electrophoresis and
immunohlotted against the anti-COOH-terminal peptide antibody (Lanes 7-12). This antibody reacted with the intact
97-kDa protein (Lane 7), which was found in the pellet after
centrifugation (Lane 9).After treatment with the Ca'+-activatedprotease,theanti-COOH-terminalantibody
reacted
with the 53-kDa fragment (Lane IO),which remained in the
supernatant after centrifugation (Lane I I ). In the same experiment, we measured the reductase activity in the 100,000
x g supernatant and pellet fractions. After proteolysis,
of the microsomal reductaseactivity was recovered in the
5.7-kDa
100,000 X g supernatant (data not shown). Thus, the
water-soluhlecatalyticallyactivefragmentrepresentsthe
COOH-terminal portionof t,he reductase. Since this fragment
can he releasedfrommicrosomes
by proteolysis, it must
project to the exterior of the microsomes, i.c. into the c-ytoplasm.
Identification of the Membrane-boundFragmcnt and
Localization of Proteolytic Clcavagc Sites
Reductasecontains a cluster of four pairs of positively
charged amino acids located in a proline-rich region between
residues 360 and 400 (Fig. 1 A ) . These basicresidues were
deemed to be likely sites for the proteolytic cleavage that
gives rise totheCOOH-terminal62-kDafragmentduring
incubation of the cells in hypotonic buffer. Multiple attempts
todeterminetheamino
acidsequence of theproteolytic
fragments after labeling of the cells with radioactive amino
acidsfailed. This failure has occurred underconditions in
which we can easily obtain sequence information from the 97kDa holoprotein (3). We believe that our failure to ohtain
sequence information on the53- and 62-kDa fragmentsis due
to the fact that the proteolytic
clip is not a clean one and that
Anti-Reductase
Anti -Peptide (874-887)
multiple amino termini around the area
of the proteol-ytic
' 1
2 3 4 5 6"7
8 9 10 1 1 1 2 ' cleavage are formed.
We, therefore, decided to localize the sites of proteolytic
rr)
I
cleavage by probing with anti-peptide antibodies.Accordingly,
9
we prepared antibodies against synthetic peptides correspondx 97 c
(379-293).
ing toamino acids (354-368) andaminoacids
which are twoneighboring stretches on the molecule that
might flank the predicted protease cleavage site (Fig. 1A ).
These two antibodies were used to localize the proteol-ytic
-m
fragments of the reductase after incubation of intact cells in
P
a hypotonic buffer, which increases the formation of the 62kDa fragment (4, 5), and after treatment in oitro with the
S P M
M S PS MP
1
Ca'+-activated protease, which reduces the 62-kDa fragment
to 53 kDa. The proposed sites of proteolysis based on the
experiments described below are shown schematically in Fig.
PROTEOLYSIS
1R.
FIG. 2. NaDodS0.-polyacrylamide gel electrophoresis and
3 (Lanes 1 - 4 ) shows a series of
immunoblotting analysisof soluble and membrane-bound pro- The left panel ofFig.
were immunoblots performedwith the anti-reductase antibody( i x .
teolyticfragments of HMG-CoA reductase.UT-1cells
disrupted by Method 3 , and a microsomal pellet was isolated as the antibody directed against the entire COOH-terminal5.7described under "Experimental Procedures." Aliquots of total micro- kDa fragment). When UT-1 cells were harvested in the ahsomes (300 pg of protein; 22-36 units of reductase activity) were sence of hypotonic incubation, the antihody stained the intact
incuhated in buffer D with proteolytic inhibitors (50 mM Tris-chloride, 5 mM dithiothreitol, 5 mM EGTA, 0.1 mM leupeptin, 0.1 mM 97-kDa protein (Lane 1 ) . When cells were suhjected to hyphenylmethanesulfonyl fluoride at pH 7.5) at 4 "C for 30 min (Lanes potonic incubation prior tosolubilization, much o f the reductase was reduced to the 62-kDa form ( I m w 2)as a result of
1-3 and 7-9)or in buffer E without proteolytic inhibitors (50 mM
Tris-chloride, 5 mM dithiothreitol, 5 mM CaCI2 at pH 7.5) supplecleavage by the endogenous protease (4, 5 ) . Previous studies
mented with 15 pg of Ca2+-activatedprotease at 24 "C for 30 min have shown that this 62-kDa fragment contains virtually all
(Lanes 4-6 and 10-12). After incubation, Na2C0,was added to 0.1 M of the catalytic activity of the intact 97-kDa enzyme (4, 5).
to remove loosely adherent proteins, and EGTA and leupeptin were
adjusted to 5 and 0.1 mM, respectively, in all tuhes. An aliquot from When microsomes were isolated and incubated with the exeach tube (125 pg of protein) was centrifuged at 100,000 x g for 20 ogenous Ca2+-activated protease, the immunoreactive reducmin at 24 'C to obtain a supernatant fraction (S)and a pellet ( P ) . tase was cleaved to the 5.7-kDa fragment, which remained in
The pellet wasresuspended to thevolume of the supernatant fraction. the supernatant after a 100,000 X g centrifugation ( I m w .?).
Aliquots (40 pl, 20 pg of protein) of microsomes before centrifugation Very little of the 53-kDa fragment was found in the pellet
( M )and aliquots (40p l ) of the supernatant fractions and pelletswere (Lane 4 ) .
solubilized in buffer A and subjected to electrophoresis and immuThe middlepanel of Fig. 3 shows aliquots of the same
noblotting with polyclonal rabbit anti-reductase IgC at 10 pg/ml or
incubation mixtures that were subjected to electrophoresis
anti-peptide (874-887) antihody (y-globulin fraction) at 100 pg/ml,
followed by '2LI-labeledgoat anti-rabbit IgG (10' cpm/ml) and auto- and immunoblottedwith the antibody directed against peptide
radiography for 18 h at -70 "C.
(354-368). Again, this antibody identified the intact 97-kDa
L
d
1
+
-
+
Domains of HMG-CoA Reductase
Anti-Peptide
Anti-Peptide
354 - 368
379 - 393
.
2i
.i
(379-393) are present on the 62-kDa fragment but are removed when this fragment is clipped further to the X-kDa
I
form.
10
The above data suggest that the endogenous protease that
9
liberates
the 62-kDa fragment after
h.ypotonic incubation cuts
X
the reductase in the region between peptides (354-368) and
(379-393) (see Fig. 1, A and H ) . This region, from residues
368 to 379, contains two pairs of basic residues (Fig. 1A ). If
thecleavageoccurredint.hisregion,thereleasedsoluble
fragment would have a mass of 56 kDa, which agrees reasonably well with the 62 kDa that was estimated bv NaDodS0,gel electrophoresis.Most of the62-kDafragmentremains
bound to the UT-1 membranes at normal ionic strength ( 5 ) .
However, much of it can be released into the supernatant if
the microsomes are washed with a solution containing 0.1 M
Na,CO:I, suggesting that the 62-kDa fragment is soluble and
(data not
bound to themembranebyionicinteractions
shown).
Fraction I C C S P
C C
S P I C C S
q
, 35The other fragment produced by the cleavage ( i . ~the
kDa fragment) remains totally memhrane hound after proteFIG. 3. NaDodS0.-polyacrylamide gel electrophoresis and
olysis, even after washing in Na,CO, buffer (see Fig. 3, 1anr.s
immunoblotting analysis of proteolytic fragments of HMG7 a n d 8 ) . This fragment is known to contain cart)ohydrate
CoA reductase. UT-1 cells were disrupted by Methods 1. 2 , o r :{ a s
it is labeled with [:'H]glucosamine ( 4 ) . If proteolysis
because
indicated helow. Cell extracts ( C ) ohtained hy Methods 1 and 2 were
occurs between residues 368 and 380 as discussed above, the
soluhilized in hufferAwithoutfurthertreatment.Microsomesohtained hv Method :%
(225 p g of protein) were incuhated in huffer E
membrane-bound NH,-terminal fragment should have an M,
without proteolytic inhihitors and supplemented with 4.5 pg of Ca2+- of 42,000, as compared with a value of 35,000 estimated by
activated protease at 24 "C for 30 min. After incuhation, Na2COn was electrophoresis. It is possible that the hydrophobic fragment
added, EGTA and leupeptin were adjusted to 5 and 0.1 mM, respecis really 42 kDa, but
it binds more NaDodSO, than the
M,
tively, and the sample was centrifuged to obtain a supernatant fraction ( S ) and pellet (1') a s descrihed in Fig. 2 . Aliquots (as in Fig. 2) standard proteins and, therefore, has an electrophoretic mobilitygreaterthanexpectedonthebasis
of its true M,.
of soluhilized cell extracts, supernatant
t'ractions, and pellets were
suhjected to electrophoresisandimmunohlottingwithpolvclonal
Alternatively, it is possible that a protease clips the protein
rahhit anti-reductase lgG at 10 pglml. anti-peptide (354-368) antiat a n exposed site near the NHZ-terminal end, reducing the
30 pg/ml, or anti-peptide (379-393)
hodv(7-glohulinfraction)at
42 to 35 kDa.
mass of the major membrane fragment from
antihody (y-globulin fraction) at 200 pg/ml followed hy ""I-laheled
We
have
attempted
to
test
this
possibility
hv
preparing an
goat anti-rahhit IgC ( 1 0 " cpm/ml) and autoradiography for 9 h ( I x n m
antibody against the extreme NH,-terminal peptide. Unfor1-4) or 24 h ( I x n r s -5-12) at -70 "C. 1mw.q I , 5 , 9, cells ( C )prepared
hy rapid soluhilization (Method 1 ) ; l a n e s 2, 6. 1 0 , cells ( C )suhjected tunately, multiple immunizations have failed to yield antihodto h.ypotonic incuhationandsoluhilizationwithZwittergent3-14
ies that react with the intact protein, perhaps because there
(Method 2); 1 m w . s .'I, 7, 11, supernatant fraction ( S ) of microsomes
are only 9 hydrophilic amino acids at the NH, terminus (see
prepared hv Method 3 and suhjected to proteolysis and centrifugation; below).
1.anr.s 4. 8. 12, pellet (1') of microsomes prepared hy Method 3 and
From the dataof Fig. 3, we can estimate the pointat which
suhiected to proteolysisandcentrifugation. (No proteinhandsare
thecleavagetothe53-kDafragmenttakesplace.
If t h e
visihle in Ixnc 11; the smudge in this lane is a hlotting artifact due
NaDodSO, gel reflects a true difference of 9 kDa hetween the
to an imperfection in the gel.)
62- and 53-kDa fragments, then the cleavage that gives rise
prot.ein (I,anc ,5). When the cells were subjectedto hypotonic to the 53-kDa fragment must occur in the region of residues
incubation, the 62-kDa form of the reductase that was visu- 4.50-470 (Fig. 1, A and R ) . Since the 53-kDa fraffment retains
full enzymatic activity, this would place the active site
of the
(Lane 2 ) wasnot
alizedwiththeanti-reductaseantibody
of residue 470.
visualized wit.h the anti-peptide (354-368) antibody (Lane 6). enzyme on the COOH-terminal side
Instead,thisanti-peptideantibodyreactedwith
a 35-kDa
fragment. When isolated microsomes were incubated with the Model for Secondap Structurc of HMC-C'oA Rrductnsc
exogenous Ca"-activated protease, this 3.5-kDa fragment was
The above proteolysis experiments reveal that the
NH,also formed. After centrifugation the 35-kDa fragment was
terminal one-third of the reductase is associated with the Eli
found in the membrane pellet (Lane 8 ) and not in the super- membrane and that the COOH-terminal two-thirds, which
natant ( I h e 7 ) . (The broad 70-kDa band Lane8
in
is believed contains the catalytic site, projects into the cvtoplasm.
To
to he a dimer of the 35-kDa fragment.) Thus, the proteases
learnmoreaboutthestructures
of thesetworegions,
we
must. be cutting the reductase on the COOH-terminal side of turnedtocomputer-modelingschemesthatareknown
to
peptide (354-368), leaving this peptide (or at least the major predict, with reasonable accuracy, the secondary structuresof
part of it) on a 35-kDa fragment that remains bound to t h e proteins based on characteristicsof the amino acid sequence.
membrane.
To locate hydrophobic regions of the sequence that span the
T o localize the cleavage site more precisely, we used the
ER membrane, we used the method of Kyteand1)oolittle
antibody against peptide (379-393). This antibody also iden- (10). This method involves calculation of the relative hydrotified the int,act 97-kDa protein
( I h e 9 ) . After hypotonic
phobicity at each amino acid residue in the reductase sequence
incubation, it recognizedthe62-kDafragment
( I m w 10). as averaged over the adjacent 21 residues. In our model we
When the microsomes were digested with the Ca2+-activated have drawn these putative membrane-spanning regions
in t h e
protease and then subject.ed to electrophoresis, the anti-pep- n-helical confimration because thevpreof the correct length,
tide (379-393) antibody no longer reacted with any
of t h e about 26 amino acids, to span a 40-A bilayer, and because in
fragments (1,nnc.s 1 1 a n d 12).Thus, the amino acids in peptide
for which
the structureof bacteriorhodopsin, the only protein
Anti -Reductase
I
I
526
Domains of HMG-CoA Reductase
263G
J /
FIG. 4
Domains of HMG-CoA Reductase
a
527
postulated membrane-spanning regions in this protein (Table
I). This membrane-spanning region is structurally similar to
membrane-spanning region 7 of bacteriorhodopsin, according
to the criteriaof McLachlan (27).
3
0.2
ALPFFLLLIDLSkASALAkFALSS (HMG-CoA reductase)
0.0
H
IETLLFMVLDVSAkVGFGLILLkS (bacteriorhodopsin) (Ref.21)
-0.2
Thismembrane-spanning region of thereductase shows
strong peaks in the Fourier transforms at
1/3.6 residues",
suggesting that it has an amphipathic a-helical
configuration
(Fig. 7). This structuremay be similar to the amphipathica- 0.64
I
I
I
helix that spans thebilayer in the acetylcholine receptor (8).
0
200
400
600
800
Several other predicted membrane-spanning segments of
the reductase contain amino acids that are usually charged.
Residue
These residues tend to cluster on
faces of the putativehelices,
FIG. 5. Hydrophobicity ( H ) plots of the amino acid sequence making ion pairing within orbetween packed helices possible
of HMG-CoA reductase. Residue hydrophobicitiesh, were averaged
(Fig. 4).There are approximately equal numbers
of positively
over windows of 21 residues. The uertical scale denotes the average
free energy in kilocalories/mol/aminoacid for transfer from a hydro- charged and negatively charged amino acids in the predicted
intramembrane regions of the reductase.
phobic to a hydrophilic environment. The residue number is plotted
on the horizontal scale. The 7 predicted membrane-spanning regions
Each of the 7 membrane-spanning regions in the reductase
are numbered 1-7. The two regions that are predicted to have 0is separated from the next
by a hydrophilic linker.As depicted
structure and that contain hydrophobic stretches are labeled bl and in Fig. 4, the linkers contain numerous chargedresidues and
b2 (see Fig. 4).
are predicted tocomprise extended 6-structure orloops. The
longest linker region extends from residues 221 to 314. It is
the structure of membrane-spanning regions is known (19postulated to lie on the luminal side of the ER membrane
23), the a-helices are oriented perpendicular to the membrane
between the sixth and seventh membrane-spanning regions.
plane (20).
residues 221 t o 246 within
To identify regions of secondary structure in the reductase The Fourier transform predicts that
this
linker
form
an
amphipathic
helix,
whereas residues 247
protein, we employed a Fourier transform analysis to search
to
307
have
an
amphipathic
extended
@-structure
(Fig. 7). A
for periodicities in hydrophobicity that would suggest the
short
stretch
of
hydrophobic
residues
(258-267)
is
present
but
presence of amphipathic a-helixor 6-pleated sheet structures
either inside or outside the hydrophobic segments (8). The is not long enough to span the membrane.
Residues 221-314 includea potentialsite for N-linked
secondary structure prediction algorithmof Garnier et al. (9)
glycosylation
at asparagine 281, the only N-linked glycosylawas also used t o help predict the secondary structure
of these
regions. This program utilizes a data base of amino acids as tion signal (i.e. asparagine-X-serine or asparagine-x-threothey occur in x-ray crystal structures. proposed
The
secondary nine) in the NH2-terminal 35-kDa membrane-bound portion
structure based on combined
the
useof these analytic methods of the protein. Since the NH,-terminal 35-kDa fragment is
known to contain N-linked carbohydrate (4), this constitutes
is presented in Fig. 4.
supportive evidence thatasparagine 281 is locatedin the
NH2-terminal Domain (Residues1 to 339)"The Kyte-Doolittle plot of amino acid hydrophobicities revealed that the lumen of the ER.
COOH-terminal Domain (Residues
340-887)"As shown by
NH2-terminal domain of the reductase (residues 1-339) is in
the Kyte-Doolittle plot inFig. 5, the COOH-terminal domain
generalmuch more hydrophobic than the COOH-terminal
general hydrophilic.This region does contain
domain (Fig. 5). Within the first339 residues, seven peaks of of reductase is in
hydrophobicity are noted, each of which extends over a dis- a few short hydrophobic segments and a long one between
tance large enough to span
a membrane bilayer (Fig. 5, regions residues 520 and 545 labeled bl in Fig. 5. None of these
1-7). The boundaries andoverall hydrophobicities of each of segments can be membrane-spanning regions since each is
these putative membrane-spanning
regions are listed in Table contained in the water-soluble 53-kDa proteolytic fragment
of the reductase. They probably form a hydrophobic core in
I.
Fig. 6 shows a detailed diagramof the predicted orientation this globular domain. The COOH-terminal domain includes
of the amino acids in the NH2-terminal domain
of the reduc- 7 long amphipathic regions with the periodicity of a-helices
(2.6 to 4 residues) interspersedwith turns, random coil structase. The seven putative membrane-spanning segments are
tures, and extended @-structures depicted
as
in Fig. 4.
in generallesshydrophobic
thantheputativemembraneResidues 351 to 365, which are just outside the seventh
spanning regions inretinalrhodopsin (19, 24) and in the
membrane-spanning region, have the unusual sequence ( x - y acetylcholine receptorsubunits (8, 25) butaresimilarin
hydrophobicity to those in bacteriorhodopsin
(21-23) and the z-proline)4.The amphipathic Fourier transformsshow a peak
lac carrier protein of Escherichia coli (26).
of periodicity at 1/6.5 residues-' (Fig. 7), which is exactly the
The fourthhydrophobic stretch of reductase, which extends frequency found for the proline-richsequence in avian panfromresidue 124to 149, isthemost
hydrophilic of the creatic polypeptide, whose crystal structure is known(28).
I
I
1
1
FIG. 4. Schematic model for the predicted secondary structure of HMG-CoA reductase. All charged
residues are marked with a + or -. Cysteine residues are indicated by S. * indicates a potential N-linked
glycosylation site at residue 281. G, glycine; N, asparagine; P, proline; Q, glutamine; R, arginine. The 40-A lipid
bilayer boundaries are drawn on either side of the predicted membrane-spanning regions. Residues 516-586 and
734-830, enclosed by the clotted lines, each contain 8 t o 11 stretches of extended structure and are rich in cysteine,
proline, and glycine residues.
L)ornains o ~ ~ ~ G - Reductase
C O A
528
TABLE
I
Hydrophobicity values for the me~~rane-spanning
regions of the
~ H ~ - t e rdomain
m i ~ ofH ~ e - e o Areductase
"
Membranespanning
region
Residues
____
Length
Average
hydrophobicity"
no.
residues
1
2
3
4
5
6
7
25
29
10-39
57-78
90-114
124-149
160-187
192-220
315-339
30
22
25
26
28
0.21
0.20
0.36
0.15
0.25
0.27
0.19
The value for each region represents the average hydrophobicity/
amino acid (kcai/mol) for the most hydrophobic set of 20 contiguous
residues.
The Fourier plots would, therefore, suggest that these residues
form a p o l ~ r o ~ i n e - l i khelix
e
similar to that found in the
proline-rich region of pancreatic polypeptide. This region is
adjacent to the paired basic residues that appear to be the site
where the active fragment of the reductase is cleaved proteolytically from the membrane-spanning domain (see above).
Two regions of extended @-structureare predicted within
the COOH-terminal domain between residues 495-595 and
between 735-825 (Fig.4). Both of these regions are quite
bland in the amphipathic transformand
have several
stretches of hydrophobic extended sequences interspersed
with short hydrophilic stretches and other characteristic
turnforming residues. Both regions are rich in glycines, prolines,
and cysteines. Together, residues 516-586 and 734-830 con-
stitute only about 19% of the sequence, yet they include 10
out of 27 cysteines in theprotein as well as nearly one-half of
the glycines. Each of these two regionshas a central relatively
hydrophobic core (bl and b2 in Fig. 5 ) that is almost symmetrically flanked by sequences with very strong amphipathic
a-helical periodicity in the Fourier transform. The amphipathic structures undoubtedly interface between the hydrophobic core and the solvent, although we cannot predict the
detailed arrangement of secondary structure elements into a
three-~mensionalstructure from results of our analysis.
It seems likely that residues from the putative @-domains
form at least part of the active site for the reductase. In this
regard, it is of interest that thereductase is known to require
high concentrations of thiol-reducing agents for activity (29)
and that there are large numbers of cysteine residues in the
predicted @-domains.The enzyme reaction by whichreductase
catalyzes the conversion of HMG-CoA to mevalonate is believed to involve thetransient protonation of a histidine
residue (30). In this regard, 7 histidines are located at the
periphery of the @-domainsat residues 474,487,634,751,860,
865, and 868.
Relation between E x ~ n - ~ n t r o~n ~ rof the
~ Gene
~ and
~ r
Domain Structure of the Protein
The reductase is encoded by a 25-kilobasegene,which
contains 20 exons that are separatedby 19 introns (31). Eight
of these introns occur within the NH2-terminal hydrophobic
domain. Fig. 8 (upper panel)relates the positions of these 8
introns to the 7 membrane-spanning regions of the reductase
protein. As indicated by the arrows, 7 of theintronsare
located near the junction of a membrane-spanning region and
pGV%[4%
Amino AcidsI@SA-COOH
Lumen
of ER
N
P
V
T
Q
FIG.6. Model for the possible orientation of the hydropho~icdomain of HMG-GOAreductase in the
membrane of the endoplasmic reticulum. Amino acid residues are shown in the singie-letter code (see legend
to Fig. I). Amino acids with positively charged side chains are shown in circks. Amino acids with negatively
charged side chains are shown in squares. The N-linked carbohydrate chain a t residue 281 is indicated between
the sixth and seventh membrane-spanningregion.
e
Domains of HMG-CoA Reductase
529
*
*
c m
H M G CoA
Reductase
cytop1a.m
ER
Crystalloid
ER
iumm
of ER
Rhodopsin
of D8.k
Rod Outer
Segment
FIG. 7. Amphipathic Fourier transforms between frequencies 0 and 112 residues" for the sequence of HMG-CoA reductase. The transforms were calculated with a moving window of 20
residues and were not normalized (8).The central residue number is
plottedonthe
horizontalanis;uerticallines
aredrawn every 10
residues on the plots. The frequency is plotted on thevertical anis. A
frequency of 1/2 corresponds to the characteristic amphipathic
pstructure frequency of 1/2 residues". A horizontal line on the plot is
drawn at the characteristic amphipathic a-helix frequency of 1/3.6
residues". Seven hydrophobic segments between residues 1 and 339
are predicted to span the membrane as a-helices; their positions are
denoted with open cylinders above the horizontal axis and are numbered 2-7. The hydrophilic sequence 220-310 is predicted to contain
an amphipathic a-helix(shaded cylinder)and a (?-strand(dotted line).
Residues 495-595 and 735-825, marked by dottedlines along the
horizontal anis, are predicted to have a P-structure. On either side of
these regions, the predicted amphipathic a-helices are marked with
shaded cylinders. The hydrophobic sequence 846-862 is predicted to
have an a-helical structure (open cylinder).
FIG. 8. Similarities in the membrane organization of HMGCoA reductase (upper panel) and rhodopsin (lower panel).
The position in the protein sequence where an intron interrupts the
coding region of the gene for each protein is indicated by an arrow.
The DNA sequence data for exon-intron junctions for the reductase
and rhodopsingenes were obtained from Refs. 31 and 36, respectively.
Of the 19 introns in the reductasegene, one occurs in the 5' untranslated region and 18 occur within the coding region (31). These 18
intronsinterruptthe
gene atthepositionscorrespondingtothe
following amino acid residues: 55, 93, 122, 150, 186, 221, 260, 314,
397,455,520, 573, 626, 661,718, 765,818, and870 (31).
only two intracellular glycoproteins whose amino acid sequences are known. They show asurprisingsimilarity
in
several respects including mode of biosynthesis, post-translational glycosylation and phosphorylation, location in membranes of intracellular organelles, organization of 7 membrane-spanning regions, and domain structures of their genes.
These similarities are discussed below and illustrated schematically in Fig. 8.
Rhodopsin consists of an apoprotein, opsin, of 348 amino
acids that bears a covalently attached polyisoprenoid chromophore, 11-cis-retinal (24).Rhodopsin is inserted cotranslationally into the rough ER even though it has no cleaved
signal sequence (33, 34). The protein is then transported to
an external loop. These introns divide the coding region of the rod-outer-segment disk where it remains embedded as an
the gene in such a way that each of the first 6 membraneinternal membrane protein (reviewed in Ref. 34). Hydrophospanning regions is separated from the adjacent membrane- bicity plots of the amino acid sequence of rhodopsin suggest
spanning region by a single intron. The coding region for the that the protein crosses the membrane 7 times (24,35).
relatively large hydrophilic domain between the sixth and
"High mannose" carbohydratechainsare
added to two
seventh membrane-spanning region is split by 1 intron.
asparagine residues near the NH2 terminus during synthesis
In theCOOH-terminal domain of the reductase, the introns of rhodopsin in the ER (34). Rhodopsin also undergoes phosgenerally fall on thehydrophilic side of predicted amphipathic phorylation on serine and threonineresidues near the COOHhelices (4 introns occur at residues 455,626, 661, and 718) or terminal endof the molecule (35). The gene encoding rhodopin predicted @-turns(4 introns occur at 573, 765, 818, and sin contains 4 introns, 3 of which interrupt the coding se870) (31). One intron occurs in the center of a predicted p- quence at points corresponding to the interface between a
strand (residue 520), and anotheroccurs in the vicinity of the postulated membrane-spanningregion and anexternal hydroproteolytic site that generates the 62-kDa fragment of the philic loop (36) (Fig. 8).
reductase (residue 397). These findings are consistent with
Like rhodopsin, the amino acid sequence of HMG-CoA
the hypothesis that introns usually occur at the surface of reductase suggests a protein with 7 membrane-spanning reproteins (32).
gions and a hydrophilic NH, terminus that projects into the
lumen of the ER(Fig. 8).Like rhodopsin, reductase is believed
Similarities between HMG-CoA Reductase and Rhodopsin
to be inserted into the ERcotranslationally (37),even though
The 339-residue membrane domain of HMG-CoA reductase it lacks a cleavable signal sequence (3, 37).
bearsa strong resemblance to rhodopsin, the 348-residue
Like rhodopsin, reductase has a high-mannose N-linked
membrane proteinof the retinalphotoreceptor that is respon- oligosaccharide that is added cotranslationally and is subsesible for light reception (24). Reductase and rhodopsin are the quently trimmed but is never processed to the mature form
530
Domains of H M ~ - ~ Reductase
oA
(4). Unlike rhodopsin, the carbohydrate of reductase is not
near the NH2terminus but appears to be located between the
sixth and seventh membrane-spanning regions. The assignment of this location is based on two observations: 1) the
carbohydrate remains with
the 35-kDa membrane-bound fragment after proteolysis (4); and 2) asparagine 281 is the only
asparagine in this 35-kDa fragment that is followed by the
sequence X-serine or X-threonine (Fig. 1).Reductase has two
other potential N-linked glycosylation signals, both of which
are found on the water-soluble 53-kDa fragment(3).However,
this fragment contains no carbohydrate (4), presumably because the asparagine residues at these two glycosylation signals are never exposed to the lumen of the ER. Like rhodopsin,
reductase is phosphorylated on two or more serine residues in
its COOH-terminal domain (38).
Like rhodopsin, the reductase is targeted to a specialized
intracellular membraneafter synthesis(Fig. 8). In UT-1cells,
the reductase is localized in an extensive network of smoothsurfaced tubules known as the crystalloid ER (12, 39). Both
the crystalloid ER (39) and therod-outer-segment disk membrane (34) are extremely poor in cholesterol. It is attractive
to speculate that the 7 membrane-spanning regions of the
reductase and rhodopsin may play a role in stabilizing the
membranes of these cholest~roi-poor
organelles.
The tendency of introns to be inserted into the genome
between exons that encode membrane-spanning regions,
which was first noted by Nathans and Hogness in rhodopsin
(36), is illustrated much more dramaticallyin HMG-CoA
reductase (Fig. 8). This finding suggests that each membranespanning region represents a separate evolutionarily distinct
domain. The amino acid sequence of each membrane-spanning region is unique and, therefore, these regions are not
repetitions of the sameexon. No homology between these and
other transmembrane sequences in rhodopsin or other proteins hasbeen identified.
One other relation between reductase and rhodopsin is
intriguing. The covalently bound 11-&-retinal, which forms
the functional unit of animal-cell rhodopsin, is a polyisoprenoid compound that is synthesized in plantsfrom mevalonate,
which is produced by HMG-CoA reductase. The similarity
between rhodopsin, which uses a polyisoprenoid, and reductase, which synthesizes the precursor of isoprenoid compounds, raises the possib~lity that the reductase might also
contain a mevalonate-derived isoprenoid bound covalently or
noncovalently to its hydrophobic domain. It is conceivable
that such a substance might function to regulate the rate of
degradation of the reductase, a reaction that i s known to be
accelerated by a nonsterol product derived from mevalonate
(1, 40).
Ackno~~edgments-Weare grateful to Richard Lerner for advice
on the strategy for preparing synthetic peptides and for generously
providing us with the peptides used in this study. Clarice Grimes,
Deborah Thompson, and Claudia Stewart provided excellent technical assistance.
REFERENCES
1. Brown, M. S., and Goldstein, J. L. (1980) J. Lipid Res. 21,505-
517
2. Chin, D. J., Luskey, K. L., Faust, J. R., MacDonald, R. J., Brown,
M. S.. and Goldstein. J. L. 11982) Proc. Natl. Acad. Sci. U. S.
A. 7 9 , 7704-7708
3. Chin. D. J.. Gil.G.. Russell, D. W.. Liscum. L.. Luskev. K. L..
Basu, S. K., Okayama, H., Berg,P.1 Goldstein, J. L., anhBrown;
M. S. (1984) Nature (Lond.)308,613-617
'
4. Liscum, L., Cummings, R. D., Anderson, R.G. W . , DeMartino,
G. N., Goldstein, J. L., and Brown, M. S. (1983) Proc. Natl.
Acad. Sci. U. S. A. 80,7165-7169
5 . Faust, J. R., Luskey, K. L., Chin, D. J., Goldstein, J. L., and
Brown, M. S. (1982) Proc. Natl. Acad. Sci. U. S. A . 79,52055209
6. Lerner, R. A. (1982) Nature (Lond.)299,592-596
7. Walter, G., and DoolittIe, R. F. (1983) in Genetic ~ ~ i ~ e r i ~ :
Principles and M e ~ (Setlow,
~ s J. K., and Hollaender, A., eds)
Vol. 5, pp. 61-91, Plenum Press, New York
8. Finer-Moore, J., and Stroud, R. M. (1984) Proc. Natl. Acad. Sci.
U. S. A . 8 1 , 155-159
9. Garnier, J., Osguthorpe, D. J., and Robson, B. (1978) J. Mol.
Biol. 1 2 0 , 97-120
10. Kyte, J., and Doolittle, R. F. (1982) J. Mol. Bwl. 157, 105-132
11. DeMartino, G.N., and Croall, D. E. (1983) Biochemistry 2 2 ,
6287-6291
12. Chin, D. J., Luskey, K. L., Anderson, R. G.W., Faust, J. R.,
Goldstein, J. L., and Brown, M. S. (1982) Proc. Natl. Acad. Sci.
U. S. A. 79,1185-1189
13. MargIin, A., and Merrifield, R.B. (1970) Annu. Rev. Biochem.
3 9 , 739-866
14. Russell, D. W . , Schneider, W. J., Yamamoto, T., Luskey, K. L.,
Brown, M. S., and Goldstein, J. L. (1984) Cell 37, 577-585
15. Garvey, J. S., Cremer, N. E., and Sussdorf, D. H. (eds) (1977) in
Methods inImmunology, 3rd Ed., pp. 218-219, W. A. Benjamin,
Inc., Reading, MA
16. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J.
(1951) J. Biol. Chem. 193,265-275
17. Eisenberg, D., Weiss, P., Terwilliger, T., and Wilcox, W. (1982)
Faraday Symp. Chem. SOC.17,109-120
18. Luskey, K. L.,Faust, J. R., Chin, D. J., Brown, M. S., and
Goldstein, J. L. (1983) J. Biol. Chem. 258,8462-8469
19. Ovchinnikov, Y. A. (1982) FEBS Lett. 148, 179-191
20. Unwin, N., and Henderson, R. (1984) Sci. Am. 250, 78-93
21. Seehra, J. S., and Khorana, H.G. (1984) J. Biol. Chem. 2 5 9 ,
4187-4193
22. Ovchinnikov, Y. A.,. Abdulaev, N. G., Feigina, M. Y., Kiselev, A.
V., and Lobanov, N. A. (1979) FEBS Lett. 100,219-224
23. Agard, D. A., and Stroud, R. M. (1982) B~ophys.J. 37,589-602
24. Hargrave, P. A.,McDowell, J. H., Curtis, D.R., Wang, J. K.,
Juszczak, E., Fong, S.-L., Rao, J. K. M., and Argos, P. (1983)
Biophys. Struct. Mech. 9 , 235-244
25. Claudio, T., Ballivet, M., Patrick, J., and Heinemann, S. (1983)
Proc. Natl. Acad. Sci. U. S. A . 80,1111-1115
26. Foster, D. L., Boublik, M., and Kaback, H. R. (1983) J. Biol.
Chem. 258,31-34
27. McLachlan, A. D. (1971) J. Mol. Biol. 61,409-424
28. Blundell, T. L.,,Pitts, J. E., Tickle, I. J., Wood, S. P., and W u , C.
W. (1981) Proc. NatL Acad. Sci. U. 5'. A. 78,4175-4179
29. Roitelman, J., and Shechter I. (1984) J. Biol. Chem. 2 5 9 , 870877
30. Rogers, D. H., Panini, S. R., and Rudney, H. (1983) in 3-Hydroxy3-Methylglutaryl Coenzyme AReductase (Sabine, J. R., ed) pp.
58-75, CRC Press, Inc., Boca Raton, FL
31. Reynolds, G., Basu, S. K., Osborne, T. F., Chin, D. J., Gil, G.,
Brown, M. S., Goldstein, J. L., and Luskey, K. L. (1984) Cell
38,275-286
32. Craik, C. S., Sprang, S., Fletterick, R., and Rutter, W. J. (1982)
Nature (Lond.)299,180-182
33. Goldman, B. M., and Blobel, G. (1981) J. Cell Bioi. 90, 236-242
34. Papermaster, D. S., and Schneider, B. G. (1982) in Cell Biology
of the Eye (McDevitt, D. S.,ed) pp. 475-531, Academic Press,
New York
35. Dratz, E. A., and Hargrave, P. A. (1983) Trends Biochem. Sci. 8 ,
128-131
36. Nathans, J., and Hogness, D. S. (1983) Cell 3 4 , 807-814
37. Brown, D. A., and Simoni, R. D. (1984) Proc. Natl. Acad. Sci. U.
S. A . 81,1674-1678
38. Keith, M.L., Kennelly, P. J., and Rodwell, V. W. (1983) J.
Protein Chem. 2,209-220
39. Orci, L., Brown, M. S., Goldstein, J. L., Garcia-Segura, L.M.,
and Anderson, R. G. W . (1984) Cell 36,335-345
40. Edwards, P. A., Lan, S.-F., Tanaka, R. D., and Fogelman, A. M.
(1983) J. Biol. Chem. 2 5 8 , 7272-7275